Annotation of mandoc/roff.c, Revision 1.373
1.373 ! schwarze 1: /* $Id: roff.c,v 1.372 2020/04/03 11:35:01 schwarze Exp $ */
1.1 kristaps 2: /*
1.372 schwarze 3: * Copyright (c) 2010-2015, 2017-2020 Ingo Schwarze <schwarze@openbsd.org>
1.267 schwarze 4: * Copyright (c) 2008-2012, 2014 Kristaps Dzonsons <kristaps@bsd.lv>
1.1 kristaps 5: *
6: * Permission to use, copy, modify, and distribute this software for any
1.66 kristaps 7: * purpose with or without fee is hereby granted, provided that the above
8: * copyright notice and this permission notice appear in all copies.
1.1 kristaps 9: *
1.106 kristaps 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.66 kristaps 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.106 kristaps 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.66 kristaps 13: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
1.372 schwarze 17: *
18: * Implementation of the roff(7) parser for mandoc(1).
1.1 kristaps 19: */
1.66 kristaps 20: #include "config.h"
1.225 schwarze 21:
22: #include <sys/types.h>
1.30 kristaps 23:
1.67 kristaps 24: #include <assert.h>
1.85 kristaps 25: #include <ctype.h>
1.245 schwarze 26: #include <limits.h>
1.295 schwarze 27: #include <stddef.h>
28: #include <stdint.h>
1.178 schwarze 29: #include <stdio.h>
1.1 kristaps 30: #include <stdlib.h>
1.67 kristaps 31: #include <string.h>
1.1 kristaps 32:
1.201 schwarze 33: #include "mandoc_aux.h"
1.295 schwarze 34: #include "mandoc_ohash.h"
1.345 schwarze 35: #include "mandoc.h"
1.265 schwarze 36: #include "roff.h"
1.349 schwarze 37: #include "mandoc_parse.h"
1.226 schwarze 38: #include "libmandoc.h"
1.266 schwarze 39: #include "roff_int.h"
1.346 schwarze 40: #include "tbl_parse.h"
1.347 schwarze 41: #include "eqn_parse.h"
1.82 kristaps 42:
1.355 schwarze 43: /*
44: * ASCII_ESC is used to signal from roff_getarg() to roff_expand()
45: * that an escape sequence resulted from copy-in processing and
46: * needs to be checked or interpolated. As it is used nowhere
47: * else, it is defined here rather than in a header file.
48: */
49: #define ASCII_ESC 27
50:
1.170 schwarze 51: /* Maximum number of string expansions per line, to break infinite loops. */
52: #define EXPAND_LIMIT 1000
53:
1.315 schwarze 54: /* Types of definitions of macros and strings. */
55: #define ROFFDEF_USER (1 << 1) /* User-defined. */
56: #define ROFFDEF_PRE (1 << 2) /* Predefined. */
57: #define ROFFDEF_REN (1 << 3) /* Renamed standard macro. */
58: #define ROFFDEF_STD (1 << 4) /* mdoc(7) or man(7) macro. */
59: #define ROFFDEF_ANY (ROFFDEF_USER | ROFFDEF_PRE | \
60: ROFFDEF_REN | ROFFDEF_STD)
1.325 schwarze 61: #define ROFFDEF_UNDEF (1 << 5) /* Completely undefined. */
1.315 schwarze 62:
1.266 schwarze 63: /* --- data types --------------------------------------------------------- */
64:
1.147 kristaps 65: /*
1.167 kristaps 66: * An incredibly-simple string buffer.
67: */
1.94 kristaps 68: struct roffstr {
1.167 kristaps 69: char *p; /* nil-terminated buffer */
70: size_t sz; /* saved strlen(p) */
1.166 kristaps 71: };
72:
73: /*
1.167 kristaps 74: * A key-value roffstr pair as part of a singly-linked list.
1.166 kristaps 75: */
76: struct roffkv {
77: struct roffstr key;
78: struct roffstr val;
79: struct roffkv *next; /* next in list */
1.94 kristaps 80: };
81:
1.180 schwarze 82: /*
83: * A single number register as part of a singly-linked list.
84: */
85: struct roffreg {
86: struct roffstr key;
1.181 schwarze 87: int val;
1.327 schwarze 88: int step;
1.180 schwarze 89: struct roffreg *next;
90: };
91:
1.295 schwarze 92: /*
93: * Association of request and macro names with token IDs.
94: */
95: struct roffreq {
96: enum roff_tok tok;
97: char name[];
98: };
99:
1.339 schwarze 100: /*
101: * A macro processing context.
102: * More than one is needed when macro calls are nested.
103: */
104: struct mctx {
105: char **argv;
106: int argc;
107: int argsz;
108: };
109:
1.67 kristaps 110: struct roff {
1.296 schwarze 111: struct roff_man *man; /* mdoc or man parser */
1.67 kristaps 112: struct roffnode *last; /* leaf of stack */
1.339 schwarze 113: struct mctx *mstack; /* stack of macro contexts */
1.223 schwarze 114: int *rstack; /* stack of inverted `ie' values */
1.295 schwarze 115: struct ohash *reqtab; /* request lookup table */
1.180 schwarze 116: struct roffreg *regtab; /* number registers */
1.166 kristaps 117: struct roffkv *strtab; /* user-defined strings & macros */
1.306 schwarze 118: struct roffkv *rentab; /* renamed strings & macros */
1.167 kristaps 119: struct roffkv *xmbtab; /* multi-byte trans table (`tr') */
120: struct roffstr *xtab; /* single-byte trans table (`tr') */
1.106 kristaps 121: const char *current_string; /* value of last called user macro */
1.118 kristaps 122: struct tbl_node *first_tbl; /* first table parsed */
123: struct tbl_node *last_tbl; /* last table parsed */
124: struct tbl_node *tbl; /* current table being parsed */
1.319 schwarze 125: struct eqn_node *last_eqn; /* equation parser */
126: struct eqn_node *eqn; /* active equation parser */
1.230 schwarze 127: int eqn_inline; /* current equation is inline */
1.223 schwarze 128: int options; /* parse options */
1.339 schwarze 129: int mstacksz; /* current size of mstack */
130: int mstackpos; /* position in mstack */
1.223 schwarze 131: int rstacksz; /* current size limit of rstack */
132: int rstackpos; /* position in rstack */
1.227 schwarze 133: int format; /* current file in mdoc or man format */
1.223 schwarze 134: char control; /* control character */
1.303 schwarze 135: char escape; /* escape character */
1.79 kristaps 136: };
137:
1.364 schwarze 138: /*
139: * A macro definition, condition, or ignored block.
140: */
1.67 kristaps 141: struct roffnode {
1.294 schwarze 142: enum roff_tok tok; /* type of node */
1.67 kristaps 143: struct roffnode *parent; /* up one in stack */
144: int line; /* parse line */
145: int col; /* parse col */
1.106 kristaps 146: char *name; /* node name, e.g. macro name */
1.364 schwarze 147: char *end; /* custom end macro of the block */
148: int endspan; /* scope to: 1=eol 2=next line -1=\} */
149: int rule; /* content is: 1=evaluated 0=skipped */
1.67 kristaps 150: };
151:
152: #define ROFF_ARGS struct roff *r, /* parse ctx */ \
1.294 schwarze 153: enum roff_tok tok, /* tok of macro */ \
1.238 schwarze 154: struct buf *buf, /* input buffer */ \
1.67 kristaps 155: int ln, /* parse line */ \
1.75 kristaps 156: int ppos, /* original pos in buffer */ \
157: int pos, /* current pos in buffer */ \
1.74 kristaps 158: int *offs /* reset offset of buffer data */
1.67 kristaps 159:
1.340 schwarze 160: typedef int (*roffproc)(ROFF_ARGS);
1.67 kristaps 161:
162: struct roffmac {
1.79 kristaps 163: roffproc proc; /* process new macro */
164: roffproc text; /* process as child text of macro */
165: roffproc sub; /* process as child of macro */
166: int flags;
167: #define ROFFMAC_STRUCT (1 << 0) /* always interpret */
1.67 kristaps 168: };
169:
1.141 kristaps 170: struct predef {
171: const char *name; /* predefined input name */
172: const char *str; /* replacement symbol */
173: };
174:
175: #define PREDEF(__name, __str) \
176: { (__name), (__str) },
177:
1.266 schwarze 178: /* --- function prototypes ------------------------------------------------ */
179:
1.340 schwarze 180: static int roffnode_cleanscope(struct roff *);
181: static int roffnode_pop(struct roff *);
1.294 schwarze 182: static void roffnode_push(struct roff *, enum roff_tok,
1.155 kristaps 183: const char *, int, int);
1.346 schwarze 184: static void roff_addtbl(struct roff_man *, int, struct tbl_node *);
1.340 schwarze 185: static int roff_als(ROFF_ARGS);
186: static int roff_block(ROFF_ARGS);
187: static int roff_block_text(ROFF_ARGS);
188: static int roff_block_sub(ROFF_ARGS);
1.364 schwarze 189: static int roff_break(ROFF_ARGS);
1.340 schwarze 190: static int roff_cblock(ROFF_ARGS);
191: static int roff_cc(ROFF_ARGS);
192: static int roff_ccond(struct roff *, int, int);
1.341 schwarze 193: static int roff_char(ROFF_ARGS);
1.340 schwarze 194: static int roff_cond(ROFF_ARGS);
195: static int roff_cond_text(ROFF_ARGS);
196: static int roff_cond_sub(ROFF_ARGS);
197: static int roff_ds(ROFF_ARGS);
198: static int roff_ec(ROFF_ARGS);
199: static int roff_eo(ROFF_ARGS);
200: static int roff_eqndelim(struct roff *, struct buf *, int);
1.372 schwarze 201: static int roff_evalcond(struct roff *, int, char *, int *);
1.234 kristaps 202: static int roff_evalnum(struct roff *, int,
203: const char *, int *, int *, int);
204: static int roff_evalpar(struct roff *, int,
1.261 schwarze 205: const char *, int *, int *, int);
1.198 schwarze 206: static int roff_evalstrcond(const char *, int *);
1.355 schwarze 207: static int roff_expand(struct roff *, struct buf *,
208: int, int, char);
1.155 kristaps 209: static void roff_free1(struct roff *);
1.180 schwarze 210: static void roff_freereg(struct roffreg *);
1.167 kristaps 211: static void roff_freestr(struct roffkv *);
1.212 schwarze 212: static size_t roff_getname(struct roff *, char **, int, int);
1.261 schwarze 213: static int roff_getnum(const char *, int *, int *, int);
1.184 schwarze 214: static int roff_getop(const char *, int *, char *);
1.327 schwarze 215: static int roff_getregn(struct roff *,
216: const char *, size_t, char);
1.273 schwarze 217: static int roff_getregro(const struct roff *,
218: const char *name);
1.325 schwarze 219: static const char *roff_getstrn(struct roff *,
1.315 schwarze 220: const char *, size_t, int *);
1.271 schwarze 221: static int roff_hasregn(const struct roff *,
222: const char *, size_t);
1.340 schwarze 223: static int roff_insec(ROFF_ARGS);
224: static int roff_it(ROFF_ARGS);
225: static int roff_line_ignore(ROFF_ARGS);
1.265 schwarze 226: static void roff_man_alloc1(struct roff_man *);
227: static void roff_man_free1(struct roff_man *);
1.340 schwarze 228: static int roff_manyarg(ROFF_ARGS);
1.358 schwarze 229: static int roff_noarg(ROFF_ARGS);
1.340 schwarze 230: static int roff_nop(ROFF_ARGS);
231: static int roff_nr(ROFF_ARGS);
232: static int roff_onearg(ROFF_ARGS);
1.294 schwarze 233: static enum roff_tok roff_parse(struct roff *, char *, int *,
1.214 schwarze 234: int, int);
1.340 schwarze 235: static int roff_parsetext(struct roff *, struct buf *,
1.305 schwarze 236: int, int *);
1.340 schwarze 237: static int roff_renamed(ROFF_ARGS);
238: static int roff_return(ROFF_ARGS);
239: static int roff_rm(ROFF_ARGS);
240: static int roff_rn(ROFF_ARGS);
241: static int roff_rr(ROFF_ARGS);
1.326 schwarze 242: static void roff_setregn(struct roff *, const char *,
1.327 schwarze 243: size_t, int, char, int);
1.94 kristaps 244: static void roff_setstr(struct roff *,
1.106 kristaps 245: const char *, const char *, int);
1.207 schwarze 246: static void roff_setstrn(struct roffkv **, const char *,
1.164 kristaps 247: size_t, const char *, size_t, int);
1.340 schwarze 248: static int roff_shift(ROFF_ARGS);
249: static int roff_so(ROFF_ARGS);
250: static int roff_tr(ROFF_ARGS);
251: static int roff_Dd(ROFF_ARGS);
252: static int roff_TE(ROFF_ARGS);
253: static int roff_TS(ROFF_ARGS);
254: static int roff_EQ(ROFF_ARGS);
255: static int roff_EN(ROFF_ARGS);
256: static int roff_T_(ROFF_ARGS);
257: static int roff_unsupp(ROFF_ARGS);
258: static int roff_userdef(ROFF_ARGS);
1.67 kristaps 259:
1.266 schwarze 260: /* --- constant data ------------------------------------------------------ */
261:
1.261 schwarze 262: #define ROFFNUM_SCALE (1 << 0) /* Honour scaling in roff_getnum(). */
263: #define ROFFNUM_WHITE (1 << 1) /* Skip whitespace in roff_evalnum(). */
264:
1.294 schwarze 265: const char *__roff_name[MAN_MAX + 1] = {
1.358 schwarze 266: "br", "ce", "fi", "ft",
267: "ll", "mc", "nf",
268: "po", "rj", "sp",
1.312 schwarze 269: "ta", "ti", NULL,
1.294 schwarze 270: "ab", "ad", "af", "aln",
271: "als", "am", "am1", "ami",
272: "ami1", "as", "as1", "asciify",
273: "backtrace", "bd", "bleedat", "blm",
274: "box", "boxa", "bp", "BP",
275: "break", "breakchar", "brnl", "brp",
1.305 schwarze 276: "brpnl", "c2", "cc",
1.294 schwarze 277: "cf", "cflags", "ch", "char",
278: "chop", "class", "close", "CL",
279: "color", "composite", "continue", "cp",
280: "cropat", "cs", "cu", "da",
281: "dch", "Dd", "de", "de1",
282: "defcolor", "dei", "dei1", "device",
283: "devicem", "di", "do", "ds",
284: "ds1", "dwh", "dt", "ec",
285: "ecr", "ecs", "el", "em",
286: "EN", "eo", "EP", "EQ",
287: "errprint", "ev", "evc", "ex",
288: "fallback", "fam", "fc", "fchar",
289: "fcolor", "fdeferlig", "feature", "fkern",
290: "fl", "flig", "fp", "fps",
291: "fschar", "fspacewidth", "fspecial", "ftr",
292: "fzoom", "gcolor", "hc", "hcode",
293: "hidechar", "hla", "hlm", "hpf",
294: "hpfa", "hpfcode", "hw", "hy",
295: "hylang", "hylen", "hym", "hypp",
296: "hys", "ie", "if", "ig",
297: "index", "it", "itc", "IX",
298: "kern", "kernafter", "kernbefore", "kernpair",
299: "lc", "lc_ctype", "lds", "length",
300: "letadj", "lf", "lg", "lhang",
301: "linetabs", "lnr", "lnrf", "lpfx",
1.304 schwarze 302: "ls", "lsm", "lt",
1.294 schwarze 303: "mediasize", "minss", "mk", "mso",
304: "na", "ne", "nh", "nhychar",
305: "nm", "nn", "nop", "nr",
306: "nrf", "nroff", "ns", "nx",
307: "open", "opena", "os", "output",
308: "padj", "papersize", "pc", "pev",
309: "pi", "PI", "pl", "pm",
1.312 schwarze 310: "pn", "pnr", "ps",
1.294 schwarze 311: "psbb", "pshape", "pso", "ptr",
312: "pvs", "rchar", "rd", "recursionlimit",
1.309 schwarze 313: "return", "rfschar", "rhang",
1.294 schwarze 314: "rm", "rn", "rnn", "rr",
315: "rs", "rt", "schar", "sentchar",
316: "shc", "shift", "sizes", "so",
317: "spacewidth", "special", "spreadwarn", "ss",
318: "sty", "substring", "sv", "sy",
1.300 schwarze 319: "T&", "tc", "TE",
1.301 schwarze 320: "TH", "tkf", "tl",
1.294 schwarze 321: "tm", "tm1", "tmc", "tr",
322: "track", "transchar", "trf", "trimat",
323: "trin", "trnt", "troff", "TS",
324: "uf", "ul", "unformat", "unwatch",
325: "unwatchn", "vpt", "vs", "warn",
326: "warnscale", "watch", "watchlength", "watchn",
327: "wh", "while", "write", "writec",
328: "writem", "xflag", ".", NULL,
1.306 schwarze 329: NULL, "text",
1.294 schwarze 330: "Dd", "Dt", "Os", "Sh",
331: "Ss", "Pp", "D1", "Dl",
332: "Bd", "Ed", "Bl", "El",
333: "It", "Ad", "An", "Ap",
334: "Ar", "Cd", "Cm", "Dv",
335: "Er", "Ev", "Ex", "Fa",
336: "Fd", "Fl", "Fn", "Ft",
337: "Ic", "In", "Li", "Nd",
338: "Nm", "Op", "Ot", "Pa",
339: "Rv", "St", "Va", "Vt",
340: "Xr", "%A", "%B", "%D",
341: "%I", "%J", "%N", "%O",
342: "%P", "%R", "%T", "%V",
343: "Ac", "Ao", "Aq", "At",
344: "Bc", "Bf", "Bo", "Bq",
345: "Bsx", "Bx", "Db", "Dc",
346: "Do", "Dq", "Ec", "Ef",
347: "Em", "Eo", "Fx", "Ms",
348: "No", "Ns", "Nx", "Ox",
349: "Pc", "Pf", "Po", "Pq",
350: "Qc", "Ql", "Qo", "Qq",
351: "Re", "Rs", "Sc", "So",
352: "Sq", "Sm", "Sx", "Sy",
353: "Tn", "Ux", "Xc", "Xo",
354: "Fo", "Fc", "Oo", "Oc",
355: "Bk", "Ek", "Bt", "Hf",
356: "Fr", "Ud", "Lb", "Lp",
357: "Lk", "Mt", "Brq", "Bro",
358: "Brc", "%C", "Es", "En",
1.299 schwarze 359: "Dx", "%Q", "%U", "Ta",
1.369 schwarze 360: "Tg", NULL,
1.294 schwarze 361: "TH", "SH", "SS", "TP",
1.332 schwarze 362: "TQ",
1.294 schwarze 363: "LP", "PP", "P", "IP",
364: "HP", "SM", "SB", "BI",
365: "IB", "BR", "RB", "R",
366: "B", "I", "IR", "RI",
367: "RE", "RS", "DT", "UC",
1.297 schwarze 368: "PD", "AT", "in",
1.333 schwarze 369: "SY", "YS", "OP",
370: "EX", "EE", "UR",
1.317 schwarze 371: "UE", "MT", "ME", NULL
1.294 schwarze 372: };
373: const char *const *roff_name = __roff_name;
374:
375: static struct roffmac roffs[TOKEN_NONE] = {
1.358 schwarze 376: { roff_noarg, NULL, NULL, 0 }, /* br */
1.305 schwarze 377: { roff_onearg, NULL, NULL, 0 }, /* ce */
1.358 schwarze 378: { roff_noarg, NULL, NULL, 0 }, /* fi */
1.297 schwarze 379: { roff_onearg, NULL, NULL, 0 }, /* ft */
1.298 schwarze 380: { roff_onearg, NULL, NULL, 0 }, /* ll */
1.304 schwarze 381: { roff_onearg, NULL, NULL, 0 }, /* mc */
1.358 schwarze 382: { roff_noarg, NULL, NULL, 0 }, /* nf */
1.312 schwarze 383: { roff_onearg, NULL, NULL, 0 }, /* po */
1.309 schwarze 384: { roff_onearg, NULL, NULL, 0 }, /* rj */
1.299 schwarze 385: { roff_onearg, NULL, NULL, 0 }, /* sp */
1.300 schwarze 386: { roff_manyarg, NULL, NULL, 0 }, /* ta */
1.301 schwarze 387: { roff_onearg, NULL, NULL, 0 }, /* ti */
1.296 schwarze 388: { NULL, NULL, NULL, 0 }, /* ROFF_MAX */
1.295 schwarze 389: { roff_unsupp, NULL, NULL, 0 }, /* ab */
390: { roff_line_ignore, NULL, NULL, 0 }, /* ad */
391: { roff_line_ignore, NULL, NULL, 0 }, /* af */
392: { roff_unsupp, NULL, NULL, 0 }, /* aln */
1.311 schwarze 393: { roff_als, NULL, NULL, 0 }, /* als */
1.295 schwarze 394: { roff_block, roff_block_text, roff_block_sub, 0 }, /* am */
395: { roff_block, roff_block_text, roff_block_sub, 0 }, /* am1 */
396: { roff_block, roff_block_text, roff_block_sub, 0 }, /* ami */
397: { roff_block, roff_block_text, roff_block_sub, 0 }, /* ami1 */
398: { roff_ds, NULL, NULL, 0 }, /* as */
399: { roff_ds, NULL, NULL, 0 }, /* as1 */
400: { roff_unsupp, NULL, NULL, 0 }, /* asciify */
401: { roff_line_ignore, NULL, NULL, 0 }, /* backtrace */
402: { roff_line_ignore, NULL, NULL, 0 }, /* bd */
403: { roff_line_ignore, NULL, NULL, 0 }, /* bleedat */
404: { roff_unsupp, NULL, NULL, 0 }, /* blm */
405: { roff_unsupp, NULL, NULL, 0 }, /* box */
406: { roff_unsupp, NULL, NULL, 0 }, /* boxa */
407: { roff_line_ignore, NULL, NULL, 0 }, /* bp */
408: { roff_unsupp, NULL, NULL, 0 }, /* BP */
1.364 schwarze 409: { roff_break, NULL, NULL, 0 }, /* break */
1.295 schwarze 410: { roff_line_ignore, NULL, NULL, 0 }, /* breakchar */
411: { roff_line_ignore, NULL, NULL, 0 }, /* brnl */
1.358 schwarze 412: { roff_noarg, NULL, NULL, 0 }, /* brp */
1.295 schwarze 413: { roff_line_ignore, NULL, NULL, 0 }, /* brpnl */
414: { roff_unsupp, NULL, NULL, 0 }, /* c2 */
415: { roff_cc, NULL, NULL, 0 }, /* cc */
416: { roff_insec, NULL, NULL, 0 }, /* cf */
417: { roff_line_ignore, NULL, NULL, 0 }, /* cflags */
418: { roff_line_ignore, NULL, NULL, 0 }, /* ch */
1.341 schwarze 419: { roff_char, NULL, NULL, 0 }, /* char */
1.295 schwarze 420: { roff_unsupp, NULL, NULL, 0 }, /* chop */
421: { roff_line_ignore, NULL, NULL, 0 }, /* class */
422: { roff_insec, NULL, NULL, 0 }, /* close */
423: { roff_unsupp, NULL, NULL, 0 }, /* CL */
424: { roff_line_ignore, NULL, NULL, 0 }, /* color */
425: { roff_unsupp, NULL, NULL, 0 }, /* composite */
426: { roff_unsupp, NULL, NULL, 0 }, /* continue */
427: { roff_line_ignore, NULL, NULL, 0 }, /* cp */
428: { roff_line_ignore, NULL, NULL, 0 }, /* cropat */
429: { roff_line_ignore, NULL, NULL, 0 }, /* cs */
430: { roff_line_ignore, NULL, NULL, 0 }, /* cu */
431: { roff_unsupp, NULL, NULL, 0 }, /* da */
432: { roff_unsupp, NULL, NULL, 0 }, /* dch */
433: { roff_Dd, NULL, NULL, 0 }, /* Dd */
434: { roff_block, roff_block_text, roff_block_sub, 0 }, /* de */
435: { roff_block, roff_block_text, roff_block_sub, 0 }, /* de1 */
436: { roff_line_ignore, NULL, NULL, 0 }, /* defcolor */
437: { roff_block, roff_block_text, roff_block_sub, 0 }, /* dei */
438: { roff_block, roff_block_text, roff_block_sub, 0 }, /* dei1 */
439: { roff_unsupp, NULL, NULL, 0 }, /* device */
440: { roff_unsupp, NULL, NULL, 0 }, /* devicem */
441: { roff_unsupp, NULL, NULL, 0 }, /* di */
442: { roff_unsupp, NULL, NULL, 0 }, /* do */
443: { roff_ds, NULL, NULL, 0 }, /* ds */
