=================================================================== RCS file: /cvs/mandoc/read.c,v retrieving revision 1.204 retrieving revision 1.210 diff -u -p -r1.204 -r1.210 --- mandoc/read.c 2018/12/14 01:24:49 1.204 +++ mandoc/read.c 2018/12/31 04:55:47 1.210 @@ -1,4 +1,4 @@ -/* $Id: read.c,v 1.204 2018/12/14 01:24:49 schwarze Exp $ */ +/* $Id: read.c,v 1.210 2018/12/31 04:55:47 schwarze Exp $ */ /* * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons * Copyright (c) 2010-2018 Ingo Schwarze @@ -47,7 +47,6 @@ struct mparse { struct roff *roff; /* roff parser (!NULL) */ struct roff_man *man; /* man parser */ - char *sodest; /* filename pointed to by .so */ struct buf *primary; /* buffer currently being parsed */ struct buf *secondary; /* copy of top level input */ struct buf *loop; /* open .while request line */ @@ -65,8 +64,6 @@ static void resize_buf(struct buf *, size_t); static int mparse_buf_r(struct mparse *, struct buf, size_t, int); static int read_whole_file(struct mparse *, int, struct buf *, int *); static void mparse_end(struct mparse *); -static void mparse_parse_buffer(struct mparse *, struct buf, - const char *); static void @@ -125,15 +122,15 @@ choose_parser(struct mparse *curp) } if (format == MPARSE_MDOC) { - curp->man->macroset = MACROSET_MDOC; + curp->man->meta.macroset = MACROSET_MDOC; if (curp->man->mdocmac == NULL) curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX); } else { - curp->man->macroset = MACROSET_MAN; + curp->man->meta.macroset = MACROSET_MAN; if (curp->man->manmac == NULL) curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX); } - curp->man->first->tok = TOKEN_NONE; + curp->man->meta.first->tok = TOKEN_NONE; } /* @@ -210,7 +207,7 @@ mparse_buf_r(struct mparse *curp, struct buf blk, size if (c & 0x80) { if ( ! (curp->filenc && preconv_encode( &blk, &i, &ln, &pos, &curp->filenc))) { - mandoc_vmsg(MANDOCERR_CHAR_BAD, curp, + mandoc_msg(MANDOCERR_CHAR_BAD, curp->line, pos, "0x%x", c); ln.buf[pos++] = '?'; i++; @@ -223,10 +220,10 @@ mparse_buf_r(struct mparse *curp, struct buf blk, size */ if (c == 0x7f || (c < 0x20 && c != 0x09)) { - mandoc_vmsg(c == 0x00 || c == 0x04 || + mandoc_msg(c == 0x00 || c == 0x04 || c > 0x0a ? MANDOCERR_CHAR_BAD : MANDOCERR_CHAR_UNSUPP, - curp, curp->line, pos, "0x%x", c); + curp->line, pos, "0x%x", c); i++; if (c != '\r') ln.buf[pos++] = '?'; @@ -300,7 +297,7 @@ rerun: if (loop == curp->loop) break; mandoc_msg(MANDOCERR_WHILE_NEST, - curp, curp->line, pos, NULL); + curp->line, pos, NULL); } curp->loop = thisln; loop = NULL; @@ -310,12 +307,12 @@ rerun: case ROFF_LOOPEXIT: if (curp->loop == NULL) { mandoc_msg(MANDOCERR_WHILE_FAIL, - curp, curp->line, pos, NULL); + curp->line, pos, NULL); break; } if (inloop == 0) { mandoc_msg(MANDOCERR_WHILE_INTO, - curp, curp->line, pos, NULL); + curp->line, pos, NULL); curp->loop = loop = NULL; break; } @@ -336,9 +333,9 @@ rerun: case ROFF_IGN: break; case ROFF_CONT: - if (curp->man->macroset == MACROSET_NONE) + if (curp->man->meta.macroset == MACROSET_NONE) choose_parser(curp); - if ((curp->man->macroset == MACROSET_MDOC ? + if ((curp->man->meta.macroset == MACROSET_MDOC ? mdoc_parseln(curp->man, curp->line, ln.buf, of) : man_parseln(curp->man, curp->line, ln.buf, of) ) == 2) @@ -350,7 +347,7 @@ rerun: if (++curp->reparse_count > REPARSE_LIMIT) { /* Abort and return to the top level. */ result = ROFF_IGN; - mandoc_msg(MANDOCERR_ROFFLOOP, curp, + mandoc_msg(MANDOCERR_ROFFLOOP, curp->line, pos, NULL); goto out; } @@ -367,16 +364,17 @@ rerun: case ROFF_SO: if ( ! (curp->options & MPARSE_SO) && (i >= blk.sz || blk.buf[i] == '\0')) { - curp->sodest = mandoc_strdup(ln.buf + of); + curp->man->meta.sodest = + mandoc_strdup(ln.buf + of); goto out; } if ((fd = mparse_open(curp, ln.buf + of)) != -1) { mparse_readfd(curp, fd, ln.buf + of); close(fd); } else { - mandoc_vmsg(MANDOCERR_SO_FAIL, - curp, curp->line, pos, - ".so %s", ln.buf + of); + mandoc_msg(MANDOCERR_SO_FAIL, + curp->line, of, ".so %s: %s", + ln.buf + of, strerror(errno)); ln.sz = mandoc_asprintf(&cp, ".sp\nSee the file %s.