Removed TODO that was fixed by parsing escapes in offset strings.

[mandoc.git] / term.c

diff --git a/term.c b/term.c
index 4ab338d9d60942668d50ed4a9cc6e5fdf01f84ed..b609b0e8615d65391ec2d53ca41f2509eb298d13 100644 (file)
--- a/term.c
+++ b/term.c
@@ -1,6 +1,7 @@
-/*     $Id: term.c,v 1.82 2009/06/22 12:38:07 kristaps Exp $ */
+/*     $Id: term.c,v 1.172 2010/09/23 20:26:00 schwarze Exp $ */

 /*
 /*

- * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
+ * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons <kristaps@bsd.lv>
+ * Copyright (c) 2010 Ingo Schwarze <schwarze@openbsd.org>

  *
  * Permission to use, copy, modify, and distribute this software for any
  * purpose with or without fee is hereby granted, provided that the above
  *
  * Permission to use, copy, modify, and distribute this software for any
  * purpose with or without fee is hereby granted, provided that the above


@@ -14,156 +15,79 @@
  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  */
  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  */

+#ifdef HAVE_CONFIG_H
+#include "config.h"
+#endif
+
+#include <sys/types.h>
+

 #include <assert.h>
 #include <assert.h>

-#include <err.h>
+#include <ctype.h>
+#include <stdint.h>

 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
 
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
 

+#include "mandoc.h"
+#include "chars.h"
+#include "out.h"

 #include "term.h"
 #include "term.h"

-#include "man.h"
-#include "mdoc.h"
-
-extern int               man_run(struct termp *, 
-                               const struct man *);
-extern int               mdoc_run(struct termp *, 
-                               const struct mdoc *);
-
-static struct termp     *term_alloc(enum termenc);
-static void              term_free(struct termp *);
-static void              term_pword(struct termp *, const char *, int);
-static void              term_pescape(struct termp *, 
-                               const char *, int *, int);
-static void              term_nescape(struct termp *,
-                               const char *, size_t);
-static void              term_chara(struct termp *, char);
-static void              term_encodea(struct termp *, char);
-static int               term_isopendelim(const char *, int);
-static int               term_isclosedelim(const char *, int);
-
-
-void *
-ascii_alloc(void)
-{
-
-       return(term_alloc(TERMENC_ASCII));
-}
-
+#include "main.h"

 
 

-int
-terminal_man(void *arg, const struct man *man)
-{
-       struct termp    *p;
-
-       p = (struct termp *)arg;
-       if (NULL == p->symtab)
-               p->symtab = term_ascii2htab();
-
-       return(man_run(p, man));
-}
-
-
-int
-terminal_mdoc(void *arg, const struct mdoc *mdoc)
-{
-       struct termp    *p;
-
-       p = (struct termp *)arg;
-       if (NULL == p->symtab)
-               p->symtab = term_ascii2htab();
-
-       return(mdoc_run(p, mdoc));
-}
+static void              spec(struct termp *, enum roffdeco,
+                               const char *, size_t);
+static void              res(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);

 
 
 void
 
 
 void

-terminal_free(void *arg)
-{
-
-       term_free((struct termp *)arg);
-}
-
-
-static void

 term_free(struct termp *p)
 {
 
        if (p->buf)
                free(p->buf);
 term_free(struct termp *p)
 {
 
        if (p->buf)
                free(p->buf);

-       if (TERMENC_ASCII == p->enc && p->symtab)
-               term_asciifree(p->symtab);
+       if (p->symtab)
+               chars_free(p->symtab);

 
        free(p);
 }
 
 
 
        free(p);
 }
 
 

-static struct termp *
-term_alloc(enum termenc enc)
+void
+term_begin(struct termp *p, term_margin head, 
+               term_margin foot, const void *arg)

 {
 {

-       struct termp *p;

 
 

-       if (NULL == (p = malloc(sizeof(struct termp))))
-               err(1, "malloc");
-       bzero(p, sizeof(struct termp));
-       p->maxrmargin = 78;
-       p->enc = enc;
-       return(p);
+       p->headf = head;
+       p->footf = foot;
+       p->argf = arg;
+       (*p->begin)(p);

 }
 
 
 }
 
 

