=================================================================== RCS file: /cvs/texi2mdoc/main.c,v retrieving revision 1.1 retrieving revision 1.12 diff -u -p -r1.1 -r1.12 --- texi2mdoc/main.c 2015/02/16 22:24:43 1.1 +++ texi2mdoc/main.c 2015/02/18 23:08:16 1.12 @@ -1,4 +1,4 @@ -/* $Id: main.c,v 1.1 2015/02/16 22:24:43 kristaps Exp $ */ +/* $Id: main.c,v 1.12 2015/02/18 23:08:16 kristaps Exp $ */ /* * Copyright (c) 2015 Kristaps Dzonsons * @@ -21,10 +21,14 @@ #include #include #include +#include +#include #include #include #include #include +#include +#include /* * This defines each one of the Texinfo commands that we understand. @@ -32,55 +36,115 @@ * different story. */ enum texicmd { + TEXICMD_ACRONYM, TEXICMD_A4PAPER, TEXICMD_ANCHOR, + TEXICMD_APPENDIX, + TEXICMD_APPENDIXSEC, + TEXICMD_ASTERISK, TEXICMD_AT, + TEXICMD_AUTHOR, + TEXICMD_B, + TEXICMD_BANG, + TEXICMD_BULLET, TEXICMD_BYE, + TEXICMD_CENTER, TEXICMD_CHAPTER, TEXICMD_CINDEX, + TEXICMD_CITE, TEXICMD_CODE, + TEXICMD_COLON, TEXICMD_COMMAND, TEXICMD_COMMENT, + TEXICMD_COMMENT_LONG, TEXICMD_CONTENTS, TEXICMD_COPYING, TEXICMD_COPYRIGHT, + TEXICMD_DEFTYPEFN, + TEXICMD_DEFTYPEFNX, + TEXICMD_DEFTYPEFUN, + TEXICMD_DEFTYPEFUNX, + TEXICMD_DEFTYPEVAR, + TEXICMD_DEFTYPEVR, TEXICMD_DETAILMENU, + TEXICMD_DFN, TEXICMD_DIRCATEGORY, TEXICMD_DIRENTRY, + TEXICMD_DISPLAY, + TEXICMD_DOTS, TEXICMD_EMAIL, TEXICMD_EMPH, TEXICMD_END, + TEXICMD_ENUMERATE, + TEXICMD_ENV, TEXICMD_EXAMPLE, TEXICMD_FILE, + TEXICMD_GROUP, + TEXICMD_HEADING, + TEXICMD_HEADINGS, + TEXICMD_HYPHEN, TEXICMD_I, + TEXICMD_IFCLEAR, TEXICMD_IFHTML, + TEXICMD_IFINFO, TEXICMD_IFNOTTEX, TEXICMD_IFTEX, + TEXICMD_IFSET, TEXICMD_IMAGE, + TEXICMD_INCLUDE, + TEXICMD_INSERTCOPYING, TEXICMD_ITEM, TEXICMD_ITEMIZE, TEXICMD_KBD, TEXICMD_LATEX, + TEXICMD_MATH, TEXICMD_MENU, + TEXICMD_NEWLINE, TEXICMD_NODE, + TEXICMD_NOINDENT, + TEXICMD_PXREF, + TEXICMD_QUESTIONMARK, TEXICMD_QUOTATION, + TEXICMD_PAGE, TEXICMD_PARINDENT, + TEXICMD_PRINTINDEX, + TEXICMD_R, TEXICMD_REF, TEXICMD_SAMP, + TEXICMD_SANSSERIF, + TEXICMD_SC, TEXICMD_SECTION, + TEXICMD_SET, TEXICMD_SETCHAPNEWPAGE, TEXICMD_SETFILENAME, TEXICMD_SETTITLE, + TEXICMD_SLANTED, + TEXICMD_SP, + TEXICMD_SPACE, + TEXICMD_SMALLDISPLAY, + TEXICMD_SMALLEXAMPLE, + TEXICMD_SQUIGGLE_LEFT, + TEXICMD_SQUIGGLE_RIGHT, + TEXICMD_STRONG, TEXICMD_SUBSECTION, + TEXICMD_SUBTITLE, + TEXICMD_T, + TEXICMD_TAB, TEXICMD_TABLE, TEXICMD_TEX, TEXICMD_TEXSYM, + TEXICMD_TITLE, TEXICMD_TITLEFONT, TEXICMD_TITLEPAGE, TEXICMD_TOP, TEXICMD_UNNUMBERED, + TEXICMD_UNNUMBEREDSEC, + TEXICMD_UREF, TEXICMD_URL, TEXICMD_VAR, + TEXICMD_VSKIP, + TEXICMD_W, + TEXICMD_XREF, TEXICMD__MAX }; @@ -98,6 +162,9 @@ struct texifile { struct texi; +/* + * Callback for functions implementing texi commands. + */ typedef void (*texicmdfp)(struct texi *, enum texicmd, const char *, size_t, size_t *); @@ -110,102 +177,185 @@ struct texitok { size_t len; /* strlen(tok) */ }; +enum texilist { + TEXILIST_NONE = 0, + TEXILIST_ITEM, + TEXILIST_NOITEM, +}; + /* * The main parse structure. * This keeps any necessary information handy. */ struct texi { - struct texifile files[64]; - size_t filepos; - unsigned flags; -#define TEXI_IGN 0x01 /* don't print anything */ -#define TEXI_HEADER (TEXI_IGN | 0x02) /* haven't seen @top yet */ -#define TEXI_LITERAL 0x04 /* output all whitespace */ - size_t outcol; /* column of output */ - int outmacro; /* whether output is in line macro */ - int seenws; /* whitespace has been ignored */ + 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 */ + /* + * 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 */ }; -#define ismpunct(_x) \ - ('.' == (_x) || \ - ',' == (_x) || \ - ';' == (_x)) +/* Texi disregards spaces and tabs. */ +#define isws(_x) \ + (' ' == (_x) || '\t' == (_x)) +#define ismspace(_x) \ + (isws((_x)) || '\n' == (_x)) -static void doarg1(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 *); static void dobye(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void docommand(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void doemph(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void dodeftypefun(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void dodeftypevar(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void dodisplay(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void doenumerate(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doexample(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void dofile(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void doifnottex(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void dofont(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void doignargn(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doignblock(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doignbracket(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doignline(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void