X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=src%2Fnfaxml.c;h=2771f56ac7ec2b823a1c68f1d48c7194550ead19;hp=a573eee4041c3d79cb536f28d536c28574361201;hb=e4368af2ede9079d33e7a8fd280d29b2bde1d1ad;hpb=fb61309506b52ca0dad9aea4135ceeda63707b0d

diff --git a/src/nfaxml.c b/src/nfaxml.c
index a573eee..2771f56 100644
--- a/src/nfaxml.c
+++ b/src/nfaxml.c
@@ -1,7 +1,7 @@
 /*  Copyright (C) 2006, Index Data ApS
  *  See the file LICENSE for details.
  * 
- *  $Id: nfaxml.c,v 1.1 2006-07-04 12:59:56 heikki Exp $ 
+ *  $Id: nfaxml.c,v 1.11 2006-10-09 14:22:43 heikki Exp $ 
  */
 
 /**
@@ -10,11 +10,11 @@
  *
  */
 
-#if HAVE_XML2
+#if YAZ_HAVE_XML2
 
 #include <string.h>
 
-#include <libxml/parser.h>
+/* #include <libxml/parser.h> */
 #include <libxml/tree.h>
 #include <libxml/xinclude.h>
 
@@ -22,60 +22,298 @@
 #include <yaz/nmem.h> 
 #include <yaz/yconfig.h>
 #include <yaz/nfa.h>
+#include <yaz/yaz-iconv.h>
 #include <yaz/nfaxml.h>
+#include <yaz/libxml2_error.h>
 
+/** \brief How long strings we are willing to handle here */
+#define MAXDATALEN 200 
 
+/** \brief Get content of a node, in utf16, for yaz_nfa */
+static int utf16_content(xmlNodePtr node, yaz_nfa_char *buf, int maxlen,
+        const char *filename, int rulenumber)
+{
+    int bufidx=0;
+    xmlChar *content = xmlNodeGetContent(node);
+    xmlChar *cp = content;
+    size_t conlen = strlen((char *)content);
+    while (*cp && bufidx<maxlen )
+    {
+        int error;
+        size_t no_read;
+        int res = yaz_read_UTF8_char(cp, conlen, &no_read, &error);
+        if (res == 0) {
+            /* should be caught earlier */
+            yaz_log(YLOG_FATAL,"Illegal utf-8 sequence "
+                    "%ld bytes into '%s' in %s, rule %d ",
+                    (long)(cp-content), content, filename, rulenumber);
+            xmlFree(content);
+            return -1;
+        }
+        buf[bufidx++] = res;
+        cp += no_read;
+        conlen -= no_read;
+    }
+    buf[bufidx]=0;
+    xmlFree(content);
+    return bufidx;
+}
 
+static int parse_range(xmlNodePtr node, 
+        yaz_nfa_char *range_start,
+        yaz_nfa_char *range_end,
+        const char *filename, int rulenumber )
+{
+    xmlChar *content = xmlNodeGetContent(node);
+    xmlChar *cp=content;
+    size_t conlen = strlen((char *)content);
+    size_t no_read;
+    int error;
+    int res = yaz_read_UTF8_char(cp, conlen, &no_read, &error);
+    if ( res != 0 ) {
+        *range_start=res;
+        cp += no_read;
+        conlen -= no_read;
+        res = yaz_read_UTF8_char(cp, conlen, &no_read, &error);
+        if (res != '-' )
+            res = 0;
+    }
+    if ( res != 0 ) {
+        cp += no_read;
+        conlen -= no_read;
+        res = yaz_read_UTF8_char(cp, conlen, &no_read, &error);
+    }
+    if ( res != 0) {
+        *range_end = res;
+    }
+    xmlFree(content);
+    if (res == 0) {
+        yaz_log(YLOG_FATAL,"Illegal range. '%s'. Must be like 'a-z' "
+                "'in %s, rule %d ",
+                content, filename, rulenumber);
+        return 0;
+    }
+    return 1;
+} /* parserange */
 
