=================================================================== RCS file: /cvs/texi2mdoc/main.c,v retrieving revision 1.23 retrieving revision 1.24 diff -u -p -r1.23 -r1.24 --- texi2mdoc/main.c 2015/02/19 20:55:56 1.23 +++ texi2mdoc/main.c 2015/02/20 09:58:50 1.24 @@ -1,4 +1,4 @@ -/* $Id: main.c,v 1.23 2015/02/19 20:55:56 kristaps Exp $ */ +/* $Id: main.c,v 1.24 2015/02/20 09:58:50 kristaps Exp $ */ /* * Copyright (c) 2015 Kristaps Dzonsons * @@ -30,183 +30,7 @@ #include #include -/* - * This defines each one of the Texinfo commands that we understand. - * Obviously this only refers to native commands; overriden names are a - * different story. - */ -enum texicmd { - TEXICMD_ACRONYM, - TEXICMD_ACUTE, - TEXICMD_A4PAPER, - TEXICMD_ANCHOR, - TEXICMD_APPENDIX, - TEXICMD_APPENDIXSEC, - TEXICMD_APPENDIXSUBSEC, - TEXICMD_APPENDIXSUBSUBSEC, - TEXICMD_ASTERISK, - TEXICMD_AT, - TEXICMD_AUTHOR, - TEXICMD_B, - TEXICMD_BANG, - TEXICMD_BULLET, - TEXICMD_BYE, - TEXICMD_CENTER, - TEXICMD_CHAPTER, - TEXICMD_CINDEX, - TEXICMD_CIRCUMFLEX, - TEXICMD_CITE, - TEXICMD_CODE, - TEXICMD_COLON, - TEXICMD_COLUMNFRACTIONS, - TEXICMD_COMMAND, - TEXICMD_COMMENT, - TEXICMD_COMMENT_LONG, - TEXICMD_CONTENTS, - TEXICMD_COPYING, - TEXICMD_COPYRIGHT, - TEXICMD_DEFFN, - TEXICMD_DEFFNX, - TEXICMD_DEFMAC, - TEXICMD_DEFMACX, - TEXICMD_DEFTP, - TEXICMD_DEFTPX, - TEXICMD_DEFTYPEFN, - TEXICMD_DEFTYPEFNX, - TEXICMD_DEFTYPEFUN, - TEXICMD_DEFTYPEFUNX, - TEXICMD_DEFTYPEVAR, - TEXICMD_DEFTYPEVARX, - TEXICMD_DEFTYPEVR, - TEXICMD_DEFTYPEVRX, - TEXICMD_DEFUN, - TEXICMD_DEFUNX, - TEXICMD_DEFVAR, - TEXICMD_DEFVARX, - TEXICMD_DEFVR, - TEXICMD_DEFVRX, - TEXICMD_DETAILMENU, - TEXICMD_DFN, - TEXICMD_DIRCATEGORY, - TEXICMD_DIRENTRY, - TEXICMD_DISPLAY, - TEXICMD_DOTS, - TEXICMD_EMAIL, - TEXICMD_EMPH, - TEXICMD_END, - TEXICMD_ENUMERATE, - TEXICMD_ENV, - TEXICMD_ERROR, - TEXICMD_EXAMPLE, - TEXICMD_EXPANSION, - TEXICMD_FILE, - TEXICMD_FINALOUT, - TEXICMD_FINDEX, - TEXICMD_FTABLE, - TEXICMD_FORMAT, - TEXICMD_GRAVE, - TEXICMD_GROUP, - TEXICMD_HEADING, - TEXICMD_HEADINGS, - TEXICMD_HEADITEM, - TEXICMD_HYPHEN, - TEXICMD_I, - TEXICMD_IFCLEAR, - TEXICMD_IFDOCBOOK, - TEXICMD_IFHTML, - TEXICMD_IFINFO, - TEXICMD_IFNOTDOCBOOK, - TEXICMD_IFNOTHTML, - TEXICMD_IFNOTINFO, - TEXICMD_IFNOTPLAINTEXT, - TEXICMD_IFNOTTEX, - TEXICMD_IFNOTXML, - TEXICMD_IFPLAINTEXT, - TEXICMD_IFTEX, - TEXICMD_IFSET, - TEXICMD_IFXML, - TEXICMD_IGNORE, - TEXICMD_IMAGE, - TEXICMD_INCLUDE, - TEXICMD_INDENTBLOCK, - TEXICMD_INSERTCOPYING, - TEXICMD_ITEM, - TEXICMD_ITEMIZE, - TEXICMD_ITEMX, - TEXICMD_KBD, - TEXICMD_KEY, - TEXICMD_KINDEX, - TEXICMD_LATEX, - TEXICMD_LOWERSECTIONS, - TEXICMD_MATH, - TEXICMD_MENU, - TEXICMD_MULTITABLE, - TEXICMD_NEED, - TEXICMD_NEWLINE, - TEXICMD_NODE, - TEXICMD_NOINDENT, - TEXICMD_OPTION, - TEXICMD_PXREF, - TEXICMD_QUESTIONMARK, - TEXICMD_QUOTATION, - TEXICMD_PAGE, - TEXICMD_PARINDENT, - TEXICMD_PRINTINDEX, - TEXICMD_R, - TEXICMD_RAISESECTIONS, - TEXICMD_REF, - TEXICMD_RESULT, - TEXICMD_SAMP, - TEXICMD_SANSSERIF, - TEXICMD_SC, - TEXICMD_SECTION, - TEXICMD_SET, - TEXICMD_SETCHAPNEWPAGE, - TEXICMD_SETFILENAME, - TEXICMD_SETTITLE, - TEXICMD_SLANTED, - TEXICMD_SP, - TEXICMD_SPACE, - TEXICMD_SMALLBOOK, - TEXICMD_SMALLDISPLAY, - TEXICMD_SMALLEXAMPLE, - TEXICMD_SMALLFORMAT, - TEXICMD_SMALLINDENTBLOCK, - TEXICMD_SQUIGGLE_LEFT, - TEXICMD_SQUIGGLE_RIGHT, - TEXICMD_STRONG, - TEXICMD_SUBHEADING, - TEXICMD_SUBSECTION, - TEXICMD_SUBSUBSECTION, - TEXICMD_SUBTITLE, - TEXICMD_SYNCODEINDEX, - TEXICMD_T, - TEXICMD_TAB, - TEXICMD_TABSYM, - TEXICMD_TABLE, - TEXICMD_TEX, - TEXICMD_TEXSYM, - TEXICMD_TILDE, - TEXICMD_TITLE, - TEXICMD_TITLEFONT, - TEXICMD_TITLEPAGE, - TEXICMD_TOP, - TEXICMD_UMLAUT, - TEXICMD_UNNUMBERED, - TEXICMD_UNNUMBEREDSEC, - TEXICMD_UNNUMBEREDSUBSEC, - TEXICMD_UNNUMBEREDSUBSUBSEC, - TEXICMD_UREF, - TEXICMD_URL, - TEXICMD_VAR, - TEXICMD_VERBATIMINCLUDE, - TEXICMD_VINDEX, - TEXICMD_VSKIP, - TEXICMD_VTABLE, - TEXICMD_W, - TEXICMD_XREF, - TEXICMD__MAX -}; +#include "extern.h" #define SECTSZ 4 static const char *const sects[SECTSZ] = { @@ -216,73 +40,6 @@ static const char *const sects[SECTSZ] = { "No", }; -/* - * The file currently being parsed. - * This keeps track of our location within that file. - */ -struct texifile { - const char *name; /* name of the file */ - size_t line; /* current line (from zero) */ - size_t col; /* current column in line (from zero) */ - char *map; /* mmap'd file */ - size_t mapsz; /* size of mmap */ -}; - -struct texi; - -/* - * Callback for functions implementing texi commands. - */ -typedef void (*texicmdfp)(struct texi *, - enum texicmd, const char *, size_t, size_t *); - -/* - * Describes Texinfo commands, whether native or overriden. - */ -struct texitok { - texicmdfp fp; /* callback (or NULL if none) */ - const char *tok; /* name of the token */ - size_t len; /* strlen(tok) */ -}; - -enum texilist { - TEXILIST_NONE = 0, - TEXILIST_ITEM, - TEXILIST_NOITEM, - TEXILIST_TABLE -}; - -/* - * The main parse structure. - * This keeps any necessary information handy. - */ -struct texi { - struct texifile files[64]; /* stack of open files */ - size_t filepos; /* number of open files */ - size_t outcol; /* column in output line */ - char **dirs; /* texi directories */ - size_t dirsz; /* number of texi directories */ - char *title; /* title of document */ - char *subtitle; /* subtitle of document */ - int secoffs; /* see sectioner() */ - /* - * The following control what we output to the screen. - * The complexity is required to accomodate for mdoc(7). - */ - enum texilist list; /* current list (set recursively) */ - int outmacro; /* if >0, output is in line macro */ - int seenws; /* ws has been seen (and ignored) */ - int seenvs; /* newline has been Pp'd */ - int ign; /* if >0, don't print anything */ - int literal; /* if >0, literal context */ -}; - -/* Texi disregards spaces and tabs. */ -#define isws(_x) \ - (' ' == (_x) || '\t' == (_x)) -#define ismspace(_x) \ - (isws((_x)) || '\n' == (_x)) - static void doaccent(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doblock(struct texi *, enum texicmd, const char *, size_t, size_t *); static void dobracket(struct texi *, enum texicmd, const char *, size_t, size_t *); @@ -313,9 +70,10 @@ static void dosubsubsection(struct texi *, enum texicm static void dosymbol(struct texi *, enum texicmd, const char *, size_t, size_t *); static void dotab(struct texi *, enum texicmd, const char *, size_t, size_t *); static void dotitle(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void dovalue(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doverbinclude(struct texi *, enum texicmd, const char *, size_t, size_t *); -static const struct texitok texitoks[TEXICMD__MAX] = { +static const struct texitok __texitoks[TEXICMD__MAX] = { /* TEXICMD__BEGIN */ { doignargn, "acronym", 7 }, /* TEXICMD_ACRONYM */ { doaccent, "'", 1 }, /* TEXICMD_ACUTE */ @@ -336,8 +94,9 @@ static const struct texitok texitoks[TEXICMD__MAX] = { { dosection, "chapter", 7 }, /* TEXICMD_CHAPTER */ { doignline, "cindex", 6 }, /* TEXICMD_CINDEX */ { doaccent, "^", 1 }, /* TEXICMD_CIRCUMFLEX */ - { doinline, "code", 4 }, /* TEXICMD_CODE */ { doinline, "cite", 4 }, /* TEXICMD_CITE */ + { dovalue, "clear", 5 }, /* TEXICMD_CLEAR */ + { doinline, "code", 4 }, /* TEXICMD_CODE */ { dosymbol, ":", 1 }, /* TEXICMD_COLON */ { NULL, "columnfractions", 15 }, /* TEXICMD_COLUMNFRACTIONS */ { doinline, "command", 7 }, /* TEXICMD_COMMAND */ @@ -392,7 +151,7 @@ static const struct texitok texitoks[TEXICMD__MAX] = { { doitem, "headitem", 8 }, /* TEXICMD_HEADITEM */ { dosymbol, "-", 1 }, /* TEXICMD_HYPHEN */ { doinline, "i", 1 }, /* TEXICMD_I */ - { doignblock, "ifclear", 7 }, /* TEXICMD_IFCLEAR */ + { dovalue, "ifclear", 7 }, /* TEXICMD_IFCLEAR */ { doignblock, "ifdocbook", 9 }, /* TEXICMD_IFDOCBOOK */ { doignblock, "ifhtml", 6 }, /* TEXICMD_IFHTML */ { doignblock, "ifinfo", 6 }, /* TEXICMD_IFINFO */ @@ -441,7 +200,7 @@ static const struct texitok texitoks[TEXICMD__MAX] = { { doinline, "sansserif", 9 }, /* TEXICMD_SANSSERIF */ { dobracket, "sc", 2 }, /* TEXICMD_SC */ { dosection, "section", 7 }, /* TEXICMD_SECTION */ - { doignline, "set", 3 }, /* TEXICMD_SET */ + { dovalue, "set", 3 }, /* TEXICMD_SET */ { doignline, "setchapternewpage", 17 }, /* TEXICMD_SETCHAPNEWPAGE */ { doignline, "setfilename", 11 }, /* TEXICMD_SETFILENAME */ { dotitle, "settitle", 8 }, /* TEXICMD_SETTITLE */ @@ -479,6 +238,7 @@ static const struct texitok texitoks[TEXICMD__MAX] = { { dosubsubsection, "unnumberedsubsubsec", 19 }, /* TEXICMD_UNNUMBEREDSUBSUBSEC */ { dolink, "uref", 4 }, /* TEXICMD_UREF */ { dolink, "url", 3 }, /* TEXICMD_URL */ + { dovalue, "value", 5 }, /* TEXICMD_VALUE */ { doinline, "var", 3 }, /* TEXICMD_VAR */ { doverbinclude, "verbatiminclude", 15 }, /* TEXICMD_VERBATIMINCLUDE */ { doignline, "vindex", 6 }, /* TEXICMD_VINDEX */ @@ -489,742 +249,35 @@ static const struct texitok texitoks[TEXICMD__MAX] = { /* TEXICMD__END */ }; -/* - * Unmap the top-most file that we're using. - */ -static void -texifilepop(struct texi *p) -{ - struct texifile *f; +const struct texitok *const texitoks = __texitoks; - assert(p->filepos > 0); - f = &p->files[--p->filepos]; - munmap(f->map, f->mapsz); -} - -/* - * Unmap all files that we're currently using and free all resources - * that we've allocated during the parse. - * The utility should exit(...) after this is called. - */ -static void -texiexit(struct texi *p) +static char * +lineargdup(struct texi *p, const char *buf, size_t sz, size_t *pos) { - size_t i; - - if (p->outcol) - putchar('\n'); - - while (p->filepos > 0) - texifilepop(p); - - for (i = 0; i < p->dirsz; i++) - free(p->dirs[i]); - - free(p->dirs); - free(p->subtitle); - free(p->title); -} - -/* - * Fatal error: unmap all files and exit. - * The "errstring" is passed to perror(3). - */ -static void -texiabort(struct texi *p, const char *errstring) -{ - - perror(errstring); - texiexit(p); - exit(EXIT_FAILURE); -} - -/* - * Print a generic warning message (to stderr) tied to our current - * location in the parse sequence. - */ -static void -texiwarn(const struct texi *p, const char *fmt, ...) -{ - va_list ap; - - fprintf(stderr, "%s:%zu:%zu: warning: ", - p->files[p->filepos - 1].name, - p->files[p->filepos - 1].line + 1, - p->files[p->filepos - 1].col + 1); - va_start(ap, fmt); - vfprintf(stderr, fmt, ap); - va_end(ap); - fputc('\n', stderr); -} - -/* - * Print an error message (to stderr) tied to our current location in - * the parse sequence, invoke texiexit(), then die. - */ -static void -texierr(struct