doitalic(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void doinline(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void doinclude(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doitem(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doitemize(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void doliteral(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void dolink(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void domath(struct texi *, enum texicmd, const char *, size_t, size_t *); static void doquotation(struct texi *, enum texicmd, const char *, size_t, size_t *); static void dotable(struct texi *, enum texicmd, const char *, size_t, size_t *); static void dotop(struct texi *, enum texicmd, const char *, size_t, size_t *); static void dosection(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void dosh(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void dosp(struct texi *, enum texicmd, const char *, size_t, size_t *); static void dosubsection(struct texi *, enum texicmd, const char *, size_t, size_t *); static void dosymbol(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void dotitle(struct texi *, enum texicmd, const char *, size_t, size_t *); static const struct texitok texitoks[TEXICMD__MAX] = { + { doignargn, "acronym", 7 }, /* TEXICMD_ACRONYM */ { doignline, "afourpaper", 10 }, /* TEXICMD_A4PAPER */ { doignbracket, "anchor", 6 }, /* TEXICMD_ANCHOR */ + { dosection, "appendix", 8 }, /* TEXICMD_APPENDIX */ + { dosection, "appendixsec", 11 }, /* TEXICMD_APPENDIXSEC */ + { dosymbol, "*", 1 }, /* TEXICMD_ASTERISK */ { dosymbol, "@", 1 }, /* TEXICMD_AT */ + { doignline, "author", 6 }, /* TEXICMD_AUTHOR */ + { dofont, "b", 1 }, /* TEXICMD_BOLD */ + { dosymbol, "!", 1 }, /* TEXICMD_BANG */ + { dosymbol, "bullet", 6 }, /* TEXICMD_BULLET */ { dobye, "bye", 3 }, /* TEXICMD_BYE */ - { dosh, "chapter", 7 }, /* TEXICMD_CHAPTER */ + { doignline, "center", 6 }, /* TEXICMD_CENTER */ + { dosection, "chapter", 7 }, /* TEXICMD_CHAPTER */ { doignline, "cindex", 6 }, /* TEXICMD_CINDEX */ - { doliteral, "code", 4 }, /* TEXICMD_CODE */ - { docommand, "command", 7 }, /* TEXICMD_COMMAND */ + { dofont, "code", 4 }, /* TEXICMD_CODE */ + { dofont, "cite", 4 }, /* TEXICMD_CITE */ + { dosymbol, ":", 1 }, /* TEXICMD_COLON */ + { doinline, "command", 7 }, /* TEXICMD_COMMAND */ { doignline, "c", 1 }, /* TEXICMD_COMMENT */ + { doignline, "comment", 7 }, /* TEXICMD_COMMENT_LONG */ { doignline, "contents", 8 }, /* TEXICMD_CONTENTS */ { doignblock, "copying", 7 }, /* TEXICMD_COPYING */ { dosymbol, "copyright", 9 }, /* TEXICMD_COPYRIGHT */ + { dodeftypefun, "deftypefn", 9 }, /* TEXICMD_DEFTYPEFN */ + { dodeftypefun, "deftypefnx", 10 }, /* TEXICMD_DEFTYPEFNX */ + { dodeftypefun, "deftypefun", 10 }, /* TEXICMD_DEFTYPEFUN */ + { dodeftypefun, "deftypefunx", 11 }, /* TEXICMD_DEFTYPEFUNX */ + { dodeftypevar, "deftypevar", 10 }, /* TEXICMD_DEFTYPEVAR */ + { dodeftypevar, "deftypevr", 9 }, /* TEXICMD_DEFTYPEVR */ { doignblock, "detailmenu", 10 }, /* TEXICMD_DETAILMENU */ + { dofont, "dfn", 3 }, /* TEXICMD_DFN */ { doignline, "dircategory", 11 }, /* TEXICMD_DIRCATEGORY */ { doignblock, "direntry", 8 }, /* TEXICMD_DIRENTRY */ - { doarg1, "email", 5 }, /* TEXICMD_EMAIL */ - { doemph, "emph", 4 }, /* TEXICMD_EMPH */ + { dodisplay, "display", 7 }, /* TEXICMD_DISPLAY */ + { dosymbol, "dots", 4 }, /* TEXICMD_DOTS */ + { dolink, "email", 5 }, /* TEXICMD_EMAIL */ + { dofont, "emph", 4 }, /* TEXICMD_EMPH */ { NULL, "end", 3 }, /* TEXICMD_END */ + { doenumerate, "enumerate", 9 }, /* TEXICMD_ENUMERATE */ + { doinline, "env", 3 }, /* TEXICMD_ENV */ { doexample, "example", 7 }, /* TEXICMD_EXAMPLE */ - { dofile, "file", 4 }, /* TEXICMD_FILE */ - { doitalic, "i", 1 }, /* TEXICMD_I */ + { doinline, "file", 4 }, /* TEXICMD_FILE */ + { doblock, "group", 5 }, /* TEXICMD_GROUP */ + { dosection, "heading", 7 }, /* TEXICMD_HEADING */ + { doignline, "headings", 8 }, /* TEXICMD_HEADINGS */ + { dosymbol, "-", 1 }, /* TEXICMD_HYPHEN */ + { dofont, "i", 1 }, /* TEXICMD_I */ + { doignblock, "ifclear", 7 }, /* TEXICMD_IFCLEAR */ { doignblock, "ifhtml", 6 }, /* TEXICMD_IFHTML */ - { doifnottex, "ifnottex", 8 }, /* TEXICMD_IFNOTTEX */ + { doignblock, "ifinfo", 6 }, /* TEXICMD_IFINFO */ + { doblock, "ifnottex", 8 }, /* TEXICMD_IFNOTTEX */ { doignblock, "iftex", 5 }, /* TEXICMD_IFTEX */ + { doignblock, "ifset", 5 }, /* TEXICMD_IFSET */ { doignbracket, "image", 5 }, /* TEXICMD_IMAGE */ + { doinclude, "include", 7 }, /* TEXICMD_INCLUDE */ + { doignline, "insertcopying", 13 }, /* TEXICMD_INSERTCOPYING */ { doitem, "item", 4 }, /* TEXICMD_ITEM */ { doitemize, "itemize", 7 }, /* TEXICMD_ITEMIZE */ - { doliteral, "kbd", 3 }, /* TEXICMD_KBD */ + { dofont, "kbd", 3 }, /* TEXICMD_KBD */ { dosymbol, "LaTeX", 5 }, /* TEXICMD_LATEX */ + { domath, "math", 4 }, /* TEXICMD_MATH */ { doignblock, "menu", 4 }, /* TEXICMD_MENU */ + { dosymbol, "\n", 1 }, /* TEXICMD_NEWLINE */ { doignline, "node", 4 }, /* TEXICMD_NODE */ + { doignline, "noindent", 8 }, /* TEXICMD_NOINDENT */ + { dolink, "pxref", 5 }, /* TEXICMD_PXREF */ + { dosymbol, "?", 1 }, /* TEXICMD_QUESTIONMARK */ { doquotation, "quotation", 9 }, /* TEXICMD_QUOTATION */ + { doignline, "page", 4 }, /* TEXICMD_PAGE */ { doignline, "paragraphindent", 14 }, /* TEXICMD_PARINDENT */ + { doignline, "printindex", 10 }, /* TEXICMD_PRINTINDEX */ + { dofont, "r", 1 }, /* TEXICMD_R */ { dobracket, "ref", 3 }, /* TEXICMD_REF */ - { doliteral, "samp", 4 }, /* TEXICMD_SAMP */ + { dofont, "samp", 4 }, /* TEXICMD_SAMP */ + { dofont, "sansserif", 9 }, /* TEXICMD_SANSSERIF */ + { dobracket, "sc", 2 }, /* TEXICMD_SC */ { dosection, "section", 7 }, /* TEXICMD_SECTION */ + { doignline, "set", 3 }, /* TEXICMD_SET */ { doignline, "setchapternewpage", 17 }, /* TEXICMD_SETCHAPNEWPAGE */ { doignline, "setfilename", 11 }, /* TEXICMD_SETFILENAME */ - { doignline, "settitle", 8 }, /* TEXICMD_SETTITLE */ + { dotitle, "settitle", 8 }, /* TEXICMD_SETTITLE */ + { dofont, "slanted", 7 }, /* TEXICMD_SLANTED */ + { dosp, "sp", 2 }, /* TEXICMD_SP */ + { dosymbol, " ", 1 }, /* TEXICMD_SPACE */ + { dodisplay, "smalldisplay", 12 }, /* TEXICMD_SMALLDISPLAY */ + { doexample, "smallexample", 12 }, /* TEXICMD_SMALLEXAMPLE */ + { dosymbol, "{", 1 }, /* TEXICMD_SQUIGGLE_LEFT */ + { dosymbol, "}", 1 }, /* TEXICMD_SQUIGGLE_RIGHT */ + { dofont, "strong", 6 }, /* TEXICMD_STRONG */ { dosubsection, "subsection", 10 }, /* TEXICMD_SUBSECTION */ + { doignline, "subtitle", 8 }, /* TEXICMD_SUBTITLE */ + { dofont, "t", 1 }, /* TEXICMD_T */ + { dosymbol, "\t", 1 }, /* TEXICMD_TAB */ { dotable, "table", 5 }, /* TEXICMD_TABLE */ { doignblock, "tex", 3 }, /* TEXICMD_TEX */ { dosymbol, "TeX", 3 }, /* TEXICMD_TEXSYM */ + { doignline, "title", 5 }, /* TEXICMD_TITLE */ { dobracket, "titlefont", 9 }, /* TEXICMD_TITLEFONT */ { doignblock, "titlepage", 9 }, /* TEXICMD_TITLEPAGE */ { dotop, "top", 3 }, /* TEXICMD_TOP */ - { dosh, "unnumbered", 10 }, /* TEXICMD_UNNUMBERED */ - { doarg1, "url", 3 }, /* TEXICMD_URL */ - { doliteral, "var", 3 }, /* TEXICMD_VAR */ + { dosection, "unnumbered", 10 }, /* TEXICMD_UNNUMBERED */ + { dosection, "unnumberedsec", 13 }, /* TEXICMD_UNNUMBEREDSEC */ + { dolink, "uref", 4 }, /* TEXICMD_UREF */ + { dolink, "url", 3 }, /* TEXICMD_URL */ + { doinline, "var", 3 }, /* TEXICMD_VAR */ + { dosp, "vskip", 5 }, /* TEXICMD_VSKIP */ + { dobracket, "w", 1 }, /* TEXICMD_W */ + { dolink, "xref", 4 }, /* TEXICMD_XREF */ }; +/* + * Unmap the top-most file that we're using. + */ static void texifilepop(struct texi *p) { @@ -216,16 +366,36 @@ texifilepop(struct texi *p) 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) { + 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 -texifatal(struct texi *p, const char *errstring) +texiabort(struct texi *p, const char *errstring) { perror(errstring); @@ -242,7 +412,7 @@ texiwarn(const struct texi *p, const char *fmt, ...) { va_list ap; - fprintf(stderr, "%s:%zu:%zu: ", + 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); @@ -253,21 +423,40 @@ texiwarn(const struct texi *p, const char *fmt, ...) } /* - * Put a single data character. - * This MUST NOT be a mdoc(7) command: it should be free text that's - * outputted to the screen. + * 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. + */ +static void texiputchar(struct texi *p, char c) { - if (TEXI_IGN & p->flags) + if (p->ign) return; - putchar(c); + p->seenvs = 0; if ('\n' == c) { p->outcol = 0; - p->outmacro = 0; p->seenws = 0; } else p->outcol++; @@ -285,39 +474,69 @@ texiputchars(struct texi *p, const char *s) } /* - * Put an mdoc(7) command without the trailing newline. - * This should ONLY be used for mdoc(7) commands! + * Close an mdoc(7) macro opened with teximacroopen(). + * If there are no more macros on the line, prints a newline. */ static void -texifputs(struct texi *p, const char *s) +teximacroclose(struct texi *p) { - int rc; - if (TEXI_IGN & p->flags) - return; - if (p->outcol) + /* FIXME: punctuation. */ + if (0 == --p->outmacro) texiputchar(p, '\n'); - if (EOF != (rc = fputs(s, stdout))) - p->outcol += rc; } /* - * Put an mdoc(7) command with the trailing newline. - * This should ONLY be used for mdoc(7) commands! + * 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) +{ + + if (p->outcol && 0 == p->outmacro) + texiputchar(p, '\n'); + if (0 == p->outmacro) + texiputchar(p, '.'); + else + texiputchar(p, ' '); + texiputchars(p, s); + texiputchar(p, ' '); + 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 (TEXI_IGN & p->flags) - return; + if (p->outmacro) + texierr(p, "\"%s\" in open line scope!?", s); + else if (p->literal) + texierr(p, "\"%s\" in a literal scope!?", s); + if (p->outcol) texiputchar(p, '\n'); - puts(s); - p->outcol = 0; - p->seenws = 0; + + texiputchar(p, '.'); + texiputchars(p, s); + texiputchar(p, '\n'); } +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. */ @@ -334,6 +553,40 @@ advance(struct texi *p, const char *buf, size_t *pos) (*pos)++; } +static void +texipunctuate(struct texi *p, const char *buf, size_t sz, size_t *pos) +{ + 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 @@ -343,15 +596,18 @@ static size_t advancenext(struct texi *p, const char *buf, size_t sz, size_t *pos) { - if (TEXI_LITERAL & p->flags) { - while (*pos < sz && isspace(buf[*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 && isspace(buf[*pos])) { + while (*pos < sz && ismspace(buf[*pos])) { p->seenws = 1; /* * If it looks like we've printed a double-line, then @@ -359,7 +615,7 @@ advancenext(struct texi *p, const char *buf, size_t sz * FIXME: this is stupid. */ if (*pos && '\n' == buf[*pos] && '\n' == buf[*pos - 1]) - teximacro(p, ".Pp"); + texivspace(p); advance(p, buf, pos); } return(*pos); @@ -369,11 +625,15 @@ advancenext(struct texi *p, const char *buf, size_t sz * Advance to the EOLN in the input stream. */ static size_t -advanceeoln(struct texi *p, const char *buf, size_t sz, size_t *pos) +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); } @@ -396,28 +656,27 @@ advanceto(struct texi *p, const char *buf, size_t *pos * This also will advance the input stream. */ static void -texiword(struct texi *p, const char *buf, size_t sz, size_t *pos) +texiword(struct texi *p, const char *buf, + size_t sz, size_t *pos, char extra) { - /* - * XXX: if we're in literal mode, then we shouldn't do any - * reflowing of text here. - */ - if (p->outcol > 72 && ! (TEXI_LITERAL & p->flags)) + if (0 == p->outmacro && p->outcol > 72 && 0 == p->literal) texiputchar(p, '\n'); - - if (p->seenws && p->outcol && ! (TEXI_LITERAL & p->flags)) + /* FIXME: abstract this: we use it elsewhere. */ + if (p->seenws && p->outcol && 0 == p->literal) texiputchar(p, ' '); p->seenws = 0; - while (*pos < sz && ! isspace(buf[*pos])) { + 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]) { @@ -434,6 +693,11 @@ texiword(struct texi *p, const char *buf, size_t sz, s } } +/* + * 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) @@ -441,8 +705,29 @@ texicmd(struct texi *p, const char *buf, size_t i, len; assert('@' == buf[pos]); - for (*end = ++pos; *end < sz && ! isspace(buf[*end]); (*end)++) - if ('@' == buf[*end] || '{' == buf[*end]) + + 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]))) break; len = *end - pos; @@ -453,49 +738,75 @@ texicmd(struct texi *p, const char *buf, return(i); } - texiwarn(p, "bad command: %.*s", (int)len, &buf[pos]); + texiwarn(p, "bad command: @%.*s", (int)len, &buf[pos]); return(TEXICMD__MAX); } -static void -parseeof(struct texi *p, const char *buf, size_t sz) +/* + * 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 pos = 0; + size_t end; enum texicmd cmd; - size_t end; - while ((pos = advancenext(p, buf, sz, &pos)) < sz) { - switch (buf[pos]) { + 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 ('}'): - texiwarn(p, "unexpected \"}\""); - advance(p, buf, &pos); - continue; + advance(p, buf, pos); + return(0); case ('{'): - texiwarn(p, "unexpected \"{\""); - advance(p, buf, &pos); + if (0 == p->ign) + texiwarn(p, "unexpected \"{\""); + advance(p, buf, pos); continue; case ('@'): break; default: - texiword(p, buf, sz, &pos); + texiword(p, buf, sz, pos, ','); continue; } - cmd = texicmd(p, buf, pos, sz, &end); - advanceto(p, buf, &pos, end); + 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); - } + (*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); @@ -506,13 +817,14 @@ parsebracket(struct texi *p, const char *buf, size_t s advance(p, buf, pos); return; case ('{'): - texiwarn(p, "unexpected \"{\""); + if (0 == p->ign) + texiwarn(p, "unexpected \"{\""); advance(p, buf, pos); continue; case ('@'): break; default: - texiword(p, buf, sz, pos); + texiword(p, buf, sz, pos, '\0'); continue; } @@ -525,7 +837,108 @@ parsebracket(struct texi *p, const char *buf, size_t s } } +/* + * 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; + + assert(0 == p->literal); + + while (*pos < sz && '\n' != buf[*pos]) { + while (*pos < sz && isws(buf[*pos])) { + p->seenws = 1; + 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); +} + +/* + * 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) { @@ -539,36 +952,39 @@ parseto(struct texi *p, const char *buf, while ((*pos = advancenext(p, buf, sz, pos)) < sz) { switch (buf[*pos]) { case ('}'): - texiwarn(p, "unexpected \"}\""); + if (0 == p->ign) + texiwarn(p, "unexpected \"}\""); advance(p, buf, pos); continue; case ('{'): - texiwarn(p, "unexpected \"{\""); + if (0 == p->ign) + texiwarn(p, "unexpected \"{\""); advance(p, buf, pos); continue; case ('@'): break; default: - texiword(p, buf, sz, pos); + 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 && ' ' == buf[*pos]) + while (*pos < sz && isws(buf[*pos])) advance(p, buf, pos); /* - * FIXME: skip tabs and also check the full - * word, not just its initial substring! + * 