want to get rid of the last style= attributes, suggested by bentley@

[mandoc.git] / read.c

diff --git a/read.c b/read.c
index 82156065cfe3169471c756f1e9eaab0035662d46..b3858061e710c7ad19bc3c4cde1a399c694d8959 100644 (file)
--- a/read.c
+++ b/read.c
@@ -1,7 +1,7 @@
-/*     $Id: read.c,v 1.205 2018/12/14 02:16:21 schwarze Exp $ */
+/*     $Id: read.c,v 1.214 2019/07/10 19:39:01 schwarze Exp $ */
 /*
  * 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
@@ -47,7 +47,6 @@
 struct mparse {
        struct roff      *roff; /* roff parser (!NULL) */
        struct roff_man  *man; /* man parser */
-       char             *sodest; /* filename pointed to by .so */
        struct buf       *primary; /* buffer currently being parsed */
        struct buf       *secondary; /* copy of top level input */
        struct buf       *loop; /* open .while request line */
@@ -123,15 +122,15 @@ choose_parser(struct mparse *curp)
        }
 
        if (format == MPARSE_MDOC) {
-               curp->man->macroset = MACROSET_MDOC;
+               curp->man->meta.macroset = MACROSET_MDOC;
                if (curp->man->mdocmac == NULL)
                        curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
        } else {
-               curp->man->macroset = MACROSET_MAN;
+               curp->man->meta.macroset = MACROSET_MAN;
                if (curp->man->manmac == NULL)
                        curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
        }
-       curp->man->first->tok = TOKEN_NONE;
+       curp->man->meta.first->tok = TOKEN_NONE;
 }
 
 /*
@@ -158,7 +157,7 @@ mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
        ln.sz = 256;
        ln.buf = mandoc_malloc(ln.sz);
        ln.next = NULL;
-       firstln = loop = NULL;
+       firstln = lastln = loop = NULL;
        lnn = curp->line;
        pos = 0;
        inloop = 0;
@@ -208,7 +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))) {
-                                       mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
+                                       mandoc_msg(MANDOCERR_CHAR_BAD,
                                            curp->line, pos, "0x%x", c);
                                        ln.buf[pos++] = '?';
                                        i++;
@@ -221,10 +220,10 @@ mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
                         */
 
                        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++] = '?';
@@ -256,6 +255,8 @@ mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
                /* XXX Ugly hack to mark the end of the input. */
 
