Annotation of mandoc/roff.c, Revision 1.371
1.371 ! schwarze 1: /* $Id: roff.c,v 1.370 2020/02/27 01:43:52 schwarze Exp $ */
1.1 kristaps 2: /*
1.267 schwarze 3: * Copyright (c) 2008-2012, 2014 Kristaps Dzonsons <kristaps@bsd.lv>
1.369 schwarze 4: * Copyright (c) 2010-2015, 2017-2020 Ingo Schwarze <schwarze@openbsd.org>
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.
1.1 kristaps 17: */
1.66 kristaps 18: #include "config.h"
1.225 schwarze 19:
20: #include <sys/types.h>
1.30 kristaps 21:
1.67 kristaps 22: #include <assert.h>
1.85 kristaps 23: #include <ctype.h>
1.245 schwarze 24: #include <limits.h>
1.295 schwarze 25: #include <stddef.h>
26: #include <stdint.h>
1.178 schwarze 27: #include <stdio.h>
1.1 kristaps 28: #include <stdlib.h>
1.67 kristaps 29: #include <string.h>
1.1 kristaps 30:
1.201 schwarze 31: #include "mandoc_aux.h"
1.295 schwarze 32: #include "mandoc_ohash.h"
1.345 schwarze 33: #include "mandoc.h"
1.265 schwarze 34: #include "roff.h"
1.349 schwarze 35: #include "mandoc_parse.h"
1.226 schwarze 36: #include "libmandoc.h"
1.266 schwarze 37: #include "roff_int.h"
1.346 schwarze 38: #include "tbl_parse.h"
1.347 schwarze 39: #include "eqn_parse.h"
1.82 kristaps 40:
1.355 schwarze 41: /*
42: * ASCII_ESC is used to signal from roff_getarg() to roff_expand()
43: * that an escape sequence resulted from copy-in processing and
44: * needs to be checked or interpolated. As it is used nowhere
45: * else, it is defined here rather than in a header file.
46: */
47: #define ASCII_ESC 27
48:
1.170 schwarze 49: /* Maximum number of string expansions per line, to break infinite loops. */
50: #define EXPAND_LIMIT 1000
51:
1.315 schwarze 52: /* Types of definitions of macros and strings. */
53: #define ROFFDEF_USER (1 << 1) /* User-defined. */
54: #define ROFFDEF_PRE (1 << 2) /* Predefined. */
55: #define ROFFDEF_REN (1 << 3) /* Renamed standard macro. */
56: #define ROFFDEF_STD (1 << 4) /* mdoc(7) or man(7) macro. */
57: #define ROFFDEF_ANY (ROFFDEF_USER | ROFFDEF_PRE | \
58: ROFFDEF_REN | ROFFDEF_STD)
1.325 schwarze 59: #define ROFFDEF_UNDEF (1 << 5) /* Completely undefined. */
1.315 schwarze 60:
1.266 schwarze 61: /* --- data types --------------------------------------------------------- */
62:
1.147 kristaps 63: /*
1.167 kristaps 64: * An incredibly-simple string buffer.
65: */
1.94 kristaps 66: struct roffstr {
1.167 kristaps 67: char *p; /* nil-terminated buffer */
68: size_t sz; /* saved strlen(p) */
1.166 kristaps 69: };
70:
71: /*
1.167 kristaps 72: * A key-value roffstr pair as part of a singly-linked list.
1.166 kristaps 73: */
74: struct roffkv {
75: struct roffstr key;
76: struct roffstr val;
77: struct roffkv *next; /* next in list */
1.94 kristaps 78: };
79:
1.180 schwarze 80: /*
81: * A single number register as part of a singly-linked list.
82: */
83: struct roffreg {
84: struct roffstr key;
1.181 schwarze 85: int val;
1.327 schwarze 86: int step;
1.180 schwarze 87: struct roffreg *next;
88: };
89:
1.295 schwarze 90: /*
91: * Association of request and macro names with token IDs.
92: */
93: struct roffreq {
94: enum roff_tok tok;
95: char name[];
96: };
97:
1.339 schwarze 98: /*
99: * A macro processing context.
100: * More than one is needed when macro calls are nested.
101: */
102: struct mctx {
103: char **argv;
104: int argc;
105: int argsz;
106: };
107:
1.67 kristaps 108: struct roff {
1.296 schwarze 109: struct roff_man *man; /* mdoc or man parser */
1.67 kristaps 110: struct roffnode *last; /* leaf of stack */
1.339 schwarze 111: struct mctx *mstack; /* stack of macro contexts */
1.223 schwarze 112: int *rstack; /* stack of inverted `ie' values */
1.295 schwarze 113: struct ohash *reqtab; /* request lookup table */
1.180 schwarze 114: struct roffreg *regtab; /* number registers */
1.166 kristaps 115: struct roffkv *strtab; /* user-defined strings & macros */
1.306 schwarze 116: struct roffkv *rentab; /* renamed strings & macros */
1.167 kristaps 117: struct roffkv *xmbtab; /* multi-byte trans table (`tr') */
118: struct roffstr *xtab; /* single-byte trans table (`tr') */
1.106 kristaps 119: const char *current_string; /* value of last called user macro */
1.118 kristaps 120: struct tbl_node *first_tbl; /* first table parsed */
121: struct tbl_node *last_tbl; /* last table parsed */
122: struct tbl_node *tbl; /* current table being parsed */
1.319 schwarze 123: struct eqn_node *last_eqn; /* equation parser */
124: struct eqn_node *eqn; /* active equation parser */
1.230 schwarze 125: int eqn_inline; /* current equation is inline */
1.223 schwarze 126: int options; /* parse options */
1.339 schwarze 127: int mstacksz; /* current size of mstack */
128: int mstackpos; /* position in mstack */
1.223 schwarze 129: int rstacksz; /* current size limit of rstack */
130: int rstackpos; /* position in rstack */
1.227 schwarze 131: int format; /* current file in mdoc or man format */
1.223 schwarze 132: char control; /* control character */
1.303 schwarze 133: char escape; /* escape character */
1.79 kristaps 134: };
135:
1.364 schwarze 136: /*
137: * A macro definition, condition, or ignored block.
138: */
1.67 kristaps 139: struct roffnode {
1.294 schwarze 140: enum roff_tok tok; /* type of node */
1.67 kristaps 141: struct roffnode *parent; /* up one in stack */
142: int line; /* parse line */
143: int col; /* parse col */
1.106 kristaps 144: char *name; /* node name, e.g. macro name */
1.364 schwarze 145: char *end; /* custom end macro of the block */
146: int endspan; /* scope to: 1=eol 2=next line -1=\} */
147: int rule; /* content is: 1=evaluated 0=skipped */
1.67 kristaps 148: };
149:
150: #define ROFF_ARGS struct roff *r, /* parse ctx */ \
1.294 schwarze 151: enum roff_tok tok, /* tok of macro */ \
1.238 schwarze 152: struct buf *buf, /* input buffer */ \
1.67 kristaps 153: int ln, /* parse line */ \
1.75 kristaps 154: int ppos, /* original pos in buffer */ \
155: int pos, /* current pos in buffer */ \
1.74 kristaps 156: int *offs /* reset offset of buffer data */
1.67 kristaps 157:
1.340 schwarze 158: typedef int (*roffproc)(ROFF_ARGS);
1.67 kristaps 159:
160: struct roffmac {
1.79 kristaps 161: roffproc proc; /* process new macro */
162: roffproc text; /* process as child text of macro */
163: roffproc sub; /* process as child of macro */
164: int flags;
165: #define ROFFMAC_STRUCT (1 << 0) /* always interpret */
1.67 kristaps 166: };
167:
1.141 kristaps 168: struct predef {
169: const char *name; /* predefined input name */
170: const char *str; /* replacement symbol */
171: };
172:
173: #define PREDEF(__name, __str) \
174: { (__name), (__str) },
175:
1.266 schwarze 176: /* --- function prototypes ------------------------------------------------ */
177:
1.340 schwarze 178: static int roffnode_cleanscope(struct roff *);
179: static int roffnode_pop(struct roff *);
1.294 schwarze 180: static void roffnode_push(struct roff *, enum roff_tok,
1.155 kristaps 181: const char *, int, int);
1.346 schwarze 182: static void roff_addtbl(struct roff_man *, int, struct tbl_node *);
1.340 schwarze 183: static int roff_als(ROFF_ARGS);
184: static int roff_block(ROFF_ARGS);
185: static int roff_block_text(ROFF_ARGS);
186: static int roff_block_sub(ROFF_ARGS);
1.364 schwarze 187: static int roff_break(ROFF_ARGS);
1.340 schwarze 188: static int roff_cblock(ROFF_ARGS);
189: static int roff_cc(ROFF_ARGS);
190: static int roff_ccond(struct roff *, int, int);
1.341 schwarze 191: static int roff_char(ROFF_ARGS);
1.340 schwarze 192: static int roff_cond(ROFF_ARGS);
193: static int roff_cond_text(ROFF_ARGS);
194: static int roff_cond_sub(ROFF_ARGS);
195: static int roff_ds(ROFF_ARGS);
196: static int roff_ec(ROFF_ARGS);
197: static int roff_eo(ROFF_ARGS);
198: static int roff_eqndelim(struct roff *, struct buf *, int);
1.271 schwarze 199: static int roff_evalcond(struct roff *r, int, char *, int *);
1.234 kristaps 200: static int roff_evalnum(struct roff *, int,
201: const char *, int *, int *, int);
202: static int roff_evalpar(struct roff *, int,
1.261 schwarze 203: const char *, int *, int *, int);
1.198 schwarze 204: static int roff_evalstrcond(const char *, int *);
1.355 schwarze 205: static int roff_expand(struct roff *, struct buf *,
206: int, int, char);
1.155 kristaps 207: static void roff_free1(struct roff *);
1.180 schwarze 208: static void roff_freereg(struct roffreg *);
1.167 kristaps 209: static void roff_freestr(struct roffkv *);
1.212 schwarze 210: static size_t roff_getname(struct roff *, char **, int, int);
1.261 schwarze 211: static int roff_getnum(const char *, int *, int *, int);
1.184 schwarze 212: static int roff_getop(const char *, int *, char *);
1.327 schwarze 213: static int roff_getregn(struct roff *,
214: const char *, size_t, char);
1.273 schwarze 215: static int roff_getregro(const struct roff *,
216: const char *name);
1.325 schwarze 217: static const char *roff_getstrn(struct roff *,
1.315 schwarze 218: const char *, size_t, int *);
1.271 schwarze 219: static int roff_hasregn(const struct roff *,
220: const char *, size_t);
1.340 schwarze 221: static int roff_insec(ROFF_ARGS);
222: static int roff_it(ROFF_ARGS);
223: static int roff_line_ignore(ROFF_ARGS);
1.265 schwarze 224: static void roff_man_alloc1(struct roff_man *);
225: static void roff_man_free1(struct roff_man *);
1.340 schwarze 226: static int roff_manyarg(ROFF_ARGS);
1.358 schwarze 227: static int roff_noarg(ROFF_ARGS);
1.340 schwarze 228: static int roff_nop(ROFF_ARGS);
229: static int roff_nr(ROFF_ARGS);
230: static int roff_onearg(ROFF_ARGS);
1.294 schwarze 231: static enum roff_tok roff_parse(struct roff *, char *, int *,
1.214 schwarze 232: int, int);
1.340 schwarze 233: static int roff_parsetext(struct roff *, struct buf *,
1.305 schwarze 234: int, int *);
1.340 schwarze 235: static int roff_renamed(ROFF_ARGS);
236: static int roff_return(ROFF_ARGS);
237: static int roff_rm(ROFF_ARGS);
238: static int roff_rn(ROFF_ARGS);
239: static int roff_rr(ROFF_ARGS);
1.326 schwarze 240: static void roff_setregn(struct roff *, const char *,
1.327 schwarze 241: size_t, int, char, int);
1.94 kristaps 242: static void roff_setstr(struct roff *,
1.106 kristaps 243: const char *, const char *, int);
1.207 schwarze 244: static void roff_setstrn(struct roffkv **, const char *,
1.164 kristaps 245: size_t, const char *, size_t, int);
1.340 schwarze 246: static int roff_shift(ROFF_ARGS);
247: static int roff_so(ROFF_ARGS);
248: static int roff_tr(ROFF_ARGS);
249: static int roff_Dd(ROFF_ARGS);
250: static int roff_TE(ROFF_ARGS);
251: static int roff_TS(ROFF_ARGS);
252: static int roff_EQ(ROFF_ARGS);
253: static int roff_EN(ROFF_ARGS);
254: static int roff_T_(ROFF_ARGS);
255: static int roff_unsupp(ROFF_ARGS);
256: static int roff_userdef(ROFF_ARGS);
1.67 kristaps 257:
1.266 schwarze 258: /* --- constant data ------------------------------------------------------ */
259:
1.261 schwarze 260: #define ROFFNUM_SCALE (1 << 0) /* Honour scaling in roff_getnum(). */
261: #define ROFFNUM_WHITE (1 << 1) /* Skip whitespace in roff_evalnum(). */
262:
1.294 schwarze 263: const char *__roff_name[MAN_MAX + 1] = {
1.358 schwarze 264: "br", "ce", "fi", "ft",
265: "ll", "mc", "nf",
266: "po", "rj", "sp",
1.312 schwarze 267: "ta", "ti", NULL,
1.294 schwarze 268: "ab", "ad", "af", "aln",
269: "als", "am", "am1", "ami",
270: "ami1", "as", "as1", "asciify",
271: "backtrace", "bd", "bleedat", "blm",
272: "box", "boxa", "bp", "BP",
273: "break", "breakchar", "brnl", "brp",
1.305 schwarze 274: "brpnl", "c2", "cc",
1.294 schwarze 275: "cf", "cflags", "ch", "char",
276: "chop", "class", "close", "CL",
277: "color", "composite", "continue", "cp",
278: "cropat", "cs", "cu", "da",
279: "dch", "Dd", "de", "de1",
280: "defcolor", "dei", "dei1", "device",
281: "devicem", "di", "do", "ds",
282: "ds1", "dwh", "dt", "ec",
283: "ecr", "ecs", "el", "em",
284: "EN", "eo", "EP", "EQ",
285: "errprint", "ev", "evc", "ex",
286: "fallback", "fam", "fc", "fchar",
287: "fcolor", "fdeferlig", "feature", "fkern",
288: "fl", "flig", "fp", "fps",
289: "fschar", "fspacewidth", "fspecial", "ftr",
290: "fzoom", "gcolor", "hc", "hcode",
291: "hidechar", "hla", "hlm", "hpf",
292: "hpfa", "hpfcode", "hw", "hy",
293: "hylang", "hylen", "hym", "hypp",
294: "hys", "ie", "if", "ig",
295: "index", "it", "itc", "IX",
296: "kern", "kernafter", "kernbefore", "kernpair",
297: "lc", "lc_ctype", "lds", "length",
298: "letadj", "lf", "lg", "lhang",
299: "linetabs", "lnr", "lnrf", "lpfx",
1.304 schwarze 300: "ls", "lsm", "lt",
1.294 schwarze 301: "mediasize", "minss", "mk", "mso",
302: "na", "ne", "nh", "nhychar",
303: "nm", "nn", "nop", "nr",
304: "nrf", "nroff", "ns", "nx",
305: "open", "opena", "os", "output",
306: "padj", "papersize", "pc", "pev",
307: "pi", "PI", "pl", "pm",
1.312 schwarze 308: "pn", "pnr", "ps",
1.294 schwarze 309: "psbb", "pshape", "pso", "ptr",
310: "pvs", "rchar", "rd", "recursionlimit",
1.309 schwarze 311: "return", "rfschar", "rhang",
1.294 schwarze 312: "rm", "rn", "rnn", "rr",
313: "rs", "rt", "schar", "sentchar",
314: "shc", "shift", "sizes", "so",
315: "spacewidth", "special", "spreadwarn", "ss",
316: "sty", "substring", "sv", "sy",
1.300 schwarze 317: "T&", "tc", "TE",
1.301 schwarze 318: "TH", "tkf", "tl",
1.294 schwarze 319: "tm", "tm1", "tmc", "tr",
320: "track", "transchar", "trf", "trimat",
321: "trin", "trnt", "troff", "TS",
322: "uf", "ul", "unformat", "unwatch",
323: "unwatchn", "vpt", "vs", "warn",
324: "warnscale", "watch", "watchlength", "watchn",
325: "wh", "while", "write", "writec",
326: "writem", "xflag", ".", NULL,
1.306 schwarze 327: NULL, "text",
1.294 schwarze 328: "Dd", "Dt", "Os", "Sh",
329: "Ss", "Pp", "D1", "Dl",
330: "Bd", "Ed", "Bl", "El",
331: "It", "Ad", "An", "Ap",
332: "Ar", "Cd", "Cm", "Dv",
333: "Er", "Ev", "Ex", "Fa",
334: "Fd", "Fl", "Fn", "Ft",
335: "Ic", "In", "Li", "Nd",
336: "Nm", "Op", "Ot", "Pa",
337: "Rv", "St", "Va", "Vt",
338: "Xr", "%A", "%B", "%D",
339: "%I", "%J", "%N", "%O",
340: "%P", "%R", "%T", "%V",
341: "Ac", "Ao", "Aq", "At",
342: "Bc", "Bf", "Bo", "Bq",
343: "Bsx", "Bx", "Db", "Dc",
344: "Do", "Dq", "Ec", "Ef",
345: "Em", "Eo", "Fx", "Ms",
346: "No", "Ns", "Nx", "Ox",
347: "Pc", "Pf", "Po", "Pq",
348: "Qc", "Ql", "Qo", "Qq",
349: "Re", "Rs", "Sc", "So",
350: "Sq", "Sm", "Sx", "Sy",
351: "Tn", "Ux", "Xc", "Xo",
352: "Fo", "Fc", "Oo", "Oc",
353: "Bk", "Ek", "Bt", "Hf",
354: "Fr", "Ud", "Lb", "Lp",
355: "Lk", "Mt", "Brq", "Bro",
356: "Brc", "%C", "Es", "En",
1.299 schwarze 357: "Dx", "%Q", "%U", "Ta",
1.369 schwarze 358: "Tg", NULL,
1.294 schwarze 359: "TH", "SH", "SS", "TP",
1.332 schwarze 360: "TQ",
1.294 schwarze 361: "LP", "PP", "P", "IP",
362: "HP", "SM", "SB", "BI",
363: "IB", "BR", "RB", "R",
364: "B", "I", "IR", "RI",
365: "RE", "RS", "DT", "UC",
1.297 schwarze 366: "PD", "AT", "in",
1.333 schwarze 367: "SY", "YS", "OP",
368: "EX", "EE", "UR",
1.317 schwarze 369: "UE", "MT", "ME", NULL
1.294 schwarze 370: };
371: const char *const *roff_name = __roff_name;
372:
373: static struct roffmac roffs[TOKEN_NONE] = {
1.358 schwarze 374: { roff_noarg, NULL, NULL, 0 }, /* br */
1.305 schwarze 375: { roff_onearg, NULL, NULL, 0 }, /* ce */
1.358 schwarze 376: { roff_noarg, NULL, NULL, 0 }, /* fi */
1.297 schwarze 377: { roff_onearg, NULL, NULL, 0 }, /* ft */
1.298 schwarze 378: { roff_onearg, NULL, NULL, 0 }, /* ll */
1.304 schwarze 379: { roff_onearg, NULL, NULL, 0 }, /* mc */
1.358 schwarze 380: { roff_noarg, NULL, NULL, 0 }, /* nf */
1.312 schwarze 381: { roff_onearg, NULL, NULL, 0 }, /* po */
1.309 schwarze 382: { roff_onearg, NULL, NULL, 0 }, /* rj */
1.299 schwarze 383: { roff_onearg, NULL, NULL, 0 }, /* sp */
1.300 schwarze 384: { roff_manyarg, NULL, NULL, 0 }, /* ta */
1.301 schwarze 385: { roff_onearg, NULL, NULL, 0 }, /* ti */
1.296 schwarze 386: { NULL, NULL, NULL, 0 }, /* ROFF_MAX */
1.295 schwarze 387: { roff_unsupp, NULL, NULL, 0 }, /* ab */
388: { roff_line_ignore, NULL, NULL, 0 }, /* ad */
389: { roff_line_ignore, NULL, NULL, 0 }, /* af */
390: { roff_unsupp, NULL, NULL, 0 }, /* aln */
1.311 schwarze 391: { roff_als, NULL, NULL, 0 }, /* als */
1.295 schwarze 392: { roff_block, roff_block_text, roff_block_sub, 0 }, /* am */
393: { roff_block, roff_block_text, roff_block_sub, 0 }, /* am1 */
394: { roff_block, roff_block_text, roff_block_sub, 0 }, /* ami */
395: { roff_block, roff_block_text, roff_block_sub, 0 }, /* ami1 */
396: { roff_ds, NULL, NULL, 0 }, /* as */
397: { roff_ds, NULL, NULL, 0 }, /* as1 */
398: { roff_unsupp, NULL, NULL, 0 }, /* asciify */
399: { roff_line_ignore, NULL, NULL, 0 }, /* backtrace */
400: { roff_line_ignore, NULL, NULL, 0 }, /* bd */
401: { roff_line_ignore, NULL, NULL, 0 }, /* bleedat */
402: { roff_unsupp, NULL, NULL, 0 }, /* blm */
403: { roff_unsupp, NULL, NULL, 0 }, /* box */
404: { roff_unsupp, NULL, NULL, 0 }, /* boxa */
405: { roff_line_ignore, NULL, NULL, 0 }, /* bp */
406: { roff_unsupp, NULL, NULL, 0 }, /* BP */
1.364 schwarze 407: { roff_break, NULL, NULL, 0 }, /* break */
1.295 schwarze 408: { roff_line_ignore, NULL, NULL, 0 }, /* breakchar */
409: { roff_line_ignore, NULL, NULL, 0 }, /* brnl */
1.358 schwarze 410: { roff_noarg, NULL, NULL, 0 }, /* brp */
1.295 schwarze 411: { roff_line_ignore, NULL, NULL, 0 }, /* brpnl */
412: { roff_unsupp, NULL, NULL, 0 }, /* c2 */
413: { roff_cc, NULL, NULL, 0 }, /* cc */
414: { roff_insec, NULL, NULL, 0 }, /* cf */
415: { roff_line_ignore, NULL, NULL, 0 }, /* cflags */
416: { roff_line_ignore, NULL, NULL, 0 }, /* ch */
1.341 schwarze 417: { roff_char, NULL, NULL, 0 }, /* char */
1.295 schwarze 418: { roff_unsupp, NULL, NULL, 0 }, /* chop */
419: { roff_line_ignore, NULL, NULL, 0 }, /* class */
420: { roff_insec, NULL, NULL, 0 }, /* close */
421: { roff_unsupp, NULL, NULL, 0 }, /* CL */
422: { roff_line_ignore, NULL, NULL, 0 }, /* color */
423: { roff_unsupp, NULL, NULL, 0 }, /* composite */
424: { roff_unsupp, NULL, NULL, 0 }, /* continue */
425: { roff_line_ignore, NULL, NULL, 0 }, /* cp */
426: { roff_line_ignore, NULL, NULL, 0 }, /* cropat */
427: { roff_line_ignore, NULL, NULL, 0 }, /* cs */
428: { roff_line_ignore, NULL, NULL, 0 }, /* cu */
429: { roff_unsupp, NULL, NULL, 0 }, /* da */
430: { roff_unsupp, NULL, NULL, 0 }, /* dch */
431: { roff_Dd, NULL, NULL, 0 }, /* Dd */
432: { roff_block, roff_block_text, roff_block_sub, 0 }, /* de */
433: { roff_block, roff_block_text, roff_block_sub, 0 }, /* de1 */
434: { roff_line_ignore, NULL, NULL, 0 }, /* defcolor */
435: { roff_block, roff_block_text, roff_block_sub, 0 }, /* dei */
436: { roff_block, roff_block_text, roff_block_sub, 0 }, /* dei1 */
437: { roff_unsupp, NULL, NULL, 0 }, /* device */
438: { roff_unsupp, NULL, NULL, 0 }, /* devicem */
439: { roff_unsupp, NULL, NULL, 0 }, /* di */
440: { roff_unsupp, NULL, NULL, 0 }, /* do */
441: { roff_ds, NULL, NULL, 0 }, /* ds */
442: { roff_ds, NULL, NULL, 0 }, /* ds1 */
443: { roff_unsupp, NULL, NULL, 0 }, /* dwh */
444: { roff_unsupp, NULL, NULL, 0 }, /* dt */
1.303 schwarze 445: { roff_ec, NULL, NULL, 0 }, /* ec */
1.295 schwarze 446: { roff_unsupp, NULL, NULL, 0 }, /* ecr */
447: { roff_unsupp, NULL, NULL, 0 }, /* ecs */
