Annotation of mandoc/roff.c, Revision 1.372
1.372 ! schwarze 1: /* $Id: roff.c,v 1.371 2020/02/27 21:43:44 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;
1126: switch (n->tok) {
1127: case ROFF_ft:
1128: case ROFF_ll:
1129: case ROFF_mc:
1130: case ROFF_po:
1131: case ROFF_ta:
1132: case MDOC_Db:
1133: case MDOC_Es:
1134: case MDOC_Sm:
1135: case MDOC_Tg:
1136: case MAN_DT:
1137: case MAN_UC:
1138: case MAN_PD:
1139: case MAN_AT:
1140: return 1;
1141: default:
1142: return 0;
1143: }
1144: }
1145:
1146: struct roff_node *
1147: roff_node_child(struct roff_node *n)
1148: {
1149: for (n = n->child; roff_node_transparent(n); n = n->next)
1150: continue;
1151: return n;
1152: }
1153:
1154: struct roff_node *
1155: roff_node_prev(struct roff_node *n)
1156: {
1157: do {
1158: n = n->prev;
1159: } while (roff_node_transparent(n));
1160: return n;
1161: }
1162:
1163: struct roff_node *
1164: roff_node_next(struct roff_node *n)
1165: {
1166: do {
1167: n = n->next;
1168: } while (roff_node_transparent(n));
1169: return n;
1.269 schwarze 1170: }
1171:
1172: void
1173: deroff(char **dest, const struct roff_node *n)
1174: {
1175: char *cp;
1176: size_t sz;
1177:
1.371 schwarze 1178: if (n->string == NULL) {
1.269 schwarze 1179: for (n = n->child; n != NULL; n = n->next)
1180: deroff(dest, n);
1181: return;
1182: }
1183:
1.288 schwarze 1184: /* Skip leading whitespace. */
1.269 schwarze 1185:
1.288 schwarze 1186: for (cp = n->string; *cp != '\0'; cp++) {
1.289 schwarze 1187: if (cp[0] == '\\' && cp[1] != '\0' &&
1188: strchr(" %&0^|~", cp[1]) != NULL)
1.269 schwarze 1189: cp++;
1.288 schwarze 1190: else if ( ! isspace((unsigned char)*cp))
1.269 schwarze 1191: break;
1192: }
1193:
1.289 schwarze 1194: /* Skip trailing backslash. */
1195:
1196: sz = strlen(cp);
1.290 schwarze 1197: if (sz > 0 && cp[sz - 1] == '\\')
1.289 schwarze 1198: sz--;
1199:
1.269 schwarze 1200: /* Skip trailing whitespace. */
1201:
1.289 schwarze 1202: for (; sz; sz--)
1.269 schwarze 1203: if ( ! isspace((unsigned char)cp[sz-1]))
1204: break;
1205:
1206: /* Skip empty strings. */
1207:
1208: if (sz == 0)
1209: return;
1210:
1211: if (*dest == NULL) {
1212: *dest = mandoc_strndup(cp, sz);
1213: return;
1214: }
1215:
1216: mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
1217: free(*dest);
1218: *dest = cp;
1.266 schwarze 1219: }
1220:
1221: /* --- main functions of the roff parser ---------------------------------- */
1222:
1.94 kristaps 1223: /*
1.355 schwarze 1224: * In the current line, expand escape sequences that produce parsable
1225: * input text. Also check the syntax of the remaining escape sequences,
1226: * which typically produce output glyphs or change formatter state.
1.154 kristaps 1227: */
1.340 schwarze 1228: static int
1.355 schwarze 1229: roff_expand(struct roff *r, struct buf *buf, int ln, int pos, char newesc)
1.94 kristaps 1230: {
1.339 schwarze 1231: struct mctx *ctx; /* current macro call context */
1.208 schwarze 1232: char ubuf[24]; /* buffer to print the number */
1.328 schwarze 1233: struct roff_node *n; /* used for header comments */
1.205 schwarze 1234: const char *start; /* start of the string to process */
1.209 schwarze 1235: char *stesc; /* start of an escape sequence ('\\') */
1.352 schwarze 1236: const char *esct; /* type of esccape sequence */
1.328 schwarze 1237: char *ep; /* end of comment string */
1.108 schwarze 1238: const char *stnam; /* start of the name, after "[(*" */
1239: const char *cp; /* end of the name, e.g. before ']' */
1240: const char *res; /* the string to be substituted */
1.238 schwarze 1241: char *nbuf; /* new buffer to copy buf->buf to */
1.181 schwarze 1242: size_t maxl; /* expected length of the escape name */
1243: size_t naml; /* actual length of the escape name */
1.339 schwarze 1244: size_t asz; /* length of the replacement */
1245: size_t rsz; /* length of the rest of the string */
1.237 schwarze 1246: int inaml; /* length returned from mandoc_escape() */
1.181 schwarze 1247: int expand_count; /* to avoid infinite loops */
1.206 schwarze 1248: int npos; /* position in numeric expression */
1.218 schwarze 1249: int arg_complete; /* argument not interrupted by eol */
1.339 schwarze 1250: int quote_args; /* true for \\$@, false for \\$* */
1.303 schwarze 1251: int done; /* no more input available */
1.315 schwarze 1252: int deftype; /* type of definition to paste */
1.323 schwarze 1253: int rcsid; /* kind of RCS id seen */
1.352 schwarze 1254: enum mandocerr err; /* for escape sequence problems */
1.327 schwarze 1255: char sign; /* increment number register */
1.206 schwarze 1256: char term; /* character terminating the escape */
1.94 kristaps 1257:
1.303 schwarze 1258: /* Search forward for comments. */
1259:
1260: done = 0;
1261: start = buf->buf + pos;
1262: for (stesc = buf->buf + pos; *stesc != '\0'; stesc++) {
1.355 schwarze 1263: if (stesc[0] != newesc || stesc[1] == '\0')
1.303 schwarze 1264: continue;
1265: stesc++;
1266: if (*stesc != '"' && *stesc != '#')
1267: continue;
1.313 schwarze 1268:
1269: /* Comment found, look for RCS id. */
1270:
1.323 schwarze 1271: rcsid = 0;
1.313 schwarze 1272: if ((cp = strstr(stesc, "$" "OpenBSD")) != NULL) {
1.323 schwarze 1273: rcsid = 1 << MANDOC_OS_OPENBSD;
1.313 schwarze 1274: cp += 8;
1275: } else if ((cp = strstr(stesc, "$" "NetBSD")) != NULL) {
1.323 schwarze 1276: rcsid = 1 << MANDOC_OS_NETBSD;
1.313 schwarze 1277: cp += 7;
1278: }
1279: if (cp != NULL &&
1280: isalnum((unsigned char)*cp) == 0 &&
1.314 schwarze 1281: strchr(cp, '$') != NULL) {
1.323 schwarze 1282: if (r->man->meta.rcsids & rcsid)
1.350 schwarze 1283: mandoc_msg(MANDOCERR_RCS_REP, ln,
1284: (int)(stesc - buf->buf) + 1,
1285: "%s", stesc + 1);
1.323 schwarze 1286: r->man->meta.rcsids |= rcsid;
1.314 schwarze 1287: }
1.313 schwarze 1288:
1289: /* Handle trailing whitespace. */
1290:
1.328 schwarze 1291: ep = strchr(stesc--, '\0') - 1;
1292: if (*ep == '\n') {
1.303 schwarze 1293: done = 1;
1.328 schwarze 1294: ep--;
1.303 schwarze 1295: }
1.328 schwarze 1296: if (*ep == ' ' || *ep == '\t')
1.350 schwarze 1297: mandoc_msg(MANDOCERR_SPACE_EOL,
1298: ln, (int)(ep - buf->buf), NULL);
1.328 schwarze 1299:
1300: /*
1301: * Save comments preceding the title macro
1302: * in the syntax tree.
1303: */
1304:
1.367 schwarze 1305: if (newesc != ASCII_ESC && r->options & MPARSE_COMMENT) {
1.328 schwarze 1306: while (*ep == ' ' || *ep == '\t')
1307: ep--;
1308: ep[1] = '\0';
1309: n = roff_node_alloc(r->man,
1310: ln, stesc + 1 - buf->buf,
1311: ROFFT_COMMENT, TOKEN_NONE);
1312: n->string = mandoc_strdup(stesc + 2);
1313: roff_node_append(r->man, n);
1314: n->flags |= NODE_VALID | NODE_ENDED;
1315: r->man->next = ROFF_NEXT_SIBLING;
1316: }
1317:
1.334 schwarze 1318: /* Line continuation with comment. */
1319:
1320: if (stesc[1] == '#') {
1321: *stesc = '\0';
1.340 schwarze 1322: return ROFF_IGN | ROFF_APPEND;
1.334 schwarze 1323: }
1324:
1325: /* Discard normal comments. */
1.328 schwarze 1326:
1.335 schwarze 1327: while (stesc > start && stesc[-1] == ' ' &&
1328: (stesc == start + 1 || stesc[-2] != '\\'))
1.303 schwarze 1329: stesc--;
1330: *stesc = '\0';
1331: break;
1332: }
1333: if (stesc == start)
1334: return ROFF_CONT;
1335: stesc--;
1336:
1337: /* Notice the end of the input. */
1338:
1339: if (*stesc == '\n') {
1340: *stesc-- = '\0';
1341: done = 1;
1342: }
1343:
1.170 schwarze 1344: expand_count = 0;
1.303 schwarze 1345: while (stesc >= start) {
1.355 schwarze 1346: if (*stesc != newesc) {
1.170 schwarze 1347:
1.355 schwarze 1348: /*
1349: * If we have a non-standard escape character,
1350: * escape literal backslashes because all
1351: * processing in subsequent functions uses
1352: * the standard escaping rules.
1353: */
1.205 schwarze 1354:
1.355 schwarze 1355: if (newesc != ASCII_ESC && *stesc == '\\') {
1.303 schwarze 1356: *stesc = '\0';
1357: buf->sz = mandoc_asprintf(&nbuf, "%s\\e%s",
1358: buf->buf, stesc + 1) + 1;
1359: start = nbuf + pos;
1360: stesc = nbuf + (stesc - buf->buf);
1361: free(buf->buf);
1362: buf->buf = nbuf;
1363: }
1.355 schwarze 1364:
1365: /* Search backwards for the next escape. */
1366:
1.303 schwarze 1367: stesc--;
1.205 schwarze 1368: continue;
1.303 schwarze 1369: }
1.205 schwarze 1370:
1371: /* If it is escaped, skip it. */
1372:
1373: for (cp = stesc - 1; cp >= start; cp--)
1.303 schwarze 1374: if (*cp != r->escape)
1.205 schwarze 1375: break;
1376:
1.238 schwarze 1377: if ((stesc - cp) % 2 == 0) {
1.303 schwarze 1378: while (stesc > cp)
1379: *stesc-- = '\\';
1.205 schwarze 1380: continue;
1.303 schwarze 1381: } else if (stesc[1] != '\0') {
1382: *stesc = '\\';
1383: } else {
1384: *stesc-- = '\0';
1385: if (done)
1386: continue;
1387: else
1.340 schwarze 1388: return ROFF_IGN | ROFF_APPEND;
1.205 schwarze 1389: }
1.108 schwarze 1390:
1.206 schwarze 1391: /* Decide whether to expand or to check only. */
1.108 schwarze 1392:
1.206 schwarze 1393: term = '\0';
1.205 schwarze 1394: cp = stesc + 1;
1.352 schwarze 1395: if (*cp == 'E')
1396: cp++;
1397: esct = cp;
1398: switch (*esct) {
1.207 schwarze 1399: case '*':
1.339 schwarze 1400: case '$':
1.181 schwarze 1401: res = NULL;
1402: break;
1.207 schwarze 1403: case 'B':
1404: case 'w':
1.206 schwarze 1405: term = cp[1];
1406: /* FALLTHROUGH */
1.207 schwarze 1407: case 'n':
1.327 schwarze 1408: sign = cp[1];
1409: if (sign == '+' || sign == '-')
1410: cp++;
1.181 schwarze 1411: res = ubuf;
1412: break;
1413: default:
1.352 schwarze 1414: err = MANDOCERR_OK;
1415: switch(mandoc_escape(&cp, &stnam, &inaml)) {
1416: case ESCAPE_SPECIAL:
1417: if (mchars_spec2cp(stnam, inaml) >= 0)
1418: break;
1419: /* FALLTHROUGH */
1420: case ESCAPE_ERROR:
1421: err = MANDOCERR_ESC_BAD;
1422: break;
1423: case ESCAPE_UNDEF:
1424: err = MANDOCERR_ESC_UNDEF;
1425: break;
1426: case ESCAPE_UNSUPP:
1427: err = MANDOCERR_ESC_UNSUPP;
1428: break;
1429: default:
1430: break;
1431: }
1432: if (err != MANDOCERR_OK)
1433: mandoc_msg(err, ln, (int)(stesc - buf->buf),
1.219 schwarze 1434: "%.*s", (int)(cp - stesc), stesc);
1.303 schwarze 1435: stesc--;
1.205 schwarze 1436: continue;
1.152 kristaps 1437: }
1438:
1.205 schwarze 1439: if (EXPAND_LIMIT < ++expand_count) {
1.350 schwarze 1440: mandoc_msg(MANDOCERR_ROFFLOOP,
1.238 schwarze 1441: ln, (int)(stesc - buf->buf), NULL);
1.277 schwarze 1442: return ROFF_IGN;
1.205 schwarze 1443: }
1.108 schwarze 1444:
1445: /*
1446: * The third character decides the length
1.181 schwarze 1447: * of the name of the string or register.
1.108 schwarze 1448: * Save a pointer to the name.
1449: */
1450:
1.238 schwarze 1451: if (term == '\0') {
1.206 schwarze 1452: switch (*++cp) {
1.207 schwarze 1453: case '\0':
1.206 schwarze 1454: maxl = 0;
1455: break;
1.207 schwarze 1456: case '(':
1.206 schwarze 1457: cp++;
1458: maxl = 2;
1459: break;
1.207 schwarze 1460: case '[':
1.206 schwarze 1461: cp++;
1462: term = ']';
1463: maxl = 0;
1464: break;
1465: default:
1466: maxl = 1;
1467: break;
1468: }
1469: } else {
1470: cp += 2;
1.94 kristaps 1471: maxl = 0;
1472: }
1.108 schwarze 1473: stnam = cp;
1.94 kristaps 1474:
1.108 schwarze 1475: /* Advance to the end of the name. */
1.94 kristaps 1476:
1.253 schwarze 1477: naml = 0;
1.218 schwarze 1478: arg_complete = 1;
1.253 schwarze 1479: while (maxl == 0 || naml < maxl) {
1.238 schwarze 1480: if (*cp == '\0') {
1.350 schwarze 1481: mandoc_msg(MANDOCERR_ESC_BAD, ln,
1482: (int)(stesc - buf->buf), "%s", stesc);
1.218 schwarze 1483: arg_complete = 0;
1.206 schwarze 1484: break;
1.153 kristaps 1485: }
1.238 schwarze 1486: if (maxl == 0 && *cp == term) {
1.206 schwarze 1487: cp++;
1.253 schwarze 1488: break;
1489: }
1.352 schwarze 1490: if (*cp++ != '\\' || *esct != 'w') {
1.253 schwarze 1491: naml++;
1492: continue;
1493: }
1494: switch (mandoc_escape(&cp, NULL, NULL)) {
1495: case ESCAPE_SPECIAL:
1496: case ESCAPE_UNICODE:
1497: case ESCAPE_NUMBERED:
1.352 schwarze 1498: case ESCAPE_UNDEF:
1.253 schwarze 1499: case ESCAPE_OVERSTRIKE:
1500: naml++;
1501: break;
1502: default:
1.94 kristaps 1503: break;
1.206 schwarze 1504: }
1.94 kristaps 1505: }
1506:
1.108 schwarze 1507: /*
1508: * Retrieve the replacement string; if it is
1509: * undefined, resume searching for escapes.
1510: */
1511:
1.352 schwarze 1512: switch (*esct) {
1.207 schwarze 1513: case '*':
1.315 schwarze 1514: if (arg_complete) {
1515: deftype = ROFFDEF_USER | ROFFDEF_PRE;
1516: res = roff_getstrn(r, stnam, naml, &deftype);
1.331 schwarze 1517:
1518: /*
1519: * If not overriden, let \*(.T
1520: * through to the formatters.
1521: */
1522:
1523: if (res == NULL && naml == 2 &&
1524: stnam[0] == '.' && stnam[1] == 'T') {
1525: roff_setstrn(&r->strtab,
1526: ".T", 2, NULL, 0, 0);
1527: stesc--;
1528: continue;
1529: }
1.315 schwarze 1530: }
1.206 schwarze 1531: break;
1.339 schwarze 1532: case '$':
1533: if (r->mstackpos < 0) {
1.350 schwarze 1534: mandoc_msg(MANDOCERR_ARG_UNDEF, ln,
1535: (int)(stesc - buf->buf), "%.3s", stesc);
1.339 schwarze 1536: break;
1537: }
1538: ctx = r->mstack + r->mstackpos;
1.352 schwarze 1539: npos = esct[1] - '1';
1.339 schwarze 1540: if (npos >= 0 && npos <= 8) {
1541: res = npos < ctx->argc ?
