Added a function to check whether a (UTF-8) string is a valid token.

[gedcom-parse.git] / gedcom / encoding.c

diff --git a/gedcom/encoding.c b/gedcom/encoding.c
index b2c3c416955e65d4465f4dc7eff2e71e12d57ee7..f70dc4eb7ae47aeaba7e7555fade8b0825ea6c9d 100644 (file)
--- a/gedcom/encoding.c
+++ b/gedcom/encoding.c
@@ -39,7 +39,7 @@ static iconv_t cd_to_internal = (iconv_t) -1;
 static ENCODING the_enc = ONE_BYTE;
 static hash_t *encodings = NULL;
 
-char* charwidth_string[] = { "1", "2_HILO", "2_LOHI" };
+const char* charwidth_string[] = { "1", "2_HILO", "2_LOHI" };
 
 hnode_t *node_alloc(void *c __attribute__((unused)))
 {
@@ -53,7 +53,8 @@ void node_free(hnode_t *n, void *c __attribute__((unused)))
   free(n);
 }
 
-void add_encoding(char *gedcom_n, char* charwidth, char *iconv_n)
+void add_encoding(const char *gedcom_n, const char* charwidth,
+                 const char *iconv_n)
 {
   char *key, *val;
 
@@ -79,7 +80,7 @@ void add_encoding(char *gedcom_n, char* charwidth, char *iconv_n)
     MEMORY_ERROR;
 }
 
-char* get_encoding(char* gedcom_n, ENCODING enc)
+char* get_encoding(const char* gedcom_n, ENCODING enc)
 {
   char *key;
   hnode_t *node;
@@ -106,11 +107,21 @@ char* get_encoding(char* gedcom_n, ENCODING enc)
   }
 }
 
+static char *new_gconv_path;
+
 void cleanup_encodings()
 {
   hash_free(encodings);
 }
 
+void cleanup_gconv_path()
+{
+  /* Clean up environment */
+  putenv(GCONV_SEARCH_PATH);
+  if (new_gconv_path)
+    free(new_gconv_path);  
+}
+
 /* Let function be called before main() */
 void update_gconv_search_path() __attribute__ ((constructor));
 
@@ -139,7 +150,6 @@ void update_gconv_search_path()
   /* Add gedcom data directory to gconv search path */
   gconv_path = getenv(GCONV_SEARCH_PATH);
   if (gconv_path == NULL || strstr(gconv_path, PKGDATADIR) == NULL) {
-    char *new_gconv_path;
     if (gconv_path == NULL) {
       new_gconv_path = (char *)malloc(strlen(GCONV_SEARCH_PATH)
                                      + strlen(PKGDATADIR)
@@ -165,6 +175,9 @@ void update_gconv_search_path()
       abort();
     }
   }
+  if (init_called && atexit(cleanup_gconv_path) != 0) {
+    gedcom_warning(_("Could not register path cleanup function"));
+  }    
 }
 
 void init_encodings()
@@ -229,24 +242,28 @@ void set_encoding_width(ENCODING enc)
 static char conv_buf[MAXGEDCLINELEN * 2];
 static size_t conv_buf_size;
 
-int open_conv_to_internal(char* fromcode)
+int open_conv_to_internal(const char* fromcode)
 {
-  char *encoding = get_encoding(fromcode, the_enc);
-  if (cd_to_internal != (iconv_t) -1)
-    iconv_close(cd_to_internal);
+  iconv_t new_cd_to_internal;
+  const char *encoding = get_encoding(fromcode, the_enc);
   if (encoding == NULL) {
-    cd_to_internal = (iconv_t) -1;
+    new_cd_to_internal = (iconv_t) -1;
   }
   else {
     memset(conv_buf, 0, sizeof(conv_buf));
     conv_buf_size = 0;
-    cd_to_internal = iconv_open(INTERNAL_ENCODING, encoding);
-    if (cd_to_internal == (iconv_t) -1) {
+    new_cd_to_internal = iconv_open(INTERNAL_ENCODING, encoding);
+    if (new_cd_to_internal == (iconv_t) -1) {
       gedcom_error(_("Error opening conversion context for encoding %s: %s"),
                   encoding, strerror(errno));
     }
   }
-  return (cd_to_internal != (iconv_t) -1);  
+  if (new_cd_to_internal != (iconv_t) -1) {
+    if (cd_to_internal != (iconv_t) -1)
+      iconv_close(cd_to_internal);
+    cd_to_internal = new_cd_to_internal;
+  }
+  return (new_cd_to_internal != (iconv_t) -1);  
 }
 
 void close_conv_to_internal()
@@ -257,7 +274,7 @@ void close_conv_to_internal()
   cd_to_internal = (iconv_t) -1;
 }
 
-char* to_internal(char* str, size_t len,
+char* to_internal(const char* str, size_t len,
                  char* output_buffer, size_t out_len)
 {
   size_t res;