uniconv: parse //ignore and //translit extension in `to' charset

1 files changed, 43 insertions, 4 deletions

diff --git a/uniconv.c b/uniconv.c
index d38269e..b0a1ccb 100644
--- a/uniconv.c
+++ b/uniconv.c
@@ -10,9 +10,15 @@
 #include <stdio.h>
 #include <assert.h>
 
+#define UNICONV_MODE_IGNORE   (1 << 0)
+#define UNICONV_MODE_TRANSLIT (1 << 1)
+
 struct _uniconv {
     struct converter *from;
     struct converter *to;
+    char from_charset[64];
+    char to_charset[64];
+    int mode;
 };
 
 static char
@@ -23,9 +29,22 @@ uniconv_lower(char c)
     return c;
 }
 
+static char *
+uniconv_strrstr(const char *str, const char *token)
+{
+    int len = strlen(token);
+    const char *p = str + strlen(str);
+
+    while (str <= --p)
+        if (p[0] == token[0] && strncmp(p, token, len) == 0)
+            return (char *)p;
+    return NULL;
+}
+
 static const char*
 uniconv_canonical_charset(const char *charset,
-			  char *buf, size_t bufsz)
+			  char *buf, size_t bufsz,
+			  int *retmode)
 {
     size_t i;
     size_t len = strlen(charset);
@@ -39,6 +58,19 @@ uniconv_canonical_charset(const char *charset,
 	    buf[i] = uniconv_lower(charset[i]);
     }
 
+    if (retmode) {
+	char *p;
+
+	for (p = uniconv_strrstr(buf, "//"); p;
+	     p = uniconv_strrstr(buf, "//")) {
+	    if (!strcmp(buf + 2, "ignore"))
+		*retmode |= UNICONV_MODE_IGNORE;
+	    else if (!strcmp(buf + 2, "translit"))
+		*retmode |= UNICONV_MODE_TRANSLIT;
+	    *p = '\0';
+	}
+    }
+
     return get_canonical_charset(buf);
 }
 
@@ -46,18 +78,22 @@ uniconv_t*
 uniconv_open(const char *to, const char *from)
 {
     char frombuf[64];
-    char tobuf[64];
+    char tobuf[64 * 2];
+    int mode;
     uniconv_t *uc;
     struct converter *fc, *tc;
 
     if (!from || !to)
 	return NULL;
 
-    from = uniconv_canonical_charset(from, frombuf, sizeof(frombuf));
-    to = uniconv_canonical_charset(to, tobuf, sizeof(tobuf));
+    mode = 0;
+    from = uniconv_canonical_charset(from, frombuf, sizeof(frombuf), NULL);
+    to = uniconv_canonical_charset(to, tobuf, sizeof(tobuf), &mode);
     if (!from || !to)
 	return NULL;
 
+    if (strlen(from) > 63 || strlen(to) > 63)
+	return NULL;
     fc = converter_open(from);
     tc = converter_open(to);
     if (!fc || !tc)
@@ -68,6 +104,9 @@ uniconv_open(const char *to, const char *from)
 	goto close_conv;
     uc->from = fc;
     uc->to = tc;
+    uc->mode = mode;
+    strcpy(uc->from_charset, from);
+    strcpy(uc->to_charset, to);
     return uc;
  close_conv:
     converter_close(fc);