448: { roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT }, /* el */
449: { roff_unsupp, NULL, NULL, 0 }, /* em */
450: { roff_EN, NULL, NULL, 0 }, /* EN */
1.303 schwarze 451: { roff_eo, NULL, NULL, 0 }, /* eo */
1.295 schwarze 452: { roff_unsupp, NULL, NULL, 0 }, /* EP */
453: { roff_EQ, NULL, NULL, 0 }, /* EQ */
454: { roff_line_ignore, NULL, NULL, 0 }, /* errprint */
455: { roff_unsupp, NULL, NULL, 0 }, /* ev */
456: { roff_unsupp, NULL, NULL, 0 }, /* evc */
457: { roff_unsupp, NULL, NULL, 0 }, /* ex */
458: { roff_line_ignore, NULL, NULL, 0 }, /* fallback */
459: { roff_line_ignore, NULL, NULL, 0 }, /* fam */
460: { roff_unsupp, NULL, NULL, 0 }, /* fc */
461: { roff_unsupp, NULL, NULL, 0 }, /* fchar */
462: { roff_line_ignore, NULL, NULL, 0 }, /* fcolor */
463: { roff_line_ignore, NULL, NULL, 0 }, /* fdeferlig */
464: { roff_line_ignore, NULL, NULL, 0 }, /* feature */
465: { roff_line_ignore, NULL, NULL, 0 }, /* fkern */
466: { roff_line_ignore, NULL, NULL, 0 }, /* fl */
467: { roff_line_ignore, NULL, NULL, 0 }, /* flig */
468: { roff_line_ignore, NULL, NULL, 0 }, /* fp */
469: { roff_line_ignore, NULL, NULL, 0 }, /* fps */
470: { roff_unsupp, NULL, NULL, 0 }, /* fschar */
471: { roff_line_ignore, NULL, NULL, 0 }, /* fspacewidth */
472: { roff_line_ignore, NULL, NULL, 0 }, /* fspecial */
473: { roff_line_ignore, NULL, NULL, 0 }, /* ftr */
474: { roff_line_ignore, NULL, NULL, 0 }, /* fzoom */
475: { roff_line_ignore, NULL, NULL, 0 }, /* gcolor */
476: { roff_line_ignore, NULL, NULL, 0 }, /* hc */
477: { roff_line_ignore, NULL, NULL, 0 }, /* hcode */
478: { roff_line_ignore, NULL, NULL, 0 }, /* hidechar */
479: { roff_line_ignore, NULL, NULL, 0 }, /* hla */
480: { roff_line_ignore, NULL, NULL, 0 }, /* hlm */
481: { roff_line_ignore, NULL, NULL, 0 }, /* hpf */
482: { roff_line_ignore, NULL, NULL, 0 }, /* hpfa */
483: { roff_line_ignore, NULL, NULL, 0 }, /* hpfcode */
484: { roff_line_ignore, NULL, NULL, 0 }, /* hw */
485: { roff_line_ignore, NULL, NULL, 0 }, /* hy */
486: { roff_line_ignore, NULL, NULL, 0 }, /* hylang */
487: { roff_line_ignore, NULL, NULL, 0 }, /* hylen */
488: { roff_line_ignore, NULL, NULL, 0 }, /* hym */
489: { roff_line_ignore, NULL, NULL, 0 }, /* hypp */
490: { roff_line_ignore, NULL, NULL, 0 }, /* hys */
491: { roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT }, /* ie */
492: { roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT }, /* if */
493: { roff_block, roff_block_text, roff_block_sub, 0 }, /* ig */
494: { roff_unsupp, NULL, NULL, 0 }, /* index */
495: { roff_it, NULL, NULL, 0 }, /* it */
496: { roff_unsupp, NULL, NULL, 0 }, /* itc */
497: { roff_line_ignore, NULL, NULL, 0 }, /* IX */
498: { roff_line_ignore, NULL, NULL, 0 }, /* kern */
499: { roff_line_ignore, NULL, NULL, 0 }, /* kernafter */
500: { roff_line_ignore, NULL, NULL, 0 }, /* kernbefore */
501: { roff_line_ignore, NULL, NULL, 0 }, /* kernpair */
502: { roff_unsupp, NULL, NULL, 0 }, /* lc */
503: { roff_unsupp, NULL, NULL, 0 }, /* lc_ctype */
504: { roff_unsupp, NULL, NULL, 0 }, /* lds */
505: { roff_unsupp, NULL, NULL, 0 }, /* length */
506: { roff_line_ignore, NULL, NULL, 0 }, /* letadj */
507: { roff_insec, NULL, NULL, 0 }, /* lf */
508: { roff_line_ignore, NULL, NULL, 0 }, /* lg */
509: { roff_line_ignore, NULL, NULL, 0 }, /* lhang */
510: { roff_unsupp, NULL, NULL, 0 }, /* linetabs */
511: { roff_unsupp, NULL, NULL, 0 }, /* lnr */
512: { roff_unsupp, NULL, NULL, 0 }, /* lnrf */
513: { roff_unsupp, NULL, NULL, 0 }, /* lpfx */
514: { roff_line_ignore, NULL, NULL, 0 }, /* ls */
515: { roff_unsupp, NULL, NULL, 0 }, /* lsm */
516: { roff_line_ignore, NULL, NULL, 0 }, /* lt */
517: { roff_line_ignore, NULL, NULL, 0 }, /* mediasize */
518: { roff_line_ignore, NULL, NULL, 0 }, /* minss */
519: { roff_line_ignore, NULL, NULL, 0 }, /* mk */
520: { roff_insec, NULL, NULL, 0 }, /* mso */
521: { roff_line_ignore, NULL, NULL, 0 }, /* na */
522: { roff_line_ignore, NULL, NULL, 0 }, /* ne */
523: { roff_line_ignore, NULL, NULL, 0 }, /* nh */
524: { roff_line_ignore, NULL, NULL, 0 }, /* nhychar */
525: { roff_unsupp, NULL, NULL, 0 }, /* nm */
526: { roff_unsupp, NULL, NULL, 0 }, /* nn */
1.330 schwarze 527: { roff_nop, NULL, NULL, 0 }, /* nop */
1.295 schwarze 528: { roff_nr, NULL, NULL, 0 }, /* nr */
529: { roff_unsupp, NULL, NULL, 0 }, /* nrf */
530: { roff_line_ignore, NULL, NULL, 0 }, /* nroff */
531: { roff_line_ignore, NULL, NULL, 0 }, /* ns */
532: { roff_insec, NULL, NULL, 0 }, /* nx */
533: { roff_insec, NULL, NULL, 0 }, /* open */
534: { roff_insec, NULL, NULL, 0 }, /* opena */
535: { roff_line_ignore, NULL, NULL, 0 }, /* os */
536: { roff_unsupp, NULL, NULL, 0 }, /* output */
537: { roff_line_ignore, NULL, NULL, 0 }, /* padj */
538: { roff_line_ignore, NULL, NULL, 0 }, /* papersize */
539: { roff_line_ignore, NULL, NULL, 0 }, /* pc */
540: { roff_line_ignore, NULL, NULL, 0 }, /* pev */
541: { roff_insec, NULL, NULL, 0 }, /* pi */
542: { roff_unsupp, NULL, NULL, 0 }, /* PI */
543: { roff_line_ignore, NULL, NULL, 0 }, /* pl */
544: { roff_line_ignore, NULL, NULL, 0 }, /* pm */
545: { roff_line_ignore, NULL, NULL, 0 }, /* pn */
546: { roff_line_ignore, NULL, NULL, 0 }, /* pnr */
547: { roff_line_ignore, NULL, NULL, 0 }, /* ps */
548: { roff_unsupp, NULL, NULL, 0 }, /* psbb */
549: { roff_unsupp, NULL, NULL, 0 }, /* pshape */
550: { roff_insec, NULL, NULL, 0 }, /* pso */
551: { roff_line_ignore, NULL, NULL, 0 }, /* ptr */
552: { roff_line_ignore, NULL, NULL, 0 }, /* pvs */
553: { roff_unsupp, NULL, NULL, 0 }, /* rchar */
554: { roff_line_ignore, NULL, NULL, 0 }, /* rd */
555: { roff_line_ignore, NULL, NULL, 0 }, /* recursionlimit */
1.339 schwarze 556: { roff_return, NULL, NULL, 0 }, /* return */
1.295 schwarze 557: { roff_unsupp, NULL, NULL, 0 }, /* rfschar */
558: { roff_line_ignore, NULL, NULL, 0 }, /* rhang */
559: { roff_rm, NULL, NULL, 0 }, /* rm */
1.306 schwarze 560: { roff_rn, NULL, NULL, 0 }, /* rn */
1.295 schwarze 561: { roff_unsupp, NULL, NULL, 0 }, /* rnn */
562: { roff_rr, NULL, NULL, 0 }, /* rr */
563: { roff_line_ignore, NULL, NULL, 0 }, /* rs */
564: { roff_line_ignore, NULL, NULL, 0 }, /* rt */
565: { roff_unsupp, NULL, NULL, 0 }, /* schar */
566: { roff_line_ignore, NULL, NULL, 0 }, /* sentchar */
567: { roff_line_ignore, NULL, NULL, 0 }, /* shc */
1.339 schwarze 568: { roff_shift, NULL, NULL, 0 }, /* shift */
1.295 schwarze 569: { roff_line_ignore, NULL, NULL, 0 }, /* sizes */
570: { roff_so, NULL, NULL, 0 }, /* so */
571: { roff_line_ignore, NULL, NULL, 0 }, /* spacewidth */
572: { roff_line_ignore, NULL, NULL, 0 }, /* special */
573: { roff_line_ignore, NULL, NULL, 0 }, /* spreadwarn */
574: { roff_line_ignore, NULL, NULL, 0 }, /* ss */
575: { roff_line_ignore, NULL, NULL, 0 }, /* sty */
576: { roff_unsupp, NULL, NULL, 0 }, /* substring */
577: { roff_line_ignore, NULL, NULL, 0 }, /* sv */
578: { roff_insec, NULL, NULL, 0 }, /* sy */
579: { roff_T_, NULL, NULL, 0 }, /* T& */
580: { roff_unsupp, NULL, NULL, 0 }, /* tc */
581: { roff_TE, NULL, NULL, 0 }, /* TE */
1.315 schwarze 582: { roff_Dd, NULL, NULL, 0 }, /* TH */
1.295 schwarze 583: { roff_line_ignore, NULL, NULL, 0 }, /* tkf */
584: { roff_unsupp, NULL, NULL, 0 }, /* tl */
585: { roff_line_ignore, NULL, NULL, 0 }, /* tm */
586: { roff_line_ignore, NULL, NULL, 0 }, /* tm1 */
587: { roff_line_ignore, NULL, NULL, 0 }, /* tmc */
588: { roff_tr, NULL, NULL, 0 }, /* tr */
589: { roff_line_ignore, NULL, NULL, 0 }, /* track */
590: { roff_line_ignore, NULL, NULL, 0 }, /* transchar */
591: { roff_insec, NULL, NULL, 0 }, /* trf */
592: { roff_line_ignore, NULL, NULL, 0 }, /* trimat */
593: { roff_unsupp, NULL, NULL, 0 }, /* trin */
594: { roff_unsupp, NULL, NULL, 0 }, /* trnt */
595: { roff_line_ignore, NULL, NULL, 0 }, /* troff */
596: { roff_TS, NULL, NULL, 0 }, /* TS */
597: { roff_line_ignore, NULL, NULL, 0 }, /* uf */
598: { roff_line_ignore, NULL, NULL, 0 }, /* ul */
599: { roff_unsupp, NULL, NULL, 0 }, /* unformat */
600: { roff_line_ignore, NULL, NULL, 0 }, /* unwatch */
601: { roff_line_ignore, NULL, NULL, 0 }, /* unwatchn */
602: { roff_line_ignore, NULL, NULL, 0 }, /* vpt */
603: { roff_line_ignore, NULL, NULL, 0 }, /* vs */
604: { roff_line_ignore, NULL, NULL, 0 }, /* warn */
605: { roff_line_ignore, NULL, NULL, 0 }, /* warnscale */
606: { roff_line_ignore, NULL, NULL, 0 }, /* watch */
607: { roff_line_ignore, NULL, NULL, 0 }, /* watchlength */
608: { roff_line_ignore, NULL, NULL, 0 }, /* watchn */
609: { roff_unsupp, NULL, NULL, 0 }, /* wh */
1.340 schwarze 610: { roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT }, /*while*/
1.295 schwarze 611: { roff_insec, NULL, NULL, 0 }, /* write */
612: { roff_insec, NULL, NULL, 0 }, /* writec */
613: { roff_insec, NULL, NULL, 0 }, /* writem */
614: { roff_line_ignore, NULL, NULL, 0 }, /* xflag */
615: { roff_cblock, NULL, NULL, 0 }, /* . */
1.306 schwarze 616: { roff_renamed, NULL, NULL, 0 },
1.295 schwarze 617: { roff_userdef, NULL, NULL, 0 }
1.67 kristaps 618: };
619:
1.141 kristaps 620: /* Array of injected predefined strings. */
621: #define PREDEFS_MAX 38
622: static const struct predef predefs[PREDEFS_MAX] = {
623: #include "predefs.in"
624: };
625:
1.305 schwarze 626: static int roffce_lines; /* number of input lines to center */
627: static struct roff_node *roffce_node; /* active request */
1.178 schwarze 628: static int roffit_lines; /* number of lines to delay */
629: static char *roffit_macro; /* nil-terminated macro line */
630:
1.207 schwarze 631:
1.266 schwarze 632: /* --- request table ------------------------------------------------------ */
633:
1.295 schwarze 634: struct ohash *
635: roffhash_alloc(enum roff_tok mintok, enum roff_tok maxtok)
1.85 kristaps 636: {
1.295 schwarze 637: struct ohash *htab;
638: struct roffreq *req;
639: enum roff_tok tok;
640: size_t sz;
641: unsigned int slot;
1.85 kristaps 642:
1.295 schwarze 643: htab = mandoc_malloc(sizeof(*htab));
644: mandoc_ohash_init(htab, 8, offsetof(struct roffreq, name));
1.85 kristaps 645:
1.295 schwarze 646: for (tok = mintok; tok < maxtok; tok++) {
1.296 schwarze 647: if (roff_name[tok] == NULL)
648: continue;
1.295 schwarze 649: sz = strlen(roff_name[tok]);
650: req = mandoc_malloc(sizeof(*req) + sz + 1);
651: req->tok = tok;
652: memcpy(req->name, roff_name[tok], sz + 1);
653: slot = ohash_qlookup(htab, req->name);
654: ohash_insert(htab, slot, req);
1.85 kristaps 655: }
1.295 schwarze 656: return htab;
1.85 kristaps 657: }
658:
1.295 schwarze 659: void
660: roffhash_free(struct ohash *htab)
1.67 kristaps 661: {
1.295 schwarze 662: struct roffreq *req;
663: unsigned int slot;
1.67 kristaps 664:
1.295 schwarze 665: if (htab == NULL)
666: return;
667: for (req = ohash_first(htab, &slot); req != NULL;
668: req = ohash_next(htab, &slot))
669: free(req);
670: ohash_delete(htab);
671: free(htab);
672: }
673:
674: enum roff_tok
675: roffhash_find(struct ohash *htab, const char *name, size_t sz)
676: {
677: struct roffreq *req;
678: const char *end;
679:
680: if (sz) {
681: end = name + sz;
682: req = ohash_find(htab, ohash_qlookupi(htab, name, &end));
683: } else
684: req = ohash_find(htab, ohash_qlookup(htab, name));
685: return req == NULL ? TOKEN_NONE : req->tok;
1.67 kristaps 686: }
687:
1.266 schwarze 688: /* --- stack of request blocks -------------------------------------------- */
689:
1.67 kristaps 690: /*
691: * Pop the current node off of the stack of roff instructions currently
1.364 schwarze 692: * pending. Return 1 if it is a loop or 0 otherwise.
1.67 kristaps 693: */
1.340 schwarze 694: static int
1.67 kristaps 695: roffnode_pop(struct roff *r)
696: {
697: struct roffnode *p;
1.340 schwarze 698: int inloop;
1.67 kristaps 699:
1.207 schwarze 700: p = r->last;
1.340 schwarze 701: inloop = p->tok == ROFF_while;
702: r->last = p->parent;
1.106 kristaps 703: free(p->name);
704: free(p->end);
1.67 kristaps 705: free(p);
1.340 schwarze 706: return inloop;
1.67 kristaps 707: }
708:
709: /*
710: * Push a roff node onto the instruction stack. This must later be
711: * removed with roffnode_pop().
712: */
1.98 schwarze 713: static void
1.294 schwarze 714: roffnode_push(struct roff *r, enum roff_tok tok, const char *name,
1.106 kristaps 715: int line, int col)
1.67 kristaps 716: {
717: struct roffnode *p;
718:
1.98 schwarze 719: p = mandoc_calloc(1, sizeof(struct roffnode));
1.67 kristaps 720: p->tok = tok;
1.106 kristaps 721: if (name)
722: p->name = mandoc_strdup(name);
1.67 kristaps 723: p->parent = r->last;
724: p->line = line;
725: p->col = col;
1.198 schwarze 726: p->rule = p->parent ? p->parent->rule : 0;
1.67 kristaps 727:
728: r->last = p;
729: }
730:
1.266 schwarze 731: /* --- roff parser state data management ---------------------------------- */
732:
1.67 kristaps 733: static void
734: roff_free1(struct roff *r)
735: {
1.167 kristaps 736: int i;
1.67 kristaps 737:
1.346 schwarze 738: tbl_free(r->first_tbl);
1.113 kristaps 739: r->first_tbl = r->last_tbl = r->tbl = NULL;
740:
1.347 schwarze 741: eqn_free(r->last_eqn);
1.319 schwarze 742: r->last_eqn = r->eqn = NULL;
1.125 kristaps 743:
1.339 schwarze 744: while (r->mstackpos >= 0)
745: roff_userret(r);
746:
1.67 kristaps 747: while (r->last)
748: roffnode_pop(r);
1.109 kristaps 749:
1.223 schwarze 750: free (r->rstack);
751: r->rstack = NULL;
752: r->rstacksz = 0;
753: r->rstackpos = -1;
754:
755: roff_freereg(r->regtab);
756: r->regtab = NULL;
757:
1.167 kristaps 758: roff_freestr(r->strtab);
1.306 schwarze 759: roff_freestr(r->rentab);
1.167 kristaps 760: roff_freestr(r->xmbtab);
1.306 schwarze 761: r->strtab = r->rentab = r->xmbtab = NULL;
1.167 kristaps 762:
763: if (r->xtab)
764: for (i = 0; i < 128; i++)
765: free(r->xtab[i].p);
766: free(r->xtab);
767: r->xtab = NULL;
1.67 kristaps 768: }
769:
770: void
771: roff_reset(struct roff *r)
772: {
773: roff_free1(r);
1.367 schwarze 774: r->options |= MPARSE_COMMENT;
1.227 schwarze 775: r->format = r->options & (MPARSE_MDOC | MPARSE_MAN);
1.303 schwarze 776: r->control = '\0';
777: r->escape = '\\';
1.307 schwarze 778: roffce_lines = 0;
779: roffce_node = NULL;
780: roffit_lines = 0;
781: roffit_macro = NULL;
1.67 kristaps 782: }
783:
784: void
785: roff_free(struct roff *r)
786: {
1.366 schwarze 787: int i;
1.339 schwarze 788:
1.67 kristaps 789: roff_free1(r);
1.339 schwarze 790: for (i = 0; i < r->mstacksz; i++)
791: free(r->mstack[i].argv);
792: free(r->mstack);
1.295 schwarze 793: roffhash_free(r->reqtab);
1.67 kristaps 794: free(r);
795: }
796:
797: struct roff *
1.351 schwarze 798: roff_alloc(int options)
1.67 kristaps 799: {
800: struct roff *r;
801:
1.98 schwarze 802: r = mandoc_calloc(1, sizeof(struct roff));
1.328 schwarze 803: r->reqtab = roffhash_alloc(0, ROFF_RENAMED);
1.367 schwarze 804: r->options = options | MPARSE_COMMENT;
1.227 schwarze 805: r->format = options & (MPARSE_MDOC | MPARSE_MAN);
1.339 schwarze 806: r->mstackpos = -1;
1.82 kristaps 807: r->rstackpos = -1;
1.303 schwarze 808: r->escape = '\\';
1.277 schwarze 809: return r;
1.265 schwarze 810: }
811:
1.266 schwarze 812: /* --- syntax tree state data management ---------------------------------- */
813:
1.265 schwarze 814: static void
815: roff_man_free1(struct roff_man *man)
816: {
1.356 schwarze 817: if (man->meta.first != NULL)
818: roff_node_delete(man, man->meta.first);
1.265 schwarze 819: free(man->meta.msec);
820: free(man->meta.vol);
821: free(man->meta.os);
822: free(man->meta.arch);
823: free(man->meta.title);
824: free(man->meta.name);
825: free(man->meta.date);
1.356 schwarze 826: free(man->meta.sodest);
1.265 schwarze 827: }
828:
1.357 schwarze 829: void
830: roff_state_reset(struct roff_man *man)
831: {
832: man->last = man->meta.first;
833: man->last_es = NULL;
834: man->flags = 0;
835: man->lastsec = man->lastnamed = SEC_NONE;
836: man->next = ROFF_NEXT_CHILD;
837: roff_setreg(man->roff, "nS", 0, '=');
838: }
839:
1.265 schwarze 840: static void
841: roff_man_alloc1(struct roff_man *man)
842: {
843: memset(&man->meta, 0, sizeof(man->meta));
1.356 schwarze 844: man->meta.first = mandoc_calloc(1, sizeof(*man->meta.first));
845: man->meta.first->type = ROFFT_ROOT;
846: man->meta.macroset = MACROSET_NONE;
1.357 schwarze 847: roff_state_reset(man);
1.265 schwarze 848: }
849:
850: void
851: roff_man_reset(struct roff_man *man)
852: {
853: roff_man_free1(man);
854: roff_man_alloc1(man);
855: }
856:
857: void
858: roff_man_free(struct roff_man *man)
859: {
860: roff_man_free1(man);
861: free(man);
862: }
863:
864: struct roff_man *
1.351 schwarze 865: roff_man_alloc(struct roff *roff, const char *os_s, int quick)
1.265 schwarze 866: {
867: struct roff_man *man;
868:
869: man = mandoc_calloc(1, sizeof(*man));
870: man->roff = roff;
1.316 schwarze 871: man->os_s = os_s;
1.265 schwarze 872: man->quick = quick;
873: roff_man_alloc1(man);
1.296 schwarze 874: roff->man = man;
1.277 schwarze 875: return man;
1.67 kristaps 876: }
877:
1.266 schwarze 878: /* --- syntax tree handling ----------------------------------------------- */
879:
880: struct roff_node *
881: roff_node_alloc(struct roff_man *man, int line, int pos,
882: enum roff_type type, int tok)
883: {
884: struct roff_node *n;
885:
886: n = mandoc_calloc(1, sizeof(*n));
887: n->line = line;
888: n->pos = pos;
889: n->tok = tok;
890: n->type = type;
891: n->sec = man->lastsec;
892:
893: if (man->flags & MDOC_SYNOPSIS)
1.285 schwarze 894: n->flags |= NODE_SYNPRETTY;
1.266 schwarze 895: else
1.285 schwarze 896: n->flags &= ~NODE_SYNPRETTY;
1.360 schwarze 897: if ((man->flags & (ROFF_NOFILL | ROFF_NONOFILL)) == ROFF_NOFILL)
1.359 schwarze 898: n->flags |= NODE_NOFILL;
899: else
900: n->flags &= ~NODE_NOFILL;
1.266 schwarze 901: if (man->flags & MDOC_NEWLINE)
1.285 schwarze 902: n->flags |= NODE_LINE;
1.266 schwarze 903: man->flags &= ~MDOC_NEWLINE;
904:
1.277 schwarze 905: return n;
1.266 schwarze 906: }
907:
908: void
909: roff_node_append(struct roff_man *man, struct roff_node *n)
910: {
911:
912: switch (man->next) {
913: case ROFF_NEXT_SIBLING:
1.281 schwarze 914: if (man->last->next != NULL) {
915: n->next = man->last->next;
916: man->last->next->prev = n;
917: } else
918: man->last->parent->last = n;
1.266 schwarze 919: man->last->next = n;
920: n->prev = man->last;
921: n->parent = man->last->parent;
922: break;
923: case ROFF_NEXT_CHILD:
1.287 schwarze 924: if (man->last->child != NULL) {
925: n->next = man->last->child;
926: man->last->child->prev = n;
927: } else
928: man->last->last = n;
1.266 schwarze 929: man->last->child = n;
930: n->parent = man->last;
931: break;
932: default:
933: abort();
934: }
1.282 schwarze 935: man->last = n;
1.266 schwarze 936:
937: switch (n->type) {
938: case ROFFT_HEAD:
939: n->parent->head = n;
940: break;
941: case ROFFT_BODY:
1.282 schwarze 942: if (n->end != ENDBODY_NOT)
943: return;
1.266 schwarze 944: n->parent->body = n;
945: break;
946: case ROFFT_TAIL:
947: n->parent->tail = n;
948: break;
949: default:
1.282 schwarze 950: return;
1.266 schwarze 951: }
1.282 schwarze 952:
953: /*
954: * Copy over the normalised-data pointer of our parent. Not
955: * everybody has one, but copying a null pointer is fine.
