]> git.cameronkatri.com Git - mandoc.git/blobdiff - read.c
git.ckatri.com / mandoc.git / blobdiff
? search:
summary | shortlog | log | commit | commitdiff | tree
raw | inline | side by side
libmdoc.h no longer needs mdoc.h
[mandoc.git] / read.c
diff --git a/read.c b/read.c
index 96c7ac53b21ec626bb35ac01a5fd2cca1bf47723..37862e38e9f48847caadb34a6019b76c6fe2f32e 100644 (file)
--- a/read.c
+++ b/read.c
@@ -1,7 +1,7 @@
-/*     $Id: read.c,v 1.145 2015/10/30 19:04:16 schwarze Exp $ */
+/*     $Id: read.c,v 1.201 2018/12/13 06:18:20 schwarze Exp $ */
 /*
  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
 /*
  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
- * Copyright (c) 2010-2015 Ingo Schwarze <schwarze@openbsd.org>
+ * Copyright (c) 2010-2018 Ingo Schwarze <schwarze@openbsd.org>
  * Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
  *
  * Permission to use, copy, modify, and distribute this software for any
  * Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
  *
  * Permission to use, copy, modify, and distribute this software for any
@@ -19,18 +19,14 @@
 #include "config.h"
 
 #include <sys/types.h>
 #include "config.h"
 
 #include <sys/types.h>
-#if HAVE_MMAP
 #include <sys/mman.h>
 #include <sys/stat.h>
 #include <sys/mman.h>
 #include <sys/stat.h>
-#endif
 
 #include <assert.h>
 #include <ctype.h>
 
 #include <assert.h>
 #include <ctype.h>
-#include <err.h>
 #include <errno.h>
 #include <fcntl.h>
 #include <stdarg.h>
 #include <errno.h>
 #include <fcntl.h>
 #include <stdarg.h>
-#include <stdint.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
@@ -48,16 +44,17 @@
 #define        REPARSE_LIMIT   1000
 
 struct mparse {
 #define        REPARSE_LIMIT   1000
 
 struct mparse {
-       struct roff_man  *man; /* man parser */
        struct roff      *roff; /* roff parser (!NULL) */
        struct roff      *roff; /* roff parser (!NULL) */
+       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 */
        char             *sodest; /* filename pointed to by .so */
        const char       *file; /* filename of current input file */
        struct buf       *primary; /* buffer currently being parsed */
-       struct buf       *secondary; /* preprocessed copy of input */
-       const char       *defos; /* default operating system */
+       struct buf       *secondary; /* copy of top level input */
+       struct buf       *loop; /* open .while request line */
+       const char       *os_s; /* default operating system */
        mandocmsg         mmsg; /* warning/error message handler */
        enum mandoclevel  file_status; /* status of current parse */
        mandocmsg         mmsg; /* warning/error message handler */
        enum mandoclevel  file_status; /* status of current parse */
-       enum mandoclevel  wlevel; /* ignore messages below this */
+       enum mandocerr    mmin; /* ignore messages below this */
        int               options; /* parser options */
        int               gzip; /* current input file is gzipped */
        int               filenc; /* encoding of the current file */
        int               options; /* parser options */
        int               gzip; /* current input file is gzipped */
        int               filenc; /* encoding of the current file */
@@ -66,8 +63,9 @@ struct        mparse {
 };
 
 static void      choose_parser(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      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 *);
 static int       read_whole_file(struct mparse *, const char *, int,
                                struct buf *, int *);
 static void      mparse_end(struct mparse *);
@@ -76,7 +74,7 @@ static        void      mparse_parse_buffer(struct mparse *, struct buf,
 
 static const enum mandocerr    mandoclimits[MANDOCLEVEL_MAX] = {
        MANDOCERR_OK,
 
