===================================================================
RCS file: /cvs/mandoc/term.c,v
retrieving revision 1.121
retrieving revision 1.140
diff -u -p -r1.121 -r1.140
--- mandoc/term.c	2009/11/05 07:21:02	1.121
+++ mandoc/term.c	2010/05/25 12:37:20	1.140
@@ -1,4 +1,4 @@
-/*	$Id: term.c,v 1.121 2009/11/05 07:21:02 kristaps Exp $ */
+/*	$Id: term.c,v 1.140 2010/05/25 12:37:20 kristaps Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -14,12 +14,20 @@
  * 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 <ctype.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
 #include <time.h>
 
+#include "mandoc.h"
 #include "chars.h"
 #include "out.h"
 #include "term.h"
@@ -27,26 +35,21 @@
 #include "mdoc.h"
 #include "main.h"
 
-/* FIXME: accomodate non-breaking, non-collapsing white-space. */
-/* FIXME: accomodate non-breaking, collapsing white-space. */
-
-static	struct termp	 *term_alloc(enum termenc);
+static	struct termp	 *term_alloc(enum termenc, size_t);
 static	void		  term_free(struct termp *);
+static	void		  spec(struct termp *, 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);
 
-static	void		  do_escaped(struct termp *, const char **);
-static	void		  do_special(struct termp *,
-				const char *, size_t);
-static	void		  do_reserved(struct termp *,
-				const char *, size_t);
-static	void		  buffer(struct termp *, char);
-static	void		  encode(struct termp *, char);
 
-
 void *
-ascii_alloc(void)
+ascii_alloc(size_t width)
 {
 
-	return(term_alloc(TERMENC_ASCII));
+	return(term_alloc(TERMENC_ASCII, width));
 }
 
 
@@ -72,7 +75,7 @@ term_free(struct termp *p)
 
 
 static struct termp *
-term_alloc(enum termenc enc)
+term_alloc(enum termenc enc, size_t width)
 {
 	struct termp *p;
 
@@ -81,8 +84,12 @@ term_alloc(enum termenc enc)
 		perror(NULL);
 		exit(EXIT_FAILURE);
 	}
-	p->maxrmargin = 78;
+	p->tabwidth = 5;
 	p->enc = enc;
+	/* Enforce some lower boundary. */
+	if (width < 60)
+		width = 60;
+	p->defrmargin = width - 2;
 	return(p);
 }
 
@@ -91,13 +98,11 @@ 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.
@@ -130,11 +135,11 @@ 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 */
+	int		 jhy;	/* last hyphen before line overflow */
 	size_t		 maxvis, mmax;
-	static int	 overstep = 0;
 
 	/*
 	 * First, establish the maximum columns of "visible" content.
@@ -145,34 +150,39 @@ term_flushln(struct termp *p)
 
 	assert(p->offset < p->rmargin);
 
-	maxvis = (int)(p->rmargin - p->offset) - overstep < 0 ?
+	maxvis = (int)(p->rmargin - p->offset) - p->overstep < 0 ?
 		/* LINTED */ 
-		0 : p->rmargin - p->offset - overstep;
-	mmax = (int)(p->maxrmargin - p->offset) - overstep < 0 ?
+		0 : p->rmargin - p->offset - p->overstep;
+	mmax = (int)(p->maxrmargin - p->offset) - p->overstep < 0 ?
 		/* LINTED */
-		0 : p->maxrmargin - p->offset - overstep;
+		0 : p->maxrmargin - p->offset - p->overstep;
 
 	bp = TERMP_NOBREAK & p->flags ? mmax : maxvis;
 
+	/*
+	 * Indent the first line of a paragraph.
+	 */
+	vbl = p->flags & TERMP_NOLPAD ? 0 : p->offset;
+
 	/* 
 	 * FIXME: if bp is zero, we still output the first word before
 	 * breaking the line.
 	 */
 
