X-Git-Url: https://git.cameronkatri.com/mandoc.git/blobdiff_plain/4cd2606cca8b6f78a9d90bec6f4fa0c45c2dcd0f..f7ae141bdbc537700c2e9f197c05fefe73126d5c:/term.c

diff --git a/term.c b/term.c
index 180c5075..1cd4f9d5 100644
--- a/term.c
+++ b/term.c
@@ -1,234 +1,163 @@
-/* $Id: term.c,v 1.71 2009/03/31 13:50:19 kristaps Exp $ */
+/*	$Id: term.c,v 1.220 2014/04/05 21:18:19 schwarze Exp $ */
 /*
- * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@openbsd.org>
+ * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
+ * Copyright (c) 2010-2014 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 copyright notice and this permission notice appear in all
- * copies.
+ * purpose with or without fee is hereby granted, provided that the above
+ * copyright notice and this permission notice appear in all copies.
  *
- * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL
- * WARRANTIES WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED
- * WARRANTIES OF MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE
- * AUTHOR BE LIABLE FOR ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL
- * DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR
- * PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
- * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
- * PERFORMANCE OF THIS SOFTWARE.
+ * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
+ * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
+ * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
+ * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
+ * 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 <err.h>
+#include <ctype.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
 
+#include "mandoc.h"
+#include "mandoc_aux.h"
+#include "out.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_stringa(struct termp *, 
-				const char *, size_t);
-static	int		  term_isopendelim(const char *, int);
-static	int		  term_isclosedelim(const char *, int);
-
-
-void *
-ascii_alloc(void)
-{
-
-	return(term_alloc(TERMENC_ASCII));
-}
-
-
-int
-terminal_run(void *arg, const struct man *man,
-		const struct mdoc *mdoc)
-{
-	struct termp	*p;
-
-	p = (struct termp *)arg;
-
-	if (NULL == p->symtab)
-		p->symtab = term_ascii2htab();
-
-	if (man)
-		return(man_run(p, man));
-	if (mdoc)
-		return(mdoc_run(p, mdoc));
-
-	return(1);
-}
+#include "main.h"
 
+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
-terminal_free(void *arg)
-{
-
-	term_free((struct termp *)arg);
-}
-
-
-static void
 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)
+		mchars_free(p->symtab);
 
 	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, int len)
+void
+term_begin(struct termp *p, term_margin head, 
+		term_margin foot, const void *arg)
 {
 
-	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->headf = head;
+	p->footf = foot;
+	p->argf = arg;
+	(*p->begin)(p);
 }
 
 
-static int
-term_isopendelim(const char *p, int len)
+void
+term_end(struct termp *p)
 {
 
-	if (1 != len)
-		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.
+ * 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:
- *
- *  - 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.
+ * The following flags may be specified:
  *
  *  - 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.
+ *    columns.  In short: don't print a newline and instead expect the
+ *    next call to do the padding up to the start of the next column.
+ *    p->trailspace may be set to 0, 1, or 2, depending on how many
+ *    space characters are required at the end of the column.
+ *
+ *  - 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_NONOBREAK: don't newline when TERMP_NOBREAK is specified.
+ *  - TERMP_HANG: like TERMP_DANGLE, but doesn't newline when
+ *    overrunning, 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).
+ *  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)
 {
-	int		 i, j;
-	size_t		 vsz, vis, maxvis, mmax, bp;
+	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 */
+	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
 	 * 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.
 	 */
+	assert  (p->rmargin >= p->offset);
+	dv     = p->rmargin - p->offset;
+	maxvis = (int)dv > p->overstep ? dv - (size_t)p->overstep : 0;
+	dv     = p->maxrmargin - p->offset;
+	mmax   = (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 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;
+
+	vis = vend = 0;
+	i = 0;
 
-	if ( ! (p->flags & TERMP_NOLPAD))
-		/* LINTED */
-		for (j = 0; j < (int)p->offset; j++)
-			putchar(' ');
+	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
@@ -236,84 +165,144 @@ term_flushln(struct termp *p)
 		 * 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;
-			else if (8 == p->buf[j])
-				j += 1;
-			else
-				vsz++;
-		}
 
