===================================================================
RCS file: /cvs/mandoc/term.c,v
retrieving revision 1.129
retrieving revision 1.181
diff -u -p -r1.129 -r1.181
--- mandoc/term.c	2010/03/23 12:42:22	1.129
+++ mandoc/term.c	2011/03/22 10:13:01	1.181
@@ -1,6 +1,7 @@
-/*	$Id: term.c,v 1.129 2010/03/23 12:42:22 kristaps Exp $ */
+/*	$Id: term.c,v 1.181 2011/03/22 10:13:01 kristaps Exp $ */
 /*
- * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
+ * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons <kristaps@bsd.lv>
+ * Copyright (c) 2010, 2011 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
@@ -22,67 +23,63 @@
 
 #include <assert.h>
 #include <ctype.h>
+#include <stdint.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
-#include <time.h>
 
-#include "chars.h"
+#include "mandoc.h"
 #include "out.h"
 #include "term.h"
-#include "man.h"
-#include "mdoc.h"
 #include "main.h"
 
-static	struct termp	 *term_alloc(enum termenc);
-static	void		  term_free(struct termp *);
-static	void		  spec(struct termp *, const char *, size_t);
+static	void		  spec(struct termp *, enum roffdeco,
+				const char *, size_t);
 static	void		  res(struct termp *, const char *, size_t);
-static	void		  buffera(struct termp *, const char *, size_t);
 static	void		  bufferc(struct termp *, char);
 static	void		  adjbuf(struct termp *p, size_t);
 static	void		  encode(struct termp *, const char *, size_t);
 
 
-void *
-ascii_alloc(void)
+void
+term_free(struct termp *p)
 {
 
-	return(term_alloc(TERMENC_ASCII));
+	if (p->buf)
+		free(p->buf);
+	if (p->symtab)
+		chars_free(p->symtab);
+
+	free(p);
 }
 
 
 void
-terminal_free(void *arg)
+term_begin(struct termp *p, term_margin head, 
+		term_margin foot, const void *arg)
 {
 
-	term_free((struct termp *)arg);
+	p->headf = head;
+	p->footf = foot;
+	p->argf = arg;
+	(*p->begin)(p);
 }
 
 
-static void
-term_free(struct termp *p)
+void
+term_end(struct termp *p)
 {
 
-	if (p->buf)
-		free(p->buf);
-	if (p->symtab)
-		chars_free(p->symtab);
-
-	free(p);
+	(*p->end)(p);
 }
 
 
-static struct termp *
+struct termp *
 term_alloc(enum termenc enc)
 {
-	struct termp *p;
+	struct termp	*p;
 
-	p = calloc(1, sizeof(struct termp));
-	if (NULL == p) {
-		perror(NULL);
-		exit(EXIT_FAILURE);
-	}
+	p = mandoc_calloc(1, sizeof(struct termp));
 	p->enc = enc;
 	return(p);
 }
@@ -92,13 +89,11 @@ term_alloc(enum termenc enc)
  * 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.
+ * fragment of a columnar list (`Bl -tag' or `Bl -column'), which does
+ * not have a trailing newline.
  *
- * Specifically, a line is whatever's in p->buf of length p->col, which
- * is zeroed after this function returns.
+ * The following flags may be specified:
  *
- * 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.
@@ -131,10 +126,13 @@ term_flushln(struct termp *p)
 	int		 i;     /* current input position in p->buf */
 	size_t		 vis;   /* current visual position on output */
 	size_t		 vbl;   /* number of blanks to prepend to output */
-	size_t		 vsz;   /* visual characters to write to output */
+	size_t		 vend;	/* end of word visual position on output */
 	size_t		 bp;    /* visual right border position */
-	int		 j;     /* temporary loop index */
-	size_t		 maxvis, mmax;
+	size_t		 dv;    /* temporary for visual pos calculations */
+	int		 j;     /* temporary loop index for p->buf */
+	int		 jhy;	/* last hyph before overflow w/r/t j */
+	size_t		 maxvis; /* output position of visible boundary */
+	size_t		 mmax; /* used in calculating bp */
 
 	/*
 	 * First, establish the maximum columns of "visible" content.
@@ -142,115 +140,143 @@ term_flushln(struct termp *p)
 	 * an indentation, but can be, for tagged lists or columns, a
 	 * small set of values. 
 	 */
+	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 = (int)(p->rmargin - p->offset) - p->overstep < 0 ?
-		/* LINTED */ 
-		0 : p->rmargin - p->offset - p->overstep;
-	mmax = (int)(p->maxrmargin - p->offset) - p->overstep < 0 ?
-		/* LINTED */
-		0 : p->maxrmargin - p->offset - p->overstep;
-
 	bp = TERMP_NOBREAK & p->flags ? mmax : maxvis;
 
