added tests for bug #883 'Need an 'ignore' value for the z:type

attribute in the canonical indexing format'
resolved bug #883
tested as well on gutenberg collection
zebra-setup/gutenberg
case closed, see
http://bugzilla.indexdata.dk/show_bug.cgi?id=883

diff --git a/index/mod_dom.c b/index/mod_dom.c
index fca04eb..a2e83c7 100644 (file)
--- a/index/mod_dom.c
+++ b/index/mod_dom.c
@@ -1,4 +1,5 @@
-/* $Id: mod_dom.c,v 1.26 2007-03-03 21:39:10 adam Exp $
+
+/* $Id: mod_dom.c,v 1.27 2007-03-05 13:02:11 marc Exp $
    Copyright (C) 1995-2007
    Index Data ApS
 
@@ -1057,7 +1058,6 @@ static void extract_dom_doc_node(struct filter_info *tinfo,
     RecWord recword;
     (*extctr->init)(extctr, &recword);
 
-    tinfo->record_info_invoked = 0;
     process_xml_element_node(tinfo, extctr, &recword, (xmlNodePtr)doc);
 }
 
@@ -1076,6 +1076,14 @@ static int convert_extract_doc(struct filter_info *tinfo,
     xsltStylesheetPtr last_xsp = 0;
     xmlDocPtr store_doc = 0;
 
+    /* per default do not ingest record */
+    tinfo->record_info_invoked = 0;
+
+    /* exit if empty document given */
+    if (!doc)
+        return RECCTRL_EXTRACT_SKIP;
+
+    /* we actuallu have a document which needs to be processed further */
     params[0] = 0;
     set_param_str(params, "schema", zebra_dom_ns, tinfo->odr_record);
 
@@ -1090,14 +1098,12 @@ static int convert_extract_doc(struct filter_info *tinfo,
                         params, &store_doc, &last_xsp);
     }
     
+    /* saving either store doc or original doc in case no store doc exists */
     if (last_xsp)
         xsltSaveResultToString(&buf_out, &len_out, 
                                store_doc ? store_doc : doc, last_xsp);
     else
         xmlDocDumpMemory(store_doc ? store_doc : doc, &buf_out, &len_out);
-  
-    /* if (p->flagShowRecords)
-       fwrite(buf_out, len_out, 1, stdout); */
 
     (*p->setStoreData)(p, buf_out, len_out);
     xmlFree(buf_out);
@@ -1108,15 +1114,17 @@ static int convert_extract_doc(struct filter_info *tinfo,
     /* extract conversion */
     perform_convert(tinfo, p, tinfo->extract->convert, params, &doc, 0);
 
+
     /* finally, do the indexing */
-    if (doc)
+    if (doc){
         extract_dom_doc_node(tinfo, p, doc);
-
-    if (doc)
        xmlFreeDoc(doc);
-
+    }
+    
+    /* there was nothing to index, so there is no inserted/updated record */
     if (tinfo->record_info_invoked == 0)
         return RECCTRL_EXTRACT_SKIP;
+
     return RECCTRL_EXTRACT_OK;
 }
 
@@ -1145,13 +1153,20 @@ static int extract_xml_split(struct filter_info *tinfo,
     {
         int type = xmlTextReaderNodeType(input->u.xmlreader.reader);
         int depth = xmlTextReaderDepth(input->u.xmlreader.reader);
+
         if (type == XML_READER_TYPE_ELEMENT && 
             input->u.xmlreader.split_level == depth)
         {
-            xmlNodePtr ptr
+            /* per default do not ingest record */
+            tinfo->record_info_invoked = 0;
+
+           xmlNodePtr ptr
                 = xmlTextReaderExpand(input->u.xmlreader.reader);
+
             if (ptr)
-            {
+                {
+                /* we have a new document */
+
                 xmlNodePtr ptr2 = xmlCopyNode(ptr, 1);
                 xmlDocPtr doc = xmlNewDoc((const xmlChar*) "1.0");
                 

diff --git a/test/xslt/Makefile.am b/test/xslt/Makefile.am
index b886d7a..38b6d1f 100644 (file)
--- a/test/xslt/Makefile.am
+++ b/test/xslt/Makefile.am
@@ -1,15 +1,18 @@
-# $Id: Makefile.am,v 1.12 2007-02-18 21:53:22 adam Exp $
+# $Id: Makefile.am,v 1.13 2007-03-05 13:02:11 marc Exp $
 
 check_PROGRAMS = xslt1 xslt2 xslt3 xslt4 xslt5 dom1
 TESTS = $(check_PROGRAMS)
 
