]> git.cameronkatri.com Git - mandoc.git/blobdiff - term.c
enclosures sometimes cause bogus end-of-sentence
[mandoc.git] / term.c
diff --git a/term.c b/term.c
index dd7aac5054ce127db5d537fb58a20205e63dd566..71ab3a359abe4c9b671ef554bb8f8d960bad4f83 100644 (file)
--- a/term.c
+++ b/term.c
@@ -1,4 +1,4 @@
-/*     $Id: term.c,v 1.163 2010/07/21 20:35:03 kristaps Exp $ */
+/*     $Id: term.c,v 1.171 2010/09/15 14:36:16 kristaps Exp $ */
 /*
  * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2010 Ingo Schwarze <schwarze@openbsd.org>
@@ -37,7 +37,6 @@
 static void              spec(struct termp *, enum roffdeco,
                                const char *, size_t);
 static void              res(struct termp *, const char *, size_t);
-static void              buffera(struct termp *, const char *, size_t);
 static void              bufferc(struct termp *, char);
 static void              adjbuf(struct termp *p, size_t);
 static void              encode(struct termp *, const char *, size_t);
@@ -84,7 +83,7 @@ term_alloc(enum termenc enc)
        p = calloc(1, sizeof(struct termp));
        if (NULL == p) {
                perror(NULL);
-               exit(EXIT_FAILURE);
+               exit((int)MANDOCLEVEL_SYSERR);
        }
 
        p->enc = enc;
@@ -170,12 +169,11 @@ term_flushln(struct termp *p)
                 * Handle literal tab characters: collapse all
                 * subsequent tabs into a single huge set of spaces.
                 */
-               for (j = i; j < (int)p->col; j++) {
-                       if ('\t' != p->buf[j])
-                               break;
+               while (i < (int)p->col && '\t' == p->buf[i]) {
                        vend = (vis / p->tabwidth + 1) * p->tabwidth;
                        vbl += vend - vis;
                        vis = vend;
+                       i++;
                }
 
                /*
@@ -186,7 +184,7 @@ term_flushln(struct termp *p)
                 */
 
                /* LINTED */
-               for (jhy = 0; j < (int)p->col; j++) {
+               for (j = i, jhy = 0; j < (int)p->col; j++) {
                        if ((j && ' ' == p->buf[j]) || '\t' == p->buf[j])
                                break;
 
@@ -229,12 +227,6 @@ term_flushln(struct termp *p)
                        p->overstep = 0;
                }
 
-               /*
-                * Skip leading tabs, they were handled above.
-                */
-               while (i < (int)p->col && '\t' == p->buf[i])
-                       i++;
-
                /* Write out the [remaining] word. */
                for ( ; i < (int)p->col; i++) {
                        if (vend > bp && jhy > 0 && i > jhy)
@@ -242,10 +234,10 @@ term_flushln(struct termp *p)
                        if ('\t' == p->buf[i])
                                break;
                        if (' ' == p->buf[i]) {
-                               while (' ' == p->buf[i]) {
-                                       vbl += (*p->width)(p, p->buf[i]);
+                               j = i;
+                               while (' ' == p->buf[i])
                                        i++;
-                               }
+                               vbl += (i - j) * (*p->width)(p, ' ');
                                break;
                        }
                        if (ASCII_NBRSP == p->buf[i]) {
@@ -276,6 +268,12 @@ term_flushln(struct termp *p)
                vis = vend;
        }
 
+       /*
+        * If there was trailing white space, it was not printed;
+        * so reset the cursor position accordingly.
+        */
+       vis -= vbl;
+
        p->col = 0;
        p->overstep = 0;
 
@@ -460,7 +458,6 @@ void
 term_word(struct termp *p, const char *word)
 {
        const char      *sv, *seq;
-       int              sz;
        size_t           ssz;
        enum roffdeco    deco;
 
@@ -503,6 +500,8 @@ term_word(struct termp *p, const char *word)
 
        if ( ! (p->flags & TERMP_NONOSPACE))
                p->flags &= ~TERMP_NOSPACE;
+       else
+               p->flags |= TERMP_NOSPACE;
 
        p->flags &= ~TERMP_SENTENCE;
 
