===================================================================
RCS file: /cvs/mandoc/term.c,v
retrieving revision 1.1
retrieving revision 1.8
diff -u -p -r1.1 -r1.8
--- mandoc/term.c	2009/02/20 11:04:23	1.1
+++ mandoc/term.c	2009/02/22 15:50:45	1.8
@@ -1,6 +1,6 @@
-/* $Id: term.c,v 1.1 2009/02/20 11:04:23 kristaps Exp $ */
+/* $Id: term.c,v 1.8 2009/02/22 15:50:45 kristaps Exp $ */
 /*
- * Copyright (c) 2008 Kristaps Dzonsons <kristaps@kth.se>
+ * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
  * Permission to use, copy, modify, and distribute this software for any
  * purpose with or without fee is hereby granted, provided that the
@@ -17,273 +17,406 @@
  * PERFORMANCE OF THIS SOFTWARE.
  */
 #include <assert.h>
-#include <curses.h>
+#include <ctype.h>
 #include <err.h>
 #include <stdlib.h>
 #include <stdio.h>
 #include <string.h>
-#include <term.h>
 #include <unistd.h>
 
-#include "mdoc.h"
+#ifdef __linux__
+#include <time.h>
+#endif
 
+#include "term.h"
 
-static	int		 termprint_r(size_t, size_t, 
+enum	termstyle {
+	STYLE_CLEAR,
+	STYLE_BOLD,
+	STYLE_UNDERLINE
+};
+
+static	void		  termprint_r(struct termp *,
+				const struct mdoc_meta *,
 				const struct mdoc_node *);
-static	void		 termprint_head(size_t, 
+static	void		  termprint_header(struct termp *,
 				const struct mdoc_meta *);
-static	void		 termprint_tail(size_t, 
+static	void		  termprint_footer(struct termp *,
 				const struct mdoc_meta *);
 
-static	char 		*arch2a(enum mdoc_arch);
-static	char 		*vol2a(enum mdoc_vol);
-static	char 		*msec2a(enum mdoc_msec);
+static	void		  pword(struct termp *, const char *, size_t);
+static	void		  pescape(struct termp *, 
+				const char *, size_t *, size_t);
+static	void		  chara(struct termp *, char);
+static	void		  style(struct termp *, enum termstyle);
 
-static	size_t 		 ttitle2a(char *, enum mdoc_vol, enum mdoc_msec,
-				enum mdoc_arch, size_t);
+#ifdef __linux__
+extern	size_t		  strlcat(char *, const char *, size_t);
+extern	size_t		  strlcpy(char *, const char *, size_t);
+#endif
 
-
-static char *
-arch2a(enum mdoc_arch arch)
+void
+flushln(struct termp *p)
 {
+	size_t		 i, j, vsz, vis, maxvis;
 
-	switch (arch) {
-	case (ARCH_alpha):
-		return("Alpha");
-	case (ARCH_amd64):
-		return("AMD64");
-	case (ARCH_amiga):
-		return("Amiga");
-	case (ARCH_arc):
-		return("ARC");
-	case (ARCH_arm):
-		return("ARM");
-	case (ARCH_armish):
-		return("ARMISH");
-	case (ARCH_aviion):
-		return("AViion");
-	case (ARCH_hp300):
-		return("HP300");
-	case (ARCH_hppa):
-		return("HPPA");
-	case (ARCH_hppa64):
-		return("HPPA64");
-	case (ARCH_i386):
-		return("i386");
-	case (ARCH_landisk):
-		return("LANDISK");
-	case (ARCH_luna88k):
-		return("Luna88k");
-	case (ARCH_mac68k):
-		return("Mac68k");
-	case (ARCH_macppc):
-		return("MacPPC");
-	case (ARCH_mvme68k):
-		return("MVME68k");
-	case (ARCH_mvme88k):
-		return("MVME88k");
-	case (ARCH_mvmeppc):
-		return("MVMEPPC");
-	case (ARCH_pmax):
-		return("PMAX");
-	case (ARCH_sgi):
-		return("SGI");
-	case (ARCH_socppc):
-		return("SOCPPC");
-	case (ARCH_sparc):
-		return("SPARC");
-	case (ARCH_sparc64):
-		return("SPARC64");
-	case (ARCH_sun3):
-		return("Sun3");
-	case (ARCH_vax):
-		return("VAX");
-	case (ARCH_zaurus):
-		return("Zaurus");
-	default:	 	
-		break;
+	/*
+	 * 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.
+	 */
+
+	assert(p->offset < p->rmargin);
+	maxvis = p->rmargin - p->offset;
+	vis = 0;
+
+	/*
+	 * If in the standard case (left-justified), then begin with our
+	 * indentation, otherwise (columns, etc.) just start spitting
+	 * out text.
+	 */
+
+	if ( ! (p->flags & TERMP_NOLPAD))
+		/* LINTED */
+		for (j = 0; j < p->offset; j++)
+			putchar(' ');
+
+	/*
+	 * If we're literal, print out verbatim.
+	 */
+	if (p->flags & TERMP_LITERAL) {
+		/* FIXME: count non-printing chars. */
+		for (i = 0; i < p->col; i++)
+			putchar(p->buf[i]);
+		putchar('\n');
+		p->col = 0;
+		return;
 	}
 