-EXTRA_DIST=zebra.cfg zebrastaticrank.cfg \
+EXTRA_DIST= \
+ zebra.cfg zebrastaticrank.cfg zebra-dom.cfg \
  marc-col.xml marc-one.xml marc-col.mrc \
  marc-missing-ns.xml \
  index.xsl id.xsl \
  dom-index-element.xsl  dom-index-pi.xsl \
  marcschema-col.xml marcschema-one.xml snippet.xsl \
- dom-config-col.xml dom-config-one.xml dom-config-marc.xml
+ dom-config-col.xml dom-config-one.xml dom-config-marc.xml \
+ dom-config-skipped.xml
+
 
 xslt1_SOURCES = xslt1.c
 xslt2_SOURCES = xslt2.c

diff --git a/test/xslt/dom-config-skipped.xml b/test/xslt/dom-config-skipped.xml
new file mode 100644 (file)
index 0000000..26caa42
--- /dev/null
+++ b/test/xslt/dom-config-skipped.xml
@@ -0,0 +1,12 @@
+<dom>
+  <!-- $Id: dom-config-skipped.xml,v 1.1 2007-03-05 13:02:11 marc Exp $ -->
+  <extract name="index">
+      <xslt stylesheet="dom-index-skipped.xsl"/>
+  </extract>
+  <retrieve name="F">
+    <xslt stylesheet="id.xsl"/>
+  </retrieve>
+  <input syntax="xml">
+    <xmlreader level="1"/>
+  </input>
+</dom>

diff --git a/test/xslt/dom-index-skipped.xsl b/test/xslt/dom-index-skipped.xsl
new file mode 100644 (file)
index 0000000..1545f0d
--- /dev/null
+++ b/test/xslt/dom-index-skipped.xsl
@@ -0,0 +1,41 @@
+<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
+  xmlns:m="http://www.loc.gov/MARC21/slim"
+  xmlns:z="http://indexdata.com/zebra-2.0"
+  exclude-result-prefixes="m z"
+  version="1.0">
+  <!-- $Id: dom-index-skipped.xsl,v 1.1 2007-03-05 13:02:11 marc Exp $ -->
+  <xsl:output indent="yes" method="xml" version="1.0" encoding="UTF-8"/>
+  
+
+  <xsl:template match="text()"/>
+
+  <!-- skipping exactly first record with controlfield '11224466' -->
+  <xsl:template match="/m:record[normalize-space(m:controlfield[@tag='001']) =
+    '11224466']"/>
+
+  <!-- skipping exactly last record with controlfield '73090924' -->
+  <xsl:template match="/m:record[normalize-space(m:controlfield[@tag='001']) =
+    '73090924']"/>
+
+  <!-- and allowing one specific middle record -->
+  <xsl:template match="/m:record[normalize-space(m:controlfield[@tag='001']) =
+    '11224467']">
+    <z:record z:id="{normalize-space(m:controlfield[@tag='001'])}"
+        z:rank="{normalize-space(m:rank)}">
+      <xsl:apply-templates/>
+    </z:record>
+  </xsl:template>
+
+  <xsl:template match="m:controlfield[@tag='001']">
+    <z:index name="control">
+      <xsl:value-of select="normalize-space(.)"/>
+    </z:index>
+  </xsl:template>
+  
+  <xsl:template match="m:datafield[@tag='245']/m:subfield[@code='a']">
+    <z:index name="title:w title:p title:s any:w">
+      <xsl:value-of select="."/>
+    </z:index>
+  </xsl:template>
+
+</xsl:stylesheet>

diff --git a/test/xslt/dom1.c b/test/xslt/dom1.c
index efa7edc..236d1c5 100644 (file)
--- a/test/xslt/dom1.c
+++ b/test/xslt/dom1.c
@@ -1,4 +1,4 @@
-/* $Id: dom1.c,v 1.1 2007-02-07 12:08:54 adam Exp $
+/* $Id: dom1.c,v 1.2 2007-03-05 13:02:11 marc Exp $
    Copyright (C) 1995-2007
    Index Data ApS
 