 static const enum mandocerr    mandoclimits[MANDOCLEVEL_MAX] = {
        MANDOCERR_OK,
-       MANDOCERR_WARNING,
+       MANDOCERR_OK,
        MANDOCERR_WARNING,
        MANDOCERR_ERROR,
        MANDOCERR_UNSUPP,
        MANDOCERR_WARNING,
        MANDOCERR_ERROR,
        MANDOCERR_UNSUPP,
@@ -87,20 +85,48 @@ static      const enum mandocerr    mandoclimits[MANDOCLEVEL_MAX] = {
 static const char * const      mandocerrs[MANDOCERR_MAX] = {
        "ok",
 
 static const char * const      mandocerrs[MANDOCERR_MAX] = {
        "ok",
 
+       "base system convention",
+
+       "Mdocdate found",
+       "Mdocdate missing",
+       "unknown architecture",
+       "operating system explicitly specified",
+       "RCS id missing",
+       "referenced manual not found",
+
+       "generic style suggestion",
+
+       "legacy man(7) date format",
+       "normalizing date format to",
+       "lower case character in document title",
+       "duplicate RCS id",
+       "possible typo in section name",
+       "unterminated quoted argument",
+       "useless macro",
+       "consider using OS macro",
+       "errnos out of order",
+       "duplicate errno",
+       "trailing delimiter",
+       "no blank before trailing delimiter",
+       "fill mode already enabled, skipping",
+       "fill mode already disabled, skipping",
+       "verbatim \"--\", maybe consider using \\(em",
+       "function name without markup",
+       "whitespace at end of input line",
+       "bad comment style",
+
        "generic warning",
 
        /* related to the prologue */
        "missing manual title, using UNTITLED",
        "missing manual title, using \"\"",
        "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 manual section, using \"\"",
        "unknown manual section",
        "missing date, using today's date",
        "cannot parse date, using it verbatim",
+       "date in the future, using it anyway",
        "missing Os macro, using \"\"",
        "missing Os macro, using \"\"",
-       "duplicate prologue macro",
        "late prologue macro",
        "late prologue macro",
-       "skipping late title macro",
        "prologue macros out of order",
 
        /* related to document structure */
        "prologue macros out of order",
 
        /* related to document structure */
@@ -108,14 +134,17 @@ static    const char * const      mandocerrs[MANDOCERR_MAX] = {
        "no document body",
        "content before first section header",
        "first section is not \"NAME\"",
        "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",
        "NAME section without description",
        "description not at the end of NAME",
        "bad NAME section content",
+       "missing comma before name",
        "missing description line, using \"\"",
        "missing description line, using \"\"",
+       "description line outside NAME section",
        "sections out of conventional order",
        "duplicate section title",
        "unexpected section",
        "sections out of conventional order",
        "duplicate section title",
        "unexpected section",
+       "cross reference to self",
        "unusual Xr order",
        "unusual Xr punctuation",
        "AUTHORS section without An macro",
        "unusual Xr order",
        "unusual Xr punctuation",
        "AUTHORS section without An macro",
@@ -129,9 +158,9 @@ static      const char * const      mandocerrs[MANDOCERR_MAX] = {
        "blocks badly nested",
        "nested displays are not portable",
        "moving content out of list",
        "blocks badly nested",
        "nested displays are not portable",
        "moving content out of list",
-       "fill mode already enabled, skipping",
-       "fill mode already disabled, skipping",
+       "first macro on line",
        "line scope broken",
        "line scope broken",
+       "skipping blank line in line scope",
 
