Better interface for encoding.

Bugfix in use of iconv: there can be characters left...

diff --git a/encoding.c b/encoding.c
index ba5aa1d73395778815776000c01a48889aff7b46..6c342e0ccba4fe2817eab3f553287efe89a47c72 100644 (file)
--- a/encoding.c
+++ b/encoding.c
@@ -12,25 +12,29 @@
 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(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)) {
@@ -39,11 +43,13 @@ 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[MAXBUF + 1];
+  sprintf(buffer, "%s(%s)", gedcom_n, charwidth_string[enc]);
+  search_node.gedcom_name = buffer;
   datum = tfind(&search_node, &encoding_mapping, node_compare);
   if (datum == NULL) {
     gedcom_error("No encoding found for '%s'", gedcom_n);
@@ -60,6 +66,7 @@ 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) {
@@ -70,8 +77,13 @@ void init_encodings()
          return;
        }
        else if (buffer[0] != '#') {
-         if (sscanf(buffer, "%s %s", gedcom_n, iconv_n) == 2) {
-           add_encoding(gedcom_n, iconv_n);
+         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;
          }
        }
       }
@@ -84,15 +96,25 @@ void init_encodings()
   }
 }
 
+void set_encoding_width(ENCODING enc)
+{
+  the_enc = enc;
+}
+
+static char conv_buf[MAXGEDCLINELEN * 2];
+static int 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);
   }
   return (cd_to_internal != (iconv_t) -1);  
@@ -106,12 +128,23 @@ void close_conv_to_internal()
 
 char* to_internal(char* str, size_t len)
 {
-  size_t insize = len;
+  size_t insize;
   size_t outsize = MAXGEDCLINELEN * 2;
   char *wrptr = int_buf;
-  char *rdptr = str;
+  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;
+  insize = conv_buf_size;
+  /* set up output buffer (empty it) */
   memset(int_buf, 0, sizeof(int_buf));
+  /* do the conversion */
   iconv(cd_to_internal, &rdptr, &insize, &wrptr, &outsize);
+  /* then shift what is left over to the head of the input buffer */
+  memmove(conv_buf, rdptr, insize);
+  memset(conv_buf + insize, 0, sizeof(conv_buf) - insize);
+  conv_buf_size = insize;
   return int_buf;
 }