Working in progress: refactor the xml_write to handle two formats
[yaz-moved-to-github.git] / src / marcdisp.c
index 33ec6f9..811b84d 100644 (file)
@@ -1,8 +1,6 @@
-/*
- * Copyright (C) 1995-2007, Index Data ApS
+/* This file is part of the YAZ toolkit.
+ * Copyright (C) 1995-2010 Index Data
  * See the file LICENSE for details.
- *
- * $Id: marcdisp.c,v 1.49 2007-03-20 21:37:32 adam Exp $
  */
 
 /**
@@ -20,6 +18,7 @@
 
 #include <stdarg.h>
 #include <stdio.h>
+#include <stdlib.h>
 #include <string.h>
 #include <ctype.h>
 #include <yaz/marcdisp.h>
 #include <libxml/tree.h>
 #endif
 
+enum yaz_collection_state {
+    no_collection,
+    collection_first,
+    collection_second
+};
+   
 /** \brief node types for yaz_marc_node */
 enum YAZ_MARC_NODE_TYPE
 { 
@@ -85,6 +90,7 @@ struct yaz_marc_t_ {
     int xml;
     int debug;
     int write_using_libxml2;
+    enum yaz_collection_state enable_collection;
     yaz_iconv_t iconv_cd;
     char subfield_str[8];
     char endline_str[8];
@@ -100,6 +106,7 @@ yaz_marc_t yaz_marc_create(void)
     mt->xml = YAZ_MARC_LINE;
     mt->debug = 0;
     mt->write_using_libxml2 = 0;
+    mt->enable_collection = no_collection;
     mt->m_wr = wrbuf_alloc();
     mt->iconv_cd = 0;
     mt->leader_spec = 0;
@@ -137,7 +144,8 @@ static int marc_exec_leader(const char *leader_spec, char *leader,
 
 static struct yaz_marc_node *yaz_marc_add_node(yaz_marc_t mt)
 {
-    struct yaz_marc_node *n = nmem_malloc(mt->nmem, sizeof(*n));
+    struct yaz_marc_node *n = (struct yaz_marc_node *)
+        nmem_malloc(mt->nmem, sizeof(*n));
     n->next = 0;
     *mt->nodes_pp = n;
     mt->nodes_pp = &n->next;
@@ -256,7 +264,8 @@ void yaz_marc_add_subfield(yaz_marc_t mt,
 
     if (mt->subfield_pp)
     {
-        struct yaz_marc_subfield *n = nmem_malloc(mt->nmem, sizeof(*n));
+        struct yaz_marc_subfield *n = (struct yaz_marc_subfield *)
+            nmem_malloc(mt->nmem, sizeof(*n));
         n->code_data = nmem_strdupn(mt->nmem, code_data, code_data_len);
         n->next = 0;
         /* mark subfield_pp to point to this one, so we append here next */
@@ -265,16 +274,6 @@ void yaz_marc_add_subfield(yaz_marc_t mt,
     }
 }
 
-int atoi_n_check(const char *buf, int size, int *val)
-{
-    int i;
-    for (i = 0; i < size; i++)
-        if (!isdigit(i[(const unsigned char *) buf]))
-            return 0;
-    *val = atoi_n(buf, size);
-    return 1;
-}
-
 void yaz_marc_set_leader(yaz_marc_t mt, const char *leader_c,
                          int *indicator_length,
                          int *identifier_length,
@@ -416,7 +415,7 @@ int yaz_marc_write_check(yaz_marc_t mt, WRBUF wr)
         case YAZ_MARC_COMMENT:
             wrbuf_iconv_write(wr, mt->iconv_cd, 
                               n->u.comment, strlen(n->u.comment));
-            wrbuf_puts(wr, ")\n");
+            wrbuf_puts(wr, "\n");
             break;
         default:
             break;
@@ -425,6 +424,17 @@ int yaz_marc_write_check(yaz_marc_t mt, WRBUF wr)
     return 0;
 }
 