-	/* 
-	 * FIXME: if bp is zero, we still output the first word before
-	 * breaking the line.
-	 */
-
-	vis = 0;
-
 	/*
-	 * If in the standard case (left-justified), then begin with our
-	 * indentation, otherwise (columns, etc.) just start spitting
-	 * out text.
+	 * Indent the first line of a paragraph.
 	 */
+	vbl = p->flags & TERMP_NOLPAD ? (size_t)0 : p->offset;
 
-	if ( ! (p->flags & TERMP_NOLPAD))
-		/* LINTED */
-		for (j = 0; j < (int)p->offset; j++)
-			putchar(' ');
+	vis = vend = 0;
+	i = 0;
 
-	for (i = 0; i < (int)p->col; i++) {
+	while (i < (int)p->col) {
 		/*
+		 * Handle literal tab characters: collapse all
+		 * subsequent tabs into a single huge set of spaces.
+		 */
+		while (i < (int)p->col && '\t' == p->buf[i]) {
+			vend = (vis / p->tabwidth + 1) * p->tabwidth;
+			vbl += vend - vis;
+			vis = vend;
+			i++;
+		}
+
+		/*
 		 * 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).
 		 */
 
-		/* LINTED */
-		for (j = i, vsz = 0; j < (int)p->col; j++) {
-			if (j && ' ' == p->buf[j]) 
+		for (j = i, jhy = 0; j < (int)p->col; j++) {
+			if ((j && ' ' == p->buf[j]) || '\t' == p->buf[j])
 				break;
-			else if (8 == p->buf[j])
-				vsz--;
-			else
-				vsz++;
+
+			/* Back over the the last printed character. */
+			if (8 == p->buf[j]) {
+				assert(j);
+				vend -= (*p->width)(p, p->buf[j - 1]);
+				continue;
+			}
+
+			/* Regular word. */
+			/* Break at the hyphen point if we overrun. */
+			if (vend > vis && vend < bp && 
+					ASCII_HYPH == p->buf[j])
+				jhy = j;
+
+			vend += (*p->width)(p, p->buf[j]);
 		}
 
 		/*
-		 * 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);
-
-		/*
 		 * Find out whether we would exceed the right margin.
-		 * If so, break to the next line.  (TODO: hyphenate)
-		 * Otherwise, write the chosen number of blanks now.
+		 * If so, break to the next line.
 		 */
-		if (vis && vis + vbl + vsz > bp) {
-			putchar('\n');
+		if (vend > bp && 0 == jhy && vis > 0) {
+			vend -= vis;
+			(*p->endline)(p);
 			if (TERMP_NOBREAK & p->flags) {
-				for (j = 0; j < (int)p->rmargin; j++)
-					putchar(' ');
-				vis = p->rmargin - p->offset;
+				p->viscol = p->rmargin;
+				(*p->advance)(p, p->rmargin);
+				vend += p->rmargin - p->offset;
 			} else {
-				for (j = 0; j < (int)p->offset; j++)
-					putchar(' ');
-				vis = 0;
+				p->viscol = 0;
+				vbl = p->offset;
 			}
+
 			/* Remove the p->overstep width. */
-			bp += (int)/* LINTED */
-				p->overstep;
+
+			bp += (size_t)p->overstep;
 			p->overstep = 0;
-		} else {
-			for (j = 0; j < (int)vbl; j++)
-				putchar(' ');
-			vis += vbl;
 		}
 
-		/*
-		 * Finally, write out the word.
-		 */
+		/* Write out the [remaining] word. */
 		for ( ; i < (int)p->col; i++) {
-			if (' ' == p->buf[i])
+			if (vend > bp && jhy > 0 && i > jhy)
 				break;
+			if ('\t' == p->buf[i])
+				break;
+			if (' ' == p->buf[i]) {
+				j = i;
+				while (' ' == p->buf[i])
+					i++;
+				dv = (size_t)(i - j) * (*p->width)(p, ' ');
+				vbl += dv;
+				vend += dv;
+				break;
+			}
+			if (ASCII_NBRSP == p->buf[i]) {
+				vbl += (*p->width)(p, ' ');
+				continue;
+			}
 