956: */
957:
958: n->norm = n->parent->norm;
959: assert(n->parent->type == ROFFT_BLOCK);
1.266 schwarze 960: }
961:
1.267 schwarze 962: void
963: roff_word_alloc(struct roff_man *man, int line, int pos, const char *word)
964: {
965: struct roff_node *n;
966:
967: n = roff_node_alloc(man, line, pos, ROFFT_TEXT, TOKEN_NONE);
968: n->string = roff_strdup(man->roff, word);
969: roff_node_append(man, n);
1.286 schwarze 970: n->flags |= NODE_VALID | NODE_ENDED;
1.267 schwarze 971: man->next = ROFF_NEXT_SIBLING;
972: }
973:
974: void
975: roff_word_append(struct roff_man *man, const char *word)
976: {
977: struct roff_node *n;
978: char *addstr, *newstr;
979:
980: n = man->last;
981: addstr = roff_strdup(man->roff, word);
982: mandoc_asprintf(&newstr, "%s %s", n->string, addstr);
983: free(addstr);
984: free(n->string);
985: n->string = newstr;
986: man->next = ROFF_NEXT_SIBLING;
1.268 schwarze 987: }
988:
989: void
990: roff_elem_alloc(struct roff_man *man, int line, int pos, int tok)
991: {
992: struct roff_node *n;
993:
994: n = roff_node_alloc(man, line, pos, ROFFT_ELEM, tok);
995: roff_node_append(man, n);
996: man->next = ROFF_NEXT_CHILD;
997: }
998:
999: struct roff_node *
1000: roff_block_alloc(struct roff_man *man, int line, int pos, int tok)
1001: {
1002: struct roff_node *n;
1003:
1004: n = roff_node_alloc(man, line, pos, ROFFT_BLOCK, tok);
1005: roff_node_append(man, n);
1006: man->next = ROFF_NEXT_CHILD;
1.277 schwarze 1007: return n;
1.267 schwarze 1008: }
1009:
1.266 schwarze 1010: struct roff_node *
1011: roff_head_alloc(struct roff_man *man, int line, int pos, int tok)
1012: {
1013: struct roff_node *n;
1014:
1015: n = roff_node_alloc(man, line, pos, ROFFT_HEAD, tok);
1016: roff_node_append(man, n);
1017: man->next = ROFF_NEXT_CHILD;
1.277 schwarze 1018: return n;
1.266 schwarze 1019: }
1020:
1021: struct roff_node *
1022: roff_body_alloc(struct roff_man *man, int line, int pos, int tok)
1023: {
1024: struct roff_node *n;
1025:
1026: n = roff_node_alloc(man, line, pos, ROFFT_BODY, tok);
1027: roff_node_append(man, n);
1028: man->next = ROFF_NEXT_CHILD;
1.277 schwarze 1029: return n;
1.267 schwarze 1030: }
1031:
1.321 schwarze 1032: static void
1.346 schwarze 1033: roff_addtbl(struct roff_man *man, int line, struct tbl_node *tbl)
1.267 schwarze 1034: {
1035: struct roff_node *n;
1.346 schwarze 1036: struct tbl_span *span;
1.267 schwarze 1037:
1.356 schwarze 1038: if (man->meta.macroset == MACROSET_MAN)
1.302 schwarze 1039: man_breakscope(man, ROFF_TS);
1.321 schwarze 1040: while ((span = tbl_span(tbl)) != NULL) {
1.346 schwarze 1041: n = roff_node_alloc(man, line, 0, ROFFT_TBL, TOKEN_NONE);
1.321 schwarze 1042: n->span = span;
1043: roff_node_append(man, n);
1044: n->flags |= NODE_VALID | NODE_ENDED;
1045: man->next = ROFF_NEXT_SIBLING;
1046: }
1.266 schwarze 1047: }
1048:
1049: void
1050: roff_node_unlink(struct roff_man *man, struct roff_node *n)
1051: {
1052:
1053: /* Adjust siblings. */
1054:
1055: if (n->prev)
1056: n->prev->next = n->next;
1057: if (n->next)
1058: n->next->prev = n->prev;
1059:
1060: /* Adjust parent. */
1061:
1062: if (n->parent != NULL) {
1063: if (n->parent->child == n)
1064: n->parent->child = n->next;
1065: if (n->parent->last == n)
1066: n->parent->last = n->prev;
1067: }
1068:
1069: /* Adjust parse point. */
1070:
1071: if (man == NULL)
1072: return;
1073: if (man->last == n) {
1074: if (n->prev == NULL) {
1075: man->last = n->parent;
1076: man->next = ROFF_NEXT_CHILD;
1077: } else {
1078: man->last = n->prev;
1079: man->next = ROFF_NEXT_SIBLING;
1080: }
1081: }
1.356 schwarze 1082: if (man->meta.first == n)
1083: man->meta.first = NULL;
1.344 schwarze 1084: }
1085:
1086: void
1087: roff_node_relink(struct roff_man *man, struct roff_node *n)
1088: {
1089: roff_node_unlink(man, n);
1090: n->prev = n->next = NULL;
1091: roff_node_append(man, n);
1.266 schwarze 1092: }
1093:
1094: void
1095: roff_node_free(struct roff_node *n)
1096: {
1097:
1098: if (n->args != NULL)
1099: mdoc_argv_free(n->args);
1100: if (n->type == ROFFT_BLOCK || n->type == ROFFT_ELEM)
1101: free(n->norm);
1.347 schwarze 1102: eqn_box_free(n->eqn);
1.266 schwarze 1103: free(n->string);
1104: free(n);
1105: }
1106:
1107: void
1108: roff_node_delete(struct roff_man *man, struct roff_node *n)
1109: {
1110:
1111: while (n->child != NULL)
1112: roff_node_delete(man, n->child);
1113: roff_node_unlink(man, n);
1114: roff_node_free(n);
1.370 schwarze 1115: }
1116:
1117: int
1118: roff_node_transparent(struct roff_node *n)
1119: {
1120: if (n == NULL)
1121: return 0;
1122: if (n->type == ROFFT_COMMENT || n->flags & NODE_NOPRT)
1123: return 1;
1124: switch (n->tok) {
1125: case ROFF_ft:
1126: case ROFF_ll:
1127: case ROFF_mc:
1128: case ROFF_po:
1129: case ROFF_ta:
1130: case MDOC_Db:
1131: case MDOC_Es:
1132: case MDOC_Sm:
1133: case MDOC_Tg:
1134: case MAN_DT:
1135: case MAN_UC:
1136: case MAN_PD:
1137: case MAN_AT:
1138: return 1;
1139: default:
1140: return 0;
1141: }
1142: }
1143:
1144: struct roff_node *
1145: roff_node_child(struct roff_node *n)
1146: {
1147: for (n = n->child; roff_node_transparent(n); n = n->next)
1148: continue;
1149: return n;
1150: }
1151:
1152: struct roff_node *
1153: roff_node_prev(struct roff_node *n)
1154: {
1155: do {
1156: n = n->prev;
1157: } while (roff_node_transparent(n));
1158: return n;
1159: }
1160:
1161: struct roff_node *
1162: roff_node_next(struct roff_node *n)
1163: {
1164: do {
1165: n = n->next;
1166: } while (roff_node_transparent(n));
1167: return n;
1.269 schwarze 1168: }
1169:
1170: void
1171: deroff(char **dest, const struct roff_node *n)
1172: {
1173: char *cp;
1174: size_t sz;
1175:
1.371 ! schwarze 1176: if (n->string == NULL) {
1.269 schwarze 1177: for (n = n->child; n != NULL; n = n->next)
1178: deroff(dest, n);
1179: return;
1180: }
1181:
1.288 schwarze 1182: /* Skip leading whitespace. */
1.269 schwarze 1183:
1.288 schwarze 1184: for (cp = n->string; *cp != '\0'; cp++) {
1.289 schwarze 1185: if (cp[0] == '\\' && cp[1] != '\0' &&
1186: strchr(" %&0^|~", cp[1]) != NULL)
1.269 schwarze 1187: cp++;
1.288 schwarze 1188: else if ( ! isspace((unsigned char)*cp))
1.269 schwarze 1189: break;
1190: }
1191:
1.289 schwarze 1192: /* Skip trailing backslash. */
1193:
1194: sz = strlen(cp);
1.290 schwarze 1195: if (sz > 0 && cp[sz - 1] == '\\')
1.289 schwarze 1196: sz--;
1197:
1.269 schwarze 1198: /* Skip trailing whitespace. */
1199:
1.289 schwarze 1200: for (; sz; sz--)
1.269 schwarze 1201: if ( ! isspace((unsigned char)cp[sz-1]))
1202: break;
1203:
1204: /* Skip empty strings. */
1205:
1206: if (sz == 0)
1207: return;
1208:
1209: if (*dest == NULL) {
1210: *dest = mandoc_strndup(cp, sz);
1211: return;
1212: }
1213:
1214: mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
1215: free(*dest);
1216: *dest = cp;
1.266 schwarze 1217: }
1218:
1219: /* --- main functions of the roff parser ---------------------------------- */
1220:
1.94 kristaps 1221: /*
1.355 schwarze 1222: * In the current line, expand escape sequences that produce parsable
1223: * input text. Also check the syntax of the remaining escape sequences,
1224: * which typically produce output glyphs or change formatter state.
1.154 kristaps 1225: */
1.340 schwarze 1226: static int
1.355 schwarze 1227: roff_expand(struct roff *r, struct buf *buf, int ln, int pos, char newesc)
1.94 kristaps 1228: {
1.339 schwarze 1229: struct mctx *ctx; /* current macro call context */
1.208 schwarze 1230: char ubuf[24]; /* buffer to print the number */
1.328 schwarze 1231: struct roff_node *n; /* used for header comments */
1.205 schwarze 1232: const char *start; /* start of the string to process */
1.209 schwarze 1233: char *stesc; /* start of an escape sequence ('\\') */
1.352 schwarze 1234: const char *esct; /* type of esccape sequence */
1.328 schwarze 1235: char *ep; /* end of comment string */
1.108 schwarze 1236: const char *stnam; /* start of the name, after "[(*" */
1237: const char *cp; /* end of the name, e.g. before ']' */
1238: const char *res; /* the string to be substituted */
1.238 schwarze 1239: char *nbuf; /* new buffer to copy buf->buf to */
1.181 schwarze 1240: size_t maxl; /* expected length of the escape name */
1241: size_t naml; /* actual length of the escape name */
1.339 schwarze 1242: size_t asz; /* length of the replacement */
1243: size_t rsz; /* length of the rest of the string */
1.237 schwarze 1244: int inaml; /* length returned from mandoc_escape() */
1.181 schwarze 1245: int expand_count; /* to avoid infinite loops */
1.206 schwarze 1246: int npos; /* position in numeric expression */
1.218 schwarze 1247: int arg_complete; /* argument not interrupted by eol */
1.339 schwarze 1248: int quote_args; /* true for \\$@, false for \\$* */
1.303 schwarze 1249: int done; /* no more input available */
1.315 schwarze 1250: int deftype; /* type of definition to paste */
1.323 schwarze 1251: int rcsid; /* kind of RCS id seen */
1.352 schwarze 1252: enum mandocerr err; /* for escape sequence problems */
1.327 schwarze 1253: char sign; /* increment number register */
1.206 schwarze 1254: char term; /* character terminating the escape */
1.94 kristaps 1255:
1.303 schwarze 1256: /* Search forward for comments. */
1257:
1258: done = 0;
1259: start = buf->buf + pos;
1260: for (stesc = buf->buf + pos; *stesc != '\0'; stesc++) {
1.355 schwarze 1261: if (stesc[0] != newesc || stesc[1] == '\0')
1.303 schwarze 1262: continue;
1263: stesc++;
1264: if (*stesc != '"' && *stesc != '#')
1265: continue;
1.313 schwarze 1266:
1267: /* Comment found, look for RCS id. */
1268:
1.323 schwarze 1269: rcsid = 0;
1.313 schwarze 1270: if ((cp = strstr(stesc, "$" "OpenBSD")) != NULL) {
1.323 schwarze 1271: rcsid = 1 << MANDOC_OS_OPENBSD;
1.313 schwarze 1272: cp += 8;
1273: } else if ((cp = strstr(stesc, "$" "NetBSD")) != NULL) {
1.323 schwarze 1274: rcsid = 1 << MANDOC_OS_NETBSD;
1.313 schwarze 1275: cp += 7;
1276: }
1277: if (cp != NULL &&
1278: isalnum((unsigned char)*cp) == 0 &&
1.314 schwarze 1279: strchr(cp, '$') != NULL) {
1.323 schwarze 1280: if (r->man->meta.rcsids & rcsid)
1.350 schwarze 1281: mandoc_msg(MANDOCERR_RCS_REP, ln,
1282: (int)(stesc - buf->buf) + 1,
1283: "%s", stesc + 1);
1.323 schwarze 1284: r->man->meta.rcsids |= rcsid;
1.314 schwarze 1285: }
1.313 schwarze 1286:
1287: /* Handle trailing whitespace. */
1288:
1.328 schwarze 1289: ep = strchr(stesc--, '\0') - 1;
1290: if (*ep == '\n') {
1.303 schwarze 1291: done = 1;
1.328 schwarze 1292: ep--;
1.303 schwarze 1293: }
1.328 schwarze 1294: if (*ep == ' ' || *ep == '\t')
1.350 schwarze 1295: mandoc_msg(MANDOCERR_SPACE_EOL,
1296: ln, (int)(ep - buf->buf), NULL);
1.328 schwarze 1297:
1298: /*
1299: * Save comments preceding the title macro
1300: * in the syntax tree.
1301: */
1302:
1.367 schwarze 1303: if (newesc != ASCII_ESC && r->options & MPARSE_COMMENT) {
1.328 schwarze 1304: while (*ep == ' ' || *ep == '\t')
1305: ep--;
1306: ep[1] = '\0';
1307: n = roff_node_alloc(r->man,
1308: ln, stesc + 1 - buf->buf,
1309: ROFFT_COMMENT, TOKEN_NONE);
1310: n->string = mandoc_strdup(stesc + 2);
1311: roff_node_append(r->man, n);
1312: n->flags |= NODE_VALID | NODE_ENDED;
1313: r->man->next = ROFF_NEXT_SIBLING;
1314: }
1315:
1.334 schwarze 1316: /* Line continuation with comment. */
1317:
1318: if (stesc[1] == '#') {
1319: *stesc = '\0';
1.340 schwarze 1320: return ROFF_IGN | ROFF_APPEND;
1.334 schwarze 1321: }
1322:
1323: /* Discard normal comments. */
1.328 schwarze 1324:
1.335 schwarze 1325: while (stesc > start && stesc[-1] == ' ' &&
1326: (stesc == start + 1 || stesc[-2] != '\\'))
1.303 schwarze 1327: stesc--;
1328: *stesc = '\0';
1329: break;
1330: }
1331: if (stesc == start)
1332: return ROFF_CONT;
1333: stesc--;
1334:
1335: /* Notice the end of the input. */
1336:
1337: if (*stesc == '\n') {
1338: *stesc-- = '\0';
1339: done = 1;
1340: }
1341:
1.170 schwarze 1342: expand_count = 0;
1.303 schwarze 1343: while (stesc >= start) {
1.355 schwarze 1344: if (*stesc != newesc) {
1.170 schwarze 1345:
1.355 schwarze 1346: /*
1347: * If we have a non-standard escape character,
1348: * escape literal backslashes because all
1349: * processing in subsequent functions uses
1350: * the standard escaping rules.
1351: */
1.205 schwarze 1352:
1.355 schwarze 1353: if (newesc != ASCII_ESC && *stesc == '\\') {
1.303 schwarze 1354: *stesc = '\0';
1355: buf->sz = mandoc_asprintf(&nbuf, "%s\\e%s",
1356: buf->buf, stesc + 1) + 1;
1357: start = nbuf + pos;
1358: stesc = nbuf + (stesc - buf->buf);
1359: free(buf->buf);
1360: buf->buf = nbuf;
1361: }
1.355 schwarze 1362:
1363: /* Search backwards for the next escape. */
1364:
1.303 schwarze 1365: stesc--;
1.205 schwarze 1366: continue;
1.303 schwarze 1367: }
1.205 schwarze 1368:
1369: /* If it is escaped, skip it. */
1370:
1371: for (cp = stesc - 1; cp >= start; cp--)
1.303 schwarze 1372: if (*cp != r->escape)
1.205 schwarze 1373: break;
1374:
1.238 schwarze 1375: if ((stesc - cp) % 2 == 0) {
1.303 schwarze 1376: while (stesc > cp)
1377: *stesc-- = '\\';
1.205 schwarze 1378: continue;
1.303 schwarze 1379: } else if (stesc[1] != '\0') {
1380: *stesc = '\\';
1381: } else {
1382: *stesc-- = '\0';
1383: if (done)
1384: continue;
1385: else
1.340 schwarze 1386: return ROFF_IGN | ROFF_APPEND;
1.205 schwarze 1387: }
1.108 schwarze 1388:
1.206 schwarze 1389: /* Decide whether to expand or to check only. */
1.108 schwarze 1390:
1.206 schwarze 1391: term = '\0';
1.205 schwarze 1392: cp = stesc + 1;
1.352 schwarze 1393: if (*cp == 'E')
1394: cp++;
1395: esct = cp;
1396: switch (*esct) {
1.207 schwarze 1397: case '*':
1.339 schwarze 1398: case '$':
1.181 schwarze 1399: res = NULL;
1400: break;
1.207 schwarze 1401: case 'B':
1402: case 'w':
1.206 schwarze 1403: term = cp[1];
1404: /* FALLTHROUGH */
1.207 schwarze 1405: case 'n':
1.327 schwarze 1406: sign = cp[1];
1407: if (sign == '+' || sign == '-')
1408: cp++;
1.181 schwarze 1409: res = ubuf;
1410: break;
1411: default:
1.352 schwarze 1412: err = MANDOCERR_OK;
1413: switch(mandoc_escape(&cp, &stnam, &inaml)) {
1414: case ESCAPE_SPECIAL:
1415: if (mchars_spec2cp(stnam, inaml) >= 0)
1416: break;
1417: /* FALLTHROUGH */
1418: case ESCAPE_ERROR:
1419: err = MANDOCERR_ESC_BAD;
1420: break;
1421: case ESCAPE_UNDEF:
1422: err = MANDOCERR_ESC_UNDEF;
1423: break;
1424: case ESCAPE_UNSUPP:
1425: err = MANDOCERR_ESC_UNSUPP;
1426: break;
1427: default:
1428: break;
1429: }
1430: if (err != MANDOCERR_OK)
1431: mandoc_msg(err, ln, (int)(stesc - buf->buf),
1.219 schwarze 1432: "%.*s", (int)(cp - stesc), stesc);
1.303 schwarze 1433: stesc--;
1.205 schwarze 1434: continue;
1.152 kristaps 1435: }
1436:
1.205 schwarze 1437: if (EXPAND_LIMIT < ++expand_count) {
1.350 schwarze 1438: mandoc_msg(MANDOCERR_ROFFLOOP,
1.238 schwarze 1439: ln, (int)(stesc - buf->buf), NULL);
1.277 schwarze 1440: return ROFF_IGN;
1.205 schwarze 1441: }
1.108 schwarze 1442:
1443: /*
1444: * The third character decides the length
1.181 schwarze 1445: * of the name of the string or register.
1.108 schwarze 1446: * Save a pointer to the name.
1447: */
1448:
1.238 schwarze 1449: if (term == '\0') {
1.206 schwarze 1450: switch (*++cp) {
1.207 schwarze 1451: case '\0':
1.206 schwarze 1452: maxl = 0;
1453: break;
1.207 schwarze 1454: case '(':
1.206 schwarze 1455: cp++;
1456: maxl = 2;
1457: break;
1.207 schwarze 1458: case '[':
1.206 schwarze 1459: cp++;
1460: term = ']';
1461: maxl = 0;
1462: break;
1463: default:
1464: maxl = 1;
1465: break;
1466: }
1467: } else {
1468: cp += 2;
1.94 kristaps 1469: maxl = 0;
1470: }
1.108 schwarze 1471: stnam = cp;
1.94 kristaps 1472:
1.108 schwarze 1473: /* Advance to the end of the name. */
1.94 kristaps 1474:
1.253 schwarze 1475: naml = 0;
1.218 schwarze 1476: arg_complete = 1;
1.253 schwarze 1477: while (maxl == 0 || naml < maxl) {
1.238 schwarze 1478: if (*cp == '\0') {
1.350 schwarze 1479: mandoc_msg(MANDOCERR_ESC_BAD, ln,
1480: (int)(stesc - buf->buf), "%s", stesc);
1.218 schwarze 1481: arg_complete = 0;
1.206 schwarze 1482: break;
1.153 kristaps 1483: }
1.238 schwarze 1484: if (maxl == 0 && *cp == term) {
1.206 schwarze 1485: cp++;
1.253 schwarze 1486: break;
1487: }
1.352 schwarze 1488: if (*cp++ != '\\' || *esct != 'w') {
1.253 schwarze 1489: naml++;
1490: continue;
1491: }
1492: switch (mandoc_escape(&cp, NULL, NULL)) {
1493: case ESCAPE_SPECIAL:
1494: case ESCAPE_UNICODE:
1495: case ESCAPE_NUMBERED:
1.352 schwarze 1496: case ESCAPE_UNDEF:
1.253 schwarze 1497: case ESCAPE_OVERSTRIKE:
1498: naml++;
1499: break;
1500: default:
1.94 kristaps 1501: break;
1.206 schwarze 1502: }
1.94 kristaps 1503: }
1504:
1.108 schwarze 1505: /*
1506: * Retrieve the replacement string; if it is
1507: * undefined, resume searching for escapes.