-	return(NULL);
+	for (i = 0; i < p->col; i++) {
+		/*
+		 * Count up visible word characters.  Control sequences
+		 * (starting with the CSI) aren't counted. 
+		 */
+		assert( ! isspace(p->buf[i]));
+
+		/* LINTED */
+		for (j = i, vsz = 0; j < p->col; j++) {
+			if (isspace(p->buf[j]))
+				break;
+			else if (27 == p->buf[j]) {
+				assert(j + 4 <= p->col);
+				j += 3;
+			} else
+				vsz++;
+		}
+		assert(vsz > 0);
+
+		/*
+		 * If a word is too long and we're within a line, put it
+		 * on the next line.  Puke if we're being asked to write
+		 * something that will exceed the right margin (i.e.,
+		 * from a fresh line or when we're not allowed to break
+		 * the line with TERMP_NOBREAK).
+		 */
+
+		if (vis && vis + vsz >= maxvis) {
+			/* FIXME */
+			if (p->flags & TERMP_NOBREAK)
+				errx(1, "word breaks right margin");
+			putchar('\n');
+			for (j = 0; j < p->offset; j++)
+				putchar(' ');
+			vis = 0;
+		} else if (vis + vsz >= maxvis) {
+			/* FIXME */
+			errx(1, "word breaks right margin");
+		}
+
+		/* 
+		 * Write out the word and a trailing space.  Omit the
+		 * space if we're the last word in the line.
+		 */
+
+		for ( ; i < p->col; i++) {
+			if (isspace(p->buf[i]))
+				break;
+			putchar(p->buf[i]);
+		}
+		vis += vsz;
+		if (i < p->col) {
+			putchar(' ');
+			vis++;
+		}
+	}
+
+	/*
+	 * If we're not to right-marginalise it (newline), then instead
+	 * pad to the right margin and stay off.
+	 */
+
+	if (p->flags & TERMP_NOBREAK) {
+		for ( ; vis <= maxvis; vis++)
+			putchar(' ');
+	} else
+		putchar('\n');
+
+	p->col = 0;
 }
 
 
-static char *
-vol2a(enum mdoc_vol vol)
+void
+newln(struct termp *p)
 {
 
-	switch (vol) {
-	case (VOL_AMD):
-		return("OpenBSD Ancestral Manual Documents");
-	case (VOL_IND):
-		return("OpenBSD Manual Master Index");
-	case (VOL_KM):
-		return("OpenBSD Kernel Manual");
-	case (VOL_LOCAL):
-		return("OpenBSD Local Manual");
-	case (VOL_PRM):
-		return("OpenBSD Programmer's Manual");
-	case (VOL_PS1):
-		return("OpenBSD Programmer's Supplementary Documents");
-	case (VOL_SMM):
-		return("OpenBSD System Manager's Manual");
-	case (VOL_URM):
-		return("OpenBSD Reference Manual");
-	case (VOL_USD):
-		return("OpenBSD User's Supplementary Documents");
-	default:
-		break;
-	}
+	/* 
+	 * A newline only breaks an existing line; it won't assert
+	 * vertical space.
+	 */
+	p->flags |= TERMP_NOSPACE;
+	if (0 == p->col) 
+		return;
+	flushln(p);
+}
 
-	return(NULL);
+
+void
+vspace(struct termp *p)
+{
+
+	/*
+	 * Asserts a vertical space (a full, empty line-break between
+	 * lines).
+	 */
+	newln(p);
+	putchar('\n');
 }
 
 
-static char *
-msec2a(enum mdoc_msec msec)
+static void
+chara(struct termp *p, char c)
 {
 