444: { roff_ds, NULL, NULL, 0 }, /* ds1 */
445: { roff_unsupp, NULL, NULL, 0 }, /* dwh */
446: { roff_unsupp, NULL, NULL, 0 }, /* dt */
1.303 schwarze 447: { roff_ec, NULL, NULL, 0 }, /* ec */
1.295 schwarze 448: { roff_unsupp, NULL, NULL, 0 }, /* ecr */
449: { roff_unsupp, NULL, NULL, 0 }, /* ecs */
450: { roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT }, /* el */
451: { roff_unsupp, NULL, NULL, 0 }, /* em */
452: { roff_EN, NULL, NULL, 0 }, /* EN */
1.303 schwarze 453: { roff_eo, NULL, NULL, 0 }, /* eo */
1.295 schwarze 454: { roff_unsupp, NULL, NULL, 0 }, /* EP */
455: { roff_EQ, NULL, NULL, 0 }, /* EQ */
456: { roff_line_ignore, NULL, NULL, 0 }, /* errprint */
457: { roff_unsupp, NULL, NULL, 0 }, /* ev */
458: { roff_unsupp, NULL, NULL, 0 }, /* evc */
459: { roff_unsupp, NULL, NULL, 0 }, /* ex */
460: { roff_line_ignore, NULL, NULL, 0 }, /* fallback */
461: { roff_line_ignore, NULL, NULL, 0 }, /* fam */
462: { roff_unsupp, NULL, NULL, 0 }, /* fc */
463: { roff_unsupp, NULL, NULL, 0 }, /* fchar */
464: { roff_line_ignore, NULL, NULL, 0 }, /* fcolor */
465: { roff_line_ignore, NULL, NULL, 0 }, /* fdeferlig */
466: { roff_line_ignore, NULL, NULL, 0 }, /* feature */
467: { roff_line_ignore, NULL, NULL, 0 }, /* fkern */
468: { roff_line_ignore, NULL, NULL, 0 }, /* fl */
469: { roff_line_ignore, NULL, NULL, 0 }, /* flig */
470: { roff_line_ignore, NULL, NULL, 0 }, /* fp */
471: { roff_line_ignore, NULL, NULL, 0 }, /* fps */
472: { roff_unsupp, NULL, NULL, 0 }, /* fschar */
473: { roff_line_ignore, NULL, NULL, 0 }, /* fspacewidth */
474: { roff_line_ignore, NULL, NULL, 0 }, /* fspecial */
475: { roff_line_ignore, NULL, NULL, 0 }, /* ftr */
476: { roff_line_ignore, NULL, NULL, 0 }, /* fzoom */
477: { roff_line_ignore, NULL, NULL, 0 }, /* gcolor */
478: { roff_line_ignore, NULL, NULL, 0 }, /* hc */
479: { roff_line_ignore, NULL, NULL, 0 }, /* hcode */
480: { roff_line_ignore, NULL, NULL, 0 }, /* hidechar */
481: { roff_line_ignore, NULL, NULL, 0 }, /* hla */
482: { roff_line_ignore, NULL, NULL, 0 }, /* hlm */
483: { roff_line_ignore, NULL, NULL, 0 }, /* hpf */
484: { roff_line_ignore, NULL, NULL, 0 }, /* hpfa */
485: { roff_line_ignore, NULL, NULL, 0 }, /* hpfcode */
486: { roff_line_ignore, NULL, NULL, 0 }, /* hw */
487: { roff_line_ignore, NULL, NULL, 0 }, /* hy */
488: { roff_line_ignore, NULL, NULL, 0 }, /* hylang */
489: { roff_line_ignore, NULL, NULL, 0 }, /* hylen */
490: { roff_line_ignore, NULL, NULL, 0 }, /* hym */
491: { roff_line_ignore, NULL, NULL, 0 }, /* hypp */
492: { roff_line_ignore, NULL, NULL, 0 }, /* hys */
493: { roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT }, /* ie */
494: { roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT }, /* if */
495: { roff_block, roff_block_text, roff_block_sub, 0 }, /* ig */
496: { roff_unsupp, NULL, NULL, 0 }, /* index */
497: { roff_it, NULL, NULL, 0 }, /* it */
498: { roff_unsupp, NULL, NULL, 0 }, /* itc */
499: { roff_line_ignore, NULL, NULL, 0 }, /* IX */
500: { roff_line_ignore, NULL, NULL, 0 }, /* kern */
501: { roff_line_ignore, NULL, NULL, 0 }, /* kernafter */
502: { roff_line_ignore, NULL, NULL, 0 }, /* kernbefore */
503: { roff_line_ignore, NULL, NULL, 0 }, /* kernpair */
504: { roff_unsupp, NULL, NULL, 0 }, /* lc */
505: { roff_unsupp, NULL, NULL, 0 }, /* lc_ctype */
506: { roff_unsupp, NULL, NULL, 0 }, /* lds */
507: { roff_unsupp, NULL, NULL, 0 }, /* length */
508: { roff_line_ignore, NULL, NULL, 0 }, /* letadj */
509: { roff_insec, NULL, NULL, 0 }, /* lf */
510: { roff_line_ignore, NULL, NULL, 0 }, /* lg */
511: { roff_line_ignore, NULL, NULL, 0 }, /* lhang */
512: { roff_unsupp, NULL, NULL, 0 }, /* linetabs */
513: { roff_unsupp, NULL, NULL, 0 }, /* lnr */
514: { roff_unsupp, NULL, NULL, 0 }, /* lnrf */
515: { roff_unsupp, NULL, NULL, 0 }, /* lpfx */
516: { roff_line_ignore, NULL, NULL, 0 }, /* ls */
517: { roff_unsupp, NULL, NULL, 0 }, /* lsm */
518: { roff_line_ignore, NULL, NULL, 0 }, /* lt */
519: { roff_line_ignore, NULL, NULL, 0 }, /* mediasize */
520: { roff_line_ignore, NULL, NULL, 0 }, /* minss */
521: { roff_line_ignore, NULL, NULL, 0 }, /* mk */
522: { roff_insec, NULL, NULL, 0 }, /* mso */
523: { roff_line_ignore, NULL, NULL, 0 }, /* na */
524: { roff_line_ignore, NULL, NULL, 0 }, /* ne */
525: { roff_line_ignore, NULL, NULL, 0 }, /* nh */
526: { roff_line_ignore, NULL, NULL, 0 }, /* nhychar */
527: { roff_unsupp, NULL, NULL, 0 }, /* nm */
528: { roff_unsupp, NULL, NULL, 0 }, /* nn */
1.330 schwarze 529: { roff_nop, NULL, NULL, 0 }, /* nop */
1.295 schwarze 530: { roff_nr, NULL, NULL, 0 }, /* nr */
531: { roff_unsupp, NULL, NULL, 0 }, /* nrf */
532: { roff_line_ignore, NULL, NULL, 0 }, /* nroff */
533: { roff_line_ignore, NULL, NULL, 0 }, /* ns */
534: { roff_insec, NULL, NULL, 0 }, /* nx */
535: { roff_insec, NULL, NULL, 0 }, /* open */
536: { roff_insec, NULL, NULL, 0 }, /* opena */
537: { roff_line_ignore, NULL, NULL, 0 }, /* os */
538: { roff_unsupp, NULL, NULL, 0 }, /* output */
539: { roff_line_ignore, NULL, NULL, 0 }, /* padj */
540: { roff_line_ignore, NULL, NULL, 0 }, /* papersize */
541: { roff_line_ignore, NULL, NULL, 0 }, /* pc */
542: { roff_line_ignore, NULL, NULL, 0 }, /* pev */
543: { roff_insec, NULL, NULL, 0 }, /* pi */
544: { roff_unsupp, NULL, NULL, 0 }, /* PI */
545: { roff_line_ignore, NULL, NULL, 0 }, /* pl */
546: { roff_line_ignore, NULL, NULL, 0 }, /* pm */
547: { roff_line_ignore, NULL, NULL, 0 }, /* pn */
548: { roff_line_ignore, NULL, NULL, 0 }, /* pnr */
549: { roff_line_ignore, NULL, NULL, 0 }, /* ps */
550: { roff_unsupp, NULL, NULL, 0 }, /* psbb */
551: { roff_unsupp, NULL, NULL, 0 }, /* pshape */
552: { roff_insec, NULL, NULL, 0 }, /* pso */
553: { roff_line_ignore, NULL, NULL, 0 }, /* ptr */
554: { roff_line_ignore, NULL, NULL, 0 }, /* pvs */
555: { roff_unsupp, NULL, NULL, 0 }, /* rchar */
556: { roff_line_ignore, NULL, NULL, 0 }, /* rd */
557: { roff_line_ignore, NULL, NULL, 0 }, /* recursionlimit */
1.339 schwarze 558: { roff_return, NULL, NULL, 0 }, /* return */
1.295 schwarze 559: { roff_unsupp, NULL, NULL, 0 }, /* rfschar */
560: { roff_line_ignore, NULL, NULL, 0 }, /* rhang */
561: { roff_rm, NULL, NULL, 0 }, /* rm */
1.306 schwarze 562: { roff_rn, NULL, NULL, 0 }, /* rn */
1.295 schwarze 563: { roff_unsupp, NULL, NULL, 0 }, /* rnn */
564: { roff_rr, NULL, NULL, 0 }, /* rr */
565: { roff_line_ignore, NULL, NULL, 0 }, /* rs */
566: { roff_line_ignore, NULL, NULL, 0 }, /* rt */
567: { roff_unsupp, NULL, NULL, 0 }, /* schar */
568: { roff_line_ignore, NULL, NULL, 0 }, /* sentchar */
569: { roff_line_ignore, NULL, NULL, 0 }, /* shc */
1.339 schwarze 570: { roff_shift, NULL, NULL, 0 }, /* shift */
1.295 schwarze 571: { roff_line_ignore, NULL, NULL, 0 }, /* sizes */
572: { roff_so, NULL, NULL, 0 }, /* so */
573: { roff_line_ignore, NULL, NULL, 0 }, /* spacewidth */
574: { roff_line_ignore, NULL, NULL, 0 }, /* special */
575: { roff_line_ignore, NULL, NULL, 0 }, /* spreadwarn */
576: { roff_line_ignore, NULL, NULL, 0 }, /* ss */
577: { roff_line_ignore, NULL, NULL, 0 }, /* sty */
578: { roff_unsupp, NULL, NULL, 0 }, /* substring */
579: { roff_line_ignore, NULL, NULL, 0 }, /* sv */
580: { roff_insec, NULL, NULL, 0 }, /* sy */
581: { roff_T_, NULL, NULL, 0 }, /* T& */
582: { roff_unsupp, NULL, NULL, 0 }, /* tc */
583: { roff_TE, NULL, NULL, 0 }, /* TE */
1.315 schwarze 584: { roff_Dd, NULL, NULL, 0 }, /* TH */
1.295 schwarze 585: { roff_line_ignore, NULL, NULL, 0 }, /* tkf */
586: { roff_unsupp, NULL, NULL, 0 }, /* tl */
587: { roff_line_ignore, NULL, NULL, 0 }, /* tm */
588: { roff_line_ignore, NULL, NULL, 0 }, /* tm1 */
589: { roff_line_ignore, NULL, NULL, 0 }, /* tmc */
590: { roff_tr, NULL, NULL, 0 }, /* tr */
591: { roff_line_ignore, NULL, NULL, 0 }, /* track */
592: { roff_line_ignore, NULL, NULL, 0 }, /* transchar */
593: { roff_insec, NULL, NULL, 0 }, /* trf */
594: { roff_line_ignore, NULL, NULL, 0 }, /* trimat */
595: { roff_unsupp, NULL, NULL, 0 }, /* trin */
596: { roff_unsupp, NULL, NULL, 0 }, /* trnt */
597: { roff_line_ignore, NULL, NULL, 0 }, /* troff */
598: { roff_TS, NULL, NULL, 0 }, /* TS */
599: { roff_line_ignore, NULL, NULL, 0 }, /* uf */
600: { roff_line_ignore, NULL, NULL, 0 }, /* ul */
601: { roff_unsupp, NULL, NULL, 0 }, /* unformat */
602: { roff_line_ignore, NULL, NULL, 0 }, /* unwatch */
603: { roff_line_ignore, NULL, NULL, 0 }, /* unwatchn */
604: { roff_line_ignore, NULL, NULL, 0 }, /* vpt */
605: { roff_line_ignore, NULL, NULL, 0 }, /* vs */
606: { roff_line_ignore, NULL, NULL, 0 }, /* warn */
607: { roff_line_ignore, NULL, NULL, 0 }, /* warnscale */
608: { roff_line_ignore, NULL, NULL, 0 }, /* watch */
609: { roff_line_ignore, NULL, NULL, 0 }, /* watchlength */
610: { roff_line_ignore, NULL, NULL, 0 }, /* watchn */
611: { roff_unsupp, NULL, NULL, 0 }, /* wh */
1.340 schwarze 612: { roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT }, /*while*/
1.295 schwarze 613: { roff_insec, NULL, NULL, 0 }, /* write */
614: { roff_insec, NULL, NULL, 0 }, /* writec */
615: { roff_insec, NULL, NULL, 0 }, /* writem */
616: { roff_line_ignore, NULL, NULL, 0 }, /* xflag */
617: { roff_cblock, NULL, NULL, 0 }, /* . */
1.306 schwarze 618: { roff_renamed, NULL, NULL, 0 },
1.295 schwarze 619: { roff_userdef, NULL, NULL, 0 }
1.67 kristaps 620: };
621:
1.141 kristaps 622: /* Array of injected predefined strings. */
623: #define PREDEFS_MAX 38
624: static const struct predef predefs[PREDEFS_MAX] = {
625: #include "predefs.in"
626: };
627:
1.305 schwarze 628: static int roffce_lines; /* number of input lines to center */
629: static struct roff_node *roffce_node; /* active request */
1.178 schwarze 630: static int roffit_lines; /* number of lines to delay */
631: static char *roffit_macro; /* nil-terminated macro line */
632:
1.207 schwarze 633:
1.266 schwarze 634: /* --- request table ------------------------------------------------------ */
635:
1.295 schwarze 636: struct ohash *
637: roffhash_alloc(enum roff_tok mintok, enum roff_tok maxtok)
1.85 kristaps 638: {
1.295 schwarze 639: struct ohash *htab;
640: struct roffreq *req;
641: enum roff_tok tok;
642: size_t sz;
643: unsigned int slot;
1.85 kristaps 644:
1.295 schwarze 645: htab = mandoc_malloc(sizeof(*htab));
646: mandoc_ohash_init(htab, 8, offsetof(struct roffreq, name));
1.85 kristaps 647:
1.295 schwarze 648: for (tok = mintok; tok < maxtok; tok++) {
1.296 schwarze 649: if (roff_name[tok] == NULL)
650: continue;
1.295 schwarze 651: sz = strlen(roff_name[tok]);
652: req = mandoc_malloc(sizeof(*req) + sz + 1);
653: req->tok = tok;
654: memcpy(req->name, roff_name[tok], sz + 1);
655: slot = ohash_qlookup(htab, req->name);
656: ohash_insert(htab, slot, req);
1.85 kristaps 657: }
1.295 schwarze 658: return htab;
1.85 kristaps 659: }
660:
1.295 schwarze 661: void
662: roffhash_free(struct ohash *htab)
1.67 kristaps 663: {
1.295 schwarze 664: struct roffreq *req;
665: unsigned int slot;
1.67 kristaps 666:
1.295 schwarze 667: if (htab == NULL)
668: return;
669: for (req = ohash_first(htab, &slot); req != NULL;
670: req = ohash_next(htab, &slot))
671: free(req);
672: ohash_delete(htab);
673: free(htab);
674: }
675:
676: enum roff_tok
677: roffhash_find(struct ohash *htab, const char *name, size_t sz)
678: {
679: struct roffreq *req;
680: const char *end;
681:
682: if (sz) {
683: end = name + sz;
684: req = ohash_find(htab, ohash_qlookupi(htab, name, &end));
685: } else
686: req = ohash_find(htab, ohash_qlookup(htab, name));
687: return req == NULL ? TOKEN_NONE : req->tok;
1.67 kristaps 688: }
689:
1.266 schwarze 690: /* --- stack of request blocks -------------------------------------------- */
691:
1.67 kristaps 692: /*
693: * Pop the current node off of the stack of roff instructions currently
1.364 schwarze 694: * pending. Return 1 if it is a loop or 0 otherwise.
1.67 kristaps 695: */
1.340 schwarze 696: static int
1.67 kristaps 697: roffnode_pop(struct roff *r)
698: {
699: struct roffnode *p;
1.340 schwarze 700: int inloop;
1.67 kristaps 701:
1.207 schwarze 702: p = r->last;
1.340 schwarze 703: inloop = p->tok == ROFF_while;
704: r->last = p->parent;
1.106 kristaps 705: free(p->name);
706: free(p->end);
1.67 kristaps 707: free(p);
1.340 schwarze 708: return inloop;
1.67 kristaps 709: }
710:
711: /*
712: * Push a roff node onto the instruction stack. This must later be
713: * removed with roffnode_pop().
714: */
1.98 schwarze 715: static void
1.294 schwarze 716: roffnode_push(struct roff *r, enum roff_tok tok, const char *name,
1.106 kristaps 717: int line, int col)
1.67 kristaps 718: {
719: struct roffnode *p;
720:
1.98 schwarze 721: p = mandoc_calloc(1, sizeof(struct roffnode));
1.67 kristaps 722: p->tok = tok;
1.106 kristaps 723: if (name)
724: p->name = mandoc_strdup(name);
1.67 kristaps 725: p->parent = r->last;
726: p->line = line;
727: p->col = col;
1.198 schwarze 728: p->rule = p->parent ? p->parent->rule : 0;
1.67 kristaps 729:
730: r->last = p;
731: }
732:
1.266 schwarze 733: /* --- roff parser state data management ---------------------------------- */
734:
1.67 kristaps 735: static void
736: roff_free1(struct roff *r)
737: {
1.167 kristaps 738: int i;
1.67 kristaps 739:
1.346 schwarze 740: tbl_free(r->first_tbl);
1.113 kristaps 741: r->first_tbl = r->last_tbl = r->tbl = NULL;
742:
1.347 schwarze 743: eqn_free(r->last_eqn);
1.319 schwarze 744: r->last_eqn = r->eqn = NULL;
1.125 kristaps 745:
1.339 schwarze 746: while (r->mstackpos >= 0)
747: roff_userret(r);
748:
1.67 kristaps 749: while (r->last)
750: roffnode_pop(r);
1.109 kristaps 751:
1.223 schwarze 752: free (r->rstack);
753: r->rstack = NULL;
754: r->rstacksz = 0;
755: r->rstackpos = -1;
756:
757: roff_freereg(r->regtab);
758: r->regtab = NULL;
759:
1.167 kristaps 760: roff_freestr(r->strtab);
1.306 schwarze 761: roff_freestr(r->rentab);
1.167 kristaps 762: roff_freestr(r->xmbtab);
1.306 schwarze 763: r->strtab = r->rentab = r->xmbtab = NULL;
1.167 kristaps 764:
765: if (r->xtab)
766: for (i = 0; i < 128; i++)
767: free(r->xtab[i].p);
768: free(r->xtab);
769: r->xtab = NULL;
1.67 kristaps 770: }
771:
772: void
773: roff_reset(struct roff *r)
774: {
775: roff_free1(r);
1.367 schwarze 776: r->options |= MPARSE_COMMENT;
1.227 schwarze 777: r->format = r->options & (MPARSE_MDOC | MPARSE_MAN);
1.303 schwarze 778: r->control = '\0';
779: r->escape = '\\';
1.307 schwarze 780: roffce_lines = 0;
781: roffce_node = NULL;
782: roffit_lines = 0;
783: roffit_macro = NULL;
1.67 kristaps 784: }
785:
786: void
787: roff_free(struct roff *r)
788: {
1.366 schwarze 789: int i;
1.339 schwarze 790:
1.67 kristaps 791: roff_free1(r);
1.339 schwarze 792: for (i = 0; i < r->mstacksz; i++)
793: free(r->mstack[i].argv);
794: free(r->mstack);
1.295 schwarze 795: roffhash_free(r->reqtab);
1.67 kristaps 796: free(r);
797: }
798:
799: struct roff *
1.351 schwarze 800: roff_alloc(int options)
1.67 kristaps 801: {
802: struct roff *r;
803:
1.98 schwarze 804: r = mandoc_calloc(1, sizeof(struct roff));
1.328 schwarze 805: r->reqtab = roffhash_alloc(0, ROFF_RENAMED);
1.367 schwarze 806: r->options = options | MPARSE_COMMENT;
1.227 schwarze 807: r->format = options & (MPARSE_MDOC | MPARSE_MAN);
1.339 schwarze 808: r->mstackpos = -1;
1.82 kristaps 809: r->rstackpos = -1;
1.303 schwarze 810: r->escape = '\\';
1.277 schwarze 811: return r;
1.265 schwarze 812: }
813:
1.266 schwarze 814: /* --- syntax tree state data management ---------------------------------- */
815:
1.265 schwarze 816: static void
817: roff_man_free1(struct roff_man *man)
818: {
1.356 schwarze 819: if (man->meta.first != NULL)
820: roff_node_delete(man, man->meta.first);
1.265 schwarze 821: free(man->meta.msec);
822: free(man->meta.vol);
823: free(man->meta.os);
824: free(man->meta.arch);
825: free(man->meta.title);
826: free(man->meta.name);
827: free(man->meta.date);
1.356 schwarze 828: free(man->meta.sodest);
1.265 schwarze 829: }
830:
1.357 schwarze 831: void
832: roff_state_reset(struct roff_man *man)
833: {
834: man->last = man->meta.first;
835: man->last_es = NULL;
836: man->flags = 0;
837: man->lastsec = man->lastnamed = SEC_NONE;
838: man->next = ROFF_NEXT_CHILD;
839: roff_setreg(man->roff, "nS", 0, '=');
840: }
841:
1.265 schwarze 842: static void
843: roff_man_alloc1(struct roff_man *man)
844: {
845: memset(&man->meta, 0, sizeof(man->meta));
1.356 schwarze 846: man->meta.first = mandoc_calloc(1, sizeof(*man->meta.first));
847: man->meta.first->type = ROFFT_ROOT;
848: man->meta.macroset = MACROSET_NONE;
1.357 schwarze 849: roff_state_reset(man);
1.265 schwarze 850: }
851:
852: void
853: roff_man_reset(struct roff_man *man)
854: {
855: roff_man_free1(man);
856: roff_man_alloc1(man);
857: }
858:
859: void
860: roff_man_free(struct roff_man *man)
861: {
862: roff_man_free1(man);
863: free(man);
864: }
865:
866: struct roff_man *
1.351 schwarze 867: roff_man_alloc(struct roff *roff, const char *os_s, int quick)
1.265 schwarze 868: {
869: struct roff_man *man;
870:
871: man = mandoc_calloc(1, sizeof(*man));
872: man->roff = roff;
1.316 schwarze 873: man->os_s = os_s;
1.265 schwarze 874: man->quick = quick;
875: roff_man_alloc1(man);
1.296 schwarze 876: roff->man = man;
1.277 schwarze 877: return man;
1.67 kristaps 878: }
879:
1.266 schwarze 880: /* --- syntax tree handling ----------------------------------------------- */
881:
882: struct roff_node *
883: roff_node_alloc(struct roff_man *man, int line, int pos,
884: enum roff_type type, int tok)
885: {
886: struct roff_node *n;
887:
888: n = mandoc_calloc(1, sizeof(*n));
889: n->line = line;
890: n->pos = pos;
891: n->tok = tok;
892: n->type = type;
893: n->sec = man->lastsec;
894:
895: if (man->flags & MDOC_SYNOPSIS)
1.285 schwarze 896: n->flags |= NODE_SYNPRETTY;
1.266 schwarze 897: else
1.285 schwarze 898: n->flags &= ~NODE_SYNPRETTY;
1.360 schwarze 899: if ((man->flags & (ROFF_NOFILL | ROFF_NONOFILL)) == ROFF_NOFILL)
1.359 schwarze 900: n->flags |= NODE_NOFILL;
901: else
902: n->flags &= ~NODE_NOFILL;
1.266 schwarze 903: if (man->flags & MDOC_NEWLINE)
1.285 schwarze 904: n->flags |= NODE_LINE;
1.266 schwarze 905: man->flags &= ~MDOC_NEWLINE;
906:
1.277 schwarze 907: return n;
1.266 schwarze 908: }
909:
910: void
911: roff_node_append(struct roff_man *man, struct roff_node *n)
912: {
913:
914: switch (man->next) {
915: case ROFF_NEXT_SIBLING:
1.281 schwarze 916: if (man->last->next != NULL) {
917: n->next = man->last->next;
918: man->last->next->prev = n;
919: } else
920: man->last->parent->last = n;
1.266 schwarze 921: man->last->next = n;
922: n->prev = man->last;
923: n->parent = man->last->parent;
924: break;
925: case ROFF_NEXT_CHILD:
1.287 schwarze 926: if (man->last->child != NULL) {
927: n->next = man->last->child;
928: man->last->child->prev = n;
929: } else
930: man->last->last = n;
1.266 schwarze 931: man->last->child = n;
932: n->parent = man->last;
933: break;
934: default:
935: abort();
936: }
1.282 schwarze 937: man->last = n;
1.266 schwarze 938:
939: switch (n->type) {
940: case ROFFT_HEAD:
941: n->parent->head = n;
942: break;
943: case ROFFT_BODY:
1.282 schwarze 944: if (n->end != ENDBODY_NOT)
945: return;
1.266 schwarze 946: n->parent->body = n;
947: break;
948: case ROFFT_TAIL:
949: n->parent->tail = n;
950: break;
951: default:
1.282 schwarze 952: return;
1.266 schwarze 953: }
1.282 schwarze 954:
955: /*
956: * Copy over the normalised-data pointer of our parent. Not
957: * everybody has one, but copying a null pointer is fine.