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); + advanceeoln(p, buf, sz, pos, 0); break; } - texiwarn(p, "unexpected \"end\""); - advanceeoln(p, buf, sz, pos); + 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) @@ -576,116 +992,250 @@ parseto(struct texi *p, const char *buf, } } +/* + * Memory-map the file "fname" and begin parsing it. + * This can be called in a nested context. + */ static void -doignblock(struct texi *p, enum texicmd cmd, +parsefile(struct texi *p, const char *fname) +{ + struct texifile *f; + int fd; + struct stat st; + + 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++; + parseeof(p, f->map, f->mapsz); + texifilepop(p); +} + +static void +dodeftypevar(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - unsigned int sv = p->flags; - const char *blockname; - - p->flags |= TEXI_IGN; + const char *blk; + + blk = TEXICMD_DEFTYPEVR == cmd ? + "deftypevr" : "deftypevar"; + + if (p->ign) { + parseto(p, buf, sz, pos, blk); + return; + } + + texivspace(p); + if (TEXICMD_DEFTYPEVR == cmd) { + parsebracket(p, buf, sz, pos); + texiputchars(p, ":\n"); + } + teximacroopen(p, "Vt"); + parseeoln(p, buf, sz, pos); + teximacroclose(p); + texivspace(p); + parseto(p, buf, sz, pos, blk); +} + +static void +dodeftypefun(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) +{ + const char *blk; + + blk = NULL; switch (cmd) { - case (TEXICMD_COPYING): - blockname = "copying"; + case (TEXICMD_DEFTYPEFN): + case (TEXICMD_DEFTYPEFUN): + blk = texitoks[cmd].tok; break; - case (TEXICMD_DETAILMENU): - blockname = "detailmenu"; + default: break; - case (TEXICMD_DIRENTRY): - blockname = "direntry"; + } + + if (p->ign) { + if (NULL != blk) + parseto(p, buf, sz, pos, blk); + return; + } + + switch (cmd) { + case (TEXICMD_DEFTYPEFN): + case (TEXICMD_DEFTYPEFUN): + texivspace(p); break; - case (TEXICMD_IFHTML): - blockname = "ifhtml"; - break; - case (TEXICMD_IFTEX): - blockname = "iftex"; - break; - case (TEXICMD_MENU): - blockname = "menu"; - break; - case (TEXICMD_TEX): - blockname = "tex"; - break; - case (TEXICMD_TITLEPAGE): - blockname = "titlepage"; - break; default: - abort(); + break; } - parseto(p, buf, sz, pos, blockname); - p->flags = sv; + if (TEXICMD_DEFTYPEFN == cmd || + TEXICMD_DEFTYPEFNX == cmd) { + parsebracket(p, buf, sz, pos); + texiputchars(p, ":\n"); + } + teximacroopen(p, "Ft"); + parsesingle(p, buf, sz, pos); + teximacroclose(p); + teximacroopen(p, "Fn"); + parsesingle(p, buf, sz, pos); + teximacroclose(p); + teximacroopen(p, "Li"); + parseeoln(p, buf, sz, pos); + teximacroclose(p); + texivspace(p); + if (NULL != blk) + parseto(p, buf, sz, pos, blk); } static void -doifnottex(struct texi *p, enum texicmd cmd, +doignblock(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - parseto(p, buf, sz, pos, "ifnottex"); + p->ign++; + parseto(p, buf, sz, pos, texitoks[cmd].tok); + p->ign--; } static void -doinline(struct texi *p, const char *buf, - size_t sz, size_t *pos, const char *macro) +doblock(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) { - - if ( ! p->outmacro) - texifputs(p, "."); - texiputchars(p, macro); - texiputchar(p, ' '); - p->seenws = 0; - p->outmacro++; - parsebracket(p, buf, sz, pos); - p->outmacro--; - if (*pos < sz - 1 && - ismpunct(buf[*pos]) && - isspace(buf[*pos + 1])) { - texiputchar(p, ' '); - texiputchar(p, buf[*pos]); - advance(p, buf, pos); - } - if ( ! p->outmacro) - texiputchar(p, '\n'); + + parseto(p, buf, sz, pos, texitoks[cmd].tok); } static void -doitalic(struct texi *p, enum texicmd cmd, +doinline(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { + const char *macro; - texiputchars(p, "\\fI"); + switch (cmd) { + case (TEXICMD_COMMAND): + macro = "Xr"; + break; + case (TEXICMD_ENV): + macro = "Ev"; + break; + case (TEXICMD_FILE): + macro = "Pa"; + break; + case (TEXICMD_VAR): + macro = "Va"; + break; + default: + abort(); + } + + if (p->literal) { + parsebracket(p, buf, sz, pos); + return; + } + + teximacroopen(p, macro); + p->seenws = 0; parsebracket(p, buf, sz, pos); - texiputchars(p, "\\fP"); + texipunctuate(p, buf, sz, pos); + teximacroclose(p); } static void -doliteral(struct texi *p, enum texicmd cmd, +doinclude(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { + char fname[PATH_MAX], path[PATH_MAX]; + size_t i; + int rc; - if (TEXI_LITERAL & p->flags) - parsebracket(p, buf, sz, pos); - else - doinline(p, buf, sz, pos, "Li"); -} + while (*pos < sz && ' ' == buf[*pos]) + advance(p, buf, pos); -static void -doemph(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) -{ + /* Read in the filename. */ + for (i = 0; *pos < sz && '\n' != buf[*pos]; i++) { + if (i == sizeof(fname) - 1) + break; + fname[i] = buf[*pos]; + advance(p, buf, pos); + } - if (TEXI_LITERAL & p->flags) - doitalic(p, cmd, buf, sz, pos); - else - doinline(p, buf, sz, pos, "Em"); + if (i == 0) + texierr(p, "path too short"); + else if ('\n' != buf[*pos]) + texierr(p, "path too long"); + else if ('/' == fname[0]) + texierr(p, "no absolute