        /* related to missing macro arguments */
        "skipping empty request",
 
        /* related to missing macro arguments */
        "skipping empty request",
@@ -141,22 +170,23 @@ static    const char * const      mandocerrs[MANDOCERR_MAX] = {
        "empty argument, using 0n",
        "missing display type, using -ragged",
        "list type is not the first argument",
        "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",
        "empty list item",
        "missing utility name, using \"\"",
        "missing function name, using \"\"",
        "empty head in list item",
        "empty list item",
+       "missing argument, using next line",
        "missing font type, using \\fR",
        "unknown font type, using \\fR",
        "nothing follows prefix",
        "empty reference block",
        "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 */
        "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",
        "duplicate argument",
        "skipping duplicate argument",
        "skipping duplicate display type",
@@ -166,16 +196,17 @@ static    const char * const      mandocerrs[MANDOCERR_MAX] = {
        "unknown AT&T UNIX version",
        "comma in function argument",
        "parenthesis in function name",
        "unknown AT&T UNIX version",
        "comma in function argument",
        "parenthesis in function name",
+       "unknown library name",
        "invalid content in Rs block",
        "invalid Boolean argument",
        "invalid content in Rs block",
        "invalid Boolean argument",
+       "argument contains two font escapes",
        "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",
        "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",
+       "new sentence, new line",
        "invalid escape sequence",
        "undefined string, using \"\"",
 
        "invalid escape sequence",
        "undefined string, using \"\"",
 
@@ -201,9 +232,12 @@ static     const char * const      mandocerrs[MANDOCERR_MAX] = {
 
        /* related to document structure and macros */
        NULL,
 
        /* related to document structure and macros */
        NULL,
+       "duplicate prologue macro",
+       "skipping late title macro",
        "input stack limit exceeded, infinite loop?",
        "skipping bad character",
        "skipping unknown macro",
        "input stack limit exceeded, infinite loop?",
        "skipping bad character",
        "skipping unknown macro",
+       "ignoring request outside macro",
        "skipping insecure request",
        "skipping item outside list",
        "skipping column outside column list",
        "skipping insecure request",
        "skipping item outside list",
        "skipping column outside column list",
@@ -214,13 +248,18 @@ static    const char * const      mandocerrs[MANDOCERR_MAX] = {
 
        /* related to request and macro arguments */
        "escaped character not allowed in a name",
 
        /* related to request and macro arguments */
        "escaped character not allowed in a name",
+       "using macro argument outside macro",
+       "argument number is not numeric",
        "NOT IMPLEMENTED: Bd -file",
        "skipping display without arguments",
        "missing list type, using -item",
        "NOT IMPLEMENTED: Bd -file",
        "skipping display without arguments",
        "missing list type, using -item",
+       "argument is not numeric, using 1",
+       "argument is not a character",
        "missing manual name, using \"\"",
        "uname(3) system call failed, using UNKNOWN",
        "unknown standard specifier",
        "skipping request without numeric argument",
        "missing manual name, using \"\"",
        "uname(3) system call failed, using UNKNOWN",
        "unknown standard specifier",
        "skipping request without numeric argument",
+       "excessive shift",
        "NOT IMPLEMENTED: .so with absolute path or \"..\"",
        ".so request failed",
        "skipping all arguments",
        "NOT IMPLEMENTED: .so with absolute path or \"..\"",
        ".so request failed",
        "skipping all arguments",
@@ -231,6 +270,10 @@ static     const char * const      mandocerrs[MANDOCERR_MAX] = {
        "input too large",
        "unsupported control character",
        "unsupported roff request",
        "input too large",
        "unsupported control character",
        "unsupported roff request",
+       "nested .while loops",
+       "end of scope with open .while loop",
+       "end of .while loop in inner scope",
+       "cannot continue this .while loop",
        "eqn delim option in tbl",
        "unsupported tbl layout modifier",
        "ignoring macro in table",
        "eqn delim option in tbl",
        "unsupported tbl layout modifier",
        "ignoring macro in table",
@@ -238,7 +281,7 @@ static      const char * const      mandocerrs[MANDOCERR_MAX] = {
 
 static const char * const      mandoclevels[MANDOCLEVEL_MAX] = {
        "SUCCESS",
 
 static const char * const      mandoclevels[MANDOCLEVEL_MAX] = {
        "SUCCESS",
-       "RESERVED",
+       "STYLE",
        "WARNING",
        "ERROR",
        "UNSUPP",
        "WARNING",
        "ERROR",
        "UNSUPP",
@@ -255,6 +298,19 @@ resize_buf(struct buf *buf, size_t initial)
        buf->buf = mandoc_realloc(buf->buf, buf->sz);
 }
 
        buf->buf = mandoc_realloc(buf->buf, buf->sz);
 }
 
+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)
 {
 static void
 choose_parser(struct mparse *curp)
 {
@@ -289,22 +345,16 @@ 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) {
        if (format == MPARSE_MDOC) {
-               mdoc_hash_init();
                curp->man->macroset = MACROSET_MDOC;
                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 {
        } else {
-               man_hash_init();
                curp->man->macroset = MACROSET_MAN;
                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;
 }
 
