===================================================================
RCS file: /cvs/mandoc/Attic/mdocterm.c,v
retrieving revision 1.1
retrieving revision 1.12
diff -u -p -r1.1 -r1.12
--- mandoc/Attic/mdocterm.c	2009/02/21 21:00:06	1.1
+++ mandoc/Attic/mdocterm.c	2009/02/25 15:12:26	1.12
@@ -1,4 +1,4 @@
-	/* $Id: mdocterm.c,v 1.1 2009/02/21 21:00:06 kristaps Exp $ */
+/* $Id: mdocterm.c,v 1.12 2009/02/25 15:12:26 kristaps Exp $ */
 /*
  * Copyright (c) 2008 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -16,274 +16,603 @@
  * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
  * PERFORMANCE OF THIS SOFTWARE.
  */
-#include <sys/stat.h>
-#include <sys/param.h>
-
 #include <assert.h>
-#include <fcntl.h>
+#include <ctype.h>
 #include <err.h>
 #include <getopt.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
-#include <unistd.h>
 
-#include "mdoc.h"
+#ifndef __OpenBSD__
+#include <time.h>
+#endif
+
+#include "mmain.h"
 #include "term.h"
 
-#define	MD_LINE_SZ	(256)		/* Max input line size. */
+#ifdef __NetBSD__
+#define xisspace(x) isspace((int)(x))
+#else
+#define	xisspace(x) isspace((x))
+#endif
 
-struct	md_parse {
-	int		  warn;		/* Warning flags. */
-#define	MD_WARN_SYNTAX	 (1 << 0)	/* Show syntax warnings. */
-#define	MD_WARN_COMPAT	 (1 << 1)	/* Show compat warnings. */
-#define	MD_WARN_ALL	 (0x03)		/* Show all warnings. */
-#define	MD_WARN_ERR	 (1 << 2)	/* Make warnings->errors. */
-	int		  dbg;		/* Debug level. */
-	struct mdoc	 *mdoc;		/* Active parser. */
-	char		 *buf;		/* Input buffer. */
-	u_long		  bufsz;	/* Input buffer size. */
-	char		 *in;		/* Input file name. */
-	int		  fdin;		/* Input file desc. */
+enum	termstyle {
+	STYLE_CLEAR,
+	STYLE_BOLD,
+	STYLE_UNDERLINE
 };
 
-extern	char	 	 *__progname;
+static	void		  body(struct termp *,
+				struct termpair *,
+				const struct mdoc_meta *,
+				const struct mdoc_node *);
+static	void		  header(struct termp *,
+				const struct mdoc_meta *);
+static	void		  footer(struct termp *,
+				const struct mdoc_meta *);
 
-static	void		  usage(void);
-static	int		  getsopts(struct md_parse *, char *);
-static	int		  parse(struct md_parse *);
-static	void		  msg_msg(void *, int, int, const char *);
-static	int		  msg_err(void *, int, int, const char *);
-static	int		  msg_warn(void *, int, int, 
-				enum mdoc_warn, const char *);
+static	void		  pword(struct termp *, const char *, size_t);
+static	void		  pescape(struct termp *, 
+				const char *, size_t *, size_t);
+static	void		  nescape(struct termp *, 
+				const char *, size_t);
+static	void		  chara(struct termp *, char);
+static	void		  stringa(struct termp *, const char *);
+static	void		  style(struct termp *, enum termstyle);
 
 #ifdef __linux__
-extern	int		  getsubopt(char **, char *const *, char **);
+extern	size_t		  strlcat(char *, const char *, size_t);
+extern	size_t		  strlcpy(char *, const char *, size_t);
 #endif
 
+
 int
 main(int argc, char *argv[])
 {
-	struct md_parse	 p;
-	struct mdoc_cb	 cb;
-	struct stat	 st;
-	int		 c;
-	extern char	*optarg;
-	extern int	 optind;
+	struct mmain	*p;
+	const struct mdoc *mdoc;
+	struct termp	 termp;
 