@@ -515,7 +514,7 @@ term_word(struct termp *p, const char *word)
                        continue;
 
                seq = ++word;
-               sz = a2roffdeco(&deco, &seq, &ssz);
+               word += a2roffdeco(&deco, &seq, &ssz);
 
                switch (deco) {
                case (DECO_RESERVED):
@@ -542,7 +541,6 @@ term_word(struct termp *p, const char *word)
                        break;
                }
 
-               word += sz;
                if (DECO_NOSPACE == deco && '\0' == *word)
                        p->flags |= TERMP_NOSPACE;
        }
@@ -576,23 +574,11 @@ adjbuf(struct termp *p, size_t sz)
        p->buf = realloc(p->buf, p->maxcols);
        if (NULL == p->buf) {
                perror(NULL);
-               exit(EXIT_FAILURE);
+               exit((int)MANDOCLEVEL_SYSERR);
        }
 }
 
 
-static void
-buffera(struct termp *p, const char *word, size_t sz)
-{
-
-       if (p->col + sz >= p->maxcols) 
-               adjbuf(p, p->col + sz);
-
-       memcpy(&p->buf[(int)p->col], word, sz);
-       p->col += sz;
-}
-
-
 static void
 bufferc(struct termp *p, char c)
 {
@@ -617,23 +603,31 @@ encode(struct termp *p, const char *word, size_t sz)
         */
 
        if (TERMFONT_NONE == (f = term_fonttop(p))) {
-               buffera(p, word, sz);
+               if (p->col + sz >= p->maxcols) 
+                       adjbuf(p, p->col + sz);
+               memcpy(&p->buf[(int)p->col], word, sz);
+               p->col += sz;
                return;
        }
 
+       /* Pre-buffer, assuming worst-case. */
+
+       if (p->col + 1 + (sz * 3) >= p->maxcols)
+               adjbuf(p, p->col + 1 + (sz * 3));
+
        for (i = 0; i < (int)sz; i++) {
                if ( ! isgraph((u_char)word[i])) {
-                       bufferc(p, word[i]);
+                       p->buf[(int)p->col++] = word[i];
                        continue;
                }
 
                if (TERMFONT_UNDER == f)
-                       bufferc(p, '_');
+                       p->buf[(int)p->col++] = '_';
                else
-                       bufferc(p, word[i]);
+                       p->buf[(int)p->col++] = word[i];
 
-               bufferc(p, 8);
-               bufferc(p, word[i]);
+               p->buf[(int)p->col++] = 8;
+               p->buf[(int)p->col++] = word[i];
        }
 }
 
@@ -649,10 +643,48 @@ term_len(const struct termp *p, size_t sz)
 size_t
 term_strlen(const struct termp *p, const char *cp)
 {
-       size_t           sz;
+       size_t           sz, ssz, rsz, i;
+       enum roffdeco    d;
+       const char      *seq, *rhs;
 
-       for (sz = 0; *cp; cp++)
-               sz += (*p->width)(p, *cp);
+       for (sz = 0; '\0' != *cp; )
+               /*
+                * Account for escaped sequences within string length
+                * calculations.  This follows the logic in term_word()
+                * as we must calculate the width of produced strings.
+                */
+               if ('\\' == *cp) {
+                       seq = ++cp;
+                       cp += a2roffdeco(&d, &seq, &ssz);
+
+                       switch (d) {
+                       case (DECO_RESERVED):
+                               rhs = chars_res2str
+                                       (p->symtab, seq, ssz, &rsz);
+                               break;
+                       case (DECO_SPECIAL):
+                               /* FALLTHROUGH */
+                       case (DECO_SSPECIAL):
+                               rhs = chars_spec2str
+                                       (p->symtab, seq, ssz, &rsz);
+
+                               /* Allow for one-char escapes. */
+                               if (DECO_SSPECIAL != d || rhs)
+                                       break;
+
+                               rhs = seq;
+                               rsz = ssz;
+                               break;
+                       default:
+                               rhs = NULL;
+                               break;
+                       }
+
+                       if (rhs)
+                               for (i = 0; i < rsz; i++)
+                                       sz += (*p->width)(p, *rhs++);
+               } else
+                       sz += (*p->width)(p, *cp++);
 
        return(sz);
 }