-	switch (msec) {
-	case(MSEC_1):
-		return("1");
-	case(MSEC_2):
-		return("2");
-	case(MSEC_3):
-		return("3");
-	case(MSEC_3f):
-		return("3f");
-	case(MSEC_3p):
-		return("3p");
-	case(MSEC_4):
-		return("4");
-	case(MSEC_5):
-		return("5");
-	case(MSEC_6):
-		return("6");
-	case(MSEC_7):
-		return("7");
-	case(MSEC_8):
-		return("8");
-	case(MSEC_9):
-		return("9");
-	case(MSEC_X11):
-		return("X11");
-	case(MSEC_X11R6):
-		return("X11R6");
-	case(MSEC_local):
-		return("local");
-	case(MSEC_n):
-		return("n");
-	case(MSEC_unass):
-		/* FALLTHROUGH */
-	case(MSEC_draft):
-		return("draft");
-	case(MSEC_paper):
-		return("paper");
-	default:
+	/* TODO: dynamically expand the buffer. */
+	if (p->col + 1 >= p->maxcols)
+		errx(1, "line overrun");
+	p->buf[(p->col)++] = c;
+}
+
+
+static void
+style(struct termp *p, enum termstyle esc)
+{
+
+	if (p->col + 4 >= p->maxcols)
+		errx(1, "line overrun");
+
+	p->buf[(p->col)++] = 27;
+	p->buf[(p->col)++] = '[';
+	switch (esc) {
+	case (STYLE_CLEAR):
+		p->buf[(p->col)++] = '0';
 		break;
+	case (STYLE_BOLD):
+		p->buf[(p->col)++] = '1';
+		break;
+	case (STYLE_UNDERLINE):
+		p->buf[(p->col)++] = '4';
+		break;
+	default:
+		abort();
+		/* NOTREACHED */
 	}
-	return(NULL);
+	p->buf[(p->col)++] = 'm';
 }
 
 
-static size_t
-ttitle2a(char *dst, enum mdoc_vol vol, enum mdoc_msec msec,
-		enum mdoc_arch arch, size_t sz)
+static void
+pescape(struct termp *p, const char *word, size_t *i, size_t len)
 {
-	char		*p;
-	size_t		 ssz;
 
-	if (NULL == (p = vol2a(vol)))
-		switch (msec) {
-		case (MSEC_1):
+	(*i)++;
+	assert(*i < len);
+
+	if ('(' == word[*i]) {
+		/* Two-character escapes. */
+		(*i)++;
+		assert(*i + 1 < len);
+
+		if ('r' == word[*i] && 'B' == word[*i + 1])
+			chara(p, ']');
+		else if ('l' == word[*i] && 'B' == word[*i + 1])
+			chara(p, '[');
+
+		(*i)++;
+		return;
+
+	} else if ('[' != word[*i]) {
+		/* One-character escapes. */
+		switch (word[*i]) {
+		case ('\\'):
 			/* FALLTHROUGH */
-		case (MSEC_6):
+		case ('\''):
 			/* FALLTHROUGH */
-		case (MSEC_7):
-			p = vol2a(VOL_URM);
-			break;
-		case (MSEC_8):
-			p = vol2a(VOL_SMM);
-			break;
-		case (MSEC_2):
+		case ('`'):
 			/* FALLTHROUGH */
-		case (MSEC_3):
+		case ('-'):
 			/* FALLTHROUGH */
-		case (MSEC_4):
-			/* FALLTHROUGH */
-		case (MSEC_5):
-			p = vol2a(VOL_PRM);
-			break;
-		case (MSEC_9):
-			p = vol2a(VOL_KM);
-			break;
+		case ('.'):
+			chara(p, word[*i]);
 		default:
-			/* FIXME: capitalise. */
-			if (NULL == (p = msec2a(msec)))
-				p = msec2a(MSEC_local);
 			break;
 		}
-	assert(p);
+		return;
+	}
+	/* n-character escapes. */
+}
 
-	if ((ssz = strlcpy(dst, p, sz)) >= sz)
-		return(ssz);
 
-	if ((p = arch2a(arch))) {
-		if ((ssz = strlcat(dst, " (", sz)) >= sz)
-			return(ssz);
-		if ((ssz = strlcat(dst, p, sz)) >= sz)
-			return(ssz);
-		if ((ssz = strlcat(dst, ")", sz)) >= sz)
-			return(ssz);
+static void
+pword(struct termp *p, const char *word, size_t len)
+{
+	size_t		 i;
+
+	/*assert(len > 0);*/ /* Can be, if literal. */
+
+	if ( ! (p->flags & TERMP_NOSPACE) && 
+			! (p->flags & TERMP_LITERAL))
+		chara(p, ' ');
+
+	p->flags &= ~TERMP_NOSPACE;
+
+	if (p->flags & TERMP_BOLD)
+		style(p, STYLE_BOLD);
+	if (p->flags & TERMP_UNDERLINE)
+		style(p, STYLE_UNDERLINE);
+
+	for (i = 0; i < len; i++) {
+		if ('\\' == word[i]) {
+			pescape(p, word, &i, len);
+			continue;
+		}
+		chara(p, word[i]);
 	}
 