@@ -61,6 +61,7 @@ void tst(int argc, char **argv)
     zh = index_some(zs, "dom.bad.xml", "marc-col.xml");
     zebra_close(zh);
     
+    /* testing XMLREADER input with PI stylesheet */ 
     zh = index_some(zs, "dom.dom-config-col.xml", "marc-col.xml");
     YAZ_CHECK(tl_query(zh, "@attr 1=title computer", 3));
     YAZ_CHECK(tl_query(zh, "@attr 1=control 11224466", 1));
@@ -68,6 +69,7 @@ void tst(int argc, char **argv)
     YAZ_CHECK(tl_query_x(zh, "@attr 1=4 computer", 0, 121));
     zebra_close(zh);
 
+    /* testing XMLREADER input with ELEMENT stylesheet */ 
     zh = index_some(zs, "dom.dom-config-one.xml", "marc-one.xml");
     YAZ_CHECK(tl_query(zh, "@attr 1=title computer", 1));
     YAZ_CHECK(tl_query(zh, "@attr 1=control 11224466", 1));
@@ -75,6 +77,7 @@ void tst(int argc, char **argv)
     YAZ_CHECK(tl_query_x(zh, "@attr 1=4 computer", 0, 121));
     zebra_close(zh);
 
+    /* testing MARC input with ELEMENT stylesheet */ 
     zh = index_some(zs, "dom.dom-config-marc.xml", "marc-col.mrc");
     YAZ_CHECK(tl_query(zh, "@attr 1=title computer", 3));
     YAZ_CHECK(tl_query(zh, "@attr 1=control 11224466", 1));
@@ -82,6 +85,16 @@ void tst(int argc, char **argv)
     YAZ_CHECK(tl_query_x(zh, "@attr 1=4 computer", 0, 121));
     zebra_close(zh);
 
+    /* testing XMLREADER input with ELEMENT stylesheet and skipped records */ 
+    zh = index_some(zs, "dom.dom-config-skipped.xml", "marc-col.xml");
+    YAZ_CHECK(tl_query(zh, "@attr 1=title computer", 1));
+    YAZ_CHECK(tl_query(zh, "@attr 1=control 11224466", 0));
+    YAZ_CHECK(tl_query(zh, "@attr 1=control 11224467", 1));
+    YAZ_CHECK(tl_query(zh, "@attr 1=control 73090924", 0));
+
+    zebra_close(zh);
+
+
     YAZ_CHECK(tl_close_down(0, zs));
 }
 

diff --git a/test/xslt/marc-col.xml b/test/xslt/marc-col.xml
index 71e8a77..84b6f67 100644 (file)
--- a/test/xslt/marc-col.xml
+++ b/test/xslt/marc-col.xml
@@ -71,12 +71,12 @@
 <record xmlns="http://www.loc.gov/MARC21/slim">
   <rank>3</rank>
   <leader>01369cam  2200265 i 4500</leader>
-  <controlfield tag="001">   73090924 //r82  </controlfield>
+  <controlfield tag="001">   73090924  </controlfield>
   <controlfield tag="003">DLC  </controlfield>
   <controlfield tag="005">19820524000000.0  </controlfield>
   <controlfield tag="008">760609s1974    nyua     b    10110 eng    </controlfield>
   <datafield tag="010" ind1=" " ind2=" ">
-    <subfield code="a">   73090924 //r82</subfield>
+    <subfield code="a">   73090924 </subfield>
   </datafield>
   <datafield tag="040" ind1=" " ind2=" ">
     <subfield code="a">DLC</subfield>

diff --git a/test/xslt/zebra-dom.cfg b/test/xslt/zebra-dom.cfg
new file mode 100644 (file)
index 0000000..a912991
--- /dev/null
+++ b/test/xslt/zebra-dom.cfg
@@ -0,0 +1,10 @@
+profilePath: ${srcdir:-.}:${srcdir:-.}/../../tab
+
+modulePath: ../../index/.libs
+
+recordType: dom.dom-config-skipped.xml
+#recordType: dom.dom-config-col.xml
+
+perm.anonymous: rw
+
+