Tedu support for the -xsh4.2 argument to the mdoc(7) .St macro

[mandoc.git] / term.c

diff --git a/term.c b/term.c
index 5aebbe458dd0641c82fe79abb41567f9ce2aa948..70ca20c440c57a0cfac04e7a67c9310c42581dec 100644 (file)
--- a/term.c
+++ b/term.c
@@ -1,7 +1,7 @@
-/*     $Id: term.c,v 1.265 2017/06/07 02:14:09 schwarze Exp $ */
+/* $Id: term.c,v 1.285 2022/01/10 18:01:35 schwarze Exp $ */

 /*
 /*

+ * Copyright (c) 2010-2022 Ingo Schwarze <schwarze@openbsd.org>

  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>

- * Copyright (c) 2010-2017 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


@@ -21,6 +21,7 @@
 
 #include <assert.h>
 #include <ctype.h>
 
 #include <assert.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>


@@ -32,18 +33,35 @@
 #include "main.h"
 
 static size_t           cond_width(const struct termp *, int, int *);
 #include "main.h"
 
 static size_t           cond_width(const struct termp *, int, int *);

-static void             adjbuf(struct termp *p, size_t);
+static void             adjbuf(struct termp_col *, size_t);

 static void             bufferc(struct termp *, char);
 static void             encode(struct termp *, const char *, size_t);
 static void             encode1(struct termp *, int);
 static void             endline(struct termp *);
 static void             bufferc(struct termp *, char);
 static void             encode(struct termp *, const char *, size_t);
 static void             encode1(struct termp *, int);
 static void             endline(struct termp *);

+static void             term_field(struct termp *, size_t, size_t);
+static void             term_fill(struct termp *, size_t *, size_t *,
+                               size_t);

 
 
 void
 
 
 void

-term_free(struct termp *p)
+term_setcol(struct termp *p, size_t maxtcol)

 {
 {

+       if (maxtcol > p->maxtcol) {
+               p->tcols = mandoc_recallocarray(p->tcols,
+                   p->maxtcol, maxtcol, sizeof(*p->tcols));
+               p->maxtcol = maxtcol;
+       }
+       p->lasttcol = maxtcol - 1;
+       p->tcol = p->tcols;
+}

 
 

-       free(p->buf);
+void
+term_free(struct termp *p)
+{
+       term_tab_free();
+       for (p->tcol = p->tcols; p->tcol < p->tcols + p->maxtcol; p->tcol++)
+               free(p->tcol->buf);
+       free(p->tcols);

        free(p->fontq);
        free(p);
 }
        free(p->fontq);
        free(p);
 }


@@ -70,192 +88,325 @@ term_end(struct termp *p)
  * 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
  * 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_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_BRTRSP: Consider trailing whitespace significant
- *    when deciding whether the chunk fits or not.
- *  - 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_HANG: 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_NOPAD: Start writing at the current position,
- *    do not pad with blank characters up to the offset.
+ * output line at the end of the chunk.  There are many flags modifying
+ * this behaviour, see the comments in the body of the function.

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

-       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 */
-
-       vbl = (p->flags & TERMP_NOPAD) || p->offset < p->viscol ? 0 :
-           p->offset - p->viscol;
+       size_t   vbl;      /* Number of blanks to prepend to the output. */
+       size_t   vbr;      /* Actual visual position of the end of field. */
+       size_t   vfield;   /* Desired visual field width. */
+       size_t   vtarget;  /* Desired visual position of the right margin. */
+       size_t   ic;       /* Character position in the input buffer. */
+       size_t   nbr;      /* Number of characters to print in this field. */
+
+       /*
+        * Normally, start writing at the left margin, but with the
+        * NOPAD flag, start writing at the current position instead.
+        */
+
+       vbl = (p->flags & TERMP_NOPAD) || p->tcol->offset < p->viscol ?
+           0 : p->tcol->offset - p->viscol;

        if (p->minbl && vbl < p->minbl)
                vbl = p->minbl;
        if (p->minbl && vbl < p->minbl)
                vbl = p->minbl;

-       maxvis = p->rmargin > p->viscol + vbl ?
-           p->rmargin - p->viscol - vbl : 0;
-       bp = !(p->flags & TERMP_NOBREAK) ? maxvis :
-           p->maxrmargin > p->viscol + vbl ?
-           p->maxrmargin - p->viscol - vbl : 0;
-       vis = vend = 0;
-       i = 0;
-
-       while (i < p->lastcol) {
+
+       if ((p->flags & TERMP_MULTICOL) == 0)
+               p->tcol->col = 0;
+
+       /* Loop over output lines. */
+
+       for (;;) {
+               vfield = p->tcol->rmargin > p->viscol + vbl ?
+                   p->tcol->rmargin - p->viscol - vbl : 0;
+

                /*
                /*

-                * Handle literal tab characters: collapse all
-                * subsequent tabs into a single huge set of spaces.
+                * Normally, break the line at the the right margin
+                * of the field, but with the NOBREAK flag, only
+                * break it at the max right margin of the screen,
+                * and with the BRNEVER flag, never break it at all.