-	vis = 0;
+	vis = vend = i = 0;
+	while (i < (int)p->col) {
 
-	/*
-	 * If in the standard case (left-justified), then begin with our
-	 * indentation, otherwise (columns, etc.) just start spitting
-	 * out text.
-	 */
+		/*
+		 * Handle literal tab characters.
+		 */
+		for (j = i; j < (int)p->col; j++) {
+			if ('\t' != p->buf[j])
+				break;
+			vend = (vis/p->tabwidth+1)*p->tabwidth;
+			vbl += vend - vis;
+			vis = vend;
+		}
 
-	if ( ! (p->flags & TERMP_NOLPAD))
-		/* LINTED */
-		for (j = 0; j < (int)p->offset; j++)
-			putchar(' ');
-
-	for (i = 0; i < (int)p->col; i++) {
 		/*
 		 * Count up visible word characters.  Control sequences
 		 * (starting with the CSI) aren't counted.  A space
@@ -181,75 +191,101 @@ term_flushln(struct termp *p)
 		 */
 
 		/* LINTED */
-		for (j = i, vsz = 0; j < (int)p->col; j++) {
-			if (j && ' ' == p->buf[j]) 
+		for (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++;
+			if (8 != p->buf[j]) {
+				if (vend > vis && vend < bp &&
+				    ASCII_HYPH == p->buf[j])
+					jhy = j;
+				vend++;
+			} else
+				vend--;
 		}
 
 		/*
-		 * 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) {
+		if (vend > bp && 0 == jhy && vis > 0) {
+			vend -= vis;
 			putchar('\n');
 			if (TERMP_NOBREAK & p->flags) {
+				p->viscol = p->rmargin;
 				for (j = 0; j < (int)p->rmargin; j++)
 					putchar(' ');
-				vis = p->rmargin - p->offset;
+				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 overstep width. */
+
+			/* Remove the p->overstep width. */
+
 			bp += (int)/* LINTED */
-				overstep;
-			overstep = 0;
-		} else {
-			for (j = 0; j < (int)vbl; j++)
-				putchar(' ');
-			vis += vbl;
+				p->overstep;
+			p->overstep = 0;
 		}
 
 		/*
-		 * Finally, write out the word.
+		 * Skip leading tabs, they were handled above.
 		 */
+		while (i < (int)p->col && '\t' == p->buf[i])
+			i++;
+
+		/* 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]) {
+				while (' ' == p->buf[i]) {
+					vbl++;
+					i++;
+				}
+				break;
+			}
+			if (ASCII_NBRSP == p->buf[i]) {
+				vbl++;
+				continue;
+			}
 
-			/* The unit sep. is a non-breaking space. */
-			if (31 == p->buf[i])
-				putchar(' ');
+			/*
+			 * Now we definitely know there will be
+			 * printable characters to output,
+			 * so write preceding white space now.
+			 */
+			if (vbl) {
+				for (j = 0; j < (int)vbl; j++)
+					putchar(' ');
+				p->viscol += vbl;
+				vbl = 0;
+			}
+
+			if (ASCII_HYPH == p->buf[i])
+				putchar('-');
 			else
 				putchar(p->buf[i]);
+
+			p->viscol += 1;
 		}
-		vis += vsz;
+		vend += vbl;
+		vis = vend;
 	}
 
 	p->col = 0;
-	overstep = 0;
+	p->overstep = 0;
 
 	if ( ! (TERMP_NOBREAK & p->flags)) {
+		p->viscol = 0;
 		putchar('\n');
 		return;
 	}
 
 	if (TERMP_HANG & p->flags) {
 		/* We need one blank after the tag. */
-		overstep = /* LINTED */
+		p->overstep = /* LINTED */
 			vis - maxvis + 1;
 
 		/*
@@ -262,23 +298,25 @@ term_flushln(struct termp *p)
 		 * move it one step LEFT and flag the rest of the line
 		 * to be longer.
 		 */
-		if (overstep >= -1) {
-			assert((int)maxvis + overstep >= 0);
+		if (p->overstep >= -1) {
+			assert((int)maxvis + p->overstep >= 0);
 			/* LINTED */
-			maxvis += overstep;
+			maxvis += p->overstep;
 		} else
-			overstep = 0;
+			p->overstep = 0;
 