paths"); + fname[i] = '\0'; + + if (strstr(fname, "../") || strstr(fname, "/..")) + texierr(p, "insecure path"); + + for (i = 0; i < p->dirsz; i++) { + rc = snprintf(path, sizeof(path), + "%s/%s", p->dirs[i], fname); + if (rc < 0) + texierr(p, "couldn't format path"); + else if ((size_t)rc >= sizeof(path)) + texierr(p, "path too long"); + else if (-1 == access(path, R_OK)) + continue; + + parsefile(p, path); + return; + } + + texierr(p, "couldn't find %s in includes", fname); } static void -docommand(struct texi *p, enum texicmd cmd, +dofont(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { + const char *font; - doinline(p, buf, sz, pos, "Xr"); + switch (cmd) { + case (TEXICMD_B): + case (TEXICMD_STRONG): + font = "\\fB"; + break; + case (TEXICMD_CITE): + case (TEXICMD_DFN): + case (TEXICMD_EMPH): + case (TEXICMD_I): + case (TEXICMD_SLANTED): + font = "\\fI"; + break; + case (TEXICMD_CODE): + case (TEXICMD_KBD): + case (TEXICMD_R): + case (TEXICMD_SAMP): + case (TEXICMD_SANSSERIF): + case (TEXICMD_T): + font = "\\fR"; + break; + default: + abort(); + } + + texiputchars(p, font); + parsebracket(p, buf, sz, pos); + texiputchars(p, "\\fP"); } static void @@ -697,31 +1247,33 @@ dobracket(struct texi *p, enum texicmd cmd, } static void -dofile(struct texi *p, enum texicmd cmd, +dodisplay(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - if (TEXI_LITERAL & p->flags) - parsebracket(p, buf, sz, pos); - else - doinline(p, buf, sz, pos, "Pa"); + teximacro(p, "Bd -display -offset indent"); + p->seenvs = 1; + /* FIXME: ignore and parseeoln. */ + advanceeoln(p, buf, sz, pos, 1); + parseto(p, buf, sz, pos, "display"); + teximacro(p, "Ed"); } static void doexample(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - unsigned int sv; + const char *blk; - teximacro(p, ".Bd -literal"); - advanceeoln(p, buf, sz, pos); - if ('\n' == buf[*pos]) - advance(p, buf, pos); - sv = p->flags; - p->flags |= TEXI_LITERAL; - parseto(p, buf, sz, pos, "example"); - p->flags = sv; - teximacro(p, ".Ed"); + blk = TEXICMD_EXAMPLE == cmd ? "example" : "smallexample"; + + teximacro(p, "Bd -literal -offset indent"); + /* FIXME: ignore and parseeoln. */ + advanceeoln(p, buf, sz, pos, 1); + p->literal++; + parseto(p, buf, sz, pos, blk); + p->literal--; + teximacro(p, "Ed"); } static void @@ -734,28 +1286,79 @@ dobye(struct texi *p, enum texicmd cmd, } static void +dotitle(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) +{ + size_t start, end; + + while (*pos < sz && isws(buf[*pos])) + advance(p, buf, pos); + start = end = *pos; + while (end < sz && '\n' != buf[end]) + end++; + free(p->subtitle); + p->subtitle = malloc(end - start + 1); + memcpy(p->subtitle, &buf[start], end - start); + p->subtitle[end - start] = '\0'; +} + +static void dosymbol(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { + if (p->seenws && p->outcol && 0 == p->literal) { + texiputchar(p, ' '); + p->seenws = 0; + } + switch (cmd) { + case (TEXICMD_ASTERISK): + case (TEXICMD_NEWLINE): + case (TEXICMD_SPACE): + case (TEXICMD_TAB): + texiputchar(p, ' '); + break; case (TEXICMD_AT): - texiputchars(p, "@"); + texiputchar(p, '@'); break; + case (TEXICMD_BANG): + texiputchar(p, '!'); + break; + case (TEXICMD_BULLET): + texiputchars(p, "\\(bu"); + break; case (TEXICMD_COPYRIGHT): texiputchars(p, "\\(co"); break; + case (TEXICMD_DOTS): + texiputchars(p, "..."); + break; case (TEXICMD_LATEX): texiputchars(p, "LaTeX"); break; + case (TEXICMD_QUESTIONMARK): + texiputchar(p, '?'); + break; + case (TEXICMD_SQUIGGLE_LEFT): + texiputchars(p, "{"); + break; + case (TEXICMD_SQUIGGLE_RIGHT): + texiputchars(p, "}"); + break; case (TEXICMD_TEXSYM): texiputchars(p, "TeX"); break; + case (TEXICMD_COLON): + case (TEXICMD_HYPHEN): + break; default: + texiwarn(p, "sym: %d", cmd); abort(); } - doignbracket(p, cmd, buf, sz, pos); + if (texitoks[cmd].len > 1) + doignbracket(p, cmd, buf, sz, pos); } static void @@ -763,175 +1366,271 @@ doquotation(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - teximacro(p, ".Qo"); + teximacro(p, "Qo"); parseto(p, buf, sz, pos, "quotation"); - teximacro(p, ".Qc"); + teximacro(p, "Qc"); } static void -doarg1(struct texi *p, enum texicmd cmd, +domath(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { + size_t nest; + /* + * Math handling is different from everything else. + * We don't allow any subcomponents, and we ignore the rules in + * terms of @-commands. + * This departs from GNU's rules, but whatever. + */ + while (*pos < sz && isws(buf[*pos])) + advance(p, buf, pos); if (*pos == sz || '{' != buf[*pos]) return; advance(p, buf, pos); - if ( ! p->outmacro) - texifputs(p, "."); + if (p->seenws && p->outcol && 0 == p->literal) + texiputchar(p, ' '); + p->seenws = 0; + for (nest = 1; *pos < sz && nest > 0; ) { + if ('{' == buf[*pos]) + nest++; + else if ('}' == buf[*pos]) + if (0 == --nest) + continue; + texiputchar(p, buf[*pos]); + advance(p, buf, pos); + } + if (*pos == sz) + return; + assert('}' == buf[*pos]); + advance(p, buf, pos); +} + +static void +dolink(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) +{ + int c; + switch (cmd) { case (TEXICMD_EMAIL): - texiputchars(p, "Lk "); + teximacroopen(p, "Mt"); break; + case (TEXICMD_UREF): case (TEXICMD_URL): - texiputchars(p, "Mt "); + teximacroopen(p, "Lk"); break; + case (TEXICMD_XREF): + texiputchars(p, "See Section"); + teximacroopen(p, "Qq"); + break; + case (TEXICMD_PXREF): + texiputchars(p, "see Section"); + teximacroopen(p, "Qq"); + break; default: abort(); } - while (*pos < sz && '}' != buf[*pos] && ',' != buf[*pos]) { - texiputchar(p, buf[*pos]); - advance(p, buf, pos); - } - while (*pos < sz && '}' != buf[*pos]) - advance(p, buf, pos); - if (*pos < sz) - advance(p, buf, pos); - if (*pos < sz - 1 && - ismpunct(buf[*pos]) && - isspace(buf[*pos + 1])) { - texiputchar(p, ' '); - texiputchar(p, buf[*pos]); - advance(p, buf, pos); - } - if ( ! p->outmacro) - texiputchar(p, '\n'); + + c = parsearg(p, buf, sz, pos, 0); + p->ign++; + while (c > 0) + c = parsearg(p, buf, sz, pos, 1); + p->ign--; + + texipunctuate(p, buf, sz, pos); + teximacroclose(p); } static void +doignargn(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) +{ + int c; + + c = parsearg(p, buf, sz, pos, 0); + p->ign++; + while (c > 0) + c = parsearg(p, buf, sz, pos, 1); + p->ign--; +} + +static void dosubsection(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - if (TEXI_IGN & p->flags) { - advanceeoln(p, buf, sz, pos); - return; - } - while (*pos < sz && ' ' == buf[*pos]) - advance(p, buf, pos); - texifputs(p, ".Pp"); - while (*pos < sz && '\n' != buf[*pos]) { - texiputchar(p, buf[*pos]); - advance(p, buf, pos); - } - texifputs(p, ".Pp"); + texivspace(p); + teximacroopen(p, "Em"); + parseeoln(p, buf, sz, pos); + teximacroclose(p); + texivspace(p); } static void dosection(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { + const char *blk; - if (TEXI_IGN & p->flags) { - advanceeoln(p, buf, sz, pos); - return; + switch (cmd) { + case (TEXICMD_APPENDIX): + case (TEXICMD_CHAPTER): + case (TEXICMD_TOP): + case (TEXICMD_UNNUMBERED): + blk = "Sh"; + break; + case (TEXICMD_APPENDIXSEC): + case (TEXICMD_HEADING): + case (TEXICMD_SECTION): + case (TEXICMD_UNNUMBEREDSEC): + blk = "Ss"; + break; + default: + abort(); } - while (*pos < sz && ' ' == buf[*pos]) - advance(p, buf, pos); - texifputs(p, ".Ss "); - while (*pos < sz && '\n' != buf[*pos]) { - texiputchar(p, buf[*pos]); - advance(p, buf, pos); - } - texiputchar(p, '\n'); + + if (p->outmacro) + texierr(p, "%s in open line scope!?", blk); + else if (p->literal) + texierr(p, "%s in a literal scope!?", blk); + + teximacroopen(p, blk); + parseeoln(p, buf, sz, pos); + teximacroclose(p); + p->seenvs = 1; } static void -dosh(struct texi *p, enum texicmd cmd, +dosp(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - if (TEXI_IGN & p->flags) { - advanceeoln(p, buf, sz, pos); - return; - } - while (*pos < sz && ' ' == buf[*pos]) - advance(p, buf, pos); - texifputs(p, ".Sh "); - while (*pos < sz && '\n' != buf[*pos]) { - texiputchar(p, toupper(buf[*pos])); - advance(p, buf, pos); - } - texiputchar(p, '\n'); + texivspace(p); + /* FIXME: ignore and parseeoln. */ + advanceeoln(p, buf, sz, pos, 1); } static void dotop(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { + const char *cp; + time_t t; + char date[32]; - p->flags &= ~TEXI_HEADER; - advanceeoln(p, buf, sz, pos); - teximacro(p, ".Dd $Mdocdate: February 16 2015 $"); - teximacro(p, ".Dt SOMETHING 7"); - teximacro(p, ".Os"); - teximacro(p, ".Sh NAME"); - teximacro(p, ".Nm Something"); - teximacro(p, ".Nd Something"); + t = time(NULL); + strftime(date, sizeof(date), "%F", localtime(&t)); + + p->ign--; + teximacroopen(p, "Dd"); + texiputchars(p, date); + teximacroclose(p); + teximacroopen(p, "Dt"); + for (cp = p->title; '\0' != *cp; cp++) + texiputchar(p, toupper(*cp)); + texiputchars(p, " 7"); + teximacroclose(p); + teximacro(p, "Os"); + teximacro(p, "Sh NAME"); + teximacroopen(p, "Nm"); + texiputchars(p, p->title); + teximacroclose(p); + teximacroopen(p, "Nd"); + texiputchars(p, NULL != p->subtitle ? + p->subtitle : "Unknown description"); + teximacroclose(p); + p->seenvs = 1; + dosection(p, cmd, buf, sz, pos); } static void doitem(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - size_t end; - /* See if we have arguments... */ - for (end = *pos; end < sz; end++) - if (' ' != buf[end] && '\t' != buf[end]) - break; + if (p->outmacro) + texierr(p, "item in open line scope!?"); + else if (p->literal) + texierr(p, "item in a literal scope!?"); - /* If we have arguments, print them too. */ - if ('\n' != buf[end]) { - texifputs(p, ".It"); - /* FIXME: process commands. */ - while (*pos < sz && '\n' != buf[*pos]) { - texiputchar(p, buf[*pos]); - advance(p, buf, pos); - } + switch (p->list) { + case (TEXILIST_ITEM): + teximacroopen(p, "It"); + break; + case (TEXILIST_NOITEM): + teximacro(p, "It"); + break; + default: + texivspace(p); + break; + } + p->seenvs = 1; + + parseeoln(p, buf, sz, pos); + + if (TEXILIST_ITEM == p->list) + teximacroclose(p); + else if (p->outcol > 0) texiputchar(p, '\n'); - } else - teximacro(p, ".It"); } static void dotable(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { + enum texilist sv = p->list; - teximacro(p, ".Bl -tag -width Ds"); + p->list = TEXILIST_ITEM; + teximacro(p, "Bl -tag -width Ds"); + /* FIXME: ignore and parseeoln. */ + advanceeoln(p, buf, sz, pos, 1); + p->seenvs = 1; parseto(p, buf, sz, pos, "table"); - teximacro(p, ".El"); + teximacro(p, "El"); + p->list = sv; } static void +doenumerate(struct texi *p, enum texicmd cmd, + const char *buf, size_t sz, size_t *pos) +{ + enum texilist sv = p->list; + + p->list = TEXILIST_NOITEM; + teximacro(p, "Bl -enum"); + p->seenvs = 1; + /* FIXME: ignore and parseeoln. */ + advanceeoln(p, buf, sz, pos, 1); + parseto(p, buf, sz, pos, "enumerate"); + teximacro(p, "El"); + p->list = sv; +} + +static void doitemize(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { + enum texilist sv = p->list; - teximacro(p, ".Bl -bullet"); + p->list = TEXILIST_ITEM; + teximacro(p, "Bl -bullet"); + p->seenvs = 1; + /* FIXME: ignore and parseeoln. */ + advanceeoln(p, buf, sz, pos, 1); parseto(p, buf, sz, pos, "itemize"); - teximacro(p, ".El"); + teximacro(p, "El"); + p->list = sv; } static void doignbracket(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - unsigned int sv = p->flags; - p->flags |= TEXI_IGN; + p->ign++; parsebracket(p, buf, sz, pos); - p->flags = sv; + p->ign--; } static void @@ -939,52 +1638,63 @@ doignline(struct texi *p, enum texicmd cmd, const char *buf, size_t sz, size_t *pos) { - advanceeoln(p, buf, sz, pos); - if (*pos < sz) - advance(p, buf, pos); + /* FIXME: ignore and parseeoln. */ + advanceeoln(p, buf, sz, pos, 1); } -static int -parsefile(struct texi *p, const char *fname) +/* + * Parse colon-separated directories from "cp" (if not NULL) and returns + * the array of pointers. + * Prepends "base" to the array. + * This does NOT sanitise the directories! + */ +static char ** +parsedirs(const char *base, const char *cp, size_t *sz) { - struct texifile *f; - int fd; - struct stat st; + char *tok, *str, *tofree; + const char *cpp; + size_t i; + char **dirs; - assert(p->filepos < 64); - f = &p->files[p->filepos]; - memset(f, 0, sizeof(struct texifile)); + *sz = NULL != (cpp = cp) ? 2 : 1; + if (*sz > 1) + for ( ; NULL != (cpp = strchr(cpp, ':')); (*sz)++) + cpp++; - f->name = fname; - if (-1 == (fd = open(fname, O_RDONLY, 0))) { - texifatal(p, fname); - } else if (-1 == fstat(fd, &st)) { - close(fd); - texifatal(p, fname); + dirs = calloc(*sz, sizeof(char *)); + if (NULL == dirs) { + perror(NULL); + exit(EXIT_FAILURE); + } else if (NULL == (dirs[0] = strdup(base))) { + perror(NULL); + exit(EXIT_FAILURE); } - f->mapsz = st.st_size; - f->map = mmap(NULL, f->mapsz, - PROT_READ, MAP_SHARED, fd, 0); - close(fd); - - if (MAP_FAILED == f->map) { - texifatal(p, fname); - return(0); + if (NULL == cp) + return(dirs); + + if (NULL == (tofree = tok = str = strdup(cp))) { + perror(NULL); + exit(EXIT_FAILURE); } - p->filepos++; - parseeof(p, f->map, f->mapsz); - texifilepop(p); - return(1); + for (i = 1; NULL != (tok = strsep(&str, ":")); i++) + if (NULL == (dirs[i] = strdup(tok))) { + perror(NULL); + exit(EXIT_FAILURE); + } + + free(tofree); + return(dirs); } int main(int argc, char *argv[]) { struct texi texi; - int c, rc; - const char *progname; + int c; + char *path, *dir; + const char *progname, *Idir, *cp; progname = strrchr(argv[0], '/'); if (progname == NULL) @@ -992,8 +1702,14 @@ main(int argc, char *argv[]) else ++progname; - while (-1 != (c = getopt(argc, argv, ""))) + memset(&texi, 0, sizeof(struct texi)); + Idir = NULL; + + while (-1 != (c = getopt(argc, argv, "I:"))) switch (c) { + case ('I'): + Idir = optarg; + break; default: goto usage; } @@ -1002,12 +1718,34 @@ main(int argc, char *argv[]) if (0 == (argc -= optind)) goto usage; - memset(&texi, 0, sizeof(struct texi)); - texi.flags = TEXI_HEADER; - rc = parsefile(&texi, argv[0]); - return(rc ? EXIT_SUCCESS : EXIT_FAILURE); + if (NULL == (path = strdup(argv[0]))) { + perror(NULL); + exit(EXIT_FAILURE); + } else if (NULL == (dir = dirname(path))) { + perror(argv[0]); + free(path); + exit(EXIT_FAILURE); + } + free(path); + if (NULL != (cp = strrchr(argv[0], '/'))) + texi.title = strdup(cp + 1); + else + texi.title = strdup(argv[0]); + + if (NULL == texi.title) { + perror(NULL); + exit(EXIT_FAILURE); + } else if (NULL != (path = strchr(texi.title, '.'))) + *path = '\0'; + + texi.ign = 1; + texi.dirs = parsedirs(dir, Idir, &texi.dirsz); + parsefile(&texi, argv[0]); + /* We shouldn't get here. */ + texiexit(&texi); + return(EXIT_FAILURE); usage: - fprintf(stderr, "usage: %s file\n", progname); + fprintf(stderr, "usage: %s [-Idirs] file\n", progname); return(EXIT_FAILURE); }