Annotation of texi2mdoc/main.c, Revision 1.16
1.16 ! kristaps 1: /* $Id: main.c,v 1.15 2015/02/19 09:28:42 kristaps Exp $ */
1.1 kristaps 2: /*
3: * Copyright (c) 2015 Kristaps Dzonsons <kristaps@bsd.lv>
4: *
5: * Permission to use, copy, modify, and distribute this software for any
6: * purpose with or without fee is hereby granted, provided that the above
7: * copyright notice and this permission notice appear in all copies.
8: *
9: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16: */
17: #include <sys/mman.h>
18: #include <sys/stat.h>
19:
20: #include <assert.h>
21: #include <ctype.h>
22: #include <fcntl.h>
23: #include <getopt.h>
1.2 kristaps 24: #include <libgen.h>
25: #include <limits.h>
1.1 kristaps 26: #include <stdarg.h>
27: #include <stdio.h>
28: #include <stdlib.h>
29: #include <string.h>
1.10 kristaps 30: #include <time.h>
1.6 kristaps 31: #include <unistd.h>
1.1 kristaps 32:
33: /*
34: * This defines each one of the Texinfo commands that we understand.
35: * Obviously this only refers to native commands; overriden names are a
36: * different story.
37: */
38: enum texicmd {
1.2 kristaps 39: TEXICMD_ACRONYM,
1.16 ! kristaps 40: TEXICMD_ACUTE,
1.1 kristaps 41: TEXICMD_A4PAPER,
42: TEXICMD_ANCHOR,
1.2 kristaps 43: TEXICMD_APPENDIX,
44: TEXICMD_APPENDIXSEC,
1.3 kristaps 45: TEXICMD_ASTERISK,
1.1 kristaps 46: TEXICMD_AT,
1.3 kristaps 47: TEXICMD_AUTHOR,
1.12 kristaps 48: TEXICMD_B,
1.3 kristaps 49: TEXICMD_BANG,
1.7 kristaps 50: TEXICMD_BULLET,
1.1 kristaps 51: TEXICMD_BYE,
1.5 kristaps 52: TEXICMD_CENTER,
1.1 kristaps 53: TEXICMD_CHAPTER,
54: TEXICMD_CINDEX,
1.16 ! kristaps 55: TEXICMD_CIRCUMFLEX,
1.3 kristaps 56: TEXICMD_CITE,
1.1 kristaps 57: TEXICMD_CODE,
1.3 kristaps 58: TEXICMD_COLON,
1.1 kristaps 59: TEXICMD_COMMAND,
60: TEXICMD_COMMENT,
1.2 kristaps 61: TEXICMD_COMMENT_LONG,
1.1 kristaps 62: TEXICMD_CONTENTS,
63: TEXICMD_COPYING,
64: TEXICMD_COPYRIGHT,
1.15 kristaps 65: TEXICMD_DEFFN,
66: TEXICMD_DEFFNX,
67: TEXICMD_DEFMAC,
68: TEXICMD_DEFMACX,
69: TEXICMD_DEFTP,
70: TEXICMD_DEFTPX,
1.3 kristaps 71: TEXICMD_DEFTYPEFN,
72: TEXICMD_DEFTYPEFNX,
73: TEXICMD_DEFTYPEFUN,
74: TEXICMD_DEFTYPEFUNX,
75: TEXICMD_DEFTYPEVAR,
1.15 kristaps 76: TEXICMD_DEFTYPEVARX,
1.3 kristaps 77: TEXICMD_DEFTYPEVR,
1.15 kristaps 78: TEXICMD_DEFTYPEVRX,
79: TEXICMD_DEFUN,
80: TEXICMD_DEFUNX,
81: TEXICMD_DEFVAR,
82: TEXICMD_DEFVARX,
83: TEXICMD_DEFVR,
84: TEXICMD_DEFVRX,
1.1 kristaps 85: TEXICMD_DETAILMENU,
1.3 kristaps 86: TEXICMD_DFN,
1.1 kristaps 87: TEXICMD_DIRCATEGORY,
88: TEXICMD_DIRENTRY,
1.3 kristaps 89: TEXICMD_DISPLAY,
1.2 kristaps 90: TEXICMD_DOTS,
1.1 kristaps 91: TEXICMD_EMAIL,
92: TEXICMD_EMPH,
93: TEXICMD_END,
1.2 kristaps 94: TEXICMD_ENUMERATE,
1.3 kristaps 95: TEXICMD_ENV,
1.15 kristaps 96: TEXICMD_ERROR,
1.1 kristaps 97: TEXICMD_EXAMPLE,
98: TEXICMD_FILE,
1.16 ! kristaps 99: TEXICMD_GRAVE,
1.3 kristaps 100: TEXICMD_GROUP,
1.2 kristaps 101: TEXICMD_HEADING,
1.3 kristaps 102: TEXICMD_HEADINGS,
103: TEXICMD_HYPHEN,
1.1 kristaps 104: TEXICMD_I,
1.3 kristaps 105: TEXICMD_IFCLEAR,
1.14 kristaps 106: TEXICMD_IFDOCBOOK,
1.1 kristaps 107: TEXICMD_IFHTML,
1.3 kristaps 108: TEXICMD_IFINFO,
1.14 kristaps 109: TEXICMD_IFNOTDOCBOOK,
110: TEXICMD_IFNOTHTML,
111: TEXICMD_IFNOTINFO,
112: TEXICMD_IFNOTPLAINTEXT,
1.1 kristaps 113: TEXICMD_IFNOTTEX,
1.14 kristaps 114: TEXICMD_IFNOTXML,
115: TEXICMD_IFPLAINTEXT,
1.1 kristaps 116: TEXICMD_IFTEX,
1.3 kristaps 117: TEXICMD_IFSET,
1.14 kristaps 118: TEXICMD_IFXML,
1.1 kristaps 119: TEXICMD_IMAGE,
1.2 kristaps 120: TEXICMD_INCLUDE,
1.13 kristaps 121: TEXICMD_INDENTBLOCK,
1.5 kristaps 122: TEXICMD_INSERTCOPYING,
1.1 kristaps 123: TEXICMD_ITEM,
124: TEXICMD_ITEMIZE,
125: TEXICMD_KBD,
126: TEXICMD_LATEX,
1.3 kristaps 127: TEXICMD_MATH,
1.1 kristaps 128: TEXICMD_MENU,
1.15 kristaps 129: TEXICMD_NEED,
1.3 kristaps 130: TEXICMD_NEWLINE,
1.1 kristaps 131: TEXICMD_NODE,
1.3 kristaps 132: TEXICMD_NOINDENT,
1.16 ! kristaps 133: TEXICMD_OPTION,
1.8 kristaps 134: TEXICMD_PXREF,
1.3 kristaps 135: TEXICMD_QUESTIONMARK,
1.1 kristaps 136: TEXICMD_QUOTATION,
1.3 kristaps 137: TEXICMD_PAGE,
1.1 kristaps 138: TEXICMD_PARINDENT,
1.2 kristaps 139: TEXICMD_PRINTINDEX,
1.12 kristaps 140: TEXICMD_R,
1.1 kristaps 141: TEXICMD_REF,
1.15 kristaps 142: TEXICMD_RESULT,
1.1 kristaps 143: TEXICMD_SAMP,
1.12 kristaps 144: TEXICMD_SANSSERIF,
1.7 kristaps 145: TEXICMD_SC,
1.1 kristaps 146: TEXICMD_SECTION,
1.3 kristaps 147: TEXICMD_SET,
1.1 kristaps 148: TEXICMD_SETCHAPNEWPAGE,
149: TEXICMD_SETFILENAME,
150: TEXICMD_SETTITLE,
1.12 kristaps 151: TEXICMD_SLANTED,
1.3 kristaps 152: TEXICMD_SP,
153: TEXICMD_SPACE,
1.12 kristaps 154: TEXICMD_SMALLDISPLAY,
1.3 kristaps 155: TEXICMD_SMALLEXAMPLE,
1.13 kristaps 156: TEXICMD_SMALLINDENTBLOCK,
1.3 kristaps 157: TEXICMD_SQUIGGLE_LEFT,
158: TEXICMD_SQUIGGLE_RIGHT,
1.8 kristaps 159: TEXICMD_STRONG,
1.1 kristaps 160: TEXICMD_SUBSECTION,
1.3 kristaps 161: TEXICMD_SUBTITLE,
1.12 kristaps 162: TEXICMD_T,
1.3 kristaps 163: TEXICMD_TAB,
1.1 kristaps 164: TEXICMD_TABLE,
165: TEXICMD_TEX,
166: TEXICMD_TEXSYM,
1.16 ! kristaps 167: TEXICMD_TILDE,
1.3 kristaps 168: TEXICMD_TITLE,
1.1 kristaps 169: TEXICMD_TITLEFONT,
170: TEXICMD_TITLEPAGE,
171: TEXICMD_TOP,
1.16 ! kristaps 172: TEXICMD_UMLAUT,
1.1 kristaps 173: TEXICMD_UNNUMBERED,
1.2 kristaps 174: TEXICMD_UNNUMBEREDSEC,
1.3 kristaps 175: TEXICMD_UREF,
1.1 kristaps 176: TEXICMD_URL,
177: TEXICMD_VAR,
1.16 ! kristaps 178: TEXICMD_VERBATIMINCLUDE,
1.9 kristaps 179: TEXICMD_VSKIP,
1.3 kristaps 180: TEXICMD_W,
1.8 kristaps 181: TEXICMD_XREF,
1.1 kristaps 182: TEXICMD__MAX
183: };
184:
185: /*
186: * The file currently being parsed.
187: * This keeps track of our location within that file.
188: */
189: struct texifile {
190: const char *name; /* name of the file */
191: size_t line; /* current line (from zero) */
192: size_t col; /* current column in line (from zero) */
193: char *map; /* mmap'd file */
194: size_t mapsz; /* size of mmap */
195: };
196:
197: struct texi;
198:
1.2 kristaps 199: /*
200: * Callback for functions implementing texi commands.
201: */
1.1 kristaps 202: typedef void (*texicmdfp)(struct texi *,
203: enum texicmd, const char *, size_t, size_t *);
204:
205: /*
206: * Describes Texinfo commands, whether native or overriden.
207: */
208: struct texitok {
209: texicmdfp fp; /* callback (or NULL if none) */
210: const char *tok; /* name of the token */
211: size_t len; /* strlen(tok) */
212: };
213:
1.3 kristaps 214: enum texilist {
215: TEXILIST_NONE = 0,
216: TEXILIST_ITEM,
217: TEXILIST_NOITEM,
218: };
219:
1.1 kristaps 220: /*
221: * The main parse structure.
222: * This keeps any necessary information handy.
223: */
224: struct texi {
1.5 kristaps 225: struct texifile files[64]; /* stack of open files */
226: size_t filepos; /* number of open files */
227: size_t outcol; /* column in output line */
228: char **dirs; /* texi directories */
229: size_t dirsz; /* number of texi directories */
1.11 kristaps 230: char *title; /* title of document */
231: char *subtitle; /* subtitle of document */
232: /*
233: * The following control what we output to the screen.
234: * The complexity is required to accomodate for mdoc(7).