1508: */
1509:
1.352 schwarze 1510: switch (*esct) {
1.207 schwarze 1511: case '*':
1.315 schwarze 1512: if (arg_complete) {
1513: deftype = ROFFDEF_USER | ROFFDEF_PRE;
1514: res = roff_getstrn(r, stnam, naml, &deftype);
1.331 schwarze 1515:
1516: /*
1517: * If not overriden, let \*(.T
1518: * through to the formatters.
1519: */
1520:
1521: if (res == NULL && naml == 2 &&
1522: stnam[0] == '.' && stnam[1] == 'T') {
1523: roff_setstrn(&r->strtab,
1524: ".T", 2, NULL, 0, 0);
1525: stesc--;
1526: continue;
1527: }
1.315 schwarze 1528: }
1.206 schwarze 1529: break;
1.339 schwarze 1530: case '$':
1531: if (r->mstackpos < 0) {
1.350 schwarze 1532: mandoc_msg(MANDOCERR_ARG_UNDEF, ln,
1533: (int)(stesc - buf->buf), "%.3s", stesc);
1.339 schwarze 1534: break;
1535: }
1536: ctx = r->mstack + r->mstackpos;
1.352 schwarze 1537: npos = esct[1] - '1';
1.339 schwarze 1538: if (npos >= 0 && npos <= 8) {
1539: res = npos < ctx->argc ?
1540: ctx->argv[npos] : "";
1541: break;
1542: }
1.352 schwarze 1543: if (esct[1] == '*')
1.339 schwarze 1544: quote_args = 0;
1.352 schwarze 1545: else if (esct[1] == '@')
1.339 schwarze 1546: quote_args = 1;
1547: else {
1.350 schwarze 1548: mandoc_msg(MANDOCERR_ARG_NONUM, ln,
1549: (int)(stesc - buf->buf), "%.3s", stesc);
1.339 schwarze 1550: break;
1551: }
1552: asz = 0;
1553: for (npos = 0; npos < ctx->argc; npos++) {
1554: if (npos)
1555: asz++; /* blank */
1556: if (quote_args)
1557: asz += 2; /* quotes */
1558: asz += strlen(ctx->argv[npos]);
1559: }
1560: if (asz != 3) {
1561: rsz = buf->sz - (stesc - buf->buf) - 3;
1562: if (asz < 3)
1563: memmove(stesc + asz, stesc + 3, rsz);
1564: buf->sz += asz - 3;
1565: nbuf = mandoc_realloc(buf->buf, buf->sz);
1566: start = nbuf + pos;
1567: stesc = nbuf + (stesc - buf->buf);
1568: buf->buf = nbuf;
1569: if (asz > 3)
1570: memmove(stesc + asz, stesc + 3, rsz);
1571: }
1572: for (npos = 0; npos < ctx->argc; npos++) {
1573: if (npos)
1574: *stesc++ = ' ';
1575: if (quote_args)
1576: *stesc++ = '"';
1577: cp = ctx->argv[npos];
1578: while (*cp != '\0')
1579: *stesc++ = *cp++;
1580: if (quote_args)
1581: *stesc++ = '"';
1582: }
1583: continue;
1.207 schwarze 1584: case 'B':
1.206 schwarze 1585: npos = 0;
1.218 schwarze 1586: ubuf[0] = arg_complete &&
1.261 schwarze 1587: roff_evalnum(r, ln, stnam, &npos,
1588: NULL, ROFFNUM_SCALE) &&
1.218 schwarze 1589: stnam + npos + 1 == cp ? '1' : '0';
1.206 schwarze 1590: ubuf[1] = '\0';
1591: break;
1.207 schwarze 1592: case 'n':
1.218 schwarze 1593: if (arg_complete)
1594: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.327 schwarze 1595: roff_getregn(r, stnam, naml, sign));
1.218 schwarze 1596: else
1597: ubuf[0] = '\0';
1.206 schwarze 1598: break;
1.207 schwarze 1599: case 'w':
1.218 schwarze 1600: /* use even incomplete args */
1.208 schwarze 1601: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.206 schwarze 1602: 24 * (int)naml);
1603: break;
1604: }
1.94 kristaps 1605:
1.238 schwarze 1606: if (res == NULL) {
1.352 schwarze 1607: if (*esct == '*')
1.350 schwarze 1608: mandoc_msg(MANDOCERR_STR_UNDEF,
1609: ln, (int)(stesc - buf->buf),
1.339 schwarze 1610: "%.*s", (int)naml, stnam);
1.142 kristaps 1611: res = "";
1.246 schwarze 1612: } else if (buf->sz + strlen(res) > SHRT_MAX) {
1.350 schwarze 1613: mandoc_msg(MANDOCERR_ROFFLOOP,
1.246 schwarze 1614: ln, (int)(stesc - buf->buf), NULL);
1.277 schwarze 1615: return ROFF_IGN;
1.94 kristaps 1616: }
1617:
1.108 schwarze 1618: /* Replace the escape sequence by the string. */
1619:
1.209 schwarze 1620: *stesc = '\0';
1.238 schwarze 1621: buf->sz = mandoc_asprintf(&nbuf, "%s%s%s",
1622: buf->buf, res, cp) + 1;
1.94 kristaps 1623:
1.205 schwarze 1624: /* Prepare for the next replacement. */
1.94 kristaps 1625:
1.205 schwarze 1626: start = nbuf + pos;
1.238 schwarze 1627: stesc = nbuf + (stesc - buf->buf) + strlen(res);
1628: free(buf->buf);
1629: buf->buf = nbuf;
1.154 kristaps 1630: }
1.277 schwarze 1631: return ROFF_CONT;
1.154 kristaps 1632: }
1.353 schwarze 1633:
1634: /*
1635: * Parse a quoted or unquoted roff-style request or macro argument.
1636: * Return a pointer to the parsed argument, which is either the original
1637: * pointer or advanced by one byte in case the argument is quoted.
1638: * NUL-terminate the argument in place.
1639: * Collapse pairs of quotes inside quoted arguments.
1640: * Advance the argument pointer to the next argument,
1641: * or to the NUL byte terminating the argument line.
1642: */
1643: char *
1.355 schwarze 1644: roff_getarg(struct roff *r, char **cpp, int ln, int *pos)
1.353 schwarze 1645: {
1.355 schwarze 1646: struct buf buf;
1.366 schwarze 1647: char *cp, *start;
1.355 schwarze 1648: int newesc, pairs, quoted, white;
1.353 schwarze 1649:
1650: /* Quoting can only start with a new word. */
1651: start = *cpp;
1652: quoted = 0;
1653: if ('"' == *start) {
1654: quoted = 1;
1655: start++;
1656: }
1657:
1.355 schwarze 1658: newesc = pairs = white = 0;
1.353 schwarze 1659: for (cp = start; '\0' != *cp; cp++) {
1660:
1661: /*
1662: * Move the following text left
1663: * after quoted quotes and after "\\" and "\t".
1664: */
1665: if (pairs)
1666: cp[-pairs] = cp[0];
1667:
1668: if ('\\' == cp[0]) {
1669: /*
1670: * In copy mode, translate double to single
1671: * backslashes and backslash-t to literal tabs.
1672: */
1673: switch (cp[1]) {
1674: case 'a':
1675: case 't':
1.354 schwarze 1676: cp[-pairs] = '\t';
1.355 schwarze 1677: pairs++;
1678: cp++;
1679: break;
1.353 schwarze 1680: case '\\':
1.355 schwarze 1681: newesc = 1;
1682: cp[-pairs] = ASCII_ESC;
1.353 schwarze 1683: pairs++;
1684: cp++;
1685: break;
1686: case ' ':
1687: /* Skip escaped blanks. */
1688: if (0 == quoted)
1689: cp++;
1690: break;
1691: default:
1692: break;
1693: }
1694: } else if (0 == quoted) {
1695: if (' ' == cp[0]) {
1696: /* Unescaped blanks end unquoted args. */
1697: white = 1;
1698: break;
1699: }
1700: } else if ('"' == cp[0]) {
1701: if ('"' == cp[1]) {
1702: /* Quoted quotes collapse. */
1703: pairs++;
1704: cp++;
1705: } else {
1706: /* Unquoted quotes end quoted args. */
1707: quoted = 2;
1708: break;
1709: }
1710: }
1711: }
1712:
1713: /* Quoted argument without a closing quote. */
1714: if (1 == quoted)
1715: mandoc_msg(MANDOCERR_ARG_QUOTE, ln, *pos, NULL);
1716:
1717: /* NUL-terminate this argument and move to the next one. */
1718: if (pairs)
1719: cp[-pairs] = '\0';
1720: if ('\0' != *cp) {
1721: *cp++ = '\0';
1722: while (' ' == *cp)
1723: cp++;
1724: }
1725: *pos += (int)(cp - start) + (quoted ? 1 : 0);
1726: *cpp = cp;
1727:
1728: if ('\0' == *cp && (white || ' ' == cp[-1]))
1729: mandoc_msg(MANDOCERR_SPACE_EOL, ln, *pos, NULL);
1730:
1.355 schwarze 1731: start = mandoc_strdup(start);
1732: if (newesc == 0)
1733: return start;
1734:
1735: buf.buf = start;
1736: buf.sz = strlen(start) + 1;
1737: buf.next = NULL;
1738: if (roff_expand(r, &buf, ln, 0, ASCII_ESC) & ROFF_IGN) {
1739: free(buf.buf);
1740: buf.buf = mandoc_strdup("");
1741: }
1742: return buf.buf;
1.353 schwarze 1743: }
1744:
1.154 kristaps 1745:
1746: /*
1.275 schwarze 1747: * Process text streams.
1.154 kristaps 1748: */
1.340 schwarze 1749: static int
1.305 schwarze 1750: roff_parsetext(struct roff *r, struct buf *buf, int pos, int *offs)
1.154 kristaps 1751: {
1752: size_t sz;
1753: const char *start;
1.178 schwarze 1754: char *p;
1755: int isz;
1.154 kristaps 1756: enum mandoc_esc esc;
1757:
1.275 schwarze 1758: /* Spring the input line trap. */
1759:
1760: if (roffit_lines == 1) {
1761: isz = mandoc_asprintf(&p, "%s\n.%s", buf->buf, roffit_macro);
1762: free(buf->buf);
1763: buf->buf = p;
1764: buf->sz = isz + 1;
1765: *offs = 0;
1766: free(roffit_macro);
1767: roffit_lines = 0;
1.277 schwarze 1768: return ROFF_REPARSE;
1.275 schwarze 1769: } else if (roffit_lines > 1)
1770: --roffit_lines;
1771:
1.305 schwarze 1772: if (roffce_node != NULL && buf->buf[pos] != '\0') {
1773: if (roffce_lines < 1) {
1774: r->man->last = roffce_node;
1775: r->man->next = ROFF_NEXT_SIBLING;
1776: roffce_lines = 0;
1777: roffce_node = NULL;
1778: } else
1779: roffce_lines--;
1780: }
1781:
1.275 schwarze 1782: /* Convert all breakable hyphens into ASCII_HYPH. */
1783:
1.238 schwarze 1784: start = p = buf->buf + pos;
1.154 kristaps 1785:
1.238 schwarze 1786: while (*p != '\0') {
1.154 kristaps 1787: sz = strcspn(p, "-\\");
1788: p += sz;
1789:
1.238 schwarze 1790: if (*p == '\0')
1.159 kristaps 1791: break;
1792:
1.238 schwarze 1793: if (*p == '\\') {
1.154 kristaps 1794: /* Skip over escapes. */
1795: p++;
1.189 schwarze 1796: esc = mandoc_escape((const char **)&p, NULL, NULL);
1.238 schwarze 1797: if (esc == ESCAPE_ERROR)
1.154 kristaps 1798: break;
1.264 schwarze 1799: while (*p == '-')
1800: p++;
1.155 kristaps 1801: continue;
1.159 kristaps 1802: } else if (p == start) {
1.158 kristaps 1803: p++;
1.155 kristaps 1804: continue;
1.158 kristaps 1805: }
1.155 kristaps 1806:
1.171 schwarze 1807: if (isalpha((unsigned char)p[-1]) &&
1808: isalpha((unsigned char)p[1]))
1.155 kristaps 1809: *p = ASCII_HYPH;
1810: p++;
1.94 kristaps 1811: }
1.277 schwarze 1812: return ROFF_CONT;
1.94 kristaps 1813: }
1814:
1.340 schwarze 1815: int
1.238 schwarze 1816: roff_parseln(struct roff *r, int ln, struct buf *buf, int *offs)
1.67 kristaps 1817: {
1.294 schwarze 1818: enum roff_tok t;
1.340 schwarze 1819: int e;
1.238 schwarze 1820: int pos; /* parse point */
1.243 schwarze 1821: int spos; /* saved parse point for messages */
1.238 schwarze 1822: int ppos; /* original offset in buf->buf */
1823: int ctl; /* macro line (boolean) */
1824:
1825: ppos = pos = *offs;
1.79 kristaps 1826:
1.230 schwarze 1827: /* Handle in-line equation delimiters. */
1828:
1.236 schwarze 1829: if (r->tbl == NULL &&
1830: r->last_eqn != NULL && r->last_eqn->delim &&
1.230 schwarze 1831: (r->eqn == NULL || r->eqn_inline)) {
1.238 schwarze 1832: e = roff_eqndelim(r, buf, pos);
1.230 schwarze 1833: if (e == ROFF_REPARSE)
1.277 schwarze 1834: return e;
1.230 schwarze 1835: assert(e == ROFF_CONT);
1836: }
1837:
1838: /* Expand some escape sequences. */
1.94 kristaps 1839:
1.355 schwarze 1840: e = roff_expand(r, buf, ln, pos, r->escape);
1.340 schwarze 1841: if ((e & ROFF_MASK) == ROFF_IGN)
1.277 schwarze 1842: return e;
1.238 schwarze 1843: assert(e == ROFF_CONT);
1.94 kristaps 1844:
1.238 schwarze 1845: ctl = roff_getcontrol(r, buf->buf, &pos);
1.130 kristaps 1846:
1.94 kristaps 1847: /*
1.79 kristaps 1848: * First, if a scope is open and we're not a macro, pass the
1.252 schwarze 1849: * text through the macro's filter.
1850: * Equations process all content themselves.
1851: * Tables process almost all content themselves, but we want
1852: * to warn about macros before passing it there.
1.79 kristaps 1853: */
1.74 kristaps 1854:
1.252 schwarze 1855: if (r->last != NULL && ! ctl) {
1.78 kristaps 1856: t = r->last->tok;
1.238 schwarze 1857: e = (*roffs[t].text)(r, t, buf, ln, pos, pos, offs);
1.340 schwarze 1858: if ((e & ROFF_MASK) == ROFF_IGN)
1.277 schwarze 1859: return e;
1.340 schwarze 1860: e &= ~ROFF_MASK;
1861: } else
1862: e = ROFF_IGN;
1.319 schwarze 1863: if (r->eqn != NULL && strncmp(buf->buf + ppos, ".EN", 3)) {
1864: eqn_read(r->eqn, buf->buf + ppos);
1.340 schwarze 1865: return e;
1.319 schwarze 1866: }
1.321 schwarze 1867: if (r->tbl != NULL && (ctl == 0 || buf->buf[pos] == '\0')) {
1868: tbl_read(r->tbl, ln, buf->buf, ppos);
1.346 schwarze 1869: roff_addtbl(r->man, ln, r->tbl);
1.340 schwarze 1870: return e;
1.321 schwarze 1871: }
1.367 schwarze 1872: if ( ! ctl) {
1873: r->options &= ~MPARSE_COMMENT;
1.340 schwarze 1874: return roff_parsetext(r, buf, pos, offs) | e;
1.367 schwarze 1875: }
1.228 schwarze 1876:
1877: /* Skip empty request lines. */
1878:
1.238 schwarze 1879: if (buf->buf[pos] == '"') {
1.350 schwarze 1880: mandoc_msg(MANDOCERR_COMMENT_BAD, ln, pos, NULL);
1.277 schwarze 1881: return ROFF_IGN;
1.238 schwarze 1882: } else if (buf->buf[pos] == '\0')
1.277 schwarze 1883: return ROFF_IGN;
1.67 kristaps 1884:
1.79 kristaps 1885: /*
1886: * If a scope is open, go to the child handler for that macro,
1887: * as it may want to preprocess before doing anything with it.
1.125 kristaps 1888: * Don't do so if an equation is open.
1.79 kristaps 1889: */
1.78 kristaps 1890:
1.79 kristaps 1891: if (r->last) {
1892: t = r->last->tok;
1.277 schwarze 1893: return (*roffs[t].sub)(r, t, buf, ln, ppos, pos, offs);
1.79 kristaps 1894: }
1.78 kristaps 1895:
1.243 schwarze 1896: /* No scope is open. This is a new request or macro. */
1897:
1.367 schwarze 1898: r->options &= ~MPARSE_COMMENT;
1.243 schwarze 1899: spos = pos;
1900: t = roff_parse(r, buf->buf, &pos, ln, ppos);
1901:
1902: /* Tables ignore most macros. */
1903:
1.308 schwarze 1904: if (r->tbl != NULL && (t == TOKEN_NONE || t == ROFF_TS ||
1.309 schwarze 1905: t == ROFF_br || t == ROFF_ce || t == ROFF_rj || t == ROFF_sp)) {
1.350 schwarze 1906: mandoc_msg(MANDOCERR_TBLMACRO,
1907: ln, pos, "%s", buf->buf + spos);
1.308 schwarze 1908: if (t != TOKEN_NONE)
1.277 schwarze 1909: return ROFF_IGN;
1.257 schwarze 1910: while (buf->buf[pos] != '\0' && buf->buf[pos] != ' ')
1911: pos++;
1.291 schwarze 1912: while (buf->buf[pos] == ' ')
1.257 schwarze 1913: pos++;
1.321 schwarze 1914: tbl_read(r->tbl, ln, buf->buf, pos);
1.346 schwarze 1915: roff_addtbl(r->man, ln, r->tbl);
1.321 schwarze 1916: return ROFF_IGN;
1.243 schwarze 1917: }
1918:
1.305 schwarze 1919: /* For now, let high level macros abort .ce mode. */
1920:
1921: if (ctl && roffce_node != NULL &&
1.324 schwarze 1922: (t == TOKEN_NONE || t == ROFF_Dd || t == ROFF_EQ ||
1923: t == ROFF_TH || t == ROFF_TS)) {
1.305 schwarze 1924: r->man->last = roffce_node;
1925: r->man->next = ROFF_NEXT_SIBLING;
1926: roffce_lines = 0;
1927: roffce_node = NULL;
1928: }
1929:
1.79 kristaps 1930: /*
1.243 schwarze 1931: * This is neither a roff request nor a user-defined macro.
1932: * Let the standard macro set parsers handle it.
1.79 kristaps 1933: */
1.67 kristaps 1934:
1.294 schwarze 1935: if (t == TOKEN_NONE)
1.277 schwarze 1936: return ROFF_CONT;
1.243 schwarze 1937:
1938: /* Execute a roff request or a user defined macro. */
1.67 kristaps 1939:
1.296 schwarze 1940: return (*roffs[t].proc)(r, t, buf, ln, spos, pos, offs);
1.74 kristaps 1941: }
1942:
1.339 schwarze 1943: /*
1944: * Internal interface function to tell the roff parser that execution
1945: * of the current macro ended. This is required because macro
1946: * definitions usually do not end with a .return request.
1947: */
1948: void
1949: roff_userret(struct roff *r)
1950: {
1951: struct mctx *ctx;
1952: int i;
1953:
1954: assert(r->mstackpos >= 0);
1955: ctx = r->mstack + r->mstackpos;
1956: for (i = 0; i < ctx->argc; i++)
1957: free(ctx->argv[i]);
1958: ctx->argc = 0;
1959: r->mstackpos--;
1960: }
1961:
1.117 kristaps 1962: void
1.74 kristaps 1963: roff_endparse(struct roff *r)
1964: {
1.321 schwarze 1965: if (r->last != NULL)
1.350 schwarze 1966: mandoc_msg(MANDOCERR_BLK_NOEND, r->last->line,
1967: r->last->col, "%s", roff_name[r->last->tok]);
1.117 kristaps 1968:
1.321 schwarze 1969: if (r->eqn != NULL) {
1.350 schwarze 1970: mandoc_msg(MANDOCERR_BLK_NOEND,
1.319 schwarze 1971: r->eqn->node->line, r->eqn->node->pos, "EQ");
1972: eqn_parse(r->eqn);
1973: r->eqn = NULL;
1.125 kristaps 1974: }
1975:
1.321 schwarze 1976: if (r->tbl != NULL) {
1.346 schwarze 1977: tbl_end(r->tbl, 1);
1.321 schwarze 1978: r->tbl = NULL;
1.117 kristaps 1979: }
1.67 kristaps 1980: }
1981:
1982: /*
1983: * Parse a roff node's type from the input buffer. This must be in the
1984: * form of ".foo xxx" in the usual way.
1985: */
1.294 schwarze 1986: static enum roff_tok
1.214 schwarze 1987: roff_parse(struct roff *r, char *buf, int *pos, int ln, int ppos)
1.67 kristaps 1988: {
1.214 schwarze 1989: char *cp;
1.106 kristaps 1990: const char *mac;
1991: size_t maclen;
1.315 schwarze 1992: int deftype;
1.294 schwarze 1993: enum roff_tok t;
1.67 kristaps 1994:
1.214 schwarze 1995: cp = buf + *pos;
1996:
1997: if ('\0' == *cp || '"' == *cp || '\t' == *cp || ' ' == *cp)
1.294 schwarze 1998: return TOKEN_NONE;
1.67 kristaps 1999:
1.214 schwarze 2000: mac = cp;
2001: maclen = roff_getname(r, &cp, ln, ppos);
1.67 kristaps 2002:
1.315 schwarze 2003: deftype = ROFFDEF_USER | ROFFDEF_REN;
2004: r->current_string = roff_getstrn(r, mac, maclen, &deftype);
2005: switch (deftype) {
2006: case ROFFDEF_USER:
2007: t = ROFF_USERDEF;
2008: break;
2009: case ROFFDEF_REN:
2010: t = ROFF_RENAMED;
2011: break;
2012: default:
2013: t = roffhash_find(r->reqtab, mac, maclen);
2014: break;
2015: }
1.294 schwarze 2016: if (t != TOKEN_NONE)
1.214 schwarze 2017: *pos = cp - buf;
1.325 schwarze 2018: else if (deftype == ROFFDEF_UNDEF) {
2019: /* Using an undefined macro defines it to be empty. */
2020: roff_setstrn(&r->strtab, mac, maclen, "", 0, 0);
2021: roff_setstrn(&r->rentab, mac, maclen, NULL, 0, 0);
2022: }
1.277 schwarze 2023: return t;
1.67 kristaps 2024: }
2025:
1.266 schwarze 2026: /* --- handling of request blocks ----------------------------------------- */
2027:
1.340 schwarze 2028: static int
1.76 kristaps 2029: roff_cblock(ROFF_ARGS)
1.67 kristaps 2030: {
2031:
1.79 kristaps 2032: /*
2033: * A block-close `..' should only be invoked as a child of an
2034: * ignore macro, otherwise raise a warning and just ignore it.
2035: */
2036:
1.238 schwarze 2037: if (r->last == NULL) {
1.350 schwarze 2038: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "..");
1.277 schwarze 2039: return ROFF_IGN;
1.76 kristaps 2040: }
1.67 kristaps 2041:
1.81 kristaps 2042: switch (r->last->tok) {
1.207 schwarze 2043: case ROFF_am:
1.220 schwarze 2044: /* ROFF_am1 is remapped to ROFF_am in roff_block(). */
1.207 schwarze 2045: case ROFF_ami:
2046: case ROFF_de:
1.108 schwarze 2047: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.207 schwarze 2048: case ROFF_dei:
2049: case ROFF_ig:
1.81 kristaps 2050: break;
2051: default:
1.350 schwarze 2052: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "..");
1.277 schwarze 2053: return ROFF_IGN;
1.76 kristaps 2054: }
1.67 kristaps 2055:
1.238 schwarze 2056: if (buf->buf[pos] != '\0')
1.350 schwarze 2057: mandoc_msg(MANDOCERR_ARG_SKIP, ln, pos,
1.238 schwarze 2058: ".. %s", buf->buf + pos);
1.71 kristaps 2059:
2060: roffnode_pop(r);
1.76 kristaps 2061: roffnode_cleanscope(r);
1.277 schwarze 2062: return ROFF_IGN;
1.71 kristaps 2063:
1.67 kristaps 2064: }
2065:
1.364 schwarze 2066: /*
2067: * Pop all nodes ending at the end of the current input line.
