Cope with another one of the many kinds of DocBook stupidity:

[mandoc.git] / term.c

diff --git a/term.c b/term.c
index 4d33ce8dd426529193fbff066dd8f9d4bbca5944..75c12f4af21e9576cfead9ad040785a20dd39886 100644 (file)
--- a/term.c
+++ b/term.c
@@ -1,6 +1,7 @@
-/*     $Id: term.c,v 1.90 2009/07/19 21:26:27 kristaps Exp $ */
+/*     $Id: term.c,v 1.244 2015/01/31 00:12:41 schwarze Exp $ */

 /*
 /*

- * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
+ * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
+ * Copyright (c) 2010-2015 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


@@ -14,238 +15,138 @@
  * 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.
  */

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

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

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

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

+#include "mandoc.h"
+#include "mandoc_aux.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_pescape(struct termp *, const char **);
-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 *);
-static int               term_isclosedelim(const char *);
-
-
-void *
-ascii_alloc(void)
-{
-
-       return(term_alloc(TERMENC_ASCII));
-}
-
-
-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();
+#include "main.h"

 
 

-       return(mdoc_run(p, mdoc));
-}
+static size_t           cond_width(const struct termp *, int, int *);
+static void             adjbuf(struct termp *p, size_t);
+static void             bufferc(struct termp *, char);
+static void             encode(struct termp *, const char *, size_t);
+static void             encode1(struct termp *, int);

 
 
 void
 
 
 void

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

 term_free(struct termp *p)
 {
 
 term_free(struct termp *p)
 {
 

-       if (p->buf)
-               free(p->buf);
-       if (TERMENC_ASCII == p->enc && p->symtab)
-               term_asciifree(p->symtab);
-
+       free(p->buf);
+       free(p->fontq);

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

-
-static struct termp *
-term_alloc(enum termenc enc)
-{
-       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);
-}
-
-
-static int
-term_isclosedelim(const char *p)
+void
+term_begin(struct termp *p, term_margin head,
+               term_margin foot, const void *arg)

 {
 
 {
 

-       if ( ! (*p && 0 == *(p + 1)))
-               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->headf = head;
+       p->footf = foot;
+       p->argf = arg;
+       (*p->begin)(p);

 }
 
 }
 

-
-static int
-term_isopendelim(const char *p)
+void
+term_end(struct termp *p)

 {
 
 {
 

-       if ( ! (*p && 0 == *(p + 1)))
-               return(0);
-
-       switch (*p) {
-       case('('):
-               /* FALLTHROUGH */
-       case('['):
-               /* FALLTHROUGH */
-       case('{'):
-               return(1);
-       default:
-               break;
-       }
-
-       return(0);
+       (*p->end)(p);

 }
 
 }
 

-

 /*
 /*

- * 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.
- *
- * The usage of termp:flags is as follows:
- *
- *  - TERMP_NOLPAD: when beginning to write the line, don't left-pad the
- *    offset value.  This is useful when doing columnar lists where the
- *    prior column has right-padded.
- *
- *  - TERMP_NOBREAK: this is the most important and is used when making
- *    columns.  In short: don't print a newline and instead pad to the
- *    right margin.  Used in conjunction with TERMP_NOLPAD.
+ * Flush a chunk of text.  By default, break the output line each time
+ * the right margin is reached, and continue output on the next line
+ * at the same offset as the chunk itself.  By default, also break the
+ * output line at the end of the chunk.
+ * The following flags may be specified:

  *
  *

- *  - 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
- *  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 (try `Bl -tag')
- *
- *  FIXME: there's a newline error where a `Bl -diag' will have a
- *  trailing newline if the line is exactly 73 chars long.
+ *  - TERMP_NOBREAK: Do not break the output line at the right margin,
+ *    but only at the max right margin.  Also, do not break the output
+ *    line at the end of the chunk, such that the next call can pad to
+ *    the next column.  However, if less than p->trailspace blanks,
+ *    which can be 0, 1, or 2, remain to the right margin, the line
+ *    will be broken.
+ *  - TERMP_BRIND: If the chunk does not fit and the output line has
+ *    to be broken, start the next line at the right margin instead
+ *    of at the offset.  Used together with TERMP_NOBREAK for the tags
+ *    in various kinds of tagged lists.
+ *  - TERMP_DANGLE: Do not break the output line at the right margin,
+ *    append the next chunk after it even if this one is too long.
+ *    To be used together with TERMP_NOBREAK.
+ *  - TERMP_HANG: Like TERMP_DANGLE, and also suppress padding before
+ *    the next chunk if this column is not full.

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

-       int              i, j;
-       size_t           vbl, vsz, vis, maxvis, mmax, bp;
-       static int       sv = -1;
+       size_t           i;     /* current input position in p->buf */
+       int              ntab;  /* number of tabs to prepend */
+       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 */
+       size_t           j;     /* temporary loop index for p->buf */
+       size_t           jhy;   /* last hyph before overflow w/r/t j */
+       size_t           maxvis; /* output position of visible boundary */

 
        /*
         * 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.
 
        /*
         * 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.

+        *
+        * The following unsigned-signed subtractions look strange,
+        * but they are actually correct.  If the int p->overstep
+        * is negative, it gets sign extended.  Subtracting that
+        * very large size_t effectively adds a small number to dv.

         */
         */

+       dv = p->rmargin > p->offset ? p->rmargin - p->offset : 0;
+       maxvis = (int)dv > p->overstep ? dv - (size_t)p->overstep : 0;

 
 

-       assert(p->offset < p->rmargin);
-       maxvis = p->rmargin - p->offset;
-       mmax = p->maxrmargin - p->offset;
-       bp = TERMP_NOBREAK & p->flags ? mmax : maxvis;
-       vis = 0;
-
-       if (sv >= 0) {
-               vis = (size_t)sv;
-               sv = -1;
-       }
+       if (p->flags & TERMP_NOBREAK) {
+               dv = p->maxrmargin > p->offset ?
+                    p->maxrmargin - p->offset : 0;
+               bp = (int)dv > p->overstep ?
+                    dv - (size_t)p->overstep : 0;
+       } else
+               bp = maxvis;

 
        /*
 
        /*

-        * If in the standard case (left-justified), then begin with our
-        * indentation, otherwise (columns, etc.) just start spitting
-        * out text.
+        * Calculate the required amount of padding.

         */
         */

+       vbl = p->offset + p->overstep > p->viscol ?
+             p->offset + p->overstep - p->viscol : 0;

 
 

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


@@ -253,91 +154,150 @@ term_flushln(struct termp *p)
                 * space is printed according to regular spacing rules).
                 */
 
                 * space is printed according to regular spacing rules).
                 */
 

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

                                break;
                                break;

-                       else if (8 == p->buf[j])
-                               vsz--;
-                       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] ||
+                            ASCII_BREAK == p->buf[j]))
+                               jhy = j;
+
+                       /*
+                        * Hyphenation now decided, put back a real
+                        * hyphen such that we get the correct width.
+                        */
+                       if (ASCII_HYPH == p->buf[j])
+                               p->buf[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 (TERMP_NOBREAK & p->flags) {
-                               for (j = 0; j < (int)p->rmargin; j++)
-                                       putchar(' ');
-                               vis = p->rmargin - p->offset;
-                       } else {
-                               for (j = 0; j < (int)p->offset; j++)
-                                       putchar(' ');
-                               vis = 0;
-                       }
-               } else {
-                       for (j = 0; j < (int)vbl; j++)
-                               putchar(' ');
-                       vis += vbl;
+               if (vend > bp && 0 == jhy && vis > 0) {
+                       vend -= vis;
+                       (*p->endline)(p);
+                       p->viscol = 0;
+                       if (TERMP_BRIND & p->flags) {
+                               vbl = p->rmargin;
+                               vend += p->rmargin;
+                               vend -= p->offset;
+                       } else
+                               vbl = p->offset;
+
+                       /* use pending tabs on the new line */
+
+                       if (0 < ntab)
+                               vbl += ntab * p->tabwidth;
+
+                       /*
+                        * Remove the p->overstep width.
+                        * Again, if p->overstep is negative,
+                        * sign extension does the right thing.
+                        */
+
+                       bp += (size_t)p->overstep;
+                       p->overstep = 0;