1542: ctx->argv[npos] : "";
1543: break;
1544: }
1.352 schwarze 1545: if (esct[1] == '*')
1.339 schwarze 1546: quote_args = 0;
1.352 schwarze 1547: else if (esct[1] == '@')
1.339 schwarze 1548: quote_args = 1;
1549: else {
1.350 schwarze 1550: mandoc_msg(MANDOCERR_ARG_NONUM, ln,
1551: (int)(stesc - buf->buf), "%.3s", stesc);
1.339 schwarze 1552: break;
1553: }
1554: asz = 0;
1555: for (npos = 0; npos < ctx->argc; npos++) {
1556: if (npos)
1557: asz++; /* blank */
1558: if (quote_args)
1559: asz += 2; /* quotes */
1560: asz += strlen(ctx->argv[npos]);
1561: }
1562: if (asz != 3) {
1563: rsz = buf->sz - (stesc - buf->buf) - 3;
1564: if (asz < 3)
1565: memmove(stesc + asz, stesc + 3, rsz);
1566: buf->sz += asz - 3;
1567: nbuf = mandoc_realloc(buf->buf, buf->sz);
1568: start = nbuf + pos;
1569: stesc = nbuf + (stesc - buf->buf);
1570: buf->buf = nbuf;
1571: if (asz > 3)
1572: memmove(stesc + asz, stesc + 3, rsz);
1573: }
1574: for (npos = 0; npos < ctx->argc; npos++) {
1575: if (npos)
1576: *stesc++ = ' ';
1577: if (quote_args)
1578: *stesc++ = '"';
1579: cp = ctx->argv[npos];
1580: while (*cp != '\0')
1581: *stesc++ = *cp++;
1582: if (quote_args)
1583: *stesc++ = '"';
1584: }
1585: continue;
1.207 schwarze 1586: case 'B':
1.206 schwarze 1587: npos = 0;
1.218 schwarze 1588: ubuf[0] = arg_complete &&
1.261 schwarze 1589: roff_evalnum(r, ln, stnam, &npos,
1590: NULL, ROFFNUM_SCALE) &&
1.218 schwarze 1591: stnam + npos + 1 == cp ? '1' : '0';
1.206 schwarze 1592: ubuf[1] = '\0';
1593: break;
1.207 schwarze 1594: case 'n':
1.218 schwarze 1595: if (arg_complete)
1596: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.327 schwarze 1597: roff_getregn(r, stnam, naml, sign));
1.218 schwarze 1598: else
1599: ubuf[0] = '\0';
1.206 schwarze 1600: break;
1.207 schwarze 1601: case 'w':
1.218 schwarze 1602: /* use even incomplete args */
1.208 schwarze 1603: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.206 schwarze 1604: 24 * (int)naml);
1605: break;
1606: }
1.94 kristaps 1607:
1.238 schwarze 1608: if (res == NULL) {
1.352 schwarze 1609: if (*esct == '*')
1.350 schwarze 1610: mandoc_msg(MANDOCERR_STR_UNDEF,
1611: ln, (int)(stesc - buf->buf),
1.339 schwarze 1612: "%.*s", (int)naml, stnam);
1.142 kristaps 1613: res = "";
1.246 schwarze 1614: } else if (buf->sz + strlen(res) > SHRT_MAX) {
1.350 schwarze 1615: mandoc_msg(MANDOCERR_ROFFLOOP,
1.246 schwarze 1616: ln, (int)(stesc - buf->buf), NULL);
1.277 schwarze 1617: return ROFF_IGN;
1.94 kristaps 1618: }
1619:
1.108 schwarze 1620: /* Replace the escape sequence by the string. */
1621:
1.209 schwarze 1622: *stesc = '\0';
1.238 schwarze 1623: buf->sz = mandoc_asprintf(&nbuf, "%s%s%s",
1624: buf->buf, res, cp) + 1;
1.94 kristaps 1625:
1.205 schwarze 1626: /* Prepare for the next replacement. */
1.94 kristaps 1627:
1.205 schwarze 1628: start = nbuf + pos;
1.238 schwarze 1629: stesc = nbuf + (stesc - buf->buf) + strlen(res);
1630: free(buf->buf);
1631: buf->buf = nbuf;
1.154 kristaps 1632: }
1.277 schwarze 1633: return ROFF_CONT;
1.154 kristaps 1634: }
1.353 schwarze 1635:
1636: /*
1637: * Parse a quoted or unquoted roff-style request or macro argument.
1638: * Return a pointer to the parsed argument, which is either the original
1639: * pointer or advanced by one byte in case the argument is quoted.
1640: * NUL-terminate the argument in place.
1641: * Collapse pairs of quotes inside quoted arguments.
1642: * Advance the argument pointer to the next argument,
1643: * or to the NUL byte terminating the argument line.
1644: */
1645: char *
1.355 schwarze 1646: roff_getarg(struct roff *r, char **cpp, int ln, int *pos)
1.353 schwarze 1647: {
1.355 schwarze 1648: struct buf buf;
1.366 schwarze 1649: char *cp, *start;
1.355 schwarze 1650: int newesc, pairs, quoted, white;
1.353 schwarze 1651:
1652: /* Quoting can only start with a new word. */
1653: start = *cpp;
1654: quoted = 0;
1655: if ('"' == *start) {
1656: quoted = 1;
1657: start++;
1658: }
1659:
1.355 schwarze 1660: newesc = pairs = white = 0;
1.353 schwarze 1661: for (cp = start; '\0' != *cp; cp++) {
1662:
1663: /*
1664: * Move the following text left
1665: * after quoted quotes and after "\\" and "\t".
1666: */
1667: if (pairs)
1668: cp[-pairs] = cp[0];
1669:
1670: if ('\\' == cp[0]) {
1671: /*
1672: * In copy mode, translate double to single
1673: * backslashes and backslash-t to literal tabs.
1674: */
1675: switch (cp[1]) {
1676: case 'a':
1677: case 't':
1.354 schwarze 1678: cp[-pairs] = '\t';
1.355 schwarze 1679: pairs++;
1680: cp++;
1681: break;
1.353 schwarze 1682: case '\\':
1.355 schwarze 1683: newesc = 1;
1684: cp[-pairs] = ASCII_ESC;
1.353 schwarze 1685: pairs++;
1686: cp++;
1687: break;
1688: case ' ':
1689: /* Skip escaped blanks. */
1690: if (0 == quoted)
1691: cp++;
1692: break;
1693: default:
1694: break;
1695: }
1696: } else if (0 == quoted) {
1697: if (' ' == cp[0]) {
1698: /* Unescaped blanks end unquoted args. */
1699: white = 1;
1700: break;
1701: }
1702: } else if ('"' == cp[0]) {
1703: if ('"' == cp[1]) {
1704: /* Quoted quotes collapse. */
1705: pairs++;
1706: cp++;
1707: } else {
1708: /* Unquoted quotes end quoted args. */
1709: quoted = 2;
1710: break;
1711: }
1712: }
1713: }
1714:
1715: /* Quoted argument without a closing quote. */
1716: if (1 == quoted)
1717: mandoc_msg(MANDOCERR_ARG_QUOTE, ln, *pos, NULL);
1718:
1719: /* NUL-terminate this argument and move to the next one. */
1720: if (pairs)
1721: cp[-pairs] = '\0';
1722: if ('\0' != *cp) {
1723: *cp++ = '\0';
1724: while (' ' == *cp)
1725: cp++;
1726: }
1727: *pos += (int)(cp - start) + (quoted ? 1 : 0);
1728: *cpp = cp;
1729:
1730: if ('\0' == *cp && (white || ' ' == cp[-1]))
1731: mandoc_msg(MANDOCERR_SPACE_EOL, ln, *pos, NULL);
1732:
1.355 schwarze 1733: start = mandoc_strdup(start);
1734: if (newesc == 0)
1735: return start;
1736:
1737: buf.buf = start;
1738: buf.sz = strlen(start) + 1;
1739: buf.next = NULL;
1740: if (roff_expand(r, &buf, ln, 0, ASCII_ESC) & ROFF_IGN) {
1741: free(buf.buf);
1742: buf.buf = mandoc_strdup("");
1743: }
1744: return buf.buf;
1.353 schwarze 1745: }
1746:
1.154 kristaps 1747:
1748: /*
1.275 schwarze 1749: * Process text streams.
1.154 kristaps 1750: */
1.340 schwarze 1751: static int
1.305 schwarze 1752: roff_parsetext(struct roff *r, struct buf *buf, int pos, int *offs)
1.154 kristaps 1753: {
1754: size_t sz;
1755: const char *start;
1.178 schwarze 1756: char *p;
1757: int isz;
1.154 kristaps 1758: enum mandoc_esc esc;
1759:
1.275 schwarze 1760: /* Spring the input line trap. */
1761:
1762: if (roffit_lines == 1) {
1763: isz = mandoc_asprintf(&p, "%s\n.%s", buf->buf, roffit_macro);
1764: free(buf->buf);
1765: buf->buf = p;
1766: buf->sz = isz + 1;
1767: *offs = 0;
1768: free(roffit_macro);
1769: roffit_lines = 0;
1.277 schwarze 1770: return ROFF_REPARSE;
1.275 schwarze 1771: } else if (roffit_lines > 1)
1772: --roffit_lines;
1773:
1.305 schwarze 1774: if (roffce_node != NULL && buf->buf[pos] != '\0') {
1775: if (roffce_lines < 1) {
1776: r->man->last = roffce_node;
1777: r->man->next = ROFF_NEXT_SIBLING;
1778: roffce_lines = 0;
1779: roffce_node = NULL;
1780: } else
1781: roffce_lines--;
1782: }
1783:
1.275 schwarze 1784: /* Convert all breakable hyphens into ASCII_HYPH. */
1785:
1.238 schwarze 1786: start = p = buf->buf + pos;
1.154 kristaps 1787:
1.238 schwarze 1788: while (*p != '\0') {
1.154 kristaps 1789: sz = strcspn(p, "-\\");
1790: p += sz;
1791:
1.238 schwarze 1792: if (*p == '\0')
1.159 kristaps 1793: break;
1794:
1.238 schwarze 1795: if (*p == '\\') {
1.154 kristaps 1796: /* Skip over escapes. */
1797: p++;
1.189 schwarze 1798: esc = mandoc_escape((const char **)&p, NULL, NULL);
1.238 schwarze 1799: if (esc == ESCAPE_ERROR)
1.154 kristaps 1800: break;
1.264 schwarze 1801: while (*p == '-')
1802: p++;
1.155 kristaps 1803: continue;
1.159 kristaps 1804: } else if (p == start) {
1.158 kristaps 1805: p++;
1.155 kristaps 1806: continue;
1.158 kristaps 1807: }
1.155 kristaps 1808:
1.171 schwarze 1809: if (isalpha((unsigned char)p[-1]) &&
1810: isalpha((unsigned char)p[1]))
1.155 kristaps 1811: *p = ASCII_HYPH;
1812: p++;
1.94 kristaps 1813: }
1.277 schwarze 1814: return ROFF_CONT;
1.94 kristaps 1815: }
1816:
1.340 schwarze 1817: int
1.238 schwarze 1818: roff_parseln(struct roff *r, int ln, struct buf *buf, int *offs)
1.67 kristaps 1819: {
1.294 schwarze 1820: enum roff_tok t;
1.340 schwarze 1821: int e;
1.238 schwarze 1822: int pos; /* parse point */
1.243 schwarze 1823: int spos; /* saved parse point for messages */
1.238 schwarze 1824: int ppos; /* original offset in buf->buf */
1825: int ctl; /* macro line (boolean) */
1826:
1827: ppos = pos = *offs;
1.79 kristaps 1828:
1.230 schwarze 1829: /* Handle in-line equation delimiters. */
1830:
1.236 schwarze 1831: if (r->tbl == NULL &&
1832: r->last_eqn != NULL && r->last_eqn->delim &&
1.230 schwarze 1833: (r->eqn == NULL || r->eqn_inline)) {
1.238 schwarze 1834: e = roff_eqndelim(r, buf, pos);
1.230 schwarze 1835: if (e == ROFF_REPARSE)
1.277 schwarze 1836: return e;
1.230 schwarze 1837: assert(e == ROFF_CONT);
1838: }
1839:
1840: /* Expand some escape sequences. */
1.94 kristaps 1841:
1.355 schwarze 1842: e = roff_expand(r, buf, ln, pos, r->escape);
1.340 schwarze 1843: if ((e & ROFF_MASK) == ROFF_IGN)
1.277 schwarze 1844: return e;
1.238 schwarze 1845: assert(e == ROFF_CONT);
1.94 kristaps 1846:
1.238 schwarze 1847: ctl = roff_getcontrol(r, buf->buf, &pos);
1.130 kristaps 1848:
1.94 kristaps 1849: /*
1.79 kristaps 1850: * First, if a scope is open and we're not a macro, pass the
1.252 schwarze 1851: * text through the macro's filter.
1852: * Equations process all content themselves.
1853: * Tables process almost all content themselves, but we want
1854: * to warn about macros before passing it there.
1.79 kristaps 1855: */
1.74 kristaps 1856:
1.252 schwarze 1857: if (r->last != NULL && ! ctl) {
1.78 kristaps 1858: t = r->last->tok;
1.238 schwarze 1859: e = (*roffs[t].text)(r, t, buf, ln, pos, pos, offs);
1.340 schwarze 1860: if ((e & ROFF_MASK) == ROFF_IGN)
1.277 schwarze 1861: return e;
1.340 schwarze 1862: e &= ~ROFF_MASK;
1863: } else
1864: e = ROFF_IGN;
1.319 schwarze 1865: if (r->eqn != NULL && strncmp(buf->buf + ppos, ".EN", 3)) {
1866: eqn_read(r->eqn, buf->buf + ppos);
1.340 schwarze 1867: return e;
1.319 schwarze 1868: }
1.321 schwarze 1869: if (r->tbl != NULL && (ctl == 0 || buf->buf[pos] == '\0')) {
1870: tbl_read(r->tbl, ln, buf->buf, ppos);
1.346 schwarze 1871: roff_addtbl(r->man, ln, r->tbl);
1.340 schwarze 1872: return e;
1.321 schwarze 1873: }
1.367 schwarze 1874: if ( ! ctl) {
1875: r->options &= ~MPARSE_COMMENT;
1.340 schwarze 1876: return roff_parsetext(r, buf, pos, offs) | e;
1.367 schwarze 1877: }
1.228 schwarze 1878:
1879: /* Skip empty request lines. */
1880:
1.238 schwarze 1881: if (buf->buf[pos] == '"') {
1.350 schwarze 1882: mandoc_msg(MANDOCERR_COMMENT_BAD, ln, pos, NULL);
1.277 schwarze 1883: return ROFF_IGN;
1.238 schwarze 1884: } else if (buf->buf[pos] == '\0')
1.277 schwarze 1885: return ROFF_IGN;
1.67 kristaps 1886:
1.79 kristaps 1887: /*
1888: * If a scope is open, go to the child handler for that macro,
1889: * as it may want to preprocess before doing anything with it.
1.125 kristaps 1890: * Don't do so if an equation is open.
1.79 kristaps 1891: */
1.78 kristaps 1892:
1.79 kristaps 1893: if (r->last) {
1894: t = r->last->tok;
1.277 schwarze 1895: return (*roffs[t].sub)(r, t, buf, ln, ppos, pos, offs);
1.79 kristaps 1896: }
1.78 kristaps 1897:
1.243 schwarze 1898: /* No scope is open. This is a new request or macro. */
1899:
1.367 schwarze 1900: r->options &= ~MPARSE_COMMENT;
1.243 schwarze 1901: spos = pos;
1902: t = roff_parse(r, buf->buf, &pos, ln, ppos);
1903:
1904: /* Tables ignore most macros. */
1905:
1.308 schwarze 1906: if (r->tbl != NULL && (t == TOKEN_NONE || t == ROFF_TS ||
1.309 schwarze 1907: t == ROFF_br || t == ROFF_ce || t == ROFF_rj || t == ROFF_sp)) {
1.350 schwarze 1908: mandoc_msg(MANDOCERR_TBLMACRO,
1909: ln, pos, "%s", buf->buf + spos);
1.308 schwarze 1910: if (t != TOKEN_NONE)
1.277 schwarze 1911: return ROFF_IGN;
1.257 schwarze 1912: while (buf->buf[pos] != '\0' && buf->buf[pos] != ' ')
1913: pos++;
1.291 schwarze 1914: while (buf->buf[pos] == ' ')
1.257 schwarze 1915: pos++;
1.321 schwarze 1916: tbl_read(r->tbl, ln, buf->buf, pos);
1.346 schwarze 1917: roff_addtbl(r->man, ln, r->tbl);
1.321 schwarze 1918: return ROFF_IGN;
1.243 schwarze 1919: }
1920:
1.305 schwarze 1921: /* For now, let high level macros abort .ce mode. */
1922:
1923: if (ctl && roffce_node != NULL &&
1.324 schwarze 1924: (t == TOKEN_NONE || t == ROFF_Dd || t == ROFF_EQ ||
1925: t == ROFF_TH || t == ROFF_TS)) {
1.305 schwarze 1926: r->man->last = roffce_node;
1927: r->man->next = ROFF_NEXT_SIBLING;
1928: roffce_lines = 0;
1929: roffce_node = NULL;
1930: }
1931:
1.79 kristaps 1932: /*
1.243 schwarze 1933: * This is neither a roff request nor a user-defined macro.