235: */
1.8 kristaps 236: enum texilist list; /* current list (set recursively) */
237: int outmacro; /* if >0, output is in line macro */
238: int seenws; /* ws has been seen (and ignored) */
1.11 kristaps 239: int seenvs; /* newline has been Pp'd */
1.8 kristaps 240: int ign; /* if >0, don't print anything */
241: int literal; /* if >0, literal context */
1.1 kristaps 242: };
243:
1.8 kristaps 244: /* Texi disregards spaces and tabs. */
1.2 kristaps 245: #define isws(_x) \
246: (' ' == (_x) || '\t' == (_x))
1.9 kristaps 247: #define ismspace(_x) \
1.10 kristaps 248: (isws((_x)) || '\n' == (_x))
1.1 kristaps 249:
1.16 ! kristaps 250: static void doaccent(struct texi *, enum texicmd, const char *, size_t, size_t *);
1.3 kristaps 251: static void doblock(struct texi *, enum texicmd, const char *, size_t, size_t *);
1.1 kristaps 252: static void dobracket(struct texi *, enum texicmd, const char *, size_t, size_t *);
253: static void dobye(struct texi *, enum texicmd, const char *, size_t, size_t *);
1.15 kristaps 254: static void dodefn(struct texi *, enum texicmd, const char *, size_t, size_t *);
1.3 kristaps 255: static void dodisplay(struct texi *, enum texicmd, const char *, size_t, size_t *);
1.2 kristaps 256: static void doenumerate(struct texi *, enum texicmd, const char *, size_t, size_t *);
1.1 kristaps 257: static void doexample(struct texi *, enum texicmd, const char *, size_t, size_t *);
1.12 kristaps 258: static void dofont(struct texi *, enum texicmd, const char *, size_t, size_t *);
1.8 kristaps 259: static void doignargn(struct texi *, enum texicmd, const char *, size_t, size_t *);
1.1 kristaps 260: static void doignblock(struct texi *, enum texicmd, const char *, size_t, size_t *);
261: static void doignbracket(struct texi *, enum texicmd, const char *, size_t, size_t *);
262: static void doignline(struct texi *, enum texicmd, const char *, size_t, size_t *);
1.12 kristaps 263: static void doinline(struct texi *, enum texicmd, const char *, size_t, size_t *);
1.2 kristaps 264: static void doinclude(struct texi *, enum texicmd, const char *, size_t, size_t *);
1.1 kristaps 265: static void doitem(struct texi *, enum texicmd, const char *, size_t, size_t *);
266: static void doitemize(struct texi *, enum texicmd, const char *, size_t, size_t *);
1.8 kristaps 267: static void dolink(struct texi *, enum texicmd, const char *, size_t, size_t *);
1.3 kristaps 268: static void domath(struct texi *, enum texicmd, const char *, size_t, size_t *);
1.1 kristaps 269: static void doquotation(struct texi *, enum texicmd, const char *, size_t, size_t *);
270: static void dotable(struct texi *, enum texicmd, const char *, size_t, size_t *);
271: static void dotop(struct texi *, enum texicmd, const char *, size_t, size_t *);
272: static void dosection(struct texi *, enum texicmd, const char *, size_t, size_t *);
1.3 kristaps 273: static void dosp(struct texi *, enum texicmd, const char *, size_t, size_t *);
1.1 kristaps 274: static void dosubsection(struct texi *, enum texicmd, const char *, size_t, size_t *);
275: static void dosymbol(struct texi *, enum texicmd, const char *, size_t, size_t *);
1.10 kristaps 276: static void dotitle(struct texi *, enum texicmd, const char *, size_t, size_t *);
1.16 ! kristaps 277: static void doverbinclude(struct texi *, enum texicmd, const char *, size_t, size_t *);
1.1 kristaps 278:
279: static const struct texitok texitoks[TEXICMD__MAX] = {
1.8 kristaps 280: { doignargn, "acronym", 7 }, /* TEXICMD_ACRONYM */
1.16 ! kristaps 281: { doaccent, "'", 1 }, /* TEXICMD_ACUTE */
1.1 kristaps 282: { doignline, "afourpaper", 10 }, /* TEXICMD_A4PAPER */
283: { doignbracket, "anchor", 6 }, /* TEXICMD_ANCHOR */
1.12 kristaps 284: { dosection, "appendix", 8 }, /* TEXICMD_APPENDIX */
285: { dosection, "appendixsec", 11 }, /* TEXICMD_APPENDIXSEC */
1.3 kristaps 286: { dosymbol, "*", 1 }, /* TEXICMD_ASTERISK */
1.1 kristaps 287: { dosymbol, "@", 1 }, /* TEXICMD_AT */
1.3 kristaps 288: { doignline, "author", 6 }, /* TEXICMD_AUTHOR */
1.12 kristaps 289: { dofont, "b", 1 }, /* TEXICMD_BOLD */
1.3 kristaps 290: { dosymbol, "!", 1 }, /* TEXICMD_BANG */
1.7 kristaps 291: { dosymbol, "bullet", 6 }, /* TEXICMD_BULLET */
1.1 kristaps 292: { dobye, "bye", 3 }, /* TEXICMD_BYE */
1.12 kristaps 293: { doignline, "center", 6 }, /* TEXICMD_CENTER */
294: { dosection, "chapter", 7 }, /* TEXICMD_CHAPTER */
1.1 kristaps 295: { doignline, "cindex", 6 }, /* TEXICMD_CINDEX */
1.16 ! kristaps 296: { doaccent, "^", 1 }, /* TEXICMD_CIRCUMFLEX */
1.12 kristaps 297: { dofont, "code", 4 }, /* TEXICMD_CODE */
298: { dofont, "cite", 4 }, /* TEXICMD_CITE */
1.3 kristaps 299: { dosymbol, ":", 1 }, /* TEXICMD_COLON */
1.12 kristaps 300: { doinline, "command", 7 }, /* TEXICMD_COMMAND */
1.1 kristaps 301: { doignline, "c", 1 }, /* TEXICMD_COMMENT */
1.2 kristaps 302: { doignline, "comment", 7 }, /* TEXICMD_COMMENT_LONG */
1.1 kristaps 303: { doignline, "contents", 8 }, /* TEXICMD_CONTENTS */
304: { doignblock, "copying", 7 }, /* TEXICMD_COPYING */
305: { dosymbol, "copyright", 9 }, /* TEXICMD_COPYRIGHT */
1.15 kristaps 306: { dodefn, "deffn", 5 }, /* TEXICMD_DEFFN */
307: { dodefn, "deffnx", 6 }, /* TEXICMD_DEFFNX */
308: { dodefn, "defmac", 6 }, /* TEXICMD_DEFMAC */
309: { dodefn, "defmacx", 7 }, /* TEXICMD_DEFMACX */
310: { dodefn, "deftp", 5 }, /* TEXICMD_DEFTP */
311: { dodefn, "deftpx", 6 }, /* TEXICMD_DEFTPX */
312: { dodefn, "deftypefn", 9 }, /* TEXICMD_DEFTYPEFN */
313: { dodefn, "deftypefnx", 10 }, /* TEXICMD_DEFTYPEFNX */
314: { dodefn, "deftypefun", 10 }, /* TEXICMD_DEFTYPEFUN */
315: { dodefn, "deftypefunx", 11 }, /* TEXICMD_DEFTYPEFUNX */
316: { dodefn, "deftypevar", 10 }, /* TEXICMD_DEFTYPEVAR */
317: { dodefn, "deftypevarx", 11 }, /* TEXICMD_DEFTYPEVARX */
318: { dodefn, "deftypevr", 9 }, /* TEXICMD_DEFTYPEVR */
319: { dodefn, "deftypevrx", 10 }, /* TEXICMD_DEFTYPEVRX */
320: { dodefn, "defun", 5 }, /* TEXICMD_DEFUN */
321: { dodefn, "defunx", 6 }, /* TEXICMD_DEFUNX */
322: { dodefn, "defvar", 6 }, /* TEXICMD_DEFVAR */
323: { dodefn, "defvarx", 7 }, /* TEXICMD_DEFVARX */
324: { dodefn, "defvr", 5 }, /* TEXICMD_DEFVR */
325: { dodefn, "defvrx", 6 }, /* TEXICMD_DEFVRX */
1.1 kristaps 326: { doignblock, "detailmenu", 10 }, /* TEXICMD_DETAILMENU */
1.12 kristaps 327: { dofont, "dfn", 3 }, /* TEXICMD_DFN */
1.1 kristaps 328: { doignline, "dircategory", 11 }, /* TEXICMD_DIRCATEGORY */
329: { doignblock, "direntry", 8 }, /* TEXICMD_DIRENTRY */
1.3 kristaps 330: { dodisplay, "display", 7 }, /* TEXICMD_DISPLAY */
1.2 kristaps 331: { dosymbol, "dots", 4 }, /* TEXICMD_DOTS */
1.8 kristaps 332: { dolink, "email", 5 }, /* TEXICMD_EMAIL */
1.12 kristaps 333: { dofont, "emph", 4 }, /* TEXICMD_EMPH */
1.1 kristaps 334: { NULL, "end", 3 }, /* TEXICMD_END */
1.2 kristaps 335: { doenumerate, "enumerate", 9 }, /* TEXICMD_ENUMERATE */
1.12 kristaps 336: { doinline, "env", 3 }, /* TEXICMD_ENV */
1.15 kristaps 337: { dosymbol, "error", 5 }, /* TEXICMD_ERROR */
1.1 kristaps 338: { doexample, "example", 7 }, /* TEXICMD_EXAMPLE */
1.12 kristaps 339: { doinline, "file", 4 }, /* TEXICMD_FILE */
1.16 ! kristaps 340: { doaccent, "`", 1 }, /* TEXICMD_GRAVE */
1.3 kristaps 341: { doblock, "group", 5 }, /* TEXICMD_GROUP */
1.2 kristaps 342: { dosection, "heading", 7 }, /* TEXICMD_HEADING */
1.3 kristaps 343: { doignline, "headings", 8 }, /* TEXICMD_HEADINGS */
344: { dosymbol, "-", 1 }, /* TEXICMD_HYPHEN */
1.12 kristaps 345: { dofont, "i", 1 }, /* TEXICMD_I */
1.3 kristaps 346: { doignblock, "ifclear", 7 }, /* TEXICMD_IFCLEAR */
1.14 kristaps 347: { doignblock, "ifdocbook", 9 }, /* TEXICMD_IFDOCBOOK */
1.1 kristaps 348: { doignblock, "ifhtml", 6 }, /* TEXICMD_IFHTML */
1.3 kristaps 349: { doignblock, "ifinfo", 6 }, /* TEXICMD_IFINFO */
1.14 kristaps 350: { doblock, "ifnotdocbook", 12 }, /* TEXICMD_IFNOTDOCBOOK */
351: { doblock, "ifnothtml", 9 }, /* TEXICMD_IFNOTHTML */
352: { doblock, "ifnotinfo", 9 }, /* TEXICMD_IFNOTINFO */
353: { doignblock, "ifnotplaintext", 14 }, /* TEXICMD_IFNOTPLAINTEXT */