                }
 
                }
 

-               /*
-                * Finally, write out the word.
-                */
-               for ( ; i < (int)p->col; i++) {
-                       if (' ' == p->buf[i])
+               /* Write out the [remaining] word. */
+               for ( ; i < p->col; i++) {
+                       if (vend > bp && jhy > 0 && i > jhy)
+                               break;
+                       if ('\t' == p->buf[i])
+                               break;
+                       if (' ' == p->buf[i]) {
+                               j = i;
+                               while (i < p->col && ' ' == p->buf[i])
+                                       i++;
+                               dv = (i - j) * (*p->width)(p, ' ');
+                               vbl += dv;
+                               vend += dv;

                                break;
                                break;

-                       putchar(p->buf[i]);
+                       }
+                       if (ASCII_NBRSP == p->buf[i]) {
+                               vbl += (*p->width)(p, ' ');
+                               continue;
+                       }
+                       if (ASCII_BREAK == p->buf[i])
+                               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;
+                       }
+
+                       (*p->letter)(p, p->buf[i]);
+                       if (8 == p->buf[i])
+                               p->viscol -= (*p->width)(p, p->buf[i-1]);
+                       else
+                               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.

         */
         */

+       if (vis > vbl)
+               vis -= vbl;
+       else
+               vis = 0;

 
 

-       if ((TERMP_NOBREAK & p->flags) && vis >= maxvis) {
-               if ( ! (TERMP_DANGLE & p->flags) &&
-                               ! (TERMP_HANG & p->flags)) {
-                       putchar('\n');
-                       for (i = 0; i < (int)p->rmargin; i++)
-                               putchar(' ');
-               }
-               if (TERMP_HANG & p->flags)
-                       sv = (int)(vis - maxvis);
-               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) {
+               p->overstep += (int)(p->offset + vis - p->rmargin +
+                   p->trailspace * (*p->width)(p, ' '));

 
 

-       if (p->flags & TERMP_NOBREAK) {
-               if ( ! (TERMP_DANGLE & p->flags))
-                       for ( ; vis < maxvis; vis++)
-                               putchar(' ');
-       } else
-               putchar('\n');
+               /*
+                * If we have overstepped the margin, temporarily move
+                * it to the right and flag the rest of the line to be
+                * shorter.
+                * If there is a request to keep the columns together,
+                * allow negative overstep when the column is not full.
+                */
+               if (p->trailspace && p->overstep < 0)
+                       p->overstep = 0;
+               return;

 
 

-       p->col = 0;
-}
+       } else if (TERMP_DANGLE & p->flags)
+               return;

 
 

+       /* If the column was overrun, break the line. */
+       if (maxvis < vis + p->trailspace * (*p->width)(p, ' ')) {
+               (*p->endline)(p);
+               p->viscol = 0;
+       }
+}

 
 

-/* 
+/*

  * A newline only breaks an existing line; it won't assert vertical
  * space.  All data in the output buffer is flushed prior to the newline
  * assertion.
  * A newline only breaks an existing line; it won't assert vertical
  * space.  All data in the output buffer is flushed prior to the newline
  * assertion.