1934: * Let the standard macro set parsers handle it.
1.79 kristaps 1935: */
1.67 kristaps 1936:
1.294 schwarze 1937: if (t == TOKEN_NONE)
1.277 schwarze 1938: return ROFF_CONT;
1.243 schwarze 1939:
1940: /* Execute a roff request or a user defined macro. */
1.67 kristaps 1941:
1.296 schwarze 1942: return (*roffs[t].proc)(r, t, buf, ln, spos, pos, offs);
1.74 kristaps 1943: }
1944:
1.339 schwarze 1945: /*
1946: * Internal interface function to tell the roff parser that execution
1947: * of the current macro ended. This is required because macro
1948: * definitions usually do not end with a .return request.
1949: */
1950: void
1951: roff_userret(struct roff *r)
1952: {
1953: struct mctx *ctx;
1954: int i;
1955:
1956: assert(r->mstackpos >= 0);
1957: ctx = r->mstack + r->mstackpos;
1958: for (i = 0; i < ctx->argc; i++)
1959: free(ctx->argv[i]);
1960: ctx->argc = 0;
1961: r->mstackpos--;
1962: }
1963:
1.117 kristaps 1964: void
1.74 kristaps 1965: roff_endparse(struct roff *r)
1966: {
1.321 schwarze 1967: if (r->last != NULL)
1.350 schwarze 1968: mandoc_msg(MANDOCERR_BLK_NOEND, r->last->line,
1969: r->last->col, "%s", roff_name[r->last->tok]);
1.117 kristaps 1970:
1.321 schwarze 1971: if (r->eqn != NULL) {
1.350 schwarze 1972: mandoc_msg(MANDOCERR_BLK_NOEND,
1.319 schwarze 1973: r->eqn->node->line, r->eqn->node->pos, "EQ");
1974: eqn_parse(r->eqn);
1975: r->eqn = NULL;
1.125 kristaps 1976: }
1977:
1.321 schwarze 1978: if (r->tbl != NULL) {
1.346 schwarze 1979: tbl_end(r->tbl, 1);
1.321 schwarze 1980: r->tbl = NULL;
1.117 kristaps 1981: }
1.67 kristaps 1982: }
1983:
1984: /*
1985: * Parse a roff node's type from the input buffer. This must be in the
1986: * form of ".foo xxx" in the usual way.
1987: */
1.294 schwarze 1988: static enum roff_tok
1.214 schwarze 1989: roff_parse(struct roff *r, char *buf, int *pos, int ln, int ppos)
1.67 kristaps 1990: {
1.214 schwarze 1991: char *cp;
1.106 kristaps 1992: const char *mac;
1993: size_t maclen;
1.315 schwarze 1994: int deftype;
1.294 schwarze 1995: enum roff_tok t;
1.67 kristaps 1996:
1.214 schwarze 1997: cp = buf + *pos;
1998:
1999: if ('\0' == *cp || '"' == *cp || '\t' == *cp || ' ' == *cp)
1.294 schwarze 2000: return TOKEN_NONE;
1.67 kristaps 2001:
1.214 schwarze 2002: mac = cp;
2003: maclen = roff_getname(r, &cp, ln, ppos);
1.67 kristaps 2004:
1.315 schwarze 2005: deftype = ROFFDEF_USER | ROFFDEF_REN;
2006: r->current_string = roff_getstrn(r, mac, maclen, &deftype);
2007: switch (deftype) {
2008: case ROFFDEF_USER:
2009: t = ROFF_USERDEF;
2010: break;
2011: case ROFFDEF_REN:
2012: t = ROFF_RENAMED;
2013: break;
2014: default:
2015: t = roffhash_find(r->reqtab, mac, maclen);
2016: break;
2017: }
1.294 schwarze 2018: if (t != TOKEN_NONE)
1.214 schwarze 2019: *pos = cp - buf;
1.325 schwarze 2020: else if (deftype == ROFFDEF_UNDEF) {
2021: /* Using an undefined macro defines it to be empty. */
2022: roff_setstrn(&r->strtab, mac, maclen, "", 0, 0);
2023: roff_setstrn(&r->rentab, mac, maclen, NULL, 0, 0);
2024: }
1.277 schwarze 2025: return t;
1.67 kristaps 2026: }
2027:
1.266 schwarze 2028: /* --- handling of request blocks ----------------------------------------- */
2029:
1.340 schwarze 2030: static int
1.76 kristaps 2031: roff_cblock(ROFF_ARGS)
1.67 kristaps 2032: {
2033:
1.79 kristaps 2034: /*
2035: * A block-close `..' should only be invoked as a child of an
2036: * ignore macro, otherwise raise a warning and just ignore it.
2037: */
2038:
1.238 schwarze 2039: if (r->last == NULL) {
1.350 schwarze 2040: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "..");
1.277 schwarze 2041: return ROFF_IGN;
1.76 kristaps 2042: }
1.67 kristaps 2043:
1.81 kristaps 2044: switch (r->last->tok) {
1.207 schwarze 2045: case ROFF_am:
1.220 schwarze 2046: /* ROFF_am1 is remapped to ROFF_am in roff_block(). */
1.207 schwarze 2047: case ROFF_ami:
2048: case ROFF_de:
1.108 schwarze 2049: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.207 schwarze 2050: case ROFF_dei:
2051: case ROFF_ig:
1.81 kristaps 2052: break;
2053: default:
1.350 schwarze 2054: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "..");
1.277 schwarze 2055: return ROFF_IGN;
1.76 kristaps 2056: }
1.67 kristaps 2057:
1.238 schwarze 2058: if (buf->buf[pos] != '\0')
1.350 schwarze 2059: mandoc_msg(MANDOCERR_ARG_SKIP, ln, pos,
1.238 schwarze 2060: ".. %s", buf->buf + pos);
1.71 kristaps 2061:
2062: roffnode_pop(r);
1.76 kristaps 2063: roffnode_cleanscope(r);
1.277 schwarze 2064: return ROFF_IGN;
1.71 kristaps 2065:
1.67 kristaps 2066: }
2067:
1.364 schwarze 2068: /*
2069: * Pop all nodes ending at the end of the current input line.
2070: * Return the number of loops ended.
2071: */
1.340 schwarze 2072: static int
1.76 kristaps 2073: roffnode_cleanscope(struct roff *r)
1.67 kristaps 2074: {
1.340 schwarze 2075: int inloop;
1.67 kristaps 2076:
1.340 schwarze 2077: inloop = 0;
2078: while (r->last != NULL) {
1.173 schwarze 2079: if (--r->last->endspan != 0)
1.76 kristaps 2080: break;
1.340 schwarze 2081: inloop += roffnode_pop(r);
1.76 kristaps 2082: }
1.340 schwarze 2083: return inloop;
1.67 kristaps 2084: }
2085:
1.364 schwarze 2086: /*
2087: * Handle the closing \} of a conditional block.
2088: * Apart from generating warnings, this only pops nodes.
2089: * Return the number of loops ended.
2090: */
1.340 schwarze 2091: static int
1.195 schwarze 2092: roff_ccond(struct roff *r, int ln, int ppos)
1.74 kristaps 2093: {
1.76 kristaps 2094: if (NULL == r->last) {
1.350 schwarze 2095: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "\\}");
1.340 schwarze 2096: return 0;
1.76 kristaps 2097: }
2098:
1.82 kristaps 2099: switch (r->last->tok) {
1.207 schwarze 2100: case ROFF_el:
2101: case ROFF_ie:
2102: case ROFF_if:
1.340 schwarze 2103: case ROFF_while:
1.82 kristaps 2104: break;
2105: default:
1.350 schwarze 2106: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "\\}");
1.340 schwarze 2107: return 0;
1.75 kristaps 2108: }
2109:
1.76 kristaps 2110: if (r->last->endspan > -1) {
1.350 schwarze 2111: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "\\}");
1.340 schwarze 2112: return 0;
1.76 kristaps 2113: }
2114:
1.340 schwarze 2115: return roffnode_pop(r) + roffnode_cleanscope(r);
1.76 kristaps 2116: }
2117:
1.340 schwarze 2118: static int
1.80 kristaps 2119: roff_block(ROFF_ARGS)
1.76 kristaps 2120: {
1.315 schwarze 2121: const char *name, *value;
2122: char *call, *cp, *iname, *rname;
2123: size_t csz, namesz, rsz;
2124: int deftype;
1.106 kristaps 2125:
1.220 schwarze 2126: /* Ignore groff compatibility mode for now. */
1.76 kristaps 2127:
1.238 schwarze 2128: if (tok == ROFF_de1)
1.220 schwarze 2129: tok = ROFF_de;
1.251 schwarze 2130: else if (tok == ROFF_dei1)
2131: tok = ROFF_dei;
1.238 schwarze 2132: else if (tok == ROFF_am1)
1.220 schwarze 2133: tok = ROFF_am;
1.251 schwarze 2134: else if (tok == ROFF_ami1)
2135: tok = ROFF_ami;
1.220 schwarze 2136:
2137: /* Parse the macro name argument. */
2138:
1.238 schwarze 2139: cp = buf->buf + pos;
2140: if (tok == ROFF_ig) {
1.220 schwarze 2141: iname = NULL;
2142: namesz = 0;
2143: } else {
2144: iname = cp;
2145: namesz = roff_getname(r, &cp, ln, ppos);
2146: iname[namesz] = '\0';
2147: }
1.107 kristaps 2148:
1.220 schwarze 2149: /* Resolve the macro name argument if it is indirect. */
1.107 kristaps 2150:
1.238 schwarze 2151: if (namesz && (tok == ROFF_dei || tok == ROFF_ami)) {
1.315 schwarze 2152: deftype = ROFFDEF_USER;
2153: name = roff_getstrn(r, iname, namesz, &deftype);
2154: if (name == NULL) {
1.350 schwarze 2155: mandoc_msg(MANDOCERR_STR_UNDEF,
2156: ln, (int)(iname - buf->buf),
1.220 schwarze 2157: "%.*s", (int)namesz, iname);
2158: namesz = 0;
2159: } else
2160: namesz = strlen(name);
2161: } else
2162: name = iname;
1.107 kristaps 2163:
1.238 schwarze 2164: if (namesz == 0 && tok != ROFF_ig) {
1.350 schwarze 2165: mandoc_msg(MANDOCERR_REQ_EMPTY,
2166: ln, ppos, "%s", roff_name[tok]);
1.277 schwarze 2167: return ROFF_IGN;
1.220 schwarze 2168: }
1.80 kristaps 2169:
1.106 kristaps 2170: roffnode_push(r, tok, name, ln, ppos);
2171:
2172: /*
2173: * At the beginning of a `de' macro, clear the existing string
2174: * with the same name, if there is one. New content will be
1.193 schwarze 2175: * appended from roff_block_text() in multiline mode.
1.106 kristaps 2176: */
1.107 kristaps 2177:
1.311 schwarze 2178: if (tok == ROFF_de || tok == ROFF_dei) {
1.213 schwarze 2179: roff_setstrn(&r->strtab, name, namesz, "", 0, 0);
1.311 schwarze 2180: roff_setstrn(&r->rentab, name, namesz, NULL, 0, 0);
1.315 schwarze 2181: } else if (tok == ROFF_am || tok == ROFF_ami) {
2182: deftype = ROFFDEF_ANY;
2183: value = roff_getstrn(r, iname, namesz, &deftype);
2184: switch (deftype) { /* Before appending, ... */
2185: case ROFFDEF_PRE: /* copy predefined to user-defined. */
2186: roff_setstrn(&r->strtab, name, namesz,
2187: value, strlen(value), 0);
2188: break;
2189: case ROFFDEF_REN: /* call original standard macro. */
2190: csz = mandoc_asprintf(&call, ".%.*s \\$* \\\"\n",
2191: (int)strlen(value), value);
2192: roff_setstrn(&r->strtab, name, namesz, call, csz, 0);
2193: roff_setstrn(&r->rentab, name, namesz, NULL, 0, 0);
2194: free(call);
2195: break;
2196: case ROFFDEF_STD: /* rename and call standard macro. */
2197: rsz = mandoc_asprintf(&rname, "__%s_renamed", name);
2198: roff_setstrn(&r->rentab, rname, rsz, name, namesz, 0);
2199: csz = mandoc_asprintf(&call, ".%.*s \\$* \\\"\n",
2200: (int)rsz, rname);
2201: roff_setstrn(&r->strtab, name, namesz, call, csz, 0);
2202: free(call);
2203: free(rname);
2204: break;
2205: default:
2206: break;
2207: }
1.311 schwarze 2208: }
1.76 kristaps 2209:
1.238 schwarze 2210: if (*cp == '\0')
1.277 schwarze 2211: return ROFF_IGN;
1.78 kristaps 2212:
1.220 schwarze 2213: /* Get the custom end marker. */
1.107 kristaps 2214:
1.220 schwarze 2215: iname = cp;
1.213 schwarze 2216: namesz = roff_getname(r, &cp, ln, ppos);
1.220 schwarze 2217:
2218: /* Resolve the end marker if it is indirect. */
2219:
1.238 schwarze 2220: if (namesz && (tok == ROFF_dei || tok == ROFF_ami)) {
1.315 schwarze 2221: deftype = ROFFDEF_USER;
2222: name = roff_getstrn(r, iname, namesz, &deftype);
2223: if (name == NULL) {
1.350 schwarze 2224: mandoc_msg(MANDOCERR_STR_UNDEF,
2225: ln, (int)(iname - buf->buf),
1.220 schwarze 2226: "%.*s", (int)namesz, iname);
2227: namesz = 0;
2228: } else
2229: namesz = strlen(name);
2230: } else
2231: name = iname;
2232:
1.213 schwarze 2233: if (namesz)
2234: r->last->end = mandoc_strndup(name, namesz);
1.78 kristaps 2235:
1.238 schwarze 2236: if (*cp != '\0')
1.350 schwarze 2237: mandoc_msg(MANDOCERR_ARG_EXCESS,
1.295 schwarze 2238: ln, pos, ".%s ... %s", roff_name[tok], cp);
1.74 kristaps 2239:
1.277 schwarze 2240: return ROFF_IGN;
1.78 kristaps 2241: }
2242:
1.340 schwarze 2243: static int
1.80 kristaps 2244: roff_block_sub(ROFF_ARGS)
1.79 kristaps 2245: {
1.294 schwarze 2246: enum roff_tok t;
1.79 kristaps 2247: int i, j;
2248:
2249: /*
2250: * First check whether a custom macro exists at this level. If
2251: * it does, then check against it. This is some of groff's
2252: * stranger behaviours. If we encountered a custom end-scope
2253: * tag and that tag also happens to be a "real" macro, then we
2254: * need to try interpreting it again as a real macro. If it's
2255: * not, then return ignore. Else continue.
2256: */
2257:
2258: if (r->last->end) {
1.130 kristaps 2259: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.238 schwarze 2260: if (buf->buf[i] != r->last->end[j])
1.79 kristaps 2261: break;
2262:
1.238 schwarze 2263: if (r->last->end[j] == '\0' &&
2264: (buf->buf[i] == '\0' ||
2265: buf->buf[i] == ' ' ||
2266: buf->buf[i] == '\t')) {
1.79 kristaps 2267: roffnode_pop(r);
2268: roffnode_cleanscope(r);
2269:
1.238 schwarze 2270: while (buf->buf[i] == ' ' || buf->buf[i] == '\t')
1.130 kristaps 2271: i++;
2272:
2273: pos = i;
1.238 schwarze 2274: if (roff_parse(r, buf->buf, &pos, ln, ppos) !=
1.294 schwarze 2275: TOKEN_NONE)
1.277 schwarze 2276: return ROFF_RERUN;
2277: return ROFF_IGN;
1.79 kristaps 2278: }
2279: }
2280:
2281: /*
2282: * If we have no custom end-query or lookup failed, then try
2283: * pulling it out of the hashtable.
