Extracted element name encoding into function. Moved indicators back as attributes...
[yaz-moved-to-github.git] / src / marcdisp.c
index 75e6d3c..53d4e81 100644 (file)
@@ -91,7 +91,6 @@ struct yaz_marc_t_ {
     int output_format;
     int debug;
     int write_using_libxml2;
-    int turbo_format;
     enum yaz_collection_state enable_collection;
     yaz_iconv_t iconv_cd;
     char subfield_str[8];
@@ -164,7 +163,7 @@ void yaz_marc_add_controlfield_xml(yaz_marc_t mt, const xmlNode *ptr_tag,
     n->u.controlfield.data = nmem_text_node_cdata(ptr_data, mt->nmem);
 }
 
-void yaz_marc_add_controlfield_turbo_xml(yaz_marc_t mt, const char *tag,
+void yaz_marc_add_controlfield_turbo_xml(yaz_marc_t mt, char *tag,
                                    const xmlNode *ptr_data)
 {
     struct yaz_marc_node *n = yaz_marc_add_node(mt);
@@ -242,6 +241,33 @@ void yaz_marc_add_datafield(yaz_marc_t mt, const char *tag,
     mt->subfield_pp = &n->u.datafield.subfields;
 }
 
+char *element_name_encode(yaz_marc_t mt, WRBUF buffer, char *code_data, size_t code_len) {
+       // TODO Map special codes to something possible for XML ELEMENT names
+
+       int encode = 0;
+       int index = 0;
+       for (index = 0; index < code_len; index++) {
+               if (!((code_data[index] >= '0' && code_data[index] <= '9') ||
+                         (code_data[index] >= 'a' && code_data[index] <= 'z') ||
+                         (code_data[index] >= 'A' && code_data[index] <= 'Z')))
+                       encode = 1;
+       }
+       if (!encode) {
+               wrbuf_iconv_write(buffer, mt->iconv_cd, code_data, code_len);
+       }
+       else {
+               char temp[2*code_len + 1];
+               wrbuf_puts(buffer, "-");
+               int index;
+               for (index = 0; index < code_len; index++) {
+                       sprintf(temp, "%02X", (unsigned char) code_data[index] & 0xFF);
+                       temp[2] = 0;
+                       wrbuf_puts(buffer, temp);
+               };
+               yaz_log(YLOG_WARN, "Using numeric value in element name: %s", wrbuf_cstr(buffer));
+       }
+}
+
 #if YAZ_HAVE_XML2
 void yaz_marc_add_datafield_xml(yaz_marc_t mt, const xmlNode *ptr_tag,
                                 const char *indicator, size_t indicator_len)
@@ -572,6 +598,15 @@ int yaz_marc_write_mode(yaz_marc_t mt, WRBUF wr)
     return -1;
 }
 
+const char *collection_name[2]  = { "collection", "collection"};
+const char *record_name[2]     = { "record", "r"};
+const char *leader_name[2]     = { "leader", "l"};
+const char *controlfield_name[2]= { "controlfield", "c"};
+const char *datafield_name[2]          = { "datafield", "d"};
+const char *indicator_name[2]          = { "ind", "i"};
+const char *subfield_name[2]   = { "subfield", "s"};
+
+
 /** \brief common MARC XML/Xchange writer
     \param mt handle
     \param wr WRBUF output
@@ -588,6 +623,146 @@ static int yaz_marc_write_marcxml_ns1(yaz_marc_t mt, WRBUF wr,
     int identifier_length;
     const char *leader = 0;
 
