=================================================================== RCS file: /cvs/mandoc/read.c,v retrieving revision 1.157 retrieving revision 1.210 diff -u -p -r1.157 -r1.210 --- mandoc/read.c 2017/01/09 01:37:03 1.157 +++ mandoc/read.c 2018/12/31 04:55:47 1.210 @@ -1,7 +1,7 @@ -/* $Id: read.c,v 1.157 2017/01/09 01:37:03 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) 2010-2017 Ingo Schwarze + * Copyright (c) 2010-2018 Ingo Schwarze * Copyright (c) 2010, 2012 Joerg Sonnenberger * * Permission to use, copy, modify, and distribute this software for any @@ -24,13 +24,9 @@ #include #include -#if HAVE_ERR -#include -#endif #include #include #include -#include #include #include #include @@ -42,22 +38,19 @@ #include "roff.h" #include "mdoc.h" #include "man.h" +#include "mandoc_parse.h" #include "libmandoc.h" #include "roff_int.h" #define REPARSE_LIMIT 1000 struct mparse { - struct roff_man *man; /* man parser */ struct roff *roff; /* roff parser (!NULL) */ - char *sodest; /* filename pointed to by .so */ - const char *file; /* filename of current input file */ + struct roff_man *man; /* man parser */ struct buf *primary; /* buffer currently being parsed */ - struct buf *secondary; /* preprocessed copy of input */ - const char *defos; /* default operating system */ - mandocmsg mmsg; /* warning/error message handler */ - enum mandoclevel file_status; /* status of current parse */ - enum mandoclevel wlevel; /* ignore messages below this */ + struct buf *secondary; /* copy of top level input */ + struct buf *loop; /* open .while request line */ + const char *os_s; /* default operating system */ int options; /* parser options */ int gzip; /* current input file is gzipped */ int filenc; /* encoding of the current file */ @@ -66,189 +59,13 @@ struct mparse { }; static void choose_parser(struct mparse *); +static void free_buf_list(struct buf *); static void resize_buf(struct buf *, size_t); -static void mparse_buf_r(struct mparse *, struct buf, size_t, int); -static int read_whole_file(struct mparse *, const char *, int, - struct buf *, int *); +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 const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = { - MANDOCERR_OK, - MANDOCERR_WARNING, - MANDOCERR_WARNING, - MANDOCERR_ERROR, - MANDOCERR_UNSUPP, - MANDOCERR_MAX, - MANDOCERR_MAX -}; -static const char * const mandocerrs[MANDOCERR_MAX] = { - "ok", - - "generic warning", - - /* related to the prologue */ - "missing manual title, using UNTITLED", - "missing manual title, using \"\"", - "lower case character in document title", - "missing manual section, using \"\"", - "unknown manual section", - "missing date, using today's date", - "cannot parse date, using it verbatim", - "missing Os macro, using \"\"", - "duplicate prologue macro", - "late prologue macro", - "skipping late title macro", - "prologue macros out of order", - - /* related to document structure */ - ".so is fragile, better use ln(1)", - "no document body", - "content before first section header", - "first section is not \"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 \"\"", - "sections out of conventional order", - "duplicate section title", - "unexpected section", - "unusual Xr order", - "unusual Xr punctuation", - "AUTHORS section without An macro", - - /* related to macros and nesting */ - "obsolete macro", - "macro neither callable nor escaped", - "skipping paragraph macro", - "moving paragraph macro out of list", - "skipping no-space macro", - "blocks badly nested", - "nested displays are not portable", - "moving content out of list", - "fill mode already enabled, skipping", - "fill mode already disabled, skipping", - "line