958: */
959:
960: n->norm = n->parent->norm;
961: assert(n->parent->type == ROFFT_BLOCK);
1.266 schwarze 962: }
963:
1.267 schwarze 964: void
965: roff_word_alloc(struct roff_man *man, int line, int pos, const char *word)
966: {
967: struct roff_node *n;
968:
969: n = roff_node_alloc(man, line, pos, ROFFT_TEXT, TOKEN_NONE);
970: n->string = roff_strdup(man->roff, word);
971: roff_node_append(man, n);
1.286 schwarze 972: n->flags |= NODE_VALID | NODE_ENDED;
1.267 schwarze 973: man->next = ROFF_NEXT_SIBLING;
974: }
975:
976: void
977: roff_word_append(struct roff_man *man, const char *word)
978: {
979: struct roff_node *n;
980: char *addstr, *newstr;
981:
982: n = man->last;
983: addstr = roff_strdup(man->roff, word);
984: mandoc_asprintf(&newstr, "%s %s", n->string, addstr);
985: free(addstr);
986: free(n->string);
987: n->string = newstr;
988: man->next = ROFF_NEXT_SIBLING;
1.268 schwarze 989: }
990:
991: void
992: roff_elem_alloc(struct roff_man *man, int line, int pos, int tok)
993: {
994: struct roff_node *n;
995:
996: n = roff_node_alloc(man, line, pos, ROFFT_ELEM, tok);
997: roff_node_append(man, n);
998: man->next = ROFF_NEXT_CHILD;
999: }
1000:
1001: struct roff_node *
1002: roff_block_alloc(struct roff_man *man, int line, int pos, int tok)
1003: {
1004: struct roff_node *n;
1005:
1006: n = roff_node_alloc(man, line, pos, ROFFT_BLOCK, tok);
1007: roff_node_append(man, n);
1008: man->next = ROFF_NEXT_CHILD;
1.277 schwarze 1009: return n;
1.267 schwarze 1010: }
1011:
1.266 schwarze 1012: struct roff_node *
1013: roff_head_alloc(struct roff_man *man, int line, int pos, int tok)
1014: {
1015: struct roff_node *n;
1016:
1017: n = roff_node_alloc(man, line, pos, ROFFT_HEAD, tok);
1018: roff_node_append(man, n);
1019: man->next = ROFF_NEXT_CHILD;
1.277 schwarze 1020: return n;
1.266 schwarze 1021: }
1022:
1023: struct roff_node *
1024: roff_body_alloc(struct roff_man *man, int line, int pos, int tok)
1025: {
1026: struct roff_node *n;
1027:
1028: n = roff_node_alloc(man, line, pos, ROFFT_BODY, tok);
1029: roff_node_append(man, n);
1030: man->next = ROFF_NEXT_CHILD;
1.277 schwarze 1031: return n;
1.267 schwarze 1032: }
1033:
1.321 schwarze 1034: static void
1.346 schwarze 1035: roff_addtbl(struct roff_man *man, int line, struct tbl_node *tbl)
1.267 schwarze 1036: {
1037: struct roff_node *n;
1.346 schwarze 1038: struct tbl_span *span;
1.267 schwarze 1039:
1.356 schwarze 1040: if (man->meta.macroset == MACROSET_MAN)
1.302 schwarze 1041: man_breakscope(man, ROFF_TS);
1.321 schwarze 1042: while ((span = tbl_span(tbl)) != NULL) {
1.346 schwarze 1043: n = roff_node_alloc(man, line, 0, ROFFT_TBL, TOKEN_NONE);
1.321 schwarze 1044: n->span = span;
1045: roff_node_append(man, n);
1046: n->flags |= NODE_VALID | NODE_ENDED;
1047: man->next = ROFF_NEXT_SIBLING;
1048: }
1.266 schwarze 1049: }
1050:
1051: void
1052: roff_node_unlink(struct roff_man *man, struct roff_node *n)
1053: {
1054:
1055: /* Adjust siblings. */
1056:
1057: if (n->prev)
1058: n->prev->next = n->next;
1059: if (n->next)
1060: n->next->prev = n->prev;
1061:
1062: /* Adjust parent. */
1063:
1064: if (n->parent != NULL) {
1065: if (n->parent->child == n)
1066: n->parent->child = n->next;
1067: if (n->parent->last == n)
1068: n->parent->last = n->prev;
1069: }
1070:
1071: /* Adjust parse point. */
1072:
1073: if (man == NULL)
1074: return;
1075: if (man->last == n) {
1076: if (n->prev == NULL) {
1077: man->last = n->parent;
1078: man->next = ROFF_NEXT_CHILD;
1079: } else {
1080: man->last = n->prev;
1081: man->next = ROFF_NEXT_SIBLING;
1082: }
1083: }
1.356 schwarze 1084: if (man->meta.first == n)
1085: man->meta.first = NULL;
1.344 schwarze 1086: }
1087:
1088: void
1089: roff_node_relink(struct roff_man *man, struct roff_node *n)
1090: {
1091: roff_node_unlink(man, n);
1092: n->prev = n->next = NULL;
1093: roff_node_append(man, n);
1.266 schwarze 1094: }
1095:
1096: void
1097: roff_node_free(struct roff_node *n)
1098: {
1099:
1100: if (n->args != NULL)
1101: mdoc_argv_free(n->args);
1102: if (n->type == ROFFT_BLOCK || n->type == ROFFT_ELEM)
1103: free(n->norm);
1.347 schwarze 1104: eqn_box_free(n->eqn);
1.266 schwarze 1105: free(n->string);
1106: free(n);
1107: }
1108:
1109: void
1110: roff_node_delete(struct roff_man *man, struct roff_node *n)
1111: {
1112:
1113: while (n->child != NULL)
1114: roff_node_delete(man, n->child);
1115: roff_node_unlink(man, n);
1116: roff_node_free(n);
1.370 schwarze 1117: }
1118:
1119: int
1120: roff_node_transparent(struct roff_node *n)
1121: {
1122: if (n == NULL)
1123: return 0;
1124: if (n->type == ROFFT_COMMENT || n->flags & NODE_NOPRT)
1125: return 1;
1.373 ! schwarze 1126: return roff_tok_transparent(n->tok);
! 1127: }
! 1128:
! 1129: int
! 1130: roff_tok_transparent(enum roff_tok tok)
! 1131: {
! 1132: switch (tok) {
1.370 schwarze 1133: case ROFF_ft:
1134: case ROFF_ll:
1135: case ROFF_mc:
1136: case ROFF_po:
1137: case ROFF_ta:
1138: case MDOC_Db:
1139: case MDOC_Es:
1140: case MDOC_Sm:
1141: case MDOC_Tg:
1142: case MAN_DT:
1143: case MAN_UC:
1144: case MAN_PD:
1145: case MAN_AT:
1146: return 1;
1147: default:
1148: return 0;
1149: }
1150: }
1151:
1152: struct roff_node *
1153: roff_node_child(struct roff_node *n)
1154: {
1155: for (n = n->child; roff_node_transparent(n); n = n->next)
1156: continue;
1157: return n;
1158: }
1159:
1160: struct roff_node *
1161: roff_node_prev(struct roff_node *n)
1162: {
1163: do {
1164: n = n->prev;
1165: } while (roff_node_transparent(n));
1166: return n;
1167: }
1168:
1169: struct roff_node *
1170: roff_node_next(struct roff_node *n)
1171: {
1172: do {
1173: n = n->next;
1174: } while (roff_node_transparent(n));
1175: return n;
1.269 schwarze 1176: }
1177:
1178: void
1179: deroff(char **dest, const struct roff_node *n)
1180: {
1181: char *cp;
1182: size_t sz;
1183:
1.371 schwarze 1184: if (n->string == NULL) {
1.269 schwarze 1185: for (n = n->child; n != NULL; n = n->next)
1186: deroff(dest, n);
1187: return;
1188: }
1189:
1.288 schwarze 1190: /* Skip leading whitespace. */
1.269 schwarze 1191:
1.288 schwarze 1192: for (cp = n->string; *cp != '\0'; cp++) {
1.289 schwarze 1193: if (cp[0] == '\\' && cp[1] != '\0' &&
1194: strchr(" %&0^|~", cp[1]) != NULL)
1.269 schwarze 1195: cp++;
1.288 schwarze 1196: else if ( ! isspace((unsigned char)*cp))
1.269 schwarze 1197: break;
1198: }
1199:
1.289 schwarze 1200: /* Skip trailing backslash. */
1201:
1202: sz = strlen(cp);
1.290 schwarze 1203: if (sz > 0 && cp[sz - 1] == '\\')
1.289 schwarze 1204: sz--;
1205:
1.269 schwarze 1206: /* Skip trailing whitespace. */
1207:
1.289 schwarze 1208: for (; sz; sz--)
1.269 schwarze 1209: if ( ! isspace((unsigned char)cp[sz-1]))
1210: break;
1211:
1212: /* Skip empty strings. */
1213:
1214: if (sz == 0)
1215: return;
1216:
1217: if (*dest == NULL) {
1218: *dest = mandoc_strndup(cp, sz);
1219: return;
1220: }
1221:
1222: mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
1223: free(*dest);
1224: *dest = cp;
1.266 schwarze 1225: }
1226:
1227: /* --- main functions of the roff parser ---------------------------------- */
1228:
1.94 kristaps 1229: /*
1.355 schwarze 1230: * In the current line, expand escape sequences that produce parsable
1231: * input text. Also check the syntax of the remaining escape sequences,
1232: * which typically produce output glyphs or change formatter state.
1.154 kristaps 1233: */
1.340 schwarze 1234: static int
1.355 schwarze 1235: roff_expand(struct roff *r, struct buf *buf, int ln, int pos, char newesc)
1.94 kristaps 1236: {
1.339 schwarze 1237: struct mctx *ctx; /* current macro call context */
1.208 schwarze 1238: char ubuf[24]; /* buffer to print the number */
1.328 schwarze 1239: struct roff_node *n; /* used for header comments */
1.205 schwarze 1240: const char *start; /* start of the string to process */
1.209 schwarze 1241: char *stesc; /* start of an escape sequence ('\\') */
1.352 schwarze 1242: const char *esct; /* type of esccape sequence */
1.328 schwarze 1243: char *ep; /* end of comment string */
1.108 schwarze 1244: const char *stnam; /* start of the name, after "[(*" */
1245: const char *cp; /* end of the name, e.g. before ']' */
1246: const char *res; /* the string to be substituted */
1.238 schwarze 1247: char *nbuf; /* new buffer to copy buf->buf to */
1.181 schwarze 1248: size_t maxl; /* expected length of the escape name */
1249: size_t naml; /* actual length of the escape name */
1.339 schwarze 1250: size_t asz; /* length of the replacement */
1251: size_t rsz; /* length of the rest of the string */
1.237 schwarze 1252: int inaml; /* length returned from mandoc_escape() */
1.181 schwarze 1253: int expand_count; /* to avoid infinite loops */
1.206 schwarze 1254: int npos; /* position in numeric expression */
1.218 schwarze 1255: int arg_complete; /* argument not interrupted by eol */
1.339 schwarze 1256: int quote_args; /* true for \\$@, false for \\$* */
1.303 schwarze 1257: int done; /* no more input available */
1.315 schwarze 1258: int deftype; /* type of definition to paste */
1.323 schwarze 1259: int rcsid; /* kind of RCS id seen */
1.352 schwarze 1260: enum mandocerr err; /* for escape sequence problems */
1.327 schwarze 1261: char sign; /* increment number register */
1.206 schwarze 1262: char term; /* character terminating the escape */
1.94 kristaps 1263:
1.303 schwarze 1264: /* Search forward for comments. */
1265:
1266: done = 0;
1267: start = buf->buf + pos;
1268: for (stesc = buf->buf + pos; *stesc != '\0'; stesc++) {
1.355 schwarze 1269: if (stesc[0] != newesc || stesc[1] == '\0')
1.303 schwarze 1270: continue;
1271: stesc++;
1272: if (*stesc != '"' && *stesc != '#')
1273: continue;
1.313 schwarze 1274:
1275: /* Comment found, look for RCS id. */
1276:
1.323 schwarze 1277: rcsid = 0;
1.313 schwarze 1278: if ((cp = strstr(stesc, "$" "OpenBSD")) != NULL) {
1.323 schwarze 1279: rcsid = 1 << MANDOC_OS_OPENBSD;
1.313 schwarze 1280: cp += 8;
1281: } else if ((cp = strstr(stesc, "$" "NetBSD")) != NULL) {
1.323 schwarze 1282: rcsid = 1 << MANDOC_OS_NETBSD;
1.313 schwarze 1283: cp += 7;
1284: }
1285: if (cp != NULL &&
1286: isalnum((unsigned char)*cp) == 0 &&
1.314 schwarze 1287: strchr(cp, '$') != NULL) {
1.323 schwarze 1288: if (r->man->meta.rcsids & rcsid)
1.350 schwarze 1289: mandoc_msg(MANDOCERR_RCS_REP, ln,
1290: (int)(stesc - buf->buf) + 1,
1291: "%s", stesc + 1);
1.323 schwarze 1292: r->man->meta.rcsids |= rcsid;
1.314 schwarze 1293: }
1.313 schwarze 1294:
1295: /* Handle trailing whitespace. */
1296:
1.328 schwarze 1297: ep = strchr(stesc--, '\0') - 1;
1298: if (*ep == '\n') {
1.303 schwarze 1299: done = 1;
1.328 schwarze 1300: ep--;
1.303 schwarze 1301: }
1.328 schwarze 1302: if (*ep == ' ' || *ep == '\t')
1.350 schwarze 1303: mandoc_msg(MANDOCERR_SPACE_EOL,
1304: ln, (int)(ep - buf->buf), NULL);
1.328 schwarze 1305:
1306: /*
1307: * Save comments preceding the title macro
1308: * in the syntax tree.
1309: */
1310:
1.367 schwarze 1311: if (newesc != ASCII_ESC && r->options & MPARSE_COMMENT) {
1.328 schwarze 1312: while (*ep == ' ' || *ep == '\t')
1313: ep--;
1314: ep[1] = '\0';
1315: n = roff_node_alloc(r->man,
1316: ln, stesc + 1 - buf->buf,
1317: ROFFT_COMMENT, TOKEN_NONE);
1318: n->string = mandoc_strdup(stesc + 2);
1319: roff_node_append(r->man, n);
1320: n->flags |= NODE_VALID | NODE_ENDED;
1321: r->man->next = ROFF_NEXT_SIBLING;
1322: }
1323:
1.334 schwarze 1324: /* Line continuation with comment. */
1325:
1326: if (stesc[1] == '#') {
1327: *stesc = '\0';
1.340 schwarze 1328: return ROFF_IGN | ROFF_APPEND;
1.334 schwarze 1329: }
1330:
1331: /* Discard normal comments. */
1.328 schwarze 1332:
1.335 schwarze 1333: while (stesc > start && stesc[-1] == ' ' &&
1334: (stesc == start + 1 || stesc[-2] != '\\'))
1.303 schwarze 1335: stesc--;
1336: *stesc = '\0';
1337: break;
1338: }
1339: if (stesc == start)
1340: return ROFF_CONT;
1341: stesc--;
1342:
1343: /* Notice the end of the input. */
1344:
1345: if (*stesc == '\n') {
1346: *stesc-- = '\0';
1347: done = 1;
1348: }
1349:
1.170 schwarze 1350: expand_count = 0;
1.303 schwarze 1351: while (stesc >= start) {
1.355 schwarze 1352: if (*stesc != newesc) {
1.170 schwarze 1353:
1.355 schwarze 1354: /*
1355: * If we have a non-standard escape character,
1356: * escape literal backslashes because all
1357: * processing in subsequent functions uses
1358: * the standard escaping rules.
1359: */
1.205 schwarze 1360:
1.355 schwarze 1361: if (newesc != ASCII_ESC && *stesc == '\\') {
1.303 schwarze 1362: *stesc = '\0';
1363: buf->sz = mandoc_asprintf(&nbuf, "%s\\e%s",
1364: buf->buf, stesc + 1) + 1;
1365: start = nbuf + pos;
1366: stesc = nbuf + (stesc - buf->buf);
1367: free(buf->buf);
1368: buf->buf = nbuf;
1369: }
1.355 schwarze 1370:
1371: /* Search backwards for the next escape. */
1372:
1.303 schwarze 1373: stesc--;
1.205 schwarze 1374: continue;
1.303 schwarze 1375: }
1.205 schwarze 1376:
1377: /* If it is escaped, skip it. */
1378:
1379: for (cp = stesc - 1; cp >= start; cp--)
1.303 schwarze 1380: if (*cp != r->escape)
1.205 schwarze 1381: break;
1382:
1.238 schwarze 1383: if ((stesc - cp) % 2 == 0) {
1.303 schwarze 1384: while (stesc > cp)
1385: *stesc-- = '\\';
1.205 schwarze 1386: continue;
1.303 schwarze 1387: } else if (stesc[1] != '\0') {
1388: *stesc = '\\';
1389: } else {
1390: *stesc-- = '\0';
1391: if (done)
1392: continue;
1393: else
1.340 schwarze 1394: return ROFF_IGN | ROFF_APPEND;
1.205 schwarze 1395: }
1.108 schwarze 1396:
1.206 schwarze 1397: /* Decide whether to expand or to check only. */
1.108 schwarze 1398:
1.206 schwarze 1399: term = '\0';
1.205 schwarze 1400: cp = stesc + 1;
1.352 schwarze 1401: if (*cp == 'E')
1402: cp++;
1403: esct = cp;
1404: switch (*esct) {
1.207 schwarze 1405: case '*':
1.339 schwarze 1406: case '$':
1.181 schwarze 1407: res = NULL;
1408: break;
1.207 schwarze 1409: case 'B':
1410: case 'w':
1.206 schwarze 1411: term = cp[1];
1412: /* FALLTHROUGH */
1.207 schwarze 1413: case 'n':
1.327 schwarze 1414: sign = cp[1];
1415: if (sign == '+' || sign == '-')
1416: cp++;
1.181 schwarze 1417: res = ubuf;
1418: break;
1419: default:
1.352 schwarze 1420: err = MANDOCERR_OK;
1421: switch(mandoc_escape(&cp, &stnam, &inaml)) {
1422: case ESCAPE_SPECIAL:
1423: if (mchars_spec2cp(stnam, inaml) >= 0)
1424: break;
1425: /* FALLTHROUGH */
1426: case ESCAPE_ERROR:
1427: err = MANDOCERR_ESC_BAD;
1428: break;
1429: case ESCAPE_UNDEF:
1430: err = MANDOCERR_ESC_UNDEF;
1431: break;
1432: case ESCAPE_UNSUPP:
1433: err = MANDOCERR_ESC_UNSUPP;
1434: break;
1435: default:
1436: break;
1437: }
1438: if (err != MANDOCERR_OK)
1439: mandoc_msg(err, ln, (int)(stesc - buf->buf),
1.219 schwarze 1440: "%.*s", (int)(cp - stesc), stesc);
1.303 schwarze 1441: stesc--;
1.205 schwarze 1442: continue;
1.152 kristaps 1443: }
1444:
1.205 schwarze 1445: if (EXPAND_LIMIT < ++expand_count) {
1.350 schwarze 1446: mandoc_msg(MANDOCERR_ROFFLOOP,
1.238 schwarze 1447: ln, (int)(stesc - buf->buf), NULL);
1.277 schwarze 1448: return ROFF_IGN;
1.205 schwarze 1449: }
1.108 schwarze 1450:
1451: /*
1452: * The third character decides the length
1.181 schwarze 1453: * of the name of the string or register.
1.108 schwarze 1454: * Save a pointer to the name.
1455: */
1456:
1.238 schwarze 1457: if (term == '\0') {
1.206 schwarze 1458: switch (*++cp) {
1.207 schwarze 1459: case '\0':
1.206 schwarze 1460: maxl = 0;
1461: break;
1.207 schwarze 1462: case '(':
1.206 schwarze 1463: cp++;
1464: maxl = 2;
1465: break;
1.207 schwarze 1466: case '[':
1.206 schwarze 1467: cp++;
1468: term = ']';
1469: maxl = 0;
1470: break;
1471: default:
1472: maxl = 1;
1473: break;
1474: }
1475: } else {
1476: cp += 2;
1.94 kristaps 1477: maxl = 0;
1478: }
1.108 schwarze 1479: stnam = cp;
1.94 kristaps 1480:
1.108 schwarze 1481: /* Advance to the end of the name. */
1.94 kristaps 1482:
1.253 schwarze 1483: naml = 0;
1.218 schwarze 1484: arg_complete = 1;
1.253 schwarze 1485: while (maxl == 0 || naml < maxl) {
1.238 schwarze 1486: if (*cp == '\0') {
1.350 schwarze 1487: mandoc_msg(MANDOCERR_ESC_BAD, ln,
1488: (int)(stesc - buf->buf), "%s", stesc);
1.218 schwarze 1489: arg_complete = 0;
1.206 schwarze 1490: break;
1.153 kristaps 1491: }
1.238 schwarze 1492: if (maxl == 0 && *cp == term) {
1.206 schwarze 1493: cp++;
1.253 schwarze 1494: break;
1495: }
1.352 schwarze 1496: if (*cp++ != '\\' || *esct != 'w') {
1.253 schwarze 1497: naml++;
1498: continue;
1499: }
1500: switch (mandoc_escape(&cp, NULL, NULL)) {
1501: case ESCAPE_SPECIAL:
1502: case ESCAPE_UNICODE:
1503: case ESCAPE_NUMBERED:
1.352 schwarze 1504: case ESCAPE_UNDEF:
1.253 schwarze 1505: case ESCAPE_OVERSTRIKE:
1506: naml++;
1507: break;
1508: default:
1.94 kristaps 1509: break;
1.206 schwarze 1510: }
1.94 kristaps 1511: }
1512:
1.108 schwarze 1513: /*
1514: * Retrieve the replacement string; if it is
1515: * undefined, resume searching for escapes.
