Ugly fix for `Bl' or `Bd' causing badness when nested in `Bl -hang' lists.

[mandoc.git] / main.c

diff --git a/main.c b/main.c
index 517dbc490a8d8f53e00d5b94ad6527e5a6723313..26988773e7bf647641613f7338ba8750fd7e60e9 100644 (file)
--- a/main.c
+++ b/main.c
@@ -1,4 +1,4 @@
-/*     $Id: main.c,v 1.27 2009/06/11 12:07:49 kristaps Exp $ */
+/*     $Id: main.c,v 1.38 2009/07/07 09:52:08 kristaps Exp $ */

 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *


@@ -65,21 +65,19 @@ struct      curparse {
        const char       *file;         /* Current parse. */
        int               fd;           /* Current parse. */
        int               wflags;
        const char       *file;         /* Current parse. */
        int               fd;           /* Current parse. */
        int               wflags;

-#define        WARN_WALL         0x03          /* All-warnings mask. */
-#define        WARN_WCOMPAT     (1 << 0)       /* Compatibility warnings. */
-#define        WARN_WSYNTAX     (1 << 1)       /* Syntax warnings. */
+#define        WARN_WALL        (1 << 0)       /* All-warnings mask. */

 #define        WARN_WERR        (1 << 2)       /* Warnings->errors. */
        int               fflags;
 #define        IGN_SCOPE        (1 << 0)       /* Ignore scope errors. */
 #define        NO_IGN_ESCAPE    (1 << 1)       /* Don't ignore bad escapes. */
 #define        NO_IGN_MACRO     (1 << 2)       /* Don't ignore bad macros. */
 #define        NO_IGN_CHARS     (1 << 3)       /* Don't ignore bad chars. */
 #define        WARN_WERR        (1 << 2)       /* Warnings->errors. */
        int               fflags;
 #define        IGN_SCOPE        (1 << 0)       /* Ignore scope errors. */
 #define        NO_IGN_ESCAPE    (1 << 1)       /* Don't ignore bad escapes. */
 #define        NO_IGN_MACRO     (1 << 2)       /* Don't ignore bad macros. */
 #define        NO_IGN_CHARS     (1 << 3)       /* Don't ignore bad chars. */

-       enum intt         inttype;      /* Input parsers. */
+       enum intt         inttype;      /* Input parsers... */

        struct man       *man;
        struct man       *lastman;
        struct mdoc      *mdoc;
        struct mdoc      *lastmdoc;
        struct man       *man;
        struct man       *lastman;
        struct mdoc      *mdoc;
        struct mdoc      *lastmdoc;

-       enum outt         outtype;      /* Output devices. */
+       enum outt         outtype;      /* Output devices... */

        out_mdoc          outmdoc;
        out_man           outman;
        out_free          outfree;
        out_mdoc          outmdoc;
        out_man           outman;
        out_free          outfree;


@@ -98,11 +96,7 @@ static       int               toptions(enum outt *, char *);
 static int               moptions(enum intt *, char *);
 static int               woptions(int *, char *);
 static int               merr(void *, int, int, const char *);
 static int               moptions(enum intt *, char *);
 static int               woptions(int *, char *);
 static int               merr(void *, int, int, const char *);

-static int               manwarn(void *, int, int, const char *);
-static int               mdocwarn(void *, int, int, 
-                               enum mdoc_warn, const char *);
-static int               fstdin(struct buf *, struct buf *, 
-                               struct curparse *);
+static int               mwarn(void *, int, int, const char *);

 static int               ffile(struct buf *, struct buf *, 
                                const char *, struct curparse *);
 static int               fdesc(struct buf *, struct buf *,
 static int               ffile(struct buf *, struct buf *, 
                                const char *, struct curparse *);
 static int               fdesc(struct buf *, struct buf *,


@@ -134,19 +128,19 @@ main(int argc, char *argv[])
                switch (c) {
                case ('f'):
                        if ( ! foptions(&curp.fflags, optarg))
                switch (c) {
                case ('f'):
                        if ( ! foptions(&curp.fflags, optarg))

-                               return(0);
+                               return(EXIT_FAILURE);

                        break;
                case ('m'):
                        if ( ! moptions(&curp.inttype, optarg))
                        break;
                case ('m'):
                        if ( ! moptions(&curp.inttype, optarg))

-                               return(0);
+                               return(EXIT_FAILURE);

                        break;
                case ('T'):
                        if ( ! toptions(&curp.outtype, optarg))
                        break;
                case ('T'):
                        if ( ! toptions(&curp.outtype, optarg))