1.3 kristaps 354: { doblock, "ifnottex", 8 }, /* TEXICMD_IFNOTTEX */
1.14 kristaps 355: { doblock, "ifnotxml", 8 }, /* TEXICMD_IFNOTXML */
356: { doblock, "ifplaintext", 11 }, /* TEXICMD_IFPLAINTEXT */
1.1 kristaps 357: { doignblock, "iftex", 5 }, /* TEXICMD_IFTEX */
1.3 kristaps 358: { doignblock, "ifset", 5 }, /* TEXICMD_IFSET */
1.14 kristaps 359: { doignblock, "ifxml", 5 }, /* TEXICMD_IFXML */
1.1 kristaps 360: { doignbracket, "image", 5 }, /* TEXICMD_IMAGE */
1.2 kristaps 361: { doinclude, "include", 7 }, /* TEXICMD_INCLUDE */
1.13 kristaps 362: { dodisplay, "indentblock", 11 }, /* TEXICMD_INDENTBLOCK */
1.5 kristaps 363: { doignline, "insertcopying", 13 }, /* TEXICMD_INSERTCOPYING */
1.1 kristaps 364: { doitem, "item", 4 }, /* TEXICMD_ITEM */
365: { doitemize, "itemize", 7 }, /* TEXICMD_ITEMIZE */
1.12 kristaps 366: { dofont, "kbd", 3 }, /* TEXICMD_KBD */
1.1 kristaps 367: { dosymbol, "LaTeX", 5 }, /* TEXICMD_LATEX */
1.3 kristaps 368: { domath, "math", 4 }, /* TEXICMD_MATH */
1.1 kristaps 369: { doignblock, "menu", 4 }, /* TEXICMD_MENU */
1.15 kristaps 370: { doignline, "need", 4 }, /* TEXICMD_NEED */
1.3 kristaps 371: { dosymbol, "\n", 1 }, /* TEXICMD_NEWLINE */
1.1 kristaps 372: { doignline, "node", 4 }, /* TEXICMD_NODE */
1.3 kristaps 373: { doignline, "noindent", 8 }, /* TEXICMD_NOINDENT */
1.16 ! kristaps 374: { doinline, "option", 6 }, /* TEXICMD_OPTION */
1.8 kristaps 375: { dolink, "pxref", 5 }, /* TEXICMD_PXREF */
1.3 kristaps 376: { dosymbol, "?", 1 }, /* TEXICMD_QUESTIONMARK */
1.1 kristaps 377: { doquotation, "quotation", 9 }, /* TEXICMD_QUOTATION */
1.3 kristaps 378: { doignline, "page", 4 }, /* TEXICMD_PAGE */
379: { doignline, "paragraphindent", 14 }, /* TEXICMD_PARINDENT */
1.2 kristaps 380: { doignline, "printindex", 10 }, /* TEXICMD_PRINTINDEX */
1.12 kristaps 381: { dofont, "r", 1 }, /* TEXICMD_R */
1.1 kristaps 382: { dobracket, "ref", 3 }, /* TEXICMD_REF */
1.15 kristaps 383: { dosymbol, "result", 6 }, /* TEXICMD_RESULT */
1.12 kristaps 384: { dofont, "samp", 4 }, /* TEXICMD_SAMP */
385: { dofont, "sansserif", 9 }, /* TEXICMD_SANSSERIF */
1.7 kristaps 386: { dobracket, "sc", 2 }, /* TEXICMD_SC */
1.1 kristaps 387: { dosection, "section", 7 }, /* TEXICMD_SECTION */
1.3 kristaps 388: { doignline, "set", 3 }, /* TEXICMD_SET */
1.1 kristaps 389: { doignline, "setchapternewpage", 17 }, /* TEXICMD_SETCHAPNEWPAGE */
390: { doignline, "setfilename", 11 }, /* TEXICMD_SETFILENAME */
1.10 kristaps 391: { dotitle, "settitle", 8 }, /* TEXICMD_SETTITLE */
1.12 kristaps 392: { dofont, "slanted", 7 }, /* TEXICMD_SLANTED */
1.3 kristaps 393: { dosp, "sp", 2 }, /* TEXICMD_SP */
394: { dosymbol, " ", 1 }, /* TEXICMD_SPACE */
1.12 kristaps 395: { dodisplay, "smalldisplay", 12 }, /* TEXICMD_SMALLDISPLAY */
1.3 kristaps 396: { doexample, "smallexample", 12 }, /* TEXICMD_SMALLEXAMPLE */
1.13 kristaps 397: { dodisplay, "smallindentblock", 16 }, /* TEXICMD_SMALLINDENTBLOCK */
1.3 kristaps 398: { dosymbol, "{", 1 }, /* TEXICMD_SQUIGGLE_LEFT */
399: { dosymbol, "}", 1 }, /* TEXICMD_SQUIGGLE_RIGHT */
1.12 kristaps 400: { dofont, "strong", 6 }, /* TEXICMD_STRONG */
1.1 kristaps 401: { dosubsection, "subsection", 10 }, /* TEXICMD_SUBSECTION */
1.3 kristaps 402: { doignline, "subtitle", 8 }, /* TEXICMD_SUBTITLE */
1.12 kristaps 403: { dofont, "t", 1 }, /* TEXICMD_T */
1.3 kristaps 404: { dosymbol, "\t", 1 }, /* TEXICMD_TAB */
1.1 kristaps 405: { dotable, "table", 5 }, /* TEXICMD_TABLE */
406: { doignblock, "tex", 3 }, /* TEXICMD_TEX */
407: { dosymbol, "TeX", 3 }, /* TEXICMD_TEXSYM */
1.16 ! kristaps 408: { doaccent, "~", 1 }, /* TEXICMD_TILDE */
1.3 kristaps 409: { doignline, "title", 5 }, /* TEXICMD_TITLE */
1.1 kristaps 410: { dobracket, "titlefont", 9 }, /* TEXICMD_TITLEFONT */
411: { doignblock, "titlepage", 9 }, /* TEXICMD_TITLEPAGE */
412: { dotop, "top", 3 }, /* TEXICMD_TOP */
1.16 ! kristaps 413: { doaccent, "\"", 1 }, /* TEXICMD_UMLAUT */
1.12 kristaps 414: { dosection, "unnumbered", 10 }, /* TEXICMD_UNNUMBERED */
1.2 kristaps 415: { dosection, "unnumberedsec", 13 }, /* TEXICMD_UNNUMBEREDSEC */
1.8 kristaps 416: { dolink, "uref", 4 }, /* TEXICMD_UREF */
417: { dolink, "url", 3 }, /* TEXICMD_URL */
1.12 kristaps 418: { doinline, "var", 3 }, /* TEXICMD_VAR */
1.16 ! kristaps 419: { doverbinclude, "verbatiminclude", 15 }, /* TEXICMD_VERBATIMINCLUDE */
1.9 kristaps 420: { dosp, "vskip", 5 }, /* TEXICMD_VSKIP */
1.3 kristaps 421: { dobracket, "w", 1 }, /* TEXICMD_W */
1.8 kristaps 422: { dolink, "xref", 4 }, /* TEXICMD_XREF */
1.1 kristaps 423: };
424:
1.2 kristaps 425: /*
426: * Unmap the top-most file that we're using.
427: */
1.1 kristaps 428: static void
429: texifilepop(struct texi *p)
430: {
431: struct texifile *f;
432:
433: assert(p->filepos > 0);
434: f = &p->files[--p->filepos];
435: munmap(f->map, f->mapsz);
436: }
437:
1.2 kristaps 438: /*
1.8 kristaps 439: * Unmap all files that we're currently using and free all resources
440: * that we've allocated during the parse.
1.2 kristaps 441: * The utility should exit(...) after this is called.
442: */
1.1 kristaps 443: static void
444: texiexit(struct texi *p)
445: {
1.5 kristaps 446: size_t i;
447:
448: if (p->outcol)
449: putchar('\n');
1.1 kristaps 450:
451: while (p->filepos > 0)
452: texifilepop(p);
1.5 kristaps 453:
454: for (i = 0; i < p->dirsz; i++)
455: free(p->dirs[i]);
1.10 kristaps 456:
1.5 kristaps 457: free(p->dirs);
1.10 kristaps 458: free(p->subtitle);
459: free(p->title);
1.1 kristaps 460: }
461:
1.2 kristaps 462: /*
463: * Fatal error: unmap all files and exit.
464: * The "errstring" is passed to perror(3).
465: */
1.1 kristaps 466: static void
1.2 kristaps 467: texiabort(struct texi *p, const char *errstring)
1.1 kristaps 468: {
469:
470: perror(errstring);
471: texiexit(p);
472: exit(EXIT_FAILURE);
473: }
474:
475: /*
476: * Print a generic warning message (to stderr) tied to our current
477: * location in the parse sequence.
478: */
479: static void
480: texiwarn(const struct texi *p, const char *fmt, ...)
481: {
482: va_list ap;
483:
1.2 kristaps 484: fprintf(stderr, "%s:%zu:%zu: warning: ",
1.1 kristaps 485: p->files[p->filepos - 1].name,
486: p->files[p->filepos - 1].line + 1,
487: p->files[p->filepos - 1].col + 1);
488: va_start(ap, fmt);
489: vfprintf(stderr, fmt, ap);
490: va_end(ap);
491: fputc('\n', stderr);
492: }
493:
1.8 kristaps 494: /*
495: * Print an error message (to stderr) tied to our current location in
496: * the parse sequence, invoke texiexit(), then die.
497: */
1.2 kristaps 498: static void
499: texierr(struct texi *p, const char *fmt, ...)
500: {
501: va_list ap;
502:
503: fprintf(stderr, "%s:%zu:%zu: error: ",
504: p->files[p->filepos - 1].name,
505: p->files[p->filepos - 1].line + 1,
506: p->files[p->filepos - 1].col + 1);
507: va_start(ap, fmt);
508: vfprintf(stderr, fmt, ap);
509: va_end(ap);
510: fputc('\n', stderr);
511: texiexit(p);
512: exit(EXIT_FAILURE);
513: }
514:
1.1 kristaps 515: /*
1.8 kristaps 516: * Put a single data character to the output if we're not ignoring.
517: * Adjusts our output status.
1.1 kristaps 518: */
519: static void
520: texiputchar(struct texi *p, char c)
521: {
522:
1.3 kristaps 523: if (p->ign)
1.1 kristaps 524: return;
525: putchar(c);
1.11 kristaps 526: p->seenvs = 0;
1.1 kristaps 527: if ('\n' == c) {
528: p->outcol = 0;
529: p->seenws = 0;
530: } else
531: p->outcol++;
532: }
533:
534: /*
535: * Put multiple characters (see texiputchar()).
536: */
537: static void
538: texiputchars(struct texi *p, const char *s)
539: {
540:
541: while ('\0' != *s)
542: texiputchar(p, *s++);
543: }
544:
545: /*
1.8 kristaps 546: * Close an mdoc(7) macro opened with teximacroopen().
547: * If there are no more macros on the line, prints a newline.
1.1 kristaps 548: */
549: static void
1.3 kristaps 550: teximacroclose(struct texi *p)
551: {
552:
1.8 kristaps 553: /* FIXME: punctuation. */
1.5 kristaps 554: if (0 == --p->outmacro)
555: texiputchar(p, '\n');
1.3 kristaps 556: }
557:
558: /*
1.8 kristaps 559: * Open a mdoc(7) macro.
560: * This is used for line macros, e.g., Qq [foo bar baz].