 /*
 }
 
 /*
@@ -314,29 +364,31 @@ choose_parser(struct mparse *curp)
  * macros, inline equations, and input line traps)
  * and indirectly (for .so file inclusion).
  */
  * 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)
 {
 mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
 {
-       const struct tbl_span   *span;
        struct buf       ln;
        struct buf       ln;
+       struct buf      *firstln, *lastln, *thisln, *loop;
        const char      *save_file;
        char            *cp;
        size_t           pos; /* byte number in the ln buffer */
        const char      *save_file;
        char            *cp;
        size_t           pos; /* byte number in the ln buffer */
-       enum rofferr     rr;
+       int              line_result, result;
        int              of;
        int              lnn; /* line number in the real file */
        int              fd;
        int              of;
        int              lnn; /* line number in the real file */
        int              fd;
+       int              inloop; /* Saw .while on this level. */
        unsigned char    c;
 
        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;
        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;
                if (start) {
                        curp->line = lnn;
                        curp->reparse_count = 0;
@@ -366,10 +418,10 @@ mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
 
                        /*
                         * Make sure we have space for the worst
 
                        /*
                         * 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);
 
                        /*
                                resize_buf(&ln, 256);
 
                        /*
@@ -403,73 +455,34 @@ mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
                                continue;
                        }
 
                                continue;
                        }
 
-                       /* Trailing backslash = a plain char. */
-
-                       if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
-                               ln.buf[pos++] = blk.buf[i++];
-                               continue;
-                       }
-
-                       /*
-                        * 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.
-                        */
-
-                       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;
-                       }
-
-                       if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
-                               i += 2;
-                               /* Comment, skip to end of line */
-                               for (; i < blk.sz; ++i) {
-                                       if ('\n' == blk.buf[i]) {
-                                               ++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;
-                       }
+                       ln.buf[pos++] = blk.buf[i++];
+               }
+               ln.buf[pos] = '\0';
 
 
-                       /* Some other escape sequence, copy & cont. */
+               /*
+                * Maintain a lookaside buffer of all lines.
+                * parsed from this input source.
+                */
 
 
-                       ln.buf[pos++] = blk.buf[i++];
-                       ln.buf[pos++] = blk.buf[i++];
+               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 (pos >= ln.sz)
-                       resize_buf(&ln, 256);
+               /* XXX Ugly hack to mark the end of the input. */
 
 
-               ln.buf[pos] = '\0';
+               if (i == blk.sz || blk.buf[i] == '\0') {
+                       ln.buf[pos++] = '\n';
+                       ln.buf[pos] = '\0';
+               }
 
                /*
                 * A significant amount of complexity is contained by
 
                /*
                 * A significant amount of complexity is contained by
@@ -481,65 +494,108 @@ mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
                 */
 
                of = 0;
                 */
 
                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 (line_result & ROFF_APPEND)
+                       assert(line_result == (ROFF_IGN | ROFF_APPEND));
 