1516: */
1517:
1.352 schwarze 1518: switch (*esct) {
1.207 schwarze 1519: case '*':
1.315 schwarze 1520: if (arg_complete) {
1521: deftype = ROFFDEF_USER | ROFFDEF_PRE;
1522: res = roff_getstrn(r, stnam, naml, &deftype);
1.331 schwarze 1523:
1524: /*
1525: * If not overriden, let \*(.T
1526: * through to the formatters.
1527: */
1528:
1529: if (res == NULL && naml == 2 &&
1530: stnam[0] == '.' && stnam[1] == 'T') {
1531: roff_setstrn(&r->strtab,
1532: ".T", 2, NULL, 0, 0);
1533: stesc--;
1534: continue;
1535: }
1.315 schwarze 1536: }
1.206 schwarze 1537: break;
1.339 schwarze 1538: case '$':
1539: if (r->mstackpos < 0) {
1.350 schwarze 1540: mandoc_msg(MANDOCERR_ARG_UNDEF, ln,
1541: (int)(stesc - buf->buf), "%.3s", stesc);
1.339 schwarze 1542: break;
1543: }
1544: ctx = r->mstack + r->mstackpos;
1.352 schwarze 1545: npos = esct[1] - '1';
1.339 schwarze 1546: if (npos >= 0 && npos <= 8) {
1547: res = npos < ctx->argc ?
1548: ctx->argv[npos] : "";
1549: break;
1550: }
1.352 schwarze 1551: if (esct[1] == '*')
1.339 schwarze 1552: quote_args = 0;
1.352 schwarze 1553: else if (esct[1] == '@')
1.339 schwarze 1554: quote_args = 1;
1555: else {
1.350 schwarze 1556: mandoc_msg(MANDOCERR_ARG_NONUM, ln,
1557: (int)(stesc - buf->buf), "%.3s", stesc);
1.339 schwarze 1558: break;
1559: }
1560: asz = 0;
1561: for (npos = 0; npos < ctx->argc; npos++) {
1562: if (npos)
1563: asz++; /* blank */
1564: if (quote_args)
1565: asz += 2; /* quotes */
1566: asz += strlen(ctx->argv[npos]);
1567: }
1568: if (asz != 3) {
1569: rsz = buf->sz - (stesc - buf->buf) - 3;
1570: if (asz < 3)
1571: memmove(stesc + asz, stesc + 3, rsz);
1572: buf->sz += asz - 3;
1573: nbuf = mandoc_realloc(buf->buf, buf->sz);
1574: start = nbuf + pos;
1575: stesc = nbuf + (stesc - buf->buf);
1576: buf->buf = nbuf;
1577: if (asz > 3)
1578: memmove(stesc + asz, stesc + 3, rsz);
1579: }
1580: for (npos = 0; npos < ctx->argc; npos++) {
1581: if (npos)
1582: *stesc++ = ' ';
1583: if (quote_args)
1584: *stesc++ = '"';
1585: cp = ctx->argv[npos];
1586: while (*cp != '\0')
1587: *stesc++ = *cp++;
1588: if (quote_args)
1589: *stesc++ = '"';
1590: }
1591: continue;
1.207 schwarze 1592: case 'B':
1.206 schwarze 1593: npos = 0;
1.218 schwarze 1594: ubuf[0] = arg_complete &&
1.261 schwarze 1595: roff_evalnum(r, ln, stnam, &npos,
1596: NULL, ROFFNUM_SCALE) &&
1.218 schwarze 1597: stnam + npos + 1 == cp ? '1' : '0';
1.206 schwarze 1598: ubuf[1] = '\0';
1599: break;
1.207 schwarze 1600: case 'n':
1.218 schwarze 1601: if (arg_complete)
1602: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.327 schwarze 1603: roff_getregn(r, stnam, naml, sign));
1.218 schwarze 1604: else
1605: ubuf[0] = '\0';
1.206 schwarze 1606: break;
1.207 schwarze 1607: case 'w':
1.218 schwarze 1608: /* use even incomplete args */
1.208 schwarze 1609: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.206 schwarze 1610: 24 * (int)naml);
1611: break;
1612: }
1.94 kristaps 1613:
1.238 schwarze 1614: if (res == NULL) {
1.352 schwarze 1615: if (*esct == '*')
1.350 schwarze 1616: mandoc_msg(MANDOCERR_STR_UNDEF,
1617: ln, (int)(stesc - buf->buf),
1.339 schwarze 1618: "%.*s", (int)naml, stnam);
1.142 kristaps 1619: res = "";
1.246 schwarze 1620: } else if (buf->sz + strlen(res) > SHRT_MAX) {
1.350 schwarze 1621: mandoc_msg(MANDOCERR_ROFFLOOP,
1.246 schwarze 1622: ln, (int)(stesc - buf->buf), NULL);
1.277 schwarze 1623: return ROFF_IGN;
1.94 kristaps 1624: }
1625:
1.108 schwarze 1626: /* Replace the escape sequence by the string. */
1627:
1.209 schwarze 1628: *stesc = '\0';
1.238 schwarze 1629: buf->sz = mandoc_asprintf(&nbuf, "%s%s%s",
1630: buf->buf, res, cp) + 1;
1.94 kristaps 1631:
1.205 schwarze 1632: /* Prepare for the next replacement. */
1.94 kristaps 1633:
1.205 schwarze 1634: start = nbuf + pos;
1.238 schwarze 1635: stesc = nbuf + (stesc - buf->buf) + strlen(res);
1636: free(buf->buf);
1637: buf->buf = nbuf;
1.154 kristaps 1638: }
1.277 schwarze 1639: return ROFF_CONT;
1.154 kristaps 1640: }
1.353 schwarze 1641:
1642: /*
1643: * Parse a quoted or unquoted roff-style request or macro argument.
1644: * Return a pointer to the parsed argument, which is either the original
1645: * pointer or advanced by one byte in case the argument is quoted.
1646: * NUL-terminate the argument in place.
1647: * Collapse pairs of quotes inside quoted arguments.
1648: * Advance the argument pointer to the next argument,
1649: * or to the NUL byte terminating the argument line.
1650: */
1651: char *
1.355 schwarze 1652: roff_getarg(struct roff *r, char **cpp, int ln, int *pos)
1.353 schwarze 1653: {
1.355 schwarze 1654: struct buf buf;
1.366 schwarze 1655: char *cp, *start;
1.355 schwarze 1656: int newesc, pairs, quoted, white;
1.353 schwarze 1657:
1658: /* Quoting can only start with a new word. */
1659: start = *cpp;
1660: quoted = 0;
1661: if ('"' == *start) {
1662: quoted = 1;
1663: start++;
1664: }
1665:
1.355 schwarze 1666: newesc = pairs = white = 0;
1.353 schwarze 1667: for (cp = start; '\0' != *cp; cp++) {
1668:
1669: /*
1670: * Move the following text left
1671: * after quoted quotes and after "\\" and "\t".
1672: */
1673: if (pairs)
1674: cp[-pairs] = cp[0];
1675:
1676: if ('\\' == cp[0]) {
1677: /*
1678: * In copy mode, translate double to single
1679: * backslashes and backslash-t to literal tabs.
1680: */
1681: switch (cp[1]) {
1682: case 'a':
1683: case 't':
1.354 schwarze 1684: cp[-pairs] = '\t';
1.355 schwarze 1685: pairs++;
1686: cp++;
1687: break;
1.353 schwarze 1688: case '\\':
1.355 schwarze 1689: newesc = 1;
1690: cp[-pairs] = ASCII_ESC;
1.353 schwarze 1691: pairs++;
1692: cp++;
1693: break;
1694: case ' ':
1695: /* Skip escaped blanks. */
1696: if (0 == quoted)
1697: cp++;
1698: break;
1699: default:
1700: break;
1701: }
1702: } else if (0 == quoted) {
1703: if (' ' == cp[0]) {
1704: /* Unescaped blanks end unquoted args. */
1705: white = 1;
1706: break;
1707: }
1708: } else if ('"' == cp[0]) {
1709: if ('"' == cp[1]) {
1710: /* Quoted quotes collapse. */
1711: pairs++;
1712: cp++;
1713: } else {
1714: /* Unquoted quotes end quoted args. */
1715: quoted = 2;
1716: break;
1717: }
1718: }
1719: }
1720:
1721: /* Quoted argument without a closing quote. */
1722: if (1 == quoted)
1723: mandoc_msg(MANDOCERR_ARG_QUOTE, ln, *pos, NULL);
1724:
1725: /* NUL-terminate this argument and move to the next one. */
1726: if (pairs)
1727: cp[-pairs] = '\0';
1728: if ('\0' != *cp) {
1729: *cp++ = '\0';
1730: while (' ' == *cp)
1731: cp++;
1732: }
1733: *pos += (int)(cp - start) + (quoted ? 1 : 0);
1734: *cpp = cp;
1735:
1736: if ('\0' == *cp && (white || ' ' == cp[-1]))
1737: mandoc_msg(MANDOCERR_SPACE_EOL, ln, *pos, NULL);
1738:
1.355 schwarze 1739: start = mandoc_strdup(start);
1740: if (newesc == 0)
1741: return start;
1742:
1743: buf.buf = start;
1744: buf.sz = strlen(start) + 1;
1745: buf.next = NULL;
1746: if (roff_expand(r, &buf, ln, 0, ASCII_ESC) & ROFF_IGN) {
1747: free(buf.buf);
1748: buf.buf = mandoc_strdup("");
1749: }
1750: return buf.buf;
1.353 schwarze 1751: }
1752:
1.154 kristaps 1753:
1754: /*
1.275 schwarze 1755: * Process text streams.
1.154 kristaps 1756: */
1.340 schwarze 1757: static int
1.305 schwarze 1758: roff_parsetext(struct roff *r, struct buf *buf, int pos, int *offs)
1.154 kristaps 1759: {
1760: size_t sz;
1761: const char *start;
1.178 schwarze 1762: char *p;
1763: int isz;
1.154 kristaps 1764: enum mandoc_esc esc;
1765:
1.275 schwarze 1766: /* Spring the input line trap. */
1767:
1768: if (roffit_lines == 1) {
1769: isz = mandoc_asprintf(&p, "%s\n.%s", buf->buf, roffit_macro);
1770: free(buf->buf);
1771: buf->buf = p;
1772: buf->sz = isz + 1;
1773: *offs = 0;
1774: free(roffit_macro);
1775: roffit_lines = 0;
1.277 schwarze 1776: return ROFF_REPARSE;
1.275 schwarze 1777: } else if (roffit_lines > 1)
1778: --roffit_lines;
1779:
1.305 schwarze 1780: if (roffce_node != NULL && buf->buf[pos] != '\0') {
1781: if (roffce_lines < 1) {
1782: r->man->last = roffce_node;
1783: r->man->next = ROFF_NEXT_SIBLING;
1784: roffce_lines = 0;
1785: roffce_node = NULL;
1786: } else
1787: roffce_lines--;
1788: }
1789:
1.275 schwarze 1790: /* Convert all breakable hyphens into ASCII_HYPH. */
1791:
1.238 schwarze 1792: start = p = buf->buf + pos;
1.154 kristaps 1793:
1.238 schwarze 1794: while (*p != '\0') {
1.154 kristaps 1795: sz = strcspn(p, "-\\");
1796: p += sz;
1797:
1.238 schwarze 1798: if (*p == '\0')
1.159 kristaps 1799: break;
1800:
1.238 schwarze 1801: if (*p == '\\') {
1.154 kristaps 1802: /* Skip over escapes. */
1803: p++;
1.189 schwarze 1804: esc = mandoc_escape((const char **)&p, NULL, NULL);
1.238 schwarze 1805: if (esc == ESCAPE_ERROR)
1.154 kristaps 1806: break;
1.264 schwarze 1807: while (*p == '-')
1808: p++;
1.155 kristaps 1809: continue;
1.159 kristaps 1810: } else if (p == start) {
1.158 kristaps 1811: p++;
1.155 kristaps 1812: continue;
1.158 kristaps 1813: }
1.155 kristaps 1814:
1.171 schwarze 1815: if (isalpha((unsigned char)p[-1]) &&
1816: isalpha((unsigned char)p[1]))
1.155 kristaps 1817: *p = ASCII_HYPH;
1818: p++;
1.94 kristaps 1819: }
1.277 schwarze 1820: return ROFF_CONT;
1.94 kristaps 1821: }
1822:
1.340 schwarze 1823: int
1.238 schwarze 1824: roff_parseln(struct roff *r, int ln, struct buf *buf, int *offs)
1.67 kristaps 1825: {
1.294 schwarze 1826: enum roff_tok t;
1.340 schwarze 1827: int e;
1.238 schwarze 1828: int pos; /* parse point */
1.243 schwarze 1829: int spos; /* saved parse point for messages */
1.238 schwarze 1830: int ppos; /* original offset in buf->buf */
1831: int ctl; /* macro line (boolean) */
1832:
1833: ppos = pos = *offs;
1.79 kristaps 1834:
1.230 schwarze 1835: /* Handle in-line equation delimiters. */
1836:
1.236 schwarze 1837: if (r->tbl == NULL &&
1838: r->last_eqn != NULL && r->last_eqn->delim &&
1.230 schwarze 1839: (r->eqn == NULL || r->eqn_inline)) {
1.238 schwarze 1840: e = roff_eqndelim(r, buf, pos);
1.230 schwarze 1841: if (e == ROFF_REPARSE)
1.277 schwarze 1842: return e;
1.230 schwarze 1843: assert(e == ROFF_CONT);
1844: }
1845:
1846: /* Expand some escape sequences. */
1.94 kristaps 1847:
1.355 schwarze 1848: e = roff_expand(r, buf, ln, pos, r->escape);
1.340 schwarze 1849: if ((e & ROFF_MASK) == ROFF_IGN)
1.277 schwarze 1850: return e;
1.238 schwarze 1851: assert(e == ROFF_CONT);
1.94 kristaps 1852:
1.238 schwarze 1853: ctl = roff_getcontrol(r, buf->buf, &pos);
1.130 kristaps 1854:
1.94 kristaps 1855: /*
1.79 kristaps 1856: * First, if a scope is open and we're not a macro, pass the
1.252 schwarze 1857: * text through the macro's filter.
1858: * Equations process all content themselves.
1859: * Tables process almost all content themselves, but we want
1860: * to warn about macros before passing it there.
1.79 kristaps 1861: */
1.74 kristaps 1862:
1.252 schwarze 1863: if (r->last != NULL && ! ctl) {
1.78 kristaps 1864: t = r->last->tok;
1.238 schwarze 1865: e = (*roffs[t].text)(r, t, buf, ln, pos, pos, offs);
1.340 schwarze 1866: if ((e & ROFF_MASK) == ROFF_IGN)
1.277 schwarze 1867: return e;
1.340 schwarze 1868: e &= ~ROFF_MASK;
1869: } else
1870: e = ROFF_IGN;
1.319 schwarze 1871: if (r->eqn != NULL && strncmp(buf->buf + ppos, ".EN", 3)) {
1872: eqn_read(r->eqn, buf->buf + ppos);
1.340 schwarze 1873: return e;
1.319 schwarze 1874: }
1.321 schwarze 1875: if (r->tbl != NULL && (ctl == 0 || buf->buf[pos] == '\0')) {
1876: tbl_read(r->tbl, ln, buf->buf, ppos);
1.346 schwarze 1877: roff_addtbl(r->man, ln, r->tbl);
1.340 schwarze 1878: return e;
1.321 schwarze 1879: }
1.367 schwarze 1880: if ( ! ctl) {
1881: r->options &= ~MPARSE_COMMENT;
1.340 schwarze 1882: return roff_parsetext(r, buf, pos, offs) | e;
1.367 schwarze 1883: }
1.228 schwarze 1884:
1885: /* Skip empty request lines. */
1886:
1.238 schwarze 1887: if (buf->buf[pos] == '"') {
1.350 schwarze 1888: mandoc_msg(MANDOCERR_COMMENT_BAD, ln, pos, NULL);
1.277 schwarze 1889: return ROFF_IGN;
1.238 schwarze 1890: } else if (buf->buf[pos] == '\0')
1.277 schwarze 1891: return ROFF_IGN;
1.67 kristaps 1892:
1.79 kristaps 1893: /*
1894: * If a scope is open, go to the child handler for that macro,
1895: * as it may want to preprocess before doing anything with it.
1.125 kristaps 1896: * Don't do so if an equation is open.
1.79 kristaps 1897: */
1.78 kristaps 1898:
1.79 kristaps 1899: if (r->last) {
1900: t = r->last->tok;
1.277 schwarze 1901: return (*roffs[t].sub)(r, t, buf, ln, ppos, pos, offs);
1.79 kristaps 1902: }
1.78 kristaps 1903:
1.243 schwarze 1904: /* No scope is open. This is a new request or macro. */
1905:
1.367 schwarze 1906: r->options &= ~MPARSE_COMMENT;
1.243 schwarze 1907: spos = pos;
1908: t = roff_parse(r, buf->buf, &pos, ln, ppos);
1909:
1910: /* Tables ignore most macros. */
1911:
1.308 schwarze 1912: if (r->tbl != NULL && (t == TOKEN_NONE || t == ROFF_TS ||
1.309 schwarze 1913: t == ROFF_br || t == ROFF_ce || t == ROFF_rj || t == ROFF_sp)) {
1.350 schwarze 1914: mandoc_msg(MANDOCERR_TBLMACRO,
1915: ln, pos, "%s", buf->buf + spos);
1.308 schwarze 1916: if (t != TOKEN_NONE)
1.277 schwarze 1917: return ROFF_IGN;
1.257 schwarze 1918: while (buf->buf[pos] != '\0' && buf->buf[pos] != ' ')
1919: pos++;
1.291 schwarze 1920: while (buf->buf[pos] == ' ')
1.257 schwarze 1921: pos++;
1.321 schwarze 1922: tbl_read(r->tbl, ln, buf->buf, pos);
1.346 schwarze 1923: roff_addtbl(r->man, ln, r->tbl);
1.321 schwarze 1924: return ROFF_IGN;
1.243 schwarze 1925: }
1926:
1.305 schwarze 1927: /* For now, let high level macros abort .ce mode. */
1928:
1929: if (ctl && roffce_node != NULL &&
1.324 schwarze 1930: (t == TOKEN_NONE || t == ROFF_Dd || t == ROFF_EQ ||
1931: t == ROFF_TH || t == ROFF_TS)) {
1.305 schwarze 1932: r->man->last = roffce_node;
1933: r->man->next = ROFF_NEXT_SIBLING;
1934: roffce_lines = 0;
1935: roffce_node = NULL;
1936: }
1937:
1.79 kristaps 1938: /*
1.243 schwarze 1939: * This is neither a roff request nor a user-defined macro.
1940: * Let the standard macro set parsers handle it.
1.79 kristaps 1941: */
1.67 kristaps 1942:
1.294 schwarze 1943: if (t == TOKEN_NONE)
1.277 schwarze 1944: return ROFF_CONT;
1.243 schwarze 1945:
1946: /* Execute a roff request or a user defined macro. */
1.67 kristaps 1947:
1.296 schwarze 1948: return (*roffs[t].proc)(r, t, buf, ln, spos, pos, offs);
1.74 kristaps 1949: }
1950:
1.339 schwarze 1951: /*
1952: * Internal interface function to tell the roff parser that execution
1953: * of the current macro ended. This is required because macro
1954: * definitions usually do not end with a .return request.
1955: */
1956: void
1957: roff_userret(struct roff *r)
1958: {
1959: struct mctx *ctx;
1960: int i;
1961:
1962: assert(r->mstackpos >= 0);
1963: ctx = r->mstack + r->mstackpos;
1964: for (i = 0; i < ctx->argc; i++)
1965: free(ctx->argv[i]);
1966: ctx->argc = 0;
1967: r->mstackpos--;
1968: }
1969:
1.117 kristaps 1970: void
1.74 kristaps 1971: roff_endparse(struct roff *r)
1972: {
1.321 schwarze 1973: if (r->last != NULL)
1.350 schwarze 1974: mandoc_msg(MANDOCERR_BLK_NOEND, r->last->line,
1975: r->last->col, "%s", roff_name[r->last->tok]);
1.117 kristaps 1976:
1.321 schwarze 1977: if (r->eqn != NULL) {
1.350 schwarze 1978: mandoc_msg(MANDOCERR_BLK_NOEND,
1.319 schwarze 1979: r->eqn->node->line, r->eqn->node->pos, "EQ");
1980: eqn_parse(r->eqn);
1981: r->eqn = NULL;
1.125 kristaps 1982: }
1983:
1.321 schwarze 1984: if (r->tbl != NULL) {
1.346 schwarze 1985: tbl_end(r->tbl, 1);
1.321 schwarze 1986: r->tbl = NULL;
1.117 kristaps 1987: }
1.67 kristaps 1988: }
1989:
1990: /*
1991: * Parse a roff node's type from the input buffer. This must be in the
1992: * form of ".foo xxx" in the usual way.
1993: */
1.294 schwarze 1994: static enum roff_tok
1.214 schwarze 1995: roff_parse(struct roff *r, char *buf, int *pos, int ln, int ppos)
1.67 kristaps 1996: {
1.214 schwarze 1997: char *cp;
1.106 kristaps 1998: const char *mac;
1999: size_t maclen;
1.315 schwarze 2000: int deftype;
1.294 schwarze 2001: enum roff_tok t;
1.67 kristaps 2002:
1.214 schwarze 2003: cp = buf + *pos;
2004:
2005: if ('\0' == *cp || '"' == *cp || '\t' == *cp || ' ' == *cp)
1.294 schwarze 2006: return TOKEN_NONE;
1.67 kristaps 2007:
1.214 schwarze 2008: mac = cp;
2009: maclen = roff_getname(r, &cp, ln, ppos);
1.67 kristaps 2010:
1.315 schwarze 2011: deftype = ROFFDEF_USER | ROFFDEF_REN;
2012: r->current_string = roff_getstrn(r, mac, maclen, &deftype);
2013: switch (deftype) {
2014: case ROFFDEF_USER:
2015: t = ROFF_USERDEF;
2016: break;
2017: case ROFFDEF_REN:
2018: t = ROFF_RENAMED;
2019: break;
2020: default:
2021: t = roffhash_find(r->reqtab, mac, maclen);
2022: break;
2023: }
1.294 schwarze 2024: if (t != TOKEN_NONE)
1.214 schwarze 2025: *pos = cp - buf;
1.325 schwarze 2026: else if (deftype == ROFFDEF_UNDEF) {
2027: /* Using an undefined macro defines it to be empty. */
2028: roff_setstrn(&r->strtab, mac, maclen, "", 0, 0);
2029: roff_setstrn(&r->rentab, mac, maclen, NULL, 0, 0);
2030: }
1.277 schwarze 2031: return t;
1.67 kristaps 2032: }
2033:
1.266 schwarze 2034: /* --- handling of request blocks ----------------------------------------- */
2035:
1.340 schwarze 2036: static int
1.76 kristaps 2037: roff_cblock(ROFF_ARGS)
1.67 kristaps 2038: {
2039:
1.79 kristaps 2040: /*
2041: * A block-close `..' should only be invoked as a child of an
2042: * ignore macro, otherwise raise a warning and just ignore it.
