Fixed bug #2352: yaz-marcdump crashes for certain record.

[yaz-moved-to-github.git] / src / marcdisp.c

diff --git a/src/marcdisp.c b/src/marcdisp.c
index 33ec6f9..a5084ce 100644 (file)
--- a/src/marcdisp.c
+++ b/src/marcdisp.c
@@ -1,8 +1,6 @@
-/*
- * Copyright (C) 1995-2007, Index Data ApS
+/* This file is part of the YAZ toolkit.
+ * Copyright (C) 1995-2008 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>
@@ -33,6 +32,12 @@
 #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 */
@@ -458,7 +467,7 @@ int yaz_marc_write_line(yaz_marc_t mt, WRBUF wr)
                    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
+                    (identifier_length > 2) ? identifier_length - 1
                     :
                     cdata_one_character(mt, s->code_data);
                 
@@ -494,6 +503,28 @@ 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 +570,18 @@ 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)
@@ -575,7 +616,7 @@ static int yaz_marc_write_marcxml_ns1(yaz_marc_t mt, WRBUF wr,
                    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
+                    (identifier_length > 2) ? identifier_length - 1
                     :
                     cdata_one_character(mt, s->code_data);
                 
@@ -597,7 +638,9 @@ 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 +700,8 @@ 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",
@@ -673,12 +718,12 @@ int yaz_marc_write_marcxchange(yaz_marc_t mt, WRBUF wr,
 }
 
 
+#if YAZ_HAVE_XML2
 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;
@@ -741,7 +786,7 @@ int yaz_marc_write_xml(yaz_marc_t mt, xmlNode **root_ptr,
                    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
+                    (identifier_length > 2) ? identifier_length - 1
                     :
                     cdata_one_character(mt, s->code_data);
 
@@ -783,10 +828,8 @@ int yaz_marc_write_xml(yaz_marc_t mt, xmlNode **root_ptr,
     }
     wrbuf_destroy(wr_cdata);
     return 0;
-#else
-    return -1;
-#endif
 }
+#endif
 
 int yaz_marc_write_iso2709(yaz_marc_t mt, WRBUF wr)
 {
@@ -964,6 +1007,11 @@ 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 +1051,7 @@ 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 == '\'')