-static int
-term_isclosedelim(const char *p, int len)
+void
+term_end(struct termp *p)

 {
 
 {
 

-       if (1 != len)
-               return(0);
-
-       switch (*p) {
-       case('.'):
-               /* FALLTHROUGH */
-       case(','):
-               /* FALLTHROUGH */
-       case(';'):
-               /* FALLTHROUGH */
-       case(':'):
-               /* FALLTHROUGH */
-       case('?'):
-               /* FALLTHROUGH */
-       case('!'):
-               /* FALLTHROUGH */
-       case(')'):
-               /* FALLTHROUGH */
-       case(']'):
-               /* FALLTHROUGH */
-       case('}'):
-               return(1);
-       default:
-               break;
-       }
-
-       return(0);
+       (*p->end)(p);

 }
 
 
 }
 
 

-static int
-term_isopendelim(const char *p, int len)
+struct termp *
+term_alloc(enum termenc enc)

 {
 {

+       struct termp    *p;

 
 

-       if (1 != len)
-               return(0);
-
-       switch (*p) {
-       case('('):
-               /* FALLTHROUGH */
-       case('['):
-               /* FALLTHROUGH */
-       case('{'):
-               return(1);
-       default:
-               break;
+       p = calloc(1, sizeof(struct termp));
+       if (NULL == p) {
+               perror(NULL);
+               exit((int)MANDOCLEVEL_SYSERR);

        }
 
        }
 

-       return(0);
+       p->enc = enc;
+       return(p);

 }
 
 
 }
 
 


@@ -171,13 +95,10 @@ term_isopendelim(const char *p, int len)
  * Flush a line of text.  A "line" is loosely defined as being something
  * that should be followed by a newline, regardless of whether it's
  * broken apart by newlines getting there.  A line can also be a
  * Flush a line of text.  A "line" is loosely defined as being something
  * that should be followed by a newline, regardless of whether it's
  * broken apart by newlines getting there.  A line can also be a

- * fragment of a columnar list.
- *
- * Specifically, a line is whatever's in p->buf of length p->col, which
- * is zeroed after this function returns.
+ * fragment of a columnar list (`Bl -tag' or `Bl -column'), which does
+ * not have a trailing newline.

  *
  *

- * The variables TERMP_NOLPAD, TERMP_LITERAL and TERMP_NOBREAK are of
- * critical importance here.  Their behaviour follows:
+ * The following flags may be specified:

  *
  *  - TERMP_NOLPAD: when beginning to write the line, don't left-pad the
  *    offset value.  This is useful when doing columnar lists where the
  *
  *  - TERMP_NOLPAD: when beginning to write the line, don't left-pad the
  *    offset value.  This is useful when doing columnar lists where the


@@ -187,54 +108,75 @@ term_isopendelim(const char *p, int len)
  *    columns.  In short: don't print a newline and instead pad to the
  *    right margin.  Used in conjunction with TERMP_NOLPAD.
  *
  *    columns.  In short: don't print a newline and instead pad to the
  *    right margin.  Used in conjunction with TERMP_NOLPAD.
  *

- *  - TERMP_NONOBREAK: don't newline when TERMP_NOBREAK is specified.
+ *  - TERMP_TWOSPACE: when padding, make sure there are at least two
+ *    space characters of padding.  Otherwise, rather break the line.
+ *
+ *  - TERMP_DANGLE: don't newline when TERMP_NOBREAK is specified and
+ *    the line is overrun, and don't pad-right if it's underrun.
+ *
+ *  - TERMP_HANG: like TERMP_DANGLE, but doesn't newline when
+ *    overruning, instead save the position and continue at that point
+ *    when the next invocation.

  *
  *  In-line line breaking:
  *
  *  If TERMP_NOBREAK is specified and the line overruns the right
  *  margin, it will break and pad-right to the right margin after
  *  writing.  If maxrmargin is violated, it will break and continue
  *
  *  In-line line breaking:
  *
  *  If TERMP_NOBREAK is specified and the line overruns the right
  *  margin, it will break and pad-right to the right margin after
  *  writing.  If maxrmargin is violated, it will break and continue

- *  writing from the right-margin, which will lead to the above
- *  scenario upon exit.
- *
- *  Otherwise, the line will break at the right margin.  Extremely long
- *  lines will cause the system to emit a warning (TODO: hyphenate, if
- *  possible).
- *
- *  FIXME: newline breaks occur (in groff) also occur when a single
- *  space follows a NOBREAK!
+ *  writing from the right-margin, which will lead to the above scenario
+ *  upon exit.  Otherwise, the line will break at the right margin.