-	(void)memset(&p, 0, sizeof(struct md_parse));
+	p = mmain_alloc();
 
-	while (-1 != (c = getopt(argc, argv, "vW:")))
-		switch (c) {
-		case ('v'):
-			p.dbg++;
-			break;
-		case ('W'):
-			if ( ! getsopts(&p, optarg))
-				return(0);
-			break;
-		default:
-			usage();
-			return(0);
+	if ( ! mmain_getopt(p, argc, argv, NULL, NULL, NULL, NULL))
+		mmain_exit(p, 1);
+
+	if (NULL == (mdoc = mmain_mdoc(p)))
+		mmain_exit(p, 1);
+
+	termp.maxrmargin = 80; /* XXX */
+	termp.rmargin = termp.maxrmargin;
+	termp.maxcols = 1024;
+	termp.offset = termp.col = 0;
+	termp.flags = TERMP_NOSPACE;
+
+	if (NULL == (termp.buf = malloc(termp.maxcols)))
+		err(1, "malloc");
+
+	header(&termp, mdoc_meta(mdoc));
+	body(&termp, NULL, mdoc_meta(mdoc), mdoc_node(mdoc));
+	footer(&termp, mdoc_meta(mdoc));
+
+	free(termp.buf);
+
+	mmain_exit(p, 0);
+	/* NOTREACHED */
+}
+
+
+void
+flushln(struct termp *p)
+{
+	size_t		 i, j, vsz, vis, maxvis;
+
+	/*
+	 * 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;
+	}
+
+	for (i = 0; i < p->col; i++) {
+		/*
+		 * Count up visible word characters.  Control sequences
+		 * (starting with the CSI) aren't counted. 
+		 */
+		assert( ! xisspace(p->buf[i]));
+
+		/* LINTED */
+		for (j = i, vsz = 0; j < p->col; j++) {
+			if (xisspace(p->buf[j]))
+				break;
+			else if (27 == p->buf[j]) {
+				assert(j + 4 <= p->col);
+				j += 3;
+			} else
+				vsz++;
 		}
+		assert(vsz > 0);
 
-	argv += optind;
-	argc -= optind;
+		/*
+		 * 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).
+		 */
 
-	/* Initialise the input file. */
+		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");
 
-	p.in = "-";
-	p.fdin = STDIN_FILENO;
+		/* 
+		 * Write out the word and a trailing space.  Omit the
+		 * space if we're the last word in the line.
+		 */
 
-	if (argc > 0) {
-		p.in = *argv++;
-		p.fdin = open(p.in, O_RDONLY, 0);
-		if (-1 == p.fdin)
-			err(1, "%s", p.in);
+		for ( ; i < p->col; i++) {
+			if (xisspace(p->buf[i]))
+				break;
+			putchar(p->buf[i]);
+		}
+		vis += vsz;
+		if (i < p->col) {
+			putchar(' ');
+			vis++;
+		}
 	}
 
-	/* Allocate a buffer to be BUFSIZ/block size. */
+	/*
+	 * If we're not to right-marginalise it (newline), then instead
+	 * pad to the right margin and stay off.
+	 */
 
-	if (-1 == fstat(p.fdin, &st)) {
-		warn("%s", p.in);
-		p.bufsz = BUFSIZ;
-	} else 
-		p.bufsz = MAX(st.st_blksize, BUFSIZ);
+	if (p->flags & TERMP_NOBREAK) {
+		for ( ; vis < maxvis; vis++)
+			putchar(' ');
+	} else
+		putchar('\n');
 
-	p.buf = malloc(p.bufsz);
-	if (NULL == p.buf)
-		err(1, "malloc");
+	p->col = 0;
+}
 
-	/* Allocate the parser. */
 
-	cb.mdoc_err = msg_err;
-	cb.mdoc_warn = msg_warn;
-	cb.mdoc_msg = msg_msg;
+void
+newln(struct termp *p)
+{
 