\n.sp", ln.buf + of); @@ -411,7 +409,7 @@ rerun: out: if (inloop) { if (result != ROFF_USERRET) - mandoc_msg(MANDOCERR_WHILE_OUTOF, curp, + mandoc_msg(MANDOCERR_WHILE_OUTOF, curp->line, pos, NULL); curp->loop = NULL; } @@ -431,7 +429,7 @@ read_whole_file(struct mparse *curp, int fd, struct bu int gzerrnum, retval; if (fstat(fd, &st) == -1) { - mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, + mandoc_msg(MANDOCERR_FILE, 0, 0, "fstat: %s", strerror(errno)); return 0; } @@ -445,7 +443,7 @@ read_whole_file(struct mparse *curp, int fd, struct bu if (curp->gzip == 0 && S_ISREG(st.st_mode)) { if (st.st_size > 0x7fffffff) { - mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL); + mandoc_msg(MANDOCERR_TOOLARGE, 0, 0, NULL); return 0; } *with_mmap = 1; @@ -464,12 +462,12 @@ read_whole_file(struct mparse *curp, int fd, struct bu * which this function must not do. */ if ((fd = dup(fd)) == -1) { - mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, + mandoc_msg(MANDOCERR_FILE, 0, 0, "dup: %s", strerror(errno)); return 0; } if ((gz = gzdopen(fd, "rb")) == NULL) { - mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, + mandoc_msg(MANDOCERR_FILE, 0, 0, "gzdopen: %s", strerror(errno)); close(fd); return 0; @@ -490,8 +488,7 @@ read_whole_file(struct mparse *curp, int fd, struct bu for (;;) { if (off == fb->sz) { if (fb->sz == (1U << 31)) { - mandoc_msg(MANDOCERR_TOOLARGE, curp, - 0, 0, NULL); + mandoc_msg(MANDOCERR_TOOLARGE, 0, 0, NULL); break; } resize_buf(fb, 65536); @@ -507,7 +504,7 @@ read_whole_file(struct mparse *curp, int fd, struct bu if (ssz == -1) { if (curp->gzip) (void)gzerror(gz, &gzerrnum); - mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, "read: %s", + mandoc_msg(MANDOCERR_FILE, 0, 0, "read: %s", curp->gzip && gzerrnum != Z_ERRNO ? zError(gzerrnum) : strerror(errno)); break; @@ -516,7 +513,7 @@ read_whole_file(struct mparse *curp, int fd, struct bu } if (curp->gzip && (gzerrnum = gzclose(gz)) != Z_OK) - mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, "gzclose: %s", + mandoc_msg(MANDOCERR_FILE, 0, 0, "gzclose: %s", gzerrnum == Z_ERRNO ? strerror(errno) : zError(gzerrnum)); if (retval == 0) { @@ -529,35 +526,51 @@ read_whole_file(struct mparse *curp, int fd, struct bu static void mparse_end(struct mparse *curp) { - if (curp->man->macroset == MACROSET_NONE) - curp->man->macroset = MACROSET_MAN; - if (curp->man->macroset == MACROSET_MDOC) + if (curp->man->meta.macroset == MACROSET_NONE) + curp->man->meta.macroset = MACROSET_MAN; + if (curp->man->meta.macroset == MACROSET_MDOC) mdoc_endparse(curp->man); else man_endparse(curp->man); roff_endparse(curp->roff); } -static void -mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file) +/* + * Read the whole file into memory and call the parsers. + * Called recursively when an .so request is encountered. + */ +void +mparse_readfd(struct mparse *curp, int fd, const char *filename) { - struct buf *svprimary; - const char *svfile; - size_t offset; static int recursion_depth; - if (64 < recursion_depth) { - mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL); + struct buf blk; + struct buf *save_primary; + const char *save_filename; + size_t offset; + int save_filenc, save_lineno; + int with_mmap; + + if (recursion_depth > 64) { + mandoc_msg(MANDOCERR_ROFFLOOP, curp->line, 0, NULL); return; } + if (read_whole_file(curp, fd, &blk, &with_mmap) == 0) + return; - /* Line number is per-file. */ - svfile = mandoc_msg_getinfilename(); - mandoc_msg_setinfilename(file); - svprimary = curp->primary; + /* + * Save some properties of the parent file. + */ + + save_primary = curp->primary; + save_filenc = curp->filenc; + save_lineno = curp->line; + save_filename = mandoc_msg_getinfilename(); + curp->primary = &blk; + curp->filenc = curp->options & (MPARSE_UTF8 | MPARSE_LATIN1); curp->line = 1; - recursion_depth++; + mandoc_msg_setinfilename(filename); /* Skip an UTF-8 byte order mark. */ if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 && @@ -569,38 +582,25 @@ mparse_parse_buffer(struct mparse *curp, struct buf bl } else offset = 0; + recursion_depth++; mparse_buf_r(curp, blk, offset, 1); - if (--recursion_depth == 0) mparse_end(curp); - curp->primary = svprimary; - if (svfile != NULL) - mandoc_msg_setinfilename(svfile); -} + /* + * Clean up and restore saved parent properties. + */ -/* - * Read the whole file into memory and call the parsers. - * Called recursively when an .so request is encountered. - */ -void -mparse_readfd(struct mparse *curp, int fd, const char *file) -{ - struct buf blk; - int with_mmap; - int save_filenc; + if (with_mmap) + munmap(blk.buf, blk.sz); + else + free(blk.buf); - if (read_whole_file(curp, fd, &blk, &with_mmap)) { - save_filenc = curp->filenc; - curp->filenc = curp->options & - (MPARSE_UTF8 | MPARSE_LATIN1); - mparse_parse_buffer(curp, blk, file); - curp->filenc = save_filenc; - if (with_mmap) - munmap(blk.buf, blk.sz); - else - free(blk.buf); - } + curp->primary = save_primary; + curp->filenc = save_filenc; + curp->line = save_lineno; + if (save_filename != NULL) + mandoc_msg_setinfilename(save_filename); } int @@ -634,7 +634,6 @@ mparse_open(struct mparse *curp, const char *file) /* Neither worked, give up. */ - mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno)); return -1; } @@ -648,19 +647,19 @@ mparse_alloc(int options, enum mandoc_os os_e, const c curp->options = options; curp->os_s = os_s; - curp->roff = roff_alloc(curp, options); - curp->man = roff_man_alloc(curp->roff, curp, curp->os_s, + curp->roff = roff_alloc(options); + curp->man = roff_man_alloc(curp->roff, curp->os_s, curp->options & MPARSE_QUICK ? 1 : 0); if (curp->options & MPARSE_MDOC) { - curp->man->macroset = MACROSET_MDOC; + curp->man->meta.macroset = MACROSET_MDOC; if (curp->man->mdocmac == NULL) curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX); } else if (curp->options & MPARSE_MAN) { - curp->man->macroset = MACROSET_MAN; + curp->man->meta.macroset = MACROSET_MAN; if (curp->man->manmac == NULL) curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX); } - curp->man->first->tok = TOKEN_NONE; + curp->man->meta.first->tok = TOKEN_NONE; curp->man->meta.os_e = os_e; return curp; } @@ -672,8 +671,6 @@ mparse_reset(struct mparse *curp) roff_man_reset(curp->man); free_buf_list(curp->secondary); curp->secondary = NULL; - free(curp->sodest); - curp->sodest = NULL; curp->gzip = 0; } @@ -685,21 +682,20 @@ mparse_free(struct mparse *curp) roff_man_free(curp->man); roff_free(curp->roff); free_buf_list(curp->secondary); - free(curp->sodest); free(curp); } -void -mparse_result(struct mparse *curp, struct roff_man **man, - char **sodest) +struct roff_meta * +mparse_result(struct mparse *curp) { - - if (sodest && NULL != (*sodest = curp->sodest)) { - *man = NULL; - return; + roff_state_reset(curp->man); + if (curp->options & MPARSE_VALIDATE) { + if (curp->man->meta.macroset == MACROSET_MDOC) + mdoc_validate(curp->man); + else + man_validate(curp->man); } - if (man) - *man = curp->man; + return &curp->man->meta; } void