 	} else if (TERMP_DANGLE & p->flags)
 		return;
 
 	/* Right-pad. */
 	if (maxvis > vis + /* LINTED */
-			((TERMP_TWOSPACE & p->flags) ? 1 : 0))  
+			((TERMP_TWOSPACE & p->flags) ? 1 : 0)) {
+		p->viscol += maxvis - vis;
 		for ( ; vis < maxvis; vis++)
 			putchar(' ');
-	else {	/* ...or newline break. */
+	} else {	/* ...or newline break. */
 		putchar('\n');
+		p->viscol = p->rmargin;
 		for (i = 0; i < (int)p->rmargin; i++)
 			putchar(' ');
 	}
@@ -295,7 +333,7 @@ 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;
 	}
@@ -315,160 +353,100 @@ term_vspace(struct termp *p)
 {
 
 	term_newln(p);
+	p->viscol = 0;
 	putchar('\n');
 }
 
 
 static void
-do_special(struct termp *p, const char *word, size_t len)
+spec(struct termp *p, const char *word, size_t len)
 {
 	const char	*rhs;
 	size_t		 sz;
-	int		 i;
 
 	rhs = chars_a2ascii(p->symtab, word, len, &sz);
-
-	if (NULL == rhs) {
-#if 0
-		fputs("Unknown special character: ", stderr);
-		for (i = 0; i < (int)len; i++)
-			fputc(word[i], stderr);
-		fputc('\n', stderr);
-#endif
-		return;
-	}
-	for (i = 0; i < (int)sz; i++) 
-		encode(p, rhs[i]);
+	if (rhs) 
+		encode(p, rhs, sz);
 }
 
 
 static void
-do_reserved(struct termp *p, const char *word, size_t len)
+res(struct termp *p, const char *word, size_t len)
 {
 	const char	*rhs;
 	size_t		 sz;
-	int		 i;
 
 	rhs = chars_a2res(p->symtab, word, len, &sz);
+	if (rhs)
+		encode(p, rhs, sz);
+}
 
-	if (NULL == rhs) {
-#if 0
-		fputs("Unknown reserved word: ", stderr);
-		for (i = 0; i < (int)len; i++)
-			fputc(word[i], stderr);
-		fputc('\n', stderr);
-#endif
-		return;
-	}
-	for (i = 0; i < (int)sz; i++) 
-		encode(p, rhs[i]);
+
+void
+term_fontlast(struct termp *p)
+{
+	enum termfont	 f;
+
+	f = p->fontl;
+	p->fontl = p->fontq[p->fonti];
+	p->fontq[p->fonti] = f;
 }
 
 
-/*
- * Handle an escape sequence: determine its length and pass it to the
- * escape-symbol look table.  Note that we assume mdoc(3) has validated
- * the escape sequence (we assert upon badly-formed escape sequences).
- */
-static void
-do_escaped(struct termp *p, const char **word)
+void
+term_fontrepl(struct termp *p, enum termfont f)
 {
-	int		 j, type;
-	const char	*wp;
 
-	wp = *word;
-	type = 1;
+	p->fontl = p->fontq[p->fonti];
+	p->fontq[p->fonti] = f;
+}
 
-	if (0 == *(++wp)) {
-		*word = wp;
-		return;
-	}
 
-	if ('(' == *wp) {
-		wp++;
-		if (0 == *wp || 0 == *(wp + 1)) {
-			*word = 0 == *wp ? wp : wp + 1;
-			return;
-		}
+void
+term_fontpush(struct termp *p, enum termfont f)
+{
 
-		do_special(p, wp, 2);
-		*word = ++wp;
-		return;
+	assert(p->fonti + 1 < 10);
+	p->fontl = p->fontq[p->fonti];
+	p->fontq[++p->fonti] = f;
+}
 