@@ -347,15 +307,10 @@ term_newln(struct termp *p)
 {
 
        p->flags |= TERMP_NOSPACE;
 {
 
        p->flags |= TERMP_NOSPACE;

-       if (0 == p->col) {
-               p->flags &= ~TERMP_NOLPAD;
-               return;
-       }
-       term_flushln(p);
-       p->flags &= ~TERMP_NOLPAD;
+       if (p->col || p->viscol)
+               term_flushln(p);

 }
 
 }
 

-

 /*
  * Asserts a vertical space (a full, empty line-break between lines).
  * Note that if used twice, this will cause two blank spaces and so on.
 /*
  * Asserts a vertical space (a full, empty line-break between lines).
  * Note that if used twice, this will cause two blank spaces and so on.


@@ -367,201 +322,519 @@ term_vspace(struct termp *p)
 {
 
        term_newln(p);
 {
 
        term_newln(p);

-       putchar('\n');
+       p->viscol = 0;
+       if (0 < p->skipvsp)
+               p->skipvsp--;
+       else
+               (*p->endline)(p);

 }
 
 }
 

+/* Swap current and previous font; for \fP and .ft P */
+void
+term_fontlast(struct termp *p)
+{
+       enum termfont    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)
+       f = p->fontl;
+       p->fontl = p->fontq[p->fonti];
+       p->fontq[p->fonti] = f;
+}
+
+/* Set font, save current, discard previous; for \f, .ft, .B etc. */
+void
+term_fontrepl(struct termp *p, enum termfont f)

 {
 {

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

 
 

-       rhs = term_a2ascii(p->symtab, word, len, &sz);
+       p->fontl = p->fontq[p->fonti];
+       p->fontq[p->fonti] = f;
+}
+
+/* Set font, save previous. */
+void
+term_fontpush(struct termp *p, enum termfont f)
+{

 
 

-       if (rhs)
-               for (i = 0; i < (int)sz; i++) 
-                       term_encodea(p, rhs[i]);
+       p->fontl = p->fontq[p->fonti];
+       if (++p->fonti == p->fontsz) {
+               p->fontsz += 8;
+               p->fontq = mandoc_reallocarray(p->fontq,
+                   p->fontsz, sizeof(enum termfont *));
+       }
+       p->fontq[p->fonti] = f;

 }
 
 }
 

+/* Flush to make the saved pointer current again. */
+void
+term_fontpopq(struct termp *p, int i)
+{

 
 

-/*
- * 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)
+       assert(i >= 0);
+       if (p->fonti > i)
+               p->fonti = i;
+}
+
+/* Pop one font off the stack. */
+void
+term_fontpop(struct termp *p)

 {
 {

-       int              j;
-       const char      *wp;

 
 

-       wp = *word;
+       assert(p->fonti);
+       p->fonti--;
+}

 
 

-       if (0 == *(++wp)) {
-               *word = wp;
-               return;
+/*
+ * 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       nbrsp[2] = { ASCII_NBRSP, 0 };
+       const char      *seq, *cp;
+       int              sz, uc;
+       size_t           ssz;
+       enum mandoc_esc  esc;
+
+       if ( ! (TERMP_NOSPACE & p->flags)) {
+               if ( ! (TERMP_KEEP & p->flags)) {
+                       bufferc(p, ' ');
+                       if (TERMP_SENTENCE & p->flags)
+                               bufferc(p, ' ');
+               } else
+                       bufferc(p, ASCII_NBRSP);

        }
        }

+       if (TERMP_PREKEEP & p->flags)
+               p->flags |= TERMP_KEEP;

 
 

-       if ('(' == *wp) {
-               wp++;
-               if (0 == *wp || 0 == *(wp + 1)) {
-                       *word = 0 == *wp ? wp : wp + 1;
-                       return;
-               }
-
-               term_nescape(p, wp, 2);
-               *word = ++wp;
-               return;
+       if ( ! (p->flags & TERMP_NONOSPACE))
+               p->flags &= ~TERMP_NOSPACE;
+       else
+               p->flags |= TERMP_NOSPACE;