                 */
                 */

-               ntab = 0;
-               while (i < p->lastcol && p->buf[i] == '\t') {
-                       vend = term_tab_next(vis);
-                       vbl += vend - vis;
-                       vis = vend;
-                       ntab++;
-                       i++;
-               }
+
+               vtarget = (p->flags & TERMP_NOBREAK) == 0 ? vfield :
+                   p->maxrmargin > p->viscol + vbl ?
+                   p->maxrmargin - p->viscol - vbl : 0;

 
                /*
 
                /*

-                * Count up visible word characters.  Control sequences
-                * (starting with the CSI) aren't counted.  A space
-                * generates a non-printing word, which is valid (the
-                * space is printed according to regular spacing rules).
+                * Figure out how much text will fit in the field.
+                * If there is whitespace only, print nothing.

                 */
 
                 */
 

-               for (j = i, jhy = 0; j < p->lastcol; j++) {
-                       if (' ' == p->buf[j] || '\t' == p->buf[j])
-                               break;
+               term_fill(p, &nbr, &vbr,
+                   p->flags & TERMP_BRNEVER ? SIZE_MAX : vtarget);
+               if (nbr == 0)
+                       break;

 
 

-                       /* Back over the last printed character. */
-                       if (8 == p->buf[j]) {
-                               assert(j);
-                               vend -= (*p->width)(p, p->buf[j - 1]);
+               /*
+                * With the CENTER or RIGHT flag, increase the indentation
+                * to center the text between the left and right margins
+                * or to adjust it to the right margin, respectively.
+                */
+
+               if (vbr < vtarget) {
+                       if (p->flags & TERMP_CENTER)
+                               vbl += (vtarget - vbr) / 2;
+                       else if (p->flags & TERMP_RIGHT)
+                               vbl += vtarget - vbr;
+               }
+
+               /* Finally, print the field content. */
+
+               term_field(p, vbl, nbr);
+
+               /*
+                * If there is no text left in the field, exit the loop.
+                * If the BRTRSP flag is set, consider trailing
+                * whitespace significant when deciding whether
+                * the field fits or not.
+                */
+
+               for (ic = p->tcol->col; ic < p->tcol->lastcol; ic++) {
+                       switch (p->tcol->buf[ic]) {
+                       case '\t':
+                               if (p->flags & TERMP_BRTRSP)
+                                       vbr = term_tab_next(vbr);
+                               continue;
+                       case ' ':
+                               if (p->flags & TERMP_BRTRSP)
+                                       vbr += (*p->width)(p, ' ');
+                               continue;
+                       case '\n':
+                       case ASCII_BREAK:

                                continue;
                                continue;

+                       default:
+                               break;

                        }
                        }

+                       break;
+               }
+               if (ic == p->tcol->lastcol)
+                       break;
+
+               /*
+                * At the location of an automtic line break, input
+                * space characters are consumed by the line break.
+                */

 
 

-                       /* 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;
+               while (p->tcol->col < p->tcol->lastcol &&
+                   p->tcol->buf[p->tcol->col] == ' ')
+                       p->tcol->col++;

 
 

-                       /*
-                        * Hyphenation now decided, put back a real
-                        * hyphen such that we get the correct width.
-                        */
-                       if (ASCII_HYPH == p->buf[j])
-                               p->buf[j] = '-';
+               /*
+                * In multi-column mode, leave the rest of the text
+                * in the buffer to be handled by a subsequent
+                * invocation, such that the other columns of the
+                * table can be handled first.
+                * In single-column mode, simply break the line.
+                */

 
 

-                       vend += (*p->width)(p, p->buf[j]);
-               }
+               if (p->flags & TERMP_MULTICOL)
+                       return;
+
+               endline(p);
+               p->viscol = 0;

 
                /*
 
                /*

-                * Find out whether we would exceed the right margin.
-                * If so, break to the next line.
+                * Normally, start the next line at the same indentation
+                * as this one, but with the BRIND flag, start it at the
+                * right margin instead.  This is used together with
+                * NOBREAK for the tags in various kinds of tagged lists.