-	} else if ('*' == *wp) {
-		if (0 == *(++wp)) {
-			*word = wp;
-			return;
-		}
 
-		switch (*wp) {
-		case ('('):
-			wp++;
-			if (0 == *wp || 0 == *(wp + 1)) {
-				*word = 0 == *wp ? wp : wp + 1;
-				return;
-			}
+const void *
+term_fontq(struct termp *p)
+{
 
-			do_reserved(p, wp, 2);
-			*word = ++wp;
-			return;
-		case ('['):
-			type = 0;
-			break;
-		default:
-			do_reserved(p, wp, 1);
-			*word = wp;
-			return;
-		}
-	
-	} else if ('f' == *wp) {
-		if (0 == *(++wp)) {
-			*word = wp;
-			return;
-		}
+	return(&p->fontq[p->fonti]);
+}
 
-		switch (*wp) {
-		case ('B'):
-			p->bold++;
-			break;
-		case ('I'):
-			p->under++;
-			break;
-		case ('P'):
-			/* FALLTHROUGH */
-		case ('R'):
-			p->bold = p->under = 0;
-			break;
-		default:
-			break;
-		}
 
-		*word = wp;
-		return;
+enum termfont
+term_fonttop(struct termp *p)
+{
 
-	} else if ('[' != *wp) {
-		do_special(p, wp, 1);
-		*word = wp;
-		return;
-	}
+	return(p->fontq[p->fonti]);
+}
 
-	wp++;
-	for (j = 0; *wp && ']' != *wp; wp++, j++)
-		/* Loop... */ ;
 
-	if (0 == *wp) {
-		*word = wp;
-		return;
-	}
+void
+term_fontpopq(struct termp *p, const void *key)
+{
 
-	if (type)
-		do_special(p, wp - j, (size_t)j);
-	else
-		do_reserved(p, wp - j, (size_t)j);
-	*word = wp;
+	while (p->fonti >= 0 && key != &p->fontq[p->fonti])
+		p->fonti--;
+	assert(p->fonti >= 0);
 }
 
 
+void
+term_fontpop(struct termp *p)
+{
+
+	assert(p->fonti);
+	p->fonti--;
+}
+
+
 /*
  * 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
@@ -477,11 +455,14 @@ do_escaped(struct termp *p, const char **word)
 void
 term_word(struct termp *p, const char *word)
 {
-	const char	 *sv;
+	const char	*sv, *seq;
+	int		 sz;
+	size_t		 ssz;
+	enum roffdeco	 deco;
 
 	sv = word;
 
-	if (word[0] && 0 == word[1])
+	if (word[0] && '\0' == word[1])
 		switch (word[0]) {
 		case('.'):
 			/* FALLTHROUGH */
@@ -498,8 +479,6 @@ term_word(struct termp *p, const char *word)
 		case(')'):
 			/* FALLTHROUGH */
 		case(']'):
-			/* FALLTHROUGH */
-		case('}'):
 			if ( ! (TERMP_IGNDELIM & p->flags))
 				p->flags |= TERMP_NOSPACE;
 			break;
@@ -507,25 +486,66 @@ term_word(struct termp *p, const char *word)
 			break;
 		}
 
-	if ( ! (TERMP_NOSPACE & p->flags))
-		buffer(p, ' ');
+	if ( ! (TERMP_NOSPACE & p->flags)) {
+		bufferc(p, ' ');
+		if (TERMP_SENTENCE & p->flags)
+			bufferc(p, ' ');
+	}
 
 	if ( ! (p->flags & TERMP_NONOSPACE))
 		p->flags &= ~TERMP_NOSPACE;
 
-	for ( ; *word; word++)
-		if ('\\' != *word)
-			encode(p, *word);
-		else
-			do_escaped(p, &word);
+	p->flags &= ~TERMP_SENTENCE;
 