 
-               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_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) {
+               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, curp->line, pos, NULL);
+                       }
+                       curp->loop = thisln;
+                       loop = NULL;
+                       inloop = 1;
+                       break;
+               case ROFF_LOOPCONT:
+               case ROFF_LOOPEXIT:
+                       if (curp->loop == NULL) {
+                               mandoc_msg(MANDOCERR_WHILE_FAIL,
+                                   curp, curp->line, pos, NULL);
+                               break;
+                       }
+                       if (inloop == 0) {
+                               mandoc_msg(MANDOCERR_WHILE_INTO,
+                                   curp, 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->macroset == MACROSET_NONE)
+                               choose_parser(curp);
+                       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)
+                               goto out;
+                       break;
+               case ROFF_RERUN:
+                       goto rerun;
                case ROFF_REPARSE:
                case ROFF_REPARSE:
-                       if (REPARSE_LIMIT >= ++curp->reparse_count)
-                               mparse_buf_r(curp, ln, of, 0);
-                       else
+                       if (++curp->reparse_count > REPARSE_LIMIT) {
+                               /* Abort and return to the top level. */
+                               result = ROFF_IGN;
                                mandoc_msg(MANDOCERR_ROFFLOOP, curp,
                                    curp->line, pos, NULL);
                                mandoc_msg(MANDOCERR_ROFFLOOP, curp,
                                    curp->line, pos, NULL);
-                       pos = 0;
-                       continue;
-               case ROFF_APPEND:
-                       pos = strlen(ln.buf);
-                       continue;
-               case ROFF_RERUN:
-                       goto rerun;
-               case ROFF_IGN:
-                       pos = 0;
-                       continue;
+                               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);
                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;
+                               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;
                        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);
                                mparse_readfd(curp, fd, ln.buf + of);
+                               close(fd);
                                curp->file = save_file;
                        } else {
                                curp->file = save_file;
                                curp->file = save_file;
                        } else {
                                curp->file = save_file;
@@ -554,68 +610,57 @@ rerun:
                                of = 0;
                                mparse_buf_r(curp, ln, of, 0);
                        }
                                of = 0;
                                mparse_buf_r(curp, ln, of, 0);
                        }
-                       pos = 0;
-                       continue;
-               default:
                        break;
                        break;
+               default:
+                       abort();
                }
 
                }
 
-               /*
-                * 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)
-                       choose_parser(curp);
-
-               /*
-                * 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 (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;
-
-               /* Temporary buffers typically are not full. */
-
-               if (0 == start && '\0' == blk.buf[i])
-                       break;
-
                /* Start the next input line. */
 
                /* Start the next input line. */
 
-               pos = 0;
-       }
+               if (loop != NULL &&
+                   (line_result & ROFF_LOOPMASK) == ROFF_IGN)
+                       loop = loop->next;
+
+               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;
+               }
 
 
+               pos = (line_result & ROFF_APPEND) ? strlen(ln.buf) : 0;
+       }
+out:
+       if (inloop) {
+               if (result != ROFF_USERRET)
+                       mandoc_msg(MANDOCERR_WHILE_OUTOF, curp,
+                           curp->line, pos, NULL);
+               curp->loop = NULL;
+       }
        free(ln.buf);
        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)
 {
 }
 
 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;
        gzFile           gz;
        size_t           off;
        ssize_t          ssz;
+       int              gzerrnum, retval;
 
 
-#if HAVE_MMAP
-       struct stat      st;
-
-       if (fstat(fd, &st) == -1)
-               err((int)MANDOCLEVEL_SYSERR, "%s", file);
+       if (fstat(fd, &st) == -1) {
+               mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
+                   "fstat: %s", strerror(errno));
+               return 0;
+       }
 
        /*
         * If we're a regular file, try just reading in the whole entry
 
        /*
         * If we're a regular file, try just reading in the whole entry
@@ -635,11 +680,26 @@ read_whole_file(struct mparse *curp, const char *file, int fd,
                if (fb->buf != MAP_FAILED)
                        return 1;
        }
                if (fb->buf != MAP_FAILED)
                        return 1;
        }
-#endif
 
        if (curp->gzip) {
 
        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_vmsg(MANDOCERR_FILE, curp, 0, 0,
+                           "dup: %s", strerror(errno));
+                       return 0;
+               }
+               if ((gz = gzdopen(fd, "rb")) == NULL) {
+                       mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
+                           "gzdopen: %s", strerror(errno));
+                       close(fd);
+                       return 0;
+               }
        } else
                gz = NULL;
 
