Edmund Grimley Evans' UTF-8 patch.

1 files changed, 167 insertions, 0 deletions

diff --git a/gettext.c b/gettext.c
new file mode 100644
index 00000000..338e7c5d
--- /dev/null
+++ b/gettext.c
@@ -0,0 +1,167 @@
+
+#include <string.h>
+
+#include "mutt.h"
+#include "iconv.h"
+#include "lib.h"
+#include "charset.h"
+
+/*
+ * One day, gettext will return strings in the appropriate
+ * encoding. In the meantime, we use this code to handle
+ * the conversion.
+ */
+
+struct gt_hash_elem
+{
+  const char *key;
+  char *data;
+  struct gt_hash_elem *next;
+};
+
+#define gt_hash_size 127
+
+static char *get_charset (const char *header)
+{
+  /* FIXME: the comparison should at least be case-insensitive */
+  const char f[] = "\nContent-Type: text/plain; charset=";
+  char *charset, *i, *j;
+
+  charset = 0;
+  i = strstr (header, f);
+  if (i)
+  {
+    i += sizeof (f)-1;
+    for (j = i; *j >= 32; j++)
+      ;
+    charset = safe_malloc (j-i+1);
+    memcpy (charset, i, j-i);
+    charset[j-i] = '\0';
+  }
+  return charset;
+}
+
+char *mutt_gettext (const char *message)
+{
+  static struct gt_hash_elem **messages = 0;
+  static char *po_header = 0;
+  static char *po_charset = 0;
+  static char *message_charset = 0;
+  static char *outrepl = "?";
+  static iconv_t cd = (iconv_t)-1;
+  int change_cd = 0;
+  char *t, *orig;
+  char *header_msgid = "";
+
+  /* gettext ("") doesn't work due to __builtin_constant_p optimisation */
+  if ((t = gettext (header_msgid)) != po_header)
+  {
+    po_header = t;
+    t = get_charset (po_header);
+    if (t != po_charset &&
+	(!t || !po_charset || strcmp (t, po_charset)))
+    {
+      free (po_charset);
+      po_charset = t;
+      change_cd = 1;
+    }
+    else
+      free (t);
+  }
+
+  if (message_charset != Charset &&
+      (!message_charset || !Charset || strcmp (message_charset, Charset)))
+  {
+    free (message_charset);
+    if (Charset)
+    {
+      int n = strlen (Charset);
+      message_charset = safe_malloc (n+1);
+      memcpy (message_charset, Charset, n+1);
+    }
+    else
+      message_charset = 0;
+    outrepl = mutt_is_utf8 (message_charset) ? "\357\277\275" : "?";
+    change_cd = 1;
+  }
+
+  if (change_cd)
+  {
+    if (cd != (iconv_t)-1)
+      iconv_close (cd);
+    if (message_charset)
+      cd = iconv_open (message_charset, po_charset ? po_charset : "UTF-8");
+    else
+      cd = (iconv_t)-1;
+
+    if (messages)
+    {
+      int i;
+      struct gt_hash_elem *p, *pn;
+
+      for (i = 0; i < gt_hash_size; i++)
+	for (p = messages[i]; p; p = pn)
+	{
+	  pn = p->next;
+	  free (p);
+	}
+      free (messages);
+      messages = 0;
+    }
+  }
+
+  orig = gettext (message);
+
+  if (cd == (iconv_t)-1)
+    return orig;
+  else
+  {
+    struct gt_hash_elem *p;
+    int hash;
+    char *s, *t;
+    int n, nn;
+    const char *ib;
+    char *ob;
+    size_t ibl, obl;
+
+    if (!messages)
+    {
+      messages = safe_malloc (gt_hash_size * sizeof (*messages));
+      memset (messages, 0, gt_hash_size * sizeof (*messages));
+    }
+    hash = (long int)orig % gt_hash_size; /* not very clever */
+    for (p = messages[hash]; p && p->key != orig; p = p->next)
+      ;
+    if (p)
+      return p->data;
+
+    n = strlen (orig);
+    nn = n + 1;
+    t = safe_malloc (nn);
+
+    ib = orig, ibl = n;
+    ob = t, obl = n;
+    for (;;)
+    {
+      mutt_iconv (cd, &ib, &ibl, &ob, &obl, 0, outrepl);
+      if (!ibl || obl > 256)
+	break;
+      s = t;
+      safe_realloc ((void **)&t, nn += n);
+      ob += t - s;
+      obl += n;
+    }
+    *ob = '\0';
+    n = strlen (t);
+    s = safe_malloc (n+1);
+    memcpy (s, t, n+1);
+    free (t);
+
+    p = safe_malloc (sizeof (struct gt_hash_elem));
+    p->key = orig;
+    p->data = s;
+    p->next = messages[hash];
+    messages[hash] = p;
+    return s;
+  }
+}