=================================================================== RCS file: /cvs/mandoc/read.c,v retrieving revision 1.141 retrieving revision 1.169 diff -u -p -r1.141 -r1.169 --- mandoc/read.c 2015/09/14 15:36:14 1.141 +++ mandoc/read.c 2017/06/03 15:55:24 1.169 @@ -1,7 +1,7 @@ -/* $Id: read.c,v 1.141 2015/09/14 15:36:14 schwarze Exp $ */ +/* $Id: read.c,v 1.169 2017/06/03 15:55:24 schwarze Exp $ */ /* * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons - * Copyright (c) 2010-2015 Ingo Schwarze + * Copyright (c) 2010-2017 Ingo Schwarze * Copyright (c) 2010, 2012 Joerg Sonnenberger * * Permission to use, copy, modify, and distribute this software for any @@ -19,17 +19,17 @@ #include "config.h" #include -#if HAVE_MMAP #include #include -#endif #include #include +#if HAVE_ERR +#include +#endif #include #include #include -#include #include #include #include @@ -47,9 +47,8 @@ #define REPARSE_LIMIT 1000 struct mparse { - struct roff_man *man; /* man parser */ struct roff *roff; /* roff parser (!NULL) */ - const struct mchars *mchars; /* character table */ + struct roff_man *man; /* man parser */ char *sodest; /* filename pointed to by .so */ const char *file; /* filename of current input file */ struct buf *primary; /* buffer currently being parsed */ @@ -67,7 +66,7 @@ struct mparse { static void choose_parser(struct mparse *); static void resize_buf(struct buf *, size_t); -static void mparse_buf_r(struct mparse *, struct buf, size_t, int); +static int mparse_buf_r(struct mparse *, struct buf, size_t, int); static int read_whole_file(struct mparse *, const char *, int, struct buf *, int *); static void mparse_end(struct mparse *); @@ -76,8 +75,8 @@ static void mparse_parse_buffer(struct mparse *, str static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = { MANDOCERR_OK, + MANDOCERR_STYLE, MANDOCERR_WARNING, - MANDOCERR_WARNING, MANDOCERR_ERROR, MANDOCERR_UNSUPP, MANDOCERR_MAX, @@ -87,6 +86,12 @@ static const enum mandocerr mandoclimits[MANDOCLEVEL_M static const char * const mandocerrs[MANDOCERR_MAX] = { "ok", + "generic style suggestion", + + "useless macro", + "consider using OS macro", + "description line ends with a full stop", + "generic warning", /* related to the prologue */ @@ -108,11 +113,13 @@ static const char * const mandocerrs[MANDOCERR_MAX] = "no document body", "content before first section header", "first section is not \"NAME\"", - "NAME section without name", + "NAME section without Nm before Nd", "NAME section without description", "description not at the end of NAME", "bad NAME section content", + "missing comma before name", "missing description line, using \"\"", + "description line outside NAME section", "sections out of conventional order", "duplicate section title", "unexpected section", @@ -132,6 +139,7 @@ static const char * const mandocerrs[MANDOCERR_MAX] = "fill mode already enabled, skipping", "fill mode already disabled, skipping", "line scope broken", + "skipping blank line in line scope", /* related to missing macro arguments */ "skipping empty request", @@ -141,7 +149,7 @@ static const char * const mandocerrs[MANDOCERR_MAX] = "empty argument, using 0n", "missing display type, using -ragged", "list type is not the first argument", - "missing -width in -tag list, using 8n", + "missing -width in -tag list, using 6n", "missing utility name, using \"\"", "missing function name, using \"\"", "empty head in list item", @@ -150,6 +158,7 @@ static const char * const mandocerrs[MANDOCERR_MAX] = "unknown font type, using \\fR", "nothing follows prefix", "empty reference block", + "missing section argument", "missing -std argument, adding it", "missing option string, using \"\"", "missing resource identifier, using \"\"", @@ -175,6 +184,7 @@ static const char * const mandocerrs[MANDOCERR_MAX] = "blank line in fill mode, using .sp", "tab in filled text", "whitespace at end of input line", + "new sentence, new line", "bad comment style", "invalid escape sequence", "undefined string, using \"\"", @@ -215,6 +225,7 @@ static const char * const mandocerrs[MANDOCERR_MAX] = /* related to request and macro arguments */ "escaped character not allowed in a name", "NOT IMPLEMENTED: Bd -file", + "skipping display without arguments", "missing list type, using -item", "missing manual name, using \"\"", "uname(3) system call failed, using UNKNOWN", @@ -237,7 +248,7 @@ static const char * const mandocerrs[MANDOCERR_MAX] = static const char * const mandoclevels[MANDOCLEVEL_MAX] = { "SUCCESS", - "RESERVED", + "STYLE", "WARNING", "ERROR", "UNSUPP", @@ -288,22 +299,16 @@ choose_parser(struct mparse *curp) } } - if (curp->man == NULL) { - curp->man = roff_man_alloc(curp->roff, curp, curp->defos, - curp->options & MPARSE_QUICK ? 