Use a character-table for quick per-character substitution in `tr'. As

[mandoc.git] / term_ascii.c

diff --git a/term_ascii.c b/term_ascii.c
index 55877f42b23276907bf09e0f5e791c3b71de1703..e65f590a712b7b012952d9fff8e39609d26341ae 100644 (file)
--- a/term_ascii.c
+++ b/term_ascii.c
@@ -1,4 +1,4 @@
-/*     $Id: term_ascii.c,v 1.10 2010/09/04 20:18:53 kristaps Exp $ */
+/*     $Id: term_ascii.c,v 1.17 2011/05/20 15:48:22 kristaps Exp $ */
 /*
  * Copyright (c) 2010 Kristaps Dzonsons <kristaps@bsd.lv>
  *
@@ -21,47 +21,89 @@
 #include <sys/types.h>
 
 #include <assert.h>
+#ifdef USE_WCHAR
+# include <locale.h>
+#endif
 #include <stdint.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <unistd.h>
+#ifdef USE_WCHAR
+# include <wchar.h>
+#endif
 
+#include "mandoc.h"
 #include "out.h"
 #include "term.h"
 #include "main.h"
 
+/* 
+ * Sadly, this doesn't seem to be defined on systems even when they
+ * support it.  For the time being, remove it and let those compiling
+ * the software decide for themselves what to use.
+ */
+#if 0
+#if ! defined(__STDC_ISO_10646__)
+# undef USE_WCHAR
+#endif
+#endif
+
+static struct termp     *ascii_init(enum termenc, char *);
 static double            ascii_hspan(const struct termp *,
                                const struct roffsu *);
-static size_t            ascii_width(const struct termp *, char);
+static size_t            ascii_width(const struct termp *, int);
 static void              ascii_advance(struct termp *, size_t);
 static void              ascii_begin(struct termp *);
 static void              ascii_end(struct termp *);
 static void              ascii_endline(struct termp *);
-static void              ascii_letter(struct termp *, char);
+static void              ascii_letter(struct termp *, int);
 
+#ifdef USE_WCHAR
+static void              locale_advance(struct termp *, size_t);
+static void              locale_endline(struct termp *);
+static void              locale_letter(struct termp *, int);
+static size_t            locale_width(const struct termp *, int);
+#endif
 
-void *
-ascii_alloc(char *outopts)
+static struct termp *
+ascii_init(enum termenc enc, char *outopts)
 {
-       struct termp    *p;
        const char      *toks[2];
        char            *v;
+       struct termp    *p;
 
-       if (NULL == (p = term_alloc(TERMENC_ASCII)))
-               return(NULL);
+       p = mandoc_calloc(1, sizeof(struct termp));
+       p->enc = enc;
 
        p->tabwidth = 5;
        p->defrmargin = 78;
 
-       p->advance = ascii_advance;
        p->begin = ascii_begin;
        p->end = ascii_end;
-       p->endline = ascii_endline;
        p->hspan = ascii_hspan;
-       p->letter = ascii_letter;
        p->type = TERMTYPE_CHAR;
+
+       p->enc = TERMENC_ASCII;
+       p->advance = ascii_advance;
+       p->endline = ascii_endline;
+       p->letter = ascii_letter;
        p->width = ascii_width;
 
+#ifdef USE_WCHAR
+       if (TERMENC_ASCII != enc) {
+               v = TERMENC_LOCALE == enc ?
+                       setlocale(LC_ALL, "") :
+                       setlocale(LC_CTYPE, "UTF-8");
+               if (NULL != v && MB_CUR_MAX > 1) {
+                       p->enc = enc;
+                       p->advance = locale_advance;
+                       p->endline = locale_endline;
+                       p->letter = locale_letter;
+                       p->width = locale_width;
+               }
+       }
+#endif
+
        toks[0] = "width";
        toks[1] = NULL;
 
@@ -81,16 +123,36 @@ ascii_alloc(char *outopts)
        return(p);
 }
 
+void *
+ascii_alloc(char *outopts)
+{
+
+       return(ascii_init(TERMENC_ASCII, outopts));
+}
+
+void *
+utf8_alloc(char *outopts)
+{
+
+       return(ascii_init(TERMENC_UTF8, outopts));
+}
+
+
+void *
+locale_alloc(char *outopts)
+{
+
+       return(ascii_init(TERMENC_LOCALE, outopts));
+}
 
 /* ARGSUSED */
 static size_t
-ascii_width(const struct termp *p, char c)
+ascii_width(const struct termp *p, int c)
 {
 
        return(1);
 }
 
-
 void
 ascii_free(void *arg)
 {
@@ -98,17 +160,14 @@ ascii_free(void *arg)
        term_free((struct termp *)arg);
 }
 
-
 /* ARGSUSED */
 static void
-ascii_letter(struct termp *p, char c)
+ascii_letter(struct termp *p, int c)
 {
        
-       /* LINTED */
        putchar(c);
 }
 
-
 static void
 ascii_begin(struct termp *p)
 {
@@ -116,7 +175,6 @@ ascii_begin(struct termp *p)
        (*p->headf)(p, p->argf);
 }
 
-
 static void
 ascii_end(struct termp *p)
 {
@@ -124,7 +182,6 @@ ascii_end(struct termp *p)
        (*p->footf)(p, p->argf);
 }
 
-
 /* ARGSUSED */
 static void
 ascii_endline(struct termp *p)
@@ -133,19 +190,16 @@ ascii_endline(struct termp *p)
        putchar('\n');
 }
 
-
 /* ARGSUSED */
 static void
 ascii_advance(struct termp *p, size_t len)
 {
        size_t          i;
 
-       /* Just print whitespace on the terminal. */
        for (i = 0; i < len; i++)
                putchar(' ');
 }
 
-
 /* ARGSUSED */
 static double
 ascii_hspan(const struct termp *p, const struct roffsu *su)
@@ -184,3 +238,39 @@ ascii_hspan(const struct termp *p, const struct roffsu *su)
        return(r);
 }
 
+#ifdef USE_WCHAR
+/* ARGSUSED */
+static size_t
+locale_width(const struct termp *p, int c)
+{
+       int             rc;
+
+       return((rc = wcwidth(c)) < 0 ? 0 : rc);
+}
+
+/* ARGSUSED */
+static void
+locale_advance(struct termp *p, size_t len)
+{
+       size_t          i;
+
+       for (i = 0; i < len; i++)
+               putwchar(L' ');
+}
+
+/* ARGSUSED */
+static void
+locale_endline(struct termp *p)
+{
+
+       putwchar(L'\n');
+}
+
+/* ARGSUSED */
+static void
+locale_letter(struct termp *p, int c)
+{
+       
+       putwchar(c);
+}
+#endif