Use libutf8.

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

diff --git a/gedcom/encoding.c b/gedcom/encoding.c
index 07a95891e6bfa29c9a52c6fbc619479e29975031..469854b0d8528aa78203db911c594f70cb8a547c 100644 (file)
--- a/gedcom/encoding.c
+++ b/gedcom/encoding.c
@@ -30,23 +30,26 @@
 #include "gedcom.h"
 #include "encoding.h"
 #include "hash.h"
+#include "utf8.h"
 
 #define ENCODING_CONF_FILE "gedcom.enc"
 #define GCONV_SEARCH_PATH "GCONV_PATH"
 #define MAXBUF 255
 
+/*
 static iconv_t cd_to_internal = (iconv_t) -1;
+*/
 static ENCODING the_enc = ONE_BYTE;
 static hash_t *encodings = NULL;
 
 const char* charwidth_string[] = { "1", "2_HILO", "2_LOHI" };
 
-hnode_t *node_alloc(void *c __attribute__((unused)))
+hnode_t *node_alloc(void *c UNUSED)
 {
   return (hnode_t *)malloc(sizeof *node_alloc(NULL));
 }
 
-void node_free(hnode_t *n, void *c __attribute__((unused)))
+void node_free(hnode_t *n, void *c UNUSED)
 {
   free((void*)hnode_getkey(n));
   free(hnode_get(n));
@@ -112,9 +115,23 @@ void cleanup_encodings()
   hash_free(encodings);
 }
 
+#ifdef USE_GLIBC_ICONV
+
+static char *new_gconv_path;
+
+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));
 
+#endif /* USE_GLIBC_ICONV */
+
 /* Note:
 
    The environment variable GCONV_PATH has to be adjusted before the very
@@ -136,11 +153,11 @@ void update_gconv_search_path() __attribute__ ((constructor));
 
 void update_gconv_search_path()
 {
+#ifdef USE_GLIBC_ICONV
   char *gconv_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)
@@ -166,6 +183,10 @@ void update_gconv_search_path()
       abort();
     }
   }
+  if (init_called && atexit(cleanup_gconv_path) != 0) {
+    gedcom_warning(_("Could not register path cleanup function"));
+  }    
+#endif /* USE_GLIBC_ICONV */
 }
 
 void init_encodings()
@@ -227,70 +248,44 @@ void set_encoding_width(ENCODING enc)
   the_enc = enc;
 }
 
-static char conv_buf[MAXGEDCLINELEN * 2];
-static size_t conv_buf_size;
+static convert_t to_int = NULL;
+static char* error_value = "<error>";
 
 int open_conv_to_internal(const char* fromcode)
 {
+  convert_t new_to_int = NULL;
   const char *encoding = get_encoding(fromcode, the_enc);
-  if (cd_to_internal != (iconv_t) -1)
-    iconv_close(cd_to_internal);
-  if (encoding == NULL) {
-    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) {
+  
+  if (encoding != NULL) {
+    new_to_int = initialize_utf8_conversion(encoding, 1);
+    if (new_to_int == NULL) {
       gedcom_error(_("Error opening conversion context for encoding %s: %s"),
                   encoding, strerror(errno));
     }
   }
-  return (cd_to_internal != (iconv_t) -1);  
+
+  if (new_to_int != NULL) {
+    if (to_int != NULL)
+      cleanup_utf8_conversion(to_int);
+    to_int = new_to_int;
+  }
+
+  return (new_to_int != NULL);
 }
 
 void close_conv_to_internal()
 {
-  if (iconv_close(cd_to_internal) != 0) {
-    gedcom_warning(_("Error closing conversion context: %s"), strerror(errno));
+  if (to_int != NULL) {
+    cleanup_utf8_conversion(to_int);
+    to_int = NULL;
   }
-  cd_to_internal = (iconv_t) -1;
 }
 
-char* to_internal(const char* str, size_t len,
-                 char* output_buffer, size_t out_len)
+
+char* to_internal(const char* str, size_t len, struct conv_buffer* output_buf)
 {
-  size_t res;
-  size_t outsize = out_len;
-  char *wrptr = output_buffer;
-  char *rdptr = conv_buf;
-  char *retval = output_buffer;
-  /* set up input buffer (concatenate to what was left previous time) */
-  /* can't use strcpy, because possible null bytes from unicode */
-  memcpy(conv_buf + conv_buf_size, str, len);
-  conv_buf_size += len;
-  /* set up output buffer (empty it) */
-  memset(output_buffer, 0, out_len);
-  /* do the conversion */
-  res = iconv(cd_to_internal, &rdptr, &conv_buf_size, &wrptr, &outsize);
-  if (res == (size_t)-1) {
-    if (errno == EILSEQ) {
-      /* restart from an empty state and return NULL */
-      iconv(cd_to_internal, NULL, NULL, NULL, NULL);
-      retval = NULL;
-      rdptr++;
-      conv_buf_size--;
-    }
-    else if (errno == EINVAL) {
-      /* Do nothing, leave it to next iteration */
-    }
-    else {
-      gedcom_error(_("Error in converting characters: %s"), strerror(errno));
-    }
-  }
-  /* then shift what is left over to the head of the input buffer */
-  memmove(conv_buf, rdptr, conv_buf_size);
-  memset(conv_buf + conv_buf_size, 0, sizeof(conv_buf) - conv_buf_size);
-  return retval;
+  if (conversion_set_output_buffer(to_int, output_buf))
+    return convert_to_utf8_incremental(to_int, str, len);
+  else
+    return error_value;
 }