        } else
                gz = NULL;
 
@@ -650,6 +710,7 @@ read_whole_file(struct mparse *curp, const char *file, int fd,
 
        *with_mmap = 0;
        off = 0;
 
        *with_mmap = 0;
        off = 0;
+       retval = 0;
        fb->sz = 0;
        fb->buf = NULL;
        for (;;) {
        fb->sz = 0;
        fb->buf = NULL;
        for (;;) {
@@ -666,25 +727,34 @@ read_whole_file(struct mparse *curp, const char *file, int fd,
                    read(fd, fb->buf + (int)off, fb->sz - off);
                if (ssz == 0) {
                        fb->sz = off;
                    read(fd, fb->buf + (int)off, fb->sz - off);
                if (ssz == 0) {
                        fb->sz = off;
-                       return 1;
+                       retval = 1;
+                       break;
+               }
+               if (ssz == -1) {
+                       if (curp->gzip)
+                               (void)gzerror(gz, &gzerrnum);
+                       mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, "read: %s",
+                           curp->gzip && gzerrnum != Z_ERRNO ?
+                           zError(gzerrnum) : strerror(errno));
+                       break;
                }
                }
-               if (ssz == -1)
-                       err((int)MANDOCLEVEL_SYSERR, "%s", file);
                off += (size_t)ssz;
        }
 
                off += (size_t)ssz;
        }
 
-       free(fb->buf);
-       fb->buf = NULL;
-       return 0;
+       if (curp->gzip && (gzerrnum = gzclose(gz)) != Z_OK)
+               mandoc_vmsg(MANDOCERR_FILE, curp, 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)
 {
 }
 
 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)
        if (curp->man->macroset == MACROSET_NONE)
                curp->man->macroset = MACROSET_MAN;
        if (curp->man->macroset == MACROSET_MDOC)
@@ -764,24 +834,19 @@ mparse_readfd(struct mparse *curp, int fd, const char *file)
                    (MPARSE_UTF8 | MPARSE_LATIN1);
                mparse_parse_buffer(curp, blk, file);
                curp->filenc = save_filenc;
                    (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
                if (with_mmap)
                        munmap(blk.buf, blk.sz);
                else
-#endif
                        free(blk.buf);
        }
                        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;
 {
        char             *cp;
+       int               fd;
 
        curp->file = file;
        cp = strrchr(file, '.');
 
        curp->file = file;
        cp = strrchr(file, '.');
@@ -789,8 +854,8 @@ mparse_open(struct mparse *curp, int *fd, const char *file)
 
        /* First try to use the filename as it is. */
 
 
        /* 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
 
        /*
         * If that doesn't work and the filename doesn't
@@ -799,75 +864,73 @@ mparse_open(struct mparse *curp, int *fd, const char *file)
 
        if ( ! curp->gzip) {
                mandoc_asprintf(&cp, "%s.gz", file);
 
        if ( ! curp->gzip) {
                mandoc_asprintf(&cp, "%s.gz", file);
-               *fd = open(file, O_RDONLY);
+               fd = open(cp, O_RDONLY);
                free(cp);
                free(cp);
-               if (*fd != -1) {
+               if (fd != -1) {
                        curp->gzip = 1;
                        curp->gzip = 1;
-                       return MANDOCLEVEL_OK;
+                       return fd;
                }
        }
 
        /* Neither worked, give up. */
 
        mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
                }
        }
 