561: * It can be invoked for nested macros, e.g., Qq Li foo .
1.3 kristaps 562: */
563: static void
564: teximacroopen(struct texi *p, const char *s)
1.1 kristaps 565: {
566:
1.5 kristaps 567: if (p->outcol && 0 == p->outmacro)
568: texiputchar(p, '\n');
569: if (0 == p->outmacro)
570: texiputchar(p, '.');
571: else
572: texiputchar(p, ' ');
573: texiputchars(p, s);
574: texiputchar(p, ' ');
1.3 kristaps 575: p->outmacro++;
1.5 kristaps 576: p->seenws = 0;
1.1 kristaps 577: }
578:
579: /*
1.8 kristaps 580: * Put a stadnalone mdoc(7) command with the trailing newline.
1.1 kristaps 581: */
582: static void
583: teximacro(struct texi *p, const char *s)
584: {
585:
1.4 kristaps 586: if (p->outmacro)
587: texierr(p, "\"%s\" in open line scope!?", s);
588: else if (p->literal)
589: texierr(p, "\"%s\" in a literal scope!?", s);
590:
1.1 kristaps 591: if (p->outcol)
592: texiputchar(p, '\n');
1.5 kristaps 593:
594: texiputchar(p, '.');
595: texiputchars(p, s);
596: texiputchar(p, '\n');
1.1 kristaps 597: }
598:
1.11 kristaps 599: static void
600: texivspace(struct texi *p)
601: {
602:
603: if (p->seenvs)
604: return;
605: teximacro(p, "Pp");
606: p->seenvs = 1;
607: }
608:
1.1 kristaps 609: /*
610: * Advance by a single byte in the input stream.
611: */
612: static void
613: advance(struct texi *p, const char *buf, size_t *pos)
614: {
615:
616: if ('\n' == buf[*pos]) {
617: p->files[p->filepos - 1].line++;
618: p->files[p->filepos - 1].col = 0;
619: } else
620: p->files[p->filepos - 1].col++;
621:
622: (*pos)++;
623: }
624:
1.11 kristaps 625: static void
626: texipunctuate(struct texi *p, const char *buf, size_t sz, size_t *pos)
627: {
628: size_t start, end;
629:
630: if (1 != p->outmacro)
631: return;
632:
633: for (start = end = *pos; end < sz; end++) {
634: switch (buf[end]) {
635: case (','):
636: case (')'):
637: case ('.'):
638: case ('"'):
639: case (':'):
640: case ('!'):
641: case ('?'):
642: continue;
643: default:
644: break;
645: }
646: break;
647: }
648: if (end == *pos)
649: return;
650: if (end + 1 == sz || ' ' == buf[end] || '\n' == buf[end]) {
651: for ( ; start < end; start++) {
652: texiputchar(p, ' ');
653: texiputchar(p, buf[start]);
654: advance(p, buf, pos);
655: }
656: }
657: }
658:
1.1 kristaps 659: /*
660: * Advance to the next non-whitespace word in the input stream.
661: * If we're in literal mode, then print all of the whitespace as we're
662: * doing so.
663: */
664: static size_t
665: advancenext(struct texi *p, const char *buf, size_t sz, size_t *pos)
666: {
667:
1.3 kristaps 668: if (p->literal) {
1.9 kristaps 669: while (*pos < sz && ismspace(buf[*pos])) {
1.5 kristaps 670: if (*pos && '\n' == buf[*pos] &&
671: '\\' == buf[*pos - 1])
672: texiputchar(p, 'e');
1.1 kristaps 673: texiputchar(p, buf[*pos]);
674: advance(p, buf, pos);
675: }
676: return(*pos);
677: }
678:
1.9 kristaps 679: while (*pos < sz && ismspace(buf[*pos])) {
1.1 kristaps 680: p->seenws = 1;
681: /*
682: * If it looks like we've printed a double-line, then
683: * output a paragraph.
684: * FIXME: this is stupid.
685: */
1.11 kristaps 686: if (*pos && '\n' == buf[*pos] && '\n' == buf[*pos - 1])
687: texivspace(p);
1.1 kristaps 688: advance(p, buf, pos);
689: }
690: return(*pos);
691: }
692:
693: /*
694: * Advance to the EOLN in the input stream.
695: */
696: static size_t
1.3 kristaps 697: advanceeoln(struct texi *p, const char *buf,
698: size_t sz, size_t *pos, int consumenl)
1.1 kristaps 699: {
700:
1.8 kristaps 701: /* FIXME: disregards @NEWLINE. */
1.1 kristaps 702: while (*pos < sz && '\n' != buf[*pos])
703: advance(p, buf, pos);
1.3 kristaps 704: if (*pos < sz && consumenl)
705: advance(p, buf, pos);
1.1 kristaps 706: return(*pos);
707: }
708:
709: /*
710: * Advance to position "end", which is an absolute position in the
711: * current buffer greater than or equal to the current position.
712: */
713: static void
714: advanceto(struct texi *p, const char *buf, size_t *pos, size_t end)
715: {
716:
717: assert(*pos <= end);
718: while (*pos < end)
719: advance(p, buf, pos);
720: }
721:
722: /*
723: * Output a free-form word in the input stream, progressing to the next
724: * command or white-space.
725: * This also will advance the input stream.
726: */
727: static void
1.8 kristaps 728: texiword(struct texi *p, const char *buf,
729: size_t sz, size_t *pos, char extra)
1.1 kristaps 730: {
731:
1.3 kristaps 732: if (0 == p->outmacro && p->outcol > 72 && 0 == p->literal)
1.1 kristaps 733: texiputchar(p, '\n');
1.8 kristaps 734: /* FIXME: abstract this: we use it elsewhere. */
1.3 kristaps 735: if (p->seenws && p->outcol && 0 == p->literal)
1.1 kristaps 736: texiputchar(p, ' ');
737:
738: p->seenws = 0;
739:
1.9 kristaps 740: while (*pos < sz && ! ismspace(buf[*pos])) {
1.1 kristaps 741: switch (buf[*pos]) {
742: case ('@'):
743: case ('}'):
744: case ('{'):
745: return;
746: }
1.8 kristaps 747: if ('\0' != extra && buf[*pos] == extra)
748: return;
1.1 kristaps 749: if (*pos < sz - 1 &&
750: '`' == buf[*pos] &&
751: '`' == buf[*pos + 1]) {
752: texiputchars(p, "\\(lq");
753: advance(p, buf, pos);
754: } else if (*pos < sz - 1 &&
755: '\'' == buf[*pos] &&
756: '\'' == buf[*pos + 1]) {
757: texiputchars(p, "\\(rq");
758: advance(p, buf, pos);
759: } else
760: texiputchar(p, buf[*pos]);
761: advance(p, buf, pos);
762: }
763: }
764:
1.8 kristaps 765: /*
766: * Look up the command at position "pos" in the buffer, returning it (or
767: * TEXICMD__MAX if none found) and setting "end" to be the absolute
768: * index after the command name.
769: */
1.1 kristaps 770: static enum texicmd
771: texicmd(struct texi *p, const char *buf,
772: size_t pos, size_t sz, size_t *end)
773: {
774: size_t i, len;
775:
776: assert('@' == buf[pos]);
1.3 kristaps 777:
1.9 kristaps 778: if ((*end = pos) == sz)
779: return(TEXICMD__MAX);
780: else if ((*end = ++pos) == sz)
1.3 kristaps 781: return(TEXICMD__MAX);
782:
783: /* Alphabetic commands are special. */
784: if ( ! isalpha(buf[pos])) {
1.9 kristaps 785: if ((*end = pos + 1) == sz)
786: return(TEXICMD__MAX);
1.3 kristaps 787: for (i = 0; i < TEXICMD__MAX; i++) {
788: if (1 != texitoks[i].len)
789: continue;
790: if (0 == strncmp(texitoks[i].tok, &buf[pos], 1))
791: return(i);
792: }
793: texiwarn(p, "bad command: @%c", buf[pos]);
794: return(TEXICMD__MAX);
795: }
796:
1.9 kristaps 797: for (*end = pos; *end < sz && ! ismspace(buf[*end]); (*end)++)
1.3 kristaps 798: if ((*end > pos && ('@' == buf[*end] ||
799: '{' == buf[*end] || '}' == buf[*end])))
1.1 kristaps 800: break;
801:
802: len = *end - pos;
803: for (i = 0; i < TEXICMD__MAX; i++) {
804: if (len != texitoks[i].len)
805: continue;
806: if (0 == strncmp(texitoks[i].tok, &buf[pos], len))
807: return(i);
808: }
809:
1.3 kristaps 810: texiwarn(p, "bad command: @%.*s", (int)len, &buf[pos]);
1.1 kristaps 811: return(TEXICMD__MAX);
812: }
813:
1.8 kristaps 814: /*
815: * Parse an argument from a bracketed command, e.g., @url{foo, baz}.
816: * Num should be set to the argument we're currently parsing, although
817: * it suffixes for it to be zero or non-zero.
818: * This will return 1 if there are more arguments, 0 otherwise.
819: * This will stop (returning 0) in the event of EOF or if we're not at a
820: * bracket for the zeroth parse.
821: */
822: static int
823: parsearg(struct texi *p, const char *buf,
824: size_t sz, size_t *pos, size_t num)
825: {
826: size_t end;
827: enum texicmd cmd;
828:
1.9 kristaps 829: while (*pos < sz && ismspace(buf[*pos]))
1.8 kristaps 830: advance(p, buf, pos);
831: if (*pos == sz || (0 == num && '{' != buf[*pos]))
832: return(0);
833: if (0 == num)
834: advance(p, buf, pos);
835:
836: while ((*pos = advancenext(p, buf, sz, pos)) < sz) {
837: switch (buf[*pos]) {
838: case (','):
839: advance(p, buf, pos);
840: return(1);
841: case ('}'):
842: advance(p, buf, pos);
843: return(0);
844: case ('{'):
845: if (0 == p->ign)
846: texiwarn(p, "unexpected \"{\"");
847: advance(p, buf, pos);
848: continue;
849: case ('@'):
850: break;
851: default:
852: texiword(p, buf, sz, pos, ',');
853: continue;
854: }
855:
856: cmd = texicmd(p, buf, *pos, sz, &end);
857: advanceto(p, buf, pos, end);
858: if (TEXICMD__MAX == cmd)
859: continue;
860: if (NULL != texitoks[cmd].fp)
861: (*texitoks[cmd].fp)(p, cmd, buf, sz, pos);
862: }
863: return(0);
864: }
865:
866: /*
867: * Parse until the end of a bracketed statement, e.g., @foo{bar baz}.
868: * This will stop in the event of EOF or if we're not at a bracket.