-/** \brief Log XML errors in yaz_log 
- *
- */
 
-static void log_xml_error(int errlevel, char *msg) {
-    xmlErrorPtr e=xmlGetLastError();
-    if (!e)  /* no error happened */
-        return;  
-    if (!errlevel)
-        errlevel=YLOG_FATAL;
-    yaz_log(errlevel,"%s %d/%d: %s:%d: '%s' ",
-            msg, e->domain, e->code, e->file, e->line, e->message);
-    if (e->str1 || e->str2 || e->str3 ) 
-    yaz_log(errlevel,"extra info: '%s' '%s' '%s' %d %d",
-            e->str1, e->str2, e->str3, e->int1, e->int2 );
-}
+/** \brief Parse a fromstring clause */
+static yaz_nfa_state *parse_fromstring(yaz_nfa *nfa, 
+        xmlNodePtr node, const char *filename, int rulenumber )
+{
+    yaz_nfa_char buf[MAXDATALEN];
+    yaz_nfa_state *state;
+    int bufidx=utf16_content(node, buf, MAXDATALEN, filename, rulenumber);
+    if (bufidx<0) 
+        return 0;
+    state=yaz_nfa_add_sequence(nfa, 0, buf, bufidx);
+    return state;
+} /* parse_fromstring */
+
+/** \brief Parse a tostring clause */
+static yaz_nfa_converter *parse_tostring(yaz_nfa *nfa,
+                xmlNodePtr node, const char *filename, int rulenumber )
+{
+    yaz_nfa_char buf[MAXDATALEN];
+    yaz_nfa_converter *conv;
+    int bufidx=utf16_content(node, buf, MAXDATALEN, filename, rulenumber);
+    if (bufidx<0) 
+        return 0;
+    conv=yaz_nfa_create_string_converter(nfa, buf, bufidx);
+    return conv;
+} /* parse_tostring */
+
+static yaz_nfa_state * parse_fromrange(yaz_nfa *nfa,
+                xmlNodePtr node, 
+                yaz_nfa_char *from_begin,
+                yaz_nfa_char *from_end,
+                const char *filename, int rulenumber )
+{
+    yaz_nfa_char begin;
+    yaz_nfa_char end;
+    yaz_nfa_state *state;
+    int rc;
+    rc=parse_range(node, &begin, &end, filename, rulenumber);
+    if (!rc)
+        return 0;
+    *from_begin=begin;
+    *from_end=end; /* save for calculating the to-range */
+    state=yaz_nfa_add_range(nfa, 0, begin, end);
+    return state;
+} /* parse_fromrange */
+
+static yaz_nfa_converter *parse_torange(yaz_nfa *nfa,
+             xmlNodePtr node, yaz_nfa_char from_begin, yaz_nfa_char from_end,
+             const char *filename, int rulenumber )
+{
+    yaz_nfa_char begin;
+    yaz_nfa_char end;
+    yaz_nfa_converter *conv;
+    int rc;
+    rc=parse_range(node, &begin, &end, filename, rulenumber);
+    if (!rc)
+        return 0;
+    if ( from_end - from_begin != end - begin ) {
+        yaz_log(YLOG_FATAL,"From-range not as long as to-range: "
+                "from=%x-%x to=%x-%x in rule %d in %s",
+                from_begin, from_end,  begin, end, rulenumber, filename);
+        return 0;
+    }
+    conv=yaz_nfa_create_range_converter(nfa, 0, from_begin, begin);
+    return conv;
+} /* parse_torange */
+
+/** \brief Parse one rule from an XML node */
+static int parse_rule(yaz_nfa *nfa, xmlNodePtr rulenode, 
+        const char *filename, int rulenumber ) 
+{
+    yaz_nfa_state *state=0;
+    yaz_nfa_converter *conv=0;
+    yaz_nfa_char range_begin=0, range_end=0;
+    xmlNodePtr node;
+    int clauses=0;
+    for (node = rulenode->children; node; node = node->next)
+    {
+        if (node->type != XML_ELEMENT_NODE)
+            continue;
+        clauses++;
+        if (!strcmp((const char *) node->name, "fromstring")) 
+        {
+            state = parse_fromstring(nfa, node, filename, rulenumber );
+            if (!state)
+                return 0;
+        } else if (!strcmp((const char *) node->name, "tostring")) 
+        {
+            conv = parse_tostring(nfa, node, filename, rulenumber );
+            if (!conv)
+                return 0;
+        } else if (!strcmp((const char *) node->name, "fromrange")) 
+        {
+            state = parse_fromrange(nfa, node, 
+                    &range_begin, &range_end, filename, rulenumber );
+            if (!state)
+                return 0;
+        } else if (!strcmp((const char *) node->name, "torange")) 
+        {
+            conv = parse_torange(nfa, node, 
+                    range_begin, range_end, filename, rulenumber );
+            if (!conv)
+                return 0;
+        } else {
+            yaz_log(YLOG_FATAL,"Unknown clause '%s' in %s rule %d",
+                    node->name, filename,rulenumber);
+            return 0;
+        }
+    } /* for child */
+    if (!state) {
+        yaz_log(YLOG_FATAL,"No 'from' clause in a rule %d in %s", 
+                rulenumber,filename);
+        return 0;
+    }
+    if (!conv) {
+        yaz_log(YLOG_FATAL,"No 'to' clause in a rule %d in %s",
+                rulenumber,filename);
+        return 0;
+    }
+    if (clauses != 2) {
+        yaz_log(YLOG_FATAL,"Must have exactly one 'from' and one 'to' clause "
+                "in rule %d in %s", rulenumber,filename);
+        return 0;
+    }
+    if ( YAZ_NFA_SUCCESS == yaz_nfa_set_result(nfa,state,conv))
+        return 1; 
+    yaz_log(YLOG_FATAL,"Conflicting rules in %s rule %d",
+            filename, rulenumber);
+    return 0;
+} /* parse_rule */
+
 
