===================================================================
RCS file: /cvs/mandoc/term.c,v
retrieving revision 1.87
retrieving revision 1.129
diff -u -p -r1.87 -r1.129
--- mandoc/term.c	2009/07/15 15:53:57	1.87
+++ mandoc/term.c	2010/03/23 12:42:22	1.129
@@ -1,4 +1,4 @@
-/*	$Id: term.c,v 1.87 2009/07/15 15:53:57 kristaps Exp $ */
+/*	$Id: term.c,v 1.129 2010/03/23 12:42:22 kristaps Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -14,30 +14,34 @@
  * 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 <time.h>
 
+#include "chars.h"
+#include "out.h"
 #include "term.h"
 #include "man.h"
 #include "mdoc.h"
+#include "main.h"
 
-extern	int		  man_run(struct termp *, 
-				const struct man *);
-extern	int		  mdoc_run(struct termp *, 
-				const struct mdoc *);
-
 static	struct termp	 *term_alloc(enum termenc);
 static	void		  term_free(struct termp *);
-static	void		  term_pescape(struct termp *, const char **);
-static	void		  term_nescape(struct termp *,
-				const char *, size_t);
-static	void		  term_chara(struct termp *, char);
-static	void		  term_encodea(struct termp *, char);
-static	int		  term_isopendelim(const char *);
-static	int		  term_isclosedelim(const char *);
+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);
 
 
 void *
@@ -48,32 +52,6 @@ ascii_alloc(void)
 }
 
 
-int
-terminal_man(void *arg, const struct man *man)
-{
-	struct termp	*p;
-
-	p = (struct termp *)arg;
-	if (NULL == p->symtab)
-		p->symtab = term_ascii2htab();
-
-	return(man_run(p, man));
-}
-
-
-int
-terminal_mdoc(void *arg, const struct mdoc *mdoc)
-{
-	struct termp	*p;
-
-	p = (struct termp *)arg;
-	if (NULL == p->symtab)
-		p->symtab = term_ascii2htab();
-
-	return(mdoc_run(p, mdoc));
-}
-
-
 void
 terminal_free(void *arg)
 {
@@ -88,8 +66,8 @@ 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)
+		chars_free(p->symtab);
 
 	free(p);
 }
@@ -100,71 +78,16 @@ 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 = calloc(1, sizeof(struct termp));
+	if (NULL == p) {
+		perror(NULL);
+		exit(EXIT_FAILURE);
+	}
 	p->enc = enc;
 	return(p);
 }
 
 
-static int
-term_isclosedelim(const char *p)
-{
-
-	if ( ! (*p && 0 == *(p + 1)))
-		return(0);
-
-	switch (*p) {
-	case('.'):
-		/* FALLTHROUGH */
-	case(','):
-		/* FALLTHROUGH */
-	case(';'):
-		/* FALLTHROUGH */
-	case(':'):
-		/* FALLTHROUGH */
-	case('?'):
-		/* FALLTHROUGH */
-	case('!'):
-		/* FALLTHROUGH */
-	case(')'):
-		/* FALLTHROUGH */
-	case(']'):
-		/* FALLTHROUGH */
-	case('}'):
-		return(1);
-	default:
-		break;
-	}
-
-	return(0);
-}
-
-
-static int
-term_isopendelim(const char *p)
-{
-
-	if ( ! (*p && 0 == *(p + 1)))
-		return(0);
-
-	switch (*p) {
-	case('('):
-		/* FALLTHROUGH */
-	case('['):
-		/* FALLTHROUGH */
-	case('{'):
-		return(1);
-	default:
-		break;
-	}
-
-	return(0);
-}
-
-
 /*
  * 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
@@ -184,6 +107,9 @@ term_isopendelim(const char *p)
  *    columns.  In short: don't print a newline and instead pad to the
  *    right margin.  Used in conjunction with TERMP_NOLPAD.
  *
+ *  - TERMP_TWOSPACE: when padding, make sure there are at least two
+ *    space characters of padding.  Otherwise, rather break the line.
+ *
  *  - TERMP_DANGLE: don't newline when TERMP_NOBREAK is specified and
  *    the line is overrun, and don't pad-right if it's underrun.
  *
@@ -196,41 +122,45 @@ term_isopendelim(const char *p)
  *  If TERMP_NOBREAK is specified and the line overruns the right
  *  margin, it will break and pad-right to the right margin after
  *  writing.  If maxrmargin is violated, it will break and continue
- *  writing from the right-margin, which will lead to the above
- *  scenario upon exit.
- *
- *  Otherwise, the line will break at the right margin.  Extremely long
- *  lines will cause the system to emit a warning (TODO: hyphenate, if
- *  possible).
- *
- *  FIXME: newline breaks occur (in groff) also occur when a single
- *  space follows a NOBREAK!
+ *  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		 vbl, vsz, vis, maxvis, mmax, bp;
-	static int	 sv = -1;
+	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		 bp;    /* visual right border position */
+	int		 j;     /* temporary loop index */
+	size_t		 maxvis, mmax;
 