869: */
1.1 kristaps 870: static void
871: parsebracket(struct texi *p, const char *buf, size_t sz, size_t *pos)
872: {
873: size_t end;
874: enum texicmd cmd;
875:
1.9 kristaps 876: while (*pos < sz && ismspace(buf[*pos]))
1.3 kristaps 877: advance(p, buf, pos);
878:
1.1 kristaps 879: if (*pos == sz || '{' != buf[*pos])
880: return;
881: advance(p, buf, pos);
882:
883: while ((*pos = advancenext(p, buf, sz, pos)) < sz) {
884: switch (buf[*pos]) {
885: case ('}'):
886: advance(p, buf, pos);
887: return;
888: case ('{'):
1.3 kristaps 889: if (0 == p->ign)
890: texiwarn(p, "unexpected \"{\"");
891: advance(p, buf, pos);
892: continue;
893: case ('@'):
894: break;
895: default:
1.8 kristaps 896: texiword(p, buf, sz, pos, '\0');
1.3 kristaps 897: continue;
898: }
899:
900: cmd = texicmd(p, buf, *pos, sz, &end);
901: advanceto(p, buf, pos, end);
902: if (TEXICMD__MAX == cmd)
903: continue;
904: if (NULL != texitoks[cmd].fp)
905: (*texitoks[cmd].fp)(p, cmd, buf, sz, pos);
906: }
907: }
908:
909: /*
910: * This should be invoked when we're on a macro line and want to process
911: * to the end of the current input line, doing all of our macros along
912: * the way.
913: */
914: static void
915: parseeoln(struct texi *p, const char *buf, size_t sz, size_t *pos)
916: {
917: size_t end;
918: enum texicmd cmd;
919:
920: assert(0 == p->literal);
921:
922: while (*pos < sz && '\n' != buf[*pos]) {
923: while (*pos < sz && isws(buf[*pos])) {
924: p->seenws = 1;
925: advance(p, buf, pos);
926: }
927: switch (buf[*pos]) {
928: case ('}'):
929: if (0 == p->ign)
930: texiwarn(p, "unexpected \"}\"");
931: advance(p, buf, pos);
932: continue;
933: case ('{'):
934: if (0 == p->ign)
935: texiwarn(p, "unexpected \"{\"");
1.1 kristaps 936: advance(p, buf, pos);
937: continue;
938: case ('@'):
939: break;
940: default:
1.8 kristaps 941: texiword(p, buf, sz, pos, '\0');
1.1 kristaps 942: continue;
943: }
944:
945: cmd = texicmd(p, buf, *pos, sz, &end);
946: advanceto(p, buf, pos, end);
947: if (TEXICMD__MAX == cmd)
948: continue;
949: if (NULL != texitoks[cmd].fp)
950: (*texitoks[cmd].fp)(p, cmd, buf, sz, pos);
951: }
952: }
953:
1.8 kristaps 954: /*
955: * Parse a single word or command.
956: * This will return immediately at the EOF.
957: */
1.1 kristaps 958: static void
1.3 kristaps 959: parsesingle(struct texi *p, const char *buf, size_t sz, size_t *pos)
960: {
961: size_t end;
962: enum texicmd cmd;
963:
964: if ((*pos = advancenext(p, buf, sz, pos)) >= sz)
965: return;
966:
967: switch (buf[*pos]) {
968: case ('}'):
969: if (0 == p->ign)
970: texiwarn(p, "unexpected \"}\"");
971: advance(p, buf, pos);
972: return;
973: case ('{'):
974: if (0 == p->ign)
975: texiwarn(p, "unexpected \"{\"");
976: advance(p, buf, pos);
977: return;
978: case ('@'):
979: break;
980: default:
1.8 kristaps 981: texiword(p, buf, sz, pos, '\0');
1.3 kristaps 982: return;
983: }
984:
985: cmd = texicmd(p, buf, *pos, sz, &end);
986: advanceto(p, buf, pos, end);
987: if (TEXICMD__MAX == cmd)
988: return;
989: if (NULL != texitoks[cmd].fp)
990: (*texitoks[cmd].fp)(p, cmd, buf, sz, pos);
991: }
992:
1.15 kristaps 993: static int
994: parselinearg(struct texi *p, const char *buf, size_t sz, size_t *pos)
995: {
996:
997: while (*pos < sz && isws(buf[*pos])) {
998: p->seenws = 1;
999: advance(p, buf, pos);
1000: }
1001:
1002: if (*pos < sz && '{' == buf[*pos])
1003: parsebracket(p, buf, sz, pos);
1004: else if ('\n' != buf[*pos])
1005: parsesingle(p, buf, sz, pos);
1006: else
1007: return(0);
1008:
1009: return(1);
1010: }
1011:
1.8 kristaps 1012: /*
1013: * Parse til the end of the buffer.
1014: */
1.3 kristaps 1015: static void
1.7 kristaps 1016: parseeof(struct texi *p, const char *buf, size_t sz)
1017: {
1018: size_t pos;
1019:
1020: for (pos = 0; pos < sz; )
1021: parsesingle(p, buf, sz, &pos);
1022: }
1023:
1.8 kristaps 1024: /*
1025: * Parse a block sequence until we have the "@end endtoken" command
1026: * invocation.
1027: * This will return immediately at EOF.
1028: */
1.7 kristaps 1029: static void
1.1 kristaps 1030: parseto(struct texi *p, const char *buf,
1031: size_t sz, size_t *pos, const char *endtoken)
1032: {
1033: size_t end;
1034: enum texicmd cmd;
1035: size_t endtoksz;
1036:
1037: endtoksz = strlen(endtoken);
1038: assert(endtoksz > 0);
1039:
1040: while ((*pos = advancenext(p, buf, sz, pos)) < sz) {
1041: switch (buf[*pos]) {
1042: case ('}'):
1.3 kristaps 1043: if (0 == p->ign)
1044: texiwarn(p, "unexpected \"}\"");
1.1 kristaps 1045: advance(p, buf, pos);
1046: continue;
1047: case ('{'):
1.3 kristaps 1048: if (0 == p->ign)
1049: texiwarn(p, "unexpected \"{\"");
1.1 kristaps 1050: advance(p, buf, pos);
1051: continue;
1052: case ('@'):
1053: break;
1054: default:
1.8 kristaps 1055: texiword(p, buf, sz, pos, '\0');
1.1 kristaps 1056: continue;
1057: }
1058:
1059: cmd = texicmd(p, buf, *pos, sz, &end);
1060: advanceto(p, buf, pos, end);
1061: if (TEXICMD_END == cmd) {
1.2 kristaps 1062: while (*pos < sz && isws(buf[*pos]))
1.1 kristaps 1063: advance(p, buf, pos);
1064: /*
1.8 kristaps 1065: * FIXME: check the full word, not just its
1066: * initial substring!
1.1 kristaps 1067: */
1068: if (sz - *pos >= endtoksz && 0 == strncmp
1069: (&buf[*pos], endtoken, endtoksz)) {
1.3 kristaps 1070: advanceeoln(p, buf, sz, pos, 0);
1.1 kristaps 1071: break;
1072: }
1.3 kristaps 1073: if (0 == p->ign)
1074: texiwarn(p, "unexpected \"end\"");
1075: advanceeoln(p, buf, sz, pos, 0);
1.1 kristaps 1076: continue;
1077: } else if (TEXICMD__MAX != cmd)
1078: if (NULL != texitoks[cmd].fp)
1079: (*texitoks[cmd].fp)(p, cmd, buf, sz, pos);
1080: }
1081: }
1082:
1.8 kristaps 1083: /*
1084: * Memory-map the file "fname" and begin parsing it.
1085: * This can be called in a nested context.
1086: */
1.1 kristaps 1087: static void
1.16 ! kristaps 1088: parsefile(struct texi *p, const char *fname, int parse)
1.2 kristaps 1089: {
1.16 ! kristaps 1090: struct texifile *f;
! 1091: int fd;
! 1092: struct stat st;
! 1093: size_t i;
1.2 kristaps 1094:
1095: assert(p->filepos < 64);
1096: f = &p->files[p->filepos];
1097: memset(f, 0, sizeof(struct texifile));
1098:
1099: f->name = fname;
1100: if (-1 == (fd = open(fname, O_RDONLY, 0))) {
1101: texiabort(p, fname);
1102: } else if (-1 == fstat(fd, &st)) {
1103: close(fd);
1104: texiabort(p, fname);
1105: }
1106:
1107: f->mapsz = st.st_size;
1108: f->map = mmap(NULL, f->mapsz,
1109: PROT_READ, MAP_SHARED, fd, 0);
1110: close(fd);
1111:
1112: if (MAP_FAILED == f->map)
1113: texiabort(p, fname);
1114:
1115: p->filepos++;
1.16 ! kristaps 1116: if ( ! parse) {
! 1117: /*
! 1118: * We're printing verbatim output.
! 1119: * Make sure it doesn't get interpreted as mdoc by
! 1120: * escaping escapes and making sure leading dots don't
! 1121: * trigger mdoc(7) expansion.
! 1122: */
! 1123: for (i = 0; i < f->mapsz; i++) {
! 1124: if (i > 0 && '.' == f->map[i])
! 1125: if ('\n' == f->map[i - 1])
! 1126: fputs("\\&", stdout);
! 1127: putchar(f->map[i]);
! 1128: if ('\\' == f->map[i])
! 1129: putchar('e');
! 1130: }
! 1131: } else
! 1132: parseeof(p, f->map, f->mapsz);
1.2 kristaps 1133: texifilepop(p);
1134: }
1135:
1136: static void
1.15 kristaps 1137: dodefn(struct texi *p, enum texicmd cmd,
1.3 kristaps 1138: const char *buf, size_t sz, size_t *pos)
1139: {
1140: const char *blk;
1141:
1.5 kristaps 1142: blk = NULL;
1.3 kristaps 1143: switch (cmd) {
1.15 kristaps 1144: case (TEXICMD_DEFFN):
1145: case (TEXICMD_DEFTP):
1.3 kristaps 1146: case (TEXICMD_DEFTYPEFN):
1147: case (TEXICMD_DEFTYPEFUN):
1.15 kristaps 1148: case (TEXICMD_DEFTYPEVAR):
1149: case (TEXICMD_DEFTYPEVR):
1150: case (TEXICMD_DEFUN):
1151: case (TEXICMD_DEFVAR):
1152: case (TEXICMD_DEFVR):
1.5 kristaps 1153: blk = texitoks[cmd].tok;
1.3 kristaps 1154: break;
1.5 kristaps 1155: default:
1.3 kristaps 1156: break;
1157: }
1158:
1159: if (p->ign) {
1.15 kristaps 1160: NULL != blk ?