                if (i == blk.sz || blk.buf[i] == '\0') {
+                       if (pos + 2 > ln.sz)
+                               resize_buf(&ln, 256);
                        ln.buf[pos++] = '\n';
                        ln.buf[pos] = '\0';
                }
@@ -298,7 +299,7 @@ rerun:
                                if (loop == curp->loop)
                                        break;
                                mandoc_msg(MANDOCERR_WHILE_NEST,
-                                   curp, curp->line, pos, NULL);
+                                   curp->line, pos, NULL);
                        }
                        curp->loop = thisln;
                        loop = NULL;
@@ -308,12 +309,12 @@ rerun:
                case ROFF_LOOPEXIT:
                        if (curp->loop == NULL) {
                                mandoc_msg(MANDOCERR_WHILE_FAIL,
-                                   curp, curp->line, pos, NULL);
+                                   curp->line, pos, NULL);
                                break;
                        }
                        if (inloop == 0) {
                                mandoc_msg(MANDOCERR_WHILE_INTO,
-                                   curp, curp->line, pos, NULL);
+                                   curp->line, pos, NULL);
                                curp->loop = loop = NULL;
                                break;
                        }
@@ -334,9 +335,9 @@ rerun:
                case ROFF_IGN:
                        break;
                case ROFF_CONT:
-                       if (curp->man->macroset == MACROSET_NONE)
+                       if (curp->man->meta.macroset == MACROSET_NONE)
                                choose_parser(curp);
-                       if ((curp->man->macroset == MACROSET_MDOC ?
+                       if ((curp->man->meta.macroset == MACROSET_MDOC ?
                             mdoc_parseln(curp->man, curp->line, ln.buf, of) :
                             man_parseln(curp->man, curp->line, ln.buf, of)
                            ) == 2)
@@ -348,7 +349,7 @@ rerun:
                        if (++curp->reparse_count > REPARSE_LIMIT) {
                                /* Abort and return to the top level. */
                                result = ROFF_IGN;
-                               mandoc_msg(MANDOCERR_ROFFLOOP, curp,
+                               mandoc_msg(MANDOCERR_ROFFLOOP,
                                    curp->line, pos, NULL);
                                goto out;
                        }
@@ -365,16 +366,17 @@ rerun:
                case ROFF_SO:
                        if ( ! (curp->options & MPARSE_SO) &&
                            (i >= blk.sz || blk.buf[i] == '\0')) {
-                               curp->sodest = mandoc_strdup(ln.buf + of);
+                               curp->man->meta.sodest =
+                                   mandoc_strdup(ln.buf + of);
                                goto out;
                        }
                        if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
                                mparse_readfd(curp, fd, ln.buf + of);
                                close(fd);
                        } else {
-                               mandoc_vmsg(MANDOCERR_SO_FAIL,
-                                   curp, curp->line, pos,
-                                   ".so %s", ln.buf + of);
+                               mandoc_msg(MANDOCERR_SO_FAIL,
+                                   curp->line, of, ".so %s: %s",
+                                   ln.buf + of, strerror(errno));
                                ln.sz = mandoc_asprintf(&cp,
                                    ".sp\nSee the file %s.\n.sp",
                                    ln.buf + of);
@@ -409,7 +411,7 @@ rerun:
 out:
        if (inloop) {
                if (result != ROFF_USERRET)
-                       mandoc_msg(MANDOCERR_WHILE_OUTOF, curp,
+                       mandoc_msg(MANDOCERR_WHILE_OUTOF,
                            curp->line, pos, NULL);
                curp->loop = NULL;
        }
@@ -429,9 +431,8 @@ read_whole_file(struct mparse *curp, int fd, struct buf *fb, int *with_mmap)
        int              gzerrnum, retval;
 
        if (fstat(fd, &st) == -1) {
-               mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
-                   "fstat: %s", strerror(errno));
-               return 0;
+               mandoc_msg(MANDOCERR_FSTAT, 0, 0, "%s", strerror(errno));
+               return -1;
        }
 
        /*
@@ -443,14 +444,14 @@ 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) {
-                       mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
-                       return 0;
+                       mandoc_msg(MANDOCERR_TOOLARGE, 0, 0, NULL);
+                       return -1;
                }
                *with_mmap = 1;
                fb->sz = (size_t)st.st_size;
                fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
                if (fb->buf != MAP_FAILED)
-                       return 1;
+                       return 0;
        }
 
        if (curp->gzip) {
@@ -462,15 +463,15 @@ 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) {
-                       mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
-                           "dup: %s", strerror(errno));
-                       return 0;
+                       mandoc_msg(MANDOCERR_DUP, 0, 0,
+                           "%s", strerror(errno));
+                       return -1;
                }
                if ((gz = gzdopen(fd, "rb")) == NULL) {
-                       mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
-                           "gzdopen: %s", strerror(errno));
+                       mandoc_msg(MANDOCERR_GZDOPEN, 0, 0,
+                           "%s", strerror(errno));
                        close(fd);
-                       return 0;
+                       return -1;
                }
        } else
                gz = NULL;
@@ -482,14 +483,13 @@ read_whole_file(struct mparse *curp, int fd, struct buf *fb, int *with_mmap)
 