+    int turbo = yaz_marc_get_write_format(mt) == YAZ_MARC_TMARCXML;
+
+    for (n = mt->nodes; n; n = n->next)
+        if (n->which == YAZ_MARC_LEADER)
+        {
+            leader = n->u.leader;
+            break;
+        }
+    
+    if (!leader)
+        return -1;
+    if (!atoi_n_check(leader+11, 1, &identifier_length))
+        return -1;
+    
+    if (mt->enable_collection != no_collection)
+    {
+        if (mt->enable_collection == collection_first) {
+            wrbuf_printf(wr, "<%s xmlns=\"%s\">\n", collection_name[turbo], ns);
+            mt->enable_collection = collection_second;
+        }
+        wrbuf_printf(wr, "<%s", record_name[turbo]);
+    }
+    else
+    {
+        wrbuf_printf(wr, "<%s xmlns=\"%s\"", record_name[turbo], ns);
+    }
+    if (format)
+        wrbuf_printf(wr, " format=\"%.80s\"", format);
+    if (type)
+        wrbuf_printf(wr, " type=\"%.80s\"", type);
+    wrbuf_printf(wr, ">\n");
+    for (n = mt->nodes; n; n = n->next)
+    {
+        struct yaz_marc_subfield *s;
+
+        switch(n->which)
+        {
+        case YAZ_MARC_DATAFIELD:
+
+               wrbuf_printf(wr, "  <%s", datafield_name[turbo]);
+            if (!turbo)
+               wrbuf_printf(wr, " tag=\"");
+            wrbuf_iconv_write_cdata(wr, mt->iconv_cd, n->u.datafield.tag,
+                                    strlen(n->u.datafield.tag));
+               if (!turbo)
+                       wrbuf_printf(wr, "\"");
+           if (n->u.datafield.indicator)
+           {
+               int i;
+               for (i = 0; n->u.datafield.indicator[i]; i++)
+               {
+                       wrbuf_printf(wr, " %s%d=\"", indicator_name[turbo], i+1);
+                       wrbuf_iconv_write_cdata(wr, mt->iconv_cd,
+                                       n->u.datafield.indicator+i, 1);
+                       wrbuf_iconv_puts(wr, mt->iconv_cd, "\"");
+                               }
+            }
+               wrbuf_printf(wr, ">\n");
+            for (s = n->u.datafield.subfields; s; s = s->next)
+            {
+                size_t using_code_len = get_subfield_len(mt, s->code_data,
+                                                         identifier_length);
+                wrbuf_printf(wr, "    <%s", subfield_name[turbo]);
+                               if (!turbo) {
+                                       wrbuf_printf(wr, " code=\"");
+                       wrbuf_iconv_write_cdata(wr, mt->iconv_cd,
+                                        s->code_data, using_code_len);
+                       wrbuf_iconv_puts(wr, mt->iconv_cd, "\">");
+                               } else {
+                               element_name_encode(mt, wr, s->code_data, using_code_len);
+                                       wrbuf_puts(wr, ">");
+                               }
+                wrbuf_iconv_write_cdata(wr, mt->iconv_cd,
+                                        s->code_data + using_code_len,
+                                        strlen(s->code_data + using_code_len));
+                marc_iconv_reset(mt, wr);
+                               wrbuf_printf(wr, "</%s", subfield_name[turbo]);
+               if (turbo)
+                       wrbuf_iconv_write_cdata(wr, mt->iconv_cd,
+                                       s->code_data, using_code_len);
+                wrbuf_puts(wr, ">\n");
+            }
+            wrbuf_printf(wr, "  </%s", datafield_name[turbo]);
+               //TODO Not CDATA
+            if (turbo)
+               wrbuf_iconv_write_cdata(wr, mt->iconv_cd, n->u.datafield.tag,
+                               strlen(n->u.datafield.tag));
+            wrbuf_printf(wr, ">\n", datafield_name[turbo]);
+            break;
+        case YAZ_MARC_CONTROLFIELD:
+               wrbuf_printf(wr, "  <%s", controlfield_name[turbo]);
+               if (!turbo) {
+               wrbuf_printf(wr, " tag=\"");
+                       wrbuf_iconv_write_cdata(wr, mt->iconv_cd, n->u.controlfield.tag,
+                                       strlen(n->u.controlfield.tag));
+                       wrbuf_iconv_puts(wr, mt->iconv_cd, "\">");
+               }
+               else {
+                       //TODO convert special
+                       wrbuf_iconv_write_cdata(wr, mt->iconv_cd, n->u.controlfield.tag,
+                                       strlen(n->u.controlfield.tag));
+                       wrbuf_iconv_puts(wr, mt->iconv_cd, ">");
+               }
+               wrbuf_iconv_write_cdata(wr, mt->iconv_cd,
+                                                                       n->u.controlfield.data,
+                                                                       strlen(n->u.controlfield.data));
+               marc_iconv_reset(mt, wr);
+               wrbuf_printf(wr, "</%s", controlfield_name[turbo]);
+               //TODO convert special
+               if (turbo)
+                       wrbuf_iconv_write_cdata(wr, mt->iconv_cd, n->u.controlfield.tag,
+                                       strlen(n->u.controlfield.tag));
+               wrbuf_puts(wr, ">\n");
+            break;
+        case YAZ_MARC_COMMENT:
+            wrbuf_printf(wr, "<!-- ");
+            wrbuf_puts(wr, n->u.comment);
+            wrbuf_printf(wr, " -->\n");
+            break;
+        case YAZ_MARC_LEADER:
+            wrbuf_printf(wr, "  <%s>", leader_name[turbo]);
+            wrbuf_iconv_write_cdata(wr, 
+                                    0 /* no charset conversion for leader */,
+                                    n->u.leader, strlen(n->u.leader));
+            wrbuf_printf(wr, "</%s>\n", leader_name[turbo]);
+        }
+    }
+    wrbuf_printf(wr, "</%s>\n", record_name[turbo]);
+    return 0;
+}
+
+static int yaz_marc_write_marcxml_ns2(yaz_marc_t mt, WRBUF wr,
+                                      const char *ns, 
+                                      const char *format,
+                                      const char *type)
+{
+    struct yaz_marc_node *n;
+    int identifier_length;
+    const char *leader = 0;
+
     for (n = mt->nodes; n; n = n->next)
         if (n->which == YAZ_MARC_LEADER)
         {
@@ -686,6 +861,7 @@ static int yaz_marc_write_marcxml_ns1(yaz_marc_t mt, WRBUF wr,
     return 0;
 }
 