+static size_t get_subfield_len(yaz_marc_t mt, const char *data,
+                               int identifier_length)
+{
+    /* if identifier length is 2 (most MARCs) or less (probably an error),
+       the code is a single character .. However we've
+       seen multibyte codes, so see how big it really is */
+    if (identifier_length > 2)
+        return identifier_length - 1;
+    else
+        return cdata_one_character(mt, data);
+}
 
 int yaz_marc_write_line(yaz_marc_t mt, WRBUF wr)
 {
@@ -454,13 +464,8 @@ int yaz_marc_write_line(yaz_marc_t mt, WRBUF wr)
                          n->u.datafield.indicator);
             for (s = n->u.datafield.subfields; s; s = s->next)
             {
-                /* if identifier length is 2 (most MARCs),
-                   the code is a single character .. However we've
-                   seen multibyte codes, so see how big it really is */
-                size_t using_code_len = 
-                    (identifier_length != 2) ? identifier_length - 1
-                    :
-                    cdata_one_character(mt, s->code_data);
+                size_t using_code_len = get_subfield_len(mt, s->code_data,
+                                                         identifier_length);
                 
                 wrbuf_puts (wr, mt->subfield_str); 
                 wrbuf_iconv_write(wr, mt->iconv_cd, s->code_data, 
@@ -494,6 +499,28 @@ int yaz_marc_write_line(yaz_marc_t mt, WRBUF wr)
     return 0;
 }
 
+int yaz_marc_write_trailer(yaz_marc_t mt, WRBUF wr)
+{
+    if (mt->enable_collection == collection_second)
+    {
+        switch(mt->xml)
+        {
+        case YAZ_MARC_MARCXML:
+            wrbuf_printf(wr, "</collection>\n");
+            break;
+        case YAZ_MARC_XCHANGE:
+            wrbuf_printf(wr, "</collection>\n");
+            break;
+        }
+    }
+    return 0;
+}
+
+void yaz_marc_enable_collection(yaz_marc_t mt)
+{
+    mt->enable_collection = collection_first;
+}
+
 int yaz_marc_write_mode(yaz_marc_t mt, WRBUF wr)
 {
     switch(mt->xml)
@@ -539,8 +566,18 @@ static int yaz_marc_write_marcxml_ns1(yaz_marc_t mt, WRBUF wr,
         return -1;
     if (!atoi_n_check(leader+11, 1, &identifier_length))
         return -1;
-
-    wrbuf_printf(wr, "<record xmlns=\"%s\"", ns);
+    
+    if (mt->enable_collection != no_collection)
+    {
+        if (mt->enable_collection == collection_first)
+            wrbuf_printf(wr, "<collection xmlns=\"%s\">\n", ns);
+        mt->enable_collection = collection_second;
+        wrbuf_printf(wr, "<record");
+    }
+    else
+    {
+        wrbuf_printf(wr, "<record xmlns=\"%s\"", ns);
+    }
     if (format)
         wrbuf_printf(wr, " format=\"%.80s\"", format);
     if (type)
@@ -571,14 +608,8 @@ static int yaz_marc_write_marcxml_ns1(yaz_marc_t mt, WRBUF wr,
             wrbuf_printf(wr, ">\n");
             for (s = n->u.datafield.subfields; s; s = s->next)
             {
-                /* if identifier length is 2 (most MARCs),
-                   the code is a single character .. However we've
-                   seen multibyte codes, so see how big it really is */
-                size_t using_code_len = 
-                    (identifier_length != 2) ? identifier_length - 1
-                    :
-                    cdata_one_character(mt, s->code_data);
-                
+                size_t using_code_len = get_subfield_len(mt, s->code_data,
+                                                         identifier_length);
                 wrbuf_iconv_puts(wr, mt->iconv_cd, "    <subfield code=\"");
                 wrbuf_iconv_write_cdata(wr, mt->iconv_cd,
                                         s->code_data, using_code_len);
@@ -597,7 +628,9 @@ static int yaz_marc_write_marcxml_ns1(yaz_marc_t mt, WRBUF wr,
             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_puts(wr, mt->iconv_cd, n->u.controlfield.data);
+            wrbuf_iconv_write_cdata(wr, mt->iconv_cd,
+                                    n->u.controlfield.data,
+                                    strlen(n->u.controlfield.data));
 
             marc_iconv_reset(mt, wr);
             wrbuf_iconv_puts(wr, mt->iconv_cd, "</controlfield>");
@@ -657,6 +690,8 @@ static int yaz_marc_write_marcxml_ns(yaz_marc_t mt, WRBUF wr,
 