  */
 void
 term_flushln(struct termp *p)
 {
  */
 void
 term_flushln(struct termp *p)
 {

-       int              i, j;
-       size_t           vbl, vsz, vis, maxvis, mmax, bp;
+       int              i;     /* current input position in p->buf */
+       size_t           vis;   /* current visual position on output */
+       size_t           vbl;   /* number of blanks to prepend to output */
+       size_t           vend;  /* end of word visual position on output */
+       size_t           bp;    /* visual right border position */
+       size_t           dv;    /* temporary for visual pos calculations */
+       int              j;     /* temporary loop index for p->buf */
+       int              jhy;   /* last hyph before overflow w/r/t j */
+       size_t           maxvis; /* output position of visible boundary */
+       size_t           mmax; /* used in calculating bp */

 
        /*
         * First, establish the maximum columns of "visible" content.
         * This is usually the difference between the right-margin and
         * an indentation, but can be, for tagged lists or columns, a
 
        /*
         * First, establish the maximum columns of "visible" content.
         * This is usually the difference between the right-margin and
         * an indentation, but can be, for tagged lists or columns, a

-        * small set of values.
+        * small set of values. 

         */
 
        assert(p->offset < p->rmargin);
         */
 
        assert(p->offset < p->rmargin);

-       maxvis = p->rmargin - p->offset;
-       mmax = p->maxrmargin - p->offset;
+
+       maxvis = (int)(p->rmargin - p->offset) - p->overstep < 0 ?
+               /* LINTED */ 
+               0 : p->rmargin - p->offset - p->overstep;
+       mmax = (int)(p->maxrmargin - p->offset) - p->overstep < 0 ?
+               /* LINTED */
+               0 : p->maxrmargin - p->offset - p->overstep;
+

        bp = TERMP_NOBREAK & p->flags ? mmax : maxvis;
        bp = TERMP_NOBREAK & p->flags ? mmax : maxvis;

-       vis = 0;

 
        /*
 
        /*

-        * If in the standard case (left-justified), then begin with our
-        * indentation, otherwise (columns, etc.) just start spitting
-        * out text.
+        * Indent the first line of a paragraph.

         */
         */

+       vbl = p->flags & TERMP_NOLPAD ? 0 : p->offset;

 
 

-       if ( ! (p->flags & TERMP_NOLPAD))
-               /* LINTED */
-               for (j = 0; j < (int)p->offset; j++)
-                       putchar(' ');
+       vis = vend = i = 0;
+
+       while (i < (int)p->col) {
+               /*
+                * Handle literal tab characters: collapse all
+                * subsequent tabs into a single huge set of spaces.
+                */
+               while (i < (int)p->col && '\t' == p->buf[i]) {
+                       vend = (vis / p->tabwidth + 1) * p->tabwidth;
+                       vbl += vend - vis;
+                       vis = vend;
+                       i++;
+               }

 
 

-       for (i = 0; i < (int)p->col; i++) {

                /*
                 * Count up visible word characters.  Control sequences
                 * (starting with the CSI) aren't counted.  A space
                /*
                 * Count up visible word characters.  Control sequences
                 * (starting with the CSI) aren't counted.  A space


@@ -243,83 +185,143 @@ term_flushln(struct termp *p)
                 */
 
                /* LINTED */
                 */
 
                /* LINTED */

-               for (j = i, vsz = 0; j < (int)p->col; j++) {
-                       if (' ' == p->buf[j])
+               for (j = i, jhy = 0; j < (int)p->col; j++) {
+                       if ((j && ' ' == p->buf[j]) || '\t' == p->buf[j])

                                break;
                                break;

-                       else if (8 == p->buf[j])
-                               j += 1;
-                       else
-                               vsz++;
-               }

 
 