2068: * Return the number of loops ended.
2069: */
1.340 schwarze 2070: static int
1.76 kristaps 2071: roffnode_cleanscope(struct roff *r)
1.67 kristaps 2072: {
1.340 schwarze 2073: int inloop;
1.67 kristaps 2074:
1.340 schwarze 2075: inloop = 0;
2076: while (r->last != NULL) {
1.173 schwarze 2077: if (--r->last->endspan != 0)
1.76 kristaps 2078: break;
1.340 schwarze 2079: inloop += roffnode_pop(r);
1.76 kristaps 2080: }
1.340 schwarze 2081: return inloop;
1.67 kristaps 2082: }
2083:
1.364 schwarze 2084: /*
2085: * Handle the closing \} of a conditional block.
2086: * Apart from generating warnings, this only pops nodes.
2087: * Return the number of loops ended.
2088: */
1.340 schwarze 2089: static int
1.195 schwarze 2090: roff_ccond(struct roff *r, int ln, int ppos)
1.74 kristaps 2091: {
1.76 kristaps 2092: if (NULL == r->last) {
1.350 schwarze 2093: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "\\}");
1.340 schwarze 2094: return 0;
1.76 kristaps 2095: }
2096:
1.82 kristaps 2097: switch (r->last->tok) {
1.207 schwarze 2098: case ROFF_el:
2099: case ROFF_ie:
2100: case ROFF_if:
1.340 schwarze 2101: case ROFF_while:
1.82 kristaps 2102: break;
2103: default:
1.350 schwarze 2104: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "\\}");
1.340 schwarze 2105: return 0;
1.75 kristaps 2106: }
2107:
1.76 kristaps 2108: if (r->last->endspan > -1) {
1.350 schwarze 2109: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "\\}");
1.340 schwarze 2110: return 0;
1.76 kristaps 2111: }
2112:
1.340 schwarze 2113: return roffnode_pop(r) + roffnode_cleanscope(r);
1.76 kristaps 2114: }
2115:
1.340 schwarze 2116: static int
1.80 kristaps 2117: roff_block(ROFF_ARGS)
1.76 kristaps 2118: {
1.315 schwarze 2119: const char *name, *value;
2120: char *call, *cp, *iname, *rname;
2121: size_t csz, namesz, rsz;
2122: int deftype;
1.106 kristaps 2123:
1.220 schwarze 2124: /* Ignore groff compatibility mode for now. */
1.76 kristaps 2125:
1.238 schwarze 2126: if (tok == ROFF_de1)
1.220 schwarze 2127: tok = ROFF_de;
1.251 schwarze 2128: else if (tok == ROFF_dei1)
2129: tok = ROFF_dei;
1.238 schwarze 2130: else if (tok == ROFF_am1)
1.220 schwarze 2131: tok = ROFF_am;
1.251 schwarze 2132: else if (tok == ROFF_ami1)
2133: tok = ROFF_ami;
1.220 schwarze 2134:
2135: /* Parse the macro name argument. */
2136:
1.238 schwarze 2137: cp = buf->buf + pos;
2138: if (tok == ROFF_ig) {
1.220 schwarze 2139: iname = NULL;
2140: namesz = 0;
2141: } else {
2142: iname = cp;
2143: namesz = roff_getname(r, &cp, ln, ppos);
2144: iname[namesz] = '\0';
2145: }
1.107 kristaps 2146:
1.220 schwarze 2147: /* Resolve the macro name argument if it is indirect. */
1.107 kristaps 2148:
1.238 schwarze 2149: if (namesz && (tok == ROFF_dei || tok == ROFF_ami)) {
1.315 schwarze 2150: deftype = ROFFDEF_USER;
2151: name = roff_getstrn(r, iname, namesz, &deftype);
2152: if (name == NULL) {
1.350 schwarze 2153: mandoc_msg(MANDOCERR_STR_UNDEF,
2154: ln, (int)(iname - buf->buf),
1.220 schwarze 2155: "%.*s", (int)namesz, iname);
2156: namesz = 0;
2157: } else
2158: namesz = strlen(name);
2159: } else
2160: name = iname;
1.107 kristaps 2161:
1.238 schwarze 2162: if (namesz == 0 && tok != ROFF_ig) {
1.350 schwarze 2163: mandoc_msg(MANDOCERR_REQ_EMPTY,
2164: ln, ppos, "%s", roff_name[tok]);
1.277 schwarze 2165: return ROFF_IGN;
1.220 schwarze 2166: }
1.80 kristaps 2167:
1.106 kristaps 2168: roffnode_push(r, tok, name, ln, ppos);
2169:
2170: /*
2171: * At the beginning of a `de' macro, clear the existing string
2172: * with the same name, if there is one. New content will be
1.193 schwarze 2173: * appended from roff_block_text() in multiline mode.
1.106 kristaps 2174: */
1.107 kristaps 2175:
1.311 schwarze 2176: if (tok == ROFF_de || tok == ROFF_dei) {
1.213 schwarze 2177: roff_setstrn(&r->strtab, name, namesz, "", 0, 0);
1.311 schwarze 2178: roff_setstrn(&r->rentab, name, namesz, NULL, 0, 0);
1.315 schwarze 2179: } else if (tok == ROFF_am || tok == ROFF_ami) {
2180: deftype = ROFFDEF_ANY;
2181: value = roff_getstrn(r, iname, namesz, &deftype);
2182: switch (deftype) { /* Before appending, ... */
2183: case ROFFDEF_PRE: /* copy predefined to user-defined. */
2184: roff_setstrn(&r->strtab, name, namesz,
2185: value, strlen(value), 0);
2186: break;
2187: case ROFFDEF_REN: /* call original standard macro. */
2188: csz = mandoc_asprintf(&call, ".%.*s \\$* \\\"\n",
2189: (int)strlen(value), value);
2190: roff_setstrn(&r->strtab, name, namesz, call, csz, 0);
2191: roff_setstrn(&r->rentab, name, namesz, NULL, 0, 0);
2192: free(call);
2193: break;
2194: case ROFFDEF_STD: /* rename and call standard macro. */
2195: rsz = mandoc_asprintf(&rname, "__%s_renamed", name);
2196: roff_setstrn(&r->rentab, rname, rsz, name, namesz, 0);
2197: csz = mandoc_asprintf(&call, ".%.*s \\$* \\\"\n",
2198: (int)rsz, rname);
2199: roff_setstrn(&r->strtab, name, namesz, call, csz, 0);
2200: free(call);
2201: free(rname);
2202: break;
2203: default:
2204: break;
2205: }
1.311 schwarze 2206: }
1.76 kristaps 2207:
1.238 schwarze 2208: if (*cp == '\0')
1.277 schwarze 2209: return ROFF_IGN;
1.78 kristaps 2210:
1.220 schwarze 2211: /* Get the custom end marker. */
1.107 kristaps 2212:
1.220 schwarze 2213: iname = cp;
1.213 schwarze 2214: namesz = roff_getname(r, &cp, ln, ppos);
1.220 schwarze 2215:
2216: /* Resolve the end marker if it is indirect. */
2217:
1.238 schwarze 2218: if (namesz && (tok == ROFF_dei || tok == ROFF_ami)) {
1.315 schwarze 2219: deftype = ROFFDEF_USER;
2220: name = roff_getstrn(r, iname, namesz, &deftype);
2221: if (name == NULL) {
1.350 schwarze 2222: mandoc_msg(MANDOCERR_STR_UNDEF,
2223: ln, (int)(iname - buf->buf),
1.220 schwarze 2224: "%.*s", (int)namesz, iname);
2225: namesz = 0;
2226: } else
2227: namesz = strlen(name);
2228: } else
2229: name = iname;
2230:
1.213 schwarze 2231: if (namesz)
2232: r->last->end = mandoc_strndup(name, namesz);
1.78 kristaps 2233:
1.238 schwarze 2234: if (*cp != '\0')
1.350 schwarze 2235: mandoc_msg(MANDOCERR_ARG_EXCESS,
1.295 schwarze 2236: ln, pos, ".%s ... %s", roff_name[tok], cp);
1.74 kristaps 2237:
1.277 schwarze 2238: return ROFF_IGN;
1.78 kristaps 2239: }
2240:
1.340 schwarze 2241: static int
1.80 kristaps 2242: roff_block_sub(ROFF_ARGS)
1.79 kristaps 2243: {
1.294 schwarze 2244: enum roff_tok t;
1.79 kristaps 2245: int i, j;
2246:
2247: /*
2248: * First check whether a custom macro exists at this level. If
2249: * it does, then check against it. This is some of groff's
2250: * stranger behaviours. If we encountered a custom end-scope
2251: * tag and that tag also happens to be a "real" macro, then we
2252: * need to try interpreting it again as a real macro. If it's
2253: * not, then return ignore. Else continue.
2254: */
2255:
2256: if (r->last->end) {
1.130 kristaps 2257: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.238 schwarze 2258: if (buf->buf[i] != r->last->end[j])
1.79 kristaps 2259: break;
2260:
1.238 schwarze 2261: if (r->last->end[j] == '\0' &&
2262: (buf->buf[i] == '\0' ||
2263: buf->buf[i] == ' ' ||
2264: buf->buf[i] == '\t')) {
1.79 kristaps 2265: roffnode_pop(r);
2266: roffnode_cleanscope(r);
2267:
1.238 schwarze 2268: while (buf->buf[i] == ' ' || buf->buf[i] == '\t')
1.130 kristaps 2269: i++;
2270:
2271: pos = i;
1.238 schwarze 2272: if (roff_parse(r, buf->buf, &pos, ln, ppos) !=
1.294 schwarze 2273: TOKEN_NONE)
1.277 schwarze 2274: return ROFF_RERUN;
2275: return ROFF_IGN;
1.79 kristaps 2276: }
2277: }
2278:
2279: /*
2280: * If we have no custom end-query or lookup failed, then try
2281: * pulling it out of the hashtable.
2282: */
2283:
1.238 schwarze 2284: t = roff_parse(r, buf->buf, &pos, ln, ppos);
1.79 kristaps 2285:
1.238 schwarze 2286: if (t != ROFF_cblock) {
2287: if (tok != ROFF_ig)
2288: roff_setstr(r, r->last->name, buf->buf + ppos, 2);
1.277 schwarze 2289: return ROFF_IGN;
1.106 kristaps 2290: }
1.79 kristaps 2291:
1.277 schwarze 2292: return (*roffs[t].proc)(r, t, buf, ln, ppos, pos, offs);
1.79 kristaps 2293: }
2294:
1.340 schwarze 2295: static int
1.80 kristaps 2296: roff_block_text(ROFF_ARGS)
1.78 kristaps 2297: {
2298:
1.238 schwarze 2299: if (tok != ROFF_ig)
2300: roff_setstr(r, r->last->name, buf->buf + pos, 2);
1.106 kristaps 2301:
1.277 schwarze 2302: return ROFF_IGN;
1.78 kristaps 2303: }
2304:
1.340 schwarze 2305: static int
1.82 kristaps 2306: roff_cond_sub(ROFF_ARGS)
2307: {
1.364 schwarze 2308: struct roffnode *bl;
1.340 schwarze 2309: char *ep;
2310: int endloop, irc, rr;
1.294 schwarze 2311: enum roff_tok t;
1.82 kristaps 2312:
1.340 schwarze 2313: irc = ROFF_IGN;
1.82 kristaps 2314: rr = r->last->rule;
1.340 schwarze 2315: endloop = tok != ROFF_while ? ROFF_IGN :
2316: rr ? ROFF_LOOPCONT : ROFF_LOOPEXIT;
2317: if (roffnode_cleanscope(r))
2318: irc |= endloop;
1.144 kristaps 2319:
1.196 schwarze 2320: /*
2321: * If `\}' occurs on a macro line without a preceding macro,
2322: * drop the line completely.
2323: */
2324:
1.238 schwarze 2325: ep = buf->buf + pos;
2326: if (ep[0] == '\\' && ep[1] == '}')
1.198 schwarze 2327: rr = 0;
1.196 schwarze 2328:
1.343 schwarze 2329: /*
2330: * The closing delimiter `\}' rewinds the conditional scope
2331: * but is otherwise ignored when interpreting the line.
2332: */
1.144 kristaps 2333:
1.238 schwarze 2334: while ((ep = strchr(ep, '\\')) != NULL) {
1.318 schwarze 2335: switch (ep[1]) {
2336: case '}':
2337: memmove(ep, ep + 2, strlen(ep + 2) + 1);
1.340 schwarze 2338: if (roff_ccond(r, ln, ep - buf->buf))
2339: irc |= endloop;
1.318 schwarze 2340: break;
2341: case '\0':
2342: ++ep;
2343: break;
2344: default:
2345: ep += 2;
2346: break;
1.197 schwarze 2347: }
1.177 schwarze 2348: }
1.318 schwarze 2349:
1.368 schwarze 2350: t = roff_parse(r, buf->buf, &pos, ln, ppos);
2351:
2352: /* For now, let high level macros abort .ce mode. */
2353:
2354: if (roffce_node != NULL &&
2355: (t == TOKEN_NONE || t == ROFF_Dd || t == ROFF_EQ ||
2356: t == ROFF_TH || t == ROFF_TS)) {
2357: r->man->last = roffce_node;
2358: r->man->next = ROFF_NEXT_SIBLING;
2359: roffce_lines = 0;
2360: roffce_node = NULL;
2361: }
2362:
1.318 schwarze 2363: /*
2364: * Fully handle known macros when they are structurally
2365: * required or when the conditional evaluated to true.
2366: */
2367:
1.364 schwarze 2368: if (t == ROFF_break) {
2369: if (irc & ROFF_LOOPMASK)
2370: irc = ROFF_IGN | ROFF_LOOPEXIT;
2371: else if (rr) {
2372: for (bl = r->last; bl != NULL; bl = bl->parent) {
2373: bl->rule = 0;
2374: if (bl->tok == ROFF_while)
2375: break;
2376: }
2377: }
2378: } else if (t != TOKEN_NONE &&
2379: (rr || roffs[t].flags & ROFFMAC_STRUCT))
2380: irc |= (*roffs[t].proc)(r, t, buf, ln, ppos, pos, offs);
2381: else
2382: irc |= rr ? ROFF_CONT : ROFF_IGN;
1.340 schwarze 2383: return irc;
1.82 kristaps 2384: }
2385:
1.340 schwarze 2386: static int
1.82 kristaps 2387: roff_cond_text(ROFF_ARGS)
1.78 kristaps 2388: {
1.140 kristaps 2389: char *ep;
1.340 schwarze 2390: int endloop, irc, rr;
1.82 kristaps 2391:
1.340 schwarze 2392: irc = ROFF_IGN;
1.82 kristaps 2393: rr = r->last->rule;
1.340 schwarze 2394: endloop = tok != ROFF_while ? ROFF_IGN :
2395: rr ? ROFF_LOOPCONT : ROFF_LOOPEXIT;
2396: if (roffnode_cleanscope(r))
2397: irc |= endloop;
1.82 kristaps 2398:
1.343 schwarze 2399: /*
2400: * If `\}' occurs on a text line with neither preceding
2401: * nor following characters, drop the line completely.
2402: */
2403:
1.238 schwarze 2404: ep = buf->buf + pos;
1.343 schwarze 2405: if (strcmp(ep, "\\}") == 0)
2406: rr = 0;
2407:
2408: /*
2409: * The closing delimiter `\}' rewinds the conditional scope
2410: * but is otherwise ignored when interpreting the line.
2411: */
2412:
1.238 schwarze 2413: while ((ep = strchr(ep, '\\')) != NULL) {
1.343 schwarze 2414: switch (ep[1]) {
2415: case '}':
2416: memmove(ep, ep + 2, strlen(ep + 2) + 1);
2417: if (roff_ccond(r, ln, ep - buf->buf))
1.340 schwarze 2418: irc |= endloop;
1.343 schwarze 2419: break;
2420: case '\0':
2421: ++ep;
2422: break;
2423: default:
2424: ep += 2;
2425: break;
1.197 schwarze 2426: }
1.78 kristaps 2427: }
1.340 schwarze 2428: if (rr)
2429: irc |= ROFF_CONT;
2430: return irc;
1.74 kristaps 2431: }
2432:
1.266 schwarze 2433: /* --- handling of numeric and conditional expressions -------------------- */
2434:
1.204 schwarze 2435: /*
2436: * Parse a single signed integer number. Stop at the first non-digit.
2437: * If there is at least one digit, return success and advance the
2438: * parse point, else return failure and let the parse point unchanged.
2439: * Ignore overflows, treat them just like the C language.
2440: */
1.184 schwarze 2441: static int
1.261 schwarze 2442: roff_getnum(const char *v, int *pos, int *res, int flags)
1.184 schwarze 2443: {
1.261 schwarze 2444: int myres, scaled, n, p;
1.206 schwarze 2445:
2446: if (NULL == res)
2447: res = &myres;
1.184 schwarze 2448:
2449: p = *pos;
2450: n = v[p] == '-';
1.261 schwarze 2451: if (n || v[p] == '+')
1.184 schwarze 2452: p++;
2453:
1.261 schwarze 2454: if (flags & ROFFNUM_WHITE)
2455: while (isspace((unsigned char)v[p]))
2456: p++;
2457:
1.184 schwarze 2458: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1.204 schwarze 2459: *res = 10 * *res + v[p] - '0';
1.184 schwarze 2460: if (p == *pos + n)
2461: return 0;
2462:
2463: if (n)
2464: *res = -*res;
2465:
1.254 schwarze 2466: /* Each number may be followed by one optional scaling unit. */
2467:
2468: switch (v[p]) {
2469: case 'f':
1.261 schwarze 2470: scaled = *res * 65536;
1.254 schwarze 2471: break;
2472: case 'i':
1.261 schwarze 2473: scaled = *res * 240;
1.254 schwarze 2474: break;
2475: case 'c':
1.261 schwarze 2476: scaled = *res * 240 / 2.54;
1.254 schwarze 2477: break;
2478: case 'v':
2479: case 'P':
1.261 schwarze 2480: scaled = *res * 40;
1.254 schwarze 2481: break;
2482: case 'm':
2483: case 'n':
1.261 schwarze 2484: scaled = *res * 24;
1.254 schwarze 2485: break;
2486: case 'p':
1.261 schwarze 2487: scaled = *res * 10 / 3;
1.254 schwarze 2488: break;
2489: case 'u':
1.261 schwarze 2490: scaled = *res;
1.254 schwarze 2491: break;
2492: case 'M':
1.261 schwarze 2493: scaled = *res * 6 / 25;
1.254 schwarze 2494: break;
2495: default:
1.261 schwarze 2496: scaled = *res;
1.254 schwarze 2497: p--;
2498: break;
2499: }
1.261 schwarze 2500: if (flags & ROFFNUM_SCALE)
2501: *res = scaled;
1.254 schwarze 2502:
2503: *pos = p + 1;
1.277 schwarze 2504: return 1;
1.184 schwarze 2505: }
2506:
1.198 schwarze 2507: /*
2508: * Evaluate a string comparison condition.
2509: * The first character is the delimiter.
2510: * Succeed if the string up to its second occurrence
2511: * matches the string up to its third occurence.
2512: * Advance the cursor after the third occurrence
2513: * or lacking that, to the end of the line.
2514: */
2515: static int
2516: roff_evalstrcond(const char *v, int *pos)
2517: {
2518: const char *s1, *s2, *s3;
2519: int match;
2520:
2521: match = 0;
2522: s1 = v + *pos; /* initial delimiter */
2523: s2 = s1 + 1; /* for scanning the first string */
2524: s3 = strchr(s2, *s1); /* for scanning the second string */
2525:
2526: if (NULL == s3) /* found no middle delimiter */
2527: goto out;
2528:
2529: while ('\0' != *++s3) {
2530: if (*s2 != *s3) { /* mismatch */
2531: s3 = strchr(s3, *s1);
2532: break;
2533: }
2534: if (*s3 == *s1) { /* found the final delimiter */
2535: match = 1;
2536: break;
2537: }
2538: s2++;
2539: }
2540:
2541: out:
2542: if (NULL == s3)
2543: s3 = strchr(s2, '\0');
1.242 schwarze 2544: else if (*s3 != '\0')
1.198 schwarze 2545: s3++;
2546: *pos = s3 - v;
1.277 schwarze 2547: return match;
1.198 schwarze 2548: }
2549:
1.204 schwarze 2550: /*
2551: * Evaluate an optionally negated single character, numerical,
2552: * or string condition.
2553: */
1.198 schwarze 2554: static int
1.271 schwarze 2555: roff_evalcond(struct roff *r, int ln, char *v, int *pos)
1.88 kristaps 2556: {
1.336 schwarze 2557: const char *start, *end;
2558: char *cp, *name;
2559: size_t sz;
2560: int deftype, len, number, savepos, istrue, wanttrue;
1.88 kristaps 2561:
1.198 schwarze 2562: if ('!' == v[*pos]) {
2563: wanttrue = 0;
2564: (*pos)++;
2565: } else
2566: wanttrue = 1;
2567:
1.88 kristaps 2568: switch (v[*pos]) {
1.240 schwarze 2569: case '\0':
1.277 schwarze 2570: return 0;
1.207 schwarze 2571: case 'n':
2572: case 'o':
1.88 kristaps 2573: (*pos)++;
1.277 schwarze 2574: return wanttrue;
1.207 schwarze 2575: case 'e':
2576: case 't':
1.239 schwarze 2577: case 'v':
1.88 kristaps 2578: (*pos)++;
1.277 schwarze 2579: return !wanttrue;
1.336 schwarze 2580: case 'c':
2581: do {
2582: (*pos)++;
2583: } while (v[*pos] == ' ');
2584:
2585: /*
2586: * Quirk for groff compatibility:
2587: * The horizontal tab is neither available nor unavailable.
2588: */
2589:
2590: if (v[*pos] == '\t') {
2591: (*pos)++;
2592: return 0;
2593: }
2594:
2595: /* Printable ASCII characters are available. */
2596:
2597: if (v[*pos] != '\\') {
2598: (*pos)++;
2599: return wanttrue;
2600: }
2601:
2602: end = v + ++*pos;
2603: switch (mandoc_escape(&end, &start, &len)) {
2604: case ESCAPE_SPECIAL:
2605: istrue = mchars_spec2cp(start, len) != -1;
2606: break;
2607: case ESCAPE_UNICODE:
2608: istrue = 1;
2609: break;
2610: case ESCAPE_NUMBERED:
2611: istrue = mchars_num2char(start, len) != -1;
2612: break;
2613: default:
2614: istrue = !wanttrue;
2615: break;
2616: }
2617: *pos = end - v;
2618: return istrue == wanttrue;
1.310 schwarze 2619: case 'd':
1.271 schwarze 2620: case 'r':
1.310 schwarze 2621: cp = v + *pos + 1;
2622: while (*cp == ' ')
2623: cp++;
2624: name = cp;
2625: sz = roff_getname(r, &cp, ln, cp - v);
1.315 schwarze 2626: if (sz == 0)
2627: istrue = 0;
2628: else if (v[*pos] == 'r')
2629: istrue = roff_hasregn(r, name, sz);
2630: else {
2631: deftype = ROFFDEF_ANY;
2632: roff_getstrn(r, name, sz, &deftype);
2633: istrue = !!deftype;
2634: }
1.363 schwarze 2635: *pos = (name + sz) - v;
1.310 schwarze 2636: return istrue == wanttrue;
1.88 kristaps 2637: default:
2638: break;
2639: }
2640:
1.241 schwarze 2641: savepos = *pos;
1.261 schwarze 2642: if (roff_evalnum(r, ln, v, pos, &number, ROFFNUM_SCALE))
1.277 schwarze 2643: return (number > 0) == wanttrue;
1.241 schwarze 2644: else if (*pos == savepos)
1.277 schwarze 2645: return roff_evalstrcond(v, pos) == wanttrue;
1.204 schwarze 2646: else
1.277 schwarze 2647: return 0;
1.88 kristaps 2648: }
2649:
1.340 schwarze 2650: static int
1.103 kristaps 2651: roff_line_ignore(ROFF_ARGS)
1.89 kristaps 2652: {
1.123 schwarze 2653:
1.277 schwarze 2654: return ROFF_IGN;
1.89 kristaps 2655: }
2656:
1.340 schwarze 2657: static int
1.251 schwarze 2658: roff_insec(ROFF_ARGS)
2659: {
2660:
1.350 schwarze 2661: mandoc_msg(MANDOCERR_REQ_INSEC, ln, ppos, "%s", roff_name[tok]);
1.277 schwarze 2662: return ROFF_IGN;
1.251 schwarze 2663: }
2664:
1.340 schwarze 2665: static int
1.251 schwarze 2666: roff_unsupp(ROFF_ARGS)
2667: {
2668:
1.350 schwarze 2669: mandoc_msg(MANDOCERR_REQ_UNSUPP, ln, ppos, "%s", roff_name[tok]);
1.277 schwarze 2670: return ROFF_IGN;
1.251 schwarze 2671: }
2672:
1.340 schwarze 2673: static int
1.82 kristaps 2674: roff_cond(ROFF_ARGS)
1.74 kristaps 2675: {
1.340 schwarze 2676: int irc;
1.173 schwarze 2677:
2678: roffnode_push(r, tok, NULL, ln, ppos);
1.74 kristaps 2679:
1.207 schwarze 2680: /*
1.134 kristaps 2681: * An `.el' has no conditional body: it will consume the value
2682: * of the current rstack entry set in prior `ie' calls or
1.207 schwarze 2683: * defaults to DENY.