 
 

-       } else if ('*' == *wp) {
-               if (0 == *(++wp)) {
-                       *word = wp;
-                       return;
-               }
+       p->flags &= ~(TERMP_SENTENCE | TERMP_NONEWLINE);

 
 

-               switch (*wp) {
-               case ('('):
-                       wp++;
-                       if (0 == *wp || 0 == *(wp + 1)) {
-                               *word = 0 == *wp ? wp : wp + 1;
-                               return;
+       while ('\0' != *word) {
+               if ('\\' != *word) {
+                       if (TERMP_SKIPCHAR & p->flags) {
+                               p->flags &= ~TERMP_SKIPCHAR;
+                               word++;
+                               continue;

                        }
                        }

-
-                       term_nescape(p, wp, 2);
-                       *word = ++wp;
-                       return;
-               case ('['):
-                       break;
-               default:
-                       term_nescape(p, wp, 1);
-                       *word = wp;
-                       return;
-               }
-       
-       } else if ('f' == *wp) {
-               if (0 == *(++wp)) {
-                       *word = wp;
-                       return;
+                       if (TERMP_NBRWORD & p->flags) {
+                               if (' ' == *word) {
+                                       encode(p, nbrsp, 1);
+                                       word++;
+                                       continue;
+                               }
+                               ssz = strcspn(word, "\\ ");
+                       } else
+                               ssz = strcspn(word, "\\");
+                       encode(p, word, ssz);
+                       word += (int)ssz;
+                       continue;

                }
 
                }
 

-               switch (*wp) {
-               case ('B'):
-                       p->flags |= TERMP_BOLD;
+               word++;
+               esc = mandoc_escape(&word, &seq, &sz);
+               if (ESCAPE_ERROR == esc)
+                       continue;
+
+               switch (esc) {
+               case ESCAPE_UNICODE:
+                       uc = mchars_num2uc(seq + 1, sz - 1);

                        break;
                        break;

-               case ('I'):
-                       p->flags |= TERMP_UNDER;
+               case ESCAPE_NUMBERED:
+                       uc = mchars_num2char(seq, sz);
+                       if (uc < 0)
+                               continue;

                        break;
                        break;

-               case ('P'):
+               case ESCAPE_SPECIAL:
+                       if (p->enc == TERMENC_ASCII) {
+                               cp = mchars_spec2str(p->symtab,
+                                   seq, sz, &ssz);
+                               if (cp != NULL)
+                                       encode(p, cp, ssz);
+                       } else {
+                               uc = mchars_spec2cp(p->symtab, seq, sz);
+                               if (uc > 0)
+                                       encode1(p, uc);
+                       }
+                       continue;
+               case ESCAPE_FONTBOLD:
+                       term_fontrepl(p, TERMFONT_BOLD);
+                       continue;
+               case ESCAPE_FONTITALIC:
+                       term_fontrepl(p, TERMFONT_UNDER);
+                       continue;
+               case ESCAPE_FONTBI:
+                       term_fontrepl(p, TERMFONT_BI);
+                       continue;
+               case ESCAPE_FONT:

                        /* FALLTHROUGH */
                        /* FALLTHROUGH */

-               case ('R'):
-                       p->flags &= ~TERMP_STYLE;
-                       break;
+               case ESCAPE_FONTROMAN:
+                       term_fontrepl(p, TERMFONT_NONE);
+                       continue;
+               case ESCAPE_FONTPREV:
+                       term_fontlast(p);
+                       continue;
+               case ESCAPE_NOSPACE:
+                       if (TERMP_SKIPCHAR & p->flags)
+                               p->flags &= ~TERMP_SKIPCHAR;
+                       else if ('\0' == *word)
+                               p->flags |= (TERMP_NOSPACE | TERMP_NONEWLINE);
+                       continue;
+               case ESCAPE_SKIPCHAR:
+                       p->flags |= TERMP_SKIPCHAR;
+                       continue;
+               case ESCAPE_OVERSTRIKE:
+                       cp = seq + sz;
+                       while (seq < cp) {
+                               if (*seq == '\\') {
+                                       mandoc_escape(&seq, NULL, NULL);
+                                       continue;
+                               }
+                               encode1(p, *seq++);
+                               if (seq < cp)
+                                       encode(p, "\b", 1);
+                       }

                default:
                default:

-                       break;
+                       continue;

                }
 
                }
 

-               *word = wp;
-               return;
+               /*
+                * Common handling for Unicode and numbered
+                * character escape sequences.
+                */

 
 

-       } else if ('[' != *wp) {
-               term_nescape(p, wp, 1);
-               *word = wp;
-               return;
+               if (p->enc == TERMENC_ASCII) {
+                       cp = ascii_uc2str(uc);
+                       encode(p, cp, strlen(cp));
+               } else {
+                       if ((uc < 0x20 && uc != 0x09) ||
+                           (uc > 0x7E && uc < 0xA0))
+                               uc = 0xFFFD;
+                       encode1(p, uc);
+               }

        }
        }

+       p->flags &= ~TERMP_NBRWORD;
+}

 
 

-       wp++;
-       for (j = 0; *wp && ']' != *wp; wp++, j++)
-               /* Loop... */ ;
+static void
+adjbuf(struct termp *p, size_t sz)
+{

 
 

-       if (0 == *wp) {
-               *word = wp;
-               return;
-       }
+       if (0 == p->maxcols)
+               p->maxcols = 1024;
+       while (sz >= p->maxcols)
+               p->maxcols <<= 2;

 
 

-       term_nescape(p, wp - j, (size_t)j);
-       *word = wp;
+       p->buf = mandoc_reallocarray(p->buf, p->maxcols, sizeof(int));

 }
 
 }
 

+static void
+bufferc(struct termp *p, char c)
+{
+
+       if (p->col + 1 >= p->maxcols)
+               adjbuf(p, p->col + 1);
+
+       p->buf[p->col++] = c;
+}

 
 /*
 
 /*

- * 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.
+ * See encode().
+ * Do this for a single (probably unicode) value.
+ * Does not check for non-decorated glyphs.

  */
  */

-void
-term_word(struct termp *p, const char *word)
+static void
+encode1(struct termp *p, int c)

 {
 {

-       const char       *sv;
+       enum termfont     f;
+
+       if (TERMP_SKIPCHAR & p->flags) {
+               p->flags &= ~TERMP_SKIPCHAR;
+               return;
+       }

 
 

-       if (term_isclosedelim(word))
-               if ( ! (TERMP_IGNDELIM & p->flags))
-                       p->flags |= TERMP_NOSPACE;
+       if (p->col + 6 >= p->maxcols)
+               adjbuf(p, p->col + 6);

 
 

-       if ( ! (TERMP_NOSPACE & p->flags))
-               term_chara(p, ' ');
+       f = p->fontq[p->fonti];

 
 

-       if ( ! (p->flags & TERMP_NONOSPACE))
-               p->flags &= ~TERMP_NOSPACE;
+       if (TERMFONT_UNDER == f || TERMFONT_BI == f) {
+               p->buf[p->col++] = '_';
+               p->buf[p->col++] = 8;
+       }
+       if (TERMFONT_BOLD == f || TERMFONT_BI == f) {
+               if (ASCII_HYPH == c)
+                       p->buf[p->col++] = '-';
+               else
+                       p->buf[p->col++] = c;
+               p->buf[p->col++] = 8;
+       }
+       p->buf[p->col++] = c;
+}

 
 

-       /* 
-        * 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)
+{
+       size_t            i;
+
+       if (TERMP_SKIPCHAR & p->flags) {
+               p->flags &= ~TERMP_SKIPCHAR;
+               return;
+       }
+
+       /*
+        * Encode and buffer a string of characters.  If the current
+        * font mode is unset, buffer directly, else encode then buffer
+        * character by character.