-               /*
-                * Choose the number of blanks to prepend: no blank at the
-                * beginning of a line, one between words -- but do not
-                * actually write them yet.
-                */
-               vbl = (size_t)(0 == vis ? 0 : 1);
+                       /* Back over the the last printed character. */
+                       if (8 == p->buf[j]) {
+                               assert(j);
+                               vend -= (*p->width)(p, p->buf[j - 1]);
+                               continue;
+                       }
+
+                       /* Regular word. */
+                       /* Break at the hyphen point if we overrun. */
+                       if (vend > vis && vend < bp && 
+                                       ASCII_HYPH == p->buf[j])
+                               jhy = j;
+
+                       vend += (*p->width)(p, p->buf[j]);
+               }

 
                /*
                 * Find out whether we would exceed the right margin.
 
                /*
                 * Find out whether we would exceed the right margin.

-                * If so, break to the next line.  (TODO: hyphenate)
-                * Otherwise, write the chosen number of blanks now.
+                * If so, break to the next line.

                 */
                 */

-               if (vis && vis + vbl + vsz > bp) {
-                       putchar('\n');
+               if (vend > bp && 0 == jhy && vis > 0) {
+                       vend -= vis;
+                       (*p->endline)(p);

                        if (TERMP_NOBREAK & p->flags) {
                        if (TERMP_NOBREAK & p->flags) {

-                               for (j = 0; j < (int)p->rmargin; j++)
-                                       putchar(' ');
-                               vis = p->rmargin - p->offset;
+                               p->viscol = p->rmargin;
+                               (*p->advance)(p, p->rmargin);
+                               vend += p->rmargin - p->offset;

                        } else {
                        } else {

-                               for (j = 0; j < (int)p->offset; j++)
-                                       putchar(' ');
-                               vis = 0;
+                               p->viscol = 0;
+                               vbl = p->offset;

                        }
                        }

-               } else {
-                       for (j = 0; j < (int)vbl; j++)
-                               putchar(' ');
-                       vis += vbl;
+
+                       /* Remove the p->overstep width. */
+
+                       bp += (int)/* LINTED */
+                               p->overstep;
+                       p->overstep = 0;

                }
 
                }
 

-               /*
-                * Finally, write out the word.
-                */
+               /* Write out the [remaining] word. */

                for ( ; i < (int)p->col; i++) {
                for ( ; i < (int)p->col; i++) {

-                       if (' ' == p->buf[i])
+                       if (vend > bp && jhy > 0 && i > jhy)
+                               break;
+                       if ('\t' == p->buf[i])

                                break;
                                break;

-                       putchar(p->buf[i]);
+                       if (' ' == p->buf[i]) {
+                               j = i;
+                               while (' ' == p->buf[i])
+                                       i++;
+                               dv = (i - j) * (*p->width)(p, ' ');
+                               vbl += dv;
+                               vend += dv;
+                               break;
+                       }
+                       if (ASCII_NBRSP == p->buf[i]) {
+                               vbl += (*p->width)(p, ' ');
+                               continue;
+                       }
+
+                       /*
+                        * Now we definitely know there will be
+                        * printable characters to output,
+                        * so write preceding white space now.
+                        */
+                       if (vbl) {
+                               (*p->advance)(p, vbl);
+                               p->viscol += vbl;
+                               vbl = 0;
+                       }
+
+                       if (ASCII_HYPH == p->buf[i]) {
+                               (*p->letter)(p, '-');
+                               p->viscol += (*p->width)(p, '-');
+                       } else {
+                               (*p->letter)(p, p->buf[i]);
+                               p->viscol += (*p->width)(p, p->buf[i]);
+                       }

                }
                }

-               vis += vsz;
+               vis = vend;

        }
 
        /*
        }
 
        /*

-        * If we've overstepped our maximum visible no-break space, then
-        * cause a newline and offset at the right margin.
+        * If there was trailing white space, it was not printed;
+        * so reset the cursor position accordingly.