                 */
                 */

-               if (vend > bp && 0 == jhy && vis > 0 &&
-                   (p->flags & TERMP_BRNEVER) == 0) {
-                       vend -= vis;
-                       endline(p);

 
 

-                       /* Use pending tabs on the new line. */
+               vbl = p->flags & TERMP_BRIND ?
+                   p->tcol->rmargin : p->tcol->offset;
+       }

 
 

-                       vbl = 0;
-                       while (ntab--)
-                               vbl = term_tab_next(vbl);
+       /* Reset output state in preparation for the next field. */

 
 

-                       /* Re-establish indentation. */
+       p->col = p->tcol->col = p->tcol->lastcol = 0;
+       p->minbl = p->trailspace;
+       p->flags &= ~(TERMP_BACKAFTER | TERMP_BACKBEFORE | TERMP_NOPAD);

 
 

-                       if (p->flags & TERMP_BRIND)
-                               vbl += p->rmargin;
-                       else
-                               vbl += p->offset;
-                       maxvis = p->rmargin > vbl ? p->rmargin - vbl : 0;
-                       bp = !(p->flags & TERMP_NOBREAK) ? maxvis :
-                           p->maxrmargin > vbl ?  p->maxrmargin - vbl : 0;
-               }
+       if (p->flags & TERMP_MULTICOL)
+               return;

 
 

-               /* Write out the [remaining] word. */
-               for ( ; i < p->lastcol; i++) {
-                       if (vend > bp && jhy > 0 && i > jhy)
+       /*
+        * The HANG flag means that the next field
+        * always follows on the same line.
+        * The NOBREAK flag means that the next field
+        * follows on the same line unless the field was overrun.
+        * Normally, break the line at the end of each field.
+        */
+
+       if ((p->flags & TERMP_HANG) == 0 &&
+           ((p->flags & TERMP_NOBREAK) == 0 ||
+            vbr + term_len(p, p->trailspace) > vfield))
+               endline(p);
+}
+
+/*
+ * Store the number of input characters to print in this field in *nbr
+ * and their total visual width to print in *vbr.
+ * If there is only whitespace in the field, both remain zero.
+ * The desired visual width of the field is provided by vtarget.
+ * If the first word is longer, the field will be overrun.
+ */
+static void
+term_fill(struct termp *p, size_t *nbr, size_t *vbr, size_t vtarget)
+{
+       size_t   ic;        /* Character position in the input buffer. */
+       size_t   vis;       /* Visual position of the current character. */
+       size_t   vn;        /* Visual position of the next character. */
+       int      breakline; /* Break at the end of this word. */
+       int      graph;     /* Last character was non-blank. */
+
+       *nbr = *vbr = vis = 0;
+       breakline = graph = 0;
+       for (ic = p->tcol->col; ic < p->tcol->lastcol; ic++) {
+               switch (p->tcol->buf[ic]) {
+               case '\b':  /* Escape \o (overstrike) or backspace markup. */
+                       assert(ic > 0);
+                       vis -= (*p->width)(p, p->tcol->buf[ic - 1]);
+                       continue;
+
+               case '\t':  /* Normal ASCII whitespace. */
+               case ' ':
+               case ASCII_BREAK:  /* Escape \: (breakpoint). */
+                       switch (p->tcol->buf[ic]) {
+                       case '\t':
+                               vn = term_tab_next(vis);

                                break;
                                break;

-                       if ('\t' == p->buf[i])
+                       case ' ':
+                               vn = vis + (*p->width)(p, ' ');

                                break;
                                break;

-                       if (' ' == p->buf[i]) {
-                               j = i;
-                               while (i < p->lastcol && ' ' == p->buf[i])
-                                       i++;
-                               dv = (i - j) * (*p->width)(p, ' ');
-                               vbl += dv;
-                               vend += dv;
+                       case ASCII_BREAK:
+                               vn = vis;

                                break;
                                break;

+                       default:
+                               abort();

                        }
                        }

-                       if (ASCII_NBRSP == p->buf[i]) {
-                               vbl += (*p->width)(p, ' ');
-                               continue;
+                       /* Can break at the end of a word. */
+                       if (breakline || vn > vtarget)
+                               break;
+                       if (graph) {
+                               *nbr = ic;
+                               *vbr = vis;
+                               graph = 0;

                        }
                        }

-                       if (ASCII_BREAK == p->buf[i])
-                               continue;
+                       vis = vn;
+                       continue;
+
+               case '\n':  /* Escape \p (break at the end of the word). */
+                       breakline = 1;
+                       continue;

 
 

+               case ASCII_HYPH:  /* Breakable hyphen. */
+                       graph = 1;

                        /*
                        /*

-                        * Now we definitely know there will be
-                        * printable characters to output,
-                        * so write preceding white space now.
+                        * We are about to decide whether to break the
+                        * line or not, so we no longer need this hyphen
+                        * to be marked as breakable.  Put back a real
+                        * hyphen such that we get the correct width.