-                               return(0);
+                               return(EXIT_FAILURE);

                        break;
                case ('W'):
                        if ( ! woptions(&curp.wflags, optarg))
                        break;
                case ('W'):
                        if ( ! woptions(&curp.wflags, optarg))

-                               return(0);
+                               return(EXIT_FAILURE);

                        break;
                case ('V'):
                        version();
                        break;
                case ('V'):
                        version();


@@ -159,16 +153,17 @@ main(int argc, char *argv[])
        argc -= optind;
        argv += optind;
 
        argc -= optind;
        argv += optind;
 

-       /* Configure buffers. */
-

        bzero(&ln, sizeof(struct buf));
        bzero(&blk, sizeof(struct buf));
 
        rc = 1;
 
        bzero(&ln, sizeof(struct buf));
        bzero(&blk, sizeof(struct buf));
 
        rc = 1;
 

-       if (NULL == *argv)
-               if ( ! fstdin(&blk, &ln, &curp))
+       if (NULL == *argv) {
+               curp.file = "<stdin>";
+               curp.fd = STDIN_FILENO;
+               if ( ! fdesc(&blk, &ln, &curp))

                        rc = 0;
                        rc = 0;

+       }

 
        while (rc && *argv) {
                if ( ! ffile(&blk, &ln, *argv, &curp))
 
        while (rc && *argv) {
                if ( ! ffile(&blk, &ln, *argv, &curp))


@@ -190,8 +185,6 @@ main(int argc, char *argv[])
                free(blk.buf);
        if (ln.buf)
                free(ln.buf);
                free(blk.buf);
        if (ln.buf)
                free(ln.buf);

-
-       /* TODO: have a curp_free routine. */

        if (curp.outfree)
                (*curp.outfree)(curp.outdata);
        if (curp.mdoc)
        if (curp.outfree)
                (*curp.outfree)(curp.outdata);
        if (curp.mdoc)


@@ -231,19 +224,18 @@ man_init(struct curparse *curp)
        struct man_cb    mancb;
 
        mancb.man_err = merr;
        struct man_cb    mancb;
 
        mancb.man_err = merr;

-       mancb.man_warn = manwarn;
-
-       /*
-        * Default behaviour is to ignore unknown macros.  This is
-        * specified in mandoc.1.
-        */
+       mancb.man_warn = mwarn;

 
 

-       pflags = MAN_IGN_MACRO;
+       /* Defaults from mandoc.1. */

 
 

-       /* Override default behaviour... */
+       pflags = MAN_IGN_MACRO | MAN_IGN_ESCAPE | MAN_IGN_CHARS;

 
        if (curp->fflags & NO_IGN_MACRO)
                pflags &= ~MAN_IGN_MACRO;
 
        if (curp->fflags & NO_IGN_MACRO)
                pflags &= ~MAN_IGN_MACRO;

+       if (curp->fflags & NO_IGN_CHARS)
+               pflags &= ~MAN_IGN_CHARS;
+       if (curp->fflags & NO_IGN_ESCAPE)
+               pflags &= ~MAN_IGN_ESCAPE;

 
        if (NULL == (man = man_alloc(curp, pflags, &mancb)))
                warnx("memory exhausted");
 
        if (NULL == (man = man_alloc(curp, pflags, &mancb)))
                warnx("memory exhausted");


@@ -259,20 +251,13 @@ mdoc_init(struct curparse *curp)
        struct mdoc     *mdoc;
        struct mdoc_cb   mdoccb;
 
        struct mdoc     *mdoc;
        struct mdoc_cb   mdoccb;
 

-       mdoccb.mdoc_msg = NULL;

        mdoccb.mdoc_err = merr;
        mdoccb.mdoc_err = merr;

-       mdoccb.mdoc_warn = mdocwarn;
+       mdoccb.mdoc_warn = mwarn;

 
 

-       /* 
-        * Default behaviour is to ignore unknown macros, escape
-        * sequences and characters (very liberal).  This is specified
-        * in mandoc.1.
-        */
+       /* Defaults from mandoc.1. */

 
        pflags = MDOC_IGN_MACRO | MDOC_IGN_ESCAPE | MDOC_IGN_CHARS;
 
 
        pflags = MDOC_IGN_MACRO | MDOC_IGN_ESCAPE | MDOC_IGN_CHARS;
 