2284: */
2285:
1.238 schwarze 2286: t = roff_parse(r, buf->buf, &pos, ln, ppos);
1.79 kristaps 2287:
1.238 schwarze 2288: if (t != ROFF_cblock) {
2289: if (tok != ROFF_ig)
2290: roff_setstr(r, r->last->name, buf->buf + ppos, 2);
1.277 schwarze 2291: return ROFF_IGN;
1.106 kristaps 2292: }
1.79 kristaps 2293:
1.277 schwarze 2294: return (*roffs[t].proc)(r, t, buf, ln, ppos, pos, offs);
1.79 kristaps 2295: }
2296:
1.340 schwarze 2297: static int
1.80 kristaps 2298: roff_block_text(ROFF_ARGS)
1.78 kristaps 2299: {
2300:
1.238 schwarze 2301: if (tok != ROFF_ig)
2302: roff_setstr(r, r->last->name, buf->buf + pos, 2);
1.106 kristaps 2303:
1.277 schwarze 2304: return ROFF_IGN;
1.78 kristaps 2305: }
2306:
1.340 schwarze 2307: static int
1.82 kristaps 2308: roff_cond_sub(ROFF_ARGS)
2309: {
1.364 schwarze 2310: struct roffnode *bl;
1.340 schwarze 2311: char *ep;
2312: int endloop, irc, rr;
1.294 schwarze 2313: enum roff_tok t;
1.82 kristaps 2314:
1.340 schwarze 2315: irc = ROFF_IGN;
1.82 kristaps 2316: rr = r->last->rule;
1.340 schwarze 2317: endloop = tok != ROFF_while ? ROFF_IGN :
2318: rr ? ROFF_LOOPCONT : ROFF_LOOPEXIT;
2319: if (roffnode_cleanscope(r))
2320: irc |= endloop;
1.144 kristaps 2321:
1.196 schwarze 2322: /*
2323: * If `\}' occurs on a macro line without a preceding macro,
2324: * drop the line completely.
2325: */
2326:
1.238 schwarze 2327: ep = buf->buf + pos;
2328: if (ep[0] == '\\' && ep[1] == '}')
1.198 schwarze 2329: rr = 0;
1.196 schwarze 2330:
1.343 schwarze 2331: /*
2332: * The closing delimiter `\}' rewinds the conditional scope
2333: * but is otherwise ignored when interpreting the line.
2334: */
1.144 kristaps 2335:
1.238 schwarze 2336: while ((ep = strchr(ep, '\\')) != NULL) {
1.318 schwarze 2337: switch (ep[1]) {
2338: case '}':
2339: memmove(ep, ep + 2, strlen(ep + 2) + 1);
1.340 schwarze 2340: if (roff_ccond(r, ln, ep - buf->buf))
2341: irc |= endloop;
1.318 schwarze 2342: break;
2343: case '\0':
2344: ++ep;
2345: break;
2346: default:
2347: ep += 2;
2348: break;
1.197 schwarze 2349: }
1.177 schwarze 2350: }
1.318 schwarze 2351:
1.368 schwarze 2352: t = roff_parse(r, buf->buf, &pos, ln, ppos);
2353:
2354: /* For now, let high level macros abort .ce mode. */
2355:
2356: if (roffce_node != NULL &&
2357: (t == TOKEN_NONE || t == ROFF_Dd || t == ROFF_EQ ||
2358: t == ROFF_TH || t == ROFF_TS)) {
2359: r->man->last = roffce_node;
2360: r->man->next = ROFF_NEXT_SIBLING;
2361: roffce_lines = 0;
2362: roffce_node = NULL;
2363: }
2364:
1.318 schwarze 2365: /*
2366: * Fully handle known macros when they are structurally
2367: * required or when the conditional evaluated to true.
2368: */
2369:
1.364 schwarze 2370: if (t == ROFF_break) {
2371: if (irc & ROFF_LOOPMASK)
2372: irc = ROFF_IGN | ROFF_LOOPEXIT;
2373: else if (rr) {
2374: for (bl = r->last; bl != NULL; bl = bl->parent) {
2375: bl->rule = 0;
2376: if (bl->tok == ROFF_while)
2377: break;
2378: }
2379: }
2380: } else if (t != TOKEN_NONE &&
2381: (rr || roffs[t].flags & ROFFMAC_STRUCT))
2382: irc |= (*roffs[t].proc)(r, t, buf, ln, ppos, pos, offs);
2383: else
2384: irc |= rr ? ROFF_CONT : ROFF_IGN;
1.340 schwarze 2385: return irc;
1.82 kristaps 2386: }
2387:
1.340 schwarze 2388: static int
1.82 kristaps 2389: roff_cond_text(ROFF_ARGS)
1.78 kristaps 2390: {
1.140 kristaps 2391: char *ep;
1.340 schwarze 2392: int endloop, irc, rr;
1.82 kristaps 2393:
1.340 schwarze 2394: irc = ROFF_IGN;
1.82 kristaps 2395: rr = r->last->rule;
1.340 schwarze 2396: endloop = tok != ROFF_while ? ROFF_IGN :
2397: rr ? ROFF_LOOPCONT : ROFF_LOOPEXIT;
2398: if (roffnode_cleanscope(r))
2399: irc |= endloop;
1.82 kristaps 2400:
1.343 schwarze 2401: /*
2402: * If `\}' occurs on a text line with neither preceding
2403: * nor following characters, drop the line completely.
2404: */
2405:
1.238 schwarze 2406: ep = buf->buf + pos;
1.343 schwarze 2407: if (strcmp(ep, "\\}") == 0)
2408: rr = 0;
2409:
2410: /*
2411: * The closing delimiter `\}' rewinds the conditional scope
2412: * but is otherwise ignored when interpreting the line.
2413: */
2414:
1.238 schwarze 2415: while ((ep = strchr(ep, '\\')) != NULL) {
1.343 schwarze 2416: switch (ep[1]) {
2417: case '}':
2418: memmove(ep, ep + 2, strlen(ep + 2) + 1);
2419: if (roff_ccond(r, ln, ep - buf->buf))
1.340 schwarze 2420: irc |= endloop;
1.343 schwarze 2421: break;
2422: case '\0':
2423: ++ep;
2424: break;
2425: default:
2426: ep += 2;
2427: break;
1.197 schwarze 2428: }
1.78 kristaps 2429: }
1.340 schwarze 2430: if (rr)
2431: irc |= ROFF_CONT;
2432: return irc;
1.74 kristaps 2433: }
2434:
1.266 schwarze 2435: /* --- handling of numeric and conditional expressions -------------------- */
2436:
1.204 schwarze 2437: /*
2438: * Parse a single signed integer number. Stop at the first non-digit.
2439: * If there is at least one digit, return success and advance the
2440: * parse point, else return failure and let the parse point unchanged.
2441: * Ignore overflows, treat them just like the C language.
2442: */
1.184 schwarze 2443: static int
1.261 schwarze 2444: roff_getnum(const char *v, int *pos, int *res, int flags)
1.184 schwarze 2445: {
1.261 schwarze 2446: int myres, scaled, n, p;
1.206 schwarze 2447:
2448: if (NULL == res)
2449: res = &myres;
1.184 schwarze 2450:
2451: p = *pos;
2452: n = v[p] == '-';
1.261 schwarze 2453: if (n || v[p] == '+')
1.184 schwarze 2454: p++;
2455:
1.261 schwarze 2456: if (flags & ROFFNUM_WHITE)
2457: while (isspace((unsigned char)v[p]))
2458: p++;
2459:
1.184 schwarze 2460: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1.204 schwarze 2461: *res = 10 * *res + v[p] - '0';
1.184 schwarze 2462: if (p == *pos + n)
2463: return 0;
2464:
2465: if (n)
2466: *res = -*res;
2467:
1.254 schwarze 2468: /* Each number may be followed by one optional scaling unit. */
2469:
2470: switch (v[p]) {
2471: case 'f':
1.261 schwarze 2472: scaled = *res * 65536;
1.254 schwarze 2473: break;
2474: case 'i':
1.261 schwarze 2475: scaled = *res * 240;
1.254 schwarze 2476: break;
2477: case 'c':
1.261 schwarze 2478: scaled = *res * 240 / 2.54;
1.254 schwarze 2479: break;
2480: case 'v':
2481: case 'P':
1.261 schwarze 2482: scaled = *res * 40;
1.254 schwarze 2483: break;
2484: case 'm':
2485: case 'n':
1.261 schwarze 2486: scaled = *res * 24;
1.254 schwarze 2487: break;
2488: case 'p':
1.261 schwarze 2489: scaled = *res * 10 / 3;
1.254 schwarze 2490: break;
2491: case 'u':
1.261 schwarze 2492: scaled = *res;
1.254 schwarze 2493: break;
2494: case 'M':
1.261 schwarze 2495: scaled = *res * 6 / 25;
1.254 schwarze 2496: break;
2497: default:
1.261 schwarze 2498: scaled = *res;
1.254 schwarze 2499: p--;
2500: break;
2501: }
1.261 schwarze 2502: if (flags & ROFFNUM_SCALE)
2503: *res = scaled;
1.254 schwarze 2504:
2505: *pos = p + 1;
1.277 schwarze 2506: return 1;
1.184 schwarze 2507: }
2508:
1.198 schwarze 2509: /*
2510: * Evaluate a string comparison condition.
2511: * The first character is the delimiter.
2512: * Succeed if the string up to its second occurrence
2513: * matches the string up to its third occurence.
2514: * Advance the cursor after the third occurrence
2515: * or lacking that, to the end of the line.
2516: */
2517: static int
2518: roff_evalstrcond(const char *v, int *pos)
2519: {
2520: const char *s1, *s2, *s3;
2521: int match;
2522:
2523: match = 0;
2524: s1 = v + *pos; /* initial delimiter */
2525: s2 = s1 + 1; /* for scanning the first string */
2526: s3 = strchr(s2, *s1); /* for scanning the second string */
2527:
2528: if (NULL == s3) /* found no middle delimiter */
2529: goto out;
2530:
2531: while ('\0' != *++s3) {
2532: if (*s2 != *s3) { /* mismatch */
2533: s3 = strchr(s3, *s1);
2534: break;
2535: }
2536: if (*s3 == *s1) { /* found the final delimiter */
2537: match = 1;
2538: break;
2539: }
2540: s2++;
2541: }
2542:
2543: out:
2544: if (NULL == s3)
2545: s3 = strchr(s2, '\0');
1.242 schwarze 2546: else if (*s3 != '\0')
1.198 schwarze 2547: s3++;
2548: *pos = s3 - v;
1.277 schwarze 2549: return match;
1.198 schwarze 2550: }
2551:
1.204 schwarze 2552: /*
2553: * Evaluate an optionally negated single character, numerical,
2554: * or string condition.
2555: */
1.198 schwarze 2556: static int
1.271 schwarze 2557: roff_evalcond(struct roff *r, int ln, char *v, int *pos)
1.88 kristaps 2558: {
1.336 schwarze 2559: const char *start, *end;
2560: char *cp, *name;
2561: size_t sz;
2562: int deftype, len, number, savepos, istrue, wanttrue;
1.88 kristaps 2563:
1.198 schwarze 2564: if ('!' == v[*pos]) {
2565: wanttrue = 0;
2566: (*pos)++;
2567: } else
2568: wanttrue = 1;
2569:
1.88 kristaps 2570: switch (v[*pos]) {
1.240 schwarze 2571: case '\0':
1.277 schwarze 2572: return 0;
1.207 schwarze 2573: case 'n':
2574: case 'o':
1.88 kristaps 2575: (*pos)++;
1.277 schwarze 2576: return wanttrue;
1.207 schwarze 2577: case 'e':
2578: case 't':
1.239 schwarze 2579: case 'v':
1.88 kristaps 2580: (*pos)++;
1.277 schwarze 2581: return !wanttrue;
1.336 schwarze 2582: case 'c':
2583: do {
2584: (*pos)++;
2585: } while (v[*pos] == ' ');
2586:
2587: /*
2588: * Quirk for groff compatibility:
2589: * The horizontal tab is neither available nor unavailable.
2590: */
2591:
2592: if (v[*pos] == '\t') {
2593: (*pos)++;
2594: return 0;
2595: }
2596:
2597: /* Printable ASCII characters are available. */
2598:
2599: if (v[*pos] != '\\') {
2600: (*pos)++;
2601: return wanttrue;
2602: }
2603:
2604: end = v + ++*pos;
2605: switch (mandoc_escape(&end, &start, &len)) {
2606: case ESCAPE_SPECIAL:
2607: istrue = mchars_spec2cp(start, len) != -1;
2608: break;
2609: case ESCAPE_UNICODE:
2610: istrue = 1;
2611: break;
2612: case ESCAPE_NUMBERED:
2613: istrue = mchars_num2char(start, len) != -1;
2614: break;
2615: default:
2616: istrue = !wanttrue;
2617: break;
2618: }
2619: *pos = end - v;
2620: return istrue == wanttrue;
1.310 schwarze 2621: case 'd':
1.271 schwarze 2622: case 'r':
1.310 schwarze 2623: cp = v + *pos + 1;
2624: while (*cp == ' ')
2625: cp++;
2626: name = cp;
2627: sz = roff_getname(r, &cp, ln, cp - v);
1.315 schwarze 2628: if (sz == 0)
2629: istrue = 0;
2630: else if (v[*pos] == 'r')
2631: istrue = roff_hasregn(r, name, sz);
2632: else {
2633: deftype = ROFFDEF_ANY;
2634: roff_getstrn(r, name, sz, &deftype);
2635: istrue = !!deftype;
2636: }
1.363 schwarze 2637: *pos = (name + sz) - v;
1.310 schwarze 2638: return istrue == wanttrue;
1.88 kristaps 2639: default:
2640: break;
2641: }
2642:
1.241 schwarze 2643: savepos = *pos;
1.261 schwarze 2644: if (roff_evalnum(r, ln, v, pos, &number, ROFFNUM_SCALE))
1.277 schwarze 2645: return (number > 0) == wanttrue;
1.241 schwarze 2646: else if (*pos == savepos)
1.277 schwarze 2647: return roff_evalstrcond(v, pos) == wanttrue;
1.204 schwarze 2648: else
1.277 schwarze 2649: return 0;
1.88 kristaps 2650: }
2651:
1.340 schwarze 2652: static int
1.103 kristaps 2653: roff_line_ignore(ROFF_ARGS)
1.89 kristaps 2654: {
1.123 schwarze 2655:
1.277 schwarze 2656: return ROFF_IGN;
1.89 kristaps 2657: }
2658:
1.340 schwarze 2659: static int
1.251 schwarze 2660: roff_insec(ROFF_ARGS)
2661: {
2662:
1.350 schwarze 2663: mandoc_msg(MANDOCERR_REQ_INSEC, ln, ppos, "%s", roff_name[tok]);
1.277 schwarze 2664: return ROFF_IGN;
1.251 schwarze 2665: }
2666:
1.340 schwarze 2667: static int
1.251 schwarze 2668: roff_unsupp(ROFF_ARGS)
2669: {
2670:
1.350 schwarze 2671: mandoc_msg(MANDOCERR_REQ_UNSUPP, ln, ppos, "%s", roff_name[tok]);
1.277 schwarze 2672: return ROFF_IGN;
1.251 schwarze 2673: }
2674:
1.340 schwarze 2675: static int
1.82 kristaps 2676: roff_cond(ROFF_ARGS)
1.74 kristaps 2677: {
1.340 schwarze 2678: int irc;
1.173 schwarze 2679:
2680: roffnode_push(r, tok, NULL, ln, ppos);
1.74 kristaps 2681:
1.207 schwarze 2682: /*
1.134 kristaps 2683: * An `.el' has no conditional body: it will consume the value
2684: * of the current rstack entry set in prior `ie' calls or
1.207 schwarze 2685: * defaults to DENY.
1.134 kristaps 2686: *
2687: * If we're not an `el', however, then evaluate the conditional.
2688: */
1.133 kristaps 2689:
1.238 schwarze 2690: r->last->rule = tok == ROFF_el ?
1.207 schwarze 2691: (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
1.238 schwarze 2692: roff_evalcond(r, ln, buf->buf, &pos);
1.77 kristaps 2693:
1.134 kristaps 2694: /*
2695: * An if-else will put the NEGATION of the current evaluated
2696: * conditional into the stack of rules.
2697: */
2698:
1.238 schwarze 2699: if (tok == ROFF_ie) {
1.223 schwarze 2700: if (r->rstackpos + 1 == r->rstacksz) {
2701: r->rstacksz += 16;
2702: r->rstack = mandoc_reallocarray(r->rstack,
2703: r->rstacksz, sizeof(int));
1.134 kristaps 2704: }
1.198 schwarze 2705: r->rstack[++r->rstackpos] = !r->last->rule;
1.82 kristaps 2706: }
1.88 kristaps 2707:
2708: /* If the parent has false as its rule, then so do we. */
2709:
1.198 schwarze 2710: if (r->last->parent && !r->last->parent->rule)
2711: r->last->rule = 0;
1.88 kristaps 2712:
2713: /*
1.173 schwarze 2714: * Determine scope.
2715: * If there is nothing on the line after the conditional,
2716: * not even whitespace, use next-line scope.
1.340 schwarze 2717: * Except that .while does not support next-line scope.
1.88 kristaps 2718: */
1.74 kristaps 2719:
1.340 schwarze 2720: if (buf->buf[pos] == '\0' && tok != ROFF_while) {
1.173 schwarze 2721: r->last->endspan = 2;
2722: goto out;
2723: }
2724:
1.238 schwarze 2725: while (buf->buf[pos] == ' ')
1.173 schwarze 2726: pos++;
2727:
2728: /* An opening brace requests multiline scope. */
1.75 kristaps 2729:
1.238 schwarze 2730: if (buf->buf[pos] == '\\' && buf->buf[pos + 1] == '{') {
1.75 kristaps 2731: r->last->endspan = -1;
2732: pos += 2;
1.272 schwarze 2733: while (buf->buf[pos] == ' ')
2734: pos++;
1.173 schwarze 2735: goto out;
1.207 schwarze 2736: }
1.74 kristaps 2737:
1.77 kristaps 2738: /*
1.173 schwarze 2739: * Anything else following the conditional causes
2740: * single-line scope. Warn if the scope contains
2741: * nothing but trailing whitespace.
