Change chars.in HTML encoding to be a Unicode codepoint (int), which is

later formatted in html.c.

1 files changed, 65 insertions, 24 deletions

diff --git a/chars.c b/chars.c
index 126274c0..72754ca9 100644
--- a/chars.c
+++ b/chars.c
@@ -1,4 +1,4 @@
-/*	$Id: chars.c,v 1.20 2010/06/19 20:46:27 kristaps Exp $ */
+/*	$Id: chars.c,v 1.21 2010/07/16 22:33:30 kristaps Exp $ */
 /*
  * Copyright (c) 2009 Kristaps Dzonsons <kristaps@bsd.lv>
  *
@@ -32,11 +32,10 @@
 struct	ln {
 	struct ln	 *next;
 	const char	 *code;
-	const char	 *ascii;
-	const char	 *html;
 	size_t		  codesz;
+	const char	 *ascii;
 	size_t		  asciisz;
-	size_t		  htmlsz;
+	int		  unicode;
 	int		  type;
 #define	CHARS_CHAR	 (1 << 0)
 #define	CHARS_STRING	 (1 << 1)
@@ -45,12 +44,12 @@ struct	ln {
 
 #define	LINES_MAX	  370
 
-#define CHAR(w, x, y, z, a, b) \
-	{ NULL, (w), (y), (a), (x), (z), (b), CHARS_CHAR },
-#define STRING(w, x, y, z, a, b) \
-	{ NULL, (w), (y), (a), (x), (z), (b), CHARS_STRING },
-#define BOTH(w, x, y, z, a, b) \
-	{ NULL, (w), (y), (a), (x), (z), (b), CHARS_BOTH },
+#define CHAR(in, insz, ch, chsz, code) \
+	{ NULL, (in), (insz), (ch), (chsz), (code), CHARS_CHAR },
+#define STRING(in, insz, ch, chsz, code) \
+	{ NULL, (in), (insz), (ch), (chsz), (code), CHARS_STRING },
+#define BOTH(in, insz, ch, chsz, code) \
+	{ NULL, (in), (insz), (ch), (chsz), (code), CHARS_BOTH },
 
 #define	CHAR_TBL_START	  static struct ln lines[LINES_MAX] = {
 #define	CHAR_TBL_END	  };
@@ -64,8 +63,7 @@ struct	tbl {
 
 static	inline int	  match(const struct ln *,
 				const char *, size_t, int);
-static	const char	 *find(struct tbl *, const char *, 
-				size_t, size_t *, int);
+static	const struct ln	 *find(struct tbl *, const char *, size_t, int);
 
 
 void
@@ -126,24 +124,72 @@ chars_init(enum chars type)
 }
 
 
+/* 
+ * Special character to Unicode codepoint.
+ */
+int
+chars_spec2cp(void *arg, const char *p, size_t sz)
+{
+	const struct ln	*ln;
+
+	ln = find((struct tbl *)arg, p, sz, CHARS_CHAR);
+	if (NULL == ln)
+		return(-1);
+	return(ln->unicode);
+}
+
+
+/* 
+ * Reserved word to Unicode codepoint.
+ */
+int
+chars_res2cp(void *arg, const char *p, size_t sz)
+{
+	const struct ln	*ln;
+
+	ln = find((struct tbl *)arg, p, sz, CHARS_STRING);
+	if (NULL == ln)
+		return(-1);
+	return(ln->unicode);
+}
+
+
+/* 
+ * Special character to string array.
+ */
 const char *
-chars_a2ascii(void *arg, const char *p, size_t sz, size_t *rsz)
+chars_spec2str(void *arg, const char *p, size_t sz, size_t *rsz)
 {
+	const struct ln	*ln;
+
+	ln = find((struct tbl *)arg, p, sz, CHARS_CHAR);
+	if (NULL == ln)
+		return(NULL);
 
-	return(find((struct tbl *)arg, p, sz, rsz, CHARS_CHAR));
+	*rsz = ln->asciisz;
+	return(ln->ascii);
 }
 
 
+/* 
+ * Reserved word to string array.
+ */
 const char *
-chars_a2res(void *arg, const char *p, size_t sz, size_t *rsz)
+chars_res2str(void *arg, const char *p, size_t sz, size_t *rsz)
 {
+	const struct ln	*ln;
 
-	return(find((struct tbl *)arg, p, sz, rsz, CHARS_STRING));
+	ln = find((struct tbl *)arg, p, sz, CHARS_STRING);
+	if (NULL == ln)
+		return(NULL);
+
+	*rsz = ln->asciisz;
+	return(ln->ascii);
 }
 
 
-static const char *
-find(struct tbl *tab, const char *p, size_t sz, size_t *rsz, int type)
+static const struct ln *
+find(struct tbl *tab, const char *p, size_t sz, int type)
 {
 	struct ln	 *pp, *prev;
 	struct ln	**htab;
@@ -179,12 +225,7 @@ find(struct tbl *tab, const char *p, size_t sz, size_t *rsz, int type)
 			htab[hash] = pp;
 		}
 
-		if (CHARS_HTML == tab->type) {
-			*rsz = pp->htmlsz;
-			return(pp->html);
-		}
-		*rsz = pp->asciisz;
-		return(pp->ascii);
+		return(pp);
 	}
 
 	return(NULL);