X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~mdw/git/disorder/blobdiff_plain/22b9fa74de8e80471a5033ea067d3b360930b91d..f32583326d21b9a17862bdd375fb307175fd6b27:/lib/charset.c

diff --git a/lib/charset.c b/lib/charset.c
index 2a38fbf..3f0e8bc 100644
--- a/lib/charset.c
+++ b/lib/charset.c
@@ -33,6 +33,7 @@
 #include "configuration.h"
 #include "utf8.h"
 #include "vector.h"
+#include "unidata.h"
 
 /** @brief Low-level converstion routine
  * @param from Source encoding
@@ -91,6 +92,33 @@ uint32_t *utf82ucs4(const char *mb) {
   return d.vec;
 }
 
+/** @brief Convert one UCS-4 character to UTF-8
+ * @param c Character to convert
+ * @param d Dynamic string to append UTF-8 sequence to
+ * @return 0 on success, -1 on error
+ */
+int one_ucs42utf8(uint32_t c, struct dynstr *d) {
+  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 %#"PRIx32, c);
+    return -1;
+  }
+  return 0;
+}
+
 /** @brief Convert UCS-4 to UTF-8
  * @param u Pointer to 0-terminated UCS-4 string
  * @return Pointer to 0-terminated UTF-8 string
@@ -103,24 +131,8 @@ char *ucs42utf8(const uint32_t *u) {
 
   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");
+    if(one_ucs42utf8(c, &d))
       return 0;
-    }
   }
   dynstr_terminate(&d);
   return d.vec;
@@ -166,6 +178,66 @@ int ucs4cmp(const uint32_t *a, const uint32_t *b) {
   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 / 256][c % 256];
+
+    return ud->gc == unicode_gc_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;
+}
+
 /*
 Local Variables:
 c-basic-offset:2