1.134 kristaps 2684: *
2685: * If we're not an `el', however, then evaluate the conditional.
2686: */
1.133 kristaps 2687:
1.238 schwarze 2688: r->last->rule = tok == ROFF_el ?
1.207 schwarze 2689: (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
1.238 schwarze 2690: roff_evalcond(r, ln, buf->buf, &pos);
1.77 kristaps 2691:
1.134 kristaps 2692: /*
2693: * An if-else will put the NEGATION of the current evaluated
2694: * conditional into the stack of rules.
2695: */
2696:
1.238 schwarze 2697: if (tok == ROFF_ie) {
1.223 schwarze 2698: if (r->rstackpos + 1 == r->rstacksz) {
2699: r->rstacksz += 16;
2700: r->rstack = mandoc_reallocarray(r->rstack,
2701: r->rstacksz, sizeof(int));
1.134 kristaps 2702: }
1.198 schwarze 2703: r->rstack[++r->rstackpos] = !r->last->rule;
1.82 kristaps 2704: }
1.88 kristaps 2705:
2706: /* If the parent has false as its rule, then so do we. */
2707:
1.198 schwarze 2708: if (r->last->parent && !r->last->parent->rule)
2709: r->last->rule = 0;
1.88 kristaps 2710:
2711: /*
1.173 schwarze 2712: * Determine scope.
2713: * If there is nothing on the line after the conditional,
2714: * not even whitespace, use next-line scope.
1.340 schwarze 2715: * Except that .while does not support next-line scope.
1.88 kristaps 2716: */
1.74 kristaps 2717:
1.340 schwarze 2718: if (buf->buf[pos] == '\0' && tok != ROFF_while) {
1.173 schwarze 2719: r->last->endspan = 2;
2720: goto out;
2721: }
2722:
1.238 schwarze 2723: while (buf->buf[pos] == ' ')
1.173 schwarze 2724: pos++;
2725:
2726: /* An opening brace requests multiline scope. */
1.75 kristaps 2727:
1.238 schwarze 2728: if (buf->buf[pos] == '\\' && buf->buf[pos + 1] == '{') {
1.75 kristaps 2729: r->last->endspan = -1;
2730: pos += 2;
1.272 schwarze 2731: while (buf->buf[pos] == ' ')
2732: pos++;
1.173 schwarze 2733: goto out;
1.207 schwarze 2734: }
1.74 kristaps 2735:
1.77 kristaps 2736: /*
1.173 schwarze 2737: * Anything else following the conditional causes
2738: * single-line scope. Warn if the scope contains
2739: * nothing but trailing whitespace.
1.77 kristaps 2740: */
2741:
1.238 schwarze 2742: if (buf->buf[pos] == '\0')
1.350 schwarze 2743: mandoc_msg(MANDOCERR_COND_EMPTY,
2744: ln, ppos, "%s", roff_name[tok]);
1.77 kristaps 2745:
1.173 schwarze 2746: r->last->endspan = 1;
1.74 kristaps 2747:
1.173 schwarze 2748: out:
1.75 kristaps 2749: *offs = pos;
1.340 schwarze 2750: irc = ROFF_RERUN;
2751: if (tok == ROFF_while)
2752: irc |= ROFF_WHILE;
2753: return irc;
1.83 schwarze 2754: }
2755:
1.340 schwarze 2756: static int
1.92 schwarze 2757: roff_ds(ROFF_ARGS)
2758: {
1.212 schwarze 2759: char *string;
2760: const char *name;
2761: size_t namesz;
1.96 kristaps 2762:
1.251 schwarze 2763: /* Ignore groff compatibility mode for now. */
2764:
2765: if (tok == ROFF_ds1)
2766: tok = ROFF_ds;
2767: else if (tok == ROFF_as1)
2768: tok = ROFF_as;
2769:
1.96 kristaps 2770: /*
1.212 schwarze 2771: * The first word is the name of the string.
2772: * If it is empty or terminated by an escape sequence,
2773: * abort the `ds' request without defining anything.
1.96 kristaps 2774: */
1.92 schwarze 2775:
1.238 schwarze 2776: name = string = buf->buf + pos;
2777: if (*name == '\0')
1.277 schwarze 2778: return ROFF_IGN;
1.92 schwarze 2779:
1.212 schwarze 2780: namesz = roff_getname(r, &string, ln, pos);
1.363 schwarze 2781: switch (name[namesz]) {
2782: case '\\':
1.277 schwarze 2783: return ROFF_IGN;
1.363 schwarze 2784: case '\t':
2785: string = buf->buf + pos + namesz;
2786: break;
2787: default:
2788: break;
2789: }
1.212 schwarze 2790:
2791: /* Read past the initial double-quote, if any. */
1.238 schwarze 2792: if (*string == '"')
1.92 schwarze 2793: string++;
2794:
1.96 kristaps 2795: /* The rest is the value. */
1.212 schwarze 2796: roff_setstrn(&r->strtab, name, namesz, string, strlen(string),
2797: ROFF_as == tok);
1.311 schwarze 2798: roff_setstrn(&r->rentab, name, namesz, NULL, 0, 0);
1.277 schwarze 2799: return ROFF_IGN;
1.92 schwarze 2800: }
2801:
1.204 schwarze 2802: /*
2803: * Parse a single operator, one or two characters long.
2804: * If the operator is recognized, return success and advance the
2805: * parse point, else return failure and let the parse point unchanged.
2806: */
2807: static int
2808: roff_getop(const char *v, int *pos, char *res)
2809: {
2810:
2811: *res = v[*pos];
2812:
2813: switch (*res) {
1.207 schwarze 2814: case '+':
2815: case '-':
2816: case '*':
2817: case '/':
2818: case '%':
2819: case '&':
2820: case ':':
1.204 schwarze 2821: break;
2822: case '<':
2823: switch (v[*pos + 1]) {
1.207 schwarze 2824: case '=':
1.204 schwarze 2825: *res = 'l';
2826: (*pos)++;
2827: break;
1.207 schwarze 2828: case '>':
1.204 schwarze 2829: *res = '!';
2830: (*pos)++;
2831: break;
1.207 schwarze 2832: case '?':
1.204 schwarze 2833: *res = 'i';
2834: (*pos)++;
2835: break;
2836: default:
2837: break;
2838: }
2839: break;
2840: case '>':
2841: switch (v[*pos + 1]) {
1.207 schwarze 2842: case '=':
1.204 schwarze 2843: *res = 'g';
2844: (*pos)++;
2845: break;
1.207 schwarze 2846: case '?':
1.204 schwarze 2847: *res = 'a';
2848: (*pos)++;
2849: break;
2850: default:
2851: break;
2852: }
2853: break;
2854: case '=':
2855: if ('=' == v[*pos + 1])
2856: (*pos)++;
2857: break;
2858: default:
1.277 schwarze 2859: return 0;
1.204 schwarze 2860: }
2861: (*pos)++;
2862:
1.277 schwarze 2863: return *res;
1.204 schwarze 2864: }
2865:
2866: /*
2867: * Evaluate either a parenthesized numeric expression
2868: * or a single signed integer number.
2869: */
2870: static int
1.235 schwarze 2871: roff_evalpar(struct roff *r, int ln,
1.261 schwarze 2872: const char *v, int *pos, int *res, int flags)
1.204 schwarze 2873: {
2874:
2875: if ('(' != v[*pos])
1.277 schwarze 2876: return roff_getnum(v, pos, res, flags);
1.204 schwarze 2877:
2878: (*pos)++;
1.261 schwarze 2879: if ( ! roff_evalnum(r, ln, v, pos, res, flags | ROFFNUM_WHITE))
1.277 schwarze 2880: return 0;
1.204 schwarze 2881:
1.206 schwarze 2882: /*
2883: * Omission of the closing parenthesis
2884: * is an error in validation mode,
2885: * but ignored in evaluation mode.
2886: */
2887:
1.204 schwarze 2888: if (')' == v[*pos])
2889: (*pos)++;
1.206 schwarze 2890: else if (NULL == res)
1.277 schwarze 2891: return 0;
1.204 schwarze 2892:
1.277 schwarze 2893: return 1;
1.204 schwarze 2894: }
2895:
2896: /*
2897: * Evaluate a complete numeric expression.
2898: * Proceed left to right, there is no concept of precedence.
2899: */
2900: static int
1.235 schwarze 2901: roff_evalnum(struct roff *r, int ln, const char *v,
1.261 schwarze 2902: int *pos, int *res, int flags)
1.204 schwarze 2903: {
2904: int mypos, operand2;
2905: char operator;
2906:
2907: if (NULL == pos) {
2908: mypos = 0;
2909: pos = &mypos;
2910: }
2911:
1.261 schwarze 2912: if (flags & ROFFNUM_WHITE)
1.204 schwarze 2913: while (isspace((unsigned char)v[*pos]))
2914: (*pos)++;
2915:
1.261 schwarze 2916: if ( ! roff_evalpar(r, ln, v, pos, res, flags))
1.277 schwarze 2917: return 0;
1.204 schwarze 2918:
2919: while (1) {
1.261 schwarze 2920: if (flags & ROFFNUM_WHITE)
1.204 schwarze 2921: while (isspace((unsigned char)v[*pos]))
2922: (*pos)++;
2923:
2924: if ( ! roff_getop(v, pos, &operator))
2925: break;
2926:
1.261 schwarze 2927: if (flags & ROFFNUM_WHITE)
1.204 schwarze 2928: while (isspace((unsigned char)v[*pos]))
2929: (*pos)++;
2930:
1.261 schwarze 2931: if ( ! roff_evalpar(r, ln, v, pos, &operand2, flags))
1.277 schwarze 2932: return 0;
1.204 schwarze 2933:
1.261 schwarze 2934: if (flags & ROFFNUM_WHITE)
1.204 schwarze 2935: while (isspace((unsigned char)v[*pos]))
2936: (*pos)++;
1.206 schwarze 2937:
2938: if (NULL == res)
2939: continue;
1.204 schwarze 2940:
2941: switch (operator) {
1.207 schwarze 2942: case '+':
1.204 schwarze 2943: *res += operand2;
2944: break;
1.207 schwarze 2945: case '-':
1.204 schwarze 2946: *res -= operand2;
2947: break;
1.207 schwarze 2948: case '*':
1.204 schwarze 2949: *res *= operand2;
2950: break;
1.207 schwarze 2951: case '/':
1.244 schwarze 2952: if (operand2 == 0) {
1.235 schwarze 2953: mandoc_msg(MANDOCERR_DIVZERO,
1.350 schwarze 2954: ln, *pos, "%s", v);
1.234 kristaps 2955: *res = 0;
2956: break;
2957: }
1.204 schwarze 2958: *res /= operand2;
2959: break;
1.207 schwarze 2960: case '%':
1.244 schwarze 2961: if (operand2 == 0) {
2962: mandoc_msg(MANDOCERR_DIVZERO,
1.350 schwarze 2963: ln, *pos, "%s", v);
1.244 schwarze 2964: *res = 0;
2965: break;
2966: }
1.204 schwarze 2967: *res %= operand2;
2968: break;
1.207 schwarze 2969: case '<':
1.204 schwarze 2970: *res = *res < operand2;
2971: break;
1.207 schwarze 2972: case '>':
1.204 schwarze 2973: *res = *res > operand2;
2974: break;
1.207 schwarze 2975: case 'l':
1.204 schwarze 2976: *res = *res <= operand2;
2977: break;
1.207 schwarze 2978: case 'g':
1.204 schwarze 2979: *res = *res >= operand2;
2980: break;
1.207 schwarze 2981: case '=':
1.204 schwarze 2982: *res = *res == operand2;
2983: break;
1.207 schwarze 2984: case '!':
1.204 schwarze 2985: *res = *res != operand2;
2986: break;
1.207 schwarze 2987: case '&':
1.204 schwarze 2988: *res = *res && operand2;
2989: break;
1.207 schwarze 2990: case ':':
1.204 schwarze 2991: *res = *res || operand2;
2992: break;
1.207 schwarze 2993: case 'i':
1.204 schwarze 2994: if (operand2 < *res)
2995: *res = operand2;
2996: break;
1.207 schwarze 2997: case 'a':
1.204 schwarze 2998: if (operand2 > *res)
2999: *res = operand2;
3000: break;
3001: default:
3002: abort();
3003: }
3004: }
1.277 schwarze 3005: return 1;
1.204 schwarze 3006: }
3007:
1.266 schwarze 3008: /* --- register management ------------------------------------------------ */
3009:
1.180 schwarze 3010: void
1.187 schwarze 3011: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.147 kristaps 3012: {
1.327 schwarze 3013: roff_setregn(r, name, strlen(name), val, sign, INT_MIN);
1.326 schwarze 3014: }
3015:
3016: static void
3017: roff_setregn(struct roff *r, const char *name, size_t len,
1.327 schwarze 3018: int val, char sign, int step)
1.326 schwarze 3019: {
1.180 schwarze 3020: struct roffreg *reg;
3021:
3022: /* Search for an existing register with the same name. */
3023: reg = r->regtab;
3024:
1.326 schwarze 3025: while (reg != NULL && (reg->key.sz != len ||
3026: strncmp(reg->key.p, name, len) != 0))
1.180 schwarze 3027: reg = reg->next;
1.147 kristaps 3028:
1.180 schwarze 3029: if (NULL == reg) {
3030: /* Create a new register. */
3031: reg = mandoc_malloc(sizeof(struct roffreg));
1.326 schwarze 3032: reg->key.p = mandoc_strndup(name, len);
3033: reg->key.sz = len;
1.187 schwarze 3034: reg->val = 0;
1.327 schwarze 3035: reg->step = 0;
1.180 schwarze 3036: reg->next = r->regtab;
3037: r->regtab = reg;
3038: }
3039:
1.187 schwarze 3040: if ('+' == sign)
3041: reg->val += val;
3042: else if ('-' == sign)
3043: reg->val -= val;
3044: else
3045: reg->val = val;
1.327 schwarze 3046: if (step != INT_MIN)
3047: reg->step = step;
1.147 kristaps 3048: }
3049:
1.192 schwarze 3050: /*
3051: * Handle some predefined read-only number registers.
3052: * For now, return -1 if the requested register is not predefined;
3053: * in case a predefined read-only register having the value -1
3054: * were to turn up, another special value would have to be chosen.
3055: */
3056: static int
1.273 schwarze 3057: roff_getregro(const struct roff *r, const char *name)
1.192 schwarze 3058: {
3059:
3060: switch (*name) {
1.273 schwarze 3061: case '$': /* Number of arguments of the last macro evaluated. */
1.339 schwarze 3062: return r->mstackpos < 0 ? 0 : r->mstack[r->mstackpos].argc;
1.207 schwarze 3063: case 'A': /* ASCII approximation mode is always off. */
1.277 schwarze 3064: return 0;
1.207 schwarze 3065: case 'g': /* Groff compatibility mode is always on. */
1.277 schwarze 3066: return 1;
1.207 schwarze 3067: case 'H': /* Fixed horizontal resolution. */
1.277 schwarze 3068: return 24;
1.207 schwarze 3069: case 'j': /* Always adjust left margin only. */
1.277 schwarze 3070: return 0;
1.207 schwarze 3071: case 'T': /* Some output device is always defined. */
1.277 schwarze 3072: return 1;
1.207 schwarze 3073: case 'V': /* Fixed vertical resolution. */
1.277 schwarze 3074: return 40;
1.192 schwarze 3075: default:
1.277 schwarze 3076: return -1;
1.192 schwarze 3077: }
3078: }
3079:
1.181 schwarze 3080: int
1.326 schwarze 3081: roff_getreg(struct roff *r, const char *name)
1.147 kristaps 3082: {
1.327 schwarze 3083: return roff_getregn(r, name, strlen(name), '\0');
1.181 schwarze 3084: }
3085:
3086: static int
1.327 schwarze 3087: roff_getregn(struct roff *r, const char *name, size_t len, char sign)
1.181 schwarze 3088: {
3089: struct roffreg *reg;
1.192 schwarze 3090: int val;
3091:
3092: if ('.' == name[0] && 2 == len) {
1.273 schwarze 3093: val = roff_getregro(r, name + 1);
1.192 schwarze 3094: if (-1 != val)
1.277 schwarze 3095: return val;
1.192 schwarze 3096: }
1.181 schwarze 3097:
1.327 schwarze 3098: for (reg = r->regtab; reg; reg = reg->next) {
1.181 schwarze 3099: if (len == reg->key.sz &&
1.327 schwarze 3100: 0 == strncmp(name, reg->key.p, len)) {
3101: switch (sign) {
3102: case '+':
3103: reg->val += reg->step;
3104: break;
3105: case '-':
3106: reg->val -= reg->step;
3107: break;
3108: default:
3109: break;
3110: }
1.277 schwarze 3111: return reg->val;
1.327 schwarze 3112: }
3113: }
1.271 schwarze 3114:
1.327 schwarze 3115: roff_setregn(r, name, len, 0, '\0', INT_MIN);
1.277 schwarze 3116: return 0;
1.271 schwarze 3117: }
3118:
3119: static int
3120: roff_hasregn(const struct roff *r, const char *name, size_t len)
3121: {
3122: struct roffreg *reg;
3123: int val;
3124:
3125: if ('.' == name[0] && 2 == len) {
1.273 schwarze 3126: val = roff_getregro(r, name + 1);
1.271 schwarze 3127: if (-1 != val)
1.277 schwarze 3128: return 1;
1.271 schwarze 3129: }
3130:
3131: for (reg = r->regtab; reg; reg = reg->next)
3132: if (len == reg->key.sz &&
3133: 0 == strncmp(name, reg->key.p, len))
1.277 schwarze 3134: return 1;
1.147 kristaps 3135:
1.277 schwarze 3136: return 0;
1.147 kristaps 3137: }
3138:
1.180 schwarze 3139: static void
3140: roff_freereg(struct roffreg *reg)
1.147 kristaps 3141: {
1.180 schwarze 3142: struct roffreg *old_reg;
1.147 kristaps 3143:
1.180 schwarze 3144: while (NULL != reg) {
3145: free(reg->key.p);
3146: old_reg = reg;
3147: reg = reg->next;
3148: free(old_reg);
3149: }
1.147 kristaps 3150: }
1.92 schwarze 3151:
1.340 schwarze 3152: static int
1.89 kristaps 3153: roff_nr(ROFF_ARGS)
1.83 schwarze 3154: {
1.327 schwarze 3155: char *key, *val, *step;
1.212 schwarze 3156: size_t keysz;
1.327 schwarze 3157: int iv, is, len;
1.187 schwarze 3158: char sign;
1.89 kristaps 3159:
1.238 schwarze 3160: key = val = buf->buf + pos;
3161: if (*key == '\0')
1.277 schwarze 3162: return ROFF_IGN;
1.212 schwarze 3163:
3164: keysz = roff_getname(r, &val, ln, pos);
1.363 schwarze 3165: if (key[keysz] == '\\' || key[keysz] == '\t')
1.277 schwarze 3166: return ROFF_IGN;
1.89 kristaps 3167:
1.187 schwarze 3168: sign = *val;
1.238 schwarze 3169: if (sign == '+' || sign == '-')
1.187 schwarze 3170: val++;
3171:
1.327 schwarze 3172: len = 0;
3173: if (roff_evalnum(r, ln, val, &len, &iv, ROFFNUM_SCALE) == 0)
3174: return ROFF_IGN;
3175:
3176: step = val + len;
3177: while (isspace((unsigned char)*step))
3178: step++;
3179: if (roff_evalnum(r, ln, step, NULL, &is, 0) == 0)
3180: is = INT_MIN;
1.109 kristaps 3181:
1.327 schwarze 3182: roff_setregn(r, key, keysz, iv, sign, is);
1.277 schwarze 3183: return ROFF_IGN;
1.203 schwarze 3184: }
3185:
1.340 schwarze 3186: static int
1.203 schwarze 3187: roff_rr(ROFF_ARGS)
3188: {
3189: struct roffreg *reg, **prev;
1.212 schwarze 3190: char *name, *cp;
3191: size_t namesz;
1.203 schwarze 3192:
1.238 schwarze 3193: name = cp = buf->buf + pos;
3194: if (*name == '\0')
1.277 schwarze 3195: return ROFF_IGN;
1.212 schwarze 3196: namesz = roff_getname(r, &cp, ln, pos);
3197: name[namesz] = '\0';
1.203 schwarze 3198:
3199: prev = &r->regtab;
3200: while (1) {
3201: reg = *prev;
1.238 schwarze 3202: if (reg == NULL || !strcmp(name, reg->key.p))
1.203 schwarze 3203: break;
3204: prev = ®->next;
3205: }
1.238 schwarze 3206: if (reg != NULL) {
1.203 schwarze 3207: *prev = reg->next;
3208: free(reg->key.p);
3209: free(reg);
3210: }
1.277 schwarze 3211: return ROFF_IGN;
1.122 schwarze 3212: }
3213:
1.266 schwarze 3214: /* --- handler functions for roff requests -------------------------------- */
3215:
1.340 schwarze 3216: static int
1.122 schwarze 3217: roff_rm(ROFF_ARGS)
3218: {
3219: const char *name;
3220: char *cp;
1.212 schwarze 3221: size_t namesz;
1.122 schwarze 3222:
1.238 schwarze 3223: cp = buf->buf + pos;
3224: while (*cp != '\0') {
1.212 schwarze 3225: name = cp;
1.238 schwarze 3226: namesz = roff_getname(r, &cp, ln, (int)(cp - buf->buf));
1.212 schwarze 3227: roff_setstrn(&r->strtab, name, namesz, NULL, 0, 0);
1.311 schwarze 3228: roff_setstrn(&r->rentab, name, namesz, NULL, 0, 0);
1.363 schwarze 3229: if (name[namesz] == '\\' || name[namesz] == '\t')
1.212 schwarze 3230: break;
1.122 schwarze 3231: }
1.277 schwarze 3232: return ROFF_IGN;
1.178 schwarze 3233: }
3234:
1.340 schwarze 3235: static int
1.178 schwarze 3236: roff_it(ROFF_ARGS)
3237: {
3238: int iv;
3239:
3240: /* Parse the number of lines. */
1.261 schwarze 3241:
3242: if ( ! roff_evalnum(r, ln, buf->buf, &pos, &iv, 0)) {
1.350 schwarze 3243: mandoc_msg(MANDOCERR_IT_NONUM,
3244: ln, ppos, "%s", buf->buf + 1);
1.277 schwarze 3245: return ROFF_IGN;
1.178 schwarze 3246: }
3247:
1.262 schwarze 3248: while (isspace((unsigned char)buf->buf[pos]))
3249: pos++;
3250:
3251: /*
3252: * Arm the input line trap.