1.77 kristaps 2742: */
2743:
1.238 schwarze 2744: if (buf->buf[pos] == '\0')
1.350 schwarze 2745: mandoc_msg(MANDOCERR_COND_EMPTY,
2746: ln, ppos, "%s", roff_name[tok]);
1.77 kristaps 2747:
1.173 schwarze 2748: r->last->endspan = 1;
1.74 kristaps 2749:
1.173 schwarze 2750: out:
1.75 kristaps 2751: *offs = pos;
1.340 schwarze 2752: irc = ROFF_RERUN;
2753: if (tok == ROFF_while)
2754: irc |= ROFF_WHILE;
2755: return irc;
1.83 schwarze 2756: }
2757:
1.340 schwarze 2758: static int
1.92 schwarze 2759: roff_ds(ROFF_ARGS)
2760: {
1.212 schwarze 2761: char *string;
2762: const char *name;
2763: size_t namesz;
1.96 kristaps 2764:
1.251 schwarze 2765: /* Ignore groff compatibility mode for now. */
2766:
2767: if (tok == ROFF_ds1)
2768: tok = ROFF_ds;
2769: else if (tok == ROFF_as1)
2770: tok = ROFF_as;
2771:
1.96 kristaps 2772: /*
1.212 schwarze 2773: * The first word is the name of the string.
2774: * If it is empty or terminated by an escape sequence,
2775: * abort the `ds' request without defining anything.
1.96 kristaps 2776: */
1.92 schwarze 2777:
1.238 schwarze 2778: name = string = buf->buf + pos;
2779: if (*name == '\0')
1.277 schwarze 2780: return ROFF_IGN;
1.92 schwarze 2781:
1.212 schwarze 2782: namesz = roff_getname(r, &string, ln, pos);
1.363 schwarze 2783: switch (name[namesz]) {
2784: case '\\':
1.277 schwarze 2785: return ROFF_IGN;
1.363 schwarze 2786: case '\t':
2787: string = buf->buf + pos + namesz;
2788: break;
2789: default:
2790: break;
2791: }
1.212 schwarze 2792:
2793: /* Read past the initial double-quote, if any. */
1.238 schwarze 2794: if (*string == '"')
1.92 schwarze 2795: string++;
2796:
1.96 kristaps 2797: /* The rest is the value. */
1.212 schwarze 2798: roff_setstrn(&r->strtab, name, namesz, string, strlen(string),
2799: ROFF_as == tok);
1.311 schwarze 2800: roff_setstrn(&r->rentab, name, namesz, NULL, 0, 0);
1.277 schwarze 2801: return ROFF_IGN;
1.92 schwarze 2802: }
2803:
1.204 schwarze 2804: /*
2805: * Parse a single operator, one or two characters long.
2806: * If the operator is recognized, return success and advance the
2807: * parse point, else return failure and let the parse point unchanged.
2808: */
2809: static int
2810: roff_getop(const char *v, int *pos, char *res)
2811: {
2812:
2813: *res = v[*pos];
2814:
2815: switch (*res) {
1.207 schwarze 2816: case '+':
2817: case '-':
2818: case '*':
2819: case '/':
2820: case '%':
2821: case '&':
2822: case ':':
1.204 schwarze 2823: break;
2824: case '<':
2825: switch (v[*pos + 1]) {
1.207 schwarze 2826: case '=':
1.204 schwarze 2827: *res = 'l';
2828: (*pos)++;
2829: break;
1.207 schwarze 2830: case '>':
1.204 schwarze 2831: *res = '!';
2832: (*pos)++;
2833: break;
1.207 schwarze 2834: case '?':
1.204 schwarze 2835: *res = 'i';
2836: (*pos)++;
2837: break;
2838: default:
2839: break;
2840: }
2841: break;
2842: case '>':
2843: switch (v[*pos + 1]) {
1.207 schwarze 2844: case '=':
1.204 schwarze 2845: *res = 'g';
2846: (*pos)++;
2847: break;
1.207 schwarze 2848: case '?':
1.204 schwarze 2849: *res = 'a';
2850: (*pos)++;
2851: break;
2852: default:
2853: break;
2854: }
2855: break;
2856: case '=':
2857: if ('=' == v[*pos + 1])
2858: (*pos)++;
2859: break;
2860: default:
1.277 schwarze 2861: return 0;
1.204 schwarze 2862: }
2863: (*pos)++;
2864:
1.277 schwarze 2865: return *res;
1.204 schwarze 2866: }
2867:
2868: /*
2869: * Evaluate either a parenthesized numeric expression
2870: * or a single signed integer number.
2871: */
2872: static int
1.235 schwarze 2873: roff_evalpar(struct roff *r, int ln,
1.261 schwarze 2874: const char *v, int *pos, int *res, int flags)
1.204 schwarze 2875: {
2876:
2877: if ('(' != v[*pos])
1.277 schwarze 2878: return roff_getnum(v, pos, res, flags);
1.204 schwarze 2879:
2880: (*pos)++;
1.261 schwarze 2881: if ( ! roff_evalnum(r, ln, v, pos, res, flags | ROFFNUM_WHITE))
1.277 schwarze 2882: return 0;
1.204 schwarze 2883:
1.206 schwarze 2884: /*
2885: * Omission of the closing parenthesis
2886: * is an error in validation mode,
2887: * but ignored in evaluation mode.
2888: */
2889:
1.204 schwarze 2890: if (')' == v[*pos])
2891: (*pos)++;
1.206 schwarze 2892: else if (NULL == res)
1.277 schwarze 2893: return 0;
1.204 schwarze 2894:
1.277 schwarze 2895: return 1;
1.204 schwarze 2896: }
2897:
2898: /*
2899: * Evaluate a complete numeric expression.
2900: * Proceed left to right, there is no concept of precedence.
2901: */
2902: static int
1.235 schwarze 2903: roff_evalnum(struct roff *r, int ln, const char *v,
1.261 schwarze 2904: int *pos, int *res, int flags)
1.204 schwarze 2905: {
2906: int mypos, operand2;
2907: char operator;
2908:
2909: if (NULL == pos) {
2910: mypos = 0;
2911: pos = &mypos;
2912: }
2913:
1.261 schwarze 2914: if (flags & ROFFNUM_WHITE)
1.204 schwarze 2915: while (isspace((unsigned char)v[*pos]))
2916: (*pos)++;
2917:
1.261 schwarze 2918: if ( ! roff_evalpar(r, ln, v, pos, res, flags))
1.277 schwarze 2919: return 0;
1.204 schwarze 2920:
2921: while (1) {
1.261 schwarze 2922: if (flags & ROFFNUM_WHITE)
1.204 schwarze 2923: while (isspace((unsigned char)v[*pos]))
2924: (*pos)++;
2925:
2926: if ( ! roff_getop(v, pos, &operator))
2927: break;
2928:
1.261 schwarze 2929: if (flags & ROFFNUM_WHITE)
1.204 schwarze 2930: while (isspace((unsigned char)v[*pos]))
2931: (*pos)++;
2932:
1.261 schwarze 2933: if ( ! roff_evalpar(r, ln, v, pos, &operand2, flags))
1.277 schwarze 2934: return 0;
1.204 schwarze 2935:
1.261 schwarze 2936: if (flags & ROFFNUM_WHITE)
1.204 schwarze 2937: while (isspace((unsigned char)v[*pos]))
2938: (*pos)++;
1.206 schwarze 2939:
2940: if (NULL == res)
2941: continue;
1.204 schwarze 2942:
2943: switch (operator) {
1.207 schwarze 2944: case '+':
1.204 schwarze 2945: *res += operand2;
2946: break;
1.207 schwarze 2947: case '-':
1.204 schwarze 2948: *res -= operand2;
2949: break;
1.207 schwarze 2950: case '*':
1.204 schwarze 2951: *res *= operand2;
2952: break;
1.207 schwarze 2953: case '/':
1.244 schwarze 2954: if (operand2 == 0) {
1.235 schwarze 2955: mandoc_msg(MANDOCERR_DIVZERO,
1.350 schwarze 2956: ln, *pos, "%s", v);
1.234 kristaps 2957: *res = 0;
2958: break;
2959: }
1.204 schwarze 2960: *res /= operand2;
2961: break;
1.207 schwarze 2962: case '%':
1.244 schwarze 2963: if (operand2 == 0) {
2964: mandoc_msg(MANDOCERR_DIVZERO,
1.350 schwarze 2965: ln, *pos, "%s", v);
1.244 schwarze 2966: *res = 0;
2967: break;
2968: }
1.204 schwarze 2969: *res %= operand2;
2970: break;
1.207 schwarze 2971: case '<':
1.204 schwarze 2972: *res = *res < operand2;
2973: break;
1.207 schwarze 2974: case '>':
1.204 schwarze 2975: *res = *res > operand2;
2976: break;
1.207 schwarze 2977: case 'l':
1.204 schwarze 2978: *res = *res <= operand2;
2979: break;
1.207 schwarze 2980: case 'g':
1.204 schwarze 2981: *res = *res >= operand2;
2982: break;
1.207 schwarze 2983: case '=':
1.204 schwarze 2984: *res = *res == operand2;
2985: break;
1.207 schwarze 2986: case '!':
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 'i':
1.204 schwarze 2996: if (operand2 < *res)
2997: *res = operand2;
2998: break;
1.207 schwarze 2999: case 'a':
1.204 schwarze 3000: if (operand2 > *res)
3001: *res = operand2;
3002: break;
3003: default:
3004: abort();
3005: }
3006: }
1.277 schwarze 3007: return 1;
1.204 schwarze 3008: }
3009:
1.266 schwarze 3010: /* --- register management ------------------------------------------------ */
3011:
1.180 schwarze 3012: void
1.187 schwarze 3013: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.147 kristaps 3014: {
1.327 schwarze 3015: roff_setregn(r, name, strlen(name), val, sign, INT_MIN);
1.326 schwarze 3016: }
3017:
3018: static void
3019: roff_setregn(struct roff *r, const char *name, size_t len,
1.327 schwarze 3020: int val, char sign, int step)
1.326 schwarze 3021: {
1.180 schwarze 3022: struct roffreg *reg;
3023:
3024: /* Search for an existing register with the same name. */
3025: reg = r->regtab;
3026:
1.326 schwarze 3027: while (reg != NULL && (reg->key.sz != len ||
3028: strncmp(reg->key.p, name, len) != 0))
1.180 schwarze 3029: reg = reg->next;
1.147 kristaps 3030:
1.180 schwarze 3031: if (NULL == reg) {
3032: /* Create a new register. */
3033: reg = mandoc_malloc(sizeof(struct roffreg));
1.326 schwarze 3034: reg->key.p = mandoc_strndup(name, len);
3035: reg->key.sz = len;
1.187 schwarze 3036: reg->val = 0;
1.327 schwarze 3037: reg->step = 0;
1.180 schwarze 3038: reg->next = r->regtab;
3039: r->regtab = reg;
3040: }
3041:
1.187 schwarze 3042: if ('+' == sign)
3043: reg->val += val;
3044: else if ('-' == sign)
3045: reg->val -= val;
3046: else
3047: reg->val = val;
1.327 schwarze 3048: if (step != INT_MIN)
3049: reg->step = step;
1.147 kristaps 3050: }
3051:
1.192 schwarze 3052: /*
3053: * Handle some predefined read-only number registers.
3054: * For now, return -1 if the requested register is not predefined;
3055: * in case a predefined read-only register having the value -1
3056: * were to turn up, another special value would have to be chosen.
3057: */
3058: static int
1.273 schwarze 3059: roff_getregro(const struct roff *r, const char *name)
1.192 schwarze 3060: {
3061:
3062: switch (*name) {
1.273 schwarze 3063: case '$': /* Number of arguments of the last macro evaluated. */
1.339 schwarze 3064: return r->mstackpos < 0 ? 0 : r->mstack[r->mstackpos].argc;
1.207 schwarze 3065: case 'A': /* ASCII approximation mode is always off. */
1.277 schwarze 3066: return 0;
1.207 schwarze 3067: case 'g': /* Groff compatibility mode is always on. */
1.277 schwarze 3068: return 1;
1.207 schwarze 3069: case 'H': /* Fixed horizontal resolution. */
1.277 schwarze 3070: return 24;
1.207 schwarze 3071: case 'j': /* Always adjust left margin only. */
1.277 schwarze 3072: return 0;
1.207 schwarze 3073: case 'T': /* Some output device is always defined. */
1.277 schwarze 3074: return 1;
1.207 schwarze 3075: case 'V': /* Fixed vertical resolution. */
1.277 schwarze 3076: return 40;
1.192 schwarze 3077: default:
1.277 schwarze 3078: return -1;
1.192 schwarze 3079: }
3080: }
3081:
1.181 schwarze 3082: int
1.326 schwarze 3083: roff_getreg(struct roff *r, const char *name)
1.147 kristaps 3084: {
1.327 schwarze 3085: return roff_getregn(r, name, strlen(name), '\0');
1.181 schwarze 3086: }
3087:
3088: static int
1.327 schwarze 3089: roff_getregn(struct roff *r, const char *name, size_t len, char sign)
1.181 schwarze 3090: {
3091: struct roffreg *reg;
1.192 schwarze 3092: int val;
3093:
3094: if ('.' == name[0] && 2 == len) {
1.273 schwarze 3095: val = roff_getregro(r, name + 1);
1.192 schwarze 3096: if (-1 != val)
1.277 schwarze 3097: return val;
1.192 schwarze 3098: }
1.181 schwarze 3099:
1.327 schwarze 3100: for (reg = r->regtab; reg; reg = reg->next) {
1.181 schwarze 3101: if (len == reg->key.sz &&
1.327 schwarze 3102: 0 == strncmp(name, reg->key.p, len)) {
3103: switch (sign) {
3104: case '+':
3105: reg->val += reg->step;
3106: break;
3107: case '-':
3108: reg->val -= reg->step;
3109: break;
3110: default:
3111: break;
3112: }
1.277 schwarze 3113: return reg->val;
1.327 schwarze 3114: }
3115: }
1.271 schwarze 3116:
1.327 schwarze 3117: roff_setregn(r, name, len, 0, '\0', INT_MIN);
1.277 schwarze 3118: return 0;
1.271 schwarze 3119: }
3120:
3121: static int
3122: roff_hasregn(const struct roff *r, const char *name, size_t len)
3123: {
3124: struct roffreg *reg;
3125: int val;
3126:
3127: if ('.' == name[0] && 2 == len) {
1.273 schwarze 3128: val = roff_getregro(r, name + 1);
1.271 schwarze 3129: if (-1 != val)
1.277 schwarze 3130: return 1;
1.271 schwarze 3131: }
3132:
3133: for (reg = r->regtab; reg; reg = reg->next)
3134: if (len == reg->key.sz &&
3135: 0 == strncmp(name, reg->key.p, len))
1.277 schwarze 3136: return 1;
1.147 kristaps 3137:
1.277 schwarze 3138: return 0;
1.147 kristaps 3139: }
3140:
1.180 schwarze 3141: static void
3142: roff_freereg(struct roffreg *reg)
1.147 kristaps 3143: {
1.180 schwarze 3144: struct roffreg *old_reg;
1.147 kristaps 3145:
1.180 schwarze 3146: while (NULL != reg) {
3147: free(reg->key.p);
3148: old_reg = reg;
3149: reg = reg->next;
3150: free(old_reg);
3151: }
1.147 kristaps 3152: }
1.92 schwarze 3153:
1.340 schwarze 3154: static int
1.89 kristaps 3155: roff_nr(ROFF_ARGS)
1.83 schwarze 3156: {
1.327 schwarze 3157: char *key, *val, *step;
1.212 schwarze 3158: size_t keysz;
1.327 schwarze 3159: int iv, is, len;
1.187 schwarze 3160: char sign;
1.89 kristaps 3161:
1.238 schwarze 3162: key = val = buf->buf + pos;
3163: if (*key == '\0')
1.277 schwarze 3164: return ROFF_IGN;
1.212 schwarze 3165:
3166: keysz = roff_getname(r, &val, ln, pos);
1.363 schwarze 3167: if (key[keysz] == '\\' || key[keysz] == '\t')
1.277 schwarze 3168: return ROFF_IGN;
1.89 kristaps 3169:
1.187 schwarze 3170: sign = *val;
1.238 schwarze 3171: if (sign == '+' || sign == '-')
1.187 schwarze 3172: val++;
3173:
1.327 schwarze 3174: len = 0;
3175: if (roff_evalnum(r, ln, val, &len, &iv, ROFFNUM_SCALE) == 0)
3176: return ROFF_IGN;
3177:
3178: step = val + len;
3179: while (isspace((unsigned char)*step))
3180: step++;
3181: if (roff_evalnum(r, ln, step, NULL, &is, 0) == 0)
3182: is = INT_MIN;
1.109 kristaps 3183:
1.327 schwarze 3184: roff_setregn(r, key, keysz, iv, sign, is);
1.277 schwarze 3185: return ROFF_IGN;
1.203 schwarze 3186: }
3187:
1.340 schwarze 3188: static int
1.203 schwarze 3189: roff_rr(ROFF_ARGS)
3190: {
3191: struct roffreg *reg, **prev;
1.212 schwarze 3192: char *name, *cp;
3193: size_t namesz;
1.203 schwarze 3194:
1.238 schwarze 3195: name = cp = buf->buf + pos;
3196: if (*name == '\0')
1.277 schwarze 3197: return ROFF_IGN;
1.212 schwarze 3198: namesz = roff_getname(r, &cp, ln, pos);
3199: name[namesz] = '\0';
1.203 schwarze 3200:
3201: prev = &r->regtab;
3202: while (1) {
3203: reg = *prev;
1.238 schwarze 3204: if (reg == NULL || !strcmp(name, reg->key.p))
1.203 schwarze 3205: break;
3206: prev = ®->next;
3207: }
1.238 schwarze 3208: if (reg != NULL) {
1.203 schwarze 3209: *prev = reg->next;
3210: free(reg->key.p);
3211: free(reg);
3212: }
1.277 schwarze 3213: return ROFF_IGN;
1.122 schwarze 3214: }
3215:
1.266 schwarze 3216: /* --- handler functions for roff requests -------------------------------- */
3217:
1.340 schwarze 3218: static int
1.122 schwarze 3219: roff_rm(ROFF_ARGS)
3220: {
3221: const char *name;
3222: char *cp;
1.212 schwarze 3223: size_t namesz;
1.122 schwarze 3224:
1.238 schwarze 3225: cp = buf->buf + pos;
3226: while (*cp != '\0') {
1.212 schwarze 3227: name = cp;
1.238 schwarze 3228: namesz = roff_getname(r, &cp, ln, (int)(cp - buf->buf));
1.212 schwarze 3229: roff_setstrn(&r->strtab, name, namesz, NULL, 0, 0);
1.311 schwarze 3230: roff_setstrn(&r->rentab, name, namesz, NULL, 0, 0);
1.363 schwarze 3231: if (name[namesz] == '\\' || name[namesz] == '\t')
1.212 schwarze 3232: break;
1.122 schwarze 3233: }
1.277 schwarze 3234: return ROFF_IGN;
1.178 schwarze 3235: }
3236:
1.340 schwarze 3237: static int
1.178 schwarze 3238: roff_it(ROFF_ARGS)
3239: {
3240: int iv;
3241:
3242: /* Parse the number of lines. */
1.261 schwarze 3243:
3244: if ( ! roff_evalnum(r, ln, buf->buf, &pos, &iv, 0)) {
1.350 schwarze 3245: mandoc_msg(MANDOCERR_IT_NONUM,
3246: ln, ppos, "%s", buf->buf + 1);
1.277 schwarze 3247: return ROFF_IGN;
1.178 schwarze 3248: }
3249:
1.262 schwarze 3250: while (isspace((unsigned char)buf->buf[pos]))
3251: pos++;
3252:
3253: /*
3254: * Arm the input line trap.