 /** \brief Parse the NFA from a XML document 
  */
-yaz_nfa *yaz_nfa_parse_xml_doc(xmlDocPtr doc){
+yaz_nfa *yaz_nfa_parse_xml_doc(xmlDocPtr doc, const char *filename)
+{
+    xmlNodePtr node;
+    yaz_nfa *nfa;
+    int rulenumber=0;
+
     if (!doc)
         return 0;
-
-    return 0;
-}
+    libxml2_error_to_yazlog(YLOG_FATAL, "yaz_nfa_parse_doc");
+    node = xmlDocGetRootElement(doc);
+    if (!node || node->type != XML_ELEMENT_NODE ||
+        strcmp((const char *) node->name, "ruleset")) 
+    {
+        yaz_log(YLOG_FATAL,"nfa_parse_xml: Could not find root element 'ruleset' "
+                "in %s", filename);
+        return 0;
+    }
+    nfa= yaz_nfa_init();
+    if (!nfa) 
+    {
+        yaz_log(YLOG_FATAL,"nfa_parse_xml: Creating nfa failed, can't parse %s",
+                filename);
+        return 0;
+    }
+        
+    for (node = node->children; node; node = node->next)
+    {
+        if (node->type != XML_ELEMENT_NODE)
+            continue;
+         if (!strcmp((const char *) node->name, "rule")) {
+             if (!parse_rule(nfa,node,filename,rulenumber++))
+                 return 0;
+         } else {
+            yaz_log(YLOG_FATAL,"nfa_parse_xml: "
+                    "expected 'rule', found '%s' in %s", 
+                    (const char *) node->name,filename);
+            return 0;
+         }
+    } /* for */
+    return nfa;
+} /* yaz_nfa_parse_xml_doc */
 
 
 /** \brief Parse the NFA from a file 
  */
-yaz_nfa *yaz_nfa_parse_xml_file(char *filepath) {
+yaz_nfa *yaz_nfa_parse_xml_file(const char *filepath) 
+{
     int nSubst;
-    xmlDocPtr doc = xmlParseFile(filepath);
+    xmlDocPtr doc;
+    if (!filepath) 
+    {
+        yaz_log(YLOG_FATAL,"yaz_nfa_parse_xml_file called with NULL");
+        return 0;
+    }
+    libxml2_error_to_yazlog(YLOG_FATAL, "yaz_nfa_parse_xml_file");
+
+    doc = xmlParseFile(filepath);
     if (!doc) {
-        log_xml_error(YLOG_FATAL,
-                "Error in parsing charmap file");
         return 0;
     }
     nSubst=xmlXIncludeProcess(doc);
     if (nSubst==-1) {
-        log_xml_error(YLOG_FATAL,
-                "Error handling XIncludes in charmap file");
         return 0;
     }
-    return yaz_nfa_parse_xml_doc(doc);
+    return yaz_nfa_parse_xml_doc(doc, filepath);
+}
+
+/** \brief Parse the NFA from a memory buffer
+ */
+yaz_nfa *yaz_nfa_parse_xml_memory(const char *xmlbuff, const char *filename) {
+    xmlDocPtr doc;
+    if (!xmlbuff) 
+    {
+        yaz_log(YLOG_FATAL,"yaz_nfa_parse_memroy called with NULL");
+        return 0;
+    }
+    libxml2_error_to_yazlog(YLOG_FATAL, "yaz_nfa_parse_xml_memory");
+    doc = xmlParseMemory(xmlbuff, strlen(xmlbuff));
+    return yaz_nfa_parse_xml_doc(doc,filename);
 }
 
 
 
-#endif /* HAVE_XML2 */
+#endif /* YAZ_HAVE_XML2 */
 
 
 /*