3253: * Special-casing "an-trap" is an ugly workaround to cope
3254: * with DocBook stupidly fiddling with man(7) internals.
3255: */
1.261 schwarze 3256:
1.178 schwarze 3257: roffit_lines = iv;
1.262 schwarze 3258: roffit_macro = mandoc_strdup(iv != 1 ||
3259: strcmp(buf->buf + pos, "an-trap") ?
3260: buf->buf + pos : "br");
1.277 schwarze 3261: return ROFF_IGN;
1.175 schwarze 3262: }
3263:
1.340 schwarze 3264: static int
1.175 schwarze 3265: roff_Dd(ROFF_ARGS)
3266: {
1.315 schwarze 3267: int mask;
3268: enum roff_tok t, te;
1.227 schwarze 3269:
1.315 schwarze 3270: switch (tok) {
3271: case ROFF_Dd:
3272: tok = MDOC_Dd;
3273: te = MDOC_MAX;
3274: if (r->format == 0)
3275: r->format = MPARSE_MDOC;
3276: mask = MPARSE_MDOC | MPARSE_QUICK;
3277: break;
3278: case ROFF_TH:
3279: tok = MAN_TH;
3280: te = MAN_MAX;
3281: if (r->format == 0)
3282: r->format = MPARSE_MAN;
3283: mask = MPARSE_QUICK;
3284: break;
3285: default:
3286: abort();
3287: }
3288: if ((r->options & mask) == 0)
3289: for (t = tok; t < te; t++)
3290: roff_setstr(r, roff_name[t], NULL, 0);
1.277 schwarze 3291: return ROFF_CONT;
1.109 kristaps 3292: }
3293:
1.340 schwarze 3294: static int
1.109 kristaps 3295: roff_TE(ROFF_ARGS)
3296: {
1.361 schwarze 3297: r->man->flags &= ~ROFF_NONOFILL;
1.321 schwarze 3298: if (r->tbl == NULL) {
1.350 schwarze 3299: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "TE");
1.321 schwarze 3300: return ROFF_IGN;
3301: }
1.346 schwarze 3302: if (tbl_end(r->tbl, 0) == 0) {
1.321 schwarze 3303: r->tbl = NULL;
1.258 schwarze 3304: free(buf->buf);
3305: buf->buf = mandoc_strdup(".sp");
3306: buf->sz = 4;
1.329 schwarze 3307: *offs = 0;
1.277 schwarze 3308: return ROFF_REPARSE;
1.258 schwarze 3309: }
1.321 schwarze 3310: r->tbl = NULL;
1.277 schwarze 3311: return ROFF_IGN;
1.112 kristaps 3312: }
3313:
1.340 schwarze 3314: static int
1.112 kristaps 3315: roff_T_(ROFF_ARGS)
3316: {
3317:
3318: if (NULL == r->tbl)
1.350 schwarze 3319: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "T&");
1.112 kristaps 3320: else
1.296 schwarze 3321: tbl_restart(ln, ppos, r->tbl);
1.112 kristaps 3322:
1.277 schwarze 3323: return ROFF_IGN;
1.109 kristaps 3324: }
3325:
1.230 schwarze 3326: /*
3327: * Handle in-line equation delimiters.
3328: */
1.340 schwarze 3329: static int
1.238 schwarze 3330: roff_eqndelim(struct roff *r, struct buf *buf, int pos)
1.151 kristaps 3331: {
1.233 schwarze 3332: char *cp1, *cp2;
3333: const char *bef_pr, *bef_nl, *mac, *aft_nl, *aft_pr;
1.151 kristaps 3334:
1.230 schwarze 3335: /*
3336: * Outside equations, look for an opening delimiter.
3337: * If we are inside an equation, we already know it is
3338: * in-line, or this function wouldn't have been called;
3339: * so look for a closing delimiter.
3340: */
3341:
1.238 schwarze 3342: cp1 = buf->buf + pos;
1.230 schwarze 3343: cp2 = strchr(cp1, r->eqn == NULL ?
3344: r->last_eqn->odelim : r->last_eqn->cdelim);
3345: if (cp2 == NULL)
1.277 schwarze 3346: return ROFF_CONT;
1.230 schwarze 3347:
1.233 schwarze 3348: *cp2++ = '\0';
3349: bef_pr = bef_nl = aft_nl = aft_pr = "";
3350:
3351: /* Handle preceding text, protecting whitespace. */
3352:
1.238 schwarze 3353: if (*buf->buf != '\0') {
1.233 schwarze 3354: if (r->eqn == NULL)
3355: bef_pr = "\\&";
3356: bef_nl = "\n";
3357: }
3358:
3359: /*
3360: * Prepare replacing the delimiter with an equation macro
3361: * and drop leading white space from the equation.
3362: */
1.230 schwarze 3363:
1.233 schwarze 3364: if (r->eqn == NULL) {
3365: while (*cp2 == ' ')
3366: cp2++;
3367: mac = ".EQ";
3368: } else
3369: mac = ".EN";
3370:
3371: /* Handle following text, protecting whitespace. */
3372:
3373: if (*cp2 != '\0') {
3374: aft_nl = "\n";
3375: if (r->eqn != NULL)
3376: aft_pr = "\\&";
3377: }
3378:
3379: /* Do the actual replacement. */
3380:
1.238 schwarze 3381: buf->sz = mandoc_asprintf(&cp1, "%s%s%s%s%s%s%s", buf->buf,
1.233 schwarze 3382: bef_pr, bef_nl, mac, aft_nl, aft_pr, cp2) + 1;
1.238 schwarze 3383: free(buf->buf);
3384: buf->buf = cp1;
1.230 schwarze 3385:
3386: /* Toggle the in-line state of the eqn subsystem. */
3387:
3388: r->eqn_inline = r->eqn == NULL;
1.277 schwarze 3389: return ROFF_REPARSE;
1.151 kristaps 3390: }
3391:
1.340 schwarze 3392: static int
1.235 schwarze 3393: roff_EQ(ROFF_ARGS)
1.125 kristaps 3394: {
1.319 schwarze 3395: struct roff_node *n;
3396:
1.356 schwarze 3397: if (r->man->meta.macroset == MACROSET_MAN)
1.322 schwarze 3398: man_breakscope(r->man, ROFF_EQ);
1.319 schwarze 3399: n = roff_node_alloc(r->man, ln, ppos, ROFFT_EQN, TOKEN_NONE);
3400: if (ln > r->man->last->line)
3401: n->flags |= NODE_LINE;
1.348 schwarze 3402: n->eqn = eqn_box_new();
1.319 schwarze 3403: roff_node_append(r->man, n);
3404: r->man->next = ROFF_NEXT_SIBLING;
1.125 kristaps 3405:
1.238 schwarze 3406: assert(r->eqn == NULL);
1.319 schwarze 3407: if (r->last_eqn == NULL)
1.351 schwarze 3408: r->last_eqn = eqn_alloc();
1.319 schwarze 3409: else
3410: eqn_reset(r->last_eqn);
3411: r->eqn = r->last_eqn;
3412: r->eqn->node = n;
1.151 kristaps 3413:
1.238 schwarze 3414: if (buf->buf[pos] != '\0')
1.350 schwarze 3415: mandoc_msg(MANDOCERR_ARG_SKIP, ln, pos,
1.238 schwarze 3416: ".EQ %s", buf->buf + pos);
1.151 kristaps 3417:
1.277 schwarze 3418: return ROFF_IGN;
1.125 kristaps 3419: }
3420:
1.340 schwarze 3421: static int
1.125 kristaps 3422: roff_EN(ROFF_ARGS)
3423: {
1.319 schwarze 3424: if (r->eqn != NULL) {
3425: eqn_parse(r->eqn);
3426: r->eqn = NULL;
3427: } else
1.350 schwarze 3428: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, ppos, "EN");
1.319 schwarze 3429: if (buf->buf[pos] != '\0')
1.350 schwarze 3430: mandoc_msg(MANDOCERR_ARG_SKIP, ln, pos,
1.319 schwarze 3431: "EN %s", buf->buf + pos);
1.277 schwarze 3432: return ROFF_IGN;
1.125 kristaps 3433: }
3434:
1.340 schwarze 3435: static int
1.109 kristaps 3436: roff_TS(ROFF_ARGS)
3437: {
1.321 schwarze 3438: if (r->tbl != NULL) {
1.350 schwarze 3439: mandoc_msg(MANDOCERR_BLK_BROKEN, ln, ppos, "TS breaks TS");
1.346 schwarze 3440: tbl_end(r->tbl, 0);
1.115 kristaps 3441: }
1.361 schwarze 3442: r->man->flags |= ROFF_NONOFILL;
1.351 schwarze 3443: r->tbl = tbl_alloc(ppos, ln, r->last_tbl);
1.346 schwarze 3444: if (r->last_tbl == NULL)
1.321 schwarze 3445: r->first_tbl = r->tbl;
3446: r->last_tbl = r->tbl;
1.297 schwarze 3447: return ROFF_IGN;
3448: }
3449:
1.340 schwarze 3450: static int
1.358 schwarze 3451: roff_noarg(ROFF_ARGS)
3452: {
3453: if (r->man->flags & (MAN_BLINE | MAN_ELINE))
3454: man_breakscope(r->man, tok);
3455: if (tok == ROFF_brp)
3456: tok = ROFF_br;
3457: roff_elem_alloc(r->man, ln, ppos, tok);
3458: if (buf->buf[pos] != '\0')
3459: mandoc_msg(MANDOCERR_ARG_SKIP, ln, pos,
3460: "%s %s", roff_name[tok], buf->buf + pos);
3461: if (tok == ROFF_nf)
3462: r->man->flags |= ROFF_NOFILL;
3463: else if (tok == ROFF_fi)
3464: r->man->flags &= ~ROFF_NOFILL;
3465: r->man->last->flags |= NODE_LINE | NODE_VALID | NODE_ENDED;
3466: r->man->next = ROFF_NEXT_SIBLING;
3467: return ROFF_IGN;
3468: }
3469:
3470: static int
1.297 schwarze 3471: roff_onearg(ROFF_ARGS)
3472: {
3473: struct roff_node *n;
3474: char *cp;
1.305 schwarze 3475: int npos;
1.297 schwarze 3476:
1.302 schwarze 3477: if (r->man->flags & (MAN_BLINE | MAN_ELINE) &&
1.320 schwarze 3478: (tok == ROFF_ce || tok == ROFF_rj || tok == ROFF_sp ||
3479: tok == ROFF_ti))
1.302 schwarze 3480: man_breakscope(r->man, tok);
3481:
1.309 schwarze 3482: if (roffce_node != NULL && (tok == ROFF_ce || tok == ROFF_rj)) {
1.305 schwarze 3483: r->man->last = roffce_node;
3484: r->man->next = ROFF_NEXT_SIBLING;
3485: }
3486:
1.297 schwarze 3487: roff_elem_alloc(r->man, ln, ppos, tok);
3488: n = r->man->last;
3489:
3490: cp = buf->buf + pos;
3491: if (*cp != '\0') {
3492: while (*cp != '\0' && *cp != ' ')
3493: cp++;
3494: while (*cp == ' ')
3495: *cp++ = '\0';
3496: if (*cp != '\0')
1.350 schwarze 3497: mandoc_msg(MANDOCERR_ARG_EXCESS,
3498: ln, (int)(cp - buf->buf),
1.297 schwarze 3499: "%s ... %s", roff_name[tok], cp);
3500: roff_word_alloc(r->man, ln, pos, buf->buf + pos);
1.300 schwarze 3501: }
3502:
1.309 schwarze 3503: if (tok == ROFF_ce || tok == ROFF_rj) {
3504: if (r->man->last->type == ROFFT_ELEM) {
1.305 schwarze 3505: roff_word_alloc(r->man, ln, pos, "1");
3506: r->man->last->flags |= NODE_NOSRC;
3507: }
3508: npos = 0;
3509: if (roff_evalnum(r, ln, r->man->last->string, &npos,
3510: &roffce_lines, 0) == 0) {
1.350 schwarze 3511: mandoc_msg(MANDOCERR_CE_NONUM,
3512: ln, pos, "ce %s", buf->buf + pos);
1.305 schwarze 3513: roffce_lines = 1;
3514: }
3515: if (roffce_lines < 1) {
3516: r->man->last = r->man->last->parent;
3517: roffce_node = NULL;
3518: roffce_lines = 0;
3519: } else
3520: roffce_node = r->man->last->parent;
3521: } else {
3522: n->flags |= NODE_VALID | NODE_ENDED;
3523: r->man->last = n;
3524: }
3525: n->flags |= NODE_LINE;
1.300 schwarze 3526: r->man->next = ROFF_NEXT_SIBLING;
3527: return ROFF_IGN;
3528: }
3529:
1.340 schwarze 3530: static int
1.300 schwarze 3531: roff_manyarg(ROFF_ARGS)
3532: {
3533: struct roff_node *n;
3534: char *sp, *ep;
3535:
3536: roff_elem_alloc(r->man, ln, ppos, tok);
3537: n = r->man->last;
3538:
3539: for (sp = ep = buf->buf + pos; *sp != '\0'; sp = ep) {
3540: while (*ep != '\0' && *ep != ' ')
3541: ep++;
3542: while (*ep == ' ')
3543: *ep++ = '\0';
3544: roff_word_alloc(r->man, ln, sp - buf->buf, sp);
1.297 schwarze 3545: }
3546:
3547: n->flags |= NODE_LINE | NODE_VALID | NODE_ENDED;
3548: r->man->last = n;
3549: r->man->next = ROFF_NEXT_SIBLING;
1.277 schwarze 3550: return ROFF_IGN;
1.251 schwarze 3551: }
3552:
1.340 schwarze 3553: static int
1.311 schwarze 3554: roff_als(ROFF_ARGS)
3555: {
3556: char *oldn, *newn, *end, *value;
3557: size_t oldsz, newsz, valsz;
3558:
3559: newn = oldn = buf->buf + pos;
3560: if (*newn == '\0')
3561: return ROFF_IGN;
3562:
3563: newsz = roff_getname(r, &oldn, ln, pos);
1.363 schwarze 3564: if (newn[newsz] == '\\' || newn[newsz] == '\t' || *oldn == '\0')
1.311 schwarze 3565: return ROFF_IGN;
3566:
3567: end = oldn;
3568: oldsz = roff_getname(r, &end, ln, oldn - buf->buf);
3569: if (oldsz == 0)
3570: return ROFF_IGN;
3571:
1.338 schwarze 3572: valsz = mandoc_asprintf(&value, ".%.*s \\$@\\\"\n",
1.315 schwarze 3573: (int)oldsz, oldn);
1.311 schwarze 3574: roff_setstrn(&r->strtab, newn, newsz, value, valsz, 0);
3575: roff_setstrn(&r->rentab, newn, newsz, NULL, 0, 0);
3576: free(value);
1.364 schwarze 3577: return ROFF_IGN;
3578: }
3579:
3580: /*
3581: * The .break request only makes sense inside conditionals,
3582: * and that case is already handled in roff_cond_sub().
3583: */
3584: static int
3585: roff_break(ROFF_ARGS)
3586: {
3587: mandoc_msg(MANDOCERR_BLK_NOTOPEN, ln, pos, "break");
1.296 schwarze 3588: return ROFF_IGN;
1.92 schwarze 3589: }
3590:
1.340 schwarze 3591: static int
1.174 kristaps 3592: roff_cc(ROFF_ARGS)
3593: {
3594: const char *p;
3595:
1.238 schwarze 3596: p = buf->buf + pos;
1.174 kristaps 3597:
1.238 schwarze 3598: if (*p == '\0' || (r->control = *p++) == '.')
1.303 schwarze 3599: r->control = '\0';
1.174 kristaps 3600:
1.238 schwarze 3601: if (*p != '\0')
1.350 schwarze 3602: mandoc_msg(MANDOCERR_ARG_EXCESS,
1.260 schwarze 3603: ln, p - buf->buf, "cc ... %s", p);
1.174 kristaps 3604:
1.341 schwarze 3605: return ROFF_IGN;
3606: }
3607:
3608: static int
3609: roff_char(ROFF_ARGS)
3610: {
3611: const char *p, *kp, *vp;
3612: size_t ksz, vsz;
3613: int font;
3614:
3615: /* Parse the character to be replaced. */
3616:
3617: kp = buf->buf + pos;
3618: p = kp + 1;
3619: if (*kp == '\0' || (*kp == '\\' &&
3620: mandoc_escape(&p, NULL, NULL) != ESCAPE_SPECIAL) ||
3621: (*p != ' ' && *p != '\0')) {
1.350 schwarze 3622: mandoc_msg(MANDOCERR_CHAR_ARG, ln, pos, "char %s", kp);
1.341 schwarze 3623: return ROFF_IGN;
3624: }
3625: ksz = p - kp;
3626: while (*p == ' ')
3627: p++;
3628:
3629: /*
3630: * If the replacement string contains a font escape sequence,
3631: * we have to restore the font at the end.
3632: */
3633:
3634: vp = p;
3635: vsz = strlen(p);
3636: font = 0;
3637: while (*p != '\0') {
3638: if (*p++ != '\\')
3639: continue;
3640: switch (mandoc_escape(&p, NULL, NULL)) {
3641: case ESCAPE_FONT:
3642: case ESCAPE_FONTROMAN:
3643: case ESCAPE_FONTITALIC:
3644: case ESCAPE_FONTBOLD:
3645: case ESCAPE_FONTBI:
1.342 schwarze 3646: case ESCAPE_FONTCW:
1.341 schwarze 3647: case ESCAPE_FONTPREV:
3648: font++;
3649: break;
3650: default:
3651: break;
3652: }
3653: }
3654: if (font > 1)
1.350 schwarze 3655: mandoc_msg(MANDOCERR_CHAR_FONT,
3656: ln, (int)(vp - buf->buf), "%s", vp);
1.341 schwarze 3657:
3658: /*
3659: * Approximate the effect of .char using the .tr tables.
3660: * XXX In groff, .char and .tr interact differently.
3661: */
3662:
3663: if (ksz == 1) {
3664: if (r->xtab == NULL)
3665: r->xtab = mandoc_calloc(128, sizeof(*r->xtab));
3666: assert((unsigned int)*kp < 128);
3667: free(r->xtab[(int)*kp].p);
3668: r->xtab[(int)*kp].sz = mandoc_asprintf(&r->xtab[(int)*kp].p,
3669: "%s%s", vp, font ? "\fP" : "");
3670: } else {
3671: roff_setstrn(&r->xmbtab, kp, ksz, vp, vsz, 0);
3672: if (font)
3673: roff_setstrn(&r->xmbtab, kp, ksz, "\\fP", 3, 1);
3674: }
1.277 schwarze 3675: return ROFF_IGN;
1.174 kristaps 3676: }
3677:
1.340 schwarze 3678: static int
1.303 schwarze 3679: roff_ec(ROFF_ARGS)
3680: {
3681: const char *p;
3682:
3683: p = buf->buf + pos;
3684: if (*p == '\0')
3685: r->escape = '\\';
3686: else {
3687: r->escape = *p;
3688: if (*++p != '\0')
1.350 schwarze 3689: mandoc_msg(MANDOCERR_ARG_EXCESS, ln,
3690: (int)(p - buf->buf), "ec ... %s", p);
1.303 schwarze 3691: }
3692: return ROFF_IGN;
3693: }
3694:
1.340 schwarze 3695: static int
1.303 schwarze 3696: roff_eo(ROFF_ARGS)
3697: {
3698: r->escape = '\0';
3699: if (buf->buf[pos] != '\0')
1.350 schwarze 3700: mandoc_msg(MANDOCERR_ARG_SKIP,
1.303 schwarze 3701: ln, pos, "eo %s", buf->buf + pos);
3702: return ROFF_IGN;
1.330 schwarze 3703: }
3704:
1.340 schwarze 3705: static int
1.330 schwarze 3706: roff_nop(ROFF_ARGS)
3707: {
3708: while (buf->buf[pos] == ' ')
3709: pos++;
3710: *offs = pos;
3711: return ROFF_RERUN;
1.303 schwarze 3712: }
3713:
1.340 schwarze 3714: static int
1.164 kristaps 3715: roff_tr(ROFF_ARGS)
3716: {
3717: const char *p, *first, *second;
3718: size_t fsz, ssz;
3719: enum mandoc_esc esc;
3720:
1.238 schwarze 3721: p = buf->buf + pos;
1.164 kristaps 3722:
1.238 schwarze 3723: if (*p == '\0') {
1.350 schwarze 3724: mandoc_msg(MANDOCERR_REQ_EMPTY, ln, ppos, "tr");
1.277 schwarze 3725: return ROFF_IGN;
1.164 kristaps 3726: }
3727:
1.238 schwarze 3728: while (*p != '\0') {
1.164 kristaps 3729: fsz = ssz = 1;
3730:
3731: first = p++;
1.238 schwarze 3732: if (*first == '\\') {
1.164 kristaps 3733: esc = mandoc_escape(&p, NULL, NULL);
1.238 schwarze 3734: if (esc == ESCAPE_ERROR) {
1.350 schwarze 3735: mandoc_msg(MANDOCERR_ESC_BAD, ln,
3736: (int)(p - buf->buf), "%s", first);
1.277 schwarze 3737: return ROFF_IGN;
1.164 kristaps 3738: }
3739: fsz = (size_t)(p - first);
3740: }
3741:
3742: second = p++;
1.238 schwarze 3743: if (*second == '\\') {
1.164 kristaps 3744: esc = mandoc_escape(&p, NULL, NULL);
1.238 schwarze 3745: if (esc == ESCAPE_ERROR) {
1.350 schwarze 3746: mandoc_msg(MANDOCERR_ESC_BAD, ln,
3747: (int)(p - buf->buf), "%s", second);
1.277 schwarze 3748: return ROFF_IGN;
1.164 kristaps 3749: }
3750: ssz = (size_t)(p - second);
1.238 schwarze 3751: } else if (*second == '\0') {
1.350 schwarze 3752: mandoc_msg(MANDOCERR_TR_ODD, ln,
3753: (int)(first - buf->buf), "tr %s", first);
1.164 kristaps 3754: second = " ";
1.165 kristaps 3755: p--;
1.164 kristaps 3756: }
3757:
1.167 kristaps 3758: if (fsz > 1) {
1.207 schwarze 3759: roff_setstrn(&r->xmbtab, first, fsz,
3760: second, ssz, 0);
1.167 kristaps 3761: continue;
3762: }
3763:
1.238 schwarze 3764: if (r->xtab == NULL)
1.207 schwarze 3765: r->xtab = mandoc_calloc(128,
3766: sizeof(struct roffstr));
1.167 kristaps 3767:
3768: free(r->xtab[(int)*first].p);
3769: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
3770: r->xtab[(int)*first].sz = ssz;
1.164 kristaps 3771: }
3772:
1.277 schwarze 3773: return ROFF_IGN;
1.164 kristaps 3774: }
3775:
1.339 schwarze 3776: /*
3777: * Implementation of the .return request.
3778: * There is no need to call roff_userret() from here.
3779: * The read module will call that after rewinding the reader stack
3780: * to the place from where the current macro was called.