2043: */
2044:
1.238 schwarze 2045: if (r->last == NULL) {
1.350 schwarze 2046: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "..");
1.277 schwarze 2047: return ROFF_IGN;
1.76 kristaps 2048: }
1.67 kristaps 2049:
1.81 kristaps 2050: switch (r->last->tok) {
1.207 schwarze 2051: case ROFF_am:
1.220 schwarze 2052: /* ROFF_am1 is remapped to ROFF_am in roff_block(). */
1.207 schwarze 2053: case ROFF_ami:
2054: case ROFF_de:
1.108 schwarze 2055: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.207 schwarze 2056: case ROFF_dei:
2057: case ROFF_ig:
1.81 kristaps 2058: break;
2059: default:
1.350 schwarze 2060: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "..");
1.277 schwarze 2061: return ROFF_IGN;
1.76 kristaps 2062: }
1.67 kristaps 2063:
1.238 schwarze 2064: if (buf->buf[pos] != '\0')
1.350 schwarze 2065: mandoc_msg(MANDOCERR_ARG_SKIP, ln, pos,
1.238 schwarze 2066: ".. %s", buf->buf + pos);
1.71 kristaps 2067:
2068: roffnode_pop(r);
1.76 kristaps 2069: roffnode_cleanscope(r);
1.277 schwarze 2070: return ROFF_IGN;
1.71 kristaps 2071:
1.67 kristaps 2072: }
2073:
1.364 schwarze 2074: /*
2075: * Pop all nodes ending at the end of the current input line.
2076: * Return the number of loops ended.
2077: */
1.340 schwarze 2078: static int
1.76 kristaps 2079: roffnode_cleanscope(struct roff *r)
1.67 kristaps 2080: {
1.340 schwarze 2081: int inloop;
1.67 kristaps 2082:
1.340 schwarze 2083: inloop = 0;
2084: while (r->last != NULL) {
1.173 schwarze 2085: if (--r->last->endspan != 0)
1.76 kristaps 2086: break;
1.340 schwarze 2087: inloop += roffnode_pop(r);
1.76 kristaps 2088: }
1.340 schwarze 2089: return inloop;
1.67 kristaps 2090: }
2091:
1.364 schwarze 2092: /*
2093: * Handle the closing \} of a conditional block.
2094: * Apart from generating warnings, this only pops nodes.
2095: * Return the number of loops ended.
2096: */
1.340 schwarze 2097: static int
1.195 schwarze 2098: roff_ccond(struct roff *r, int ln, int ppos)
1.74 kristaps 2099: {
1.76 kristaps 2100: if (NULL == r->last) {
1.350 schwarze 2101: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "\\}");
1.340 schwarze 2102: return 0;
1.76 kristaps 2103: }
2104:
1.82 kristaps 2105: switch (r->last->tok) {
1.207 schwarze 2106: case ROFF_el:
2107: case ROFF_ie:
2108: case ROFF_if:
1.340 schwarze 2109: case ROFF_while:
1.82 kristaps 2110: break;
2111: default:
1.350 schwarze 2112: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "\\}");
1.340 schwarze 2113: return 0;
1.75 kristaps 2114: }
2115:
1.76 kristaps 2116: if (r->last->endspan > -1) {
1.350 schwarze 2117: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "\\}");
1.340 schwarze 2118: return 0;
1.76 kristaps 2119: }
2120:
1.340 schwarze 2121: return roffnode_pop(r) + roffnode_cleanscope(r);
1.76 kristaps 2122: }
2123:
1.340 schwarze 2124: static int
1.80 kristaps 2125: roff_block(ROFF_ARGS)
1.76 kristaps 2126: {
1.315 schwarze 2127: const char *name, *value;
2128: char *call, *cp, *iname, *rname;
2129: size_t csz, namesz, rsz;
2130: int deftype;
1.106 kristaps 2131:
1.220 schwarze 2132: /* Ignore groff compatibility mode for now. */
1.76 kristaps 2133:
1.238 schwarze 2134: if (tok == ROFF_de1)
1.220 schwarze 2135: tok = ROFF_de;
1.251 schwarze 2136: else if (tok == ROFF_dei1)
2137: tok = ROFF_dei;
1.238 schwarze 2138: else if (tok == ROFF_am1)
1.220 schwarze 2139: tok = ROFF_am;
1.251 schwarze 2140: else if (tok == ROFF_ami1)
2141: tok = ROFF_ami;
1.220 schwarze 2142:
2143: /* Parse the macro name argument. */
2144:
1.238 schwarze 2145: cp = buf->buf + pos;
2146: if (tok == ROFF_ig) {
1.220 schwarze 2147: iname = NULL;
2148: namesz = 0;
2149: } else {
2150: iname = cp;
2151: namesz = roff_getname(r, &cp, ln, ppos);
2152: iname[namesz] = '\0';
2153: }
1.107 kristaps 2154:
1.220 schwarze 2155: /* Resolve the macro name argument if it is indirect. */
1.107 kristaps 2156:
1.238 schwarze 2157: if (namesz && (tok == ROFF_dei || tok == ROFF_ami)) {
1.315 schwarze 2158: deftype = ROFFDEF_USER;
2159: name = roff_getstrn(r, iname, namesz, &deftype);
2160: if (name == NULL) {
1.350 schwarze 2161: mandoc_msg(MANDOCERR_STR_UNDEF,
2162: ln, (int)(iname - buf->buf),
1.220 schwarze 2163: "%.*s", (int)namesz, iname);
2164: namesz = 0;
2165: } else
2166: namesz = strlen(name);
2167: } else
2168: name = iname;
1.107 kristaps 2169:
1.238 schwarze 2170: if (namesz == 0 && tok != ROFF_ig) {
1.350 schwarze 2171: mandoc_msg(MANDOCERR_REQ_EMPTY,
2172: ln, ppos, "%s", roff_name[tok]);
1.277 schwarze 2173: return ROFF_IGN;
1.220 schwarze 2174: }
1.80 kristaps 2175:
1.106 kristaps 2176: roffnode_push(r, tok, name, ln, ppos);
2177:
2178: /*
2179: * At the beginning of a `de' macro, clear the existing string
2180: * with the same name, if there is one. New content will be
1.193 schwarze 2181: * appended from roff_block_text() in multiline mode.
1.106 kristaps 2182: */
1.107 kristaps 2183:
1.311 schwarze 2184: if (tok == ROFF_de || tok == ROFF_dei) {
1.213 schwarze 2185: roff_setstrn(&r->strtab, name, namesz, "", 0, 0);
1.311 schwarze 2186: roff_setstrn(&r->rentab, name, namesz, NULL, 0, 0);
1.315 schwarze 2187: } else if (tok == ROFF_am || tok == ROFF_ami) {
2188: deftype = ROFFDEF_ANY;
2189: value = roff_getstrn(r, iname, namesz, &deftype);
2190: switch (deftype) { /* Before appending, ... */
2191: case ROFFDEF_PRE: /* copy predefined to user-defined. */
2192: roff_setstrn(&r->strtab, name, namesz,
2193: value, strlen(value), 0);
2194: break;
2195: case ROFFDEF_REN: /* call original standard macro. */
2196: csz = mandoc_asprintf(&call, ".%.*s \\$* \\\"\n",
2197: (int)strlen(value), value);
2198: roff_setstrn(&r->strtab, name, namesz, call, csz, 0);
2199: roff_setstrn(&r->rentab, name, namesz, NULL, 0, 0);
2200: free(call);
2201: break;
2202: case ROFFDEF_STD: /* rename and call standard macro. */
2203: rsz = mandoc_asprintf(&rname, "__%s_renamed", name);
2204: roff_setstrn(&r->rentab, rname, rsz, name, namesz, 0);
2205: csz = mandoc_asprintf(&call, ".%.*s \\$* \\\"\n",
2206: (int)rsz, rname);
2207: roff_setstrn(&r->strtab, name, namesz, call, csz, 0);
2208: free(call);
2209: free(rname);
2210: break;
2211: default:
2212: break;
2213: }
1.311 schwarze 2214: }
1.76 kristaps 2215:
1.238 schwarze 2216: if (*cp == '\0')
1.277 schwarze 2217: return ROFF_IGN;
1.78 kristaps 2218:
1.220 schwarze 2219: /* Get the custom end marker. */
1.107 kristaps 2220:
1.220 schwarze 2221: iname = cp;
1.213 schwarze 2222: namesz = roff_getname(r, &cp, ln, ppos);
1.220 schwarze 2223:
2224: /* Resolve the end marker if it is indirect. */
2225:
1.238 schwarze 2226: if (namesz && (tok == ROFF_dei || tok == ROFF_ami)) {
1.315 schwarze 2227: deftype = ROFFDEF_USER;
2228: name = roff_getstrn(r, iname, namesz, &deftype);
2229: if (name == NULL) {
1.350 schwarze 2230: mandoc_msg(MANDOCERR_STR_UNDEF,
2231: ln, (int)(iname - buf->buf),
1.220 schwarze 2232: "%.*s", (int)namesz, iname);
2233: namesz = 0;
2234: } else
2235: namesz = strlen(name);
2236: } else
2237: name = iname;
2238:
1.213 schwarze 2239: if (namesz)
2240: r->last->end = mandoc_strndup(name, namesz);
1.78 kristaps 2241:
1.238 schwarze 2242: if (*cp != '\0')
1.350 schwarze 2243: mandoc_msg(MANDOCERR_ARG_EXCESS,
1.295 schwarze 2244: ln, pos, ".%s ... %s", roff_name[tok], cp);
1.74 kristaps 2245:
1.277 schwarze 2246: return ROFF_IGN;
1.78 kristaps 2247: }
2248:
1.340 schwarze 2249: static int
1.80 kristaps 2250: roff_block_sub(ROFF_ARGS)
1.79 kristaps 2251: {
1.294 schwarze 2252: enum roff_tok t;
1.79 kristaps 2253: int i, j;
2254:
2255: /*
2256: * First check whether a custom macro exists at this level. If
2257: * it does, then check against it. This is some of groff's
2258: * stranger behaviours. If we encountered a custom end-scope
2259: * tag and that tag also happens to be a "real" macro, then we
2260: * need to try interpreting it again as a real macro. If it's
2261: * not, then return ignore. Else continue.
2262: */
2263:
2264: if (r->last->end) {
1.130 kristaps 2265: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.238 schwarze 2266: if (buf->buf[i] != r->last->end[j])
1.79 kristaps 2267: break;
2268:
1.238 schwarze 2269: if (r->last->end[j] == '\0' &&
2270: (buf->buf[i] == '\0' ||
2271: buf->buf[i] == ' ' ||
2272: buf->buf[i] == '\t')) {
1.79 kristaps 2273: roffnode_pop(r);
2274: roffnode_cleanscope(r);
2275:
1.238 schwarze 2276: while (buf->buf[i] == ' ' || buf->buf[i] == '\t')
1.130 kristaps 2277: i++;
2278:
2279: pos = i;
1.238 schwarze 2280: if (roff_parse(r, buf->buf, &pos, ln, ppos) !=
1.294 schwarze 2281: TOKEN_NONE)
1.277 schwarze 2282: return ROFF_RERUN;
2283: return ROFF_IGN;
1.79 kristaps 2284: }
2285: }
2286:
2287: /*
2288: * If we have no custom end-query or lookup failed, then try
2289: * pulling it out of the hashtable.
2290: */
2291:
1.238 schwarze 2292: t = roff_parse(r, buf->buf, &pos, ln, ppos);
1.79 kristaps 2293:
1.238 schwarze 2294: if (t != ROFF_cblock) {
2295: if (tok != ROFF_ig)
2296: roff_setstr(r, r->last->name, buf->buf + ppos, 2);
1.277 schwarze 2297: return ROFF_IGN;
1.106 kristaps 2298: }
1.79 kristaps 2299:
1.277 schwarze 2300: return (*roffs[t].proc)(r, t, buf, ln, ppos, pos, offs);
1.79 kristaps 2301: }
2302:
1.340 schwarze 2303: static int
1.80 kristaps 2304: roff_block_text(ROFF_ARGS)
1.78 kristaps 2305: {
2306:
1.238 schwarze 2307: if (tok != ROFF_ig)
2308: roff_setstr(r, r->last->name, buf->buf + pos, 2);
1.106 kristaps 2309:
1.277 schwarze 2310: return ROFF_IGN;
1.78 kristaps 2311: }
2312:
1.340 schwarze 2313: static int
1.82 kristaps 2314: roff_cond_sub(ROFF_ARGS)
2315: {
1.364 schwarze 2316: struct roffnode *bl;
1.340 schwarze 2317: char *ep;
2318: int endloop, irc, rr;
1.294 schwarze 2319: enum roff_tok t;
1.82 kristaps 2320:
1.340 schwarze 2321: irc = ROFF_IGN;
1.82 kristaps 2322: rr = r->last->rule;
1.340 schwarze 2323: endloop = tok != ROFF_while ? ROFF_IGN :
2324: rr ? ROFF_LOOPCONT : ROFF_LOOPEXIT;
2325: if (roffnode_cleanscope(r))
2326: irc |= endloop;
1.144 kristaps 2327:
1.196 schwarze 2328: /*
2329: * If `\}' occurs on a macro line without a preceding macro,
2330: * drop the line completely.
2331: */
2332:
1.238 schwarze 2333: ep = buf->buf + pos;
2334: if (ep[0] == '\\' && ep[1] == '}')
1.198 schwarze 2335: rr = 0;
1.196 schwarze 2336:
1.343 schwarze 2337: /*
2338: * The closing delimiter `\}' rewinds the conditional scope
2339: * but is otherwise ignored when interpreting the line.
2340: */
1.144 kristaps 2341:
1.238 schwarze 2342: while ((ep = strchr(ep, '\\')) != NULL) {
1.318 schwarze 2343: switch (ep[1]) {
2344: case '}':
2345: memmove(ep, ep + 2, strlen(ep + 2) + 1);
1.340 schwarze 2346: if (roff_ccond(r, ln, ep - buf->buf))
2347: irc |= endloop;
1.318 schwarze 2348: break;
2349: case '\0':
2350: ++ep;
2351: break;
2352: default:
2353: ep += 2;
2354: break;
1.197 schwarze 2355: }
1.177 schwarze 2356: }
1.318 schwarze 2357:
1.368 schwarze 2358: t = roff_parse(r, buf->buf, &pos, ln, ppos);
2359:
2360: /* For now, let high level macros abort .ce mode. */
2361:
2362: if (roffce_node != NULL &&
2363: (t == TOKEN_NONE || t == ROFF_Dd || t == ROFF_EQ ||
2364: t == ROFF_TH || t == ROFF_TS)) {
2365: r->man->last = roffce_node;
2366: r->man->next = ROFF_NEXT_SIBLING;
2367: roffce_lines = 0;
2368: roffce_node = NULL;
2369: }
2370:
1.318 schwarze 2371: /*
2372: * Fully handle known macros when they are structurally
2373: * required or when the conditional evaluated to true.
2374: */
2375:
1.364 schwarze 2376: if (t == ROFF_break) {
2377: if (irc & ROFF_LOOPMASK)
2378: irc = ROFF_IGN | ROFF_LOOPEXIT;
2379: else if (rr) {
2380: for (bl = r->last; bl != NULL; bl = bl->parent) {
2381: bl->rule = 0;
2382: if (bl->tok == ROFF_while)
2383: break;
2384: }
2385: }
2386: } else if (t != TOKEN_NONE &&
2387: (rr || roffs[t].flags & ROFFMAC_STRUCT))
2388: irc |= (*roffs[t].proc)(r, t, buf, ln, ppos, pos, offs);
2389: else
2390: irc |= rr ? ROFF_CONT : ROFF_IGN;
1.340 schwarze 2391: return irc;
1.82 kristaps 2392: }
2393:
1.340 schwarze 2394: static int
1.82 kristaps 2395: roff_cond_text(ROFF_ARGS)
1.78 kristaps 2396: {
1.140 kristaps 2397: char *ep;
1.340 schwarze 2398: int endloop, irc, rr;
1.82 kristaps 2399:
1.340 schwarze 2400: irc = ROFF_IGN;
1.82 kristaps 2401: rr = r->last->rule;
1.340 schwarze 2402: endloop = tok != ROFF_while ? ROFF_IGN :
2403: rr ? ROFF_LOOPCONT : ROFF_LOOPEXIT;
2404: if (roffnode_cleanscope(r))
2405: irc |= endloop;
1.82 kristaps 2406:
1.343 schwarze 2407: /*
2408: * If `\}' occurs on a text line with neither preceding
2409: * nor following characters, drop the line completely.
2410: */
2411:
1.238 schwarze 2412: ep = buf->buf + pos;
1.343 schwarze 2413: if (strcmp(ep, "\\}") == 0)
2414: rr = 0;
2415:
2416: /*
2417: * The closing delimiter `\}' rewinds the conditional scope
2418: * but is otherwise ignored when interpreting the line.
2419: */
2420:
1.238 schwarze 2421: while ((ep = strchr(ep, '\\')) != NULL) {
1.343 schwarze 2422: switch (ep[1]) {
2423: case '}':
2424: memmove(ep, ep + 2, strlen(ep + 2) + 1);
2425: if (roff_ccond(r, ln, ep - buf->buf))
1.340 schwarze 2426: irc |= endloop;
1.343 schwarze 2427: break;
2428: case '\0':
2429: ++ep;
2430: break;
2431: default:
2432: ep += 2;
2433: break;
1.197 schwarze 2434: }
1.78 kristaps 2435: }
1.340 schwarze 2436: if (rr)
2437: irc |= ROFF_CONT;
2438: return irc;
1.74 kristaps 2439: }
2440:
1.266 schwarze 2441: /* --- handling of numeric and conditional expressions -------------------- */
2442:
1.204 schwarze 2443: /*
2444: * Parse a single signed integer number. Stop at the first non-digit.
2445: * If there is at least one digit, return success and advance the
2446: * parse point, else return failure and let the parse point unchanged.
2447: * Ignore overflows, treat them just like the C language.
2448: */
1.184 schwarze 2449: static int
1.261 schwarze 2450: roff_getnum(const char *v, int *pos, int *res, int flags)
1.184 schwarze 2451: {
1.261 schwarze 2452: int myres, scaled, n, p;
1.206 schwarze 2453:
2454: if (NULL == res)
2455: res = &myres;
1.184 schwarze 2456:
2457: p = *pos;
2458: n = v[p] == '-';
1.261 schwarze 2459: if (n || v[p] == '+')
1.184 schwarze 2460: p++;
2461:
1.261 schwarze 2462: if (flags & ROFFNUM_WHITE)
2463: while (isspace((unsigned char)v[p]))
2464: p++;
2465:
1.184 schwarze 2466: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1.204 schwarze 2467: *res = 10 * *res + v[p] - '0';
1.184 schwarze 2468: if (p == *pos + n)
2469: return 0;
2470:
2471: if (n)
2472: *res = -*res;
2473:
1.254 schwarze 2474: /* Each number may be followed by one optional scaling unit. */
2475:
2476: switch (v[p]) {
2477: case 'f':
1.261 schwarze 2478: scaled = *res * 65536;
1.254 schwarze 2479: break;
2480: case 'i':
1.261 schwarze 2481: scaled = *res * 240;
1.254 schwarze 2482: break;
2483: case 'c':
1.261 schwarze 2484: scaled = *res * 240 / 2.54;
1.254 schwarze 2485: break;
2486: case 'v':
2487: case 'P':
1.261 schwarze 2488: scaled = *res * 40;
1.254 schwarze 2489: break;
2490: case 'm':
2491: case 'n':
1.261 schwarze 2492: scaled = *res * 24;
1.254 schwarze 2493: break;
2494: case 'p':
1.261 schwarze 2495: scaled = *res * 10 / 3;
1.254 schwarze 2496: break;
2497: case 'u':
1.261 schwarze 2498: scaled = *res;
1.254 schwarze 2499: break;
2500: case 'M':
1.261 schwarze 2501: scaled = *res * 6 / 25;
1.254 schwarze 2502: break;
2503: default:
1.261 schwarze 2504: scaled = *res;
1.254 schwarze 2505: p--;
2506: break;
2507: }
1.261 schwarze 2508: if (flags & ROFFNUM_SCALE)
2509: *res = scaled;
1.254 schwarze 2510:
2511: *pos = p + 1;
1.277 schwarze 2512: return 1;
1.184 schwarze 2513: }
2514:
1.198 schwarze 2515: /*
2516: * Evaluate a string comparison condition.
2517: * The first character is the delimiter.
2518: * Succeed if the string up to its second occurrence
2519: * matches the string up to its third occurence.
2520: * Advance the cursor after the third occurrence
2521: * or lacking that, to the end of the line.
2522: */
2523: static int
2524: roff_evalstrcond(const char *v, int *pos)
2525: {
2526: const char *s1, *s2, *s3;
2527: int match;
2528:
2529: match = 0;
2530: s1 = v + *pos; /* initial delimiter */
2531: s2 = s1 + 1; /* for scanning the first string */
2532: s3 = strchr(s2, *s1); /* for scanning the second string */
2533:
2534: if (NULL == s3) /* found no middle delimiter */
2535: goto out;
2536:
2537: while ('\0' != *++s3) {
2538: if (*s2 != *s3) { /* mismatch */
2539: s3 = strchr(s3, *s1);
2540: break;
2541: }
2542: if (*s3 == *s1) { /* found the final delimiter */
2543: match = 1;
2544: break;
2545: }
2546: s2++;
2547: }
2548:
2549: out:
2550: if (NULL == s3)
2551: s3 = strchr(s2, '\0');
1.242 schwarze 2552: else if (*s3 != '\0')
1.198 schwarze 2553: s3++;
2554: *pos = s3 - v;
1.277 schwarze 2555: return match;
1.198 schwarze 2556: }
2557:
1.204 schwarze 2558: /*
2559: * Evaluate an optionally negated single character, numerical,
2560: * or string condition.
2561: */
1.198 schwarze 2562: static int
1.271 schwarze 2563: roff_evalcond(struct roff *r, int ln, char *v, int *pos)
1.88 kristaps 2564: {
1.336 schwarze 2565: const char *start, *end;
2566: char *cp, *name;
2567: size_t sz;
2568: int deftype, len, number, savepos, istrue, wanttrue;
1.88 kristaps 2569:
1.198 schwarze 2570: if ('!' == v[*pos]) {
2571: wanttrue = 0;
2572: (*pos)++;
2573: } else
2574: wanttrue = 1;
2575:
1.88 kristaps 2576: switch (v[*pos]) {
1.240 schwarze 2577: case '\0':
1.277 schwarze 2578: return 0;
1.207 schwarze 2579: case 'n':
2580: case 'o':
1.88 kristaps 2581: (*pos)++;
1.277 schwarze 2582: return wanttrue;
1.207 schwarze 2583: case 'e':
2584: case 't':
1.239 schwarze 2585: case 'v':
1.88 kristaps 2586: (*pos)++;
1.277 schwarze 2587: return !wanttrue;
1.336 schwarze 2588: case 'c':
2589: do {
2590: (*pos)++;
2591: } while (v[*pos] == ' ');
2592:
2593: /*
2594: * Quirk for groff compatibility:
2595: * The horizontal tab is neither available nor unavailable.