                         */
                         */

-                       if (vbl) {
-                               (*p->advance)(p, vbl);
-                               p->viscol += vbl;
-                               vbl = 0;
+                       p->tcol->buf[ic] = '-';
+                       vis += (*p->width)(p, '-');
+                       if (vis > vtarget) {
+                               ic++;
+                               break;

                        }
                        }

+                       *nbr = ic + 1;
+                       *vbr = vis;
+                       continue;

 
 

-                       (*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]);
+               case ASCII_NBRSP:  /* Non-breakable space. */
+                       p->tcol->buf[ic] = ' ';
+                       /* FALLTHROUGH */
+               default:  /* Printable character. */
+                       graph = 1;
+                       vis += (*p->width)(p, p->tcol->buf[ic]);
+                       if (vis > vtarget && *nbr > 0)
+                               return;
+                       continue;

                }
                }

-               vis = vend;
+               break;

        }
 
        /*
        }
 
        /*

-        * If there was trailing white space, it was not printed;
-        * so reset the cursor position accordingly.
+        * If the last word extends to the end of the field without any
+        * trailing whitespace, the loop could not check yet whether it
+        * can remain on this line.  So do the check now.

         */
         */

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

 
 

-       p->col = p->lastcol = 0;
-       p->minbl = p->trailspace;
-       p->flags &= ~(TERMP_BACKAFTER | TERMP_BACKBEFORE | TERMP_NOPAD);
+       if (graph && (vis <= vtarget || *nbr == 0)) {
+               *nbr = ic;
+               *vbr = vis;
+       }
+}

 
 

-       /* Trailing whitespace is significant in some columns. */
-       if (vis && vbl && (TERMP_BRTRSP & p->flags))
-               vis += vbl;
+/*
+ * Print the contents of one field
+ * with an indentation of       vbl      visual columns,
+ * and an input string length of nbr     characters.
+ */
+static void
+term_field(struct termp *p, size_t vbl, size_t nbr)
+{
+       size_t   ic;    /* Character position in the input buffer. */
+       size_t   vis;   /* Visual position of the current character. */
+       size_t   dv;    /* Visual width of the current character. */
+       size_t   vn;    /* Visual position of the next character. */

 
 

-       /* If the column was overrun, break the line. */
-       if ((p->flags & TERMP_NOBREAK) == 0 ||
-           ((p->flags & TERMP_HANG) == 0 &&
-            vis + p->trailspace * (*p->width)(p, ' ') > maxvis))
-               endline(p);
+       vis = 0;
+       for (ic = p->tcol->col; ic < nbr; ic++) {
+
+               /*
+                * To avoid the printing of trailing whitespace,
+                * do not print whitespace right away, only count it.
+                */
+
+               switch (p->tcol->buf[ic]) {
+               case '\n':
+               case ASCII_BREAK:
+                       continue;
+               case '\t':
+                       vn = term_tab_next(vis);
+                       vbl += vn - vis;
+                       vis = vn;
+                       continue;
+               case ' ':
+               case ASCII_NBRSP:
+                       dv = (*p->width)(p, ' ');
+                       vbl += dv;
+                       vis += dv;
+                       continue;
+               default:
+                       break;
+               }
+
+               /*
+                * We found a non-blank character to print,
+                * so write preceding white space now.
+                */
+
+               if (vbl > 0) {
+                       (*p->advance)(p, vbl);
+                       p->viscol += vbl;
+                       vbl = 0;
+               }
+
+               /* Print the character and adjust the visual position. */
+
+               (*p->letter)(p, p->tcol->buf[ic]);
+               if (p->tcol->buf[ic] == '\b') {
+                       dv = (*p->width)(p, p->tcol->buf[ic - 1]);
+                       p->viscol -= dv;
+                       vis -= dv;
+               } else {
+                       dv = (*p->width)(p, p->tcol->buf[ic]);
+                       p->viscol += dv;
+                       vis += dv;
+               }
+       }
+       p->tcol->col = nbr;

 }
 
 static void
 }
 
 static void


@@ -287,7 +438,7 @@ term_newln(struct termp *p)
 {
 
        p->flags |= TERMP_NOSPACE;
 {
 
        p->flags |= TERMP_NOSPACE;

