add arch.o

[mandoc.git] / read.c

diff --git a/read.c b/read.c
index cd0e027e74a3a38b4a517a78107c72d3b1acbef2..a3a1f982c01c1b2c02f419d5d70268970bdbe1e5 100644 (file)
--- a/read.c
+++ b/read.c
@@ -1,7 +1,7 @@
-/*     $Id: read.c,v 1.204 2018/12/14 01:24:49 schwarze Exp $ */
+/*     $Id: read.c,v 1.211 2019/01/11 17:04:44 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) 2010-2018 Ingo Schwarze <schwarze@openbsd.org>
+ * Copyright (c) 2010-2019 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


@@ -47,7 +47,6 @@
 struct mparse {
        struct roff      *roff; /* roff parser (!NULL) */
        struct roff_man  *man; /* man parser */
 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 */
        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 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
 
 
 static void


@@ -125,15 +122,15 @@ choose_parser(struct mparse *curp)
        }
 
        if (format == MPARSE_MDOC) {
        }
 
        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 {
                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);
        }
                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_t i, int start)
                        if (c & 0x80) {
                                if ( ! (curp->filenc && preconv_encode(
                                    &blk, &i, &ln, &pos, &curp->filenc))) {
                        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++;
                                            curp->line, pos, "0x%x", c);
                                        ln.buf[pos++] = '?';
                                        i++;


@@ -223,10 +220,10 @@ mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
                         */
 
                        if (c == 0x7f || (c < 0x20 && c != 0x09)) {
                         */
 
                        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,
                                    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++] = '?';
                                i++;
                                if (c != '\r')
                                        ln.buf[pos++] = '?';


@@ -300,7 +297,7 @@ rerun:
                                if (loop == curp->loop)
                                        break;
                                mandoc_msg(MANDOCERR_WHILE_NEST,
                                if (loop == curp->loop)
                                        break;
                                mandoc_msg(MANDOCERR_WHILE_NEST,

-                                   curp, curp->line, pos, NULL);
+                                   curp->line, pos, NULL);

                        }
                        curp->loop = thisln;
                        loop = NULL;
                        }
                        curp->loop = thisln;
                        loop = NULL;


@@ -310,12 +307,12 @@ rerun:
                case ROFF_LOOPEXIT:
                        if (curp->loop == NULL) {
                                mandoc_msg(MANDOCERR_WHILE_FAIL,
                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,
                                break;
                        }
                        if (inloop == 0) {
                                mandoc_msg(MANDOCERR_WHILE_INTO,

-                                   curp, curp->line, pos, NULL);
+                                   curp->line, pos, NULL);

                                curp->loop = loop = NULL;
                                break;
                        }
                                curp->loop = loop = NULL;
                                break;
                        }


@@ -336,9 +333,9 @@ rerun:
                case ROFF_IGN:
                        break;
                case ROFF_CONT:
                case ROFF_IGN:
                        break;
                case ROFF_CONT:

-                       if (curp->man->macroset == MACROSET_NONE)
+                       if (curp->man->meta.macroset == MACROSET_NONE)

                                choose_parser(curp);
                                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)
                             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;
                        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;
                        }
                                    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')) {
                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 {
                                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);
                                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)
 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;
        }
                            curp->line, pos, NULL);
                curp->loop = NULL;
        }


@@ -431,7 +429,7 @@ read_whole_file(struct mparse *curp, int fd, struct buf *fb, int *with_mmap)
        int              gzerrnum, retval;
 
        if (fstat(fd, &st) == -1) {
        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;
        }
                    "fstat: %s", strerror(errno));
                return 0;
        }


@@ -445,7 +443,7 @@ read_whole_file(struct mparse *curp, int fd, struct buf *fb, int *with_mmap)
 
        if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
                if (st.st_size > 0x7fffffff) {
 
        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;
                        return 0;
                }
                *with_mmap = 1;


@@ -464,12 +462,12 @@ read_whole_file(struct mparse *curp, int fd, struct buf *fb, int *with_mmap)
                 * which this function must not do.
                 */
                if ((fd = dup(fd)) == -1) {
                 * 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) {
                            "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;
                            "gzdopen: %s", strerror(errno));
                        close(fd);
                        return 0;


@@ -490,8 +488,7 @@ read_whole_file(struct mparse *curp, int fd, struct buf *fb, int *with_mmap)
        for (;;) {
                if (off == fb->sz) {
                        if (fb->sz == (1U << 31)) {
        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);
                                break;
                        }
                        resize_buf(fb, 65536);