-			/* The unit sep. is a non-breaking space. */
-			if (31 == p->buf[i])
-				putchar(' ');
-			else
-				putchar(p->buf[i]);
+			/*
+			 * Now we definitely know there will be
+			 * printable characters to output,
+			 * so write preceding white space now.
+			 */
+			if (vbl) {
+				(*p->advance)(p, vbl);
+				p->viscol += vbl;
+				vbl = 0;
+			}
+
+			if (ASCII_HYPH == p->buf[i]) {
+				(*p->letter)(p, '-');
+				p->viscol += (*p->width)(p, '-');
+			} else {
+				(*p->letter)(p, p->buf[i]);
+				p->viscol += (*p->width)(p, p->buf[i]);
+			}
 		}
-		vis += vsz;
+		vis = vend;
 	}
 
+	/*
+	 * If there was trailing white space, it was not printed;
+	 * so reset the cursor position accordingly.
+	 */
+	vis -= vbl;
+
 	p->col = 0;
 	p->overstep = 0;
 
 	if ( ! (TERMP_NOBREAK & p->flags)) {
-		putchar('\n');
+		p->viscol = 0;
+		(*p->endline)(p);
 		return;
 	}
 
 	if (TERMP_HANG & p->flags) {
 		/* We need one blank after the tag. */
-		p->overstep = /* LINTED */
-			vis - maxvis + 1;
+		p->overstep = (int)(vis - maxvis + (*p->width)(p, ' '));
 
 		/*
 		 * Behave exactly the same way as groff:
@@ -264,8 +290,7 @@ term_flushln(struct termp *p)
 		 */
 		if (p->overstep >= -1) {
 			assert((int)maxvis + p->overstep >= 0);
-			/* LINTED */
-			maxvis += p->overstep;
+			maxvis += (size_t)p->overstep;
 		} else
 			p->overstep = 0;
 
@@ -273,14 +298,15 @@ term_flushln(struct termp *p)
 		return;
 
 	/* Right-pad. */
-	if (maxvis > vis + /* LINTED */
-			((TERMP_TWOSPACE & p->flags) ? 1 : 0))  
-		for ( ; vis < maxvis; vis++)
-			putchar(' ');
-	else {	/* ...or newline break. */
-		putchar('\n');
-		for (i = 0; i < (int)p->rmargin; i++)
-			putchar(' ');
+	if (maxvis > vis +
+	    ((TERMP_TWOSPACE & p->flags) ? (*p->width)(p, ' ') : 0)) {
+		p->viscol += maxvis - vis;
+		(*p->advance)(p, maxvis - vis);
+		vis += (maxvis - vis);
+	} else {	/* ...or newline break. */
+		(*p->endline)(p);
+		p->viscol = p->rmargin;
+		(*p->advance)(p, p->rmargin);
 	}
 }
 
@@ -295,7 +321,7 @@ term_newln(struct termp *p)
 {
 
 	p->flags |= TERMP_NOSPACE;
-	if (0 == p->col) {
+	if (0 == p->col && 0 == p->viscol) {
 		p->flags &= ~TERMP_NOLPAD;
 		return;
 	}
@@ -315,19 +341,33 @@ term_vspace(struct termp *p)
 {
 
 	term_newln(p);
-	putchar('\n');
+	p->viscol = 0;
+	(*p->endline)(p);
 }
 
 
 static void
-spec(struct termp *p, const char *word, size_t len)
+numbered(struct termp *p, const char *word, size_t len)
 {
 	const char	*rhs;
+
+	rhs = chars_num2char(word, len);
+	if (rhs) 
+		encode(p, rhs, 1);
+}
+
+
+static void
+spec(struct termp *p, enum roffdeco d, const char *word, size_t len)
+{
+	const char	*rhs;
 	size_t		 sz;
 
-	rhs = chars_a2ascii(p->symtab, word, len, &sz);
+	rhs = chars_spec2str(p->symtab, word, len, &sz);
 	if (rhs) 
 		encode(p, rhs, sz);
+	else if (DECO_SSPECIAL == d)
+		encode(p, word, len);
 }
 
 
@@ -337,7 +377,7 @@ res(struct termp *p, const char *word, size_t len)
 	const char	*rhs;
 	size_t		 sz;
 
-	rhs = chars_a2res(p->symtab, word, len, &sz);
+	rhs = chars_res2str(p->symtab, word, len, &sz);
 	if (rhs)
 		encode(p, rhs, sz);
 }
@@ -417,62 +457,55 @@ void
 term_word(struct termp *p, const char *word)
 {
 	const char	*sv, *seq;
-	int		 sz;
 	size_t		 ssz;
 	enum roffdeco	 deco;
 
 	sv = word;
 