3255: * Special-casing "an-trap" is an ugly workaround to cope
3256: * with DocBook stupidly fiddling with man(7) internals.
3257: */
1.261 schwarze 3258:
1.178 schwarze 3259: roffit_lines = iv;
1.262 schwarze 3260: roffit_macro = mandoc_strdup(iv != 1 ||
3261: strcmp(buf->buf + pos, "an-trap") ?
3262: buf->buf + pos : "br");
1.277 schwarze 3263: return ROFF_IGN;
1.175 schwarze 3264: }
3265:
1.340 schwarze 3266: static int
1.175 schwarze 3267: roff_Dd(ROFF_ARGS)
3268: {
1.315 schwarze 3269: int mask;
3270: enum roff_tok t, te;
1.227 schwarze 3271:
1.315 schwarze 3272: switch (tok) {
3273: case ROFF_Dd:
3274: tok = MDOC_Dd;
3275: te = MDOC_MAX;
3276: if (r->format == 0)
3277: r->format = MPARSE_MDOC;
3278: mask = MPARSE_MDOC | MPARSE_QUICK;
3279: break;
3280: case ROFF_TH:
3281: tok = MAN_TH;
3282: te = MAN_MAX;
3283: if (r->format == 0)
3284: r->format = MPARSE_MAN;
3285: mask = MPARSE_QUICK;
3286: break;
3287: default:
3288: abort();
3289: }
3290: if ((r->options & mask) == 0)
3291: for (t = tok; t < te; t++)
3292: roff_setstr(r, roff_name[t], NULL, 0);
1.277 schwarze 3293: return ROFF_CONT;
1.109 kristaps 3294: }
3295:
1.340 schwarze 3296: static int
1.109 kristaps 3297: roff_TE(ROFF_ARGS)
3298: {
1.361 schwarze 3299: r->man->flags &= ~ROFF_NONOFILL;
1.321 schwarze 3300: if (r->tbl == NULL) {
1.350 schwarze 3301: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "TE");
1.321 schwarze 3302: return ROFF_IGN;
3303: }
1.346 schwarze 3304: if (tbl_end(r->tbl, 0) == 0) {
1.321 schwarze 3305: r->tbl = NULL;
1.258 schwarze 3306: free(buf->buf);
3307: buf->buf = mandoc_strdup(".sp");
3308: buf->sz = 4;
1.329 schwarze 3309: *offs = 0;
1.277 schwarze 3310: return ROFF_REPARSE;
1.258 schwarze 3311: }
1.321 schwarze 3312: r->tbl = NULL;
1.277 schwarze 3313: return ROFF_IGN;
1.112 kristaps 3314: }
3315:
1.340 schwarze 3316: static int
1.112 kristaps 3317: roff_T_(ROFF_ARGS)
3318: {
3319:
3320: if (NULL == r->tbl)
1.350 schwarze 3321: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "T&");
1.112 kristaps 3322: else
1.296 schwarze 3323: tbl_restart(ln, ppos, r->tbl);
1.112 kristaps 3324:
1.277 schwarze 3325: return ROFF_IGN;
1.109 kristaps 3326: }
3327:
1.230 schwarze 3328: /*
3329: * Handle in-line equation delimiters.
3330: */
1.340 schwarze 3331: static int
1.238 schwarze 3332: roff_eqndelim(struct roff *r, struct buf *buf, int pos)
1.151 kristaps 3333: {
1.233 schwarze 3334: char *cp1, *cp2;
3335: const char *bef_pr, *bef_nl, *mac, *aft_nl, *aft_pr;
1.151 kristaps 3336:
1.230 schwarze 3337: /*
3338: * Outside equations, look for an opening delimiter.
3339: * If we are inside an equation, we already know it is
3340: * in-line, or this function wouldn't have been called;
3341: * so look for a closing delimiter.
3342: */
3343:
1.238 schwarze 3344: cp1 = buf->buf + pos;
1.230 schwarze 3345: cp2 = strchr(cp1, r->eqn == NULL ?
3346: r->last_eqn->odelim : r->last_eqn->cdelim);
3347: if (cp2 == NULL)
1.277 schwarze 3348: return ROFF_CONT;
1.230 schwarze 3349:
1.233 schwarze 3350: *cp2++ = '\0';
3351: bef_pr = bef_nl = aft_nl = aft_pr = "";
3352:
3353: /* Handle preceding text, protecting whitespace. */
3354:
1.238 schwarze 3355: if (*buf->buf != '\0') {
1.233 schwarze 3356: if (r->eqn == NULL)
3357: bef_pr = "\\&";
3358: bef_nl = "\n";
3359: }
3360:
3361: /*
3362: * Prepare replacing the delimiter with an equation macro
3363: * and drop leading white space from the equation.
3364: */
1.230 schwarze 3365:
1.233 schwarze 3366: if (r->eqn == NULL) {
3367: while (*cp2 == ' ')
3368: cp2++;
3369: mac = ".EQ";
3370: } else
3371: mac = ".EN";
3372:
3373: /* Handle following text, protecting whitespace. */
3374:
3375: if (*cp2 != '\0') {
3376: aft_nl = "\n";
3377: if (r->eqn != NULL)
3378: aft_pr = "\\&";
3379: }
3380:
3381: /* Do the actual replacement. */
3382:
1.238 schwarze 3383: buf->sz = mandoc_asprintf(&cp1, "%s%s%s%s%s%s%s", buf->buf,
1.233 schwarze 3384: bef_pr, bef_nl, mac, aft_nl, aft_pr, cp2) + 1;
1.238 schwarze 3385: free(buf->buf);
3386: buf->buf = cp1;
1.230 schwarze 3387:
3388: /* Toggle the in-line state of the eqn subsystem. */
3389:
3390: r->eqn_inline = r->eqn == NULL;
1.277 schwarze 3391: return ROFF_REPARSE;
1.151 kristaps 3392: }
3393:
1.340 schwarze 3394: static int
1.235 schwarze 3395: roff_EQ(ROFF_ARGS)
1.125 kristaps 3396: {
1.319 schwarze 3397: struct roff_node *n;
3398:
1.356 schwarze 3399: if (r->man->meta.macroset == MACROSET_MAN)
1.322 schwarze 3400: man_breakscope(r->man, ROFF_EQ);
1.319 schwarze 3401: n = roff_node_alloc(r->man, ln, ppos, ROFFT_EQN, TOKEN_NONE);
3402: if (ln > r->man->last->line)
3403: n->flags |= NODE_LINE;
1.348 schwarze 3404: n->eqn = eqn_box_new();
1.319 schwarze 3405: roff_node_append(r->man, n);
3406: r->man->next = ROFF_NEXT_SIBLING;
1.125 kristaps 3407:
1.238 schwarze 3408: assert(r->eqn == NULL);
1.319 schwarze 3409: if (r->last_eqn == NULL)
1.351 schwarze 3410: r->last_eqn = eqn_alloc();
1.319 schwarze 3411: else
3412: eqn_reset(r->last_eqn);
3413: r->eqn = r->last_eqn;
3414: r->eqn->node = n;
1.151 kristaps 3415:
1.238 schwarze 3416: if (buf->buf[pos] != '\0')
1.350 schwarze 3417: mandoc_msg(MANDOCERR_ARG_SKIP, ln, pos,
1.238 schwarze 3418: ".EQ %s", buf->buf + pos);
1.151 kristaps 3419:
1.277 schwarze 3420: return ROFF_IGN;
1.125 kristaps 3421: }
3422:
1.340 schwarze 3423: static int
1.125 kristaps 3424: roff_EN(ROFF_ARGS)
3425: {
1.319 schwarze 3426: if (r->eqn != NULL) {
3427: eqn_parse(r->eqn);
3428: r->eqn = NULL;
3429: } else
1.350 schwarze 3430: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "EN");
1.319 schwarze 3431: if (buf->buf[pos] != '\0')
1.350 schwarze 3432: mandoc_msg(MANDOCERR_ARG_SKIP, ln, pos,
1.319 schwarze 3433: "EN %s", buf->buf + pos);
1.277 schwarze 3434: return ROFF_IGN;
1.125 kristaps 3435: }
3436:
1.340 schwarze 3437: static int
1.109 kristaps 3438: roff_TS(ROFF_ARGS)
3439: {
1.321 schwarze 3440: if (r->tbl != NULL) {
1.350 schwarze 3441: mandoc_msg(MANDOCERR_BLK_BROKEN, ln, ppos, "TS breaks TS");
1.346 schwarze 3442: tbl_end(r->tbl, 0);
1.115 kristaps 3443: }
1.361 schwarze 3444: r->man->flags |= ROFF_NONOFILL;
1.351 schwarze 3445: r->tbl = tbl_alloc(ppos, ln, r->last_tbl);
1.346 schwarze 3446: if (r->last_tbl == NULL)
1.321 schwarze 3447: r->first_tbl = r->tbl;
3448: r->last_tbl = r->tbl;
1.297 schwarze 3449: return ROFF_IGN;
3450: }
3451:
1.340 schwarze 3452: static int
1.358 schwarze 3453: roff_noarg(ROFF_ARGS)
3454: {
3455: if (r->man->flags & (MAN_BLINE | MAN_ELINE))
3456: man_breakscope(r->man, tok);
3457: if (tok == ROFF_brp)
3458: tok = ROFF_br;
3459: roff_elem_alloc(r->man, ln, ppos, tok);
3460: if (buf->buf[pos] != '\0')
3461: mandoc_msg(MANDOCERR_ARG_SKIP, ln, pos,
3462: "%s %s", roff_name[tok], buf->buf + pos);
3463: if (tok == ROFF_nf)
3464: r->man->flags |= ROFF_NOFILL;
3465: else if (tok == ROFF_fi)
3466: r->man->flags &= ~ROFF_NOFILL;
3467: r->man->last->flags |= NODE_LINE | NODE_VALID | NODE_ENDED;
3468: r->man->next = ROFF_NEXT_SIBLING;
3469: return ROFF_IGN;
3470: }
3471:
3472: static int
1.297 schwarze 3473: roff_onearg(ROFF_ARGS)
3474: {
3475: struct roff_node *n;
3476: char *cp;
1.305 schwarze 3477: int npos;
1.297 schwarze 3478:
1.302 schwarze 3479: if (r->man->flags & (MAN_BLINE | MAN_ELINE) &&
1.320 schwarze 3480: (tok == ROFF_ce || tok == ROFF_rj || tok == ROFF_sp ||
3481: tok == ROFF_ti))
1.302 schwarze 3482: man_breakscope(r->man, tok);
3483:
1.309 schwarze 3484: if (roffce_node != NULL && (tok == ROFF_ce || tok == ROFF_rj)) {
1.305 schwarze 3485: r->man->last = roffce_node;
3486: r->man->next = ROFF_NEXT_SIBLING;
3487: }
3488:
1.297 schwarze 3489: roff_elem_alloc(r->man, ln, ppos, tok);
3490: n = r->man->last;
3491:
3492: cp = buf->buf + pos;
3493: if (*cp != '\0') {
3494: while (*cp != '\0' && *cp != ' ')
3495: cp++;
3496: while (*cp == ' ')
3497: *cp++ = '\0';
3498: if (*cp != '\0')
1.350 schwarze 3499: mandoc_msg(MANDOCERR_ARG_EXCESS,
3500: ln, (int)(cp - buf->buf),
1.297 schwarze 3501: "%s ... %s", roff_name[tok], cp);
3502: roff_word_alloc(r->man, ln, pos, buf->buf + pos);
1.300 schwarze 3503: }
3504:
1.309 schwarze 3505: if (tok == ROFF_ce || tok == ROFF_rj) {
3506: if (r->man->last->type == ROFFT_ELEM) {
1.305 schwarze 3507: roff_word_alloc(r->man, ln, pos, "1");
3508: r->man->last->flags |= NODE_NOSRC;
3509: }
3510: npos = 0;
3511: if (roff_evalnum(r, ln, r->man->last->string, &npos,
3512: &roffce_lines, 0) == 0) {
1.350 schwarze 3513: mandoc_msg(MANDOCERR_CE_NONUM,
3514: ln, pos, "ce %s", buf->buf + pos);
1.305 schwarze 3515: roffce_lines = 1;
3516: }
3517: if (roffce_lines < 1) {
3518: r->man->last = r->man->last->parent;
3519: roffce_node = NULL;
3520: roffce_lines = 0;
3521: } else
3522: roffce_node = r->man->last->parent;
3523: } else {
3524: n->flags |= NODE_VALID | NODE_ENDED;
3525: r->man->last = n;
3526: }
3527: n->flags |= NODE_LINE;
1.300 schwarze 3528: r->man->next = ROFF_NEXT_SIBLING;
3529: return ROFF_IGN;
3530: }
3531:
1.340 schwarze 3532: static int
1.300 schwarze 3533: roff_manyarg(ROFF_ARGS)
3534: {
3535: struct roff_node *n;
3536: char *sp, *ep;
3537:
3538: roff_elem_alloc(r->man, ln, ppos, tok);
3539: n = r->man->last;
3540:
3541: for (sp = ep = buf->buf + pos; *sp != '\0'; sp = ep) {
3542: while (*ep != '\0' && *ep != ' ')
3543: ep++;
3544: while (*ep == ' ')
3545: *ep++ = '\0';
3546: roff_word_alloc(r->man, ln, sp - buf->buf, sp);
1.297 schwarze 3547: }
3548:
3549: n->flags |= NODE_LINE | NODE_VALID | NODE_ENDED;
3550: r->man->last = n;
3551: r->man->next = ROFF_NEXT_SIBLING;
1.277 schwarze 3552: return ROFF_IGN;
1.251 schwarze 3553: }
3554:
1.340 schwarze 3555: static int
1.311 schwarze 3556: roff_als(ROFF_ARGS)
3557: {
3558: char *oldn, *newn, *end, *value;
3559: size_t oldsz, newsz, valsz;
3560:
3561: newn = oldn = buf->buf + pos;
3562: if (*newn == '\0')
3563: return ROFF_IGN;
3564:
3565: newsz = roff_getname(r, &oldn, ln, pos);
1.363 schwarze 3566: if (newn[newsz] == '\\' || newn[newsz] == '\t' || *oldn == '\0')
1.311 schwarze 3567: return ROFF_IGN;
3568:
3569: end = oldn;
3570: oldsz = roff_getname(r, &end, ln, oldn - buf->buf);
3571: if (oldsz == 0)
3572: return ROFF_IGN;
3573:
1.338 schwarze 3574: valsz = mandoc_asprintf(&value, ".%.*s \\$@\\\"\n",
1.315 schwarze 3575: (int)oldsz, oldn);
1.311 schwarze 3576: roff_setstrn(&r->strtab, newn, newsz, value, valsz, 0);
3577: roff_setstrn(&r->rentab, newn, newsz, NULL, 0, 0);
3578: free(value);
1.364 schwarze 3579: return ROFF_IGN;
3580: }
3581:
3582: /*
3583: * The .break request only makes sense inside conditionals,
3584: * and that case is already handled in roff_cond_sub().