1 : 0); - curp->man->macroset = MACROSET_MAN; - curp->man->first->tok = TOKEN_NONE; - } - if (format == MPARSE_MDOC) { - mdoc_hash_init(); curp->man->macroset = MACROSET_MDOC; - curp->man->first->tok = TOKEN_NONE; + if (curp->man->mdocmac == NULL) + curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX); } else { - man_hash_init(); curp->man->macroset = MACROSET_MAN; - curp->man->first->tok = TOKEN_NONE; + if (curp->man->manmac == NULL) + curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX); } + curp->man->first->tok = TOKEN_NONE; } /* @@ -313,7 +318,7 @@ choose_parser(struct mparse *curp) * macros, inline equations, and input line traps) * and indirectly (for .so file inclusion). */ -static void +static int mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start) { const struct tbl_span *span; @@ -321,6 +326,7 @@ mparse_buf_r(struct mparse *curp, struct buf blk, size const char *save_file; char *cp; size_t pos; /* byte number in the ln buffer */ + size_t j; /* auxiliary byte number in the blk buffer */ enum rofferr rr; int of; int lnn; /* line number in the real file */ @@ -426,14 +432,21 @@ mparse_buf_r(struct mparse *curp, struct buf blk, size } if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) { + j = i; i += 2; /* Comment, skip to end of line */ for (; i < blk.sz; ++i) { - if ('\n' == blk.buf[i]) { - ++i; - ++lnn; - break; - } + if (blk.buf[i] != '\n') + continue; + if (blk.buf[i - 1] == ' ' || + blk.buf[i - 1] == '\t') + mandoc_msg( + MANDOCERR_SPACE_EOL, + curp, curp->line, + pos + i-1 - j, NULL); + ++i; + ++lnn; + break; } /* Backout trailing whitespaces */ @@ -506,13 +519,16 @@ rerun: switch (rr) { case ROFF_REPARSE: - if (REPARSE_LIMIT >= ++curp->reparse_count) - mparse_buf_r(curp, ln, of, 0); - else + if (++curp->reparse_count > REPARSE_LIMIT) mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, pos, NULL); - pos = 0; - continue; + else if (mparse_buf_r(curp, ln, of, 0) == 1 || + start == 1) { + pos = 0; + continue; + } + free(ln.buf); + return 0; case ROFF_APPEND: pos = strlen(ln.buf); continue; @@ -526,7 +542,7 @@ rerun: (i >= blk.sz || blk.buf[i] == '\0')) { curp->sodest = mandoc_strdup(ln.buf + of); free(ln.buf); - return; + return 1; } /* * We remove `so' clauses from our lookaside @@ -536,9 +552,9 @@ rerun: if (curp->secondary) curp->secondary->sz -= pos + 1; save_file = curp->file; - if (mparse_open(curp, &fd, ln.buf + of) == - MANDOCLEVEL_OK) { + if ((fd = mparse_open(curp, ln.buf + of)) != -1) { mparse_readfd(curp, fd, ln.buf + of); + close(fd); curp->file = save_file; } else { curp->file = save_file; @@ -559,15 +575,7 @@ rerun: break; } - /* - * If input parsers have not been allocated, do so now. - * We keep these instanced between parsers, but set them - * locally per parse routine since we can use different - * parsers with each one. - */ - - if (curp->man == NULL || - curp->man->macroset == MACROSET_NONE) + if (curp->man->macroset == MACROSET_NONE) choose_parser(curp); /* @@ -600,22 +608,20 @@ rerun: } free(ln.buf); + return 1; } static int read_whole_file(struct mparse *curp, const char *file, int fd, struct buf *fb, int *with_mmap) { + struct stat st; gzFile gz; size_t off; ssize_t ssz; -#if HAVE_MMAP - struct stat st; - if (-1 == fstat(fd, &st)) { - perror(file); - exit((int)MANDOCLEVEL_SYSERR); - } + if (fstat(fd, &st) == -1) + err((int)MANDOCLEVEL_SYSERR, "%s", file); /* * If we're a regular file, try just reading in the whole entry @@ -627,21 +633,18 @@ read_whole_file(struct mparse *curp, const char *file, if (curp->gzip == 0 && S_ISREG(st.st_mode)) { if (st.st_size > 0x7fffffff) { mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL); - return(0); + return 0; } *with_mmap = 1; fb->sz = (size_t)st.st_size; fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0); if (fb->buf != MAP_FAILED) - return(1); + return 1; } -#endif if (curp->gzip) { - if ((gz = gzdopen(fd, "rb")) == NULL) { - perror(file); - exit((int)MANDOCLEVEL_SYSERR); - } + if ((gz = gzdopen(fd, "rb")) == NULL) + err((int)MANDOCLEVEL_SYSERR, "%s", file); } else gz = NULL; @@ -668,27 +671,21 @@ read_whole_file(struct mparse *curp, const char *file, read(fd, fb->buf + (int)off, fb->sz - off); if (ssz == 0) { fb->sz = off; - return(1); + return 1; } - if (ssz == -1) { - perror(file); - exit((int)MANDOCLEVEL_SYSERR); - } + if (ssz == -1) + err((int)MANDOCLEVEL_SYSERR, "%s", file); off += (size_t)ssz; } free(fb->buf); fb->buf = NULL; - return(0); + return 0; } static void mparse_end(struct mparse *curp) { - - if (curp->man == NULL && curp->sodest == NULL) - curp->man = roff_man_alloc(curp->roff, curp, curp->defos, - curp->options & MPARSE_QUICK ? 