=================================================================== RCS file: /cvs/mandoc/main.c,v retrieving revision 1.72 retrieving revision 1.84 diff -u -p -r1.72 -r1.84 --- mandoc/main.c 2010/05/15 18:43:59 1.72 +++ mandoc/main.c 2010/06/07 10:52:44 1.84 @@ -1,4 +1,4 @@ -/* $Id: main.c,v 1.72 2010/05/15 18:43:59 kristaps Exp $ */ +/* $Id: main.c,v 1.84 2010/06/07 10:52:44 kristaps Exp $ */ /* * Copyright (c) 2008, 2009 Kristaps Dzonsons * @@ -81,28 +81,92 @@ struct curparse { #define FL_NIGN_MACRO (1 << 2) /* Don't ignore bad macros. */ #define FL_IGN_ERRORS (1 << 4) /* Ignore failed parse. */ #define FL_STRICT FL_NIGN_ESCAPE | \ - FL_NIGN_MACRO - enum intt inttype; /* Input parsers... */ - struct man *man; - struct mdoc *mdoc; - struct roff *roff; - enum outt outtype; /* Output devices... */ - out_mdoc outmdoc; - out_man outman; - out_free outfree; - void *outdata; - char outopts[BUFSIZ]; + FL_NIGN_MACRO /* ignore nothing */ + enum intt inttype; /* which parser to use */ + struct man *man; /* man parser */ + struct mdoc *mdoc; /* mdoc parser */ + struct roff *roff; /* roff parser (!NULL) */ + enum outt outtype; /* which output to use */ + out_mdoc outmdoc; /* mdoc output ptr */ + out_man outman; /* man output ptr */ + out_free outfree; /* free output ptr */ + void *outdata; /* data for output */ + char outopts[BUFSIZ]; /* buf of output opts */ }; +static const char * const mandocerrs[MANDOCERR_MAX] = { + "ok", + "text should be uppercase", + "sections out of conventional order", + "section name repeats", + "out of order prologue", + "repeated prologue entry", + "list type must come first", + "bad standard", + "bad library", + "bad escape sequence", + "unterminated quoted string", + "argument requires the width argument", + "superfluous width argument", + "bad date argument", + "bad width argument", + "unknown manual section", + "section not in conventional manual section", + "end of line whitespace", + "scope open on exit", + "NAME section must come first", + "bad Boolean value", + "child violates parent syntax", + "bad AT&T symbol", + "list type repeated", + "display type repeated", + "argument repeated", + "manual name not yet set", + "obsolete macro ignored", + "empty macro ignored", + "macro not allowed in body", + "macro not allowed in prologue", + "bad character", + "bad NAME section contents", + "no blank lines", + "no text in this context", + "bad comment style", + "unknown macro will be lost", + "line scope broken", + "scope broken", + "argument count wrong", + "request scope close w/none open", + "scope already open", + "macro requires line argument(s)", + "macro requires body argument(s)", + "macro requires argument(s)", + "no title in document", + "missing list type", + "line argument(s) will be lost", + "body argument(s) will be lost", + "column syntax is inconsistent", + "missing font type", + "missing display type", + "displays may not be nested", + "no scope to rewind: syntax violated", + "scope broken, syntax violated", + "line scope broken, syntax violated", + "argument count wrong, violates syntax", + "child violates parent syntax", + "argument count wrong, violates syntax", + "no document body", + "no document prologue", + "utsname system call failed", + "memory exhausted", +}; + static void fdesc(struct curparse *); static void ffile(const char *, struct curparse *); static int foptions(int *, char *); static struct man *man_init(struct curparse *); static struct mdoc *mdoc_init(struct curparse *); static struct roff *roff_init(struct curparse *); -static int merr(void *, int, int, const char *); /* DEPRECATED */ static int moptions(enum intt *, char *); -static int mwarn(void *, int, int, const char *); /* DEPRECATED */ static int mmsg(enum mandocerr, void *, int, int, const char *); static int pset(const char *, int, struct curparse *, @@ -220,11 +284,7 @@ static struct man * man_init(struct curparse *curp) { int pflags; - struct man_cb mancb; - mancb.man_err = merr; - mancb.man_warn = mwarn; - /* Defaults from mandoc.1. */ pflags = MAN_IGN_MACRO | MAN_IGN_ESCAPE; @@ -234,7 +294,7 @@ man_init(struct curparse *curp) if (curp->fflags & FL_NIGN_ESCAPE) pflags &= ~MAN_IGN_ESCAPE; - return(man_alloc(curp, pflags, &mancb)); + return(man_alloc(curp, pflags, mmsg)); } @@ -250,11 +310,7 @@ static struct mdoc * mdoc_init(struct curparse *curp) { int pflags; - struct mdoc_cb mdoccb; - mdoccb.mdoc_err = merr; - mdoccb.mdoc_warn = mwarn; - /* Defaults