-       if (p->lastcol || p->viscol)
+       if (p->tcol->lastcol || p->viscol)

                term_flushln(p);
 }
 
                term_flushln(p);
 }
 


@@ -415,9 +566,6 @@ term_word(struct termp *p, const char *word)
 
                word++;
                esc = mandoc_escape(&word, &seq, &sz);
 
                word++;
                esc = mandoc_escape(&word, &seq, &sz);

-               if (ESCAPE_ERROR == esc)
-                       continue;
-

                switch (esc) {
                case ESCAPE_UNICODE:
                        uc = mchars_num2uc(seq + 1, sz - 1);
                switch (esc) {
                case ESCAPE_UNICODE:
                        uc = mchars_num2uc(seq + 1, sz - 1);


@@ -438,72 +586,89 @@ term_word(struct termp *p, const char *word)
                                        encode1(p, uc);
                        }
                        continue;
                                        encode1(p, uc);
                        }
                        continue;

+               case ESCAPE_UNDEF:
+                       uc = *seq;
+                       break;

                case ESCAPE_FONTBOLD:
                case ESCAPE_FONTBOLD:

+               case ESCAPE_FONTCB:

                        term_fontrepl(p, TERMFONT_BOLD);
                        continue;
                case ESCAPE_FONTITALIC:
                        term_fontrepl(p, TERMFONT_BOLD);
                        continue;
                case ESCAPE_FONTITALIC:

+               case ESCAPE_FONTCI:

                        term_fontrepl(p, TERMFONT_UNDER);
                        continue;
                case ESCAPE_FONTBI:
                        term_fontrepl(p, TERMFONT_BI);
                        continue;
                case ESCAPE_FONT:
                        term_fontrepl(p, TERMFONT_UNDER);
                        continue;
                case ESCAPE_FONTBI:
                        term_fontrepl(p, TERMFONT_BI);
                        continue;
                case ESCAPE_FONT:

+               case ESCAPE_FONTCR:

                case ESCAPE_FONTROMAN:
                        term_fontrepl(p, TERMFONT_NONE);
                        continue;
                case ESCAPE_FONTPREV:
                        term_fontlast(p);
                        continue;
                case ESCAPE_FONTROMAN:
                        term_fontrepl(p, TERMFONT_NONE);
                        continue;
                case ESCAPE_FONTPREV:
                        term_fontlast(p);
                        continue;

+               case ESCAPE_BREAK:
+                       bufferc(p, '\n');
+                       continue;

                case ESCAPE_NOSPACE:
                        if (p->flags & TERMP_BACKAFTER)
                                p->flags &= ~TERMP_BACKAFTER;
                        else if (*word == '\0')
                                p->flags |= (TERMP_NOSPACE | TERMP_NONEWLINE);
                        continue;
                case ESCAPE_NOSPACE:
                        if (p->flags & TERMP_BACKAFTER)
                                p->flags &= ~TERMP_BACKAFTER;
                        else if (*word == '\0')
                                p->flags |= (TERMP_NOSPACE | TERMP_NONEWLINE);
                        continue;

+               case ESCAPE_DEVICE:
+                       if (p->type == TERMTYPE_PDF)
+                               encode(p, "pdf", 3);
+                       else if (p->type == TERMTYPE_PS)
+                               encode(p, "ps", 2);
+                       else if (p->enc == TERMENC_ASCII)
+                               encode(p, "ascii", 5);
+                       else
+                               encode(p, "utf8", 4);
+                       continue;

                case ESCAPE_HORIZ:
                case ESCAPE_HORIZ:

-                       if (a2roffsu(seq, &su, SCALE_EM) == 0)
+                       if (*seq == '|') {
+                               seq++;
+                               uc = -p->col;
+                       } else
+                               uc = 0;
+                       if (a2roffsu(seq, &su, SCALE_EM) == NULL)

                                continue;
                                continue;

-                       uc = term_hspan(p, &su) / 24;
-                       if (uc > 0)
-                               while (uc-- > 0)
+                       uc += term_hen(p, &su);
+                       if (uc > 0) {
+                               while (uc > 0) {

                                        bufferc(p, ASCII_NBRSP);
                                        bufferc(p, ASCII_NBRSP);

-                       else if (p->col > (size_t)(-uc))
+                                       uc -= term_len(p, 1);
+                               }
+                       } else if (p->col > (size_t)(-uc)) {

                                p->col += uc;
                                p->col += uc;

-                       else {
+                       } else {

                                uc += p->col;
                                p->col = 0;
                                uc += p->col;
                                p->col = 0;