1161: parseto(p, buf, sz, pos, blk) :
1162: parseeoln(p, buf, sz, pos);
1.3 kristaps 1163: return;
1164: }
1165:
1.15 kristaps 1166: if (NULL != blk)
1167: texivspace(p);
1168:
1.3 kristaps 1169: switch (cmd) {
1.15 kristaps 1170: case (TEXICMD_DEFMAC):
1171: case (TEXICMD_DEFMACX):
1172: texiputchars(p, "Macro");
1173: break;
1174: case (TEXICMD_DEFTYPEVAR):
1175: case (TEXICMD_DEFTYPEVARX):
1176: case (TEXICMD_DEFVAR):
1177: case (TEXICMD_DEFVARX):
1178: texiputchars(p, "Variable");
1179: break;
1.3 kristaps 1180: case (TEXICMD_DEFTYPEFUN):
1.15 kristaps 1181: case (TEXICMD_DEFTYPEFUNX):
1182: case (TEXICMD_DEFUN):
1183: case (TEXICMD_DEFUNX):
1184: texiputchars(p, "Function");
1.3 kristaps 1185: break;
1186: default:
1.15 kristaps 1187: parselinearg(p, buf, sz, pos);
1.3 kristaps 1188: break;
1189: }
1.15 kristaps 1190:
1191: texiputchars(p, ":\n");
1192:
1193: switch (cmd) {
1194: case (TEXICMD_DEFMAC):
1195: case (TEXICMD_DEFMACX):
1196: teximacroopen(p, "Dv");
1197: while (parselinearg(p, buf, sz, pos))
1198: /* Spin. */ ;
1199: teximacroclose(p);
1200: break;
1201: case (TEXICMD_DEFFN):
1202: case (TEXICMD_DEFFNX):
1203: case (TEXICMD_DEFUN):
1204: case (TEXICMD_DEFUNX):
1205: teximacroopen(p, "Fo");
1206: parselinearg(p, buf, sz, pos);
1207: teximacroclose(p);
1208: teximacroopen(p, "Fa");
1209: while (parselinearg(p, buf, sz, pos))
1210: /* Spin. */ ;
1211: teximacroclose(p);
1212: teximacro(p, "Fc");
1213: break;
1214: case (TEXICMD_DEFTYPEFUN):
1215: case (TEXICMD_DEFTYPEFUNX):
1216: case (TEXICMD_DEFTYPEFN):
1217: case (TEXICMD_DEFTYPEFNX):
1218: teximacroopen(p, "Ft");
1219: parselinearg(p, buf, sz, pos);
1220: teximacroclose(p);
1221: teximacroopen(p, "Fo");
1222: parselinearg(p, buf, sz, pos);
1223: teximacroclose(p);
1224: teximacroopen(p, "Fa");
1225: while (parselinearg(p, buf, sz, pos))
1226: /* Spin. */ ;
1227: teximacroclose(p);
1228: teximacro(p, "Fc");
1229: break;
1230: case (TEXICMD_DEFTP):
1231: case (TEXICMD_DEFTPX):
1232: case (TEXICMD_DEFTYPEVAR):
1233: case (TEXICMD_DEFTYPEVARX):
1234: case (TEXICMD_DEFTYPEVR):
1235: case (TEXICMD_DEFTYPEVRX):
1236: teximacroopen(p, "Vt");
1237: while (parselinearg(p, buf, sz, pos))
1238: /* Spin. */ ;
1239: teximacroclose(p);
1240: break;
1241: case (TEXICMD_DEFVAR):
1242: case (TEXICMD_DEFVARX):
1243: case (TEXICMD_DEFVR):
1244: case (TEXICMD_DEFVRX):
1245: teximacroopen(p, "Va");
1246: while (parselinearg(p, buf, sz, pos))
1247: /* Spin. */ ;
1248: teximacroclose(p);
1249: break;
1250: default:
1251: abort();
1.3 kristaps 1252: }
1.15 kristaps 1253:
1.11 kristaps 1254: texivspace(p);
1.3 kristaps 1255: if (NULL != blk)
1256: parseto(p, buf, sz, pos, blk);
1257: }
1258:
1259: static void
1.1 kristaps 1260: doignblock(struct texi *p, enum texicmd cmd,
1261: const char *buf, size_t sz, size_t *pos)
1262: {
1263:
1.3 kristaps 1264: p->ign++;
1.5 kristaps 1265: parseto(p, buf, sz, pos, texitoks[cmd].tok);
1.3 kristaps 1266: p->ign--;
1.1 kristaps 1267: }
1268:
1269: static void
1.3 kristaps 1270: doblock(struct texi *p, enum texicmd cmd,
1.1 kristaps 1271: const char *buf, size_t sz, size_t *pos)
1272: {
1273:
1.5 kristaps 1274: parseto(p, buf, sz, pos, texitoks[cmd].tok);
1.1 kristaps 1275: }
1276:
1277: static void
1.12 kristaps 1278: doinline(struct texi *p, enum texicmd cmd,
1279: const char *buf, size_t sz, size_t *pos)
1.1 kristaps 1280: {
1.12 kristaps 1281: const char *macro;
1282:
1283: switch (cmd) {
1284: case (TEXICMD_COMMAND):
1285: macro = "Xr";
1286: break;
1287: case (TEXICMD_ENV):
1288: macro = "Ev";
1289: break;
1290: case (TEXICMD_FILE):
1291: macro = "Pa";
1292: break;
1.16 ! kristaps 1293: case (TEXICMD_OPTION):
! 1294: macro = "Op";
! 1295: break;
1.12 kristaps 1296: case (TEXICMD_VAR):
1297: macro = "Va";
1298: break;
1299: default:
1300: abort();
1301: }
1302:
1303: if (p->literal) {
1304: parsebracket(p, buf, sz, pos);
1305: return;
1306: }
1.1 kristaps 1307:
1.5 kristaps 1308: teximacroopen(p, macro);
1.1 kristaps 1309: p->seenws = 0;
1310: parsebracket(p, buf, sz, pos);
1.11 kristaps 1311: texipunctuate(p, buf, sz, pos);
1.5 kristaps 1312: teximacroclose(p);
1.1 kristaps 1313: }
1314:
1315: static void
1.16 ! kristaps 1316: doverbinclude(struct texi *p, enum texicmd cmd,
! 1317: const char *buf, size_t sz, size_t *pos)
! 1318: {
! 1319: char fname[PATH_MAX], path[PATH_MAX];
! 1320: int rc;
! 1321: size_t i;
! 1322:
! 1323: while (*pos < sz && ' ' == buf[*pos])
! 1324: advance(p, buf, pos);
! 1325:
! 1326: /* Read in the filename. */
! 1327: for (i = 0; *pos < sz && '\n' != buf[*pos]; i++) {
! 1328: if (i == sizeof(fname) - 1)
! 1329: break;
! 1330: fname[i] = buf[*pos];
! 1331: advance(p, buf, pos);
! 1332: }
! 1333:
! 1334: if (i == 0)
! 1335: texierr(p, "path too short");
! 1336: else if ('\n' != buf[*pos])
! 1337: texierr(p, "path too long");
! 1338: else if ('/' == fname[0])
! 1339: texierr(p, "no absolute paths");
! 1340: fname[i] = '\0';
! 1341:
! 1342: if (strstr(fname, "../") || strstr(fname, "/.."))
! 1343: texierr(p, "insecure path");
! 1344:
! 1345: rc = snprintf(path, sizeof(path),
! 1346: "%s/%s", p->dirs[0], fname);
! 1347: if (rc < 0)
! 1348: texierr(p, "couldn't format path");
! 1349: else if ((size_t)rc >= sizeof(path))
! 1350: texierr(p, "path too long");
! 1351:
! 1352: parsefile(p, path, 0);
! 1353: }
! 1354:
! 1355: static void
1.2 kristaps 1356: doinclude(struct texi *p, enum texicmd cmd,
1357: const char *buf, size_t sz, size_t *pos)
1358: {
1359: char fname[PATH_MAX], path[PATH_MAX];
1360: size_t i;
1361: int rc;
1362:
1363: while (*pos < sz && ' ' == buf[*pos])
1364: advance(p, buf, pos);
1365:
1366: /* Read in the filename. */
1367: for (i = 0; *pos < sz && '\n' != buf[*pos]; i++) {
1368: if (i == sizeof(fname) - 1)
1369: break;
1370: fname[i] = buf[*pos];
1371: advance(p, buf, pos);
1372: }
1373:
1374: if (i == 0)
1375: texierr(p, "path too short");
1376: else if ('\n' != buf[*pos])
1377: texierr(p, "path too long");
1378: else if ('/' == fname[0])
1379: texierr(p, "no absolute paths");
1380: fname[i] = '\0';
1381:
1382: if (strstr(fname, "../") || strstr(fname, "/.."))
1383: texierr(p, "insecure path");
1384:
1.5 kristaps 1385: for (i = 0; i < p->dirsz; i++) {
1386: rc = snprintf(path, sizeof(path),
1387: "%s/%s", p->dirs[i], fname);
1388: if (rc < 0)
1389: texierr(p, "couldn't format path");
1390: else if ((size_t)rc >= sizeof(path))
1391: texierr(p, "path too long");
1392: else if (-1 == access(path, R_OK))
1393: continue;
1394:
1.16 ! kristaps 1395: parsefile(p, path, 1);
1.5 kristaps 1396: return;
1397: }
1.2 kristaps 1398:
1.5 kristaps 1399: texierr(p, "couldn't find %s in includes", fname);
1.2 kristaps 1400: }
1401:
1402: static void
1.12 kristaps 1403: dofont(struct texi *p, enum texicmd cmd,
1.1 kristaps 1404: const char *buf, size_t sz, size_t *pos)
1405: {
1.12 kristaps 1406: const char *font;
1407:
1408: switch (cmd) {
1409: case (TEXICMD_B):
1410: case (TEXICMD_STRONG):
1411: font = "\\fB";
1412: break;
1413: case (TEXICMD_CITE):
1414: case (TEXICMD_DFN):
1415: case (TEXICMD_EMPH):
1416: case (TEXICMD_I):
1417: case (TEXICMD_SLANTED):
1418: font = "\\fI";
1419: break;
1420: case (TEXICMD_CODE):
1421: case (TEXICMD_KBD):
1422: case (TEXICMD_R):
1423: case (TEXICMD_SAMP):
1424: case (TEXICMD_SANSSERIF):
1425: case (TEXICMD_T):
1426: font = "\\fR";
1427: break;
1428: default:
1429: abort();
1430: }
1.1 kristaps 1431:
1.12 kristaps 1432: texiputchars(p, font);
1.1 kristaps 1433: parsebracket(p, buf, sz, pos);
1434: texiputchars(p, "\\fP");
1435: }
1436:
1437: static void
1438: dobracket(struct texi *p, enum texicmd cmd,
1439: const char *buf, size_t sz, size_t *pos)
1440: {
1441:
1442: parsebracket(p, buf, sz, pos);
1443: }
1444:
1445: static void
1.3 kristaps 1446: dodisplay(struct texi *p, enum texicmd cmd,
1447: const char *buf, size_t sz, size_t *pos)
1448: {
1449:
1.13 kristaps 1450: teximacro(p, "Bd -filled -offset indent");
1.11 kristaps 1451: p->seenvs = 1;
1.12 kristaps 1452: /* FIXME: ignore and parseeoln. */
1.3 kristaps 1453: advanceeoln(p, buf, sz, pos, 1);
1.13 kristaps 1454: parseto(p, buf, sz, pos, texitoks[cmd].tok);
1.5 kristaps 1455: teximacro(p, "Ed");
1.3 kristaps 1456: }
1457:
1458: static void
1.1 kristaps 1459: doexample(struct texi *p, enum texicmd cmd,
1460: const char *buf, size_t sz, size_t *pos)
1461: {
1462:
1.5 kristaps 1463: teximacro(p, "Bd -literal -offset indent");
1.12 kristaps 1464: /* FIXME: ignore and parseeoln. */
1.3 kristaps 1465: advanceeoln(p, buf, sz, pos, 1);
1466: p->literal++;
1.13 kristaps 1467: parseto(p, buf, sz, pos, texitoks[cmd].tok);
1.3 kristaps 1468: p->literal--;
1.5 kristaps 1469: teximacro(p, "Ed");
1.1 kristaps 1470: }
1471:
1472: static void
1473: dobye(struct texi *p, enum texicmd cmd,
1474: const char *buf, size_t sz, size_t *pos)
1475: {
1476:
1477: texiexit(p);
1478: exit(EXIT_SUCCESS);
1479: }
1480:
1481: static void
1.10 kristaps 1482: dotitle(struct texi *p, enum texicmd cmd,
1483: const char *buf, size_t sz, size_t *pos)
1484: {
1485: size_t start, end;
1486:
1487: while (*pos < sz && isws(buf[*pos]))
1488: advance(p, buf, pos);
1489: start = end = *pos;
1490: while (end < sz && '\n' != buf[end])
1491: end++;
1492: free(p->subtitle);
1493: p->subtitle = malloc(end - start + 1);
1494: memcpy(p->subtitle, &buf[start], end - start);