-		/*
-		 * Do line-breaking.  If we're greater than our
-		 * break-point and already in-line, break to the next
-		 * line and start writing.  If we're at the line start,
-		 * then write out the word (TODO: hyphenate) and break
-		 * in a subsequent loop invocation.
-		 */
+			/* Back over the the last printed character. */
+			if (8 == p->buf[j]) {
+				assert(j);
+				vend -= (*p->width)(p, p->buf[j - 1]);
+				continue;
+			}
 
-		if ( ! (TERMP_NOBREAK & p->flags)) {
-			if (vis && vis + vsz > bp) {
-				putchar('\n');
-				for (j = 0; j < (int)p->offset; j++)
-					putchar(' ');
-				vis = 0;
-			} 
-		} else if (vis && vis + vsz > bp) {
-			putchar('\n');
-			for (j = 0; j < (int)p->rmargin; j++)
-				putchar(' ');
-			vis = p->rmargin - p->offset;
+			/* 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]);
 		}
 
-		/* 
-		 * Write out the word and a trailing space.  Omit the
-		 * space if we're the last word in the line or beyond
-		 * our breakpoint.
+		/*
+		 * Find out whether we would exceed the right margin.
+		 * If so, break to the next line.
 		 */
+		if (vend > bp && 0 == jhy && vis > 0) {
+			vend -= vis;
+			(*p->endline)(p);
+			p->viscol = 0;
+			if (TERMP_NOBREAK & p->flags) {
+				vbl = p->rmargin;
+				vend += p->rmargin - 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;
+		}
 
-		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;
-			putchar(p->buf[i]);
-		}
-		vis += vsz;
-		if (i < (int)p->col && vis <= bp) {
-			putchar(' ');
-			vis++;
+			if ('\t' == p->buf[i])
+				break;
+			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;
+			}
+			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 = 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)
+		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;
 	}
 
-	/*
-	 * 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)(vis - maxvis +
+				p->trailspace * (*p->width)(p, ' '));
 
-	if (p->flags & TERMP_NOBREAK) {
-		if ( ! (TERMP_NONOBREAK & 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;
+	}
 }
 
 
@@ -327,12 +316,8 @@ term_newln(struct termp *p)
 {
 
 	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);
 }
 
 
@@ -347,249 +332,496 @@ term_vspace(struct termp *p)
 {
 
 	term_newln(p);
-	putchar('\n');
+	p->viscol = 0;
+	if (0 < p->skipvsp)
+		p->skipvsp--;
+	else
+		(*p->endline)(p);
+}
+
+void
+term_fontlast(struct termp *p)
+{
+	enum termfont	 f;
+
+	f = p->fontl;
+	p->fontl = p->fontq[p->fonti];
+	p->fontq[p->fonti] = f;
 }
 
 
-/*
- * 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
-term_word(struct termp *p, const char *word)
+term_fontrepl(struct termp *p, enum termfont f)
 {
-	int 		 i, j, len;
 
-	len = (int)strlen(word);
+	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_fontpush(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);
-	}
+	assert(p->fonti + 1 < 10);
+	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)
+const void *
+term_fontq(struct termp *p)
 {
-	const char	*rhs;
-	size_t		 sz;
 
-	if (NULL == (rhs = term_a2ascii(p->symtab, word, len, &sz))) 
-		return;
-	term_stringa(p, rhs, sz);
+	return(&p->fontq[p->fonti]);
+}
+
+
+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 < (void *)(p->fontq + p->fonti))
+		p->fonti--;
+	assert(p->fonti >= 0);
 }
 
 
+void
+term_fontpop(struct termp *p)
+{
+
+	assert(p->fonti);
+	p->fonti--;
+}
+
 /*
- * 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).
+ * 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
-term_pescape(struct termp *p, const char *word, int *i, int len)
+void
+term_word(struct termp *p, const char *word)
 {
-	int		 j;
+	const char	 nbrsp[2] = { ASCII_NBRSP, 0 };
+	const char	*seq, *cp;
+	char		 c;
+	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 (++(*i) >= len)
-		return;
+	if ( ! (p->flags & TERMP_NONOSPACE))
+		p->flags &= ~TERMP_NOSPACE;
+	else
+		p->flags |= TERMP_NOSPACE;
 
-	if ('(' == word[*i]) {
-		(*i)++;
-		if (*i + 1 >= len)
-			return;
+	p->flags &= ~TERMP_SENTENCE;
 