        /* Neither worked, give up. */
 
        mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
-       return MANDOCLEVEL_ERROR;
+       return -1;
 }
 
 struct mparse *
 }
 
 struct mparse *
-mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
-    const char *defos)
+mparse_alloc(int options, enum mandocerr mmin, mandocmsg mmsg,
+    enum mandoc_os os_e, const char *os_s)
 {
        struct mparse   *curp;
 
        curp = mandoc_calloc(1, sizeof(struct mparse));
 
        curp->options = options;
 {
        struct mparse   *curp;
 
        curp = mandoc_calloc(1, sizeof(struct mparse));
 
        curp->options = options;
-       curp->wlevel = wlevel;
+       curp->mmin = mmin;
        curp->mmsg = mmsg;
        curp->mmsg = mmsg;
-       curp->defos = defos;
+       curp->os_s = os_s;
 
        curp->roff = roff_alloc(curp, options);
 
        curp->roff = roff_alloc(curp, options);
-       curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
+       curp->man = roff_man_alloc(curp->roff, curp, curp->os_s,
                curp->options & MPARSE_QUICK ? 1 : 0);
        if (curp->options & MPARSE_MDOC) {
                curp->options & MPARSE_QUICK ? 1 : 0);
        if (curp->options & MPARSE_MDOC) {
-               mdoc_hash_init();
                curp->man->macroset = MACROSET_MDOC;
                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) {
        } else if (curp->options & MPARSE_MAN) {
-               man_hash_init();
                curp->man->macroset = MACROSET_MAN;
                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;
        }
        curp->man->first->tok = TOKEN_NONE;
+       curp->man->meta.os_e = os_e;
        return curp;
 }
 
 void
 mparse_reset(struct mparse *curp)
 {
        return curp;
 }
 
 void
 mparse_reset(struct mparse *curp)
 {
-
        roff_reset(curp->roff);
        roff_reset(curp->roff);
-
-       if (curp->man != NULL)
-               roff_man_reset(curp->man);
-       if (curp->secondary)
-               curp->secondary->sz = 0;
-
-       curp->file_status = MANDOCLEVEL_OK;
+       roff_man_reset(curp->man);
+       free_buf_list(curp->secondary);
+       curp->secondary = NULL;
 
        free(curp->sodest);
        curp->sodest = NULL;
 
        free(curp->sodest);
        curp->sodest = NULL;
+
+       curp->file_status = MANDOCLEVEL_OK;
+       curp->gzip = 0;
 }
 
 void
 mparse_free(struct mparse *curp)
 {
 }
 
 void
 mparse_free(struct mparse *curp)
 {
-
+       roffhash_free(curp->man->mdocmac);
+       roffhash_free(curp->man->manmac);
        roff_man_free(curp->man);
        roff_man_free(curp->man);
-       if (curp->roff)
-               roff_free(curp->roff);
-       if (curp->secondary)
-               free(curp->secondary->buf);
-
-       free(curp->secondary);
+       roff_free(curp->roff);
+       free_buf_list(curp->secondary);
        free(curp->sodest);
        free(curp);
 }
        free(curp->sodest);
        free(curp);
 }
@@ -885,6 +948,13 @@ mparse_result(struct mparse *curp, struct roff_man **man,
                *man = curp->man;
 }
 
                *man = curp->man;
 }
 
+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, ...)
 void
 mandoc_vmsg(enum mandocerr t, struct mparse *m,
                int ln, int pos, const char *fmt, ...)
@@ -905,13 +975,13 @@ mandoc_msg(enum mandocerr er, struct mparse *m,
 {
        enum mandoclevel level;
 
 {
        enum mandoclevel level;
 
+       if (er < m->mmin && er != MANDOCERR_FILE)
+               return;
+
        level = MANDOCLEVEL_UNSUPP;
        while (er < mandoclimits[level])
                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->mmsg)
                (*m->mmsg)(er, level, m->file, ln, col, msg);
 
@@ -933,17 +1003,10 @@ mparse_strlevel(enum mandoclevel lvl)
 }
 
 void
 }
 
 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)
+mparse_copy(const struct mparse *p)
 {
 {
+       struct buf      *buf;
 
 
-       assert(p->secondary);
-       return p->secondary->sz ? p->secondary->buf : NULL;
+       for (buf = p->secondary; buf != NULL; buf = buf->next)
+               puts(buf->buf);
 }
 }
A git conversion mirror of mandoc.bsd.lv