1495: p->subtitle[end - start] = '\0';
1496: }
1497:
1498: static void
1.16 ! kristaps 1499: doaccent(struct texi *p, enum texicmd cmd,
! 1500: const char *buf, size_t sz, size_t *pos)
! 1501: {
! 1502:
! 1503: if (*pos == sz)
! 1504: return;
! 1505: advance(p, buf, pos);
! 1506: switch (cmd) {
! 1507: case (TEXICMD_ACUTE):
! 1508: switch (buf[*pos]) {
! 1509: case ('a'): case ('A'):
! 1510: case ('e'): case ('E'):
! 1511: case ('i'): case ('I'):
! 1512: case ('o'): case ('O'):
! 1513: case ('u'): case ('U'):
! 1514: texiputchars(p, "\\(\'");
! 1515: texiputchar(p, buf[*pos]);
! 1516: break;
! 1517: default:
! 1518: texiputchar(p, buf[*pos]);
! 1519: }
! 1520: break;
! 1521: case (TEXICMD_CIRCUMFLEX):
! 1522: switch (buf[*pos]) {
! 1523: case ('a'): case ('A'):
! 1524: case ('e'): case ('E'):
! 1525: case ('i'): case ('I'):
! 1526: case ('o'): case ('O'):
! 1527: case ('u'): case ('U'):
! 1528: texiputchars(p, "\\(^");
! 1529: texiputchar(p, buf[*pos]);
! 1530: break;
! 1531: default:
! 1532: texiputchar(p, buf[*pos]);
! 1533: }
! 1534: break;
! 1535: case (TEXICMD_GRAVE):
! 1536: switch (buf[*pos]) {
! 1537: case ('a'): case ('A'):
! 1538: case ('e'): case ('E'):
! 1539: case ('i'): case ('I'):
! 1540: case ('o'): case ('O'):
! 1541: case ('u'): case ('U'):
! 1542: texiputchars(p, "\\(`");
! 1543: texiputchar(p, buf[*pos]);
! 1544: break;
! 1545: default:
! 1546: texiputchar(p, buf[*pos]);
! 1547: }
! 1548: break;
! 1549: case (TEXICMD_TILDE):
! 1550: switch (buf[*pos]) {
! 1551: case ('a'): case ('A'):
! 1552: case ('n'): case ('N'):
! 1553: case ('o'): case ('O'):
! 1554: texiputchars(p, "\\(~");
! 1555: texiputchar(p, buf[*pos]);
! 1556: break;
! 1557: default:
! 1558: texiputchar(p, buf[*pos]);
! 1559: }
! 1560: break;
! 1561: case (TEXICMD_UMLAUT):
! 1562: switch (buf[*pos]) {
! 1563: case ('a'): case ('A'):
! 1564: case ('e'): case ('E'):
! 1565: case ('i'): case ('I'):
! 1566: case ('o'): case ('O'):
! 1567: case ('u'): case ('U'):
! 1568: case ('y'):
! 1569: texiputchars(p, "\\(:");
! 1570: texiputchar(p, buf[*pos]);
! 1571: break;
! 1572: default:
! 1573: texiputchar(p, buf[*pos]);
! 1574: }
! 1575: break;
! 1576: default:
! 1577: abort();
! 1578: }
! 1579: }
! 1580:
! 1581: static void
1.1 kristaps 1582: dosymbol(struct texi *p, enum texicmd cmd,
1583: const char *buf, size_t sz, size_t *pos)
1584: {
1585:
1.3 kristaps 1586: if (p->seenws && p->outcol && 0 == p->literal) {
1587: texiputchar(p, ' ');
1588: p->seenws = 0;
1589: }
1590:
1.1 kristaps 1591: switch (cmd) {
1.3 kristaps 1592: case (TEXICMD_ASTERISK):
1593: case (TEXICMD_NEWLINE):
1594: case (TEXICMD_SPACE):
1595: case (TEXICMD_TAB):
1596: texiputchar(p, ' ');
1597: break;
1.1 kristaps 1598: case (TEXICMD_AT):
1.3 kristaps 1599: texiputchar(p, '@');
1600: break;
1601: case (TEXICMD_BANG):
1602: texiputchar(p, '!');
1.7 kristaps 1603: break;
1604: case (TEXICMD_BULLET):
1605: texiputchars(p, "\\(bu");
1.1 kristaps 1606: break;
1607: case (TEXICMD_COPYRIGHT):
1608: texiputchars(p, "\\(co");
1609: break;
1.2 kristaps 1610: case (TEXICMD_DOTS):
1611: texiputchars(p, "...");
1612: break;
1.15 kristaps 1613: case (TEXICMD_ERROR):
1614: texiputchars(p, "error\\(->");
1615: break;
1.1 kristaps 1616: case (TEXICMD_LATEX):
1617: texiputchars(p, "LaTeX");
1618: break;
1.3 kristaps 1619: case (TEXICMD_QUESTIONMARK):
1620: texiputchar(p, '?');
1.15 kristaps 1621: break;
1622: case (TEXICMD_RESULT):
1623: texiputchars(p, "\\(rA");
1.3 kristaps 1624: break;
1625: case (TEXICMD_SQUIGGLE_LEFT):
1626: texiputchars(p, "{");
1627: break;
1628: case (TEXICMD_SQUIGGLE_RIGHT):
1629: texiputchars(p, "}");
1630: break;
1.1 kristaps 1631: case (TEXICMD_TEXSYM):
1632: texiputchars(p, "TeX");
1633: break;
1.3 kristaps 1634: case (TEXICMD_COLON):
1635: case (TEXICMD_HYPHEN):
1636: break;
1.1 kristaps 1637: default:
1.5 kristaps 1638: texiwarn(p, "sym: %d", cmd);
1.1 kristaps 1639: abort();
1640: }
1641:
1.5 kristaps 1642: if (texitoks[cmd].len > 1)
1643: doignbracket(p, cmd, buf, sz, pos);
1.1 kristaps 1644: }
1645:
1646: static void
1647: doquotation(struct texi *p, enum texicmd cmd,
1648: const char *buf, size_t sz, size_t *pos)
1649: {
1650:
1.5 kristaps 1651: teximacro(p, "Qo");
1.1 kristaps 1652: parseto(p, buf, sz, pos, "quotation");
1.5 kristaps 1653: teximacro(p, "Qc");
1.1 kristaps 1654: }
1655:
1.3 kristaps 1656: static void
1657: domath(struct texi *p, enum texicmd cmd,
1658: const char *buf, size_t sz, size_t *pos)
1659: {
1660: size_t nest;
1661:
1662: /*
1663: * Math handling is different from everything else.
1664: * We don't allow any subcomponents, and we ignore the rules in
1665: * terms of @-commands.
1666: * This departs from GNU's rules, but whatever.
1667: */
1668: while (*pos < sz && isws(buf[*pos]))
1669: advance(p, buf, pos);
1670: if (*pos == sz || '{' != buf[*pos])
1671: return;
1672: advance(p, buf, pos);
1673: if (p->seenws && p->outcol && 0 == p->literal)
1674: texiputchar(p, ' ');
1675: p->seenws = 0;
1676: for (nest = 1; *pos < sz && nest > 0; ) {
1677: if ('{' == buf[*pos])
1678: nest++;
1679: else if ('}' == buf[*pos])
1680: if (0 == --nest)
1681: continue;
1682: texiputchar(p, buf[*pos]);
1683: advance(p, buf, pos);
1684: }
1685: if (*pos == sz)
1686: return;
1687: assert('}' == buf[*pos]);
1688: advance(p, buf, pos);
1689: }
1690:
1.1 kristaps 1691: static void
1.8 kristaps 1692: dolink(struct texi *p, enum texicmd cmd,
1.1 kristaps 1693: const char *buf, size_t sz, size_t *pos)
1694: {
1.8 kristaps 1695: int c;
1.1 kristaps 1696:
1697: switch (cmd) {
1698: case (TEXICMD_EMAIL):
1.5 kristaps 1699: teximacroopen(p, "Mt");
1.1 kristaps 1700: break;
1.3 kristaps 1701: case (TEXICMD_UREF):
1.1 kristaps 1702: case (TEXICMD_URL):
1.5 kristaps 1703: teximacroopen(p, "Lk");
1.1 kristaps 1704: break;
1.8 kristaps 1705: case (TEXICMD_XREF):
1706: texiputchars(p, "See Section");
1707: teximacroopen(p, "Qq");
1708: break;
1709: case (TEXICMD_PXREF):
1710: texiputchars(p, "see Section");
1711: teximacroopen(p, "Qq");
1712: break;
1.1 kristaps 1713: default:
1.8 kristaps 1714: abort();
1.1 kristaps 1715: }
1.8 kristaps 1716:
1717: c = parsearg(p, buf, sz, pos, 0);
1718: p->ign++;
1719: while (c > 0)
1720: c = parsearg(p, buf, sz, pos, 1);
1721: p->ign--;
1722:
1.11 kristaps 1723: texipunctuate(p, buf, sz, pos);
1.8 kristaps 1724: teximacroclose(p);
1725: }
1726:
1727: static void
1728: doignargn(struct texi *p, enum texicmd cmd,
1729: const char *buf, size_t sz, size_t *pos)
1730: {
1731: int c;
1732:
1733: c = parsearg(p, buf, sz, pos, 0);
1734: p->ign++;
1735: while (c > 0)
1736: c = parsearg(p, buf, sz, pos, 1);
1737: p->ign--;
1.1 kristaps 1738: }
1739:
1740: static void
1741: dosubsection(struct texi *p, enum texicmd cmd,
1742: const char *buf, size_t sz, size_t *pos)
1743: {
1.13 kristaps 1744:
1745: if (p->outmacro)
1746: texierr(p, "\"Em\" in open line scope!?");
1747: else if (p->literal)
1748: texierr(p, "\"Em\" in a literal scope!?");
1.1 kristaps 1749:
1.11 kristaps 1750: texivspace(p);
1.5 kristaps 1751: teximacroopen(p, "Em");
1.3 kristaps 1752: parseeoln(p, buf, sz, pos);
1.5 kristaps 1753: teximacroclose(p);
1.11 kristaps 1754: texivspace(p);
1.1 kristaps 1755: }
1756:
1757: static void
1758: dosection(struct texi *p, enum texicmd cmd,
1759: const char *buf, size_t sz, size_t *pos)
1760: {
1.12 kristaps 1761: const char *blk;
1762:
1763: switch (cmd) {
1764: case (TEXICMD_APPENDIX):
1765: case (TEXICMD_CHAPTER):
1766: case (TEXICMD_TOP):
1767: case (TEXICMD_UNNUMBERED):
1768: blk = "Sh";
1769: break;
1770: case (TEXICMD_APPENDIXSEC):
1771: case (TEXICMD_HEADING):
1772: case (TEXICMD_SECTION):
1773: case (TEXICMD_UNNUMBEREDSEC):
1774: blk = "Ss";
1775: break;
1776: default:
1777: abort();
1778: }
1.1 kristaps 1779:
1.3 kristaps 1780: if (p->outmacro)
1.12 kristaps 1781: texierr(p, "%s in open line scope!?", blk);
1.3 kristaps 1782: else if (p->literal)
1.12 kristaps 1783: texierr(p, "%s in a literal scope!?", blk);
1.3 kristaps 1784:
1.12 kristaps 1785: teximacroopen(p, blk);
1.3 kristaps 1786: parseeoln(p, buf, sz, pos);
1787: teximacroclose(p);
1.11 kristaps 1788: p->seenvs = 1;
1.3 kristaps 1789: }
1790:
1791: static void
1792: dosp(struct texi *p, enum texicmd cmd,
1793: const char *buf, size_t sz, size_t *pos)
1794: {
1795:
1.11 kristaps 1796: texivspace(p);
1.12 kristaps 1797: /* FIXME: ignore and parseeoln. */
1.3 kristaps 1798: advanceeoln(p, buf, sz, pos, 1);
1.1 kristaps 1799: }
1800:
1801: static void
1802: dotop(struct texi *p, enum texicmd cmd,
1803: const char *buf, size_t sz, size_t *pos)
1804: {
1.10 kristaps 1805: const char *cp;
1806: time_t t;
1807: char date[32];
1808:
1809: t = time(NULL);
1810: strftime(date, sizeof(date), "%F", localtime(&t));
1.1 kristaps 1811:
1.3 kristaps 1812: p->ign--;
1.10 kristaps 1813: teximacroopen(p, "Dd");
1814: texiputchars(p, date);
1815: teximacroclose(p);
1816: teximacroopen(p, "Dt");
1817: for (cp = p->title; '\0' != *cp; cp++)
1818: texiputchar(p, toupper(*cp));
1.11 kristaps 1819: texiputchars(p, " 7");
1.10 kristaps 1820: teximacroclose(p);
1.5 kristaps 1821: teximacro(p, "Os");
1822: teximacro(p, "Sh NAME");
1.10 kristaps 1823: teximacroopen(p, "Nm");
1824: texiputchars(p, p->title);
1825: teximacroclose(p);
1826: teximacroopen(p, "Nd");
1827: texiputchars(p, NULL != p->subtitle ?