         */
 
         */
 

-       for (sv = word; *word; word++)
-               if ('\\' != *word)
-                       term_encodea(p, *word);
-               else
-                       term_pescape(p, &word);
+       if (p->fontq[p->fonti] == TERMFONT_NONE) {
+               if (p->col + sz >= p->maxcols)
+                       adjbuf(p, p->col + sz);
+               for (i = 0; i < sz; i++)
+                       p->buf[p->col++] = word[i];
+               return;
+       }

 
 

-       if (term_isopendelim(sv))
-               p->flags |= TERMP_NOSPACE;
-}
+       /* Pre-buffer, assuming worst-case. */

 
 

+       if (p->col + 1 + (sz * 5) >= p->maxcols)
+               adjbuf(p, p->col + 1 + (sz * 5));

 
 

-/*
- * 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)
+       for (i = 0; i < sz; i++) {
+               if (ASCII_HYPH == word[i] ||
+                   isgraph((unsigned char)word[i]))
+                       encode1(p, word[i]);
+               else
+                       p->buf[p->col++] = word[i];
+       }
+}
+
+void
+term_setwidth(struct termp *p, const char *wstr)

 {
 {

-       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;
+       struct roffsu    su;
+       size_t           width;
+       int              iop;
+
+       iop = 0;
+       width = 0;
+       if (NULL != wstr) {
+               switch (*wstr) {
+               case '+':
+                       iop = 1;
+                       wstr++;
+                       break;
+               case '-':
+                       iop = -1;
+                       wstr++;
+                       break;
+               default:
+                       break;
+               }
+               if (a2roffsu(wstr, &su, SCALE_MAX))
+                       width = term_hspan(p, &su);
+               else
+                       iop = 0;

        }
        }

-       p->buf[(int)(p->col)++] = c;
+       (*p->setwidth)(p, iop, width);

 }
 
 }
 

+size_t
+term_len(const struct termp *p, size_t sz)
+{

 
 

-static void
-term_encodea(struct termp *p, char c)
+       return((*p->width)(p, ' ') * sz);
+}
+
+static size_t
+cond_width(const struct termp *p, int c, int *skip)

