Add the ability to pass a NULL output buffer and/or an unlimited

output length to charset_{to,from}_unicode, permitting convenient
dry-running of conversions to determine the required output length
and/or test for the presence of difficult characters.

git-svn-id: svn://svn.tartarus.org/sgt/charset@7677 cda61777-01e9-0310-a592-d414129be87e

diff --git a/charset.h b/charset.h
index 545784f..144187f 100644 (file)
--- a/charset.h
+++ b/charset.h
@@ -135,6 +135,13 @@ extern const charset_state charset_init_state;
  * NULL, `errlen' will be ignored, and the library will choose
  * something sensible to do on its own. For Unicode, this will be
  * U+FFFD (REPLACEMENT CHARACTER).
+ * 
+ * `output' may be NULL, in which case the entire translation will
+ * be performed in theory (e.g. a dry run to work out how much
+ * space needs to be allocated for the real thing). `outlen' may
+ * also be negative, indicating an unlimited buffer length
+ * (although this is almost certainly unwise if `output' is _not_
+ * NULL).
  */
 
 int charset_to_unicode(const char **input, int *inlen,
@@ -165,6 +172,13 @@ int charset_to_unicode(const char **input, int *inlen,
  * If `input' is NULL, this routine will output the necessary bytes
  * to reset the encoding state in any way which might be required
  * at the end of an output piece of text.
+ * 
+ * `output' may be NULL, in which case the entire translation will
+ * be performed in theory (e.g. a dry run to work out how much
+ * space needs to be allocated for the real thing). `outlen' may
+ * also be negative, indicating an unlimited buffer length
+ * (although this is almost certainly unwise if `output' is _not_
+ * NULL).
  */
 
 int charset_from_unicode(const wchar_t **input, int *inlen,

diff --git a/fromucs.c b/fromucs.c
index 8090c19..da1ea64 100644 (file)
--- a/fromucs.c
+++ b/fromucs.c
@@ -8,6 +8,7 @@
 struct charset_emit_param {
     char *output;
     int outlen;
+    int writtenlen;
     int stopped;
 };
 
@@ -15,9 +16,12 @@ static void charset_emit(void *ctx, long int output)
 {
     struct charset_emit_param *param = (struct charset_emit_param *)ctx;
 
-    if (param->outlen > 0) {
-       *param->output++ = output;
-       param->outlen--;
+    if (param->outlen != 0) {
+       if (param->output)
+           *param->output++ = output;
+       if (param->outlen > 0)
+           param->outlen--;
+       param->writtenlen++;
     } else {
        param->stopped = 1;
     }
@@ -39,6 +43,7 @@ int charset_from_unicode(const wchar_t **input, int *inlen,
 
     param.output = output;
     param.outlen = outlen;
+    param.writtenlen = 0;
     param.stopped = 0;
 
     if (state)
@@ -47,7 +52,7 @@ int charset_from_unicode(const wchar_t **input, int *inlen,
        *error = FALSE;
 
     while (*inlen > 0) {
-       int lenbefore = param.output - output;
+       int lenbefore = param.writtenlen;
        int ret;
 
        if (input)
@@ -78,5 +83,5 @@ int charset_from_unicode(const wchar_t **input, int *inlen,
            (*input)++;
        (*inlen)--;
     }
-    return param.output - output;
+    return param.writtenlen;
 }

diff --git a/toucs.c b/toucs.c
index bee98ab..94689f5 100644 (file)
--- a/toucs.c
+++ b/toucs.c
@@ -8,6 +8,7 @@
 struct unicode_emit_param {
     wchar_t *output;
     int outlen;
+    int writtenlen;
     const wchar_t *errstr;
     int errlen;
     int stopped;
@@ -35,11 +36,14 @@ static void unicode_emit(void *ctx, long int output)
        outlen = 1;
     }
 
-    if (param->outlen >= outlen) {
+    if (param->outlen < 0 || param->outlen >= outlen) {
        while (outlen > 0) {
-           *param->output++ = *p++;
-           param->outlen--;
+           if (param->output)
+               *param->output++ = *p++;
+           if (param->outlen > 0)
+               param->outlen--;
            outlen--;
+           param->writtenlen++;
        }
     } else {
        param->stopped = 1;
@@ -59,13 +63,14 @@ int charset_to_unicode(const char **input, int *inlen,
     param.outlen = outlen;
     param.errstr = errstr;
     param.errlen = errlen;
+    param.writtenlen = 0;
     param.stopped = 0;
 
     if (state)
        localstate = *state;           /* structure copy */
 
     while (*inlen > 0) {
-       int lenbefore = param.output - output;
+       int lenbefore = param.writtenlen;
        spec->read(spec, (unsigned char)**input, &localstate,
                   unicode_emit, &param);
        if (param.stopped) {
@@ -83,5 +88,5 @@ int charset_to_unicode(const char **input, int *inlen,
        (*inlen)--;
     }
 
-    return param.output - output;
+    return param.writtenlen;
 }