X-Git-Url: https://git.dlugolecki.net.pl/?a=blobdiff_plain;f=encoding.c;h=2efe7ec72071ecdf01bf02a618f8c40447bc76ca;hb=471aa410dba0a5604cc40f4d23fc34efb098c778;hp=ba5aa1d73395778815776000c01a48889aff7b46;hpb=845f7ac49d5e63b96aca7a4cd1f60ac5e3230033;p=gedcom-parse.git

diff --git a/encoding.c b/encoding.c
index ba5aa1d..2efe7ec 100644
--- a/encoding.c
+++ b/encoding.c
@@ -1,3 +1,15 @@
+/*  This program is free software; you can redistribute it and/or modify  *
+ *  it under the terms of the GNU General Public License as published by  *
+ *  the Free Software Foundation; either version 2 of the License, or     *
+ *  (at your option) any later version.                                   *
+
+ (C) 2001 by The Genes Development Team
+ Original author: Peter Verthez (Peter.Verthez@advalvas.be)
+*/
+
+/* $Id$ */
+/* $Name$ */
+
 #include <string.h>
 #include <iconv.h>
 #include <search.h>
@@ -10,27 +22,31 @@
 #define MAXBUF 255
 
 static iconv_t cd_to_internal = (iconv_t) -1;
-static char int_buf[MAXGEDCLINELEN*2];
 static void *encoding_mapping = NULL;
+static ENCODING the_enc = ONE_BYTE;
 
 struct node {
   char *gedcom_name;
   char *iconv_name;
 };
 
+char* charwidth_string[] = { "1", "2_HILO", "2_LOHI" };
+
 int node_compare(const void *node1, const void *node2)
 {
   return strcmp(((const struct node *) node1)->gedcom_name,
 		((const struct node *) node2)->gedcom_name);
 }
 
-void add_encoding(char *gedcom_n, char *iconv_n)
+void add_encoding(char *gedcom_n, char* charwidth, char *iconv_n)
 {
   void **datum;
   struct node *nodeptr = (struct node *) malloc(sizeof *nodeptr);
-  nodeptr->gedcom_name = (char *) malloc(strlen(gedcom_n) + 1);
+  nodeptr->gedcom_name = (char *) malloc(strlen(gedcom_n)
+					 + strlen(charwidth) + 3);
   nodeptr->iconv_name  = (char *) malloc(strlen(iconv_n) + 1);
-  strcpy(nodeptr->gedcom_name, gedcom_n);
+  /* sprintf is safe here (malloc'ed before) */
+  sprintf(nodeptr->gedcom_name, "%s(%s)", gedcom_n, charwidth);
   strcpy(nodeptr->iconv_name, iconv_n);
   datum = tsearch(nodeptr, &encoding_mapping, node_compare);
   if ((datum == NULL) || (*datum != nodeptr)) {
@@ -39,12 +55,17 @@ void add_encoding(char *gedcom_n, char *iconv_n)
   }
 }
 
-char* get_encoding(char* gedcom_n)
+char* get_encoding(char* gedcom_n, ENCODING enc)
 {
   void **datum;
   struct node search_node;
-  search_node.gedcom_name = gedcom_n;
+  char *buffer;
+  buffer = (char*)malloc(strlen(gedcom_n) + strlen(charwidth_string[enc]) + 3);
+  /* sprintf is safe here (malloc'ed before) */
+  sprintf(buffer, "%s(%s)", gedcom_n, charwidth_string[enc]);
+  search_node.gedcom_name = buffer;
   datum = tfind(&search_node, &encoding_mapping, node_compare);
+  free(buffer);
   if (datum == NULL) {
     gedcom_error("No encoding found for '%s'", gedcom_n);
     return NULL;
@@ -60,6 +81,7 @@ void init_encodings()
     FILE *in;
     char buffer[MAXBUF + 1];
     char gedcom_n[MAXBUF + 1];
+    char charwidth[MAXBUF + 1];
     char iconv_n[MAXBUF + 1];
     in = fopen(ENCODING_CONF_FILE, "r");
     if (in != NULL) {
@@ -69,9 +91,14 @@ void init_encodings()
 		       ENCODING_CONF_FILE);
 	  return;
 	}
-	else if (buffer[0] != '#') {
-	  if (sscanf(buffer, "%s %s", gedcom_n, iconv_n) == 2) {
-	    add_encoding(gedcom_n, iconv_n);
+	else if ((buffer[0] != '#') && (strcmp(buffer, "\n") != 0)) {
+	  if (sscanf(buffer, "%s %s %s", gedcom_n, charwidth, iconv_n) == 3) {
+	    add_encoding(gedcom_n, charwidth, iconv_n);
+	  }
+	  else {
+	    gedcom_error("Missing data in encoding configuration file '%s'",
+			 ENCODING_CONF_FILE);
+	    return;
 	  }
 	}
       }
@@ -84,16 +111,30 @@ void init_encodings()
   }
 }
 
+void set_encoding_width(ENCODING enc)
+{
+  the_enc = enc;
+}
+
+static char conv_buf[MAXGEDCLINELEN * 2];
+static size_t conv_buf_size;
+
 int open_conv_to_internal(char* fromcode)
 {
-  char *encoding = get_encoding(fromcode);
+  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) {
+      gedcom_error("Error opening conversion context for encoding %s: %s",
+		   encoding, strerror(errno));
+    }
   }
   return (cd_to_internal != (iconv_t) -1);  
 }
@@ -104,14 +145,22 @@ void close_conv_to_internal()
   cd_to_internal = (iconv_t) -1;
 }
 
-char* to_internal(char* str, size_t len)
+char* to_internal(char* str, size_t len,
+		  char* output_buffer, size_t out_len)
 {
-  size_t insize = len;
-  size_t outsize = MAXGEDCLINELEN * 2;
-  char *wrptr = int_buf;
-  char *rdptr = str;
-  memset(int_buf, 0, sizeof(int_buf));
-  iconv(cd_to_internal, &rdptr, &insize, &wrptr, &outsize);
-  return int_buf;
+  size_t outsize = out_len;
+  char *wrptr = output_buffer;
+  char *rdptr = conv_buf;
+  /* 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 */
+  iconv(cd_to_internal, &rdptr, &conv_buf_size, &wrptr, &outsize);
+  /* 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 output_buffer;
 }
-