+
 static int yaz_marc_write_marcxml_ns(yaz_marc_t mt, WRBUF wr,
                                      const char *ns, 
                                      const char *format,
@@ -697,9 +873,9 @@ static int yaz_marc_write_marcxml_ns(yaz_marc_t mt, WRBUF wr,
         int ret;
         xmlNode *root_ptr;
 
-        if (!mt->turbo_format)
+        if (yaz_marc_get_write_format(mt) == YAZ_MARC_MARCXML)
                ret = yaz_marc_write_xml(mt, &root_ptr, ns, format, type);
-        else
+        else // Check for Turbo XML
                ret = yaz_marc_write_turbo_xml(mt, &root_ptr, ns, format, type);
         if (ret == 0)
         {
@@ -752,7 +928,7 @@ void add_marc_datafield_turbo_xml(yaz_marc_t mt, struct yaz_marc_node *n, xmlNod
 {
     xmlNode *ptr;
     struct yaz_marc_subfield *s;
-    int turbo = mt->turbo_format;
+    int turbo = mt->output_format == YAZ_MARC_TMARCXML;
     if (!turbo) {
         ptr = xmlNewChild(record_ptr, ns_record, BAD_CAST "datafield", 0);
         xmlNewProp(ptr, BAD_CAST "tag", BAD_CAST n->u.datafield.tag);
@@ -775,14 +951,8 @@ void add_marc_datafield_turbo_xml(yaz_marc_t mt, struct yaz_marc_node *n, xmlNod
             