-	if (word[0] && '\0' == word[1])
-		switch (word[0]) {
-		case('.'):
-			/* FALLTHROUGH */
-		case(','):
-			/* FALLTHROUGH */
-		case(';'):
-			/* FALLTHROUGH */
-		case(':'):
-			/* FALLTHROUGH */
-		case('?'):
-			/* FALLTHROUGH */
-		case('!'):
-			/* FALLTHROUGH */
-		case(')'):
-			/* FALLTHROUGH */
-		case(']'):
-			/* FALLTHROUGH */
-		case('}'):
-			if ( ! (TERMP_IGNDELIM & p->flags))
-				p->flags |= TERMP_NOSPACE;
-			break;
-		default:
-			break;
-		}
+	if (DELIM_CLOSE == mandoc_isdelim(word))
+		if ( ! (TERMP_IGNDELIM & p->flags))
+			p->flags |= TERMP_NOSPACE;
 
-	if ( ! (TERMP_NOSPACE & p->flags))
-		bufferc(p, ' ');
+	if ( ! (TERMP_NOSPACE & p->flags)) {
+		if ( ! (TERMP_KEEP & p->flags)) {
+			if (TERMP_PREKEEP & p->flags)
+				p->flags |= TERMP_KEEP;
+			bufferc(p, ' ');
+			if (TERMP_SENTENCE & p->flags)
+				bufferc(p, ' ');
+		} else
+			bufferc(p, ASCII_NBRSP);
+	}
 
 	if ( ! (p->flags & TERMP_NONOSPACE))
 		p->flags &= ~TERMP_NOSPACE;
+	else
+		p->flags |= TERMP_NOSPACE;
 
-	/* FIXME: use strcspn. */
+	p->flags &= ~(TERMP_SENTENCE | TERMP_IGNDELIM);
 
 	while (*word) {
-		if ('\\' != *word) {
-			encode(p, word, 1);
-			word++;
+		if ((ssz = strcspn(word, "\\")) > 0)
+			encode(p, word, ssz);
+
+		word += (int)ssz;
+		if ('\\' != *word)
 			continue;
-		}
 
 		seq = ++word;
-		sz = a2roffdeco(&deco, &seq, &ssz);
+		word += a2roffdeco(&deco, &seq, &ssz);
 
 		switch (deco) {
+		case (DECO_NUMBERED):
+			numbered(p, seq, ssz);
+			break;
 		case (DECO_RESERVED):
 			res(p, seq, ssz);
 			break;
 		case (DECO_SPECIAL):
-			spec(p, seq, ssz);
+			/* FALLTHROUGH */
+		case (DECO_SSPECIAL):
+			spec(p, deco, seq, ssz);
 			break;
 		case (DECO_BOLD):
 			term_fontrepl(p, TERMFONT_BOLD);
@@ -490,23 +523,12 @@ term_word(struct termp *p, const char *word)
 			break;
 		}
 
-		word += sz;
 		if (DECO_NOSPACE == deco && '\0' == *word)
 			p->flags |= TERMP_NOSPACE;
 	}
 
-	if (sv[0] && 0 == sv[1])
-		switch (sv[0]) {
-		case('('):
-			/* FALLTHROUGH */
-		case('['):
-			/* FALLTHROUGH */
-		case('{'):
-			p->flags |= TERMP_NOSPACE;
-			break;
-		default:
-			break;
-		}
+	if (DELIM_OPEN == mandoc_isdelim(sv))
+		p->flags |= TERMP_NOSPACE;
 }
 
 
@@ -519,27 +541,11 @@ adjbuf(struct termp *p, size_t sz)
 	while (sz >= p->maxcols)
 		p->maxcols <<= 2;
 
-	p->buf = realloc(p->buf, p->maxcols);
-	if (NULL == p->buf) {
-		perror(NULL);
-		exit(EXIT_FAILURE);
-	}
+	p->buf = mandoc_realloc(p->buf, p->maxcols);
 }
 
 
 static void
-buffera(struct termp *p, const char *word, size_t sz)
-{
-
-	if (p->col + sz >= p->maxcols) 
-		adjbuf(p, p->col + sz);
-
-	memcpy(&p->buf[(int)p->col], word, sz);
-	p->col += sz;
-}
-
-
-static void
 bufferc(struct termp *p, char c)
 {
 
