Pass encoding terminator to encoding state.

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

diff --git a/gedcom/encoding.c b/gedcom/encoding.c
index 469854b0d8528aa78203db911c594f70cb8a547c..4828c0a6a7d3c7bc9e9ef89537b6f17b1b39efb7 100644 (file)
--- a/gedcom/encoding.c
+++ b/gedcom/encoding.c
@@ -22,7 +22,6 @@
 /* $Name$ */
 
 #include <string.h>
-#include <iconv.h>
 #include <stdio.h>
 #include <limits.h>
 #include <stdlib.h>
@@ -30,16 +29,14 @@
 #include "gedcom.h"
 #include "encoding.h"
 #include "hash.h"
-#include "utf8.h"
+#include "utf8tools.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;
+struct encoding_state read_encoding;
+
 static hash_t *encodings = NULL;
 
 const char* charwidth_string[] = { "1", "2_HILO", "2_LOHI" };
@@ -83,10 +80,12 @@ void add_encoding(const char *gedcom_n, const char* charwidth,
     MEMORY_ERROR;
 }
 
-char* get_encoding(const char* gedcom_n, ENCODING enc)
+char* get_encoding(const char* gedcom_n, Encoding enc)
 {
   char *key;
   hnode_t *node;
+
+  if (encodings == NULL) return NULL;
   
   key = (char*)malloc(strlen(gedcom_n) + strlen(charwidth_string[enc]) + 3);
 
@@ -222,6 +221,7 @@ void init_encodings()
        if (buffer[strlen(buffer) - 1] != '\n') {
          gedcom_error(_("Line too long in encoding configuration file '%s'"),
                       ENCODING_CONF_FILE);
+         line_no = 0;
          return;
        }
        else if ((buffer[0] != '#') && (strcmp(buffer, "\n") != 0)) {
@@ -231,10 +231,12 @@ void init_encodings()
          else {
            gedcom_error(_("Missing data in encoding configuration file '%s'"),
                         ENCODING_CONF_FILE);
+           line_no = 0;
            return;
          }
        }
       }
+      line_no = 0;
       if (fclose(in) != 0) {
        gedcom_warning(_("Error closing file '%s': %s"),
                       ENCODING_CONF_FILE, strerror(errno));
@@ -243,9 +245,19 @@ void init_encodings()
   }
 }
 
-void set_encoding_width(ENCODING enc)
+void set_encoding_width(Encoding enc)
+{
+  read_encoding.width = enc;
+}
+
+void set_encoding_bom(Enc_bom bom)
+{
+  read_encoding.bom = bom;
+}
+
+void set_encoding_terminator(char* term)
 {
-  the_enc = enc;
+  strncpy(read_encoding.terminator, term, MAX_TERMINATOR_LEN);
 }
 
 static convert_t to_int = NULL;
@@ -254,7 +266,7 @@ 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);
+  const char *encoding = get_encoding(fromcode, read_encoding.width);
   
   if (encoding != NULL) {
     new_to_int = initialize_utf8_conversion(encoding, 1);
@@ -268,6 +280,16 @@ int open_conv_to_internal(const char* fromcode)
     if (to_int != NULL)
       cleanup_utf8_conversion(to_int);
     to_int = new_to_int;
+    strncpy(read_encoding.charset, fromcode, MAX_CHARSET_LEN);
+    read_encoding.encoding = encoding;
+    gedcom_debug_print("Encoding state is now: ");
+    gedcom_debug_print("  charset   : %s", read_encoding.charset);
+    gedcom_debug_print("  encoding  : %s", read_encoding.encoding);
+    gedcom_debug_print("  width     : %d", read_encoding.width);
+    gedcom_debug_print("  BOM       : %d", read_encoding.bom);
+    gedcom_debug_print("  terminator: 0x%02x 0x%02x",
+                      read_encoding.terminator[0],
+                      read_encoding.terminator[1]);
   }
 
   return (new_to_int != NULL);