+	/* FIXME: use strcspn. */
+
+	while (*word) {
+		if ('\\' != *word) {
+			encode(p, word, 1);
+			word++;
+			continue;
+		}
+
+		seq = ++word;
+		sz = a2roffdeco(&deco, &seq, &ssz);
+
+		switch (deco) {
+		case (DECO_RESERVED):
+			res(p, seq, ssz);
+			break;
+		case (DECO_SPECIAL):
+			spec(p, seq, ssz);
+			break;
+		case (DECO_BOLD):
+			term_fontrepl(p, TERMFONT_BOLD);
+			break;
+		case (DECO_ITALIC):
+			term_fontrepl(p, TERMFONT_UNDER);
+			break;
+		case (DECO_ROMAN):
+			term_fontrepl(p, TERMFONT_NONE);
+			break;
+		case (DECO_PREVIOUS):
+			term_fontlast(p);
+			break;
+		default:
+			break;
+		}
+
+		word += sz;
+		if (DECO_NOSPACE == deco && '\0' == *word)
+			p->flags |= TERMP_NOSPACE;
+	}
+
+	/* 
+	 * Note that we don't process the pipe: the parser sees it as
+	 * punctuation, but we don't in terms of typography.
+	 */
 	if (sv[0] && 0 == sv[1])
 		switch (sv[0]) {
 		case('('):
 			/* FALLTHROUGH */
 		case('['):
-			/* FALLTHROUGH */
-		case('{'):
 			p->flags |= TERMP_NOSPACE;
 			break;
 		default:
@@ -534,46 +554,77 @@ term_word(struct termp *p, const char *word)
 }
 
 
-/*
- * 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
-buffer(struct termp *p, char c)
+adjbuf(struct termp *p, size_t sz)
 {
-	size_t		 s;
 
-	if (p->col + 1 >= p->maxcols) {
-		if (0 == p->maxcols)
-			p->maxcols = 256;
-		s = p->maxcols * 2;
-		p->buf = realloc(p->buf, s);
-		if (NULL == p->buf) {
-			perror(NULL);
-			exit(EXIT_FAILURE);
-		}
-		p->maxcols = s;
+	if (0 == p->maxcols)
+		p->maxcols = 1024;
+	while (sz >= p->maxcols)
+		p->maxcols <<= 2;
+
+	p->buf = realloc(p->buf, p->maxcols);
+	if (NULL == p->buf) {
+		perror(NULL);
+		exit(EXIT_FAILURE);
 	}
-	p->buf[(int)(p->col)++] = c;
 }
 
 
 static void
-encode(struct termp *p, char c)
+buffera(struct termp *p, const char *word, size_t sz)
 {
-	
-	if (' ' != c) {
-		if (p->under) {
-			buffer(p, '_');
-			buffer(p, 8);
+
+	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)
+{
+
+	if (p->col + 1 >= p->maxcols)
+		adjbuf(p, p->col + 1);
+
+	p->buf[(int)p->col++] = c;
+}
+
+
+static void
+encode(struct termp *p, const char *word, size_t sz)
+{
+	enum termfont	  f;
+	int		  i;
+
+	/*
+	 * Encode and buffer a string of characters.  If the current
+	 * font mode is unset, buffer directly, else encode then buffer
+	 * character by character.
+	 */
+
+	if (TERMFONT_NONE == (f = term_fonttop(p))) {
+		buffera(p, word, sz);
+		return;
+	}
+
+	for (i = 0; i < (int)sz; i++) {
+		if ( ! isgraph((u_char)word[i])) {
+			bufferc(p, word[i]);
+			continue;
 		}
-		if (p->bold) {
-			buffer(p, c);
-			buffer(p, 8);
-		}
+
+		if (TERMFONT_UNDER == f)
+			bufferc(p, '_');
+		else
+			bufferc(p, word[i]);
+
+		bufferc(p, 8);
+		bufferc(p, word[i]);
 	}
-	buffer(p, c);
 }