         */
         */

+       vis -= vbl;

 
 

-       if ((TERMP_NOBREAK & p->flags) && vis >= maxvis) {
-               if ( ! (TERMP_NONOBREAK & p->flags)) {
-                       putchar('\n');
-                       for (i = 0; i < (int)p->rmargin; i++)
-                               putchar(' ');
-               }
-               p->col = 0;
+       p->col = 0;
+       p->overstep = 0;
+
+       if ( ! (TERMP_NOBREAK & p->flags)) {
+               p->viscol = 0;
+               (*p->endline)(p);

                return;
        }
 
                return;
        }
 

-       /*
-        * If we're not to right-marginalise it (newline), then instead
-        * pad to the right margin and stay off.
-        */
+       if (TERMP_HANG & p->flags) {
+               /* We need one blank after the tag. */
+               p->overstep = /* LINTED */
+                       vis - maxvis + (*p->width)(p, ' ');

 
 

-       if (p->flags & TERMP_NOBREAK) {
-               if ( ! (TERMP_NONOBREAK & p->flags))
-                       for ( ; vis < maxvis; vis++)
-                               putchar(' ');
-       } else
-               putchar('\n');
+               /*
+                * Behave exactly the same way as groff:
+                * If we have overstepped the margin, temporarily move
+                * it to the right and flag the rest of the line to be
+                * shorter.
+                * If we landed right at the margin, be happy.
+                * If we are one step before the margin, temporarily
+                * move it one step LEFT and flag the rest of the line
+                * to be longer.
+                */
+               if (p->overstep >= -1) {
+                       assert((int)maxvis + p->overstep >= 0);
+                       /* LINTED */
+                       maxvis += p->overstep;
+               } else
+                       p->overstep = 0;
+
+       } else if (TERMP_DANGLE & p->flags)
+               return;

 
 

-       p->col = 0;
+       /* Right-pad. */
+       if (maxvis > vis + /* LINTED */
+                       ((TERMP_TWOSPACE & p->flags) ? 
+                        (*p->width)(p, ' ') : 0)) {
+               p->viscol += maxvis - vis;
+               (*p->advance)(p, maxvis - vis);
+               vis += (maxvis - vis);
+       } else {        /* ...or newline break. */
+               (*p->endline)(p);
+               p->viscol = p->rmargin;
+               (*p->advance)(p, p->rmargin);
+       }

 }
 
 
 }
 
 


@@ -333,7 +335,7 @@ term_newln(struct termp *p)
 {
 
        p->flags |= TERMP_NOSPACE;
 {
 
        p->flags |= TERMP_NOSPACE;

-       if (0 == p->col) {
+       if (0 == p->col && 0 == p->viscol) {

                p->flags &= ~TERMP_NOLPAD;
                return;
        }
                p->flags &= ~TERMP_NOLPAD;
                return;
        }


@@ -353,224 +355,388 @@ term_vspace(struct termp *p)
 {
 
        term_newln(p);
 {
 
        term_newln(p);

-       putchar('\n');
+       p->viscol = 0;
+       (*p->endline)(p);
+}
+
+
+static void
+spec(struct termp *p, enum roffdeco d, const char *word, size_t len)
+{
+       const char      *rhs;
+       size_t           sz;
+
+       rhs = chars_spec2str(p->symtab, word, len, &sz);
+       if (rhs) 
+               encode(p, rhs, sz);
+       else if (DECO_SSPECIAL == d)
+               encode(p, word, len);
+}
+
+
+static void
+res(struct termp *p, const char *word, size_t len)
+{
+       const char      *rhs;
+       size_t           sz;
+
+       rhs = chars_res2str(p->symtab, word, len, &sz);
+       if (rhs)
+               encode(p, rhs, sz);

 }
 
 
 }
 
 

-/*
- * Break apart a word into "pwords" (partial-words, usually from
- * breaking up a phrase into individual words) and, eventually, put them
- * into the output buffer.  If we're a literal word, then don't break up
- * the word and put it verbatim into the output buffer.
- */

 void
 void

-term_word(struct termp *p, const char *word)
+term_fontlast(struct termp *p)

 {
 {

-       int              i, j, len;
+       enum termfont    f;

 
 

-       len = (int)strlen(word);
+       f = p->fontl;
+       p->fontl = p->fontq[p->fonti];
+       p->fontq[p->fonti] = f;
+}

 
 

-       if (p->flags & TERMP_LITERAL) {
-               term_pword(p, word, len);
-               return;
-       }

 
 