scope broken", - - /* related to missing macro arguments */ - "skipping empty request", - "conditional request controls empty scope", - "skipping empty macro", - "empty block", - "empty argument, using 0n", - "missing display type, using -ragged", - "list type is not the first argument", - "missing -width in -tag list, using 6n", - "missing utility name, using \"\"", - "missing function name, using \"\"", - "empty head in list item", - "empty list item", - "missing font type, using \\fR", - "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 \"\"", - "missing eqn box, using \"\"", - - /* related to bad macro arguments */ - "unterminated quoted argument", - "duplicate argument", - "skipping duplicate argument", - "skipping duplicate display type", - "skipping duplicate list type", - "skipping -width argument", - "wrong number of cells", - "unknown AT&T UNIX version", - "comma in function argument", - "parenthesis in function name", - "invalid content in Rs block", - "invalid Boolean argument", - "unknown font, skipping request", - "odd number of characters in request", - - /* related to plain text */ - "blank line in fill mode, using .sp", - "tab in filled text", - "whitespace at end of input line", - "bad comment style", - "invalid escape sequence", - "undefined string, using \"\"", - - /* related to tables */ - "tbl line starts with span", - "tbl column starts with span", - "skipping vertical bar in tbl layout", - - "generic error", - - /* related to tables */ - "non-alphabetic character in tbl options", - "skipping unknown tbl option", - "missing tbl option argument", - "wrong tbl option argument size", - "empty tbl layout", - "invalid character in tbl layout", - "unmatched parenthesis in tbl layout", - "tbl without any data cells", - "ignoring data in spanned tbl cell", - "ignoring extra tbl data cells", - "data block open at end of tbl", - - /* related to document structure and macros */ - NULL, - "input stack limit exceeded, infinite loop?", - "skipping bad character", - "skipping unknown macro", - "skipping insecure request", - "skipping item outside list", - "skipping column outside column list", - "skipping end of block that is not open", - "fewer RS blocks open, skipping", - "inserting missing end of block", - "appending missing end of block", - - /* 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", - "unknown standard specifier", - "skipping request without numeric argument", - "NOT IMPLEMENTED: .so with absolute path or \"..\"", - ".so request failed", - "skipping all arguments", - "skipping excess arguments", - "divide by zero", - - "unsupported feature", - "input too large", - "unsupported control character", - "unsupported roff request", - "eqn delim option in tbl", - "unsupported tbl layout modifier", - "ignoring macro in table", -}; - -static const char * const mandoclevels[MANDOCLEVEL_MAX] = { - "SUCCESS", - "RESERVED", - "WARNING", - "ERROR", - "UNSUPP", - "BADARG", - "SYSERR" -}; - - static void resize_buf(struct buf *buf, size_t initial) { @@ -258,6 +75,19 @@ resize_buf(struct buf *buf, size_t initial) } static void +free_buf_list(struct buf *buf) +{ + struct buf *tmp; + + while (buf != NULL) { + tmp = buf; + buf = tmp->next; + free(tmp->buf); + free(tmp); + } +} + +static void choose_parser(struct mparse *curp) { char *cp, *ep; @@ -292,14 +122,15 @@ choose_parser(struct mparse *curp) } if (format == MPARSE_MDOC) { - mdoc_hash_init(); - curp->man->macroset = MACROSET_MDOC; - curp->man->first->tok = TOKEN_NONE; + curp->man->meta.macroset = MACROSET_MDOC; + 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; + curp->man->meta.macroset = MACROSET_MAN; + if (curp->man->manmac == NULL) + curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX); } + curp->man->meta.first->tok = TOKEN_NONE; } /* @@ -309,30 +140,30 @@ 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; struct buf ln; - const char *save_file; + struct buf *firstln, *lastln, *thisln, *loop; 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 line_result, result; int of; int lnn; /* line number in the real file */ int fd; + int inloop; /* Saw .while on this level. */ unsigned char c; - memset(&ln, 0, sizeof(ln)); - + ln.sz = 256; + ln.buf = mandoc_malloc(ln.sz); + ln.next = NULL; + firstln = loop = NULL; lnn = curp->line; pos = 0; + inloop = 0; + result = ROFF_CONT; - while (i < blk.sz) { - if (0 == pos && '\0' == blk.buf[i]) - break; - + while (i < blk.sz && (blk.buf[i] != '\0' || pos != 0)) { if (start) { curp->line = lnn; curp->reparse_count = 0; @@ -362,10 +193,10 @@ mparse_buf_r(struct mparse *curp, struct buf blk, size /* * Make sure we have space for the worst - * case of 11 bytes: "\\[u10ffff]\0" + * case of 12 bytes: "\\[u10ffff]\n\0" */ - if (pos + 11 > ln.sz) + if (pos + 12 > ln.sz) resize_buf(&ln, 256); /* @@ -376,7 +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++; @@ -389,91 +220,45 @@ 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++] = '?'; continue; } - /* Trailing backslash = a plain char. */ + ln.buf[pos++] = blk.buf[i++]; + } + ln.buf[pos] = '\0'; - if (blk.buf[i] != '\\' || i + 1 == blk.sz) { - ln.buf[pos++] = blk.buf[i++]; - continue; - } + /* + * Maintain a lookaside buffer of all lines. + * parsed from this input source. + */ - /* - * Found escape and at least one other character. - * When it's a newline character, skip it. - * When there is a carriage return in between, - * skip that one as well. - */ + thisln = mandoc_malloc(sizeof(*thisln)); + thisln->buf = mandoc_strdup(ln.buf); + thisln->sz = strlen(ln.buf) + 1; + thisln->next = NULL; + if (firstln == NULL) { + firstln = lastln = thisln; + if (curp->secondary == NULL) + curp->secondary = firstln; + } else { + lastln->next = thisln; + lastln = thisln; + } - if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz && - '\n' == blk.buf[i + 2]) - ++i; - if ('\n' == blk.buf[i + 1]) { - i += 2; - ++lnn; - continue; - } + /* XXX Ugly hack to mark the end of the input. */ - 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 (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 */ - for (; pos > 0; --pos) { - if (ln.buf[pos - 1] != ' ') - break; - if (pos > 2 && ln.buf[pos - 2] == '\\') - break; - } - break; - } - - /* Catch escaped bogus characters. */ - - c = (unsigned char) blk.buf[i+1]; - - if ( ! (isascii(c) && - (isgraph(c) || isblank(c)))) { - mandoc_vmsg(MANDOCERR_CHAR_BAD, curp, - curp->line, pos, "0x%x", c); - i += 2; - ln.buf[pos++] = '?'; - continue; - } - - /* Some other escape sequence, copy & cont. */ - - ln.buf[pos++] = blk.buf[i++]; - ln.buf[pos++] = blk.buf[i++]; + if (i == blk.sz || blk.buf[i] == '\0') { + ln.buf[pos++] = '\n'; + ln.buf[pos] = '\0'; } - if (pos >= ln.sz) - resize_buf(&ln, 256); - - ln.buf[pos] = '\0'; - /* * A significant amount of complexity is contained by * the roff preprocessor. It's line-oriented but can be @@ -484,71 +269,112 @@ mparse_buf_r(struct mparse *curp, struct buf blk, size */ of = 0; +rerun: + line_result = roff_parseln(curp->roff, curp->line, &ln, &of); - /* - * Maintain a lookaside buffer of all parsed lines. We - * only do this if mparse_keep() has been invoked (the - * buffer may be accessed with mparse_getkeep()). - */ + /* Process options. */ - if (curp->secondary) { - curp->secondary->buf = mandoc_realloc( - curp->secondary->buf, - curp->secondary->sz + pos + 2); - memcpy(curp->secondary->buf + - curp->secondary->sz, - ln.buf, pos); - curp->secondary->sz += pos; - curp->secondary->buf - [curp->secondary->sz] = '\n'; - curp->secondary->sz++; - curp->secondary->buf - [curp->secondary->sz] = '\0'; + if (line_result & ROFF_APPEND) + assert(line_result == (ROFF_IGN | ROFF_APPEND)); + + if (line_result & ROFF_USERCALL) + assert((line_result & ROFF_MASK) == ROFF_REPARSE); + + if (line_result & ROFF_USERRET) { + assert(line_result == (ROFF_IGN | ROFF_USERRET)); + if (start == 0) { + /* Return from the current macro. */ + result = ROFF_USERRET; + goto out; + } } -rerun: - rr = roff_parseln(curp->roff, curp->line, &ln, &of); - switch (rr) { - case ROFF_REPARSE: - if (REPARSE_LIMIT >= ++curp->reparse_count) - mparse_buf_r(curp, ln, of, 0); - else - mandoc_msg(MANDOCERR_ROFFLOOP, curp, + switch (line_result & ROFF_LOOPMASK) { + case ROFF_IGN: + break; + case ROFF_WHILE: + if (curp->loop != NULL) { + if (loop == curp->loop) + break; + mandoc_msg(MANDOCERR_WHILE_NEST, curp->line, pos, NULL); - pos = 0; - continue; - case ROFF_APPEND: - pos = strlen(ln.buf); - continue; + } + curp->loop = thisln; + loop = NULL; + inloop = 1; + break; + case ROFF_LOOPCONT: + case ROFF_LOOPEXIT: + if (curp->loop == NULL) { + mandoc_msg(MANDOCERR_WHILE_FAIL, + curp->line, pos, NULL); + break; + } + if (inloop == 0) { + mandoc_msg(MANDOCERR_WHILE_INTO, + curp->line, pos, NULL); + curp->loop = loop = NULL; + break; + } + if (line_result & ROFF_LOOPCONT) + loop = curp->loop; + else { + curp->loop = loop = NULL; + inloop = 0; + } + break; + default: + abort(); + } + + /* Process the main instruction from the roff parser. */ + + switch (line_result & ROFF_MASK) { + case ROFF_IGN: + break; + case ROFF_CONT: + if (curp->man->meta.macroset == MACROSET_NONE) + choose_parser(curp); + 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) + goto out; + break; case ROFF_RERUN: goto rerun; - case ROFF_IGN: - pos = 0; - continue; + case ROFF_REPARSE: + if (++curp->reparse_count > REPARSE_LIMIT) { + /* Abort and return to the top level. */ + result = ROFF_IGN; + mandoc_msg(MANDOCERR_ROFFLOOP, + curp->line, pos, NULL); + goto out; + } + result = mparse_buf_r(curp, ln, of, 0); + if (line_result & ROFF_USERCALL) { + roff_userret(curp->roff); + /* Continue normally. */ + if (result & ROFF_USERRET) + result = ROFF_CONT; + } + if (start == 0 && result != ROFF_CONT) + goto out; + break; case ROFF_SO: if ( ! (curp->options & MPARSE_SO) && (i >= blk.sz || blk.buf[i] == '\0')) { - curp->sodest = mandoc_strdup(ln.buf + of); - free(ln.buf); - return; + curp->man->meta.sodest = + mandoc_strdup(ln.buf + of); + goto out; } - /* - * We remove `so' clauses from our lookaside - * buffer because we're going to descend into - * the file recursively. - */ - if (curp->secondary) - curp->secondary->sz -= pos + 1; - save_file = curp->file; 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; - 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); @@ -557,60 +383,57 @@ rerun: of = 0; mparse_buf_r(curp, ln, of, 0); } - pos = 0; - continue; - default: break; + default: + abort(); } - if (curp->man->macroset == MACROSET_NONE) - choose_parser(curp); + /* Start the next input line. */ - /* - * Lastly, push down into the parsers themselves. - * If libroff returns ROFF_TBL, then add it to the - * currently open parse. Since we only get here if - * there does exist data (see tbl_data.c), we're - * guaranteed that something's been allocated. - * Do the same for ROFF_EQN. - */ + if (loop != NULL && + (line_result & ROFF_LOOPMASK) == ROFF_IGN) + loop = loop->next; - if (rr == ROFF_TBL) - while ((span = roff_span(curp->roff)) != NULL) - roff_addtbl(curp->man, span); - else if (rr == ROFF_EQN) - roff_addeqn(curp->man, roff_eqn(curp->roff)); - else if ((curp->man->macroset == MACROSET_MDOC ? - mdoc_parseln(curp->man, curp->line, ln.buf, of) : - man_parseln(curp->man, curp->line, ln.buf, of)) == 2) - break; + if (loop != NULL) { + if ((line_result & ROFF_APPEND) == 0) + *ln.buf = '\0'; + if (ln.sz < loop->sz) + resize_buf(&ln, loop->sz); + (void)strlcat(ln.buf, loop->buf, ln.sz); + of = 0; + goto rerun; + } - /* Temporary buffers typically are not full. */ - - if (0 == start && '\0' == blk.buf[i]) - break; - - /* Start the next input line. */ - - pos = 0; + pos = (line_result & ROFF_APPEND) ? strlen(ln.buf) : 0; } - +out: + if (inloop) { + if (result != ROFF_USERRET) + mandoc_msg(MANDOCERR_WHILE_OUTOF, + curp->line, pos, NULL); + curp->loop = NULL; + } free(ln.buf); + if (firstln != curp->secondary) + free_buf_list(firstln); + return result; } static int -read_whole_file(struct mparse *curp, const char *file, int fd, - struct buf *fb, int *with_mmap) +read_whole_file(struct mparse *curp, int fd, struct buf *fb, int *with_mmap) { + struct stat st; gzFile gz; size_t off; ssize_t ssz; + int gzerrnum, retval; - struct stat st; + if (fstat(fd, &st) == -1) { + mandoc_msg(MANDOCERR_FILE, 0, 0, + "fstat: %s", strerror(errno)); + return 0; + } - if (fstat(fd, &st) == -1) - err((int)MANDOCLEVEL_SYSERR, "%s", file); - /* * If we're a regular file, try just reading in the whole entry * via mmap(). This is faster than reading it into blocks, and @@ -620,7 +443,7 @@ 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); + mandoc_msg(MANDOCERR_TOOLARGE, 0, 0, NULL); return 0; } *with_mmap = 1; @@ -631,8 +454,24 @@ read_whole_file(struct mparse *curp, const char *file, } if (curp->gzip) { - if ((gz = gzdopen(fd, "rb")) == NULL) - err((int)MANDOCLEVEL_SYSERR, "%s", file); + /* + * Duplicating the file descriptor is required + * because we will have to call gzclose(3) + * to free memory used internally by zlib, + * but that will also close the file descriptor, + * which this function must not do. + */ + if ((fd = dup(fd)) == -1) { + mandoc_msg(MANDOCERR_FILE, 0, 0, + "dup: %s", strerror(errno)); + return 0; + } + if ((gz = gzdopen(fd, "rb")) == NULL) { + mandoc_msg(MANDOCERR_FILE, 0, 0, + "gzdopen: %s", strerror(errno)); + close(fd); + return 0; + } } else gz = NULL; @@ -643,13 +482,13 @@ read_whole_file(struct mparse *curp, const char *file, *with_mmap = 0; off = 0; + retval = 0; fb->sz = 0; fb->buf = NULL; 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); @@ -659,50 +498,79 @@ 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; + retval = 1; + break; } - if (ssz == -1) - err((int)MANDOCLEVEL_SYSERR, "%s", file); + if (ssz == -1) { + if (curp->gzip) + (void)gzerror(gz, &gzerrnum); + mandoc_msg(MANDOCERR_FILE, 0, 0, "read: %s", + curp->gzip && gzerrnum != Z_ERRNO ? + zError(gzerrnum) : strerror(errno)); + break; + } off += (size_t)ssz; } - free(fb->buf); - fb->buf = NULL; - return 0; + if (curp->gzip && (gzerrnum = gzclose(gz)) != Z_OK) + mandoc_msg(MANDOCERR_FILE, 0, 0, "gzclose: %s", + gzerrnum == Z_ERRNO ? strerror(errno) : + zError(gzerrnum)); + if (retval == 0) { + free(fb->buf); + fb->buf = NULL; + } + return retval; } 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 = curp->file; - curp->file = 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 && @@ -714,60 +582,33 @@ 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; - curp->file = svfile; -} + /* + * Clean up and restore saved parent properties. + */ -enum mandoclevel -mparse_readmem(struct mparse *curp, void *buf, size_t len, - const char *file) -{ - struct buf blk; + if (with_mmap) + munmap(blk.buf, blk.sz); + else + free(blk.buf); - blk.buf = buf; - blk.sz = len; - - mparse_parse_buffer(curp, blk, file); - return curp->file_status; + curp->primary = save_primary; + curp->filenc = save_filenc; + curp->line = save_lineno; + if (save_filename != NULL) + mandoc_msg_setinfilename(save_filename); } -/* - * Read the whole file into memory and call the parsers. - * Called recursively when an .so request is encountered. - */ -enum mandoclevel -mparse_readfd(struct mparse *curp, int fd, const char *file) -{ - struct buf blk; - int with_mmap; - int save_filenc; - - if (read_whole_file(curp, file, 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); - } - return curp->file_status; -} - int mparse_open(struct mparse *curp, const char *file) { char *cp; int fd; - curp->file = file; cp = strrchr(file, '.'); curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz")); @@ -793,34 +634,33 @@ mparse_open(struct mparse *curp, const char *file) /* Neither worked, give up. */ - mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno)); return -1; } struct mparse * -mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg, - const char *defos) +mparse_alloc(int options, enum mandoc_os os_e, const char *os_s) { struct mparse *curp; curp = mandoc_calloc(1, sizeof(struct mparse)); curp->options = options; - curp->wlevel = wlevel; - curp->mmsg = mmsg; - curp->defos = defos; + curp->os_s = os_s; - curp->roff = roff_alloc(curp, options); - curp->man = roff_man_alloc( curp->roff, curp, curp->defos, + 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) { - mdoc_hash_init(); - 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) { - man_hash_init(); - 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; } @@ -829,109 +669,40 @@ mparse_reset(struct mparse *curp) { roff_reset(curp->roff); roff_man_reset(curp->man); - if (curp->secondary) - curp->secondary->sz = 0; - - curp->file_status = MANDOCLEVEL_OK; - - free(curp->sodest); - curp->sodest = NULL; + free_buf_list(curp->secondary); + curp->secondary = 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); - if (curp->secondary) - free(curp->secondary->buf); - - free(curp->secondary); - free(curp->sodest); + roff_free(curp->roff); + free_buf_list(curp->secondary); 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 -mparse_updaterc(struct mparse *curp, enum mandoclevel *rc) +mparse_copy(const struct mparse *p) { - if (curp->file_status > *rc) - *rc = curp->file_status; -} + struct buf *buf; -void -mandoc_vmsg(enum mandocerr t, struct mparse *m, - int ln, int pos, const char *fmt, ...) -{ - char buf[256]; - va_list ap; - - va_start(ap, fmt); - (void)vsnprintf(buf, sizeof(buf), fmt, ap); - va_end(ap); - - mandoc_msg(t, m, ln, pos, buf); -} - -void -mandoc_msg(enum mandocerr er, struct mparse *m, - int ln, int col, const char *msg) -{ - enum mandoclevel level; - - level = MANDOCLEVEL_UNSUPP; - while (er < mandoclimits[level]) - level--; - - if (level < m->wlevel && er != MANDOCERR_FILE) - return; - - if (m->mmsg) - (*m->mmsg)(er, level, m->file, ln, col, msg); - - if (m->file_status < level) - m->file_status = level; -} - -const char * -mparse_strerror(enum mandocerr er) -{ - - return mandocerrs[er]; -} - -const char * -mparse_strlevel(enum mandoclevel lvl) -{ - return mandoclevels[lvl]; -} - -void -mparse_keep(struct mparse *p) -{ - - assert(NULL == p->secondary); - p->secondary = mandoc_calloc(1, sizeof(struct buf)); -} - -const char * -mparse_getkeep(const struct mparse *p) -{ - - assert(p->secondary); - return p->secondary->sz ? p->secondary->buf : NULL; + for (buf = p->secondary; buf != NULL; buf = buf->next) + puts(buf->buf); }