3585: */
3586: static int
3587: roff_break(ROFF_ARGS)
3588: {
3589: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, pos, "break");
1.296 schwarze 3590: return ROFF_IGN;
1.92 schwarze 3591: }
3592:
1.340 schwarze 3593: static int
1.174 kristaps 3594: roff_cc(ROFF_ARGS)
3595: {
3596: const char *p;
3597:
1.238 schwarze 3598: p = buf->buf + pos;
1.174 kristaps 3599:
1.238 schwarze 3600: if (*p == '\0' || (r->control = *p++) == '.')
1.303 schwarze 3601: r->control = '\0';
1.174 kristaps 3602:
1.238 schwarze 3603: if (*p != '\0')
1.350 schwarze 3604: mandoc_msg(MANDOCERR_ARG_EXCESS,
1.260 schwarze 3605: ln, p - buf->buf, "cc ... %s", p);
1.174 kristaps 3606:
1.341 schwarze 3607: return ROFF_IGN;
3608: }
3609:
3610: static int
3611: roff_char(ROFF_ARGS)
3612: {
3613: const char *p, *kp, *vp;
3614: size_t ksz, vsz;
3615: int font;
3616:
3617: /* Parse the character to be replaced. */
3618:
3619: kp = buf->buf + pos;
3620: p = kp + 1;
3621: if (*kp == '\0' || (*kp == '\\' &&
3622: mandoc_escape(&p, NULL, NULL) != ESCAPE_SPECIAL) ||
3623: (*p != ' ' && *p != '\0')) {
1.350 schwarze 3624: mandoc_msg(MANDOCERR_CHAR_ARG, ln, pos, "char %s", kp);
1.341 schwarze 3625: return ROFF_IGN;
3626: }
3627: ksz = p - kp;
3628: while (*p == ' ')
3629: p++;
3630:
3631: /*
3632: * If the replacement string contains a font escape sequence,
3633: * we have to restore the font at the end.
3634: */
3635:
3636: vp = p;
3637: vsz = strlen(p);
3638: font = 0;
3639: while (*p != '\0') {
3640: if (*p++ != '\\')
3641: continue;
3642: switch (mandoc_escape(&p, NULL, NULL)) {
3643: case ESCAPE_FONT:
3644: case ESCAPE_FONTROMAN:
3645: case ESCAPE_FONTITALIC:
3646: case ESCAPE_FONTBOLD:
3647: case ESCAPE_FONTBI:
1.342 schwarze 3648: case ESCAPE_FONTCW:
1.341 schwarze 3649: case ESCAPE_FONTPREV:
3650: font++;
3651: break;
3652: default:
3653: break;
3654: }
3655: }
3656: if (font > 1)
1.350 schwarze 3657: mandoc_msg(MANDOCERR_CHAR_FONT,
3658: ln, (int)(vp - buf->buf), "%s", vp);
1.341 schwarze 3659:
3660: /*
3661: * Approximate the effect of .char using the .tr tables.
3662: * XXX In groff, .char and .tr interact differently.
3663: */
3664:
3665: if (ksz == 1) {
3666: if (r->xtab == NULL)
3667: r->xtab = mandoc_calloc(128, sizeof(*r->xtab));
3668: assert((unsigned int)*kp < 128);
3669: free(r->xtab[(int)*kp].p);
3670: r->xtab[(int)*kp].sz = mandoc_asprintf(&r->xtab[(int)*kp].p,
3671: "%s%s", vp, font ? "\fP" : "");
3672: } else {
3673: roff_setstrn(&r->xmbtab, kp, ksz, vp, vsz, 0);
3674: if (font)
3675: roff_setstrn(&r->xmbtab, kp, ksz, "\\fP", 3, 1);
3676: }
1.277 schwarze 3677: return ROFF_IGN;
1.174 kristaps 3678: }
3679:
1.340 schwarze 3680: static int
1.303 schwarze 3681: roff_ec(ROFF_ARGS)
3682: {
3683: const char *p;
3684:
3685: p = buf->buf + pos;
3686: if (*p == '\0')
3687: r->escape = '\\';
3688: else {
3689: r->escape = *p;
3690: if (*++p != '\0')
1.350 schwarze 3691: mandoc_msg(MANDOCERR_ARG_EXCESS, ln,
3692: (int)(p - buf->buf), "ec ... %s", p);
1.303 schwarze 3693: }
3694: return ROFF_IGN;
3695: }
3696:
1.340 schwarze 3697: static int
1.303 schwarze 3698: roff_eo(ROFF_ARGS)
3699: {
3700: r->escape = '\0';
3701: if (buf->buf[pos] != '\0')
1.350 schwarze 3702: mandoc_msg(MANDOCERR_ARG_SKIP,
1.303 schwarze 3703: ln, pos, "eo %s", buf->buf + pos);
3704: return ROFF_IGN;
1.330 schwarze 3705: }
3706:
1.340 schwarze 3707: static int
1.330 schwarze 3708: roff_nop(ROFF_ARGS)
3709: {
3710: while (buf->buf[pos] == ' ')
3711: pos++;
3712: *offs = pos;
3713: return ROFF_RERUN;
1.303 schwarze 3714: }
3715:
1.340 schwarze 3716: static int
1.164 kristaps 3717: roff_tr(ROFF_ARGS)
3718: {
3719: const char *p, *first, *second;
3720: size_t fsz, ssz;
3721: enum mandoc_esc esc;
3722:
1.238 schwarze 3723: p = buf->buf + pos;
1.164 kristaps 3724:
1.238 schwarze 3725: if (*p == '\0') {
1.350 schwarze 3726: mandoc_msg(MANDOCERR_REQ_EMPTY, ln, ppos, "tr");
1.277 schwarze 3727: return ROFF_IGN;
1.164 kristaps 3728: }
3729:
1.238 schwarze 3730: while (*p != '\0') {
1.164 kristaps 3731: fsz = ssz = 1;
3732:
3733: first = p++;
1.238 schwarze 3734: if (*first == '\\') {
1.164 kristaps 3735: esc = mandoc_escape(&p, NULL, NULL);
1.238 schwarze 3736: if (esc == ESCAPE_ERROR) {
1.350 schwarze 3737: mandoc_msg(MANDOCERR_ESC_BAD, ln,
3738: (int)(p - buf->buf), "%s", first);
1.277 schwarze 3739: return ROFF_IGN;
1.164 kristaps 3740: }
3741: fsz = (size_t)(p - first);
3742: }
3743:
3744: second = p++;
1.238 schwarze 3745: if (*second == '\\') {
1.164 kristaps 3746: esc = mandoc_escape(&p, NULL, NULL);
1.238 schwarze 3747: if (esc == ESCAPE_ERROR) {
1.350 schwarze 3748: mandoc_msg(MANDOCERR_ESC_BAD, ln,
3749: (int)(p - buf->buf), "%s", second);
1.277 schwarze 3750: return ROFF_IGN;
1.164 kristaps 3751: }
3752: ssz = (size_t)(p - second);
1.238 schwarze 3753: } else if (*second == '\0') {
1.350 schwarze 3754: mandoc_msg(MANDOCERR_TR_ODD, ln,
3755: (int)(first - buf->buf), "tr %s", first);
1.164 kristaps 3756: second = " ";
1.165 kristaps 3757: p--;
1.164 kristaps 3758: }
3759:
1.167 kristaps 3760: if (fsz > 1) {
1.207 schwarze 3761: roff_setstrn(&r->xmbtab, first, fsz,
3762: second, ssz, 0);
1.167 kristaps 3763: continue;
3764: }
3765:
1.238 schwarze 3766: if (r->xtab == NULL)
1.207 schwarze 3767: r->xtab = mandoc_calloc(128,
3768: sizeof(struct roffstr));
1.167 kristaps 3769:
3770: free(r->xtab[(int)*first].p);
3771: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
3772: r->xtab[(int)*first].sz = ssz;
1.164 kristaps 3773: }
3774:
1.277 schwarze 3775: return ROFF_IGN;
1.164 kristaps 3776: }
3777:
1.339 schwarze 3778: /*
3779: * Implementation of the .return request.
3780: * There is no need to call roff_userret() from here.
3781: * The read module will call that after rewinding the reader stack
3782: * to the place from where the current macro was called.
3783: */
1.340 schwarze 3784: static int
1.339 schwarze 3785: roff_return(ROFF_ARGS)
3786: {
3787: if (r->mstackpos >= 0)
1.340 schwarze 3788: return ROFF_IGN | ROFF_USERRET;
1.339 schwarze 3789:
1.350 schwarze 3790: mandoc_msg(MANDOCERR_REQ_NOMAC, ln, ppos, "return");
1.339 schwarze 3791: return ROFF_IGN;
3792: }
3793:
1.340 schwarze 3794: static int
1.306 schwarze 3795: roff_rn(ROFF_ARGS)
3796: {
3797: const char *value;
3798: char *oldn, *newn, *end;
3799: size_t oldsz, newsz;
1.315 schwarze 3800: int deftype;
1.306 schwarze 3801:
3802: oldn = newn = buf->buf + pos;
3803: if (*oldn == '\0')
3804: return ROFF_IGN;
3805:
3806: oldsz = roff_getname(r, &newn, ln, pos);
1.363 schwarze 3807: if (oldn[oldsz] == '\\' || oldn[oldsz] == '\t' || *newn == '\0')
1.306 schwarze 3808: return ROFF_IGN;
3809:
3810: end = newn;
3811: newsz = roff_getname(r, &end, ln, newn - buf->buf);
3812: if (newsz == 0)
3813: return ROFF_IGN;
3814:
1.315 schwarze 3815: deftype = ROFFDEF_ANY;
3816: value = roff_getstrn(r, oldn, oldsz, &deftype);
3817: switch (deftype) {
3818: case ROFFDEF_USER:
1.306 schwarze 3819: roff_setstrn(&r->strtab, newn, newsz, value, strlen(value), 0);
3820: roff_setstrn(&r->strtab, oldn, oldsz, NULL, 0, 0);
3821: roff_setstrn(&r->rentab, newn, newsz, NULL, 0, 0);
1.315 schwarze 3822: break;
3823: case ROFFDEF_PRE:
3824: roff_setstrn(&r->strtab, newn, newsz, value, strlen(value), 0);
3825: roff_setstrn(&r->rentab, newn, newsz, NULL, 0, 0);
3826: break;
3827: case ROFFDEF_REN:
1.306 schwarze 3828: roff_setstrn(&r->rentab, newn, newsz, value, strlen(value), 0);
3829: roff_setstrn(&r->rentab, oldn, oldsz, NULL, 0, 0);
1.315 schwarze 3830: roff_setstrn(&r->strtab, newn, newsz, NULL, 0, 0);
3831: break;
3832: case ROFFDEF_STD:
1.306 schwarze 3833: roff_setstrn(&r->rentab, newn, newsz, oldn, oldsz, 0);
1.315 schwarze 3834: roff_setstrn(&r->strtab, newn, newsz, NULL, 0, 0);
3835: break;
3836: default:
3837: roff_setstrn(&r->strtab, newn, newsz, NULL, 0, 0);
3838: roff_setstrn(&r->rentab, newn, newsz, NULL, 0, 0);
3839: break;
3840: }
1.306 schwarze 3841: return ROFF_IGN;
3842: }
3843:
1.340 schwarze 3844: static int
1.339 schwarze 3845: roff_shift(ROFF_ARGS)
3846: {
3847: struct mctx *ctx;
3848: int levels, i;
3849:
3850: levels = 1;
3851: if (buf->buf[pos] != '\0' &&
3852: roff_evalnum(r, ln, buf->buf, &pos, &levels, 0) == 0) {
1.350 schwarze 3853: mandoc_msg(MANDOCERR_CE_NONUM,
1.339 schwarze 3854: ln, pos, "shift %s", buf->buf + pos);
3855: levels = 1;
3856: }
3857: if (r->mstackpos < 0) {
1.350 schwarze 3858: mandoc_msg(MANDOCERR_REQ_NOMAC, ln, ppos, "shift");
1.339 schwarze 3859: return ROFF_IGN;
3860: }
3861: ctx = r->mstack + r->mstackpos;
3862: if (levels > ctx->argc) {
1.350 schwarze 3863: mandoc_msg(MANDOCERR_SHIFT,
1.339 schwarze 3864: ln, pos, "%d, but max is %d", levels, ctx->argc);
3865: levels = ctx->argc;
3866: }
3867: if (levels == 0)
3868: return ROFF_IGN;
3869: for (i = 0; i < levels; i++)
3870: free(ctx->argv[i]);
3871: ctx->argc -= levels;
3872: for (i = 0; i < ctx->argc; i++)
3873: ctx->argv[i] = ctx->argv[i + levels];
3874: return ROFF_IGN;
3875: }
3876:
1.340 schwarze 3877: static int
1.105 kristaps 3878: roff_so(ROFF_ARGS)
3879: {
1.249 schwarze 3880: char *name, *cp;
1.105 kristaps 3881:
1.238 schwarze 3882: name = buf->buf + pos;
1.350 schwarze 3883: mandoc_msg(MANDOCERR_SO, ln, ppos, "so %s", name);
1.105 kristaps 3884:
3885: /*
3886: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
3887: * opening anything that's not in our cwd or anything beneath
3888: * it. Thus, explicitly disallow traversing up the file-system
3889: * or using absolute paths.
3890: */
3891:
1.238 schwarze 3892: if (*name == '/' || strstr(name, "../") || strstr(name, "/..")) {
1.350 schwarze 3893: mandoc_msg(MANDOCERR_SO_PATH, ln, ppos, ".so %s", name);
1.249 schwarze 3894: buf->sz = mandoc_asprintf(&cp,
3895: ".sp\nSee the file %s.\n.sp", name) + 1;
3896: free(buf->buf);
3897: buf->buf = cp;
3898: *offs = 0;
1.277 schwarze 3899: return ROFF_REPARSE;
1.105 kristaps 3900: }
3901:
3902: *offs = pos;
1.277 schwarze 3903: return ROFF_SO;
1.105 kristaps 3904: }
1.92 schwarze 3905:
1.266 schwarze 3906: /* --- user defined strings and macros ------------------------------------ */
3907:
1.340 schwarze 3908: static int
1.106 kristaps 3909: roff_userdef(ROFF_ARGS)
1.99 kristaps 3910: {
1.339 schwarze 3911: struct mctx *ctx;
3912: char *arg, *ap, *dst, *src;
3913: size_t sz;
3914:
1.365 schwarze 3915: /* If the macro is empty, ignore it altogether. */
3916:
3917: if (*r->current_string == '\0')
3918: return ROFF_IGN;
3919:
1.339 schwarze 3920: /* Initialize a new macro stack context. */
3921:
3922: if (++r->mstackpos == r->mstacksz) {
3923: r->mstack = mandoc_recallocarray(r->mstack,
3924: r->mstacksz, r->mstacksz + 8, sizeof(*r->mstack));
3925: r->mstacksz += 8;
3926: }
3927: ctx = r->mstack + r->mstackpos;
3928: ctx->argsz = 0;
3929: ctx->argc = 0;
3930: ctx->argv = NULL;
3931:
3932: /*
3933: * Collect pointers to macro argument strings,
3934: * NUL-terminating them and escaping quotes.