-	return(ssz);
+	if (p->flags & TERMP_BOLD ||
+			p->flags & TERMP_UNDERLINE)
+		style(p, STYLE_CLEAR);
 }
 
 
-static int
-termprint_r(size_t cols, size_t indent, const struct mdoc_node *node)
+void
+word(struct termp *p, const char *word)
 {
+	size_t 		 i, j, len;
 
-	return(1);
+	if (p->flags & TERMP_LITERAL) {
+		pword(p, word, strlen(word));
+		return;
+	}
+
+	len = strlen(word);
+	assert(len > 0);
+
+	if (mdoc_isdelim(word))
+		p->flags |= TERMP_NOSPACE;
+
+	/* LINTED */
+	for (j = i = 0; i < len; i++) {
+		if ( ! isspace(word[i])) {
+			j++;
+			continue;
+		}
+		if (0 == j)
+			continue;
+		assert(i >= j);
+		pword(p, &word[i - j], j);
+		j = 0;
+	}
+	if (j > 0) {
+		assert(i >= j);
+		pword(p, &word[i - j], j);
+	}
 }
 
 
 static void
-termprint_tail(size_t cols, const struct mdoc_meta *meta)
+termprint_r(struct termp *p, const struct mdoc_meta *meta,
+		const struct mdoc_node *node)
 {
+	int		 dochild;
+
+	/* Pre-processing. */
+
+	dochild = 1;
+
+	if (MDOC_TEXT != node->type) {
+		if (termacts[node->tok].pre)
+			if ( ! (*termacts[node->tok].pre)(p, meta, node))
+				dochild = 0;
+	} else /* MDOC_TEXT == node->type */
+		word(p, node->data.text.string);
+
+	/* Children. */
+
+	if (dochild && node->child)
+		termprint_r(p, meta, node->child);
+
+	/* Post-processing. */
+
+	if (MDOC_TEXT != node->type)
+		if (termacts[node->tok].post)
+			(*termacts[node->tok].post)(p, meta, node);
+
+	/* Siblings. */
+
+	if (node->next)
+		termprint_r(p, meta, node->next);
+}
+
+
+static void
+termprint_footer(struct termp *p, const struct mdoc_meta *meta)
+{
 	struct tm	*tm;
 	char		*buf, *os;
 	size_t		 sz, osz, ssz, i;
 
-	if (NULL == (buf = malloc(cols)))
+	if (NULL == (buf = malloc(p->rmargin)))
 		err(1, "malloc");
-	if (NULL == (os = malloc(cols)))
+	if (NULL == (os = malloc(p->rmargin)))
 		err(1, "malloc");
 
 	tm = localtime(&meta->date);
-	if (NULL == strftime(buf, cols, "%B %d, %Y", tm))
+
+#ifdef __linux__
+	if (0 == strftime(buf, p->rmargin, "%B %d, %Y", tm))
+#else
+	if (NULL == strftime(buf, p->rmargin, "%B %d, %Y", tm))
+#endif
 		err(1, "strftime");
 
-	osz = strlcpy(os, meta->os, cols);
+	osz = strlcpy(os, meta->os, p->rmargin);
 
 	sz = strlen(buf);
 	ssz = sz + osz + 1;
 
-	if (ssz > cols) {
-		ssz -= cols;
+	if (ssz > p->rmargin) {
+		ssz -= p->rmargin;
 		assert(ssz <= osz);
 		os[osz - ssz] = 0;
 		ssz = 1;
 	} else
-		ssz = cols - ssz + 1;
+		ssz = p->rmargin - ssz + 1;
 
+	printf("\n");
 	printf("%s", os);
 	for (i = 0; i < ssz; i++)
 		printf(" ");
 
 	printf("%s\n", buf);
+	fflush(stdout);
 
 	free(buf);
 	free(os);
@@ -291,28 +424,70 @@ termprint_tail(size_t cols, const struct mdoc_meta *me
 
 
 static void
-termprint_head(size_t cols, const struct mdoc_meta *meta)
+termprint_header(struct termp *p, const struct mdoc_meta *meta)
 {
-	char		*msec, *buf, *title;
-	size_t		 ssz, tsz, ttsz, i;
+	char		*buf, *title;
+	const char	*pp, *msec;
+	size_t		 ssz, tsz, ttsz, i;;
 