-                               if (p->offset > (size_t)(-uc)) {
+                               if (p->tcol->offset > (size_t)(-uc)) {

                                        p->ti += uc;
                                        p->ti += uc;

-                                       p->offset += uc;
+                                       p->tcol->offset += uc;

                                } else {
                                } else {

-                                       p->ti -= p->offset;
-                                       p->offset = 0;
+                                       p->ti -= p->tcol->offset;
+                                       p->tcol->offset = 0;

                                }
                        }
                        continue;
                case ESCAPE_HLINE:
                                }
                        }
                        continue;
                case ESCAPE_HLINE:

-                       if (a2roffsu(seq, &su, SCALE_EM) == 0)
+                       if ((cp = a2roffsu(seq, &su, SCALE_EM)) == NULL)

                                continue;
                                continue;

-                       uc = term_hspan(p, &su) / 24;
+                       uc = term_hen(p, &su);

                        if (uc <= 0) {
                        if (uc <= 0) {

-                               if (p->rmargin <= p->offset)
+                               if (p->tcol->rmargin <= p->tcol->offset)

                                        continue;
                                        continue;

-                               lsz = p->rmargin - p->offset;
+                               lsz = p->tcol->rmargin - p->tcol->offset;

                        } else
                                lsz = uc;
                        } else
                                lsz = uc;

-                       while (sz &&
-                           strchr(" %&()*+-./0123456789:<=>", *seq)) {
-                               seq++;
-                               sz--;
-                       }
-                       if (sz && strchr("cifMmnPpuv", *seq)) {
-                               seq++;
-                               sz--;
-                       }
-                       if (sz == 0)
+                       if (*cp == seq[-1])

                                uc = -1;
                                uc = -1;

-                       else if (*seq == '\\') {
-                               seq++;
+                       else if (*cp == '\\') {
+                               seq = cp + 1;

                                esc = mandoc_escape(&seq, &cp, &sz);
                                switch (esc) {
                                case ESCAPE_UNICODE:
                                esc = mandoc_escape(&seq, &cp, &sz);
                                switch (esc) {
                                case ESCAPE_UNICODE:


@@ -515,12 +680,15 @@ term_word(struct termp *p, const char *word)
                                case ESCAPE_SPECIAL:
                                        uc = mchars_spec2cp(cp, sz);
                                        break;
                                case ESCAPE_SPECIAL:
                                        uc = mchars_spec2cp(cp, sz);
                                        break;

+                               case ESCAPE_UNDEF:
+                                       uc = *seq;
+                                       break;

                                default:
                                        uc = -1;
                                        break;
                                }
                        } else
                                default:
                                        uc = -1;
                                        break;
                                }
                        } else

-                               uc = *seq;
+                               uc = *cp;

                        if (uc < 0x20 || (uc > 0x7E && uc < 0xA0))
                                uc = '_';
                        if (p->enc == TERMENC_ASCII) {
                        if (uc < 0x20 || (uc > 0x7E && uc < 0xA0))
                                uc = '_';
                        if (p->enc == TERMENC_ASCII) {


@@ -556,12 +724,12 @@ term_word(struct termp *p, const char *word)
                                }
                        }
                        /* Trim trailing backspace/blank pair. */
                                }
                        }
                        /* Trim trailing backspace/blank pair. */

-                       if (p->lastcol > 2 &&
-                           (p->buf[p->lastcol - 1] == ' ' ||
-                            p->buf[p->lastcol - 1] == '\t'))
-                               p->lastcol -= 2;
-                       if (p->col > p->lastcol)
-                               p->col = p->lastcol;
+                       if (p->tcol->lastcol > 2 &&
+                           (p->tcol->buf[p->tcol->lastcol - 1] == ' ' ||
+                            p->tcol->buf[p->tcol->lastcol - 1] == '\t'))
+                               p->tcol->lastcol -= 2;
+                       if (p->col > p->tcol->lastcol)
+                               p->col = p->tcol->lastcol;

                        continue;
                default:
                        continue;
                        continue;
                default:
                        continue;


@@ -586,15 +754,13 @@ term_word(struct termp *p, const char *word)
 }
 
 static void
 }
 
 static void

-adjbuf(struct termp *p, size_t sz)
+adjbuf(struct termp_col *c, size_t sz)

 {
 {

-
-       if (0 == p->maxcols)
-               p->maxcols = 1024;
-       while (sz >= p->maxcols)
-               p->maxcols <<= 2;
-
-       p->buf = mandoc_reallocarray(p->buf, p->maxcols, sizeof(int));
+       if (c->maxcols == 0)
+               c->maxcols = 1024;
+       while (c->maxcols <= sz)
+               c->maxcols <<= 2;
+       c->buf = mandoc_reallocarray(c->buf, c->maxcols, sizeof(*c->buf));