2596: */
2597:
2598: if (v[*pos] == '\t') {
2599: (*pos)++;
2600: return 0;
2601: }
2602:
2603: /* Printable ASCII characters are available. */
2604:
2605: if (v[*pos] != '\\') {
2606: (*pos)++;
2607: return wanttrue;
2608: }
2609:
2610: end = v + ++*pos;
2611: switch (mandoc_escape(&end, &start, &len)) {
2612: case ESCAPE_SPECIAL:
2613: istrue = mchars_spec2cp(start, len) != -1;
2614: break;
2615: case ESCAPE_UNICODE:
2616: istrue = 1;
2617: break;
2618: case ESCAPE_NUMBERED:
2619: istrue = mchars_num2char(start, len) != -1;
2620: break;
2621: default:
2622: istrue = !wanttrue;
2623: break;
2624: }
2625: *pos = end - v;
2626: return istrue == wanttrue;
1.310 schwarze 2627: case 'd':
1.271 schwarze 2628: case 'r':
1.310 schwarze 2629: cp = v + *pos + 1;
2630: while (*cp == ' ')
2631: cp++;
2632: name = cp;
2633: sz = roff_getname(r, &cp, ln, cp - v);
1.315 schwarze 2634: if (sz == 0)
2635: istrue = 0;
2636: else if (v[*pos] == 'r')
2637: istrue = roff_hasregn(r, name, sz);
2638: else {
2639: deftype = ROFFDEF_ANY;
2640: roff_getstrn(r, name, sz, &deftype);
2641: istrue = !!deftype;
2642: }
1.363 schwarze 2643: *pos = (name + sz) - v;
1.310 schwarze 2644: return istrue == wanttrue;
1.88 kristaps 2645: default:
2646: break;
2647: }
2648:
1.241 schwarze 2649: savepos = *pos;
1.261 schwarze 2650: if (roff_evalnum(r, ln, v, pos, &number, ROFFNUM_SCALE))
1.277 schwarze 2651: return (number > 0) == wanttrue;
1.241 schwarze 2652: else if (*pos == savepos)
1.277 schwarze 2653: return roff_evalstrcond(v, pos) == wanttrue;
1.204 schwarze 2654: else
1.277 schwarze 2655: return 0;
1.88 kristaps 2656: }
2657:
1.340 schwarze 2658: static int
1.103 kristaps 2659: roff_line_ignore(ROFF_ARGS)
1.89 kristaps 2660: {
1.123 schwarze 2661:
1.277 schwarze 2662: return ROFF_IGN;
1.89 kristaps 2663: }
2664:
1.340 schwarze 2665: static int
1.251 schwarze 2666: roff_insec(ROFF_ARGS)
2667: {
2668:
1.350 schwarze 2669: mandoc_msg(MANDOCERR_REQ_INSEC, ln, ppos, "%s", roff_name[tok]);
1.277 schwarze 2670: return ROFF_IGN;
1.251 schwarze 2671: }
2672:
1.340 schwarze 2673: static int
1.251 schwarze 2674: roff_unsupp(ROFF_ARGS)
2675: {
2676:
1.350 schwarze 2677: mandoc_msg(MANDOCERR_REQ_UNSUPP, ln, ppos, "%s", roff_name[tok]);
1.277 schwarze 2678: return ROFF_IGN;
1.251 schwarze 2679: }
2680:
1.340 schwarze 2681: static int
1.82 kristaps 2682: roff_cond(ROFF_ARGS)
1.74 kristaps 2683: {
1.340 schwarze 2684: int irc;
1.173 schwarze 2685:
2686: roffnode_push(r, tok, NULL, ln, ppos);
1.74 kristaps 2687:
1.207 schwarze 2688: /*
1.134 kristaps 2689: * An `.el' has no conditional body: it will consume the value
2690: * of the current rstack entry set in prior `ie' calls or
1.207 schwarze 2691: * defaults to DENY.
1.134 kristaps 2692: *
2693: * If we're not an `el', however, then evaluate the conditional.
2694: */
1.133 kristaps 2695:
1.238 schwarze 2696: r->last->rule = tok == ROFF_el ?
1.207 schwarze 2697: (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
1.238 schwarze 2698: roff_evalcond(r, ln, buf->buf, &pos);
1.77 kristaps 2699:
1.134 kristaps 2700: /*
2701: * An if-else will put the NEGATION of the current evaluated
2702: * conditional into the stack of rules.
2703: */
2704:
1.238 schwarze 2705: if (tok == ROFF_ie) {
1.223 schwarze 2706: if (r->rstackpos + 1 == r->rstacksz) {
2707: r->rstacksz += 16;
2708: r->rstack = mandoc_reallocarray(r->rstack,
2709: r->rstacksz, sizeof(int));
1.134 kristaps 2710: }
1.198 schwarze 2711: r->rstack[++r->rstackpos] = !r->last->rule;
1.82 kristaps 2712: }
1.88 kristaps 2713:
2714: /* If the parent has false as its rule, then so do we. */
2715:
1.198 schwarze 2716: if (r->last->parent && !r->last->parent->rule)
2717: r->last->rule = 0;
1.88 kristaps 2718:
2719: /*
1.173 schwarze 2720: * Determine scope.
2721: * If there is nothing on the line after the conditional,
2722: * not even whitespace, use next-line scope.
1.340 schwarze 2723: * Except that .while does not support next-line scope.
1.88 kristaps 2724: */
1.74 kristaps 2725:
1.340 schwarze 2726: if (buf->buf[pos] == '\0' && tok != ROFF_while) {
1.173 schwarze 2727: r->last->endspan = 2;
2728: goto out;
2729: }
2730:
1.238 schwarze 2731: while (buf->buf[pos] == ' ')
1.173 schwarze 2732: pos++;
2733:
2734: /* An opening brace requests multiline scope. */
1.75 kristaps 2735:
1.238 schwarze 2736: if (buf->buf[pos] == '\\' && buf->buf[pos + 1] == '{') {
1.75 kristaps 2737: r->last->endspan = -1;
2738: pos += 2;
1.272 schwarze 2739: while (buf->buf[pos] == ' ')
2740: pos++;
1.173 schwarze 2741: goto out;
1.207 schwarze 2742: }
1.74 kristaps 2743:
1.77 kristaps 2744: /*
1.173 schwarze 2745: * Anything else following the conditional causes
2746: * single-line scope. Warn if the scope contains
2747: * nothing but trailing whitespace.
1.77 kristaps 2748: */
2749:
1.238 schwarze 2750: if (buf->buf[pos] == '\0')
1.350 schwarze 2751: mandoc_msg(MANDOCERR_COND_EMPTY,
2752: ln, ppos, "%s", roff_name[tok]);
1.77 kristaps 2753:
1.173 schwarze 2754: r->last->endspan = 1;
1.74 kristaps 2755:
1.173 schwarze 2756: out:
1.75 kristaps 2757: *offs = pos;
1.340 schwarze 2758: irc = ROFF_RERUN;
2759: if (tok == ROFF_while)
2760: irc |= ROFF_WHILE;
2761: return irc;
1.83 schwarze 2762: }
2763:
1.340 schwarze 2764: static int
1.92 schwarze 2765: roff_ds(ROFF_ARGS)
2766: {
1.212 schwarze 2767: char *string;
2768: const char *name;
2769: size_t namesz;
1.96 kristaps 2770:
1.251 schwarze 2771: /* Ignore groff compatibility mode for now. */
2772:
2773: if (tok == ROFF_ds1)
2774: tok = ROFF_ds;
2775: else if (tok == ROFF_as1)
2776: tok = ROFF_as;
2777:
1.96 kristaps 2778: /*
1.212 schwarze 2779: * The first word is the name of the string.
2780: * If it is empty or terminated by an escape sequence,
2781: * abort the `ds' request without defining anything.
1.96 kristaps 2782: */
1.92 schwarze 2783:
1.238 schwarze 2784: name = string = buf->buf + pos;
2785: if (*name == '\0')
1.277 schwarze 2786: return ROFF_IGN;
1.92 schwarze 2787:
1.212 schwarze 2788: namesz = roff_getname(r, &string, ln, pos);
1.363 schwarze 2789: switch (name[namesz]) {
2790: case '\\':
1.277 schwarze 2791: return ROFF_IGN;
1.363 schwarze 2792: case '\t':
2793: string = buf->buf + pos + namesz;
2794: break;
2795: default:
2796: break;
2797: }
1.212 schwarze 2798:
2799: /* Read past the initial double-quote, if any. */
1.238 schwarze 2800: if (*string == '"')
1.92 schwarze 2801: string++;
2802:
1.96 kristaps 2803: /* The rest is the value. */
1.212 schwarze 2804: roff_setstrn(&r->strtab, name, namesz, string, strlen(string),
2805: ROFF_as == tok);
1.311 schwarze 2806: roff_setstrn(&r->rentab, name, namesz, NULL, 0, 0);
1.277 schwarze 2807: return ROFF_IGN;
1.92 schwarze 2808: }
2809:
1.204 schwarze 2810: /*
2811: * Parse a single operator, one or two characters long.
2812: * If the operator is recognized, return success and advance the
2813: * parse point, else return failure and let the parse point unchanged.
2814: */
2815: static int
2816: roff_getop(const char *v, int *pos, char *res)
2817: {
2818:
2819: *res = v[*pos];
2820:
2821: switch (*res) {
1.207 schwarze 2822: case '+':
2823: case '-':
2824: case '*':
2825: case '/':
2826: case '%':
2827: case '&':
2828: case ':':
1.204 schwarze 2829: break;
2830: case '<':
2831: switch (v[*pos + 1]) {
1.207 schwarze 2832: case '=':
1.204 schwarze 2833: *res = 'l';
2834: (*pos)++;
2835: break;
1.207 schwarze 2836: case '>':
1.204 schwarze 2837: *res = '!';
2838: (*pos)++;
2839: break;
1.207 schwarze 2840: case '?':
1.204 schwarze 2841: *res = 'i';
2842: (*pos)++;
2843: break;
2844: default:
2845: break;
2846: }
2847: break;
2848: case '>':
2849: switch (v[*pos + 1]) {
1.207 schwarze 2850: case '=':
1.204 schwarze 2851: *res = 'g';
2852: (*pos)++;
2853: break;
1.207 schwarze 2854: case '?':
1.204 schwarze 2855: *res = 'a';
2856: (*pos)++;
2857: break;
2858: default:
2859: break;
2860: }
2861: break;
2862: case '=':
2863: if ('=' == v[*pos + 1])
2864: (*pos)++;
2865: break;
2866: default:
1.277 schwarze 2867: return 0;
1.204 schwarze 2868: }
2869: (*pos)++;
2870:
1.277 schwarze 2871: return *res;
1.204 schwarze 2872: }
2873:
2874: /*
2875: * Evaluate either a parenthesized numeric expression
2876: * or a single signed integer number.
2877: */
2878: static int
1.235 schwarze 2879: roff_evalpar(struct roff *r, int ln,
1.261 schwarze 2880: const char *v, int *pos, int *res, int flags)
1.204 schwarze 2881: {
2882:
2883: if ('(' != v[*pos])
1.277 schwarze 2884: return roff_getnum(v, pos, res, flags);
1.204 schwarze 2885:
2886: (*pos)++;
1.261 schwarze 2887: if ( ! roff_evalnum(r, ln, v, pos, res, flags | ROFFNUM_WHITE))
1.277 schwarze 2888: return 0;
1.204 schwarze 2889:
1.206 schwarze 2890: /*
2891: * Omission of the closing parenthesis
2892: * is an error in validation mode,
2893: * but ignored in evaluation mode.
2894: */
2895:
1.204 schwarze 2896: if (')' == v[*pos])
2897: (*pos)++;
1.206 schwarze 2898: else if (NULL == res)
1.277 schwarze 2899: return 0;
1.204 schwarze 2900:
1.277 schwarze 2901: return 1;
1.204 schwarze 2902: }
2903:
2904: /*
2905: * Evaluate a complete numeric expression.
2906: * Proceed left to right, there is no concept of precedence.
2907: */
2908: static int
1.235 schwarze 2909: roff_evalnum(struct roff *r, int ln, const char *v,
1.261 schwarze 2910: int *pos, int *res, int flags)
1.204 schwarze 2911: {
2912: int mypos, operand2;
2913: char operator;
2914:
2915: if (NULL == pos) {
2916: mypos = 0;
2917: pos = &mypos;
2918: }
2919:
1.261 schwarze 2920: if (flags & ROFFNUM_WHITE)
1.204 schwarze 2921: while (isspace((unsigned char)v[*pos]))
2922: (*pos)++;
2923:
1.261 schwarze 2924: if ( ! roff_evalpar(r, ln, v, pos, res, flags))
1.277 schwarze 2925: return 0;
1.204 schwarze 2926:
2927: while (1) {
1.261 schwarze 2928: if (flags & ROFFNUM_WHITE)
1.204 schwarze 2929: while (isspace((unsigned char)v[*pos]))
2930: (*pos)++;
2931:
2932: if ( ! roff_getop(v, pos, &operator))
2933: break;
2934:
1.261 schwarze 2935: if (flags & ROFFNUM_WHITE)
1.204 schwarze 2936: while (isspace((unsigned char)v[*pos]))
2937: (*pos)++;
2938:
1.261 schwarze 2939: if ( ! roff_evalpar(r, ln, v, pos, &operand2, flags))
1.277 schwarze 2940: return 0;
1.204 schwarze 2941:
1.261 schwarze 2942: if (flags & ROFFNUM_WHITE)
1.204 schwarze 2943: while (isspace((unsigned char)v[*pos]))
2944: (*pos)++;
1.206 schwarze 2945:
2946: if (NULL == res)
2947: continue;
1.204 schwarze 2948:
2949: switch (operator) {
1.207 schwarze 2950: case '+':
1.204 schwarze 2951: *res += operand2;
2952: break;
1.207 schwarze 2953: case '-':
1.204 schwarze 2954: *res -= operand2;
2955: break;
1.207 schwarze 2956: case '*':
1.204 schwarze 2957: *res *= operand2;
2958: break;
1.207 schwarze 2959: case '/':
1.244 schwarze 2960: if (operand2 == 0) {
1.235 schwarze 2961: mandoc_msg(MANDOCERR_DIVZERO,
1.350 schwarze 2962: ln, *pos, "%s", v);
1.234 kristaps 2963: *res = 0;
2964: break;
2965: }
1.204 schwarze 2966: *res /= operand2;
2967: break;
1.207 schwarze 2968: case '%':
1.244 schwarze 2969: if (operand2 == 0) {
2970: mandoc_msg(MANDOCERR_DIVZERO,
1.350 schwarze 2971: ln, *pos, "%s", v);
1.244 schwarze 2972: *res = 0;
2973: break;
2974: }
1.204 schwarze 2975: *res %= operand2;
2976: break;
1.207 schwarze 2977: case '<':
1.204 schwarze 2978: *res = *res < operand2;
2979: break;
1.207 schwarze 2980: case '>':
1.204 schwarze 2981: *res = *res > operand2;
2982: break;
1.207 schwarze 2983: case 'l':
1.204 schwarze 2984: *res = *res <= operand2;
2985: break;
1.207 schwarze 2986: case 'g':
1.204 schwarze 2987: *res = *res >= operand2;
2988: break;
1.207 schwarze 2989: case '=':
1.204 schwarze 2990: *res = *res == operand2;
2991: break;
1.207 schwarze 2992: case '!':
1.204 schwarze 2993: *res = *res != operand2;
2994: break;
1.207 schwarze 2995: case '&':
1.204 schwarze 2996: *res = *res && operand2;
2997: break;
1.207 schwarze 2998: case ':':
1.204 schwarze 2999: *res = *res || operand2;
3000: break;
1.207 schwarze 3001: case 'i':
1.204 schwarze 3002: if (operand2 < *res)
3003: *res = operand2;
3004: break;
1.207 schwarze 3005: case 'a':
1.204 schwarze 3006: if (operand2 > *res)
3007: *res = operand2;
3008: break;
3009: default:
3010: abort();
3011: }
3012: }
1.277 schwarze 3013: return 1;
1.204 schwarze 3014: }
3015:
1.266 schwarze 3016: /* --- register management ------------------------------------------------ */
3017:
1.180 schwarze 3018: void
1.187 schwarze 3019: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.147 kristaps 3020: {
1.327 schwarze 3021: roff_setregn(r, name, strlen(name), val, sign, INT_MIN);
1.326 schwarze 3022: }
3023:
3024: static void
3025: roff_setregn(struct roff *r, const char *name, size_t len,
1.327 schwarze 3026: int val, char sign, int step)
1.326 schwarze 3027: {
1.180 schwarze 3028: struct roffreg *reg;
3029:
3030: /* Search for an existing register with the same name. */
3031: reg = r->regtab;
3032:
1.326 schwarze 3033: while (reg != NULL && (reg->key.sz != len ||
3034: strncmp(reg->key.p, name, len) != 0))
1.180 schwarze 3035: reg = reg->next;
1.147 kristaps 3036:
1.180 schwarze 3037: if (NULL == reg) {
3038: /* Create a new register. */
3039: reg = mandoc_malloc(sizeof(struct roffreg));
1.326 schwarze 3040: reg->key.p = mandoc_strndup(name, len);
3041: reg->key.sz = len;
1.187 schwarze 3042: reg->val = 0;
1.327 schwarze 3043: reg->step = 0;
1.180 schwarze 3044: reg->next = r->regtab;
3045: r->regtab = reg;
3046: }
3047:
1.187 schwarze 3048: if ('+' == sign)
3049: reg->val += val;
3050: else if ('-' == sign)
3051: reg->val -= val;
3052: else
3053: reg->val = val;
1.327 schwarze 3054: if (step != INT_MIN)
3055: reg->step = step;
1.147 kristaps 3056: }
3057:
1.192 schwarze 3058: /*
3059: * Handle some predefined read-only number registers.
3060: * For now, return -1 if the requested register is not predefined;
3061: * in case a predefined read-only register having the value -1
3062: * were to turn up, another special value would have to be chosen.
3063: */
3064: static int
1.273 schwarze 3065: roff_getregro(const struct roff *r, const char *name)
1.192 schwarze 3066: {
3067:
3068: switch (*name) {
1.273 schwarze 3069: case '$': /* Number of arguments of the last macro evaluated. */
1.339 schwarze 3070: return r->mstackpos < 0 ? 0 : r->mstack[r->mstackpos].argc;
1.207 schwarze 3071: case 'A': /* ASCII approximation mode is always off. */
1.277 schwarze 3072: return 0;
1.207 schwarze 3073: case 'g': /* Groff compatibility mode is always on. */
1.277 schwarze 3074: return 1;
1.207 schwarze 3075: case 'H': /* Fixed horizontal resolution. */
1.277 schwarze 3076: return 24;
1.207 schwarze 3077: case 'j': /* Always adjust left margin only. */
1.277 schwarze 3078: return 0;
1.207 schwarze 3079: case 'T': /* Some output device is always defined. */
1.277 schwarze 3080: return 1;
1.207 schwarze 3081: case 'V': /* Fixed vertical resolution. */
1.277 schwarze 3082: return 40;
1.192 schwarze 3083: default:
1.277 schwarze 3084: return -1;
1.192 schwarze 3085: }
3086: }
3087:
1.181 schwarze 3088: int
1.326 schwarze 3089: roff_getreg(struct roff *r, const char *name)
1.147 kristaps 3090: {
1.327 schwarze 3091: return roff_getregn(r, name, strlen(name), '\0');
1.181 schwarze 3092: }
3093:
3094: static int
1.327 schwarze 3095: roff_getregn(struct roff *r, const char *name, size_t len, char sign)
1.181 schwarze 3096: {
3097: struct roffreg *reg;
1.192 schwarze 3098: int val;
3099:
3100: if ('.' == name[0] && 2 == len) {
1.273 schwarze 3101: val = roff_getregro(r, name + 1);
1.192 schwarze 3102: if (-1 != val)
1.277 schwarze 3103: return val;
1.192 schwarze 3104: }
1.181 schwarze 3105:
1.327 schwarze 3106: for (reg = r->regtab; reg; reg = reg->next) {
1.181 schwarze 3107: if (len == reg->key.sz &&
1.327 schwarze 3108: 0 == strncmp(name, reg->key.p, len)) {
3109: switch (sign) {
3110: case '+':
3111: reg->val += reg->step;
3112: break;
3113: case '-':
3114: reg->val -= reg->step;
3115: break;
3116: default:
3117: break;
3118: }
1.277 schwarze 3119: return reg->val;
1.327 schwarze 3120: }
3121: }
1.271 schwarze 3122:
1.327 schwarze 3123: roff_setregn(r, name, len, 0, '\0', INT_MIN);
1.277 schwarze 3124: return 0;
1.271 schwarze 3125: }
3126:
3127: static int
3128: roff_hasregn(const struct roff *r, const char *name, size_t len)
3129: {
3130: struct roffreg *reg;
3131: int val;
3132:
3133: if ('.' == name[0] && 2 == len) {
1.273 schwarze 3134: val = roff_getregro(r, name + 1);
1.271 schwarze 3135: if (-1 != val)
1.277 schwarze 3136: return 1;
1.271 schwarze 3137: }
3138:
3139: for (reg = r->regtab; reg; reg = reg->next)
3140: if (len == reg->key.sz &&
3141: 0 == strncmp(name, reg->key.p, len))
1.277 schwarze 3142: return 1;
1.147 kristaps 3143:
1.277 schwarze 3144: return 0;
1.147 kristaps 3145: }
3146:
1.180 schwarze 3147: static void
3148: roff_freereg(struct roffreg *reg)
1.147 kristaps 3149: {
1.180 schwarze 3150: struct roffreg *old_reg;
1.147 kristaps 3151:
1.180 schwarze 3152: while (NULL != reg) {
3153: free(reg->key.p);
3154: old_reg = reg;
3155: reg = reg->next;
3156: free(old_reg);
3157: }
1.147 kristaps 3158: }
1.92 schwarze 3159:
1.340 schwarze 3160: static int
1.89 kristaps 3161: roff_nr(ROFF_ARGS)
1.83 schwarze 3162: {
1.327 schwarze 3163: char *key, *val, *step;
1.212 schwarze 3164: size_t keysz;
1.327 schwarze 3165: int iv, is, len;
1.187 schwarze 3166: char sign;
1.89 kristaps 3167:
1.238 schwarze 3168: key = val = buf->buf + pos;
3169: if (*key == '\0')
1.277 schwarze 3170: return ROFF_IGN;
1.212 schwarze 3171:
3172: keysz = roff_getname(r, &val, ln, pos);
1.363 schwarze 3173: if (key[keysz] == '\\' || key[keysz] == '\t')
1.277 schwarze 3174: return ROFF_IGN;
1.89 kristaps 3175:
1.187 schwarze 3176: sign = *val;
1.238 schwarze 3177: if (sign == '+' || sign == '-')
1.187 schwarze 3178: val++;
3179:
1.327 schwarze 3180: len = 0;
3181: if (roff_evalnum(r, ln, val, &len, &iv, ROFFNUM_SCALE) == 0)
3182: return ROFF_IGN;
3183:
3184: step = val + len;
3185: while (isspace((unsigned char)*step))
3186: step++;
3187: if (roff_evalnum(r, ln, step, NULL, &is, 0) == 0)
3188: is = INT_MIN;
1.109 kristaps 3189:
1.327 schwarze 3190: roff_setregn(r, key, keysz, iv, sign, is);
1.277 schwarze 3191: return ROFF_IGN;
1.203 schwarze 3192: }
3193:
1.340 schwarze 3194: static int
1.203 schwarze 3195: roff_rr(ROFF_ARGS)
3196: {
3197: struct roffreg *reg, **prev;
1.212 schwarze 3198: char *name, *cp;
3199: size_t namesz;
1.203 schwarze 3200:
1.238 schwarze 3201: name = cp = buf->buf + pos;
3202: if (*name == '\0')
1.277 schwarze 3203: return ROFF_IGN;
1.212 schwarze 3204: namesz = roff_getname(r, &cp, ln, pos);
3205: name[namesz] = '\0';
1.203 schwarze 3206:
3207: prev = &r->regtab;
3208: while (1) {
3209: reg = *prev;
1.238 schwarze 3210: if (reg == NULL || !strcmp(name, reg->key.p))
1.203 schwarze 3211: break;
3212: prev = ®->next;
3213: }
1.238 schwarze 3214: if (reg != NULL) {
1.203 schwarze 3215: *prev = reg->next;
3216: free(reg->key.p);
3217: free(reg);
3218: }
1.277 schwarze 3219: return ROFF_IGN;
1.122 schwarze 3220: }
3221:
1.266 schwarze 3222: /* --- handler functions for roff requests -------------------------------- */
3223:
1.340 schwarze 3224: static int
1.122 schwarze 3225: roff_rm(ROFF_ARGS)
3226: {
3227: const char *name;
3228: char *cp;
1.212 schwarze 3229: size_t namesz;
1.122 schwarze 3230:
1.238 schwarze 3231: cp = buf->buf + pos;
3232: while (*cp != '\0') {
1.212 schwarze 3233: name = cp;
1.238 schwarze 3234: namesz = roff_getname(r, &cp, ln, (int)(cp - buf->buf));
1.212 schwarze 3235: roff_setstrn(&r->strtab, name, namesz, NULL, 0, 0);
1.311 schwarze 3236: roff_setstrn(&r->rentab, name, namesz, NULL, 0, 0);
1.363 schwarze 3237: if (name[namesz] == '\\' || name[namesz] == '\t')
1.212 schwarze 3238: break;
1.122 schwarze 3239: }
1.277 schwarze 3240: return ROFF_IGN;
1.178 schwarze 3241: }
3242:
1.340 schwarze 3243: static int
1.178 schwarze 3244: roff_it(ROFF_ARGS)
3245: {
3246: int iv;
3247:
3248: /* Parse the number of lines. */
1.261 schwarze 3249:
3250: if ( ! roff_evalnum(r, ln, buf->buf, &pos, &iv, 0)) {
1.350 schwarze 3251: mandoc_msg(MANDOCERR_IT_NONUM,
3252: ln, ppos, "%s", buf->buf + 1);
1.277 schwarze 3253: return ROFF_IGN;
1.178 schwarze 3254: }
3255:
1.262 schwarze 3256: while (isspace((unsigned char)buf->buf[pos]))
3257: pos++;
3258:
3259: /*
3260: * Arm the input line trap.