-	if (NULL == (buf = malloc(cols)))
+	if (NULL == (buf = malloc(p->rmargin)))
 		err(1, "malloc");
-	if (NULL == (title = malloc(cols)))
+	if (NULL == (title = malloc(p->rmargin)))
 		err(1, "malloc");
 
-	/* Format the manual page header. */
+	if (NULL == (pp = mdoc_vol2a(meta->vol)))
+		switch (meta->msec) {
+		case (MSEC_1):
+			/* FALLTHROUGH */
+		case (MSEC_6):
+			/* FALLTHROUGH */
+		case (MSEC_7):
+			pp = mdoc_vol2a(VOL_URM);
+			break;
+		case (MSEC_8):
+			pp = mdoc_vol2a(VOL_SMM);
+			break;
+		case (MSEC_2):
+			/* FALLTHROUGH */
+		case (MSEC_3):
+			/* FALLTHROUGH */
+		case (MSEC_4):
+			/* FALLTHROUGH */
+		case (MSEC_5):
+			pp = mdoc_vol2a(VOL_PRM);
+			break;
+		case (MSEC_9):
+			pp = mdoc_vol2a(VOL_KM);
+			break;
+		default:
+			/* FIXME: capitalise. */
+			if (NULL == (pp = mdoc_msec2a(meta->msec)))
+				pp = mdoc_msec2a(MSEC_local);
+			break;
+		}
+	assert(pp);
 
-	tsz = ttitle2a(buf, meta->vol, meta->msec, meta->arch, cols);
-	ttsz = strlcpy(title, meta->title, cols);
+	tsz = strlcpy(buf, pp, p->rmargin);
+	assert(tsz < p->rmargin);
 
-	if (NULL == (msec = msec2a(meta->msec)))
+	if ((pp = mdoc_arch2a(meta->arch))) {
+		tsz = strlcat(buf, " (", p->rmargin);
+		assert(tsz < p->rmargin);
+		tsz = strlcat(buf, pp, p->rmargin);
+		assert(tsz < p->rmargin);
+		tsz = strlcat(buf, ")", p->rmargin);
+		assert(tsz < p->rmargin);
+	}
+
+	ttsz = strlcpy(title, meta->title, p->rmargin);
+
+	if (NULL == (msec = mdoc_msec2a(meta->msec)))
 		msec = "";
 
 	ssz = (2 * (ttsz + 2 + strlen(msec))) + tsz + 2;
 
-	if (ssz > cols) {
-		if ((ssz -= cols) % 2)
+	if (ssz > p->rmargin) {
+		if ((ssz -= p->rmargin) % 2)
 			ssz++;
 		ssz /= 2;
 	
@@ -320,7 +495,7 @@ termprint_head(size_t cols, const struct mdoc_meta *me
 		title[ttsz - ssz] = 0;
 		ssz = 1;
 	} else
-		ssz = ((cols - ssz) / 2) + 1;
+		ssz = ((p->rmargin - ssz) / 2) + 1;
 
 	printf("%s(%s)", title, msec);
 
@@ -332,29 +507,34 @@ termprint_head(size_t cols, const struct mdoc_meta *me
 	for (i = 0; i < ssz; i++)
 		printf(" ");
 
-	printf("%s(%s)\n\n", title, msec);
+	printf("%s(%s)\n", title, msec);
+	fflush(stdout);
 
 	free(title);
 	free(buf);
 }
 
 
-int
+void
 termprint(const struct mdoc_node *node,
 		const struct mdoc_meta *meta)
 {
-	size_t		 cols;
+	struct termp	 p;
 
-	if (ERR == setupterm(NULL, STDOUT_FILENO, NULL))
-		return(0);
+	p.maxrmargin = 80; /* XXX */
+	p.rmargin = p.maxrmargin;
+	p.maxcols = 1024;
+	p.offset = p.col = 0;
+	p.flags = TERMP_NOSPACE;
 
-	cols = columns < 60 ? 60 : (size_t)columns;
+	if (NULL == (p.buf = malloc(p.maxcols)))
+		err(1, "malloc");
 
-	termprint_head(cols, meta);
-	if ( ! termprint_r(cols, 0, node))
-		return(0);
-	termprint_tail(cols, meta);
-	return(1);
+	termprint_header(&p, meta);
+	termprint_r(&p, meta, node);
+	termprint_footer(&p, meta);
+
+	free(p.buf);
 }