texi *p, const char *fmt, ...) -{ - va_list ap; - - fprintf(stderr, "%s:%zu:%zu: error: ", - p->files[p->filepos - 1].name, - p->files[p->filepos - 1].line + 1, - p->files[p->filepos - 1].col + 1); - va_start(ap, fmt); - vfprintf(stderr, fmt, ap); - va_end(ap); - fputc('\n', stderr); - texiexit(p); - exit(EXIT_FAILURE); -} - -/* - * Put a single data character to the output if we're not ignoring. - * Adjusts our output status. - * This shouldn't be called for macros: just for ordinary text. - */ -static void -texiputchar(struct texi *p, char c) -{ - - if (p->ign) - return; - - if ('.' == c && 0 == p->outcol) - fputs("\\&", stdout); - - putchar(c); - p->seenvs = 0; - if ('\n' == c) { - p->outcol = 0; - p->seenws = 0; - } else - p->outcol++; -} - -/* - * Put multiple characters (see texiputchar()). - * This shouldn't be called for macros: just for ordinary text. - */ -static void -texiputchars(struct texi *p, const char *s) -{ - - while ('\0' != *s) - texiputchar(p, *s++); -} - -/* - * Close an mdoc(7) macro opened with teximacroopen(). - * If there are no more macros on the line, prints a newline. - */ -static void -teximacroclose(struct texi *p) -{ - - if (p->ign) - return; - - if (0 == --p->outmacro) { - putchar('\n'); - p->outcol = p->seenws = 0; - } -} - -/* - * Open a mdoc(7) macro. - * This is used for line macros, e.g., Qq [foo bar baz]. - * It can be invoked for nested macros, e.g., Qq Li foo . - */ -static void -teximacroopen(struct texi *p, const char *s) -{ - int rc; - - if (p->ign) - return; - - if (p->outcol && 0 == p->outmacro) { - putchar('\n'); - p->outcol = 0; - } - - if (0 == p->outmacro) - putchar('.'); - else - putchar(' '); - - if (EOF != (rc = fputs(s, stdout))) - p->outcol += rc; - - putchar(' '); - p->outcol++; - p->outmacro++; - p->seenws = 0; -} - -/* - * Put a stadnalone mdoc(7) command with the trailing newline. - */ -static void -teximacro(struct texi *p, const char *s) -{ - - if (p->ign) - return; - - if (p->outmacro) - texierr(p, "\"%s\" in open line scope!?", s); - if (p->literal) - texierr(p, "\"%s\" in a literal scope!?", s); - - if (p->outcol) - putchar('\n'); - - putchar('.'); - puts(s); - p->outcol = p->seenws = 0; -} - -static void -texivspace(struct texi *p) -{ - - if (p->seenvs) - return; - teximacro(p, "Pp"); - p->seenvs = 1; -} - -/* - * Advance by a single byte in the input stream. - */ -static void -advance(struct texi *p, const char *buf, size_t *pos) -{ - - if ('\n' == buf[*pos]) { - p->files[p->filepos - 1].line++; - p->files[p->filepos - 1].col = 0; - } else - p->files[p->filepos - 1].col++; - - (*pos)++; -} - -static void -texipunctuate(struct texi *p, const char *buf, size_t sz, size_t *pos) -{ + char *key; size_t start, end; - if (1 != p->outmacro) - return; - - for (start = end = *pos; end < sz; end++) { - switch (buf[end]) { - case (','): - case (')'): - case ('.'): - case ('"'): - case (':'): - case ('!'): - case ('?'): - continue; - default: - break; - } - break; - } - if (end == *pos) - return; - if (end + 1 == sz || ' ' == buf[end] || '\n' == buf[end]) { - for ( ; start < end; start++) { - texiputchar(p, ' '); - texiputchar(p, buf[start]); - advance(p, buf, pos); - } - } -} - -/* - * Advance to the next non-whitespace word in the input stream. - * If we're in literal mode, then print all of the whitespace as we're - * doing so. - */ -static size_t -advancenext(struct texi *p, const char *buf, size_t sz, size_t *pos) -{ - - if (p->literal) { - while (*pos < sz && ismspace(buf[*pos])) { - if (*pos && '\n' == buf[*pos] && - '\\' == buf[*pos - 1]) - texiputchar(p, 'e'); - texiputchar(p, buf[*pos]); - advance(p, buf, pos); - } - return(*pos); - } - - while (*pos < sz && ismspace(buf[*pos])) { - p->seenws = 1; - /* - * If it looks like we've printed a double-line, then - * output a paragraph. - * FIXME: this is stupid. - */ - if (*pos && '\n' == buf[*pos] && '\n' == buf[*pos - 1]) - texivspace(p); + while (*pos < sz && isws(buf[*pos])) advance(p, buf, pos); - } - return(*pos); -} - -/* - * Advance to the EOLN in the input stream. - */ -static size_t -advanceeoln(struct texi *p, const char *buf, - size_t sz, size_t *pos, int consumenl) -{ - - /* FIXME: disregards @NEWLINE. */ - while (*pos < sz && '\n' != buf[*pos]) - advance(p, buf, pos); - if (*pos < sz && consumenl) - advance(p, buf, pos); - return(*pos); -} - -/* - * Advance to position "end", which is an absolute position in the - * current buffer greater than or equal to the current position. - */ -static void -advanceto(struct texi *p, const char *buf, size_t *pos, size_t end) -{ - - assert(*pos <= end); - while (*pos < end) - advance(p, buf, pos); -} - -/* - * Output a free-form word in the input stream, progressing to the next - * command or white-space. - * This also will advance the input stream. - */ -static void -texiword(struct texi *p, const char *buf, - size_t sz, size_t *pos, char extra) -{ - - if (p->seenws && 0 == p->outmacro && p->outcol > 72 && 0 == p->literal) - texiputchar(p, '\n'); - /* FIXME: abstract this: we use it elsewhere. */ - if (p->seenws && p->outcol && 0 == p->literal) - texiputchar(p, ' '); - - p->seenws = 0; - - while (*pos < sz && ! ismspace(buf[*pos])) { - switch (buf[*pos]) { - case ('@'): - case ('}'): - case ('{'): - return; - } - if ('\0' != extra && buf[*pos] == extra) - return; - if (*pos < sz - 1 && - '`' == buf[*pos] && - '`' == buf[*pos + 1]) { - texiputchars(p, "\\(lq"); - advance(p, buf, pos); - } else if (*pos < sz - 1 && - '\'' == buf[*pos] && - '\'' == buf[*pos + 1]) { - texiputchars(p, "\\(rq"); - advance(p, buf, pos); - } else - texiputchar(p, buf[*pos]); - advance(p, buf, pos); - } -} - -/* - * Look up the command at position "pos" in the buffer, returning it (or - * TEXICMD__MAX if none found) and setting "end" to be the absolute - * index after the command name. - */ -static enum texicmd -texicmd(struct texi *p, const char *buf, - size_t pos, size_t sz, size_t *end) -{ - size_t i, len; - - assert('@' == buf[pos]); - - if ((*end = pos) == sz) - return(TEXICMD__MAX); - else if ((*end = ++pos) == sz) - return(TEXICMD__MAX); - - /* Alphabetic commands are special. */ - if ( ! isalpha(buf[pos])) { - if ((*end = pos + 1) == sz) - return(TEXICMD__MAX); - for (i = 0; i < TEXICMD__MAX; i++) { - if (1 != texitoks[i].len) - continue; - if (0 == strncmp(texitoks[i].tok, &buf[pos], 1)) - return(i); - } - texiwarn(p, "bad command: @%c", buf[pos]); - return(TEXICMD__MAX); - } - - for (*end = pos; *end < sz && ! ismspace(buf[*end]); (*end)++) - if ((*end > pos && ('@' == buf[*end] || - '{' == buf[*end] || '}' == buf[*end]))) + if (*pos == sz) + return(NULL); + for (start = end = *pos; end < sz; end++) + if ('\n' == buf[end]) break; - - len = *end - pos; - for (i = 0; i < TEXICMD__MAX; i++) { - if (len != texitoks[i].len) - continue; - if (0 == strncmp(texitoks[i].tok, &buf[pos], len)) - return(i); - } - - texiwarn(p, "bad command: @%.