Ups... memory leak
[yaz-moved-to-github.git] / src / marc_read_xml.c
index 31a718b..6b3e3dd 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: marc_read_xml.c,v 1.4 2007-12-17 20:59:30 adam Exp $
  */
 
 /**
@@ -92,6 +90,81 @@ int yaz_marc_read_xml_subfields(yaz_marc_t mt, const xmlNode *ptr)
     return 0;
 }
 
+const char *tag_value_extract(const char *name, char tag_buffer[5])
+{
+    size_t length = strlen(name);
+    if (length == 3)
+    {
+        strcpy(tag_buffer, name);
+        return tag_buffer;
+    }
+    return 0;
+}
+
+// Given a xmlNode ptr,  extract a value from either a element name or from a given attribute
+char *element_attribute_value_extract(const xmlNode *ptr,
+                                      const char *attribute_name,
+                                      NMEM nmem)
+{
+    const char *name = (const char *) ptr->name;
+    size_t length = strlen(name);
+    xmlAttr *attr;
+    if (length > 1 )
+        return nmem_strdup(nmem, name+1);
+    // TODO Extract from attribute where matches attribute_name
+    for (attr = ptr->properties; attr; attr = attr->next)
+        if (!strcmp((const char *)attr->name, attribute_name))
+            return nmem_text_node_cdata(attr->children, nmem);
+    return 0;
+}
+
+
+int yaz_marc_read_turbo_xml_subfields(yaz_marc_t mt, const xmlNode *ptr)
+{
+    for (; ptr; ptr = ptr->next)
+    {
+        if (ptr->type == XML_ELEMENT_NODE)
+        {
+            if (!strncmp((const char *) ptr->name, "s", 1))
+            {
+                NMEM nmem = yaz_marc_get_nmem(mt);
+                xmlNode *p;
+               size_t ctrl_data_len = 0;
+                char *ctrl_data_buf = 0;
+                const char *tag_value = element_attribute_value_extract(ptr, "code", nmem);
+                if (!tag_value)
+                {
+                    yaz_marc_cprintf(
+                        mt, "Missing 'code' value for 'subfield'" );
+                    return -1;
+                }
+
+                ctrl_data_len = strlen((const char *) tag_value);
+                // Extract (length) from CDATA
+                for (p = ptr->children; p ; p = p->next)
+                    if (p->type == XML_TEXT_NODE)
+                        ctrl_data_len += strlen((const char *)p->content);
+                // Allocate memory for code value (1 character (can be multi-byte) and data
+                ctrl_data_buf = (char *) nmem_malloc(nmem, ctrl_data_len+1);
+                // Build a string with "<Code><data>"
+                strcpy(ctrl_data_buf, (const char *) tag_value);
+                for (p = ptr->children; p ; p = p->next)
+                    if (p->type == XML_TEXT_NODE)
+                        strcat(ctrl_data_buf, (const char *)p->content);
+                yaz_marc_add_subfield(mt, ctrl_data_buf, ctrl_data_len);
+            }
+            else
+            {
+                yaz_marc_cprintf(
+                    mt, "Expected element 'subfield', got '%.80s'", ptr->name);
+                return -1;
+            }
+        }
+    }
+    return 0;
+}
+
+
 static int yaz_marc_read_xml_leader(yaz_marc_t mt, const xmlNode **ptr_p)
 {
     int indicator_length;
@@ -106,7 +179,8 @@ static int yaz_marc_read_xml_leader(yaz_marc_t mt, const xmlNode **ptr_p)
     for(; ptr; ptr = ptr->next)
         if (ptr->type == XML_ELEMENT_NODE)
         {
-            if (!strcmp((const char *) ptr->name, "leader"))
+            if ( !strcmp( (const char *) ptr->name, "leader") ||
+                 (!strncmp((const char *) ptr->name, "l", 1) ))
             {
                 xmlNode *p = ptr->children;
                 for(; p; p = p->next)
@@ -147,7 +221,7 @@ static int yaz_marc_read_xml_fields(yaz_marc_t mt, const xmlNode *ptr)
     for(; ptr; ptr = ptr->next)
         if (ptr->type == XML_ELEMENT_NODE)
         {
-            if (!strcmp((const char *) ptr->name, "controlfield"))
+            if (!strcmp( (const char *) ptr->name, "controlfield"))
             {
                 const xmlNode *ptr_tag = 0;
                 struct _xmlAttr *attr;
@@ -217,18 +291,95 @@ static int yaz_marc_read_xml_fields(yaz_marc_t mt, const xmlNode *ptr)
         }
     return 0;
 }
+
+
+static int yaz_marc_read_turbo_xml_fields(yaz_marc_t mt, const xmlNode *ptr)
+{
+    for(; ptr; ptr = ptr->next)
+        if (ptr->type == XML_ELEMENT_NODE)
+        {
+            if (!strncmp( (const char *) ptr->name, "c", 1))
+            {
+                NMEM nmem = yaz_marc_get_nmem(mt);
+                char *tag_value = element_attribute_value_extract(ptr, "tag", nmem);
+                if (!tag_value)
+                {
+                    yaz_marc_cprintf(
+                        mt, "Missing attribute 'tag' for 'controlfield'" );
+                    return -1;
+                }
+                yaz_marc_add_controlfield_xml2(mt, tag_value, ptr->children);
+            }
+            else if (!strncmp((const char *) ptr->name, "d",1))
+            {
+                struct _xmlAttr *attr;
+                NMEM nmem = yaz_marc_get_nmem(mt);
+                char *tag_value;
+                char *indstr = nmem_malloc(nmem, 11);  /* 0(unused), 1,....9, + zero term */
+                int index = 0;
+                for (index = 0; index < 11; index++)
+                    indstr[index] = '\0';
+                tag_value = element_attribute_value_extract(ptr, "tag", nmem);
+                if (!tag_value)
+                {
+                    yaz_marc_cprintf(
+                        mt, "Missing attribute 'tag' for 'datafield'" );
+                    return -1;
+                }
+                for (attr = ptr->properties; attr; attr = attr->next)
+                    if (strlen((const char *)attr->name) == 2 &&
+                        attr->name[0] == 'i')
+                    {
+                       //extract indicator attribute from i#="Y" pattern
+                        int no = atoi((const char *)attr->name+1);
+                        if (attr->children
+                            && attr->children->type == XML_TEXT_NODE)
+                            indstr[no] = attr->children->content[0];
+                    }
+                    else
+                    {
+                        yaz_marc_cprintf(
+                            mt, "Bad attribute '%.80s' for 'datafield'",
+                            attr->name);
+                    }
+                /* note that indstr[0] is unused so we use indstr[1..] */
+                yaz_marc_add_datafield_xml2(mt, tag_value, indstr+1);
+                if (yaz_marc_read_turbo_xml_subfields(mt, ptr->children /*, indstr */))
+                    return -1;
+            }
+            else
+            {
+                yaz_marc_cprintf(mt,
+                                 "Expected element controlfield or datafield,"
+                                 " got %.80s", ptr->name);
+                return -1;
+            }
+        }
+    return 0;
+}
+
+
 #endif
 