3261: * Special-casing "an-trap" is an ugly workaround to cope
3262: * with DocBook stupidly fiddling with man(7) internals.
3263: */
1.261 schwarze 3264:
1.178 schwarze 3265: roffit_lines = iv;
1.262 schwarze 3266: roffit_macro = mandoc_strdup(iv != 1 ||
3267: strcmp(buf->buf + pos, "an-trap") ?
3268: buf->buf + pos : "br");
1.277 schwarze 3269: return ROFF_IGN;
1.175 schwarze 3270: }
3271:
1.340 schwarze 3272: static int
1.175 schwarze 3273: roff_Dd(ROFF_ARGS)
3274: {
1.315 schwarze 3275: int mask;
3276: enum roff_tok t, te;
1.227 schwarze 3277:
1.315 schwarze 3278: switch (tok) {
3279: case ROFF_Dd:
3280: tok = MDOC_Dd;
3281: te = MDOC_MAX;
3282: if (r->format == 0)
3283: r->format = MPARSE_MDOC;
3284: mask = MPARSE_MDOC | MPARSE_QUICK;
3285: break;
3286: case ROFF_TH:
3287: tok = MAN_TH;
3288: te = MAN_MAX;
3289: if (r->format == 0)
3290: r->format = MPARSE_MAN;
3291: mask = MPARSE_QUICK;
3292: break;
3293: default:
3294: abort();
3295: }
3296: if ((r->options & mask) == 0)
3297: for (t = tok; t < te; t++)
3298: roff_setstr(r, roff_name[t], NULL, 0);
1.277 schwarze 3299: return ROFF_CONT;
1.109 kristaps 3300: }
3301:
1.340 schwarze 3302: static int
1.109 kristaps 3303: roff_TE(ROFF_ARGS)
3304: {
1.361 schwarze 3305: r->man->flags &= ~ROFF_NONOFILL;
1.321 schwarze 3306: if (r->tbl == NULL) {
1.350 schwarze 3307: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "TE");
1.321 schwarze 3308: return ROFF_IGN;
3309: }
1.346 schwarze 3310: if (tbl_end(r->tbl, 0) == 0) {
1.321 schwarze 3311: r->tbl = NULL;
1.258 schwarze 3312: free(buf->buf);
3313: buf->buf = mandoc_strdup(".sp");
3314: buf->sz = 4;
1.329 schwarze 3315: *offs = 0;
1.277 schwarze 3316: return ROFF_REPARSE;
1.258 schwarze 3317: }
1.321 schwarze 3318: r->tbl = NULL;
1.277 schwarze 3319: return ROFF_IGN;
1.112 kristaps 3320: }
3321:
1.340 schwarze 3322: static int
1.112 kristaps 3323: roff_T_(ROFF_ARGS)
3324: {
3325:
3326: if (NULL == r->tbl)
1.350 schwarze 3327: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "T&");
1.112 kristaps 3328: else
1.296 schwarze 3329: tbl_restart(ln, ppos, r->tbl);
1.112 kristaps 3330:
1.277 schwarze 3331: return ROFF_IGN;
1.109 kristaps 3332: }
3333:
1.230 schwarze 3334: /*
3335: * Handle in-line equation delimiters.
3336: */
1.340 schwarze 3337: static int
1.238 schwarze 3338: roff_eqndelim(struct roff *r, struct buf *buf, int pos)
1.151 kristaps 3339: {
1.233 schwarze 3340: char *cp1, *cp2;
3341: const char *bef_pr, *bef_nl, *mac, *aft_nl, *aft_pr;
1.151 kristaps 3342:
1.230 schwarze 3343: /*
3344: * Outside equations, look for an opening delimiter.
3345: * If we are inside an equation, we already know it is
3346: * in-line, or this function wouldn't have been called;
3347: * so look for a closing delimiter.
3348: */
3349:
1.238 schwarze 3350: cp1 = buf->buf + pos;
1.230 schwarze 3351: cp2 = strchr(cp1, r->eqn == NULL ?
3352: r->last_eqn->odelim : r->last_eqn->cdelim);
3353: if (cp2 == NULL)
1.277 schwarze 3354: return ROFF_CONT;
1.230 schwarze 3355:
1.233 schwarze 3356: *cp2++ = '\0';
3357: bef_pr = bef_nl = aft_nl = aft_pr = "";
3358:
3359: /* Handle preceding text, protecting whitespace. */
3360:
1.238 schwarze 3361: if (*buf->buf != '\0') {
1.233 schwarze 3362: if (r->eqn == NULL)
3363: bef_pr = "\\&";
3364: bef_nl = "\n";
3365: }
3366:
3367: /*
3368: * Prepare replacing the delimiter with an equation macro
3369: * and drop leading white space from the equation.
3370: */
1.230 schwarze 3371:
1.233 schwarze 3372: if (r->eqn == NULL) {
3373: while (*cp2 == ' ')
3374: cp2++;
3375: mac = ".EQ";
3376: } else
3377: mac = ".EN";
3378:
3379: /* Handle following text, protecting whitespace. */
3380:
3381: if (*cp2 != '\0') {
3382: aft_nl = "\n";
3383: if (r->eqn != NULL)
3384: aft_pr = "\\&";
3385: }
3386:
3387: /* Do the actual replacement. */
3388:
1.238 schwarze 3389: buf->sz = mandoc_asprintf(&cp1, "%s%s%s%s%s%s%s", buf->buf,
1.233 schwarze 3390: bef_pr, bef_nl, mac, aft_nl, aft_pr, cp2) + 1;
1.238 schwarze 3391: free(buf->buf);
3392: buf->buf = cp1;
1.230 schwarze 3393:
3394: /* Toggle the in-line state of the eqn subsystem. */
3395:
3396: r->eqn_inline = r->eqn == NULL;
1.277 schwarze 3397: return ROFF_REPARSE;
1.151 kristaps 3398: }
3399:
1.340 schwarze 3400: static int
1.235 schwarze 3401: roff_EQ(ROFF_ARGS)
1.125 kristaps 3402: {
1.319 schwarze 3403: struct roff_node *n;
3404:
1.356 schwarze 3405: if (r->man->meta.macroset == MACROSET_MAN)
1.322 schwarze 3406: man_breakscope(r->man, ROFF_EQ);
1.319 schwarze 3407: n = roff_node_alloc(r->man, ln, ppos, ROFFT_EQN, TOKEN_NONE);
3408: if (ln > r->man->last->line)
3409: n->flags |= NODE_LINE;
1.348 schwarze 3410: n->eqn = eqn_box_new();
1.319 schwarze 3411: roff_node_append(r->man, n);
3412: r->man->next = ROFF_NEXT_SIBLING;
1.125 kristaps 3413:
1.238 schwarze 3414: assert(r->eqn == NULL);
1.319 schwarze 3415: if (r->last_eqn == NULL)
1.351 schwarze 3416: r->last_eqn = eqn_alloc();
1.319 schwarze 3417: else
3418: eqn_reset(r->last_eqn);
3419: r->eqn = r->last_eqn;
3420: r->eqn->node = n;
1.151 kristaps 3421:
1.238 schwarze 3422: if (buf->buf[pos] != '\0')
1.350 schwarze 3423: mandoc_msg(MANDOCERR_ARG_SKIP, ln, pos,
1.238 schwarze 3424: ".EQ %s", buf->buf + pos);
1.151 kristaps 3425:
1.277 schwarze 3426: return ROFF_IGN;
1.125 kristaps 3427: }
3428:
1.340 schwarze 3429: static int
1.125 kristaps 3430: roff_EN(ROFF_ARGS)
3431: {
1.319 schwarze 3432: if (r->eqn != NULL) {
3433: eqn_parse(r->eqn);
3434: r->eqn = NULL;
3435: } else
1.350 schwarze 3436: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "EN");
1.319 schwarze 3437: if (buf->buf[pos] != '\0')
1.350 schwarze 3438: mandoc_msg(MANDOCERR_ARG_SKIP, ln, pos,
1.319 schwarze 3439: "EN %s", buf->buf + pos);
1.277 schwarze 3440: return ROFF_IGN;
1.125 kristaps 3441: }
3442:
1.340 schwarze 3443: static int
1.109 kristaps 3444: roff_TS(ROFF_ARGS)
3445: {
1.321 schwarze 3446: if (r->tbl != NULL) {
1.350 schwarze 3447: mandoc_msg(MANDOCERR_BLK_BROKEN, ln, ppos, "TS breaks TS");
1.346 schwarze 3448: tbl_end(r->tbl, 0);
1.115 kristaps 3449: }
1.361 schwarze 3450: r->man->flags |= ROFF_NONOFILL;
1.351 schwarze 3451: r->tbl = tbl_alloc(ppos, ln, r->last_tbl);
1.346 schwarze 3452: if (r->last_tbl == NULL)
1.321 schwarze 3453: r->first_tbl = r->tbl;
3454: r->last_tbl = r->tbl;
1.297 schwarze 3455: return ROFF_IGN;
3456: }
3457:
1.340 schwarze 3458: static int
1.358 schwarze 3459: roff_noarg(ROFF_ARGS)
3460: {
3461: if (r->man->flags & (MAN_BLINE | MAN_ELINE))
3462: man_breakscope(r->man, tok);
3463: if (tok == ROFF_brp)
3464: tok = ROFF_br;
3465: roff_elem_alloc(r->man, ln, ppos, tok);
3466: if (buf->buf[pos] != '\0')
3467: mandoc_msg(MANDOCERR_ARG_SKIP, ln, pos,
3468: "%s %s", roff_name[tok], buf->buf + pos);
3469: if (tok == ROFF_nf)
3470: r->man->flags |= ROFF_NOFILL;
3471: else if (tok == ROFF_fi)
3472: r->man->flags &= ~ROFF_NOFILL;
3473: r->man->last->flags |= NODE_LINE | NODE_VALID | NODE_ENDED;
3474: r->man->next = ROFF_NEXT_SIBLING;
3475: return ROFF_IGN;
3476: }
3477:
3478: static int
1.297 schwarze 3479: roff_onearg(ROFF_ARGS)
3480: {
3481: struct roff_node *n;
3482: char *cp;
1.305 schwarze 3483: int npos;
1.297 schwarze 3484:
1.302 schwarze 3485: if (r->man->flags & (MAN_BLINE | MAN_ELINE) &&
1.320 schwarze 3486: (tok == ROFF_ce || tok == ROFF_rj || tok == ROFF_sp ||
3487: tok == ROFF_ti))
1.302 schwarze 3488: man_breakscope(r->man, tok);
3489:
1.309 schwarze 3490: if (roffce_node != NULL && (tok == ROFF_ce || tok == ROFF_rj)) {
1.305 schwarze 3491: r->man->last = roffce_node;
3492: r->man->next = ROFF_NEXT_SIBLING;
3493: }
3494:
1.297 schwarze 3495: roff_elem_alloc(r->man, ln, ppos, tok);
3496: n = r->man->last;
3497:
3498: cp = buf->buf + pos;
3499: if (*cp != '\0') {
3500: while (*cp != '\0' && *cp != ' ')
3501: cp++;
3502: while (*cp == ' ')
3503: *cp++ = '\0';
3504: if (*cp != '\0')
1.350 schwarze 3505: mandoc_msg(MANDOCERR_ARG_EXCESS,
3506: ln, (int)(cp - buf->buf),
1.297 schwarze 3507: "%s ... %s", roff_name[tok], cp);
3508: roff_word_alloc(r->man, ln, pos, buf->buf + pos);
1.300 schwarze 3509: }
3510:
1.309 schwarze 3511: if (tok == ROFF_ce || tok == ROFF_rj) {
3512: if (r->man->last->type == ROFFT_ELEM) {
1.305 schwarze 3513: roff_word_alloc(r->man, ln, pos, "1");
3514: r->man->last->flags |= NODE_NOSRC;
3515: }
3516: npos = 0;
3517: if (roff_evalnum(r, ln, r->man->last->string, &npos,
3518: &roffce_lines, 0) == 0) {
1.350 schwarze 3519: mandoc_msg(MANDOCERR_CE_NONUM,
3520: ln, pos, "ce %s", buf->buf + pos);
1.305 schwarze 3521: roffce_lines = 1;
3522: }
3523: if (roffce_lines < 1) {
3524: r->man->last = r->man->last->parent;
3525: roffce_node = NULL;
3526: roffce_lines = 0;
3527: } else
3528: roffce_node = r->man->last->parent;
3529: } else {
3530: n->flags |= NODE_VALID | NODE_ENDED;
3531: r->man->last = n;
3532: }
3533: n->flags |= NODE_LINE;
1.300 schwarze 3534: r->man->next = ROFF_NEXT_SIBLING;
3535: return ROFF_IGN;
3536: }
3537:
1.340 schwarze 3538: static int
1.300 schwarze 3539: roff_manyarg(ROFF_ARGS)
3540: {
3541: struct roff_node *n;
3542: char *sp, *ep;
3543:
3544: roff_elem_alloc(r->man, ln, ppos, tok);
3545: n = r->man->last;
3546:
3547: for (sp = ep = buf->buf + pos; *sp != '\0'; sp = ep) {
3548: while (*ep != '\0' && *ep != ' ')
3549: ep++;
3550: while (*ep == ' ')
3551: *ep++ = '\0';
3552: roff_word_alloc(r->man, ln, sp - buf->buf, sp);
1.297 schwarze 3553: }
3554:
3555: n->flags |= NODE_LINE | NODE_VALID | NODE_ENDED;
3556: r->man->last = n;
3557: r->man->next = ROFF_NEXT_SIBLING;
1.277 schwarze 3558: return ROFF_IGN;
1.251 schwarze 3559: }
3560:
1.340 schwarze 3561: static int
1.311 schwarze 3562: roff_als(ROFF_ARGS)
3563: {
3564: char *oldn, *newn, *end, *value;
3565: size_t oldsz, newsz, valsz;
3566:
3567: newn = oldn = buf->buf + pos;
3568: if (*newn == '\0')
3569: return ROFF_IGN;
3570:
3571: newsz = roff_getname(r, &oldn, ln, pos);
1.363 schwarze 3572: if (newn[newsz] == '\\' || newn[newsz] == '\t' || *oldn == '\0')
1.311 schwarze 3573: return ROFF_IGN;
3574:
3575: end = oldn;
3576: oldsz = roff_getname(r, &end, ln, oldn - buf->buf);
3577: if (oldsz == 0)
3578: return ROFF_IGN;
3579:
1.338 schwarze 3580: valsz = mandoc_asprintf(&value, ".%.*s \\$@\\\"\n",
1.315 schwarze 3581: (int)oldsz, oldn);
1.311 schwarze 3582: roff_setstrn(&r->strtab, newn, newsz, value, valsz, 0);
3583: roff_setstrn(&r->rentab, newn, newsz, NULL, 0, 0);
3584: free(value);
1.364 schwarze 3585: return ROFF_IGN;
3586: }
3587:
3588: /*
3589: * The .break request only makes sense inside conditionals,
3590: * and that case is already handled in roff_cond_sub().
3591: */
3592: static int
3593: roff_break(ROFF_ARGS)
3594: {
3595: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, pos, "break");
1.296 schwarze 3596: return ROFF_IGN;
1.92 schwarze 3597: }
3598:
1.340 schwarze 3599: static int
1.174 kristaps 3600: roff_cc(ROFF_ARGS)
3601: {
3602: const char *p;
3603:
1.238 schwarze 3604: p = buf->buf + pos;
1.174 kristaps 3605:
1.238 schwarze 3606: if (*p == '\0' || (r->control = *p++) == '.')
1.303 schwarze 3607: r->control = '\0';
1.174 kristaps 3608:
1.238 schwarze 3609: if (*p != '\0')
1.350 schwarze 3610: mandoc_msg(MANDOCERR_ARG_EXCESS,
1.260 schwarze 3611: ln, p - buf->buf, "cc ... %s", p);
1.174 kristaps 3612:
1.341 schwarze 3613: return ROFF_IGN;
3614: }
3615:
3616: static int
3617: roff_char(ROFF_ARGS)
3618: {
3619: const char *p, *kp, *vp;
3620: size_t ksz, vsz;
3621: int font;
3622:
3623: /* Parse the character to be replaced. */
3624:
3625: kp = buf->buf + pos;
3626: p = kp + 1;
3627: if (*kp == '\0' || (*kp == '\\' &&
3628: mandoc_escape(&p, NULL, NULL) != ESCAPE_SPECIAL) ||
3629: (*p != ' ' && *p != '\0')) {
1.350 schwarze 3630: mandoc_msg(MANDOCERR_CHAR_ARG, ln, pos, "char %s", kp);
1.341 schwarze 3631: return ROFF_IGN;
3632: }
3633: ksz = p - kp;
3634: while (*p == ' ')
3635: p++;
3636:
3637: /*
3638: * If the replacement string contains a font escape sequence,
3639: * we have to restore the font at the end.
3640: */
3641:
3642: vp = p;
3643: vsz = strlen(p);
3644: font = 0;
3645: while (*p != '\0') {
3646: if (*p++ != '\\')
3647: continue;
3648: switch (mandoc_escape(&p, NULL, NULL)) {
3649: case ESCAPE_FONT:
3650: case ESCAPE_FONTROMAN:
3651: case ESCAPE_FONTITALIC:
3652: case ESCAPE_FONTBOLD:
3653: case ESCAPE_FONTBI:
1.342 schwarze 3654: case ESCAPE_FONTCW:
1.341 schwarze 3655: case ESCAPE_FONTPREV:
3656: font++;
3657: break;
3658: default:
3659: break;
3660: }
3661: }
3662: if (font > 1)
1.350 schwarze 3663: mandoc_msg(MANDOCERR_CHAR_FONT,
3664: ln, (int)(vp - buf->buf), "%s", vp);
1.341 schwarze 3665:
3666: /*
3667: * Approximate the effect of .char using the .tr tables.
3668: * XXX In groff, .char and .tr interact differently.
3669: */
3670:
3671: if (ksz == 1) {
3672: if (r->xtab == NULL)
3673: r->xtab = mandoc_calloc(128, sizeof(*r->xtab));
3674: assert((unsigned int)*kp < 128);
3675: free(r->xtab[(int)*kp].p);
3676: r->xtab[(int)*kp].sz = mandoc_asprintf(&r->xtab[(int)*kp].p,
3677: "%s%s", vp, font ? "\fP" : "");
3678: } else {
3679: roff_setstrn(&r->xmbtab, kp, ksz, vp, vsz, 0);
3680: if (font)
3681: roff_setstrn(&r->xmbtab, kp, ksz, "\\fP", 3, 1);
3682: }
1.277 schwarze 3683: return ROFF_IGN;
1.174 kristaps 3684: }
3685:
1.340 schwarze 3686: static int
1.303 schwarze 3687: roff_ec(ROFF_ARGS)
3688: {
3689: const char *p;
3690:
3691: p = buf->buf + pos;
3692: if (*p == '\0')
3693: r->escape = '\\';
3694: else {
3695: r->escape = *p;
3696: if (*++p != '\0')
1.350 schwarze 3697: mandoc_msg(MANDOCERR_ARG_EXCESS, ln,
3698: (int)(p - buf->buf), "ec ... %s", p);
1.303 schwarze 3699: }
3700: return ROFF_IGN;
3701: }
3702:
1.340 schwarze 3703: static int
1.303 schwarze 3704: roff_eo(ROFF_ARGS)
3705: {
3706: r->escape = '\0';
3707: if (buf->buf[pos] != '\0')
1.350 schwarze 3708: mandoc_msg(MANDOCERR_ARG_SKIP,
1.303 schwarze 3709: ln, pos, "eo %s", buf->buf + pos);
3710: return ROFF_IGN;
1.330 schwarze 3711: }
3712:
1.340 schwarze 3713: static int
1.330 schwarze 3714: roff_nop(ROFF_ARGS)
3715: {
3716: while (buf->buf[pos] == ' ')
3717: pos++;
3718: *offs = pos;
3719: return ROFF_RERUN;
1.303 schwarze 3720: }
3721:
1.340 schwarze 3722: static int
1.164 kristaps 3723: roff_tr(ROFF_ARGS)
3724: {
3725: const char *p, *first, *second;
3726: size_t fsz, ssz;
3727: enum mandoc_esc esc;
3728:
1.238 schwarze 3729: p = buf->buf + pos;
1.164 kristaps 3730:
1.238 schwarze 3731: if (*p == '\0') {
1.350 schwarze 3732: mandoc_msg(MANDOCERR_REQ_EMPTY, ln, ppos, "tr");
1.277 schwarze 3733: return ROFF_IGN;
1.164 kristaps 3734: }
3735:
1.238 schwarze 3736: while (*p != '\0') {
1.164 kristaps 3737: fsz = ssz = 1;
3738:
3739: first = p++;
1.238 schwarze 3740: if (*first == '\\') {
1.164 kristaps 3741: esc = mandoc_escape(&p, NULL, NULL);
1.238 schwarze 3742: if (esc == ESCAPE_ERROR) {
1.350 schwarze 3743: mandoc_msg(MANDOCERR_ESC_BAD, ln,
3744: (int)(p - buf->buf), "%s", first);
1.277 schwarze 3745: return ROFF_IGN;
1.164 kristaps 3746: }
3747: fsz = (size_t)(p - first);
3748: }
3749:
3750: second = p++;
1.238 schwarze 3751: if (*second == '\\') {
1.164 kristaps 3752: esc = mandoc_escape(&p, NULL, NULL);
1.238 schwarze 3753: if (esc == ESCAPE_ERROR) {
1.350 schwarze 3754: mandoc_msg(MANDOCERR_ESC_BAD, ln,
3755: (int)(p - buf->buf), "%s", second);
1.277 schwarze 3756: return ROFF_IGN;
1.164 kristaps 3757: }
3758: ssz = (size_t)(p - second);
1.238 schwarze 3759: } else if (*second == '\0') {
1.350 schwarze 3760: mandoc_msg(MANDOCERR_TR_ODD, ln,
3761: (int)(first - buf->buf), "tr %s", first);
1.164 kristaps 3762: second = " ";
1.165 kristaps 3763: p--;
1.164 kristaps 3764: }
3765:
1.167 kristaps 3766: if (fsz > 1) {
1.207 schwarze 3767: roff_setstrn(&r->xmbtab, first, fsz,
3768: second, ssz, 0);
1.167 kristaps 3769: continue;
3770: }
3771:
1.238 schwarze 3772: if (r->xtab == NULL)
1.207 schwarze 3773: r->xtab = mandoc_calloc(128,
3774: sizeof(struct roffstr));
1.167 kristaps 3775:
3776: free(r->xtab[(int)*first].p);
3777: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
3778: r->xtab[(int)*first].sz = ssz;
1.164 kristaps 3779: }
3780:
1.277 schwarze 3781: return ROFF_IGN;
1.164 kristaps 3782: }
3783:
1.339 schwarze 3784: /*
3785: * Implementation of the .return request.
3786: * There is no need to call roff_userret() from here.
3787: * The read module will call that after rewinding the reader stack
3788: * to the place from where the current macro was called.