 {
 {

-       
-       if (' ' != c && 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);
+
+       if (*skip) {
+               (*skip) = 0;
+               return(0);
+       } else
+               return((*p->width)(p, c));
+}
+
+size_t
+term_strlen(const struct termp *p, const char *cp)
+{
+       size_t           sz, rsz, i;
+       int              ssz, skip, uc;
+       const char      *seq, *rhs;
+       enum mandoc_esc  esc;
+       static const char rej[] = { '\\', ASCII_NBRSP, ASCII_HYPH,
+                       ASCII_BREAK, '\0' };
+
+       /*
+        * Account for escaped sequences within string length
+        * calculations.  This follows the logic in term_word() as we
+        * must calculate the width of produced strings.
+        */
+
+       sz = 0;
+       skip = 0;
+       while ('\0' != *cp) {
+               rsz = strcspn(cp, rej);
+               for (i = 0; i < rsz; i++)
+                       sz += cond_width(p, *cp++, &skip);
+
+               switch (*cp) {
+               case '\\':
+                       cp++;
+                       esc = mandoc_escape(&cp, &seq, &ssz);
+                       if (ESCAPE_ERROR == esc)
+                               continue;
+
+                       rhs = NULL;
+
+                       switch (esc) {
+                       case ESCAPE_UNICODE:
+                               uc = mchars_num2uc(seq + 1, ssz - 1);
+                               break;
+                       case ESCAPE_NUMBERED:
+                               uc = mchars_num2char(seq, ssz);
+                               if (uc < 0)
+                                       continue;
+                               break;
+                       case ESCAPE_SPECIAL:
+                               if (p->enc == TERMENC_ASCII) {
+                                       rhs = mchars_spec2str(p->symtab,
+                                           seq, ssz, &rsz);
+                                       if (rhs != NULL)
+                                               break;
+                               } else {
+                                       uc = mchars_spec2cp(p->symtab,
+                                           seq, ssz);
+                                       if (uc > 0)
+                                               sz += cond_width(p, uc, &skip);
+                               }
+                               continue;
+                       case ESCAPE_SKIPCHAR:
+                               skip = 1;
+                               continue;
+                       case ESCAPE_OVERSTRIKE:
+                               rsz = 0;
+                               rhs = seq + ssz;
+                               while (seq < rhs) {
+                                       if (*seq == '\\') {
+                                               mandoc_escape(&seq, NULL, NULL);
+                                               continue;
+                                       }
+                                       i = (*p->width)(p, *seq++);
+                                       if (rsz < i)
+                                               rsz = i;
+                               }
+                               sz += rsz;
+                               continue;
+                       default:
+                               continue;
+                       }
+
+                       /*
+                        * Common handling for Unicode and numbered
+                        * character escape sequences.
+                        */
+
+                       if (rhs == NULL) {
+                               if (p->enc == TERMENC_ASCII) {
+                                       rhs = ascii_uc2str(uc);
+                                       rsz = strlen(rhs);
+                               } else {
+                                       if ((uc < 0x20 && uc != 0x09) ||
+                                           (uc > 0x7E && uc < 0xA0))
+                                               uc = 0xFFFD;
+                                       sz += cond_width(p, uc, &skip);
+                                       continue;
+                               }
+                       }
+
+                       if (skip) {
+                               skip = 0;
+                               break;
+                       }
+
+                       /*
+                        * Common handling for all escape sequences
+                        * printing more than one character.
+                        */
+
+                       for (i = 0; i < rsz; i++)
+                               sz += (*p->width)(p, *rhs++);
+                       break;
+               case ASCII_NBRSP:
+                       sz += cond_width(p, ' ', &skip);
+                       cp++;
+                       break;
+               case ASCII_HYPH:
+                       sz += cond_width(p, '-', &skip);
+                       cp++;
+                       /* FALLTHROUGH */
+               case ASCII_BREAK:
+                       break;
+               default:
+                       break;

                }
        }
                }
        }

-       term_chara(p, c);
+
+       return(sz);
+}
+
+int
+term_vspan(const struct termp *p, const struct roffsu *su)
+{
+       double           r;
+       int              ri;
+
+       switch (su->unit) {
+       case SCALE_BU:
+               r = su->scale / 40.0;
+               break;
+       case SCALE_CM:
+               r = su->scale * 6.0 / 2.54;
+               break;
+       case SCALE_FS:
+               r = su->scale * 65536.0 / 40.0;
+               break;
+       case SCALE_IN:
+               r = su->scale * 6.0;
+               break;
+       case SCALE_MM:
+               r = su->scale * 0.006;
+               break;
+       case SCALE_PC:
+               r = su->scale;
+               break;
+       case SCALE_PT:
+               r = su->scale / 12.0;
+               break;
+       case SCALE_EN:
+               /* FALLTHROUGH */
+       case SCALE_EM:
+               r = su->scale * 0.6;
+               break;
+       case SCALE_VS:
+               r = su->scale;
+               break;
+       default:
+               abort();
+               /* NOTREACHED */
+       }
+       ri = r > 0.0 ? r + 0.4995 : r - 0.4995;
+       return(ri < 66 ? ri : 1);
+}
+
+int
+term_hspan(const struct termp *p, const struct roffsu *su)
+{
+       double           v;
+
+       v = (*p->hspan)(p, su);
+       return(v > 0.0 ? v + 0.0005 : v - 0.0005);

 }
 }