             ind_val[0] = n->u.datafield.indicator[i];
             ind_val[1] = '\0';
-            if (!turbo) {
-                sprintf(ind_str, "ind%d", i+1);
-               xmlNewProp(ptr, BAD_CAST ind_str, BAD_CAST ind_val);
-            }
-            else {
-                sprintf(ind_str, "i%d", i+1);
-               xmlNewTextChild(ptr, ns_record, BAD_CAST ind_str, BAD_CAST ind_val);
-            }
+            sprintf(ind_str, "%s%d", indicator_name[turbo], i+1);
+                       xmlNewProp(ptr, BAD_CAST ind_str, BAD_CAST ind_val);
         }
     }
        WRBUF subfield_name = wrbuf_alloc();
@@ -807,24 +977,7 @@ void add_marc_datafield_turbo_xml(yaz_marc_t mt, struct yaz_marc_node *n, xmlNod
         else { // Turbo format
                wrbuf_rewind(subfield_name);
                wrbuf_puts(subfield_name, "s");
-               // TODO Map special codes to something possible for XML ELEMENT names
-               if ((s->code_data[0] >= '0' && s->code_data[0] <= '9') ||
-                   (s->code_data[0] >= 'a' && s->code_data[0] <= 'z') ||
-                               (s->code_data[0] >= 'A' && s->code_data[0] <= 'Z'))
-               {
-                       wrbuf_iconv_write(subfield_name, mt->iconv_cd,s->code_data, using_code_len);
-               }
-               else {
-                               char buffer[2*using_code_len + 1];
-                               int index;
-                               for (index = 0; index < using_code_len; index++) {
-                                       sprintf(buffer + 2*index, "%02X", (unsigned char) s->code_data[index] & 0xFF);
-                               };
-                               buffer[2*(index+1)] = 0;
-                               wrbuf_puts(subfield_name, "-");
-                               wrbuf_puts(subfield_name, buffer);
-                       yaz_log(YLOG_WARN, "Using numeric value in element name: %s", buffer);
-               }
+               element_name_encode(mt, subfield_name, s->code_data, using_code_len);
                ptr_subfield = xmlNewTextChild(ptr, ns_record,
                                BAD_CAST wrbuf_cstr(subfield_name),
                                BAD_CAST wrbuf_cstr(wr_cdata));
@@ -844,7 +997,7 @@ int yaz_marc_write_turbo_xml(yaz_marc_t mt, xmlNode **root_ptr,
     xmlNode *record_ptr;
     xmlNsPtr ns_record;
     WRBUF wr_cdata = 0;
-    int turbo = mt->turbo_format;
+    int turbo = mt->output_format == YAZ_MARC_TMARCXML;
     for (n = mt->nodes; n; n = n->next)
         if (n->which == YAZ_MARC_LEADER)
         {
@@ -859,7 +1012,7 @@ int yaz_marc_write_turbo_xml(yaz_marc_t mt, xmlNode **root_ptr,
 
     wr_cdata = wrbuf_alloc();
 
-    record_ptr = xmlNewNode(0, BAD_CAST "record");
+    record_ptr = xmlNewNode(0, BAD_CAST "r");
     *root_ptr = record_ptr;
 
     ns_record = xmlNewNs(record_ptr, BAD_CAST ns, 0);
@@ -1207,8 +1360,14 @@ int yaz_marc_get_read_format(yaz_marc_t mt)
 
 void yaz_marc_set_write_format(yaz_marc_t mt, int format)
 {
-    if (mt)
+    if (mt) {
         mt->output_format = format;
+/*
+        // Force using libxml2
+        if (mt->output_format == YAZ_MARC_TMARCXML)
+               mt->write_using_libxml2 = 1;
+*/
+    }
 }
 
 int yaz_marc_get_write_format(yaz_marc_t mt)
@@ -1224,8 +1383,7 @@ int yaz_marc_get_write_format(yaz_marc_t mt)
  */
 void yaz_marc_xml(yaz_marc_t mt, int xmlmode)
 {
-    if (mt)
-        mt->output_format = xmlmode;
+       yaz_marc_set_write_format(mt, xmlmode);
 }
 
 
@@ -1337,14 +1495,9 @@ void yaz_marc_write_using_libxml2(yaz_marc_t mt, int enable)
     mt->write_using_libxml2 = enable;
 }
 
-void yaz_marc_write_turbo_format(yaz_marc_t mt, int enable)
-{
-    mt->turbo_format = enable;
-}
-
 int yaz_marc_is_turbo_format(yaz_marc_t mt)
 {
-    return mt->turbo_format;
+    return mt->output_format == YAZ_MARC_TMARCXML;
 }