-       /* LINTED */
-       for (j = i = 0; i < len; i++) {
-               if (' ' != word[i]) {
-                       j++;
-                       continue;
-               } 
-               
-               /* Escaped spaces don't delimit... */
-               if (i && ' ' == word[i] && '\\' == word[i - 1]) {
-                       j++;
-                       continue;
-               }
+void
+term_fontrepl(struct termp *p, enum termfont f)
+{

 
 

-               if (0 == j)
-                       continue;
-               assert(i >= j);
-               term_pword(p, &word[i - j], j);
-               j = 0;
-       }
-       if (j > 0) {
-               assert(i >= j);
-               term_pword(p, &word[i - j], j);
-       }
+       p->fontl = p->fontq[p->fonti];
+       p->fontq[p->fonti] = f;

 }
 
 
 }
 
 

-/*
- * Determine the symbol indicated by an escape sequences, that is, one
- * starting with a backslash.  Once done, we pass this value into the
- * output buffer by way of the symbol table.
- */
-static void
-term_nescape(struct termp *p, const char *word, size_t len)
+void
+term_fontpush(struct termp *p, enum termfont f)

 {
 {

-       const char      *rhs;
-       size_t           sz;
-       int              i;

 
 

-       if ((rhs = term_a2ascii(p->symtab, word, len, &sz))) 
-               for (i = 0; i < (int)sz; i++) 
-                       term_encodea(p, rhs[i]);
+       assert(p->fonti + 1 < 10);
+       p->fontl = p->fontq[p->fonti];
+       p->fontq[++p->fonti] = f;

 }
 
 
 }
 
 

-/*
- * Handle an escape sequence: determine its length and pass it to the
- * escape-symbol look table.  Note that we assume mdoc(3) has validated
- * the escape sequence (we assert upon badly-formed escape sequences).
- */
-static void
-term_pescape(struct termp *p, const char *word, int *i, int len)
+const void *
+term_fontq(struct termp *p)

 {
 {

-       int              j;

 
 

-       if (++(*i) >= len)
-               return;
+       return(&p->fontq[p->fonti]);
+}

 
 

-       if ('(' == word[*i]) {
-               (*i)++;
-               if (*i + 1 >= len)
-                       return;

 
 

-               term_nescape(p, &word[*i], 2);
-               (*i)++;
-               return;
+enum termfont
+term_fonttop(struct termp *p)
+{
+
+       return(p->fontq[p->fonti]);
+}
+
+
+void
+term_fontpopq(struct termp *p, const void *key)
+{
+
+       while (p->fonti >= 0 && key != &p->fontq[p->fonti])
+               p->fonti--;
+       assert(p->fonti >= 0);
+}
+
+
+void
+term_fontpop(struct termp *p)
+{
+
+       assert(p->fonti);
+       p->fonti--;
+}
+

 
 

-       } else if ('*' == word[*i]) { 
-               (*i)++;
-               if (*i >= len)
-                       return;
-
-               switch (word[*i]) {
-               case ('('):
-                       (*i)++;
-                       if (*i + 1 >= len)
-                               return;
-
-                       term_nescape(p, &word[*i], 2);
-                       (*i)++;
-                       return;
-               case ('['):
+/*
+ * Handle pwords, partial words, which may be either a single word or a
+ * phrase that cannot be broken down (such as a literal string).  This
+ * handles word styling.
+ */
+void
+term_word(struct termp *p, const char *word)
+{
+       const char      *sv, *seq;
+       size_t           ssz;
+       enum roffdeco    deco;
+
+       sv = word;
+
+       if (word[0] && '\0' == word[1])
+               switch (word[0]) {
+               case('.'):
+                       /* FALLTHROUGH */
+               case(','):
+                       /* FALLTHROUGH */
+               case(';'):
+                       /* FALLTHROUGH */
+               case(':'):
+                       /* FALLTHROUGH */
+               case('?'):
+                       /* FALLTHROUGH */
+               case('!'):
+                       /* FALLTHROUGH */
+               case(')'):
+                       /* FALLTHROUGH */
+               case(']'):
+                       if ( ! (TERMP_IGNDELIM & p->flags))
+                               p->flags |= TERMP_NOSPACE;

                        break;
                default:
                        break;
                default:

-                       term_nescape(p, &word[*i], 1);
-                       return;
-               }
-       
-       } else if ('f' == word[*i]) {
-               (*i)++;
-               if (*i >= len)
-                       return;
-               switch (word[*i]) {
-               case ('B'):
-                       p->flags |= TERMP_BOLD;

                        break;
                        break;

-               case ('I'):
-                       p->flags |= TERMP_UNDER;
+               }
+
+       if ( ! (TERMP_NOSPACE & p->flags)) {
+               if ( ! (TERMP_KEEP & p->flags)) {
+                       if (TERMP_PREKEEP & p->flags)
+                               p->flags |= TERMP_KEEP;
+                       bufferc(p, ' ');
+                       if (TERMP_SENTENCE & p->flags)
+                               bufferc(p, ' ');
+               } else
+                       bufferc(p, ASCII_NBRSP);
+       }
+
+       if ( ! (p->flags & TERMP_NONOSPACE))
+               p->flags &= ~TERMP_NOSPACE;
+       else
+               p->flags |= TERMP_NOSPACE;
+
+       p->flags &= ~TERMP_SENTENCE;
+
+       while (*word) {
+               if ((ssz = strcspn(word, "\\")) > 0)
+                       encode(p, word, ssz);
+
+               word += ssz;
+               if ('\\' != *word)
+                       continue;
+
+               seq = ++word;
+               word += a2roffdeco(&deco, &seq, &ssz);
+
+               switch (deco) {
+               case (DECO_RESERVED):
+                       res(p, seq, ssz);

                        break;
                        break;

-               case ('P'):
+               case (DECO_SPECIAL):

                        /* FALLTHROUGH */
                        /* FALLTHROUGH */

-               case ('R'):
-                       p->flags &= ~TERMP_STYLE;
+               case (DECO_SSPECIAL):
+                       spec(p, deco, seq, ssz);
+                       break;
+               case (DECO_BOLD):
+                       term_fontrepl(p, TERMFONT_BOLD);
+                       break;
+               case (DECO_ITALIC):
+                       term_fontrepl(p, TERMFONT_UNDER);
+                       break;
+               case (DECO_ROMAN):
+                       term_fontrepl(p, TERMFONT_NONE);
+                       break;
+               case (DECO_PREVIOUS):
+                       term_fontlast(p);

                        break;
                default:
                        break;
                }
                        break;
                default:
                        break;
                }

-               return;

 
 

-       } else if ('[' != word[*i]) {
-               term_nescape(p, &word[*i], 1);
-               return;
+               if (DECO_NOSPACE == deco && '\0' == *word)
+                       p->flags |= TERMP_NOSPACE;

        }
 
        }
 

-       (*i)++;
-       for (j = 0; word[*i] && ']' != word[*i]; (*i)++, j++)
-               /* Loop... */ ;
+       /* 
+        * Note that we don't process the pipe: the parser sees it as
+        * punctuation, but we don't in terms of typography.
+        */
+       if (sv[0] && '\0' == sv[1])
+               switch (sv[0]) {
+               case('('):
+                       /* FALLTHROUGH */
+               case('['):
+                       p->flags |= TERMP_NOSPACE;
+                       break;
+               default:
+                       break;
+               }
+}

 
 

-       if (0 == word[*i])
-               return;

 
 

-       term_nescape(p, &word[*i - j], (size_t)j);
+static void
+adjbuf(struct termp *p, size_t sz)
+{
+
+       if (0 == p->maxcols)
+               p->maxcols = 1024;
+       while (sz >= p->maxcols)
+               p->maxcols <<= 2;
+
+       p->buf = realloc(p->buf, p->maxcols);
+       if (NULL == p->buf) {
+               perror(NULL);
+               exit((int)MANDOCLEVEL_SYSERR);
+       }

 }
 
 
 }
 
 

-/*
- * Handle pwords, partial words, which may be either a single word or a
- * phrase that cannot be broken down (such as a literal string).  This
- * handles word styling.
- */

 static void
 static void

-term_pword(struct termp *p, const char *word, int len)
+bufferc(struct termp *p, char c)

 {
 {

-       int              i;

 
 