-       /* Override default behaviour... */
-

        if (curp->fflags & IGN_SCOPE)
                pflags |= MDOC_IGN_SCOPE;
        if (curp->fflags & NO_IGN_ESCAPE)
        if (curp->fflags & IGN_SCOPE)
                pflags |= MDOC_IGN_SCOPE;
        if (curp->fflags & NO_IGN_ESCAPE)


@@ -289,16 +274,6 @@ mdoc_init(struct curparse *curp)
 }
 
 
 }
 
 

-static int
-fstdin(struct buf *blk, struct buf *ln, struct curparse *curp)
-{
-
-       curp->file = "<stdin>";
-       curp->fd = STDIN_FILENO;
-       return(fdesc(blk, ln, curp));
-}
-
-

 static int
 ffile(struct buf *blk, struct buf *ln, 
                const char *file, struct curparse *curp)
 static int
 ffile(struct buf *blk, struct buf *ln, 
                const char *file, struct curparse *curp)


@@ -326,7 +301,7 @@ fdesc(struct buf *blk, struct buf *ln, struct curparse *curp)
        size_t           sz;
        ssize_t          ssz;
        struct stat      st;
        size_t           sz;
        ssize_t          ssz;
        struct stat      st;

-       int              j, i, pos, lnn;
+       int              j, i, pos, lnn, comment;

        struct man      *man;
        struct mdoc     *mdoc;
 
        struct man      *man;
        struct mdoc     *mdoc;
 


@@ -341,7 +316,7 @@ fdesc(struct buf *blk, struct buf *ln, struct curparse *curp)
         */
 
        if (-1 == fstat(curp->fd, &st))
         */
 
        if (-1 == fstat(curp->fd, &st))

-               warnx("%s", curp->file);
+               warn("%s", curp->file);

        else if ((size_t)st.st_blksize > sz)
                sz = st.st_blksize;
 
        else if ((size_t)st.st_blksize > sz)
                sz = st.st_blksize;
 


@@ -356,7 +331,7 @@ fdesc(struct buf *blk, struct buf *ln, struct curparse *curp)
 
        /* Fill buf with file blocksize. */
 
 
        /* Fill buf with file blocksize. */
 

-       for (lnn = 0, pos = 0; ; ) {
+       for (lnn = pos = comment = 0; ; ) {

                if (-1 == (ssz = read(curp->fd, blk->buf, sz))) {
                        warn("%s", curp->file);
                        return(0);
                if (-1 == (ssz = read(curp->fd, blk->buf, sz))) {
                        warn("%s", curp->file);
                        return(0);


@@ -376,17 +351,34 @@ fdesc(struct buf *blk, struct buf *ln, struct curparse *curp)
                        }
 
                        if ('\n' != blk->buf[i]) {
                        }
 
                        if ('\n' != blk->buf[i]) {

+                               if (comment)
+                                       continue;

                                ln->buf[pos++] = blk->buf[i];
                                ln->buf[pos++] = blk->buf[i];

+
+                               /* Handle in-line `\"' comments. */
+
+                               if (1 == pos || '\"' != ln->buf[pos - 1])
+                                       continue;
+
+                               for (j = pos - 2; j >= 0; j--)
+                                       if ('\\' != ln->buf[j])
+                                               break;
+
+                               if ( ! ((pos - 2 - j) % 2))
+                                       continue;
+
+                               comment = 1;
+                               pos -= 2;

                                continue;
                                continue;

-                       }
+                       } 

 
 

-                       /* Check for CPP-escaped newline. */
+                       /* Handle escaped `\\n' newlines. */

 
 

-                       if (pos > 0 && '\\' == ln->buf[pos - 1]) {
+                       if (pos > 0 && 0 == comment && 
+                                       '\\' == ln->buf[pos - 1]) {

                                for (j = pos - 1; j >= 0; j--)
                                        if ('\\' != ln->buf[j])
                                                break;
                                for (j = pos - 1; j >= 0; j--)
                                        if ('\\' != ln->buf[j])
                                                break;

-

                                if ( ! ((pos - j) % 2)) {
                                        pos--;
                                        lnn++;
                                if ( ! ((pos - j) % 2)) {
                                        pos--;
                                        lnn++;


@@ -396,19 +388,16 @@ fdesc(struct buf *blk, struct buf *ln, struct curparse *curp)
 
                        ln->buf[pos] = 0;
                        lnn++;
 
                        ln->buf[pos] = 0;
                        lnn++;

-                       
-                       /*
-                        * If no manual parser has been assigned, then
-                        * try to assign one in pset(), which may do
-                        * nothing at all.  After this, parse the manual
-                        * line accordingly.
-                        */
+
+                       /* If unset, assign parser in pset(). */