 int yaz_marc_write_marcxml(yaz_marc_t mt, WRBUF wr)
 {
+    /* set leader 09 to 'a' for UNICODE */
+    /* http://www.loc.gov/marc/bibliographic/ecbdldrd.html#mrcblea */
     if (!mt->leader_spec)
         yaz_marc_modify_leader(mt, 9, "a");
     return yaz_marc_write_marcxml_ns(mt, wr, "http://www.loc.gov/MARC21/slim",
@@ -668,17 +703,64 @@ int yaz_marc_write_marcxchange(yaz_marc_t mt, WRBUF wr,
                                const char *type)
 {
     return yaz_marc_write_marcxml_ns(mt, wr,
-                                     "http://www.bs.dk/standards/MarcXchange",
+                                     "info:lc/xmlns/marcxchange-v1",
                                      0, 0);
 }
 
+#if YAZ_HAVE_XML2
+
+void addMarcDatafield(xmlNode *record_ptr, xmlNameSpace *ns_record, const char* datafield, int turbo, WRBUF wr_cdata) 
+{
+    if (!turbo) {
+        ptr = xmlNewChild(record_ptr, ns_record, BAD_CAST "datafield", 0);
+        xmlNewProp(ptr, BAD_CAST "tag", BAD_CAST n->u.datafield.tag);
+    }
+    else {
+        char *field = "datXXX";
+        sprintf(field +3,"%s",  n->u.datafield.tag);
+        ptr = xmlNewChild(record_ptr, ns_record, BAD_CAST "dat", 0);
+        xmlNewProp(ptr, BAD_CAST "tag", BAD_CAST n->u.datafield.tag);        
+    }
+    if (n->u.datafield.indicator)
+    {
+        int i;
+        for (i = 0; n->u.datafield.indicator[i]; i++)
+        {
+            char ind_str[6];
+            char ind_val[2];
+            
+            sprintf(ind_str, "ind%d", i+1);
+            ind_val[0] = n->u.datafield.indicator[i];
+            ind_val[1] = '\0';
+            xmlNewProp(ptr, BAD_CAST ind_str, BAD_CAST ind_val);
+        }
+    }
+    for (s = n->u.datafield.subfields; s; s = s->next)
+    {
+        xmlNode *ptr_subfield;
+        size_t using_code_len = get_subfield_len(mt, s->code_data,
+                                                 identifier_length);
+        wrbuf_rewind(wr_cdata);
+        wrbuf_iconv_puts(wr_cdata, mt->iconv_cd,
+                         s->code_data + using_code_len);
+        marc_iconv_reset(mt, wr_cdata);
+        ptr_subfield = xmlNewTextChild(
+            ptr, ns_record, 
+            BAD_CAST "subfield",  BAD_CAST wrbuf_cstr(wr_cdata));
+        
+        wrbuf_rewind(wr_cdata);
+        wrbuf_iconv_write(wr_cdata, mt->iconv_cd,
+                          s->code_data, using_code_len);
+        xmlNewProp(ptr_subfield, BAD_CAST "code",
+                   BAD_CAST wrbuf_cstr(wr_cdata));
+    }
+}
 
 int yaz_marc_write_xml(yaz_marc_t mt, xmlNode **root_ptr,
                        const char *ns, 
                        const char *format,
                        const char *type)
 {
-#if YAZ_HAVE_XML2
     struct yaz_marc_node *n;
     int identifier_length;
     const char *leader = 0;
@@ -737,14 +819,8 @@ int yaz_marc_write_xml(yaz_marc_t mt, xmlNode **root_ptr,
             for (s = n->u.datafield.subfields; s; s = s->next)
             {
                 xmlNode *ptr_subfield;
-                /* if identifier length is 2 (most MARCs),
-                   the code is a single character .. However we've
-                   seen multibyte codes, so see how big it really is */
-                size_t using_code_len = 
-                    (identifier_length != 2) ? identifier_length - 1
-                    :
-                    cdata_one_character(mt, s->code_data);
-
+                size_t using_code_len = get_subfield_len(mt, s->code_data,
+                                                         identifier_length);
                 wrbuf_rewind(wr_cdata);
                 wrbuf_iconv_puts(wr_cdata, mt->iconv_cd,
                                  s->code_data + using_code_len);
@@ -783,11 +859,84 @@ int yaz_marc_write_xml(yaz_marc_t mt, xmlNode **root_ptr,
     }
     wrbuf_destroy(wr_cdata);
     return 0;
-#else
-    return -1;
-#endif
 }
 