-       if (term_isclosedelim(word, len))
-               if ( ! (TERMP_IGNDELIM & p->flags))
-                       p->flags |= TERMP_NOSPACE;
+       if (p->col + 1 >= p->maxcols)
+               adjbuf(p, p->col + 1);

 
 

-       if ( ! (TERMP_NOSPACE & p->flags))
-               term_chara(p, ' ');
+       p->buf[(int)p->col++] = c;
+}

 
 

-       if ( ! (p->flags & TERMP_NONOSPACE))
-               p->flags &= ~TERMP_NOSPACE;

 
 

-       /* 
-        * If ANSI (word-length styling), then apply our style now,
-        * before the word.
+static void
+encode(struct termp *p, const char *word, size_t sz)
+{
+       enum termfont     f;
+       int               i;
+
+       /*
+        * Encode and buffer a string of characters.  If the current
+        * font mode is unset, buffer directly, else encode then buffer
+        * character by character.

         */
 
         */
 

-       for (i = 0; i < len; i++) 
-               if ('\\' == word[i]) 
-                       term_pescape(p, word, &i, len);
+       if (TERMFONT_NONE == (f = term_fonttop(p))) {
+               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])) {
+                       p->buf[(int)p->col++] = word[i];
+                       continue;
+               }
+
+               if (TERMFONT_UNDER == f)
+                       p->buf[(int)p->col++] = '_';

                else
                else

-                       term_encodea(p, word[i]);
+                       p->buf[(int)p->col++] = word[i];

 
 

-       if (term_isopendelim(word, len))
-               p->flags |= TERMP_NOSPACE;
+               p->buf[(int)p->col++] = 8;
+               p->buf[(int)p->col++] = word[i];
+       }

 }
 
 
 }
 
 

-/*
- * Insert a single character into the line-buffer.  If the buffer's
- * space is exceeded, then allocate more space by doubling the buffer
- * size.
- */
-static void
-term_chara(struct termp *p, char c)
+size_t
+term_len(const struct termp *p, size_t sz)

 {
 {

-       size_t           s;
-
-       if (p->col + 1 >= p->maxcols) {
-               if (0 == p->maxcols)
-                       p->maxcols = 256;
-               s = p->maxcols * 2;
-               p->buf = realloc(p->buf, s);
-               if (NULL == p->buf)
-                       err(1, "realloc");
-               p->maxcols = s;
-       }
-       p->buf[(int)(p->col)++] = c;
+
+       return((*p->width)(p, ' ') * sz);

 }
 
 
 }
 
 

-static void
-term_encodea(struct termp *p, char c)
+size_t
+term_strlen(const struct termp *p, const char *cp)

 {
 {

+       size_t           sz, ssz, rsz, i;
+       enum roffdeco    d;
+       const char      *seq, *rhs;

 
 

-       if (TERMP_STYLE & p->flags) {
-               if (TERMP_BOLD & p->flags) {
-                       term_chara(p, c);
-                       term_chara(p, 8);
-               }
-               if (TERMP_UNDER & p->flags) {
-                       term_chara(p, '_');
-                       term_chara(p, 8);
-               }
+       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);
+}
+
+
+/* ARGSUSED */
+size_t
+term_vspan(const struct termp *p, const struct roffsu *su)
+{
+       double           r;
+
+       switch (su->unit) {
+       case (SCALE_CM):
+               r = su->scale * 2;
+               break;
+       case (SCALE_IN):
+               r = su->scale * 6;
+               break;
+       case (SCALE_PC):
+               r = su->scale;
+               break;
+       case (SCALE_PT):
+               r = su->scale / 8;
+               break;
+       case (SCALE_MM):
+               r = su->scale / 1000;
+               break;
+       case (SCALE_VS):
+               r = su->scale;
+               break;
+       default:
+               r = su->scale - 1;
+               break;

        }
        }

-       term_chara(p, c);
+
+       if (r < 0.0)
+               r = 0.0;
+       return(/* LINTED */(size_t)
+                       r);
+}
+
+
+size_t
+term_hspan(const struct termp *p, const struct roffsu *su)
+{
+       double           v;
+
+       v = ((*p->hspan)(p, su));
+       if (v < 0.0)
+               v = 0.0;
+       return((size_t) /* LINTED */
+                       v);

 }
 }