-	p.mdoc = mdoc_alloc(&p, &cb);
+	/* 
+	 * A newline only breaks an existing line; it won't assert
+	 * vertical space.
+	 */
+	p->flags |= TERMP_NOSPACE;
+	if (0 == p->col) {
+		p->flags &= ~TERMP_NOLPAD;
+		return;
+	}
+	flushln(p);
+	p->flags &= ~TERMP_NOLPAD;
+}
 
-	/* Parse the input file. */
 
-	c = parse(&p);
-	free(p.buf);
+void
+vspace(struct termp *p)
+{
 
-	if (STDIN_FILENO != p.fdin && -1 == close(p.fdin))
-		warn("%s", p.in);
+	/*
+	 * Asserts a vertical space (a full, empty line-break between
+	 * lines).
+	 */
+	newln(p);
+	putchar('\n');
+}
 
-	if (0 == c) {
-		mdoc_free(p.mdoc);
-		return(EXIT_FAILURE);
+
+static void
+stringa(struct termp *p, const char *s)
+{
+
+	/* XXX - speed up if not passing to chara. */
+	for ( ; *s; s++)
+		chara(p, *s);
+}
+
+
+static void
+chara(struct termp *p, char c)
+{
+
+	/* 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 */
 	}
+	p->buf[(p->col)++] = 'm';
+}
 
-	/* If the parse succeeded, print it out. */
 
-	termprint(mdoc_node(p.mdoc), mdoc_meta(p.mdoc));
-	mdoc_free(p.mdoc);
+static void
+nescape(struct termp *p, const char *word, size_t len)
+{
 
-	return(EXIT_SUCCESS);
+	switch (len) {
+	case (2):
+		if ('r' == word[0] && 'B' == word[1])
+			chara(p, ']');
+		else if ('l' == word[0] && 'B' == word[1])
+			chara(p, '[');
+		else if ('<' == word[0] && '-' == word[1])
+			stringa(p, "<-");
+		else if ('-' == word[0] && '>' == word[1])
+			stringa(p, "->");
+		else if ('l' == word[0] && 'q' == word[1])
+			chara(p, '\"');
+		else if ('r' == word[0] && 'q' == word[1])
+			chara(p, '\"');
+		else if ('b' == word[0] && 'u' == word[1])
+			chara(p, 'o');
+		break;
+	default:
+		break;
+	}
 }
 
 
-static int
-getsopts(struct md_parse *p, char *arg)
+static void
+pescape(struct termp *p, const char *word, size_t *i, size_t len)
 {
-	char		*v;
-	char		*toks[] = { "all", "compat", 
-				"syntax", "error", NULL };
+	size_t		 j;
 
-	while (*arg) 
-		switch (getsubopt(&arg, toks, &v)) {
-		case (0):
-			p->warn |= MD_WARN_ALL;
-			break;
-		case (1):
-			p->warn |= MD_WARN_COMPAT;
-			break;
-		case (2):
-			p->warn |= MD_WARN_SYNTAX;
-			break;
-		case (3):
-			p->warn |= MD_WARN_ERR;
-			break;
+	(*i)++;
+	assert(*i < len);
+
+	if ('(' == word[*i]) {
+		/* Two-character escapes. */
+		(*i)++;
+		assert(*i + 1 < len);
+		nescape(p, &word[*i], 2);
+		(*i)++;
+		return;
+
+	} else if ('[' != word[*i]) {
+		/* One-character escapes. */
+		switch (word[*i]) {
+		case ('\\'):
+			/* FALLTHROUGH */
+		case ('\''):
+			/* FALLTHROUGH */
+		case ('`'):
+			/* FALLTHROUGH */
+		case ('-'):
+			/* FALLTHROUGH */
+		case ('.'):
+			chara(p, word[*i]);
 		default:
-			usage();
-			return(0);
+			break;
 		}
+		return;
+	}
 