3781: */
1.340 schwarze 3782: static int
1.339 schwarze 3783: roff_return(ROFF_ARGS)
3784: {
3785: if (r->mstackpos >= 0)
1.340 schwarze 3786: return ROFF_IGN | ROFF_USERRET;
1.339 schwarze 3787:
1.350 schwarze 3788: mandoc_msg(MANDOCERR_REQ_NOMAC, ln, ppos, "return");
1.339 schwarze 3789: return ROFF_IGN;
3790: }
3791:
1.340 schwarze 3792: static int
1.306 schwarze 3793: roff_rn(ROFF_ARGS)
3794: {
3795: const char *value;
3796: char *oldn, *newn, *end;
3797: size_t oldsz, newsz;
1.315 schwarze 3798: int deftype;
1.306 schwarze 3799:
3800: oldn = newn = buf->buf + pos;
3801: if (*oldn == '\0')
3802: return ROFF_IGN;
3803:
3804: oldsz = roff_getname(r, &newn, ln, pos);
1.363 schwarze 3805: if (oldn[oldsz] == '\\' || oldn[oldsz] == '\t' || *newn == '\0')
1.306 schwarze 3806: return ROFF_IGN;
3807:
3808: end = newn;
3809: newsz = roff_getname(r, &end, ln, newn - buf->buf);
3810: if (newsz == 0)
3811: return ROFF_IGN;
3812:
1.315 schwarze 3813: deftype = ROFFDEF_ANY;
3814: value = roff_getstrn(r, oldn, oldsz, &deftype);
3815: switch (deftype) {
3816: case ROFFDEF_USER:
1.306 schwarze 3817: roff_setstrn(&r->strtab, newn, newsz, value, strlen(value), 0);
3818: roff_setstrn(&r->strtab, oldn, oldsz, NULL, 0, 0);
3819: roff_setstrn(&r->rentab, newn, newsz, NULL, 0, 0);
1.315 schwarze 3820: break;
3821: case ROFFDEF_PRE:
3822: roff_setstrn(&r->strtab, newn, newsz, value, strlen(value), 0);
3823: roff_setstrn(&r->rentab, newn, newsz, NULL, 0, 0);
3824: break;
3825: case ROFFDEF_REN:
1.306 schwarze 3826: roff_setstrn(&r->rentab, newn, newsz, value, strlen(value), 0);
3827: roff_setstrn(&r->rentab, oldn, oldsz, NULL, 0, 0);
1.315 schwarze 3828: roff_setstrn(&r->strtab, newn, newsz, NULL, 0, 0);
3829: break;
3830: case ROFFDEF_STD:
1.306 schwarze 3831: roff_setstrn(&r->rentab, newn, newsz, oldn, oldsz, 0);
1.315 schwarze 3832: roff_setstrn(&r->strtab, newn, newsz, NULL, 0, 0);
3833: break;
3834: default:
3835: roff_setstrn(&r->strtab, newn, newsz, NULL, 0, 0);
3836: roff_setstrn(&r->rentab, newn, newsz, NULL, 0, 0);
3837: break;
3838: }
1.306 schwarze 3839: return ROFF_IGN;
3840: }
3841:
1.340 schwarze 3842: static int
1.339 schwarze 3843: roff_shift(ROFF_ARGS)
3844: {
3845: struct mctx *ctx;
3846: int levels, i;
3847:
3848: levels = 1;
3849: if (buf->buf[pos] != '\0' &&
3850: roff_evalnum(r, ln, buf->buf, &pos, &levels, 0) == 0) {
1.350 schwarze 3851: mandoc_msg(MANDOCERR_CE_NONUM,
1.339 schwarze 3852: ln, pos, "shift %s", buf->buf + pos);
3853: levels = 1;
3854: }
3855: if (r->mstackpos < 0) {
1.350 schwarze 3856: mandoc_msg(MANDOCERR_REQ_NOMAC, ln, ppos, "shift");
1.339 schwarze 3857: return ROFF_IGN;
3858: }
3859: ctx = r->mstack + r->mstackpos;
3860: if (levels > ctx->argc) {
1.350 schwarze 3861: mandoc_msg(MANDOCERR_SHIFT,
1.339 schwarze 3862: ln, pos, "%d, but max is %d", levels, ctx->argc);
3863: levels = ctx->argc;
3864: }
3865: if (levels == 0)
3866: return ROFF_IGN;
3867: for (i = 0; i < levels; i++)
3868: free(ctx->argv[i]);
3869: ctx->argc -= levels;
3870: for (i = 0; i < ctx->argc; i++)
3871: ctx->argv[i] = ctx->argv[i + levels];
3872: return ROFF_IGN;
3873: }
3874:
1.340 schwarze 3875: static int
1.105 kristaps 3876: roff_so(ROFF_ARGS)
3877: {
1.249 schwarze 3878: char *name, *cp;
1.105 kristaps 3879:
1.238 schwarze 3880: name = buf->buf + pos;
1.350 schwarze 3881: mandoc_msg(MANDOCERR_SO, ln, ppos, "so %s", name);
1.105 kristaps 3882:
3883: /*
3884: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
3885: * opening anything that's not in our cwd or anything beneath
3886: * it. Thus, explicitly disallow traversing up the file-system
3887: * or using absolute paths.
3888: */
3889:
1.238 schwarze 3890: if (*name == '/' || strstr(name, "../") || strstr(name, "/..")) {
1.350 schwarze 3891: mandoc_msg(MANDOCERR_SO_PATH, ln, ppos, ".so %s", name);
1.249 schwarze 3892: buf->sz = mandoc_asprintf(&cp,
3893: ".sp\nSee the file %s.\n.sp", name) + 1;
3894: free(buf->buf);
3895: buf->buf = cp;
3896: *offs = 0;
1.277 schwarze 3897: return ROFF_REPARSE;
1.105 kristaps 3898: }
3899:
3900: *offs = pos;
1.277 schwarze 3901: return ROFF_SO;
1.105 kristaps 3902: }
1.92 schwarze 3903:
1.266 schwarze 3904: /* --- user defined strings and macros ------------------------------------ */
3905:
1.340 schwarze 3906: static int
1.106 kristaps 3907: roff_userdef(ROFF_ARGS)
1.99 kristaps 3908: {
1.339 schwarze 3909: struct mctx *ctx;
3910: char *arg, *ap, *dst, *src;
3911: size_t sz;
3912:
1.365 schwarze 3913: /* If the macro is empty, ignore it altogether. */
3914:
3915: if (*r->current_string == '\0')
3916: return ROFF_IGN;
3917:
1.339 schwarze 3918: /* Initialize a new macro stack context. */
3919:
3920: if (++r->mstackpos == r->mstacksz) {
3921: r->mstack = mandoc_recallocarray(r->mstack,
3922: r->mstacksz, r->mstacksz + 8, sizeof(*r->mstack));
3923: r->mstacksz += 8;
3924: }
3925: ctx = r->mstack + r->mstackpos;
3926: ctx->argsz = 0;
3927: ctx->argc = 0;
3928: ctx->argv = NULL;
3929:
3930: /*
3931: * Collect pointers to macro argument strings,
3932: * NUL-terminating them and escaping quotes.
3933: */
3934:
3935: src = buf->buf + pos;
3936: while (*src != '\0') {
3937: if (ctx->argc == ctx->argsz) {
3938: ctx->argsz += 8;
3939: ctx->argv = mandoc_reallocarray(ctx->argv,
3940: ctx->argsz, sizeof(*ctx->argv));
3941: }
1.355 schwarze 3942: arg = roff_getarg(r, &src, ln, &pos);
1.339 schwarze 3943: sz = 1; /* For the terminating NUL. */
3944: for (ap = arg; *ap != '\0'; ap++)
3945: sz += *ap == '"' ? 4 : 1;
3946: ctx->argv[ctx->argc++] = dst = mandoc_malloc(sz);
3947: for (ap = arg; *ap != '\0'; ap++) {
3948: if (*ap == '"') {
3949: memcpy(dst, "\\(dq", 4);
3950: dst += 4;
3951: } else
3952: *dst++ = *ap;
1.106 kristaps 3953: }
1.339 schwarze 3954: *dst = '\0';
1.355 schwarze 3955: free(arg);
1.337 schwarze 3956: }
3957:
1.339 schwarze 3958: /* Replace the macro invocation by the macro definition. */
1.263 schwarze 3959:
1.238 schwarze 3960: free(buf->buf);
1.339 schwarze 3961: buf->buf = mandoc_strdup(r->current_string);
3962: buf->sz = strlen(buf->buf) + 1;
1.248 schwarze 3963: *offs = 0;
1.106 kristaps 3964:
1.365 schwarze 3965: return buf->buf[buf->sz - 2] == '\n' ?
1.340 schwarze 3966: ROFF_REPARSE | ROFF_USERCALL : ROFF_IGN | ROFF_APPEND;
1.99 kristaps 3967: }
1.121 schwarze 3968:
1.306 schwarze 3969: /*
3970: * Calling a high-level macro that was renamed with .rn.
3971: * r->current_string has already been set up by roff_parse().
3972: */
1.340 schwarze 3973: static int
1.306 schwarze 3974: roff_renamed(ROFF_ARGS)
3975: {
3976: char *nbuf;
3977:
1.315 schwarze 3978: buf->sz = mandoc_asprintf(&nbuf, ".%s%s%s", r->current_string,
3979: buf->buf[pos] == '\0' ? "" : " ", buf->buf + pos) + 1;
1.306 schwarze 3980: free(buf->buf);
3981: buf->buf = nbuf;
1.329 schwarze 3982: *offs = 0;
1.306 schwarze 3983: return ROFF_CONT;
3984: }
3985:
1.362 schwarze 3986: /*
3987: * Measure the length in bytes of the roff identifier at *cpp
3988: * and advance the pointer to the next word.
3989: */
1.212 schwarze 3990: static size_t
1.121 schwarze 3991: roff_getname(struct roff *r, char **cpp, int ln, int pos)
3992: {
3993: char *name, *cp;
1.212 schwarze 3994: size_t namesz;
1.121 schwarze 3995:
3996: name = *cpp;
1.362 schwarze 3997: if (*name == '\0')
1.277 schwarze 3998: return 0;
1.121 schwarze 3999:
1.362 schwarze 4000: /* Advance cp to the byte after the end of the name. */
4001:
1.212 schwarze 4002: for (cp = name; 1; cp++) {
1.362 schwarze 4003: namesz = cp - name;
1.363 schwarze 4004: if (*cp == '\0')
1.212 schwarze 4005: break;
1.363 schwarze 4006: if (*cp == ' ' || *cp == '\t') {
4007: cp++;
4008: break;
4009: }
1.362 schwarze 4010: if (*cp != '\\')
1.121 schwarze 4011: continue;
1.362 schwarze 4012: if (cp[1] == '{' || cp[1] == '}')
1.215 schwarze 4013: break;
1.362 schwarze 4014: if (*++cp == '\\')
1.121 schwarze 4015: continue;
1.350 schwarze 4016: mandoc_msg(MANDOCERR_NAMESC, ln, pos,
1.224 schwarze 4017: "%.*s", (int)(cp - name + 1), name);
1.212 schwarze 4018: mandoc_escape((const char **)&cp, NULL, NULL);
4019: break;
1.121 schwarze 4020: }
4021:
4022: /* Read past spaces. */
1.362 schwarze 4023:
4024: while (*cp == ' ')
1.121 schwarze 4025: cp++;
4026:
4027: *cpp = cp;
1.277 schwarze 4028: return namesz;
1.121 schwarze 4029: }
4030:
1.106 kristaps 4031: /*
4032: * Store *string into the user-defined string called *name.
4033: * To clear an existing entry, call with (*r, *name, NULL, 0).
1.193 schwarze 4034: * append == 0: replace mode
4035: * append == 1: single-line append mode
4036: * append == 2: multiline append mode, append '\n' after each call
1.106 kristaps 4037: */
1.94 kristaps 4038: static void
1.106 kristaps 4039: roff_setstr(struct roff *r, const char *name, const char *string,
1.193 schwarze 4040: int append)
1.92 schwarze 4041: {
1.315 schwarze 4042: size_t namesz;
1.164 kristaps 4043:
1.315 schwarze 4044: namesz = strlen(name);
4045: roff_setstrn(&r->strtab, name, namesz, string,
1.207 schwarze 4046: string ? strlen(string) : 0, append);
1.315 schwarze 4047: roff_setstrn(&r->rentab, name, namesz, NULL, 0, 0);
1.164 kristaps 4048: }
4049:
4050: static void
1.166 kristaps 4051: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.193 schwarze 4052: const char *string, size_t stringsz, int append)
1.164 kristaps 4053: {
1.166 kristaps 4054: struct roffkv *n;
1.164 kristaps 4055: char *c;
4056: int i;
4057: size_t oldch, newch;
1.92 schwarze 4058:
1.106 kristaps 4059: /* Search for an existing string with the same name. */
1.164 kristaps 4060: n = *r;
4061:
1.211 schwarze 4062: while (n && (namesz != n->key.sz ||
4063: strncmp(n->key.p, name, namesz)))
1.92 schwarze 4064: n = n->next;
1.94 kristaps 4065:
4066: if (NULL == n) {
1.106 kristaps 4067: /* Create a new string table entry. */
1.166 kristaps 4068: n = mandoc_malloc(sizeof(struct roffkv));
4069: n->key.p = mandoc_strndup(name, namesz);
4070: n->key.sz = namesz;
4071: n->val.p = NULL;
4072: n->val.sz = 0;
1.164 kristaps 4073: n->next = *r;
4074: *r = n;
1.193 schwarze 4075: } else if (0 == append) {
1.166 kristaps 4076: free(n->val.p);
4077: n->val.p = NULL;
4078: n->val.sz = 0;
1.106 kristaps 4079: }
4080:
4081: if (NULL == string)
4082: return;
4083:
4084: /*
4085: * One additional byte for the '\n' in multiline mode,
4086: * and one for the terminating '\0'.
4087: */
1.193 schwarze 4088: newch = stringsz + (1 < append ? 2u : 1u);
1.164 kristaps 4089:
1.166 kristaps 4090: if (NULL == n->val.p) {
4091: n->val.p = mandoc_malloc(newch);
4092: *n->val.p = '\0';
1.106 kristaps 4093: oldch = 0;
4094: } else {
1.166 kristaps 4095: oldch = n->val.sz;
4096: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.106 kristaps 4097: }
4098:
4099: /* Skip existing content in the destination buffer. */
1.166 kristaps 4100: c = n->val.p + (int)oldch;
1.106 kristaps 4101:
4102: /* Append new content to the destination buffer. */
1.164 kristaps 4103: i = 0;
4104: while (i < (int)stringsz) {
1.106 kristaps 4105: /*
4106: * Rudimentary roff copy mode:
4107: * Handle escaped backslashes.
4108: */
1.164 kristaps 4109: if ('\\' == string[i] && '\\' == string[i + 1])
4110: i++;
4111: *c++ = string[i++];
1.106 kristaps 4112: }
1.94 kristaps 4113:
1.106 kristaps 4114: /* Append terminating bytes. */
1.193 schwarze 4115: if (1 < append)
1.106 kristaps 4116: *c++ = '\n';
1.163 kristaps 4117:
1.106 kristaps 4118: *c = '\0';
1.166 kristaps 4119: n->val.sz = (int)(c - n->val.p);
1.92 schwarze 4120: }
4121:
1.94 kristaps 4122: static const char *
1.325 schwarze 4123: roff_getstrn(struct roff *r, const char *name, size_t len,
1.315 schwarze 4124: int *deftype)
1.92 schwarze 4125: {
1.315 schwarze 4126: const struct roffkv *n;
1.325 schwarze 4127: int found, i;
1.315 schwarze 4128: enum roff_tok tok;
4129:
1.325 schwarze 4130: found = 0;
4131: for (n = r->strtab; n != NULL; n = n->next) {
4132: if (strncmp(name, n->key.p, len) != 0 ||
4133: n->key.p[len] != '\0' || n->val.p == NULL)
4134: continue;
4135: if (*deftype & ROFFDEF_USER) {
4136: *deftype = ROFFDEF_USER;
4137: return n->val.p;
4138: } else {
4139: found = 1;
4140: break;
4141: }
4142: }
4143: for (n = r->rentab; n != NULL; n = n->next) {
4144: if (strncmp(name, n->key.p, len) != 0 ||
4145: n->key.p[len] != '\0' || n->val.p == NULL)
4146: continue;
4147: if (*deftype & ROFFDEF_REN) {
4148: *deftype = ROFFDEF_REN;
4149: return n->val.p;
4150: } else {
4151: found = 1;
4152: break;
1.315 schwarze 4153: }
4154: }
1.325 schwarze 4155: for (i = 0; i < PREDEFS_MAX; i++) {
4156: if (strncmp(name, predefs[i].name, len) != 0 ||
4157: predefs[i].name[len] != '\0')
4158: continue;
4159: if (*deftype & ROFFDEF_PRE) {
4160: *deftype = ROFFDEF_PRE;
4161: return predefs[i].str;
4162: } else {
4163: found = 1;
4164: break;
1.315 schwarze 4165: }
4166: }
1.356 schwarze 4167: if (r->man->meta.macroset != MACROSET_MAN) {
1.325 schwarze 4168: for (tok = MDOC_Dd; tok < MDOC_MAX; tok++) {
4169: if (strncmp(name, roff_name[tok], len) != 0 ||
4170: roff_name[tok][len] != '\0')
4171: continue;
4172: if (*deftype & ROFFDEF_STD) {
4173: *deftype = ROFFDEF_STD;
4174: return NULL;
4175: } else {
4176: found = 1;
4177: break;
1.315 schwarze 4178: }
4179: }
4180: }
1.356 schwarze 4181: if (r->man->meta.macroset != MACROSET_MDOC) {
1.325 schwarze 4182: for (tok = MAN_TH; tok < MAN_MAX; tok++) {
4183: if (strncmp(name, roff_name[tok], len) != 0 ||
4184: roff_name[tok][len] != '\0')
4185: continue;
4186: if (*deftype & ROFFDEF_STD) {
4187: *deftype = ROFFDEF_STD;
4188: return NULL;
4189: } else {
4190: found = 1;
4191: break;
1.315 schwarze 4192: }
4193: }
1.325 schwarze 4194: }
4195:
4196: if (found == 0 && *deftype != ROFFDEF_ANY) {
4197: if (*deftype & ROFFDEF_REN) {
4198: /*
4199: * This might still be a request,
4200: * so do not treat it as undefined yet.
4201: */
4202: *deftype = ROFFDEF_UNDEF;
4203: return NULL;
1.315 schwarze 4204: }
1.325 schwarze 4205:
4206: /* Using an undefined string defines it to be empty. */
4207:
4208: roff_setstrn(&r->strtab, name, len, "", 0, 0);
4209: roff_setstrn(&r->rentab, name, len, NULL, 0, 0);
1.315 schwarze 4210: }
1.325 schwarze 4211:
1.315 schwarze 4212: *deftype = 0;
1.277 schwarze 4213: return NULL;
1.92 schwarze 4214: }
4215:
1.94 kristaps 4216: static void
1.167 kristaps 4217: roff_freestr(struct roffkv *r)
1.92 schwarze 4218: {
1.166 kristaps 4219: struct roffkv *n, *nn;
1.92 schwarze 4220:
1.167 kristaps 4221: for (n = r; n; n = nn) {
1.166 kristaps 4222: free(n->key.p);
4223: free(n->val.p);
1.92 schwarze 4224: nn = n->next;
4225: free(n);
4226: }
1.114 kristaps 4227: }
1.266 schwarze 4228:
4229: /* --- accessors and utility functions ------------------------------------ */
1.164 kristaps 4230:
4231: /*
4232: * Duplicate an input string, making the appropriate character
4233: * conversations (as stipulated by `tr') along the way.
4234: * Returns a heap-allocated string with all the replacements made.
4235: */
4236: char *
4237: roff_strdup(const struct roff *r, const char *p)
4238: {
1.166 kristaps 4239: const struct roffkv *cp;
1.164 kristaps 4240: char *res;
4241: const char *pp;
4242: size_t ssz, sz;
4243: enum mandoc_esc esc;
4244:
1.167 kristaps 4245: if (NULL == r->xmbtab && NULL == r->xtab)
1.277 schwarze 4246: return mandoc_strdup(p);
1.164 kristaps 4247: else if ('\0' == *p)
1.277 schwarze 4248: return mandoc_strdup("");
1.164 kristaps 4249:
4250: /*
4251: * Step through each character looking for term matches
4252: * (remember that a `tr' can be invoked with an escape, which is
4253: * a glyph but the escape is multi-character).
4254: * We only do this if the character hash has been initialised
4255: * and the string is >0 length.
4256: */
4257:
4258: res = NULL;
4259: ssz = 0;
4260:
4261: while ('\0' != *p) {
1.289 schwarze 4262: assert((unsigned int)*p < 128);
4263: if ('\\' != *p && r->xtab && r->xtab[(unsigned int)*p].p) {
1.167 kristaps 4264: sz = r->xtab[(int)*p].sz;
4265: res = mandoc_realloc(res, ssz + sz + 1);
4266: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
4267: ssz += sz;
4268: p++;
4269: continue;
4270: } else if ('\\' != *p) {
4271: res = mandoc_realloc(res, ssz + 2);
4272: res[ssz++] = *p++;
4273: continue;
4274: }
4275:
1.164 kristaps 4276: /* Search for term matches. */
1.167 kristaps 4277: for (cp = r->xmbtab; cp; cp = cp->next)
1.166 kristaps 4278: if (0 == strncmp(p, cp->key.p, cp->key.sz))
1.164 kristaps 4279: break;
4280:
4281: if (NULL != cp) {
4282: /*
4283: * A match has been found.
4284: * Append the match to the array and move
4285: * forward by its keysize.
4286: */
1.207 schwarze 4287: res = mandoc_realloc(res,
4288: ssz + cp->val.sz + 1);
1.166 kristaps 4289: memcpy(res + ssz, cp->val.p, cp->val.sz);
4290: ssz += cp->val.sz;
4291: p += (int)cp->key.sz;
1.164 kristaps 4292: continue;
4293: }
4294:
1.167 kristaps 4295: /*
4296: * Handle escapes carefully: we need to copy
4297: * over just the escape itself, or else we might
4298: * do replacements within the escape itself.
4299: * Make sure to pass along the bogus string.
4300: */
4301: pp = p++;
4302: esc = mandoc_escape(&p, NULL, NULL);
4303: if (ESCAPE_ERROR == esc) {
4304: sz = strlen(pp);
1.164 kristaps 4305: res = mandoc_realloc(res, ssz + sz + 1);
4306: memcpy(res + ssz, pp, sz);
1.167 kristaps 4307: break;
1.164 kristaps 4308: }
1.207 schwarze 4309: /*
4310: * We bail out on bad escapes.
1.167 kristaps 4311: * No need to warn: we already did so when
1.355 schwarze 4312: * roff_expand() was called.
1.167 kristaps 4313: */
4314: sz = (int)(p - pp);
4315: res = mandoc_realloc(res, ssz + sz + 1);
4316: memcpy(res + ssz, pp, sz);
4317: ssz += sz;
1.164 kristaps 4318: }
4319:
4320: res[(int)ssz] = '\0';
1.277 schwarze 4321: return res;
1.227 schwarze 4322: }
4323:
4324: int
4325: roff_getformat(const struct roff *r)
4326: {
4327:
1.277 schwarze 4328: return r->format;
1.174 kristaps 4329: }
4330:
4331: /*
1.207 schwarze 4332: * Find out whether a line is a macro line or not.
1.174 kristaps 4333: * If it is, adjust the current position and return one; if it isn't,
4334: * return zero and don't change the current position.
4335: * If the control character has been set with `.cc', then let that grain
4336: * precedence.
4337: * This is slighly contrary to groff, where using the non-breaking
4338: * control character when `cc' has been invoked will cause the
4339: * non-breaking macro contents to be printed verbatim.
4340: */
4341: int
4342: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
4343: {
4344: int pos;
4345:
4346: pos = *ppos;
4347:
1.303 schwarze 4348: if (r->control != '\0' && cp[pos] == r->control)
1.174 kristaps 4349: pos++;
1.303 schwarze 4350: else if (r->control != '\0')
1.277 schwarze 4351: return 0;
1.174 kristaps 4352: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
4353: pos += 2;
4354: else if ('.' == cp[pos] || '\'' == cp[pos])
4355: pos++;
4356: else
1.277 schwarze 4357: return 0;
1.174 kristaps 4358:
4359: while (' ' == cp[pos] || '\t' == cp[pos])
4360: pos++;
4361:
4362: *ppos = pos;
1.277 schwarze 4363: return 1;
1.74 kristaps 4364: }
CVSweb