@@ -507,7 +504,7 @@ read_whole_file(struct mparse *curp, int fd, struct buf *fb, int *with_mmap)
                if (ssz == -1) {
                        if (curp->gzip)
                                (void)gzerror(gz, &gzerrnum);
                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;
                            curp->gzip && gzerrnum != Z_ERRNO ?
                            zError(gzerrnum) : strerror(errno));
                        break;


@@ -516,7 +513,7 @@ read_whole_file(struct mparse *curp, int fd, struct buf *fb, int *with_mmap)
        }
 
        if (curp->gzip && (gzerrnum = gzclose(gz)) != Z_OK)
        }
 
        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) {
                    gzerrnum == Z_ERRNO ? strerror(errno) :
                    zError(gzerrnum));
        if (retval == 0) {


@@ -529,35 +526,51 @@ read_whole_file(struct mparse *curp, int fd, struct buf *fb, int *with_mmap)
 static void
 mparse_end(struct mparse *curp)
 {
 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);
 }
 
                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;
 
        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;
        }
                return;
        }

+       if (read_whole_file(curp, fd, &blk, &with_mmap) == 0)
+               return;
+
+       /*
+        * Save some properties of the parent file.
+        */
+
+       save_primary = curp->primary;
+       save_filenc = curp->filenc;
+       save_lineno = curp->line;
+       save_filename = mandoc_msg_getinfilename();

 
 

-       /* Line number is per-file. */
-       svfile = mandoc_msg_getinfilename();
-       mandoc_msg_setinfilename(file);
-       svprimary = curp->primary;

        curp->primary = &blk;
        curp->primary = &blk;

+       curp->filenc = curp->options & (MPARSE_UTF8 | MPARSE_LATIN1);

        curp->line = 1;
        curp->line = 1;

-       recursion_depth++;
+       mandoc_msg_setinfilename(filename);

 
        /* Skip an UTF-8 byte order mark. */
        if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
 
        /* Skip an UTF-8 byte order mark. */
        if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&


@@ -569,45 +582,32 @@ mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
        } else
                offset = 0;
 
        } else
                offset = 0;
 

+       recursion_depth++;

        mparse_buf_r(curp, blk, offset, 1);
        mparse_buf_r(curp, blk, offset, 1);

-

        if (--recursion_depth == 0)
                mparse_end(curp);
 
        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 (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);
-       }
+       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
 mparse_open(struct mparse *curp, const char *file)
 {
        char             *cp;
 }
 
 int
 mparse_open(struct mparse *curp, const char *file)
 {
        char             *cp;

-       int               fd;
+       int               fd, save_errno;

 
        cp = strrchr(file, '.');
        curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
 
        cp = strrchr(file, '.');
        curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));


@@ -623,9 +623,11 @@ mparse_open(struct mparse *curp, const char *file)
         */
 
        if ( ! curp->gzip) {
         */
 
        if ( ! curp->gzip) {

+               save_errno = errno;

                mandoc_asprintf(&cp, "%s.gz", file);
                fd = open(cp, O_RDONLY);
                free(cp);
                mandoc_asprintf(&cp, "%s.gz", file);
                fd = open(cp, O_RDONLY);
                free(cp);

+               errno = save_errno;

                if (fd != -1) {
                        curp->gzip = 1;
                        return fd;
                if (fd != -1) {
                        curp->gzip = 1;
                        return fd;


@@ -634,7 +636,6 @@ mparse_open(struct mparse *curp, const char *file)
 
        /* Neither worked, give up. */
 
 
        /* Neither worked, give up. */
 

-       mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));

        return -1;
 }
 
        return -1;
 }
 


@@ -648,19 +649,19 @@ mparse_alloc(int options, enum mandoc_os os_e, const char *os_s)
        curp->options = options;
        curp->os_s = os_s;
 
        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->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) {
                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);
        }
                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;
 }
        curp->man->meta.os_e = os_e;
        return curp;
 }


@@ -672,8 +673,6 @@ mparse_reset(struct mparse *curp)
        roff_man_reset(curp->man);
        free_buf_list(curp->secondary);
        curp->secondary = NULL;
        roff_man_reset(curp->man);
        free_buf_list(curp->secondary);
        curp->secondary = NULL;

-       free(curp->sodest);
-       curp->sodest = NULL;

        curp->gzip = 0;
 }
 
        curp->gzip = 0;
 }
 


@@ -685,21 +684,20 @@ mparse_free(struct mparse *curp)
        roff_man_free(curp->man);
        roff_free(curp->roff);
        free_buf_list(curp->secondary);
        roff_man_free(curp->man);
        roff_free(curp->roff);
        free_buf_list(curp->secondary);

-       free(curp->sodest);

        free(curp);
 }
 
        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
 }
 
 void