3789: */
1.340 schwarze 3790: static int
1.339 schwarze 3791: roff_return(ROFF_ARGS)
3792: {
3793: if (r->mstackpos >= 0)
1.340 schwarze 3794: return ROFF_IGN | ROFF_USERRET;
1.339 schwarze 3795:
1.350 schwarze 3796: mandoc_msg(MANDOCERR_REQ_NOMAC, ln, ppos, "return");
1.339 schwarze 3797: return ROFF_IGN;
3798: }
3799:
1.340 schwarze 3800: static int
1.306 schwarze 3801: roff_rn(ROFF_ARGS)
3802: {
3803: const char *value;
3804: char *oldn, *newn, *end;
3805: size_t oldsz, newsz;
1.315 schwarze 3806: int deftype;
1.306 schwarze 3807:
3808: oldn = newn = buf->buf + pos;
3809: if (*oldn == '\0')
3810: return ROFF_IGN;
3811:
3812: oldsz = roff_getname(r, &newn, ln, pos);
1.363 schwarze 3813: if (oldn[oldsz] == '\\' || oldn[oldsz] == '\t' || *newn == '\0')
1.306 schwarze 3814: return ROFF_IGN;
3815:
3816: end = newn;
3817: newsz = roff_getname(r, &end, ln, newn - buf->buf);
3818: if (newsz == 0)
3819: return ROFF_IGN;
3820:
1.315 schwarze 3821: deftype = ROFFDEF_ANY;
3822: value = roff_getstrn(r, oldn, oldsz, &deftype);
3823: switch (deftype) {
3824: case ROFFDEF_USER:
1.306 schwarze 3825: roff_setstrn(&r->strtab, newn, newsz, value, strlen(value), 0);
3826: roff_setstrn(&r->strtab, oldn, oldsz, NULL, 0, 0);
3827: roff_setstrn(&r->rentab, newn, newsz, NULL, 0, 0);
1.315 schwarze 3828: break;
3829: case ROFFDEF_PRE:
3830: roff_setstrn(&r->strtab, newn, newsz, value, strlen(value), 0);
3831: roff_setstrn(&r->rentab, newn, newsz, NULL, 0, 0);
3832: break;
3833: case ROFFDEF_REN:
1.306 schwarze 3834: roff_setstrn(&r->rentab, newn, newsz, value, strlen(value), 0);
3835: roff_setstrn(&r->rentab, oldn, oldsz, NULL, 0, 0);
1.315 schwarze 3836: roff_setstrn(&r->strtab, newn, newsz, NULL, 0, 0);
3837: break;
3838: case ROFFDEF_STD:
1.306 schwarze 3839: roff_setstrn(&r->rentab, newn, newsz, oldn, oldsz, 0);
1.315 schwarze 3840: roff_setstrn(&r->strtab, newn, newsz, NULL, 0, 0);
3841: break;
3842: default:
3843: roff_setstrn(&r->strtab, newn, newsz, NULL, 0, 0);
3844: roff_setstrn(&r->rentab, newn, newsz, NULL, 0, 0);
3845: break;
3846: }
1.306 schwarze 3847: return ROFF_IGN;
3848: }
3849:
1.340 schwarze 3850: static int
1.339 schwarze 3851: roff_shift(ROFF_ARGS)
3852: {
3853: struct mctx *ctx;
3854: int levels, i;
3855:
3856: levels = 1;
3857: if (buf->buf[pos] != '\0' &&
3858: roff_evalnum(r, ln, buf->buf, &pos, &levels, 0) == 0) {
1.350 schwarze 3859: mandoc_msg(MANDOCERR_CE_NONUM,
1.339 schwarze 3860: ln, pos, "shift %s", buf->buf + pos);
3861: levels = 1;
3862: }
3863: if (r->mstackpos < 0) {
1.350 schwarze 3864: mandoc_msg(MANDOCERR_REQ_NOMAC, ln, ppos, "shift");
1.339 schwarze 3865: return ROFF_IGN;
3866: }
3867: ctx = r->mstack + r->mstackpos;
3868: if (levels > ctx->argc) {
1.350 schwarze 3869: mandoc_msg(MANDOCERR_SHIFT,
1.339 schwarze 3870: ln, pos, "%d, but max is %d", levels, ctx->argc);
3871: levels = ctx->argc;
3872: }
3873: if (levels == 0)
3874: return ROFF_IGN;
3875: for (i = 0; i < levels; i++)
3876: free(ctx->argv[i]);
3877: ctx->argc -= levels;
3878: for (i = 0; i < ctx->argc; i++)
3879: ctx->argv[i] = ctx->argv[i + levels];
3880: return ROFF_IGN;
3881: }
3882:
1.340 schwarze 3883: static int
1.105 kristaps 3884: roff_so(ROFF_ARGS)
3885: {
1.249 schwarze 3886: char *name, *cp;
1.105 kristaps 3887:
1.238 schwarze 3888: name = buf->buf + pos;
1.350 schwarze 3889: mandoc_msg(MANDOCERR_SO, ln, ppos, "so %s", name);
1.105 kristaps 3890:
3891: /*
3892: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
3893: * opening anything that's not in our cwd or anything beneath
3894: * it. Thus, explicitly disallow traversing up the file-system
3895: * or using absolute paths.
3896: */
3897:
1.238 schwarze 3898: if (*name == '/' || strstr(name, "../") || strstr(name, "/..")) {
1.350 schwarze 3899: mandoc_msg(MANDOCERR_SO_PATH, ln, ppos, ".so %s", name);
1.249 schwarze 3900: buf->sz = mandoc_asprintf(&cp,
3901: ".sp\nSee the file %s.\n.sp", name) + 1;
3902: free(buf->buf);
3903: buf->buf = cp;
3904: *offs = 0;
1.277 schwarze 3905: return ROFF_REPARSE;
1.105 kristaps 3906: }
3907:
3908: *offs = pos;
1.277 schwarze 3909: return ROFF_SO;
1.105 kristaps 3910: }
1.92 schwarze 3911:
1.266 schwarze 3912: /* --- user defined strings and macros ------------------------------------ */
3913:
1.340 schwarze 3914: static int
1.106 kristaps 3915: roff_userdef(ROFF_ARGS)
1.99 kristaps 3916: {
1.339 schwarze 3917: struct mctx *ctx;
3918: char *arg, *ap, *dst, *src;
3919: size_t sz;
3920:
1.365 schwarze 3921: /* If the macro is empty, ignore it altogether. */
3922:
3923: if (*r->current_string == '\0')
3924: return ROFF_IGN;
3925:
1.339 schwarze 3926: /* Initialize a new macro stack context. */
3927:
3928: if (++r->mstackpos == r->mstacksz) {
3929: r->mstack = mandoc_recallocarray(r->mstack,
3930: r->mstacksz, r->mstacksz + 8, sizeof(*r->mstack));
3931: r->mstacksz += 8;
3932: }
3933: ctx = r->mstack + r->mstackpos;
3934: ctx->argsz = 0;
3935: ctx->argc = 0;
3936: ctx->argv = NULL;
3937:
3938: /*
3939: * Collect pointers to macro argument strings,
3940: * NUL-terminating them and escaping quotes.
3941: */
3942:
3943: src = buf->buf + pos;
3944: while (*src != '\0') {
3945: if (ctx->argc == ctx->argsz) {
3946: ctx->argsz += 8;
3947: ctx->argv = mandoc_reallocarray(ctx->argv,
3948: ctx->argsz, sizeof(*ctx->argv));
3949: }
1.355 schwarze 3950: arg = roff_getarg(r, &src, ln, &pos);
1.339 schwarze 3951: sz = 1; /* For the terminating NUL. */
3952: for (ap = arg; *ap != '\0'; ap++)
3953: sz += *ap == '"' ? 4 : 1;
3954: ctx->argv[ctx->argc++] = dst = mandoc_malloc(sz);
3955: for (ap = arg; *ap != '\0'; ap++) {
3956: if (*ap == '"') {
3957: memcpy(dst, "\\(dq", 4);
3958: dst += 4;
3959: } else
3960: *dst++ = *ap;
1.106 kristaps 3961: }
1.339 schwarze 3962: *dst = '\0';
1.355 schwarze 3963: free(arg);
1.337 schwarze 3964: }
3965:
1.339 schwarze 3966: /* Replace the macro invocation by the macro definition. */
1.263 schwarze 3967:
1.238 schwarze 3968: free(buf->buf);
1.339 schwarze 3969: buf->buf = mandoc_strdup(r->current_string);
3970: buf->sz = strlen(buf->buf) + 1;
1.248 schwarze 3971: *offs = 0;
1.106 kristaps 3972:
1.365 schwarze 3973: return buf->buf[buf->sz - 2] == '\n' ?
1.340 schwarze 3974: ROFF_REPARSE | ROFF_USERCALL : ROFF_IGN | ROFF_APPEND;
1.99 kristaps 3975: }
1.121 schwarze 3976:
1.306 schwarze 3977: /*
3978: * Calling a high-level macro that was renamed with .rn.
3979: * r->current_string has already been set up by roff_parse().
3980: */
1.340 schwarze 3981: static int
1.306 schwarze 3982: roff_renamed(ROFF_ARGS)
3983: {
3984: char *nbuf;
3985:
1.315 schwarze 3986: buf->sz = mandoc_asprintf(&nbuf, ".%s%s%s", r->current_string,
3987: buf->buf[pos] == '\0' ? "" : " ", buf->buf + pos) + 1;
1.306 schwarze 3988: free(buf->buf);
3989: buf->buf = nbuf;
1.329 schwarze 3990: *offs = 0;
1.306 schwarze 3991: return ROFF_CONT;
3992: }
3993:
1.362 schwarze 3994: /*
3995: * Measure the length in bytes of the roff identifier at *cpp
3996: * and advance the pointer to the next word.
3997: */
1.212 schwarze 3998: static size_t
1.121 schwarze 3999: roff_getname(struct roff *r, char **cpp, int ln, int pos)
4000: {
4001: char *name, *cp;
1.212 schwarze 4002: size_t namesz;
1.121 schwarze 4003:
4004: name = *cpp;
1.362 schwarze 4005: if (*name == '\0')
1.277 schwarze 4006: return 0;
1.121 schwarze 4007:
1.362 schwarze 4008: /* Advance cp to the byte after the end of the name. */
4009:
1.212 schwarze 4010: for (cp = name; 1; cp++) {
1.362 schwarze 4011: namesz = cp - name;
1.363 schwarze 4012: if (*cp == '\0')
1.212 schwarze 4013: break;
1.363 schwarze 4014: if (*cp == ' ' || *cp == '\t') {
4015: cp++;
4016: break;
4017: }
1.362 schwarze 4018: if (*cp != '\\')
1.121 schwarze 4019: continue;
1.362 schwarze 4020: if (cp[1] == '{' || cp[1] == '}')
1.215 schwarze 4021: break;
1.362 schwarze 4022: if (*++cp == '\\')
1.121 schwarze 4023: continue;
1.350 schwarze 4024: mandoc_msg(MANDOCERR_NAMESC, ln, pos,
1.224 schwarze 4025: "%.*s", (int)(cp - name + 1), name);
1.212 schwarze 4026: mandoc_escape((const char **)&cp, NULL, NULL);
4027: break;
1.121 schwarze 4028: }
4029:
4030: /* Read past spaces. */
1.362 schwarze 4031:
4032: while (*cp == ' ')
1.121 schwarze 4033: cp++;
4034:
4035: *cpp = cp;
1.277 schwarze 4036: return namesz;
1.121 schwarze 4037: }
4038:
1.106 kristaps 4039: /*
4040: * Store *string into the user-defined string called *name.
4041: * To clear an existing entry, call with (*r, *name, NULL, 0).
1.193 schwarze 4042: * append == 0: replace mode
4043: * append == 1: single-line append mode
4044: * append == 2: multiline append mode, append '\n' after each call
1.106 kristaps 4045: */
1.94 kristaps 4046: static void
1.106 kristaps 4047: roff_setstr(struct roff *r, const char *name, const char *string,
1.193 schwarze 4048: int append)
1.92 schwarze 4049: {
1.315 schwarze 4050: size_t namesz;
1.164 kristaps 4051:
1.315 schwarze 4052: namesz = strlen(name);
4053: roff_setstrn(&r->strtab, name, namesz, string,
1.207 schwarze 4054: string ? strlen(string) : 0, append);
1.315 schwarze 4055: roff_setstrn(&r->rentab, name, namesz, NULL, 0, 0);
1.164 kristaps 4056: }
4057:
4058: static void
1.166 kristaps 4059: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.193 schwarze 4060: const char *string, size_t stringsz, int append)
1.164 kristaps 4061: {
1.166 kristaps 4062: struct roffkv *n;
1.164 kristaps 4063: char *c;
4064: int i;
4065: size_t oldch, newch;
1.92 schwarze 4066:
1.106 kristaps 4067: /* Search for an existing string with the same name. */
1.164 kristaps 4068: n = *r;
4069:
1.211 schwarze 4070: while (n && (namesz != n->key.sz ||
4071: strncmp(n->key.p, name, namesz)))
1.92 schwarze 4072: n = n->next;
1.94 kristaps 4073:
4074: if (NULL == n) {
1.106 kristaps 4075: /* Create a new string table entry. */
1.166 kristaps 4076: n = mandoc_malloc(sizeof(struct roffkv));
4077: n->key.p = mandoc_strndup(name, namesz);
4078: n->key.sz = namesz;
4079: n->val.p = NULL;
4080: n->val.sz = 0;
1.164 kristaps 4081: n->next = *r;
4082: *r = n;
1.193 schwarze 4083: } else if (0 == append) {
1.166 kristaps 4084: free(n->val.p);
4085: n->val.p = NULL;
4086: n->val.sz = 0;
1.106 kristaps 4087: }
4088:
4089: if (NULL == string)
4090: return;
4091:
4092: /*
4093: * One additional byte for the '\n' in multiline mode,
4094: * and one for the terminating '\0'.
4095: */
1.193 schwarze 4096: newch = stringsz + (1 < append ? 2u : 1u);
1.164 kristaps 4097:
1.166 kristaps 4098: if (NULL == n->val.p) {
4099: n->val.p = mandoc_malloc(newch);
4100: *n->val.p = '\0';
1.106 kristaps 4101: oldch = 0;
4102: } else {
1.166 kristaps 4103: oldch = n->val.sz;
4104: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.106 kristaps 4105: }
4106:
4107: /* Skip existing content in the destination buffer. */
1.166 kristaps 4108: c = n->val.p + (int)oldch;
1.106 kristaps 4109:
4110: /* Append new content to the destination buffer. */
1.164 kristaps 4111: i = 0;
4112: while (i < (int)stringsz) {
1.106 kristaps 4113: /*
4114: * Rudimentary roff copy mode:
4115: * Handle escaped backslashes.
4116: */
1.164 kristaps 4117: if ('\\' == string[i] && '\\' == string[i + 1])
4118: i++;
4119: *c++ = string[i++];
1.106 kristaps 4120: }
1.94 kristaps 4121:
1.106 kristaps 4122: /* Append terminating bytes. */
1.193 schwarze 4123: if (1 < append)
1.106 kristaps 4124: *c++ = '\n';
1.163 kristaps 4125:
1.106 kristaps 4126: *c = '\0';
1.166 kristaps 4127: n->val.sz = (int)(c - n->val.p);
1.92 schwarze 4128: }
4129:
1.94 kristaps 4130: static const char *
1.325 schwarze 4131: roff_getstrn(struct roff *r, const char *name, size_t len,
1.315 schwarze 4132: int *deftype)
1.92 schwarze 4133: {
1.315 schwarze 4134: const struct roffkv *n;
1.325 schwarze 4135: int found, i;
1.315 schwarze 4136: enum roff_tok tok;
4137:
1.325 schwarze 4138: found = 0;
4139: for (n = r->strtab; n != NULL; n = n->next) {
4140: if (strncmp(name, n->key.p, len) != 0 ||
4141: n->key.p[len] != '\0' || n->val.p == NULL)
4142: continue;
4143: if (*deftype & ROFFDEF_USER) {
4144: *deftype = ROFFDEF_USER;
4145: return n->val.p;
4146: } else {
4147: found = 1;
4148: break;
4149: }
4150: }
4151: for (n = r->rentab; n != NULL; n = n->next) {
4152: if (strncmp(name, n->key.p, len) != 0 ||
4153: n->key.p[len] != '\0' || n->val.p == NULL)
4154: continue;
4155: if (*deftype & ROFFDEF_REN) {
4156: *deftype = ROFFDEF_REN;
4157: return n->val.p;
4158: } else {
4159: found = 1;
4160: break;
1.315 schwarze 4161: }
4162: }
1.325 schwarze 4163: for (i = 0; i < PREDEFS_MAX; i++) {
4164: if (strncmp(name, predefs[i].name, len) != 0 ||
4165: predefs[i].name[len] != '\0')
4166: continue;
4167: if (*deftype & ROFFDEF_PRE) {
4168: *deftype = ROFFDEF_PRE;
4169: return predefs[i].str;
4170: } else {
4171: found = 1;
4172: break;
1.315 schwarze 4173: }
4174: }
1.356 schwarze 4175: if (r->man->meta.macroset != MACROSET_MAN) {
1.325 schwarze 4176: for (tok = MDOC_Dd; tok < MDOC_MAX; tok++) {
4177: if (strncmp(name, roff_name[tok], len) != 0 ||
4178: roff_name[tok][len] != '\0')
4179: continue;
4180: if (*deftype & ROFFDEF_STD) {
4181: *deftype = ROFFDEF_STD;
4182: return NULL;
4183: } else {
4184: found = 1;
4185: break;
1.315 schwarze 4186: }
4187: }
4188: }
1.356 schwarze 4189: if (r->man->meta.macroset != MACROSET_MDOC) {
1.325 schwarze 4190: for (tok = MAN_TH; tok < MAN_MAX; tok++) {
4191: if (strncmp(name, roff_name[tok], len) != 0 ||
4192: roff_name[tok][len] != '\0')
4193: continue;
4194: if (*deftype & ROFFDEF_STD) {
4195: *deftype = ROFFDEF_STD;
4196: return NULL;
4197: } else {
4198: found = 1;
4199: break;
1.315 schwarze 4200: }
4201: }
1.325 schwarze 4202: }
4203:
4204: if (found == 0 && *deftype != ROFFDEF_ANY) {
4205: if (*deftype & ROFFDEF_REN) {
4206: /*
4207: * This might still be a request,
4208: * so do not treat it as undefined yet.
4209: */
4210: *deftype = ROFFDEF_UNDEF;
4211: return NULL;
1.315 schwarze 4212: }
1.325 schwarze 4213:
4214: /* Using an undefined string defines it to be empty. */
4215:
4216: roff_setstrn(&r->strtab, name, len, "", 0, 0);
4217: roff_setstrn(&r->rentab, name, len, NULL, 0, 0);
1.315 schwarze 4218: }
1.325 schwarze 4219:
1.315 schwarze 4220: *deftype = 0;
1.277 schwarze 4221: return NULL;
1.92 schwarze 4222: }
4223:
1.94 kristaps 4224: static void
1.167 kristaps 4225: roff_freestr(struct roffkv *r)
1.92 schwarze 4226: {
1.166 kristaps 4227: struct roffkv *n, *nn;
1.92 schwarze 4228:
1.167 kristaps 4229: for (n = r; n; n = nn) {
1.166 kristaps 4230: free(n->key.p);
4231: free(n->val.p);
1.92 schwarze 4232: nn = n->next;
4233: free(n);
4234: }
1.114 kristaps 4235: }
1.266 schwarze 4236:
4237: /* --- accessors and utility functions ------------------------------------ */
1.164 kristaps 4238:
4239: /*
4240: * Duplicate an input string, making the appropriate character
4241: * conversations (as stipulated by `tr') along the way.
4242: * Returns a heap-allocated string with all the replacements made.
4243: */
4244: char *
4245: roff_strdup(const struct roff *r, const char *p)
4246: {
1.166 kristaps 4247: const struct roffkv *cp;
1.164 kristaps 4248: char *res;
4249: const char *pp;
4250: size_t ssz, sz;
4251: enum mandoc_esc esc;
4252:
1.167 kristaps 4253: if (NULL == r->xmbtab && NULL == r->xtab)
1.277 schwarze 4254: return mandoc_strdup(p);
1.164 kristaps 4255: else if ('\0' == *p)
1.277 schwarze 4256: return mandoc_strdup("");
1.164 kristaps 4257:
4258: /*
4259: * Step through each character looking for term matches
4260: * (remember that a `tr' can be invoked with an escape, which is
4261: * a glyph but the escape is multi-character).
4262: * We only do this if the character hash has been initialised
4263: * and the string is >0 length.
4264: */
4265:
4266: res = NULL;
4267: ssz = 0;
4268:
4269: while ('\0' != *p) {
1.289 schwarze 4270: assert((unsigned int)*p < 128);
4271: if ('\\' != *p && r->xtab && r->xtab[(unsigned int)*p].p) {
1.167 kristaps 4272: sz = r->xtab[(int)*p].sz;
4273: res = mandoc_realloc(res, ssz + sz + 1);
4274: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
4275: ssz += sz;
4276: p++;
4277: continue;
4278: } else if ('\\' != *p) {
4279: res = mandoc_realloc(res, ssz + 2);
4280: res[ssz++] = *p++;
4281: continue;
4282: }
4283:
1.164 kristaps 4284: /* Search for term matches. */
1.167 kristaps 4285: for (cp = r->xmbtab; cp; cp = cp->next)
1.166 kristaps 4286: if (0 == strncmp(p, cp->key.p, cp->key.sz))
1.164 kristaps 4287: break;
4288:
4289: if (NULL != cp) {
4290: /*
4291: * A match has been found.
4292: * Append the match to the array and move
4293: * forward by its keysize.
4294: */
1.207 schwarze 4295: res = mandoc_realloc(res,
4296: ssz + cp->val.sz + 1);
1.166 kristaps 4297: memcpy(res + ssz, cp->val.p, cp->val.sz);
4298: ssz += cp->val.sz;
4299: p += (int)cp->key.sz;
1.164 kristaps 4300: continue;
4301: }
4302:
1.167 kristaps 4303: /*
4304: * Handle escapes carefully: we need to copy
4305: * over just the escape itself, or else we might
4306: * do replacements within the escape itself.
4307: * Make sure to pass along the bogus string.
4308: */
4309: pp = p++;
4310: esc = mandoc_escape(&p, NULL, NULL);
4311: if (ESCAPE_ERROR == esc) {
4312: sz = strlen(pp);
1.164 kristaps 4313: res = mandoc_realloc(res, ssz + sz + 1);
4314: memcpy(res + ssz, pp, sz);
1.167 kristaps 4315: break;
1.164 kristaps 4316: }
1.207 schwarze 4317: /*
4318: * We bail out on bad escapes.
1.167 kristaps 4319: * No need to warn: we already did so when
1.355 schwarze 4320: * roff_expand() was called.
1.167 kristaps 4321: */
4322: sz = (int)(p - pp);
4323: res = mandoc_realloc(res, ssz + sz + 1);
4324: memcpy(res + ssz, pp, sz);
4325: ssz += sz;
1.164 kristaps 4326: }
4327:
4328: res[(int)ssz] = '\0';
1.277 schwarze 4329: return res;
1.227 schwarze 4330: }
4331:
4332: int
4333: roff_getformat(const struct roff *r)
4334: {
4335:
1.277 schwarze 4336: return r->format;
1.174 kristaps 4337: }
4338:
4339: /*
1.207 schwarze 4340: * Find out whether a line is a macro line or not.
1.174 kristaps 4341: * If it is, adjust the current position and return one; if it isn't,
4342: * return zero and don't change the current position.
4343: * If the control character has been set with `.cc', then let that grain
4344: * precedence.
4345: * This is slighly contrary to groff, where using the non-breaking
4346: * control character when `cc' has been invoked will cause the
4347: * non-breaking macro contents to be printed verbatim.
4348: */
4349: int
4350: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
4351: {
4352: int pos;
4353:
4354: pos = *ppos;
4355:
1.303 schwarze 4356: if (r->control != '\0' && cp[pos] == r->control)
1.174 kristaps 4357: pos++;
1.303 schwarze 4358: else if (r->control != '\0')
1.277 schwarze 4359: return 0;
1.174 kristaps 4360: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
4361: pos += 2;
4362: else if ('.' == cp[pos] || '\'' == cp[pos])
4363: pos++;
4364: else
1.277 schwarze 4365: return 0;
1.174 kristaps 4366:
4367: while (' ' == cp[pos] || '\t' == cp[pos])
4368: pos++;
4369:
4370: *ppos = pos;
1.277 schwarze 4371: return 1;
1.74 kristaps 4372: }
CVSweb