+#if YAZ_HAVE_XML2
 int yaz_marc_read_xml(yaz_marc_t mt, const xmlNode *ptr)
 {
-#if YAZ_HAVE_XML2
+    int format = 0;
     yaz_marc_reset(mt);
-
+    
     for(; ptr; ptr = ptr->next)
         if (ptr->type == XML_ELEMENT_NODE)
         {
             if (!strcmp((const char *) ptr->name, "record"))
+            {
+                format = YAZ_MARC_MARCXML;
                 break;
+            }
+            else if (!strcmp((const char *) ptr->name, "r"))
+            {
+                format = YAZ_MARC_TURBOMARC;
+                break;
+            }
             else
             {
                 yaz_marc_cprintf(
@@ -246,16 +397,23 @@ int yaz_marc_read_xml(yaz_marc_t mt, const xmlNode *ptr)
     ptr = ptr->children;
     if (yaz_marc_read_xml_leader(mt, &ptr))
         return -1;
-    return yaz_marc_read_xml_fields(mt, ptr->next);
-#else
+    
+    switch (format)
+    {
+    case YAZ_MARC_MARCXML:
+        return yaz_marc_read_xml_fields(mt, ptr->next);
+    case YAZ_MARC_TURBOMARC:
+        return yaz_marc_read_turbo_xml_fields(mt, ptr->next);
+    }
     return -1;
-#endif
 }
+#endif
 
 
 /*
  * Local variables:
  * c-basic-offset: 4
+ * c-file-style: "Stroustrup"
  * indent-tabs-mode: nil
  * End:
  * vim: shiftwidth=4 tabstop=8 expandtab