from mandoc.1. */ pflags = MDOC_IGN_MACRO | MDOC_IGN_ESCAPE; @@ -266,7 +322,7 @@ mdoc_init(struct curparse *curp) if (curp->fflags & FL_NIGN_MACRO) pflags &= ~MDOC_IGN_MACRO; - return(mdoc_alloc(curp, pflags, &mdoccb)); + return(mdoc_alloc(curp, pflags, mmsg)); } @@ -339,7 +395,7 @@ read_whole_file(struct curparse *curp, struct buf *fb, *with_mmap = 1; fb->sz = (size_t)st.st_size; fb->buf = mmap(NULL, fb->sz, PROT_READ, - MAP_FILE, curp->fd, 0); + MAP_FILE|MAP_SHARED, curp->fd, 0); if (fb->buf != MAP_FAILED) return(1); } @@ -386,7 +442,7 @@ static void fdesc(struct curparse *curp) { struct buf ln, blk; - int i, pos, lnn, lnn_start, with_mmap; + int i, pos, lnn, lnn_start, with_mmap, of; enum rofferr re; struct man *man; struct mdoc *mdoc; @@ -467,22 +523,42 @@ fdesc(struct curparse *curp) goto bailout; ln.buf[pos] = '\0'; - re = roff_parseln(roff, lnn_start, &ln.buf, &ln.sz); + /* + * A significant amount of complexity is contained by + * the roff preprocessor. It's line-oriented but can be + * expressed on one line, so we need at times to + * readjust our starting point and re-run it. The roff + * preprocessor can also readjust the buffers with new + * data, so we pass them in wholesale. + */ + + of = 0; + do { + re = roff_parseln(roff, lnn_start, + &ln.buf, &ln.sz, of, &of); + } while (ROFF_RERUN == re); + if (ROFF_IGN == re) continue; else if (ROFF_ERR == re) goto bailout; - /* If unset, assign parser in pset(). */ + /* + * If input parsers have not been allocated, do so now. + * We keep these instanced betwen parsers, but set them + * locally per parse routine since we can use different + * parsers with each one. + */ - if ( ! (man || mdoc) && ! pset(ln.buf, pos, curp, &man, &mdoc)) - goto bailout; + if ( ! (man || mdoc)) + if ( ! pset(ln.buf + of, pos - of, curp, &man, &mdoc)) + goto bailout; - /* Pass down into parsers. */ + /* Lastly, push down into the parsers themselves. */ - if (man && ! man_parseln(man, lnn_start, ln.buf)) + if (man && ! man_parseln(man, lnn_start, ln.buf, of)) goto bailout; - if (mdoc && ! mdoc_parseln(mdoc, lnn_start, ln.buf)) + if (mdoc && ! mdoc_parseln(mdoc, lnn_start, ln.buf, of)) goto bailout; } @@ -493,6 +569,8 @@ fdesc(struct curparse *curp) goto bailout; } + /* Clean up the parse routine ASTs. */ + if (mdoc && ! mdoc_endparse(mdoc)) goto bailout; if (man && ! man_endparse(man)) @@ -523,7 +601,7 @@ fdesc(struct curparse *curp) case (OUTT_LINT): break; default: - curp->outdata = ascii_alloc(80); + curp->outdata = ascii_alloc(curp->outopts); curp->outman = terminal_man; curp->outmdoc = terminal_mdoc; curp->outfree = terminal_free; @@ -574,7 +652,7 @@ pset(const char *buf, int pos, struct curparse *curp, * default to -man, which is more lenient. */ - if (buf[0] == '.') { + if ('.' == buf[0] || '\'' == buf[0]) { for (i = 1; buf[i]; i++) if (' ' != buf[i] && '\t' != buf[i]) break; @@ -734,58 +812,34 @@ woptions(int *wflags, char *arg) } -/* ARGSUSED */ static int -merr(void *arg, int line, int col, const char *msg) +mmsg(enum mandocerr t, void *arg, int ln, int col, const char *msg) { - struct curparse *curp; + struct curparse *cp; - curp = (struct curparse *)arg; + cp = (struct curparse *)arg; - (void)fprintf(stderr, "%s:%d:%d: error: %s\n", - curp->file, line, col + 1, msg); + if (t <= MANDOCERR_ERROR) { + if ( ! (cp->wflags & WARN_WALL)) + return(1); + with_warning = 1; + } else + with_error = 1; - with_error = 1; + fprintf(stderr, "%s:%d:%d: %s", cp->file, + ln, col + 1, mandocerrs[t]); - return(0); -} + if (msg) + fprintf(stderr, ": %s", msg); + fputc('\n', stderr); -static int -mwarn(void *arg, int line, int col, const char *msg) -{ - struct curparse *curp; - - curp = (struct curparse *)arg; - - if ( ! (curp->wflags & WARN_WALL)) - return(1); - - (void)fprintf(stderr, "%s:%d:%d: warning: %s\n", - curp->file, line, col + 1, msg); - - with_warning = 1; - if (curp->wflags & WARN_WERR) { + /* This is superfluous, but whatever. */ + if (t > MANDOCERR_ERROR) + return(0); + if (cp->wflags & WARN_WERR) { with_error = 1; return(0); } - - return(1); -} - -/* - * XXX: this is experimental code that will eventually become the - * generic means of covering all warnings and errors! - */ -/* ARGSUSED */ -static int -mmsg(enum mandocerr t, void *arg, int ln, int col, const char *msg) -{ - struct curparse *cp; - - cp = (struct curparse *)arg; - - /*fprintf(stderr, "%s:%d:%d: %s\n", cp->file, ln, col + 1, msg);*/ - return(1); }