 }
 
 static void
 }
 
 static void


@@ -604,12 +770,12 @@ bufferc(struct termp *p, char c)
                (*p->letter)(p, c);
                return;
        }
                (*p->letter)(p, c);
                return;
        }

-       if (p->col + 1 >= p->maxcols)
-               adjbuf(p, p->col + 1);
-       if (p->lastcol <= p->col || (c != ' ' && c != ASCII_NBRSP))
-               p->buf[p->col] = c;
-       if (p->lastcol < ++p->col)
-               p->lastcol = p->col;
+       if (p->col + 1 >= p->tcol->maxcols)
+               adjbuf(p->tcol, p->col + 1);
+       if (p->tcol->lastcol <= p->col || (c != ' ' && c != ASCII_NBRSP))
+               p->tcol->buf[p->col] = c;
+       if (p->tcol->lastcol < ++p->col)
+               p->tcol->lastcol = p->col;

 }
 
 /*
 }
 
 /*


@@ -627,34 +793,35 @@ encode1(struct termp *p, int c)
                return;
        }
 
                return;
        }
 

-       if (p->col + 7 >= p->maxcols)
-               adjbuf(p, p->col + 7);
+       if (p->col + 7 >= p->tcol->maxcols)
+               adjbuf(p->tcol, p->col + 7);

 
        f = (c == ASCII_HYPH || c > 127 || isgraph(c)) ?
            p->fontq[p->fonti] : TERMFONT_NONE;
 
        if (p->flags & TERMP_BACKBEFORE) {
 
        f = (c == ASCII_HYPH || c > 127 || isgraph(c)) ?
            p->fontq[p->fonti] : TERMFONT_NONE;
 
        if (p->flags & TERMP_BACKBEFORE) {

-               if (p->buf[p->col - 1] == ' ' || p->buf[p->col - 1] == '\t')
+               if (p->tcol->buf[p->col - 1] == ' ' ||
+                   p->tcol->buf[p->col - 1] == '\t')

                        p->col--;
                else
                        p->col--;
                else

-                       p->buf[p->col++] = 8;
+                       p->tcol->buf[p->col++] = '\b';

                p->flags &= ~TERMP_BACKBEFORE;
        }
                p->flags &= ~TERMP_BACKBEFORE;
        }

-       if (TERMFONT_UNDER == f || TERMFONT_BI == f) {
-               p->buf[p->col++] = '_';
-               p->buf[p->col++] = 8;
+       if (f == TERMFONT_UNDER || f == TERMFONT_BI) {
+               p->tcol->buf[p->col++] = '_';
+               p->tcol->buf[p->col++] = '\b';

        }
        }

-       if (TERMFONT_BOLD == f || TERMFONT_BI == f) {
-               if (ASCII_HYPH == c)
-                       p->buf[p->col++] = '-';
+       if (f == TERMFONT_BOLD || f == TERMFONT_BI) {
+               if (c == ASCII_HYPH)
+                       p->tcol->buf[p->col++] = '-';

                else
                else

-                       p->buf[p->col++] = c;
-               p->buf[p->col++] = 8;
+                       p->tcol->buf[p->col++] = c;
+               p->tcol->buf[p->col++] = '\b';

        }
        }

-       if (p->lastcol <= p->col || (c != ' ' && c != ASCII_NBRSP))
-               p->buf[p->col] = c;
-       if (p->lastcol < ++p->col)
-               p->lastcol = p->col;
+       if (p->tcol->lastcol <= p->col || (c != ' ' && c != ASCII_NBRSP))
+               p->tcol->buf[p->col] = c;
+       if (p->tcol->lastcol < ++p->col)
+               p->tcol->lastcol = p->col;

        if (p->flags & TERMP_BACKAFTER) {
                p->flags |= TERMP_BACKBEFORE;
                p->flags &= ~TERMP_BACKAFTER;
        if (p->flags & TERMP_BACKAFTER) {
                p->flags |= TERMP_BACKBEFORE;
                p->flags &= ~TERMP_BACKAFTER;