-		term_nescape(p, &word[*i], 2);
-		(*i)++;
-		return;
+	while ('\0' != *word) {
+		if ('\\' != *word) {
+			if (TERMP_SKIPCHAR & p->flags) {
+				p->flags &= ~TERMP_SKIPCHAR;
+				word++;
+				continue;
+			}
+			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;
+		}
 
-	} 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 ('['):
+		word++;
+		esc = mandoc_escape(&word, &seq, &sz);
+		if (ESCAPE_ERROR == esc)
 			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;
+
+		if (TERMENC_ASCII != p->enc)
+			switch (esc) {
+			case (ESCAPE_UNICODE):
+				uc = mchars_num2uc(seq + 1, sz - 1);
+				if ('\0' == uc)
+					break;
+				encode1(p, uc);
+				continue;
+			case (ESCAPE_SPECIAL):
+				uc = mchars_spec2cp(p->symtab, seq, sz);
+				if (uc <= 0)
+					break;
+				encode1(p, uc);
+				continue;
+			default:
+				break;
+			}
+
+		switch (esc) {
+		case (ESCAPE_UNICODE):
+			encode1(p, '?');
+			break;
+		case (ESCAPE_NUMBERED):
+			c = mchars_num2char(seq, sz);
+			if ('\0' != c)
+				encode(p, &c, 1);
+			break;
+		case (ESCAPE_SPECIAL):
+			cp = mchars_spec2str(p->symtab, seq, sz, &ssz);
+			if (NULL != cp) 
+				encode(p, cp, ssz);
+			else if (1 == ssz)
+				encode(p, seq, sz);
+			break;
+		case (ESCAPE_FONTBOLD):
+			term_fontrepl(p, TERMFONT_BOLD);
 			break;
-		case ('I'):
-			p->flags |= TERMP_UNDER;
+		case (ESCAPE_FONTITALIC):
+			term_fontrepl(p, TERMFONT_UNDER);
 			break;
-		case ('P'):
+		case (ESCAPE_FONTBI):
+			term_fontrepl(p, TERMFONT_BI);
+			break;
+		case (ESCAPE_FONT):
 			/* FALLTHROUGH */
-		case ('R'):
-			p->flags &= ~TERMP_STYLE;
+		case (ESCAPE_FONTROMAN):
+			term_fontrepl(p, TERMFONT_NONE);
+			break;
+		case (ESCAPE_FONTPREV):
+			term_fontlast(p);
+			break;
+		case (ESCAPE_NOSPACE):
+			if (TERMP_SKIPCHAR & p->flags)
+				p->flags &= ~TERMP_SKIPCHAR;
+			else if ('\0' == *word)
+				p->flags |= TERMP_NOSPACE;
+			break;
+		case (ESCAPE_SKIPCHAR):
+			p->flags |= TERMP_SKIPCHAR;
 			break;
 		default:
 			break;
 		}
-		return;
-
-	} else if ('[' != word[*i]) {
-		term_nescape(p, &word[*i], 1);
-		return;
 	}
+	p->flags &= ~TERMP_NBRWORD;
+}
 
-	(*i)++;
-	for (j = 0; word[*i] && ']' != word[*i]; (*i)++, j++)
-		/* Loop... */ ;
+static void
+adjbuf(struct termp *p, size_t sz)
+{
 
-	if (0 == word[*i])
-		return;
+	if (0 == p->maxcols)
+		p->maxcols = 1024;
+	while (sz >= p->maxcols)
+		p->maxcols <<= 2;
 
-	term_nescape(p, &word[*i - j], (size_t)j);
+	p->buf = mandoc_realloc(p->buf, sizeof(int) * p->maxcols);
 }
 
+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.
  */
 static void
-term_pword(struct termp *p, const char *word, int len)
+encode1(struct termp *p, int c)
 {
-	int		 i;
+	enum termfont	  f;
 
-	if (term_isclosedelim(word, len))
-		if ( ! (TERMP_IGNDELIM & p->flags))
-			p->flags |= TERMP_NOSPACE;
+	if (TERMP_SKIPCHAR & p->flags) {
+		p->flags &= ~TERMP_SKIPCHAR;
+		return;
+	}
 
-	if ( ! (TERMP_NOSPACE & p->flags))
-		term_chara(p, ' ');
+	if (p->col + 6 >= p->maxcols)
+		adjbuf(p, p->col + 6);
 