        *with_mmap = 0;
        off = 0;
-       retval = 0;
+       retval = -1;
        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);
@@ -499,13 +499,13 @@ read_whole_file(struct mparse *curp, int fd, struct buf *fb, int *with_mmap)
                    read(fd, fb->buf + (int)off, fb->sz - off);
                if (ssz == 0) {
                        fb->sz = off;
-                       retval = 1;
+                       retval = 0;
                        break;
                }
                if (ssz == -1) {
                        if (curp->gzip)
                                (void)gzerror(gz, &gzerrnum);
-                       mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, "read: %s",
+                       mandoc_msg(MANDOCERR_READ, 0, 0, "%s",
                            curp->gzip && gzerrnum != Z_ERRNO ?
                            zError(gzerrnum) : strerror(errno));
                        break;
@@ -514,10 +514,10 @@ read_whole_file(struct mparse *curp, int fd, struct buf *fb, int *with_mmap)
        }
 
        if (curp->gzip && (gzerrnum = gzclose(gz)) != Z_OK)
-               mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, "gzclose: %s",
+               mandoc_msg(MANDOCERR_GZCLOSE, 0, 0, "%s",
                    gzerrnum == Z_ERRNO ? strerror(errno) :
                    zError(gzerrnum));
-       if (retval == 0) {
+       if (retval == -1) {
                free(fb->buf);
                fb->buf = NULL;
        }
@@ -527,9 +527,9 @@ read_whole_file(struct mparse *curp, int fd, struct buf *fb, int *with_mmap)
 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);
@@ -553,10 +553,10 @@ mparse_readfd(struct mparse *curp, int fd, const char *filename)
        int              with_mmap;
 
        if (recursion_depth > 64) {
-               mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
+               mandoc_msg(MANDOCERR_ROFFLOOP, curp->line, 0, NULL);
                return;
        }
-       if (read_whole_file(curp, fd, &blk, &with_mmap) == 0)
+       if (read_whole_file(curp, fd, &blk, &with_mmap) == -1)
                return;
 
        /*
@@ -608,7 +608,7 @@ 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"));
@@ -624,9 +624,11 @@ mparse_open(struct mparse *curp, const char *file)
         */
 
        if ( ! curp->gzip) {
+               save_errno = errno;
                mandoc_asprintf(&cp, "%s.gz", file);
                fd = open(cp, O_RDONLY);
                free(cp);
+               errno = save_errno;
                if (fd != -1) {
                        curp->gzip = 1;
                        return fd;
@@ -635,7 +637,6 @@ mparse_open(struct mparse *curp, const char *file)
 
        /* Neither worked, give up. */
 
-       mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
        return -1;
 }
 
@@ -649,19 +650,19 @@ mparse_alloc(int options, enum mandoc_os os_e, const char *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->man->macroset = MACROSET_MDOC;
+               curp->man->meta.macroset = MACROSET_MDOC;
                if (curp->man->mdocmac == NULL)
                        curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
        } else if (curp->options & MPARSE_MAN) {
-               curp->man->macroset = MACROSET_MAN;
+               curp->man->meta.macroset = MACROSET_MAN;
                if (curp->man->manmac == NULL)
                        curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
        }
-       curp->man->first->tok = TOKEN_NONE;
+       curp->man->meta.first->tok = TOKEN_NONE;
        curp->man->meta.os_e = os_e;
        return curp;
 }
@@ -673,8 +674,6 @@ mparse_reset(struct mparse *curp)
        roff_man_reset(curp->man);
        free_buf_list(curp->secondary);
        curp->secondary = NULL;
-       free(curp->sodest);
-       curp->sodest = NULL;
        curp->gzip = 0;
 }
 
@@ -686,21 +685,20 @@ mparse_free(struct mparse *curp)
        roff_man_free(curp->man);
        roff_free(curp->roff);
        free_buf_list(curp->secondary);
-       free(curp->sodest);
        free(curp);
 }
 
-void
-mparse_result(struct mparse *curp, struct roff_man **man,
-       char **sodest)
+struct roff_meta *
+mparse_result(struct mparse *curp)
 {
-
-       if (sodest && NULL != (*sodest = curp->sodest)) {
-               *man = NULL;
-               return;
+       roff_state_reset(curp->man);
+       if (curp->options & MPARSE_VALIDATE) {
+               if (curp->man->meta.macroset == MACROSET_MDOC)
+                       mdoc_validate(curp->man);
+               else
+                       man_validate(curp->man);
        }
-       if (man)
-               *man = curp->man;
+       return &curp->man->meta;
 }
 
 void