@@ -563,30 +569,103 @@ encode(struct termp *p, const char *word, size_t sz)
 	 */
 
 	if (TERMFONT_NONE == (f = term_fonttop(p))) {
-		buffera(p, word, sz);
+		if (p->col + sz >= p->maxcols) 
+			adjbuf(p, p->col + sz);
+		memcpy(&p->buf[(int)p->col], word, sz);
+		p->col += sz;
 		return;
 	}
 
+	/* Pre-buffer, assuming worst-case. */
+
+	if (p->col + 1 + (sz * 3) >= p->maxcols)
+		adjbuf(p, p->col + 1 + (sz * 3));
+
 	for (i = 0; i < (int)sz; i++) {
 		if ( ! isgraph((u_char)word[i])) {
-			bufferc(p, word[i]);
+			p->buf[(int)p->col++] = word[i];
 			continue;
 		}
 
 		if (TERMFONT_UNDER == f)
-			bufferc(p, '_');
+			p->buf[(int)p->col++] = '_';
 		else
-			bufferc(p, word[i]);
+			p->buf[(int)p->col++] = word[i];
 
-		bufferc(p, 8);
-		bufferc(p, word[i]);
+		p->buf[(int)p->col++] = 8;
+		p->buf[(int)p->col++] = word[i];
 	}
 }
 
 
 size_t
-term_vspan(const struct roffsu *su)
+term_len(const struct termp *p, size_t sz)
 {
+
+	return((*p->width)(p, ' ') * sz);
+}
+
+
+size_t
+term_strlen(const struct termp *p, const char *cp)
+{
+	size_t		 sz, ssz, rsz, i;
+	enum roffdeco	 d;
+	const char	*seq, *rhs;
+
+	for (sz = 0; '\0' != *cp; )
+		/*
+		 * Account for escaped sequences within string length
+		 * calculations.  This follows the logic in term_word()
+		 * as we must calculate the width of produced strings.
+		 */
+		if ('\\' == *cp) {
+			seq = ++cp;
+			cp += a2roffdeco(&d, &seq, &ssz);
+
+			switch (d) {
+			case (DECO_RESERVED):
+				rhs = chars_res2str
+					(p->symtab, seq, ssz, &rsz);
+				break;
+			case (DECO_SPECIAL):
+				/* FALLTHROUGH */
+			case (DECO_SSPECIAL):
+				rhs = chars_spec2str
+					(p->symtab, seq, ssz, &rsz);
+
+				/* Allow for one-char escapes. */
+				if (DECO_SSPECIAL != d || rhs)
+					break;
+
+				rhs = seq;
+				rsz = ssz;
+				break;
+			default:
+				rhs = NULL;
+				break;
+			}
+
+			if (rhs)
+				for (i = 0; i < rsz; i++)
+					sz += (*p->width)(p, *rhs++);
+		} else if (ASCII_NBRSP == *cp) {
+			sz += (*p->width)(p, ' ');
+			cp++;
+		} else if (ASCII_HYPH == *cp) {
+			sz += (*p->width)(p, '-');
+			cp++;
+		} else
+			sz += (*p->width)(p, *cp++);
+
+	return(sz);
+}
+
+
+/* ARGSUSED */
+size_t
+term_vspan(const struct termp *p, const struct roffsu *su)
+{
 	double		 r;
 
 	switch (su->unit) {
@@ -621,41 +700,13 @@ term_vspan(const struct roffsu *su)
 
 
 size_t
-term_hspan(const struct roffsu *su)
+term_hspan(const struct termp *p, const struct roffsu *su)
 {
-	double		 r;
+	double		 v;
 
-	/* XXX: CM, IN, and PT are approximations. */
-
-	switch (su->unit) {
-	case (SCALE_CM):
-		r = 4 * su->scale;
-		break;
-	case (SCALE_IN):
-		/* XXX: this is an approximation. */
-		r = 10 * su->scale;
-		break;
-	case (SCALE_PC):
-		r = (10 * su->scale) / 6;
-		break;
-	case (SCALE_PT):
-		r = (10 * su->scale) / 72;
-		break;
-	case (SCALE_MM):
-		r = su->scale / 1000; /* FIXME: double-check. */
-		break;
-	case (SCALE_VS):
-		r = su->scale * 2 - 1; /* FIXME: double-check. */
-		break;
-	default:
-		r = su->scale;
-		break;
-	}
-
-	if (r < 0.0)
-		r = 0.0;
-	return((size_t)/* LINTED */
-			r);
+	v = ((*p->hspan)(p, su));
+	if (v < 0.0)
+		v = 0.0;
+	return((size_t) /* LINTED */
+			v);
 }
-
-