@@ -672,17 +839,17 @@ encode(struct termp *p, const char *word, size_t sz)
                return;
        }
 
                return;
        }
 

-       if (p->col + 2 + (sz * 5) >= p->maxcols)
-               adjbuf(p, p->col + 2 + (sz * 5));
+       if (p->col + 2 + (sz * 5) >= p->tcol->maxcols)
+               adjbuf(p->tcol, p->col + 2 + (sz * 5));

 
        for (i = 0; i < sz; i++) {
                if (ASCII_HYPH == word[i] ||
                    isgraph((unsigned char)word[i]))
                        encode1(p, word[i]);
                else {
 
        for (i = 0; i < sz; i++) {
                if (ASCII_HYPH == word[i] ||
                    isgraph((unsigned char)word[i]))
                        encode1(p, word[i]);
                else {

-                       if (p->lastcol <= p->col ||
+                       if (p->tcol->lastcol <= p->col ||

                            (word[i] != ' ' && word[i] != ASCII_NBRSP))
                            (word[i] != ' ' && word[i] != ASCII_NBRSP))

-                               p->buf[p->col] = word[i];
+                               p->tcol->buf[p->col] = word[i];

                        p->col++;
 
                        /*
                        p->col++;
 
                        /*


@@ -697,8 +864,8 @@ encode(struct termp *p, const char *word, size_t sz)
                        }
                }
        }
                        }
                }
        }

-       if (p->lastcol < p->col)
-               p->lastcol = p->col;
+       if (p->tcol->lastcol < p->col)
+               p->tcol->lastcol = p->col;

 }
 
 void
 }
 
 void


@@ -722,7 +889,7 @@ term_setwidth(struct termp *p, const char *wstr)
                default:
                        break;
                }
                default:
                        break;
                }

-               if (a2roffsu(wstr, &su, SCALE_MAX))
+               if (a2roffsu(wstr, &su, SCALE_MAX) != NULL)

                        width = term_hspan(p, &su);
                else
                        iop = 0;
                        width = term_hspan(p, &su);
                else
                        iop = 0;


@@ -774,12 +941,8 @@ term_strlen(const struct termp *p, const char *cp)
                switch (*cp) {
                case '\\':
                        cp++;
                switch (*cp) {
                case '\\':
                        cp++;

-                       esc = mandoc_escape(&cp, &seq, &ssz);
-                       if (ESCAPE_ERROR == esc)
-                               continue;
-

                        rhs = NULL;
                        rhs = NULL;

-
+                       esc = mandoc_escape(&cp, &seq, &ssz);

                        switch (esc) {
                        case ESCAPE_UNICODE:
                                uc = mchars_num2uc(seq + 1, ssz - 1);
                        switch (esc) {
                        case ESCAPE_UNICODE:
                                uc = mchars_num2uc(seq + 1, ssz - 1);


@@ -800,6 +963,24 @@ term_strlen(const struct termp *p, const char *cp)
                                                sz += cond_width(p, uc, &skip);
                                }
                                continue;
                                                sz += cond_width(p, uc, &skip);
                                }
                                continue;

+                       case ESCAPE_UNDEF:
+                               uc = *seq;
+                               break;
+                       case ESCAPE_DEVICE:
+                               if (p->type == TERMTYPE_PDF) {
+                                       rhs = "pdf";
+                                       rsz = 3;
+                               } else if (p->type == TERMTYPE_PS) {
+                                       rhs = "ps";
+                                       rsz = 2;
+                               } else if (p->enc == TERMENC_ASCII) {
+                                       rhs = "ascii";
+                                       rsz = 5;
+                               } else {
+                                       rhs = "utf8";
+                                       rsz = 4;
+                               }
+                               break;

                        case ESCAPE_SKIPCHAR:
                                skip = 1;
                                continue;
                        case ESCAPE_SKIPCHAR:
                                skip = 1;
                                continue;


@@ -911,7 +1092,7 @@ term_vspan(const struct termp *p, const struct roffsu *su)
 }
 
 /*
 }
 
 /*

- * Convert a scaling width to basic units, rounding down.
+ * Convert a scaling width to basic units, rounding towards 0.

  */
 int
 term_hspan(const struct termp *p, const struct roffsu *su)
  */
 int
 term_hspan(const struct termp *p, const struct roffsu *su)


@@ -919,3 +1100,17 @@ term_hspan(const struct termp *p, const struct roffsu *su)
 
        return (*p->hspan)(p, su);
 }
 
        return (*p->hspan)(p, su);
 }

+
+/*
+ * Convert a scaling width to basic units, rounding to closest.
+ */
+int
+term_hen(const struct termp *p, const struct roffsu *su)
+{
+       int bu;
+
+       if ((bu = (*p->hspan)(p, su)) >= 0)
+               return (bu + 11) / 24;
+       else
+               return -((-bu + 11) / 24);
+}