-	return(1);
+	(*i)++;
+	for (j = 0; word[*i] && ']' != word[*i]; (*i)++, j++)
+		/* Loop... */ ;
+
+	nescape(p, &word[*i - j], j);
 }
 
 
-static int
-parse(struct md_parse *p)
+static void
+pword(struct termp *p, const char *word, size_t len)
 {
-	ssize_t		 sz, i;
-	size_t		 pos;
-	char		 line[MD_LINE_SZ];
-	int		 lnn;
+	size_t		 i;
 
-	/*
-	 * This is a little more complicated than fgets.  TODO: have
-	 * some benchmarks that show it's faster (note that I want to
-	 * check many, many manuals simultaneously, so speed is
-	 * important).  Fill a buffer (sized to the block size) with a
-	 * single read, then parse \n-terminated lines into a line
-	 * buffer, which is passed to the parser.  Hard-code the line
-	 * buffer to a particular size -- a reasonable assumption.
-	 */
+	/*assert(len > 0);*/ /* Can be, if literal. */
 
-	for (lnn = 1, pos = 0; ; ) {
-		if (-1 == (sz = read(p->fdin, p->buf, p->bufsz))) {
-			warn("%s", p->in);
-			return(0);
-		} else if (0 == sz) 
-			break;
+	if ( ! (p->flags & TERMP_NOSPACE) && 
+			! (p->flags & TERMP_LITERAL))
+		chara(p, ' ');
 
-		for (i = 0; i < sz; i++) {
-			if ('\n' != p->buf[i]) {
-				if (pos < sizeof(line)) {
-					line[(int)pos++] = p->buf[(int)i];
-					continue;
-				}
-				warnx("%s: line %d too long", p->in, lnn);
-				return(0);
-			}
-	
-			line[(int)pos] = 0;
-			if ( ! mdoc_parseln(p->mdoc, lnn, line))
-				return(0);
+	p->flags &= ~TERMP_NOSPACE;
 
-			lnn++;
-			pos = 0;
+	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(mdoc_endparse(p->mdoc));
+	if (p->flags & TERMP_BOLD ||
+			p->flags & TERMP_UNDERLINE)
+		style(p, STYLE_CLEAR);
 }
 
 
-static int
-msg_err(void *arg, int line, int col, const char *msg)
+void
+word(struct termp *p, const char *word)
 {
-	struct md_parse	 *p;
+	size_t 		 i, j, len;
 
-	p = (struct md_parse *)arg;
+	if (p->flags & TERMP_LITERAL) {
+		pword(p, word, strlen(word));
+		return;
+	}
 
-	warnx("%s:%d: error: %s (column %d)", 
-			p->in, line, msg, col);
-	return(0);
+	len = strlen(word);
+	assert(len > 0);
+
+	if (mdoc_isdelim(word)) {
+		if ( ! (p->flags & TERMP_IGNDELIM))
+			p->flags |= TERMP_NOSPACE;
+		p->flags &= ~TERMP_IGNDELIM;
+	}
+
+	/* LINTED */
+	for (j = i = 0; i < len; i++) {
+		if ( ! xisspace(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
-msg_msg(void *arg, int line, int col, const char *msg)
+body(struct termp *p, struct termpair *ppair,
+		const struct mdoc_meta *meta,
+		const struct mdoc_node *node)
 {
-	struct md_parse	 *p;
+	int		 dochild;
+	struct termpair	 pair;
 
-	p = (struct md_parse *)arg;
+	/* Pre-processing. */
 
-	if (0 == p->dbg)
-		return;
+	dochild = 1;
+	pair.ppair = ppair;
+	pair.type = 0;
+	pair.offset = pair.rmargin = 0;
+	pair.flag = 0;
+	pair.count = 0;
 