3935: */
3936:
3937: src = buf->buf + pos;
3938: while (*src != '\0') {
3939: if (ctx->argc == ctx->argsz) {
3940: ctx->argsz += 8;
3941: ctx->argv = mandoc_reallocarray(ctx->argv,
3942: ctx->argsz, sizeof(*ctx->argv));
3943: }
1.355 schwarze 3944: arg = roff_getarg(r, &src, ln, &pos);
1.339 schwarze 3945: sz = 1; /* For the terminating NUL. */
3946: for (ap = arg; *ap != '\0'; ap++)
3947: sz += *ap == '"' ? 4 : 1;
3948: ctx->argv[ctx->argc++] = dst = mandoc_malloc(sz);
3949: for (ap = arg; *ap != '\0'; ap++) {
3950: if (*ap == '"') {
3951: memcpy(dst, "\\(dq", 4);
3952: dst += 4;
3953: } else
3954: *dst++ = *ap;
1.106 kristaps 3955: }
1.339 schwarze 3956: *dst = '\0';
1.355 schwarze 3957: free(arg);
1.337 schwarze 3958: }
3959:
1.339 schwarze 3960: /* Replace the macro invocation by the macro definition. */
1.263 schwarze 3961:
1.238 schwarze 3962: free(buf->buf);
1.339 schwarze 3963: buf->buf = mandoc_strdup(r->current_string);
3964: buf->sz = strlen(buf->buf) + 1;
1.248 schwarze 3965: *offs = 0;
1.106 kristaps 3966:
1.365 schwarze 3967: return buf->buf[buf->sz - 2] == '\n' ?
1.340 schwarze 3968: ROFF_REPARSE | ROFF_USERCALL : ROFF_IGN | ROFF_APPEND;
1.99 kristaps 3969: }
1.121 schwarze 3970:
1.306 schwarze 3971: /*
3972: * Calling a high-level macro that was renamed with .rn.
3973: * r->current_string has already been set up by roff_parse().
3974: */
1.340 schwarze 3975: static int
1.306 schwarze 3976: roff_renamed(ROFF_ARGS)
3977: {
3978: char *nbuf;
3979:
1.315 schwarze 3980: buf->sz = mandoc_asprintf(&nbuf, ".%s%s%s", r->current_string,
3981: buf->buf[pos] == '\0' ? "" : " ", buf->buf + pos) + 1;
1.306 schwarze 3982: free(buf->buf);
3983: buf->buf = nbuf;
1.329 schwarze 3984: *offs = 0;
1.306 schwarze 3985: return ROFF_CONT;
3986: }
3987:
1.362 schwarze 3988: /*
3989: * Measure the length in bytes of the roff identifier at *cpp
3990: * and advance the pointer to the next word.
3991: */
1.212 schwarze 3992: static size_t
1.121 schwarze 3993: roff_getname(struct roff *r, char **cpp, int ln, int pos)
3994: {
3995: char *name, *cp;
1.212 schwarze 3996: size_t namesz;
1.121 schwarze 3997:
3998: name = *cpp;
1.362 schwarze 3999: if (*name == '\0')
1.277 schwarze 4000: return 0;
1.121 schwarze 4001:
1.362 schwarze 4002: /* Advance cp to the byte after the end of the name. */
4003:
1.212 schwarze 4004: for (cp = name; 1; cp++) {
1.362 schwarze 4005: namesz = cp - name;
1.363 schwarze 4006: if (*cp == '\0')
1.212 schwarze 4007: break;
1.363 schwarze 4008: if (*cp == ' ' || *cp == '\t') {
4009: cp++;
4010: break;
4011: }
1.362 schwarze 4012: if (*cp != '\\')
1.121 schwarze 4013: continue;
1.362 schwarze 4014: if (cp[1] == '{' || cp[1] == '}')
1.215 schwarze 4015: break;
1.362 schwarze 4016: if (*++cp == '\\')
1.121 schwarze 4017: continue;
1.350 schwarze 4018: mandoc_msg(MANDOCERR_NAMESC, ln, pos,
1.224 schwarze 4019: "%.*s", (int)(cp - name + 1), name);
1.212 schwarze 4020: mandoc_escape((const char **)&cp, NULL, NULL);
4021: break;
1.121 schwarze 4022: }
4023:
4024: /* Read past spaces. */
1.362 schwarze 4025:
4026: while (*cp == ' ')
1.121 schwarze 4027: cp++;
4028:
4029: *cpp = cp;
1.277 schwarze 4030: return namesz;
1.121 schwarze 4031: }
4032:
1.106 kristaps 4033: /*
4034: * Store *string into the user-defined string called *name.
4035: * To clear an existing entry, call with (*r, *name, NULL, 0).
1.193 schwarze 4036: * append == 0: replace mode
4037: * append == 1: single-line append mode
4038: * append == 2: multiline append mode, append '\n' after each call
1.106 kristaps 4039: */
1.94 kristaps 4040: static void
1.106 kristaps 4041: roff_setstr(struct roff *r, const char *name, const char *string,
1.193 schwarze 4042: int append)
1.92 schwarze 4043: {
1.315 schwarze 4044: size_t namesz;
1.164 kristaps 4045:
1.315 schwarze 4046: namesz = strlen(name);
4047: roff_setstrn(&r->strtab, name, namesz, string,
1.207 schwarze 4048: string ? strlen(string) : 0, append);
1.315 schwarze 4049: roff_setstrn(&r->rentab, name, namesz, NULL, 0, 0);
1.164 kristaps 4050: }
4051:
4052: static void
1.166 kristaps 4053: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.193 schwarze 4054: const char *string, size_t stringsz, int append)
1.164 kristaps 4055: {
1.166 kristaps 4056: struct roffkv *n;
1.164 kristaps 4057: char *c;
4058: int i;
4059: size_t oldch, newch;
1.92 schwarze 4060:
1.106 kristaps 4061: /* Search for an existing string with the same name. */
1.164 kristaps 4062: n = *r;
4063:
1.211 schwarze 4064: while (n && (namesz != n->key.sz ||
4065: strncmp(n->key.p, name, namesz)))
1.92 schwarze 4066: n = n->next;
1.94 kristaps 4067:
4068: if (NULL == n) {
1.106 kristaps 4069: /* Create a new string table entry. */
1.166 kristaps 4070: n = mandoc_malloc(sizeof(struct roffkv));
4071: n->key.p = mandoc_strndup(name, namesz);
4072: n->key.sz = namesz;
4073: n->val.p = NULL;
4074: n->val.sz = 0;
1.164 kristaps 4075: n->next = *r;
4076: *r = n;
1.193 schwarze 4077: } else if (0 == append) {
1.166 kristaps 4078: free(n->val.p);
4079: n->val.p = NULL;
4080: n->val.sz = 0;
1.106 kristaps 4081: }
4082:
4083: if (NULL == string)
4084: return;
4085:
4086: /*
4087: * One additional byte for the '\n' in multiline mode,
4088: * and one for the terminating '\0'.
4089: */
1.193 schwarze 4090: newch = stringsz + (1 < append ? 2u : 1u);
1.164 kristaps 4091:
1.166 kristaps 4092: if (NULL == n->val.p) {
4093: n->val.p = mandoc_malloc(newch);
4094: *n->val.p = '\0';
1.106 kristaps 4095: oldch = 0;
4096: } else {
1.166 kristaps 4097: oldch = n->val.sz;
4098: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.106 kristaps 4099: }
4100:
4101: /* Skip existing content in the destination buffer. */
1.166 kristaps 4102: c = n->val.p + (int)oldch;
1.106 kristaps 4103:
4104: /* Append new content to the destination buffer. */
1.164 kristaps 4105: i = 0;
4106: while (i < (int)stringsz) {
1.106 kristaps 4107: /*
4108: * Rudimentary roff copy mode:
4109: * Handle escaped backslashes.
4110: */
1.164 kristaps 4111: if ('\\' == string[i] && '\\' == string[i + 1])
4112: i++;
4113: *c++ = string[i++];
1.106 kristaps 4114: }
1.94 kristaps 4115:
1.106 kristaps 4116: /* Append terminating bytes. */
1.193 schwarze 4117: if (1 < append)
1.106 kristaps 4118: *c++ = '\n';
1.163 kristaps 4119:
1.106 kristaps 4120: *c = '\0';
1.166 kristaps 4121: n->val.sz = (int)(c - n->val.p);
1.92 schwarze 4122: }
4123:
1.94 kristaps 4124: static const char *
1.325 schwarze 4125: roff_getstrn(struct roff *r, const char *name, size_t len,
1.315 schwarze 4126: int *deftype)
1.92 schwarze 4127: {
1.315 schwarze 4128: const struct roffkv *n;
1.325 schwarze 4129: int found, i;
1.315 schwarze 4130: enum roff_tok tok;
4131:
1.325 schwarze 4132: found = 0;
4133: for (n = r->strtab; n != NULL; n = n->next) {
4134: if (strncmp(name, n->key.p, len) != 0 ||
4135: n->key.p[len] != '\0' || n->val.p == NULL)
4136: continue;
4137: if (*deftype & ROFFDEF_USER) {
4138: *deftype = ROFFDEF_USER;
4139: return n->val.p;
4140: } else {
4141: found = 1;
4142: break;
4143: }
4144: }
4145: for (n = r->rentab; n != NULL; n = n->next) {
4146: if (strncmp(name, n->key.p, len) != 0 ||
4147: n->key.p[len] != '\0' || n->val.p == NULL)
4148: continue;
4149: if (*deftype & ROFFDEF_REN) {
4150: *deftype = ROFFDEF_REN;
4151: return n->val.p;
4152: } else {
4153: found = 1;
4154: break;
1.315 schwarze 4155: }
4156: }
1.325 schwarze 4157: for (i = 0; i < PREDEFS_MAX; i++) {
4158: if (strncmp(name, predefs[i].name, len) != 0 ||
4159: predefs[i].name[len] != '\0')
4160: continue;
4161: if (*deftype & ROFFDEF_PRE) {
4162: *deftype = ROFFDEF_PRE;
4163: return predefs[i].str;
4164: } else {
4165: found = 1;
4166: break;
1.315 schwarze 4167: }
4168: }
1.356 schwarze 4169: if (r->man->meta.macroset != MACROSET_MAN) {
1.325 schwarze 4170: for (tok = MDOC_Dd; tok < MDOC_MAX; tok++) {
4171: if (strncmp(name, roff_name[tok], len) != 0 ||
4172: roff_name[tok][len] != '\0')
4173: continue;
4174: if (*deftype & ROFFDEF_STD) {
4175: *deftype = ROFFDEF_STD;
4176: return NULL;
4177: } else {
4178: found = 1;
4179: break;
1.315 schwarze 4180: }
4181: }
4182: }
1.356 schwarze 4183: if (r->man->meta.macroset != MACROSET_MDOC) {
1.325 schwarze 4184: for (tok = MAN_TH; tok < MAN_MAX; tok++) {
4185: if (strncmp(name, roff_name[tok], len) != 0 ||
4186: roff_name[tok][len] != '\0')
4187: continue;
4188: if (*deftype & ROFFDEF_STD) {
4189: *deftype = ROFFDEF_STD;
4190: return NULL;
4191: } else {
4192: found = 1;
4193: break;
1.315 schwarze 4194: }
4195: }
1.325 schwarze 4196: }
4197:
4198: if (found == 0 && *deftype != ROFFDEF_ANY) {
4199: if (*deftype & ROFFDEF_REN) {
4200: /*
4201: * This might still be a request,
4202: * so do not treat it as undefined yet.
4203: */
4204: *deftype = ROFFDEF_UNDEF;
4205: return NULL;
1.315 schwarze 4206: }
1.325 schwarze 4207:
4208: /* Using an undefined string defines it to be empty. */
4209:
4210: roff_setstrn(&r->strtab, name, len, "", 0, 0);
4211: roff_setstrn(&r->rentab, name, len, NULL, 0, 0);
1.315 schwarze 4212: }
1.325 schwarze 4213:
1.315 schwarze 4214: *deftype = 0;
1.277 schwarze 4215: return NULL;
1.92 schwarze 4216: }
4217:
1.94 kristaps 4218: static void
1.167 kristaps 4219: roff_freestr(struct roffkv *r)
1.92 schwarze 4220: {
1.166 kristaps 4221: struct roffkv *n, *nn;
1.92 schwarze 4222:
1.167 kristaps 4223: for (n = r; n; n = nn) {
1.166 kristaps 4224: free(n->key.p);
4225: free(n->val.p);
1.92 schwarze 4226: nn = n->next;
4227: free(n);
4228: }
1.114 kristaps 4229: }
1.266 schwarze 4230:
4231: /* --- accessors and utility functions ------------------------------------ */
1.164 kristaps 4232:
4233: /*
4234: * Duplicate an input string, making the appropriate character
4235: * conversations (as stipulated by `tr') along the way.
4236: * Returns a heap-allocated string with all the replacements made.
4237: */
4238: char *
4239: roff_strdup(const struct roff *r, const char *p)
4240: {
1.166 kristaps 4241: const struct roffkv *cp;
1.164 kristaps 4242: char *res;
4243: const char *pp;
4244: size_t ssz, sz;
4245: enum mandoc_esc esc;
4246:
1.167 kristaps 4247: if (NULL == r->xmbtab && NULL == r->xtab)
1.277 schwarze 4248: return mandoc_strdup(p);
1.164 kristaps 4249: else if ('\0' == *p)
1.277 schwarze 4250: return mandoc_strdup("");
1.164 kristaps 4251:
4252: /*
4253: * Step through each character looking for term matches
4254: * (remember that a `tr' can be invoked with an escape, which is
4255: * a glyph but the escape is multi-character).
4256: * We only do this if the character hash has been initialised
4257: * and the string is >0 length.
4258: */
4259:
4260: res = NULL;
4261: ssz = 0;
4262:
4263: while ('\0' != *p) {
1.289 schwarze 4264: assert((unsigned int)*p < 128);
4265: if ('\\' != *p && r->xtab && r->xtab[(unsigned int)*p].p) {
1.167 kristaps 4266: sz = r->xtab[(int)*p].sz;
4267: res = mandoc_realloc(res, ssz + sz + 1);
4268: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
4269: ssz += sz;
4270: p++;
4271: continue;
4272: } else if ('\\' != *p) {
4273: res = mandoc_realloc(res, ssz + 2);
4274: res[ssz++] = *p++;
4275: continue;
4276: }
4277:
1.164 kristaps 4278: /* Search for term matches. */
1.167 kristaps 4279: for (cp = r->xmbtab; cp; cp = cp->next)
1.166 kristaps 4280: if (0 == strncmp(p, cp->key.p, cp->key.sz))
1.164 kristaps 4281: break;
4282:
4283: if (NULL != cp) {
4284: /*
4285: * A match has been found.
4286: * Append the match to the array and move
4287: * forward by its keysize.
4288: */
1.207 schwarze 4289: res = mandoc_realloc(res,
4290: ssz + cp->val.sz + 1);
1.166 kristaps 4291: memcpy(res + ssz, cp->val.p, cp->val.sz);
4292: ssz += cp->val.sz;
4293: p += (int)cp->key.sz;
1.164 kristaps 4294: continue;
4295: }
4296:
1.167 kristaps 4297: /*
4298: * Handle escapes carefully: we need to copy
4299: * over just the escape itself, or else we might
4300: * do replacements within the escape itself.
4301: * Make sure to pass along the bogus string.
4302: */
4303: pp = p++;
4304: esc = mandoc_escape(&p, NULL, NULL);
4305: if (ESCAPE_ERROR == esc) {
4306: sz = strlen(pp);
1.164 kristaps 4307: res = mandoc_realloc(res, ssz + sz + 1);
4308: memcpy(res + ssz, pp, sz);
1.167 kristaps 4309: break;
1.164 kristaps 4310: }
1.207 schwarze 4311: /*
4312: * We bail out on bad escapes.
1.167 kristaps 4313: * No need to warn: we already did so when
1.355 schwarze 4314: * roff_expand() was called.
1.167 kristaps 4315: */
4316: sz = (int)(p - pp);
4317: res = mandoc_realloc(res, ssz + sz + 1);
4318: memcpy(res + ssz, pp, sz);
4319: ssz += sz;
1.164 kristaps 4320: }
4321:
4322: res[(int)ssz] = '\0';
1.277 schwarze 4323: return res;
1.227 schwarze 4324: }
4325:
4326: int
4327: roff_getformat(const struct roff *r)
4328: {
4329:
1.277 schwarze 4330: return r->format;
1.174 kristaps 4331: }
4332:
4333: /*
1.207 schwarze 4334: * Find out whether a line is a macro line or not.
1.174 kristaps 4335: * If it is, adjust the current position and return one; if it isn't,
4336: * return zero and don't change the current position.
4337: * If the control character has been set with `.cc', then let that grain
4338: * precedence.
4339: * This is slighly contrary to groff, where using the non-breaking
4340: * control character when `cc' has been invoked will cause the
4341: * non-breaking macro contents to be printed verbatim.
4342: */
4343: int
4344: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
4345: {
4346: int pos;
4347:
4348: pos = *ppos;
4349:
1.303 schwarze 4350: if (r->control != '\0' && cp[pos] == r->control)
1.174 kristaps 4351: pos++;
1.303 schwarze 4352: else if (r->control != '\0')
1.277 schwarze 4353: return 0;
1.174 kristaps 4354: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
4355: pos += 2;
4356: else if ('.' == cp[pos] || '\'' == cp[pos])
4357: pos++;
4358: else
1.277 schwarze 4359: return 0;
1.174 kristaps 4360:
4361: while (' ' == cp[pos] || '\t' == cp[pos])
4362: pos++;
4363:
4364: *ppos = pos;
1.277 schwarze 4365: return 1;
1.74 kristaps 4366: }
CVSweb