-	if ( ! (p->flags & TERMP_NONOSPACE))
-		p->flags &= ~TERMP_NOSPACE;
+	f = term_fonttop(p);
+
+	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;
+}
+
+static void
+encode(struct termp *p, const char *word, size_t sz)
+{
+	size_t		  i;
 
-	/* 
-	 * If ANSI (word-length styling), then apply our style now,
-	 * before the word.
+	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 (i = 0; i < len; i++) {
-		if ('\\' == word[i]) {
-			term_pescape(p, word, &i, len);
-			continue;
-		}
+	if (TERMFONT_NONE == term_fonttop(p)) {
+		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 (TERMP_STYLE & p->flags) {
-			if (TERMP_BOLD & p->flags) {
-				term_chara(p, word[i]);
-				term_chara(p, 8);
-			}
-			if (TERMP_UNDER & p->flags) {
-				term_chara(p, '_');
-				term_chara(p, 8);
-			}
-		}
+	/* Pre-buffer, assuming worst-case. */
+
+	if (p->col + 1 + (sz * 5) >= p->maxcols)
+		adjbuf(p, p->col + 1 + (sz * 5));
 
-		term_chara(p, word[i]);
+	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];
 	}
+}
 
-	if (term_isopendelim(word, len))
-		p->flags |= TERMP_NOSPACE;
+void
+term_setwidth(struct termp *p, const char *wstr)
+{
+	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->setwidth)(p, iop, width);
 }
 
+size_t
+term_len(const struct termp *p, size_t sz)
+{
 
-/*
- * Like term_chara() but for arbitrary-length buffers.  Resize the
- * buffer by a factor of two (if the buffer is less than that) or the
- * buffer's size.
- */
-static void
-term_stringa(struct termp *p, const char *c, size_t sz)
+	return((*p->width)(p, ' ') * sz);
+}
+
+static size_t
+cond_width(const struct termp *p, int c, int *skip)
 {
-	size_t		 s;
 
-	if (0 == sz)
-		return;
+	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, c;
+	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)
+				return(sz);
+
+			if (TERMENC_ASCII != p->enc)
+				switch (esc) {
+				case (ESCAPE_UNICODE):
+					c = mchars_num2uc
+						(seq + 1, ssz - 1);
+					if ('\0' == c)
+						break;
+					sz += cond_width(p, c, &skip);
+					continue;
+				case (ESCAPE_SPECIAL):
+					c = mchars_spec2cp
+						(p->symtab, seq, ssz);
+					if (c <= 0)
+						break;
+					sz += cond_width(p, c, &skip);
+					continue;
+				default:
+					break;
+				}
+
+			rhs = NULL;
+
+			switch (esc) {
+			case (ESCAPE_UNICODE):
+				sz += cond_width(p, '?', &skip);
+				break;
+			case (ESCAPE_NUMBERED):
+				c = mchars_num2char(seq, ssz);
+				if ('\0' != c)
+					sz += cond_width(p, c, &skip);
+				break;
+			case (ESCAPE_SPECIAL):
+				rhs = mchars_spec2str
+					(p->symtab, seq, ssz, &rsz);
+
+				if (ssz != 1 || rhs)
+					break;
+
+				rhs = seq;
+				rsz = ssz;
+				break;
+			case (ESCAPE_SKIPCHAR):
+				skip = 1;
+				break;
+			default:
+				break;
+			}
+
+			if (NULL == rhs)
+				break;
+
+			if (skip) {
+				skip = 0;
+				break;
+			}
 
-	assert(c);
-	if (p->col + sz >= p->maxcols) {
-		if (0 == p->maxcols)
-			p->maxcols = 256;
-		s = sz > p->maxcols * 2 ? sz : p->maxcols * 2;
-		p->buf = realloc(p->buf, s);
-		if (NULL == p->buf)
-			err(1, "realloc");
-		p->maxcols = s;
+			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;
+		}
 	}
 
-	(void)memcpy(&p->buf[(int)p->col], c, sz);
-	p->col += sz;
+	return(sz);
 }
 
-
-/*
- * 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)
+/* ARGSUSED */
+size_t
+term_vspan(const struct termp *p, const struct roffsu *su)
 {
-	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;
+	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;
 	}
-	p->buf[(int)(p->col)++] = 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);
+}