*s", (int)len, &buf[pos]); - return(TEXICMD__MAX); -} - -/* - * Parse an argument from a bracketed command, e.g., @url{foo, baz}. - * Num should be set to the argument we're currently parsing, although - * it suffixes for it to be zero or non-zero. - * This will return 1 if there are more arguments, 0 otherwise. - * This will stop (returning 0) in the event of EOF or if we're not at a - * bracket for the zeroth parse. - */ -static int -parsearg(struct texi *p, const char *buf, - size_t sz, size_t *pos, size_t num) -{ - size_t end; - enum texicmd cmd; - - while (*pos < sz && ismspace(buf[*pos])) - advance(p, buf, pos); - if (*pos == sz || (0 == num && '{' != buf[*pos])) - return(0); - if (0 == num) - advance(p, buf, pos); - - while ((*pos = advancenext(p, buf, sz, pos)) < sz) { - switch (buf[*pos]) { - case (','): - advance(p, buf, pos); - return(1); - case ('}'): - advance(p, buf, pos); - return(0); - case ('{'): - if (0 == p->ign) - texiwarn(p, "unexpected \"{\""); - advance(p, buf, pos); - continue; - case ('@'): - break; - default: - texiword(p, buf, sz, pos, ','); - continue; - } - - cmd = texicmd(p, buf, *pos, sz, &end); - advanceto(p, buf, pos, end); - if (TEXICMD__MAX == cmd) - continue; - if (NULL != texitoks[cmd].fp) - (*texitoks[cmd].fp)(p, cmd, buf, sz, pos); - } - return(0); -} - -/* - * Parse until the end of a bracketed statement, e.g., @foo{bar baz}. - * This will stop in the event of EOF or if we're not at a bracket. - */ -static void -parsebracket(struct texi *p, const char *buf, size_t sz, size_t *pos) -{ - size_t end; - enum texicmd cmd; - - while (*pos < sz && ismspace(buf[*pos])) - advance(p, buf, pos); - - if (*pos == sz || '{' != buf[*pos]) - return; - advance(p, buf, pos); - - while ((*pos = advancenext(p, buf, sz, pos)) < sz) { - switch (buf[*pos]) { - case ('}'): - advance(p, buf, pos); - return; - case ('{'): - if (0 == p->ign) - texiwarn(p, "unexpected \"{\""); - advance(p, buf, pos); - continue; - case ('@'): - break; - default: - texiword(p, buf, sz, pos, '\0'); - continue; - } - - cmd = texicmd(p, buf, *pos, sz, &end); - advanceto(p, buf, pos, end); - if (TEXICMD__MAX == cmd) - continue; - if (NULL != texitoks[cmd].fp) - (*texitoks[cmd].fp)(p, cmd, buf, sz, pos); - } -} - -/* - * This should be invoked when we're on a macro line and want to process - * to the end of the current input line, doing all of our macros along - * the way. - */ -static void -parseeoln(struct texi *p, const char *buf, size_t sz, size_t *pos) -{ - size_t end; - enum texicmd cmd; - - while (*pos < sz && '\n' != buf[*pos]) { - while (*pos < sz && isws(buf[*pos])) { - p->seenws = 1; - if (p->literal) - texiputchar(p, buf[*pos]); - advance(p, buf, pos); - } - switch (buf[*pos]) { - case ('}'): - if (0 == p->ign) - texiwarn(p, "unexpected \"}\""); - advance(p, buf, pos); - continue; - case ('{'): - if (0 == p->ign) - texiwarn(p, "unexpected \"{\""); - advance(p, buf, pos); - continue; - case ('@'): - break; - default: - texiword(p, buf, sz, pos, '\0'); - continue; - } - - cmd = texicmd(p, buf, *pos, sz, &end); - advanceto(p, buf, pos, end); - if (TEXICMD__MAX == cmd) - continue; - if (NULL != texitoks[cmd].fp) - (*texitoks[cmd].fp)(p, cmd, buf, sz, pos); - } -} - -/* - * Parse a single word or command. - * This will return immediately at the EOF. - */ -static void -parsesingle(struct texi *p, const char *buf, size_t sz, size_t *pos) -{ - size_t end; - enum texicmd cmd; - - if ((*pos = advancenext(p, buf, sz, pos)) >= sz) - return; - - switch (buf[*pos]) { - case ('}'): - if (0 == p->ign) - texiwarn(p, "unexpected \"}\""); - advance(p, buf, pos); - return; - case ('{'): - if (0 == p->ign) - texiwarn(p, "unexpected \"{\""); - advance(p, buf, pos); - return; - case ('@'): - break; - default: - texiword(p, buf, sz, pos, '\0'); - return; - } - - cmd = texicmd(p, buf, *pos, sz, &end); advanceto(p, buf, pos, end); - if (TEXICMD__MAX == cmd) - return; - if (NULL != texitoks[cmd].fp) - (*texitoks[cmd].fp)(p, cmd, buf, sz, pos); -} - -static int -parselinearg(struct texi *p, const char *buf, size_t sz, size_t *pos) -{ - - while (*pos < sz && isws(buf[*pos])) { - p->seenws = 1; + if (*pos < sz) advance(p, buf, pos); + key = malloc(end - start + 1); + if (NULL == key) { + perror(NULL); + exit(EXIT_FAILURE); } - - if (*pos < sz && '{' == buf[*pos]) - parsebracket(p, buf, sz, pos); - else if ('\n' != buf[*pos]) - parsesingle(p, buf, sz, pos); - else - return(0); - - return(1); + memcpy(key, &buf[start], end - start); + key[end - start] = '\0'; + return(key); } -/* - * Parse til the end of the buffer. - */ static void -parseeof(struct texi *p, const char *buf, size_t sz) -{ - size_t pos; - - for (pos = 0; pos < sz; ) - parsesingle(p, buf, sz, &pos); -} - -/* - * Parse a block sequence until we have the "@end endtoken" command - * invocation. - * This will return immediately at EOF. - */ -static void -parseto(struct texi *p, const char *buf, - size_t sz, size_t *pos, const char *endtoken) -{ - size_t end; - enum texicmd cmd; - size_t endtoksz; - - endtoksz = strlen(endtoken); - assert(endtoksz > 0); - - while ((*pos = advancenext(p, buf, sz, pos)) < sz) { - switch (buf[*pos]) { - case ('}'): - if (0 == p->ign) - texiwarn(p, "unexpected \"}\""); - advance(p, buf, pos); - continue; - case ('{'): - if (0 == p->ign) - texiwarn(p, "unexpected \"{\""); - advance(p, buf, pos); - continue; - case ('@'): - break; - default: - texiword(p, buf, sz, pos, '\0'); - continue; - } - - cmd = texicmd(p, buf, *pos, sz, &end); - advanceto(p, buf, pos, end); - if (TEXICMD_END == cmd) { - while (*pos < sz && isws(buf[*pos])) - advance(p, buf, pos); - /* - * FIXME: check the full word, not just its - * initial substring! - */ - if (sz - *pos >= endtoksz && 0 == strncmp - (&buf[*pos], endtoken, endtoksz)) { - advanceeoln(p, buf, sz, pos, 0); - break; - } - if (0 == p->ign) - texiwarn(p, "unexpected \"end\""); - advanceeoln(p, buf, sz, pos, 0); - continue; - } else if (TEXICMD__MAX != cmd) - if (NULL != texitoks[cmd].fp) - (*texitoks[cmd].fp)(p, cmd, buf, sz, pos); - } -} - -/* - * Memory-map the file "fname" and begin parsing it. - * This can be called in a nested context. - */ -static void -parsefile(struct texi *p, const char *fname, int parse) -{ - struct texifile *f; - int fd; - struct stat st; - size_t i; - - assert(p->filepos < 64); - f = &p->files[p->filepos]; - memset(f, 0, sizeof(struct texifile)); - - f->name = fname; - if (-1 == (fd = open(fname, O_RDONLY, 0))) { - texiabort(p, fname); - } else if (-1 == fstat(fd, &st)) { - close(fd); - texiabort(p, fname); - } - - f->mapsz = st.st_size; - f->map = mmap(NULL, f->mapsz, - PROT_READ, MAP_SHARED, fd, 0); - close(fd); - - if (MAP_FAILED == f->map) - texiabort(p, fname); - - p->filepos++; - if ( ! parse) { - /* - * We're printing verbatim output. - * Make sure it doesn't get interpreted as mdoc by - * escaping escapes and making sure leading dots don't - * trigger mdoc(7) expansion. - */ - for (i = 0; i < f->mapsz; i++) { - if (i > 0 && '.' == f->map[i]) - if ('\n' == f->map[i - 1]) - fputs("\\&", stdout); - putchar(f->map[i]); - if ('\\' == f->map[i]) - putchar('e'); - } - } else - parseeof(p, f->map, f->mapsz); - texifilepop(p); -} - -static void dodefn(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { @@ -1771,6 +824,147 @@ domath(struct texi *p, enum texicmd cmd, return; assert('}' == buf[*pos]); advance(p, buf, pos); +} + +static void +dovalue(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) +{ + size_t i, start, end; + int clr; + char *key, *val; + + if (TEXICMD_SET == cmd) { + /* + * If we're going to set a value, then first parse the + * key and value from the input stream. + * Disallow empty keys. + */ + while (*pos < sz && isws(buf[*pos])) + advance(p, buf, pos); + for (start = end = *pos; end < sz; end++) + if (ismspace(buf[end])) + break; + if (start == end) + return; + + key = malloc(end - start + 1); + if (NULL == key) { + perror(NULL); + exit(EXIT_FAILURE); + } + memcpy(key, &buf[start], end - start); + key[end - start] = '\0'; + + advanceto(p, buf, pos, end); + + while (*pos < sz && isws(buf[*pos])) + advance(p, buf, pos); + for (start = end = *pos; end < sz; end++) + if ('\n' == buf[end]) + break; + + val = malloc(end - start + 1); + if (NULL == val) { + perror(NULL); + exit(EXIT_FAILURE); + } + memcpy(val, &buf[start], end - start); + val[end - start] = '\0'; + + /* + * Now look up the key in our table of existing keys. + * If we find it, then free the local key and replace + * the value (freeing the existing) of the existing. + * If not, reallocate the structure. + */ + for (i = 0; i < p->valsz; i++) + if (0 == strcmp(p->vals[i].key, key)) + break; + + if (i < p->valsz) { + free(key); + free(p->vals[i].value); + p->vals[i].value = val; + } else { + p->vals = realloc(p->vals, + (p->valsz + 1) * + sizeof(struct texivalue)); + if (NULL == p->vals) { + perror(NULL); + exit(EXIT_FAILURE); + } + p->vals[p->valsz].key = key; + p->vals[p->valsz].value = val; + p->valsz++; + } + + advanceeoln(p, buf, sz, pos, 1); + } else if (TEXICMD_VALUE == cmd) { + /* + * First, parse out the key from the stream, which is + * just the material within the quotes. + */ + while (*pos < sz && '{' != buf[*pos]) + advance(p, buf, pos); + if (*pos == sz) + return; + advance(p, buf, pos); + start = *pos; + while (*pos < sz && '}' != buf[*pos]) + advance(p, buf, pos); + if ((end = *pos) < sz) + advance(p, buf, pos); + + key = malloc(end - start + 1); + if (NULL == key) { + perror(NULL); + exit(EXIT_FAILURE); + } + memcpy(key, &buf[start], end - start); + key[end - start] = '\0'; + + /* + * Next, look up the key in our table. + * If it exists, print it out, being mindful all the + * while of prior whitespace. + * If it doesn't, print out a message. + */ + for (i = 0; i < p->valsz; i++) + if (0 == strcmp(p->vals[i].key, key)) + break; + if (p->seenws) + texiputchar(p, ' '); + p->seenws = 0; + if (i == p->valsz || NULL == p->vals[i].value) { + texiputchars(p, "{No value for \\(lq"); + texiputchars(p, key); + texiputchars(p, "\\(rq}"); + } else + texiputchars(p, p->vals[i].value); + free(key); + } else if (TEXICMD_CLEAR == cmd) { + key = lineargdup(p, buf, sz, pos); + for (i = 0; i < p->valsz; i++) + if (0 == strcmp(p->vals[i].key, key)) { + free(p->vals[i].value); + p->vals[i].value = NULL; + break; + } + free(key); + } else if (TEXICMD_IFCLEAR == cmd) { + key = lineargdup(p, buf, sz, pos); + for (i = 0; i < p->valsz; i++) + if (0 == strcmp(p->vals[i].key, key)) + break; + free(key); + clr = i < p->valsz && NULL != p->vals[i].value; + if (clr) + p->ign++; + parseto(p, buf, sz, pos, texitoks[cmd].tok); + if (clr) + p->ign--; + } } static void