X-Git-Url: https://git.distorted.org.uk/~mdw/disorder/blobdiff_plain/763d5e6ad88ef3ba1cd1d7742d060e4f1e54c6b8..5617aaff51ba333441230e3808bc697e66540492:/lib/charset.c

diff --git a/lib/charset.c b/lib/charset.c
index f101ec2..c1b07bf 100644
--- a/lib/charset.c
+++ b/lib/charset.c
@@ -17,6 +17,7 @@
  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307
  * USA
  */
+/** @file lib/charset.c @brief Character set conversion */
 
 #include <config.h>
 #include "types.h"
@@ -32,7 +33,15 @@
 #include "configuration.h"
 #include "utf8.h"
 #include "vector.h"
-
+#include "unidata.h"
+
+/** @brief Low-level converstion routine
+ * @param from Source encoding
+ * @param to Destination encoding
+ * @param ptr First byte to convert
+ * @param n Number of bytes to convert
+ * @return Converted text, 0-terminated; or NULL on error.
+ */
 static void *convert(const char *from, const char *to,
 		     const void *ptr, size_t n) {
   iconv_t i;
@@ -61,68 +70,28 @@ static void *convert(const char *from, const char *to,
   return buf;
 }
 
-/* not everybody's iconv supports UCS-4, and it's inconvenient to have to know
- * our endianness, and it's easy to convert it ourselves, so we do */
-uint32_t *utf82ucs4(const char *mb) {
-  struct dynstr_ucs4 d;
-  uint32_t c;
-
-  dynstr_ucs4_init(&d);
-  while(*mb) {
-    PARSE_UTF8(mb, c,
-	       error(0, "invalid UTF-8 sequence"); return 0;);
-    dynstr_ucs4_append(&d, c);
-  }
-  dynstr_ucs4_terminate(&d);
-  return d.vec;
-}
-
-char *ucs42utf8(const uint32_t *u) {
-  struct dynstr d;
-  uint32_t c;
-
-  dynstr_init(&d);
-  while((c = *u++)) {
-    if(c < 0x80)
-      dynstr_append(&d, c);
-    else if(c < 0x800) {
-      dynstr_append(&d, 0xC0 | (c >> 6));
-      dynstr_append(&d, 0x80 | (c & 0x3F));
-    } else if(c < 0x10000) {
-      dynstr_append(&d, 0xE0 | (c >> 12));
-      dynstr_append(&d, 0x80 | ((c >> 6) & 0x3F));
-      dynstr_append(&d, 0x80 | (c & 0x3F));
-    } else if(c < 0x110000) {
-      dynstr_append(&d, 0xF0 | (c >> 18));
-      dynstr_append(&d, 0x80 | ((c >> 12) & 0x3F));
-      dynstr_append(&d, 0x80 | ((c >> 6) & 0x3F));
-      dynstr_append(&d, 0x80 | (c & 0x3F));
-    } else {
-      error(0, "invalid UCS-4 character");
-      return 0;
-    }
-  }
-  dynstr_terminate(&d);
-  return d.vec;
-}
-
+/** @brief Convert from the local multibyte encoding to UTF-8 */
 char *mb2utf8(const char *mb) {
   return convert(nl_langinfo(CODESET), "UTF-8", mb, strlen(mb) + 1);
 }
 
+/** @brief Convert from UTF-8 to the local multibyte encoding */
 char *utf82mb(const char *utf8) {
   return convert("UTF-8", nl_langinfo(CODESET), utf8, strlen(utf8) + 1);
 }
 
+/** @brief Convert from encoding @p from to UTF-8 */
 char *any2utf8(const char *from, const char *any) {
   return convert(from, "UTF-8", any, strlen(any) + 1);
 }
 
+/** @brief Convert from encoding @p from to the local multibyte encoding */
 char *any2mb(const char *from, const char *any) {
   if(from) return convert(from, nl_langinfo(CODESET), any, strlen(any) + 1);
   else return xstrdup(any);
 }
 
+/** @brief Convert from encoding @p from to encoding @p to */
 char *any2any(const char *from,
 	      const char *to,
 	      const char *any) {
@@ -130,11 +99,64 @@ char *any2any(const char *from,
   else return xstrdup(any);
 }
 
-int ucs4cmp(const uint32_t *a, const uint32_t *b) {
-  while(*a && *b && *a == *b) ++a, ++b;
-  if(*a > *b) return 1;
-  else if(*a < *b) return -1;
-  else return 0;
+/** @brief Return nonzero if @p c is a combining character */
+static int combining(int c) {
+  if(c < UNICODE_NCHARS) {
+    const struct unidata *const ud = &unidata[c / UNICODE_MODULUS][c % UNICODE_MODULUS];
+
+    return ud->general_category == unicode_General_Category_Mn || ud->ccc != 0;
+  }
+  /* Assume unknown characters are noncombining */
+  return 0;
+}
+
+/** @brief Truncate a string for display purposes
+ * @param s Pointer to UTF-8 string
+ * @param max Maximum number of columns
+ * @return @p or truncated string (never NULL)
+ *
+ * We don't correctly support bidi or double-width characters yet, nor
+ * locate default grapheme cluster boundaries for saner truncation.
+ */
+const char *truncate_for_display(const char *s, long max) {
+  const char *t = s, *r, *cut = 0;
+  char *truncated;
+  uint32_t c;
+  long n = 0;
+
+  /* We need to discover two things: firstly whether the string is
+   * longer than @p max glyphs and secondly if it is not, where to cut
+   * the string.
+   *
+   * Combining characters follow their base character (unicode
+   * standard 5.0 s2.11), so after each base character we must 
+   */
+  while(*t) {
+    PARSE_UTF8(t, c, return s);
+    if(combining(c))
+      /* This must be an initial combining character.  We just skip it. */
+      continue;
+    /* So c must be a base character.  It may be followed by any
+     * number of combining characters.  We advance past them. */
+    do {
+      r = t;
+      PARSE_UTF8(t, c, return s);
+    } while(combining(c));
+    /* Last character wasn't a combining character so back up */
+    t = r;
+    ++n;
+    /* So now there are N glyphs before position T.  We might
+     * therefore have reached the cut position. */
+    if(n == max - 3)
+      cut = t;
+  }
+  /* If the string is short enough we return it unmodified */
+  if(n < max)
+    return s;
+  truncated = xmalloc_noptr(cut - s + 4);
+  memcpy(truncated, s, cut - s);
+  strcpy(truncated + (cut - s), "...");
+  return truncated;
 }
 
 /*