 	/*
 	 * 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.
+	 * small set of values. 
 	 */
 
 	assert(p->offset < p->rmargin);
-	maxvis = p->rmargin - p->offset;
-	mmax = p->maxrmargin - p->offset;
+
+	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;
-	vis = 0;
 
-	if (sv >= 0) {
-		vis = (size_t)sv;
-		sv = -1;
-	}
+	/* 
+	 * 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
@@ -252,10 +182,10 @@ term_flushln(struct termp *p)
 
 		/* LINTED */
 		for (j = i, vsz = 0; j < (int)p->col; j++) {
-			if (' ' == p->buf[j])
+			if (j && ' ' == p->buf[j]) 
 				break;
 			else if (8 == p->buf[j])
-				j += 1;
+				vsz--;
 			else
 				vsz++;
 		}
@@ -283,6 +213,10 @@ term_flushln(struct termp *p)
 					putchar(' ');
 				vis = 0;
 			}
+			/* Remove the p->overstep width. */
+			bp += (int)/* LINTED */
+				p->overstep;
+			p->overstep = 0;
 		} else {
 			for (j = 0; j < (int)vbl; j++)
 				putchar(' ');
@@ -295,42 +229,59 @@ term_flushln(struct termp *p)
 		for ( ; i < (int)p->col; i++) {
 			if (' ' == p->buf[i])
 				break;
-			putchar(p->buf[i]);
+
+			/* The unit sep. is a non-breaking space. */
+			if (31 == p->buf[i])
+				putchar(' ');
+			else
+				putchar(p->buf[i]);
 		}
 		vis += vsz;
 	}
 
-	/*
-	 * If we've overstepped our maximum visible no-break space, then
-	 * cause a newline and offset at the right margin.
-	 */
+	p->col = 0;
+	p->overstep = 0;
 
-	if ((TERMP_NOBREAK & p->flags) && vis >= maxvis) {
-		if ( ! (TERMP_DANGLE & p->flags) &&
-				! (TERMP_HANG & p->flags)) {
-			putchar('\n');
-			for (i = 0; i < (int)p->rmargin; i++)
-				putchar(' ');
-		}
-		if (TERMP_HANG & p->flags)
-			sv = (int)(vis - maxvis);
-		p->col = 0;
+	if ( ! (TERMP_NOBREAK & p->flags)) {
+		putchar('\n');
 		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) {
+		/* We need one blank after the tag. */
+		p->overstep = /* LINTED */
+			vis - maxvis + 1;
 
-	if (p->flags & TERMP_NOBREAK) {
-		if ( ! (TERMP_DANGLE & p->flags))
-			for ( ; vis < maxvis; vis++)
-				putchar(' ');
-	} else
-		putchar('\n');
+		/*
+		 * Behave exactly the same way as groff:
+		 * If we have overstepped the margin, temporarily move
+		 * it to the right and flag the rest of the line to be
+		 * shorter.
+		 * If we landed right at the margin, be happy.
+		 * If we are one step before the margin, temporarily
+		 * move it one step LEFT and flag the rest of the line
+		 * to be longer.
+		 */
+		if (p->overstep >= -1) {
+			assert((int)maxvis + p->overstep >= 0);
+			/* LINTED */
+			maxvis += p->overstep;
+		} else
+			p->overstep = 0;
 
-	p->col = 0;
+	} else if (TERMP_DANGLE & p->flags)
+		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(' ');
+	}
 }
 
 
@@ -368,196 +319,343 @@ term_vspace(struct termp *p)
 }
 
 
-/*
- * 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)
+spec(struct termp *p, const char *word, size_t len)
 {
 	const char	*rhs;
 	size_t		 sz;
-	int		 i;
 
-	rhs = term_a2ascii(p->symtab, word, len, &sz);
+	rhs = chars_a2ascii(p->symtab, word, len, &sz);
+	if (rhs) 
+		encode(p, rhs, sz);
+}
 
+
+static void
+res(struct termp *p, const char *word, size_t len)
+{
+	const char	*rhs;
+	size_t		 sz;
+
+	rhs = chars_a2res(p->symtab, word, len, &sz);
 	if (rhs)
-		for (i = 0; i < (int)sz; i++) 
-			term_encodea(p, rhs[i]);
+		encode(p, rhs, sz);
 }
 
 
-/*
- * Handle an escape sequence: determine its length and pass it to the
- * escape-symbol look table.  Note that we assume mdoc(3) has validated
- * the escape sequence (we assert upon badly-formed escape sequences).
- */
-static void
-term_pescape(struct termp *p, const char **word)
+void
+term_fontlast(struct termp *p)
 {
-	int		 j;
-	const char	*wp;
+	enum termfont	 f;
 
-	wp = *word;
+	f = p->fontl;
+	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_fontrepl(struct termp *p, enum termfont f)
+{
 
-		term_nescape(p, wp, 2);
-		*word = ++wp;
-		return;
+	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;
-			}
+void
+term_fontpush(struct termp *p, enum termfont f)
+{
 