1828: p->subtitle : "Unknown description");
1829: teximacroclose(p);
1.11 kristaps 1830: p->seenvs = 1;
1.12 kristaps 1831: dosection(p, cmd, buf, sz, pos);
1.1 kristaps 1832: }
1833:
1834: static void
1835: doitem(struct texi *p, enum texicmd cmd,
1836: const char *buf, size_t sz, size_t *pos)
1837: {
1838:
1.3 kristaps 1839: if (p->outmacro)
1840: texierr(p, "item in open line scope!?");
1841: else if (p->literal)
1842: texierr(p, "item in a literal scope!?");
1843:
1844: switch (p->list) {
1845: case (TEXILIST_ITEM):
1.5 kristaps 1846: teximacroopen(p, "It");
1.3 kristaps 1847: break;
1848: case (TEXILIST_NOITEM):
1.5 kristaps 1849: teximacro(p, "It");
1.3 kristaps 1850: break;
1851: default:
1.11 kristaps 1852: texivspace(p);
1.3 kristaps 1853: break;
1854: }
1.11 kristaps 1855: p->seenvs = 1;
1.3 kristaps 1856:
1857: parseeoln(p, buf, sz, pos);
1.1 kristaps 1858:
1.3 kristaps 1859: if (TEXILIST_ITEM == p->list)
1860: teximacroclose(p);
1.9 kristaps 1861: else if (p->outcol > 0)
1.1 kristaps 1862: texiputchar(p, '\n');
1863: }
1864:
1865: static void
1866: dotable(struct texi *p, enum texicmd cmd,
1867: const char *buf, size_t sz, size_t *pos)
1868: {
1.3 kristaps 1869: enum texilist sv = p->list;
1870:
1871: p->list = TEXILIST_ITEM;
1.5 kristaps 1872: teximacro(p, "Bl -tag -width Ds");
1.12 kristaps 1873: /* FIXME: ignore and parseeoln. */
1874: advanceeoln(p, buf, sz, pos, 1);
1.11 kristaps 1875: p->seenvs = 1;
1.1 kristaps 1876: parseto(p, buf, sz, pos, "table");
1.5 kristaps 1877: teximacro(p, "El");
1.3 kristaps 1878: p->list = sv;
1.1 kristaps 1879: }
1880:
1881: static void
1.2 kristaps 1882: doenumerate(struct texi *p, enum texicmd cmd,
1883: const char *buf, size_t sz, size_t *pos)
1884: {
1.3 kristaps 1885: enum texilist sv = p->list;
1.2 kristaps 1886:
1.3 kristaps 1887: p->list = TEXILIST_NOITEM;
1.5 kristaps 1888: teximacro(p, "Bl -enum");
1.11 kristaps 1889: p->seenvs = 1;
1.12 kristaps 1890: /* FIXME: ignore and parseeoln. */
1891: advanceeoln(p, buf, sz, pos, 1);
1.2 kristaps 1892: parseto(p, buf, sz, pos, "enumerate");
1.5 kristaps 1893: teximacro(p, "El");
1.3 kristaps 1894: p->list = sv;
1.2 kristaps 1895: }
1896:
1897: static void
1.1 kristaps 1898: doitemize(struct texi *p, enum texicmd cmd,
1899: const char *buf, size_t sz, size_t *pos)
1900: {
1.3 kristaps 1901: enum texilist sv = p->list;
1.1 kristaps 1902:
1.3 kristaps 1903: p->list = TEXILIST_ITEM;
1.5 kristaps 1904: teximacro(p, "Bl -bullet");
1.11 kristaps 1905: p->seenvs = 1;
1.12 kristaps 1906: /* FIXME: ignore and parseeoln. */
1907: advanceeoln(p, buf, sz, pos, 1);
1.1 kristaps 1908: parseto(p, buf, sz, pos, "itemize");
1.5 kristaps 1909: teximacro(p, "El");
1.3 kristaps 1910: p->list = sv;
1.1 kristaps 1911: }
1912:
1913: static void
1914: doignbracket(struct texi *p, enum texicmd cmd,
1915: const char *buf, size_t sz, size_t *pos)
1916: {
1917:
1.3 kristaps 1918: p->ign++;
1.1 kristaps 1919: parsebracket(p, buf, sz, pos);
1.3 kristaps 1920: p->ign--;
1.1 kristaps 1921: }
1922:
1923: static void
1924: doignline(struct texi *p, enum texicmd cmd,
1925: const char *buf, size_t sz, size_t *pos)
1926: {
1927:
1.12 kristaps 1928: /* FIXME: ignore and parseeoln. */
1.3 kristaps 1929: advanceeoln(p, buf, sz, pos, 1);
1.1 kristaps 1930: }
1931:
1.8 kristaps 1932: /*
1933: * Parse colon-separated directories from "cp" (if not NULL) and returns
1934: * the array of pointers.
1935: * Prepends "base" to the array.
1936: * This does NOT sanitise the directories!
1937: */
1.5 kristaps 1938: static char **
1939: parsedirs(const char *base, const char *cp, size_t *sz)
1940: {
1941: char *tok, *str, *tofree;
1942: const char *cpp;
1943: size_t i;
1944: char **dirs;
1945:
1946: *sz = NULL != (cpp = cp) ? 2 : 1;
1947: if (*sz > 1)
1948: for ( ; NULL != (cpp = strchr(cpp, ':')); (*sz)++)
1949: cpp++;
1950:
1951: dirs = calloc(*sz, sizeof(char *));
1952: if (NULL == dirs) {
1953: perror(NULL);
1954: exit(EXIT_FAILURE);
1955: } else if (NULL == (dirs[0] = strdup(base))) {
1956: perror(NULL);
1957: exit(EXIT_FAILURE);
1958: }
1959:
1960: if (NULL == cp)
1961: return(dirs);
1962:
1963: if (NULL == (tofree = tok = str = strdup(cp))) {
1964: perror(NULL);
1965: exit(EXIT_FAILURE);
1966: }
1967:
1968: for (i = 1; NULL != (tok = strsep(&str, ":")); i++)
1969: if (NULL == (dirs[i] = strdup(tok))) {
1970: perror(NULL);
1971: exit(EXIT_FAILURE);
1972: }
1973:
1974: free(tofree);
1975: return(dirs);
1976: }
1977:
1.1 kristaps 1978: int
1979: main(int argc, char *argv[])
1980: {
1981: struct texi texi;
1.2 kristaps 1982: int c;
1983: char *path, *dir;
1.10 kristaps 1984: const char *progname, *Idir, *cp;
1.1 kristaps 1985:
1986: progname = strrchr(argv[0], '/');
1987: if (progname == NULL)
1988: progname = argv[0];
1989: else
1990: ++progname;
1991:
1.10 kristaps 1992: memset(&texi, 0, sizeof(struct texi));
1.5 kristaps 1993: Idir = NULL;
1.10 kristaps 1994:
1.5 kristaps 1995: while (-1 != (c = getopt(argc, argv, "I:")))
1.1 kristaps 1996: switch (c) {
1.5 kristaps 1997: case ('I'):
1998: Idir = optarg;
1999: break;
1.1 kristaps 2000: default:
2001: goto usage;
2002: }
2003:
2004: argv += optind;
2005: if (0 == (argc -= optind))
2006: goto usage;
2007:
1.2 kristaps 2008: if (NULL == (path = strdup(argv[0]))) {
2009: perror(NULL);
2010: exit(EXIT_FAILURE);
2011: } else if (NULL == (dir = dirname(path))) {
2012: perror(argv[0]);
2013: free(path);
2014: exit(EXIT_FAILURE);
2015: }
2016: free(path);
2017:
1.10 kristaps 2018: if (NULL != (cp = strrchr(argv[0], '/')))
2019: texi.title = strdup(cp + 1);
2020: else
2021: texi.title = strdup(argv[0]);
2022:
2023: if (NULL == texi.title) {
2024: perror(NULL);
2025: exit(EXIT_FAILURE);
2026: } else if (NULL != (path = strchr(texi.title, '.')))
2027: *path = '\0';
2028:
1.3 kristaps 2029: texi.ign = 1;
1.5 kristaps 2030: texi.dirs = parsedirs(dir, Idir, &texi.dirsz);
1.16 ! kristaps 2031: parsefile(&texi, argv[0], 1);
1.5 kristaps 2032: /* We shouldn't get here. */
1.2 kristaps 2033: texiexit(&texi);
2034: return(EXIT_FAILURE);
1.1 kristaps 2035: usage:
1.8 kristaps 2036: fprintf(stderr, "usage: %s [-Idirs] file\n", progname);
1.1 kristaps 2037: return(EXIT_FAILURE);
2038: }
CVSweb