-	warnx("%s:%d: debug: %s (column %d)", 
-			p->in, line, msg, col);
+	if (MDOC_TEXT != node->type) {
+		if (termacts[node->tok].pre)
+			if ( ! (*termacts[node->tok].pre)(p, &pair, meta, node))
+				dochild = 0;
+	} else /* MDOC_TEXT == node->type */
+		word(p, node->data.text.string);
+
+	/* Children. */
+
+	if (TERMPAIR_FLAG & pair.type)
+		p->flags |= pair.flag;
+
+	if (dochild && node->child)
+		body(p, &pair, meta, node->child);
+
+	if (TERMPAIR_FLAG & pair.type)
+		p->flags &= ~pair.flag;
+
+	/* Post-processing. */
+
+	if (MDOC_TEXT != node->type)
+		if (termacts[node->tok].post)
+			(*termacts[node->tok].post)(p, &pair, meta, node);
+
+	/* Siblings. */
+
+	if (node->next)
+		body(p, ppair, meta, node->next);
 }
 
 
-static int
-msg_warn(void *arg, int line, int col, 
-		enum mdoc_warn type, const char *msg)
+static void
+footer(struct termp *p, const struct mdoc_meta *meta)
 {
-	struct md_parse	 *p;
+	struct tm	*tm;
+	char		*buf, *os;
+	size_t		 sz, osz, ssz, i;
 
-	p = (struct md_parse *)arg;
+	if (NULL == (buf = malloc(p->rmargin)))
+		err(1, "malloc");
+	if (NULL == (os = malloc(p->rmargin)))
+		err(1, "malloc");
 
-	switch (type) {
-	case (WARN_COMPAT):
-		if (p->warn & MD_WARN_COMPAT)
-			break;
-		return(1);
-	case (WARN_SYNTAX):
-		if (p->warn & MD_WARN_SYNTAX)
-			break;
-		return(1);
-	}
+	tm = localtime(&meta->date);
 
-	warnx("%s:%d: warning: %s (column %d)", 
-			p->in, line, msg, col);
+#ifdef __OpenBSD__
+	if (NULL == strftime(buf, p->rmargin, "%B %d, %Y", tm))
+#else
+	if (0 == strftime(buf, p->rmargin, "%B %d, %Y", tm))
+#endif
+		err(1, "strftime");
 
-	if ( ! (p->warn & MD_WARN_ERR))
-		return(1);
+	osz = strlcpy(os, meta->os, p->rmargin);
 
-	warnx("%s: considering warnings as errors", __progname);
-	return(0);
+	sz = strlen(buf);
+	ssz = sz + osz + 1;
+
+	if (ssz > p->rmargin) {
+		ssz -= p->rmargin;
+		assert(ssz <= osz);
+		os[osz - ssz] = 0;
+		ssz = 1;
+	} else
+		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);
 }
 
 
 static void
-usage(void)
+header(struct termp *p, const struct mdoc_meta *meta)
 {
+	char		*buf, *title;
+	const char	*pp, *msec;
+	size_t		 ssz, tsz, ttsz, i;;
 
-	warnx("usage: %s [-v] [-Wwarn...] [infile]", __progname);
-}
+	if (NULL == (buf = malloc(p->rmargin)))
+		err(1, "malloc");
+	if (NULL == (title = malloc(p->rmargin)))
+		err(1, "malloc");
 
+	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 = strlcpy(buf, pp, p->rmargin);
+	assert(tsz < p->rmargin);
+
+	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 > p->rmargin) {
+		if ((ssz -= p->rmargin) % 2)
+			ssz++;
+		ssz /= 2;
+	
+		assert(ssz <= ttsz);
+		title[ttsz - ssz] = 0;
+		ssz = 1;
+	} else
+		ssz = ((p->rmargin - ssz) / 2) + 1;
+
+	printf("%s(%s)", title, msec);
+
+	for (i = 0; i < ssz; i++)
+		printf(" ");
+
+	printf("%s", buf);
+
+	for (i = 0; i < ssz; i++)
+		printf(" ");
+
+	printf("%s(%s)\n", title, msec);
+	fflush(stdout);
+
+	free(title);
+	free(buf);
+}