+
+int yaz_marc_write_turbo_xml(yaz_marc_t mt, xmlNode **root_ptr,
+                       const char *ns, 
+                       const char *format,
+                       const char *type)
+{
+    struct yaz_marc_node *n;
+    int identifier_length;
+    const char *leader = 0;
+    xmlNode *record_ptr;
+    xmlNsPtr ns_record;
+    WRBUF wr_cdata = 0;
+
+    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;
+
+    wr_cdata = wrbuf_alloc();
+
+    record_ptr = xmlNewNode(0, BAD_CAST "record");
+    *root_ptr = record_ptr;
+
+    ns_record = xmlNewNs(record_ptr, BAD_CAST ns, 0);
+    xmlSetNs(record_ptr, ns_record);
+
+    if (format)
+        xmlNewProp(record_ptr, BAD_CAST "format", BAD_CAST format);
+    if (type)
+        xmlNewProp(record_ptr, BAD_CAST "type", BAD_CAST type);
+    for (n = mt->nodes; n; n = n->next)
+    {
+        struct yaz_marc_subfield *s;
+        xmlNode *ptr;
+
+        switch(n->which)
+        {
+        case YAZ_MARC_DATAFIELD:
+
+            addMarcDatafield(record_ptr, ns_record, datafield, turbo, wr_cdata);
+            break;
+        case YAZ_MARC_CONTROLFIELD:
+            wrbuf_rewind(wr_cdata);
+            wrbuf_iconv_puts(wr_cdata, mt->iconv_cd, n->u.controlfield.data);
+            marc_iconv_reset(mt, wr_cdata);
+            
+            ptr = xmlNewTextChild(record_ptr, ns_record,
+                                  BAD_CAST "controlfield",
+                                  BAD_CAST wrbuf_cstr(wr_cdata));
+            
+            xmlNewProp(ptr, BAD_CAST "tag", BAD_CAST n->u.controlfield.tag);
+            break;
+        case YAZ_MARC_COMMENT:
+            ptr = xmlNewComment(BAD_CAST n->u.comment);
+            xmlAddChild(record_ptr, ptr);
+            break;
+        case YAZ_MARC_LEADER:
+            xmlNewTextChild(record_ptr, ns_record, BAD_CAST "leader",
+                            BAD_CAST n->u.leader);
+            break;
+        }
+    }
+    wrbuf_destroy(wr_cdata);
+    return 0;
+}
+
+
+#endif
+
 int yaz_marc_write_iso2709(yaz_marc_t mt, WRBUF wr)
 {
     struct yaz_marc_node *n;
@@ -964,6 +1113,11 @@ void yaz_marc_iconv(yaz_marc_t mt, yaz_iconv_t cd)
     mt->iconv_cd = cd;
 }
 
+yaz_iconv_t yaz_marc_get_iconv(yaz_marc_t mt)
+{
+    return mt->iconv_cd;
+}
+
 void yaz_marc_modify_leader(yaz_marc_t mt, size_t off, const char *str)
 {
     struct yaz_marc_node *n;
@@ -1003,7 +1157,7 @@ static int marc_exec_leader(const char *leader_spec, char *leader, size_t size)
         no = sscanf(cp, "%d=%20[^,]%n", &pos, val, &no_read);
         if (no < 2 || no_read < 3)
             return -1;
-        if (pos < 0 || pos >= size)
+        if (pos < 0 || (size_t) pos >= size)
             return -1;
 
         if (*val == '\'')
@@ -1056,6 +1210,7 @@ void yaz_marc_write_using_libxml2(yaz_marc_t mt, int enable)
 /*
  * Local variables:
  * c-basic-offset: 4
+ * c-file-style: "Stroustrup"
  * indent-tabs-mode: nil
  * End:
  * vim: shiftwidth=4 tabstop=8 expandtab