 
                        if ( ! (man || mdoc) && ! pset(ln->buf, 
                                                pos, curp, &man, &mdoc))
                                return(0);
 
 
                        if ( ! (man || mdoc) && ! pset(ln->buf, 
                                                pos, curp, &man, &mdoc))
                                return(0);
 

-                       pos = 0;
+                       pos = comment = 0;
+
+                       /* Pass down into parsers. */

 
                        if (man && ! man_parseln(man, lnn, ln->buf))
                                return(0);
 
                        if (man && ! man_parseln(man, lnn, ln->buf))
                                return(0);


@@ -417,7 +406,7 @@ fdesc(struct buf *blk, struct buf *ln, struct curparse *curp)
                }
        }
 
                }
        }
 

-       /* Note that a parser may not have been assigned, yet. */
+       /* NOTE a parser may not have been assigned, yet. */

 
        if ( ! (man || mdoc)) {
                warnx("%s: not a manual", curp->file);
 
        if ( ! (man || mdoc)) {
                warnx("%s: not a manual", curp->file);


@@ -429,12 +418,7 @@ fdesc(struct buf *blk, struct buf *ln, struct curparse *curp)
        if (man && ! man_endparse(man))
                return(0);
 
        if (man && ! man_endparse(man))
                return(0);
 

-       /*
-        * If an output device hasn't been allocated, see if we should
-        * do so now.  Note that not all outtypes have functions, so
-        * this switch statement may be superfluous, but it's
-        * low-overhead enough not to matter very much.
-        */
+       /* If unset, allocate output dev now (if applicable). */

 
        if ( ! (curp->outman && curp->outmdoc)) {
                switch (curp->outtype) {
 
        if ( ! (curp->outman && curp->outmdoc)) {
                switch (curp->outtype) {


@@ -470,17 +454,23 @@ static int
 pset(const char *buf, int pos, struct curparse *curp,
                struct man **man, struct mdoc **mdoc)
 {
 pset(const char *buf, int pos, struct curparse *curp,
                struct man **man, struct mdoc **mdoc)
 {

+       int              i;

 
        /*
         * Try to intuit which kind of manual parser should be used.  If
         * passed in by command-line (-man, -mdoc), then use that
         * explicitly.  If passed as -mandoc, then try to guess from the
 
        /*
         * Try to intuit which kind of manual parser should be used.  If
         * passed in by command-line (-man, -mdoc), then use that
         * explicitly.  If passed as -mandoc, then try to guess from the

-        * line: either skip comments, use -mdoc when finding `.Dt', or
+        * line: either skip dot-lines, use -mdoc when finding `.Dt', or

         * default to -man, which is more lenient.
         */
 
         * default to -man, which is more lenient.
         */
 

-       if (pos >= 3 && 0 == memcmp(buf, ".\\\"", 3))
-               return(1);
+       if (buf[0] == '.') {
+               for (i = 1; buf[i]; i++)
+                       if (' ' != buf[i] && '\t' != buf[i])
+                               break;
+               if (0 == buf[i])
+                       return(1);
+       }

 
        switch (curp->inttype) {
        case (INTT_MDOC):
 
        switch (curp->inttype) {
        case (INTT_MDOC):


@@ -557,14 +547,10 @@ toptions(enum outt *tflags, char *arg)
 }
 
 
 }
 
 

-/*
- * Parse out the options for [-fopt...] setting compiler options.  These
- * can be comma-delimited or called again.
- */

 static int
 foptions(int *fflags, char *arg)
 {
 static int
 foptions(int *fflags, char *arg)
 {

-       char            *v;
+       char            *v, *o;

        char            *toks[6];
 
        toks[0] = "ign-scope";
        char            *toks[6];
 
        toks[0] = "ign-scope";


@@ -574,7 +560,8 @@ foptions(int *fflags, char *arg)
        toks[4] = "strict";
        toks[5] = NULL;
 
        toks[4] = "strict";
        toks[5] = NULL;
 

-       while (*arg) 
+       while (*arg) {
+               o = arg;

                switch (getsubopt(&arg, toks, &v)) {
                case (0):
                        *fflags |= IGN_SCOPE;
                switch (getsubopt(&arg, toks, &v)) {
                case (0):
                        *fflags |= IGN_SCOPE;


@@ -593,48 +580,39 @@ foptions(int *fflags, char *arg)
                                   NO_IGN_MACRO | NO_IGN_CHARS;
                        break;
                default:
                                   NO_IGN_MACRO | NO_IGN_CHARS;
                        break;
                default:

-                       warnx("bad argument: -f%s", arg);
+                       warnx("bad argument: -f%s", o);

                        return(0);
                }
                        return(0);
                }

+       }

 
        return(1);
 }
 
 
 
        return(1);
 }
 
 

-/* 
- * Parse out the options for [-Werr...], which sets warning modes.
- * These can be comma-delimited or called again.  
- */

 static int
 woptions(int *wflags, char *arg)
 {
 static int
 woptions(int *wflags, char *arg)
 {

-       char            *v;
-       char            *toks[5]; 
+       char            *v, *o;
+       char            *toks[3]; 

 
        toks[0] = "all";
 
        toks[0] = "all";

-       toks[1] = "compat";
-       toks[2] = "syntax";
-       toks[3] = "error";
-       toks[4] = NULL;
+       toks[1] = "error";
+       toks[2] = NULL;

 
 

-       while (*arg) 
+       while (*arg) {
+               o = arg;

                switch (getsubopt(&arg, toks, &v)) {
                case (0):
                        *wflags |= WARN_WALL;
                        break;
                case (1):
                switch (getsubopt(&arg, toks, &v)) {
                case (0):
                        *wflags |= WARN_WALL;
                        break;
                case (1):

-                       *wflags |= WARN_WCOMPAT;
-                       break;
-               case (2):
-                       *wflags |= WARN_WSYNTAX;
-                       break;
-               case (3):

                        *wflags |= WARN_WERR;
                        break;
                default:
                        *wflags |= WARN_WERR;
                        break;
                default:

-                       warnx("bad argument: -W%s", arg);
+                       warnx("bad argument: -W%s", o);

                        return(0);
                }
                        return(0);
                }

+       }

 
        return(1);
 }
 
        return(1);
 }


@@ -647,77 +625,31 @@ merr(void *arg, int line, int col, const char *msg)
        struct curparse *curp;
 
        curp = (struct curparse *)arg;
        struct curparse *curp;
 
        curp = (struct curparse *)arg;

+

        warnx("%s:%d: error: %s (column %d)", 
                        curp->file, line, msg, col);
 
        warnx("%s:%d: error: %s (column %d)", 
                        curp->file, line, msg, col);
 

-       /* Always exit on errors... */

        return(0);
 }
 
 
 static int
        return(0);
 }
 
 
 static int

-mdocwarn(void *arg, int line, int col, 
-               enum mdoc_warn type, const char *msg)
+mwarn(void *arg, int line, int col, const char *msg)

 {
        struct curparse *curp;
 {
        struct curparse *curp;

-       char            *wtype;

 
        curp = (struct curparse *)arg;
 
        curp = (struct curparse *)arg;

-       wtype = NULL;

 
 

-       switch (type) {
-       case (WARN_COMPAT):
-               wtype = "compat";
-               if (curp->wflags & WARN_WCOMPAT)
-                       break;
+       if ( ! (curp->wflags & WARN_WALL))

                return(1);
                return(1);

-       case (WARN_SYNTAX):
-               wtype = "syntax";
-               if (curp->wflags & WARN_WSYNTAX)
-                       break;
-               return(1);
-       }

 
 

-       assert(wtype);
-       warnx("%s:%d: %s warning: %s (column %d)", 
-                       curp->file, line, wtype, msg, col);
+       warnx("%s:%d: warning: %s (column %d)", 
+                       curp->file, line, msg, col);

 
        if ( ! (curp->wflags & WARN_WERR))
                return(1);
        
 
        if ( ! (curp->wflags & WARN_WERR))
                return(1);
        

-       /*
-        * If the -Werror flag is passed in, as in gcc, then all
-        * warnings are considered as errors.
-        */
-
-       warnx("%s: considering warnings as errors", 
-                       __progname);
+       warnx("considering warnings as errors");

        return(0);
 }
 
        return(0);
 }
 

-
-static int
-manwarn(void *arg, int line, int col, const char *msg)
-{
-       struct curparse *curp;
-
-       curp = (struct curparse *)arg;
-
-       if ( ! (curp->wflags & WARN_WSYNTAX))
-               return(1);
-
-       warnx("%s:%d: syntax warning: %s (column %d)", 
-                       curp->file, line, msg, col);
-
-       if ( ! (curp->wflags & WARN_WERR))
-               return(1);
-
-       /* 
-        * If the -Werror flag is passed in, as in gcc, then all
-        * warnings are considered as errors.
-        */
-
-       warnx("%s: considering warnings as errors", 
-                       __progname);
-       return(0);
-}