1 : 0); if (curp->man->macroset == MACROSET_NONE) curp->man->macroset = MACROSET_MAN; if (curp->man->macroset == MACROSET_MDOC) @@ -748,7 +745,7 @@ mparse_readmem(struct mparse *curp, void *buf, size_t blk.sz = len; mparse_parse_buffer(curp, blk, file); - return(curp->file_status); + return curp->file_status; } /* @@ -768,24 +765,19 @@ mparse_readfd(struct mparse *curp, int fd, const char (MPARSE_UTF8 | MPARSE_LATIN1); mparse_parse_buffer(curp, blk, file); curp->filenc = save_filenc; -#if HAVE_MMAP if (with_mmap) munmap(blk.buf, blk.sz); else -#endif free(blk.buf); } - - if (fd != STDIN_FILENO && close(fd) == -1) - perror(file); - - return(curp->file_status); + return curp->file_status; } -enum mandoclevel -mparse_open(struct mparse *curp, int *fd, const char *file) +int +mparse_open(struct mparse *curp, const char *file) { char *cp; + int fd; curp->file = file; cp = strrchr(file, '.'); @@ -793,8 +785,8 @@ mparse_open(struct mparse *curp, int *fd, const char * /* First try to use the filename as it is. */ - if ((*fd = open(file, O_RDONLY)) != -1) - return(MANDOCLEVEL_OK); + if ((fd = open(file, O_RDONLY)) != -1) + return fd; /* * If that doesn't work and the filename doesn't @@ -803,23 +795,23 @@ mparse_open(struct mparse *curp, int *fd, const char * if ( ! curp->gzip) { mandoc_asprintf(&cp, "%s.gz", file); - *fd = open(file, O_RDONLY); + fd = open(cp, O_RDONLY); free(cp); - if (*fd != -1) { + if (fd != -1) { curp->gzip = 1; - return(MANDOCLEVEL_OK); + return fd; } } /* Neither worked, give up. */ mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno)); - return(MANDOCLEVEL_ERROR); + return -1; } struct mparse * mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg, - const struct mchars *mchars, const char *defos) + const char *defos) { struct mparse *curp; @@ -830,45 +822,46 @@ mparse_alloc(int options, enum mandoclevel wlevel, man curp->mmsg = mmsg; curp->defos = defos; - curp->mchars = mchars; - curp->roff = roff_alloc(curp, curp->mchars, options); + curp->roff = roff_alloc(curp, options); curp->man = roff_man_alloc( curp->roff, curp, curp->defos, curp->options & MPARSE_QUICK ? 1 : 0); if (curp->options & MPARSE_MDOC) { - mdoc_hash_init(); curp->man->macroset = MACROSET_MDOC; + if (curp->man->mdocmac == NULL) + curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX); } else if (curp->options & MPARSE_MAN) { - man_hash_init(); curp->man->macroset = MACROSET_MAN; + if (curp->man->manmac == NULL) + curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX); } curp->man->first->tok = TOKEN_NONE; - return(curp); + return curp; } void mparse_reset(struct mparse *curp) { - roff_reset(curp->roff); + roff_man_reset(curp->man); - if (curp->man != NULL) - roff_man_reset(curp->man); + free(curp->sodest); + curp->sodest = NULL; + if (curp->secondary) curp->secondary->sz = 0; curp->file_status = MANDOCLEVEL_OK; - - free(curp->sodest); - curp->sodest = NULL; + curp->gzip = 0; } void mparse_free(struct mparse *curp) { + roffhash_free(curp->man->mdocmac); + roffhash_free(curp->man->manmac); roff_man_free(curp->man); - if (curp->roff) - roff_free(curp->roff); + roff_free(curp->roff); if (curp->secondary) free(curp->secondary->buf); @@ -891,6 +884,13 @@ mparse_result(struct mparse *curp, struct roff_man **m } void +mparse_updaterc(struct mparse *curp, enum mandoclevel *rc) +{ + if (curp->file_status > *rc) + *rc = curp->file_status; +} + +void mandoc_vmsg(enum mandocerr t, struct mparse *m, int ln, int pos, const char *fmt, ...) { @@ -928,13 +928,13 @@ const char * mparse_strerror(enum mandocerr er) { - return(mandocerrs[er]); + return mandocerrs[er]; } const char * mparse_strlevel(enum mandoclevel lvl) { - return(mandoclevels[lvl]); + return mandoclevels[lvl]; } void @@ -950,5 +950,5 @@ mparse_getkeep(const struct mparse *p) { assert(p->secondary); - return(p->secondary->sz ? p->secondary->buf : NULL); + return p->secondary->sz ? p->secondary->buf : NULL; }