-			term_nescape(p, wp, 2);
-			*word = ++wp;
-			return;
-		case ('['):
+	assert(p->fonti + 1 < 10);
+	p->fontl = p->fontq[p->fonti];
+	p->fontq[++p->fonti] = f;
+}
+
+
+const void *
+term_fontq(struct termp *p)
+{
+
+	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 != &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
+ * handles word styling.
+ */
+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:
-			term_nescape(p, wp, 1);
-			*word = wp;
-			return;
+			break;
 		}
-	
-	} else if ('f' == *wp) {
-		if (0 == *(++wp)) {
-			*word = wp;
-			return;
+
+	if ( ! (TERMP_NOSPACE & p->flags))
+		bufferc(p, ' ');
+
+	if ( ! (p->flags & TERMP_NONOSPACE))
+		p->flags &= ~TERMP_NOSPACE;
+
+	/* FIXME: use strcspn. */
+
+	while (*word) {
+		if ('\\' != *word) {
+			encode(p, word, 1);
+			word++;
+			continue;
 		}
 
-		switch (*wp) {
-		case ('B'):
-			p->flags |= TERMP_BOLD;
+		seq = ++word;
+		sz = a2roffdeco(&deco, &seq, &ssz);
+
+		switch (deco) {
+		case (DECO_RESERVED):
+			res(p, seq, ssz);
 			break;
-		case ('I'):
-			p->flags |= TERMP_UNDER;
+		case (DECO_SPECIAL):
+			spec(p, seq, ssz);
 			break;
-		case ('P'):
+		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;
+	}
+
+	if (sv[0] && 0 == sv[1])
+		switch (sv[0]) {
+		case('('):
 			/* FALLTHROUGH */
-		case ('R'):
-			p->flags &= ~TERMP_STYLE;
+		case('['):
+			/* FALLTHROUGH */
+		case('{'):
+			p->flags |= TERMP_NOSPACE;
 			break;
 		default:
 			break;
 		}
+}
 
-		*word = wp;
-		return;
 
-	} else if ('[' != *wp) {
-		term_nescape(p, wp, 1);
-		*word = wp;
-		return;
-	}
+static void
+adjbuf(struct termp *p, size_t sz)
+{
 
-	wp++;
-	for (j = 0; *wp && ']' != *wp; wp++, j++)
-		/* Loop... */ ;
+	if (0 == p->maxcols)
+		p->maxcols = 1024;
+	while (sz >= p->maxcols)
+		p->maxcols <<= 2;
 
-	if (0 == *wp) {
-		*word = wp;
-		return;
+	p->buf = realloc(p->buf, p->maxcols);
+	if (NULL == p->buf) {
+		perror(NULL);
+		exit(EXIT_FAILURE);
 	}
+}
 
-	term_nescape(p, wp - j, (size_t)j);
-	*word = wp;
+
+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;
 }
 
 
-/*
- * Handle pwords, partial words, which may be either a single word or a
- * phrase that cannot be broken down (such as a literal string).  This
- * handles word styling.
- */
-void
-term_word(struct termp *p, const char *word)
+static void
+bufferc(struct termp *p, char c)
 {
 
-	if (term_isclosedelim(word))
-		if ( ! (TERMP_IGNDELIM & p->flags))
-			p->flags |= TERMP_NOSPACE;
+	if (p->col + 1 >= p->maxcols)
+		adjbuf(p, p->col + 1);
 
-	if ( ! (TERMP_NOSPACE & p->flags))
-		term_chara(p, ' ');
+	p->buf[(int)p->col++] = c;
+}
 
-	if ( ! (p->flags & TERMP_NONOSPACE))
-		p->flags &= ~TERMP_NOSPACE;
 
-	/* 
-	 * If ANSI (word-length styling), then apply our style now,
-	 * before the word.
+static void
+encode(struct termp *p, const char *word, size_t sz)
+{
+	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.
 	 */
 
-	for ( ; *word; word++)
-		if ('\\' != *word)
-			term_encodea(p, *word);
+	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 (TERMFONT_UNDER == f)
+			bufferc(p, '_');
 		else
-			term_pescape(p, &word);
+			bufferc(p, word[i]);
 
-	if (term_isopendelim(word))
-		p->flags |= TERMP_NOSPACE;
+		bufferc(p, 8);
+		bufferc(p, word[i]);
+	}
 }
 
 
-/*
- * 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)
+size_t
+term_vspan(const struct roffsu *su)
 {
-	size_t		 s;
+	double		 r;
 
-	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;
+	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);
 }
 
 
-static void
-term_encodea(struct termp *p, char c)
+size_t
+term_hspan(const struct roffsu *su)
 {
+	double		 r;
 
-	if (TERMP_STYLE & p->flags) {
-		if (TERMP_BOLD & p->flags) {
-			term_chara(p, c);
-			term_chara(p, 8);
-		}
-		if (TERMP_UNDER & p->flags) {
-			term_chara(p, '_');
-			term_chara(p, 8);
-		}
+	/* 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;
 	}
-	term_chara(p, c);
+
+	if (r < 0.0)
+		r = 0.0;
+	return((size_t)/* LINTED */
+			r);
 }
+
+