schematron.c

xml开源解析代码.版本为libxml2-2.6.29,可支持GB3212.网络消息发送XML时很有用.

	return;
    } else {
	ruleptr = xmlSchematronAddRule(ctxt, ctxt->schema, pattern,
	                               rule, context, NULL);
	if (ruleptr == NULL) {
	    xmlFree(context);
	    return;
	}
    }

    cur = rule->children;
    NEXT_SCHEMATRON(cur);
    while (cur != NULL) {
	if (IS_SCHEMATRON(cur, "assert")) {
	    nbChecks++;
	    test = xmlGetNoNsProp(cur, BAD_CAST "test");
	    if (test == NULL) {
		xmlSchematronPErr(ctxt, cur,
		    XML_SCHEMAP_NOROOT,
		    "assert has no test attribute",
		    NULL, NULL);
	    } else if (test[0] == 0) {
		xmlSchematronPErr(ctxt, cur,
		    XML_SCHEMAP_NOROOT,
		    "assert has an empty test attribute",
		    NULL, NULL);
		xmlFree(test);
	    } else {
		/* TODO will need dynamic processing instead */
		report = xmlNodeGetContent(cur);

		testptr = xmlSchematronAddTest(ctxt, XML_SCHEMATRON_ASSERT,
		                               ruleptr, cur, test, report);
		if (testptr == NULL)
		    xmlFree(test);
	    }
	} else if (IS_SCHEMATRON(cur, "report")) {
	    nbChecks++;
	    test = xmlGetNoNsProp(cur, BAD_CAST "test");
	    if (test == NULL) {
		xmlSchematronPErr(ctxt, cur,
		    XML_SCHEMAP_NOROOT,
		    "assert has no test attribute",
		    NULL, NULL);
	    } else if (test[0] == 0) {
		xmlSchematronPErr(ctxt, cur,
		    XML_SCHEMAP_NOROOT,
		    "assert has an empty test attribute",
		    NULL, NULL);
		xmlFree(test);
	    } else {
		/* TODO will need dynamic processing instead */
		report = xmlNodeGetContent(cur);

		testptr = xmlSchematronAddTest(ctxt, XML_SCHEMATRON_REPORT,
		                               ruleptr, cur, test, report);
		if (testptr == NULL)
		    xmlFree(test);
	    }
	} else {
	    xmlSchematronPErr(ctxt, cur,
		XML_SCHEMAP_NOROOT,
		"Expecting an assert or a report element instead of %s",
		cur->name, NULL);
	}
	cur = cur->next;
	NEXT_SCHEMATRON(cur);
    }
    if (nbChecks == 0) {
	xmlSchematronPErr(ctxt, rule,
	    XML_SCHEMAP_NOROOT,
	    "rule has no assert nor report element", NULL, NULL);
    }
}

/**
 * xmlSchematronParsePattern:
 * @ctxt:  a schema validation context
 * @pat:  the pattern node
 *
 * parse a pattern element
 */
static void
xmlSchematronParsePattern(xmlSchematronParserCtxtPtr ctxt, xmlNodePtr pat)
{
    xmlNodePtr cur;
    xmlSchematronPatternPtr pattern;
    int nbRules = 0;
    xmlChar *id;

    if ((ctxt == NULL) || (pat == NULL)) return;

    id = xmlGetNoNsProp(pat, BAD_CAST "id");
    if (id == NULL) {
	id = xmlGetNoNsProp(pat, BAD_CAST "name");
    }
    pattern = xmlSchematronAddPattern(ctxt, ctxt->schema, pat, id);
    if (pattern == NULL) {
	if (id != NULL)
	    xmlFree(id);
        return;
    }
    cur = pat->children;
    NEXT_SCHEMATRON(cur);
    while (cur != NULL) {
	if (IS_SCHEMATRON(cur, "rule")) {
	    xmlSchematronParseRule(ctxt, pattern, cur);
	    nbRules++;
	} else {
	    xmlSchematronPErr(ctxt, cur,
		XML_SCHEMAP_NOROOT,
		"Expecting a rule element instead of %s", cur->name, NULL);
	}
	cur = cur->next;
	NEXT_SCHEMATRON(cur);
    }
    if (nbRules == 0) {
	xmlSchematronPErr(ctxt, pat,
	    XML_SCHEMAP_NOROOT,
	    "Pattern has no rule element", NULL, NULL);
    }
}

#if 0
/**
 * xmlSchematronLoadInclude:
 * @ctxt:  a schema validation context
 * @cur:  the include element
 *
 * Load the include document, Push the current pointer
 *
 * Returns the updated node pointer
 */
static xmlNodePtr
xmlSchematronLoadInclude(xmlSchematronParserCtxtPtr ctxt, xmlNodePtr cur)
{
    xmlNodePtr ret = NULL;
    xmlDocPtr doc = NULL;
    xmlChar *href = NULL;
    xmlChar *base = NULL;
    xmlChar *URI = NULL;

    if ((ctxt == NULL) || (cur == NULL))
        return(NULL);

    href = xmlGetNoNsProp(cur, BAD_CAST "href");
    if (href == NULL) {
	xmlSchematronPErr(ctxt, cur,
	    XML_SCHEMAP_NOROOT,
	    "Include has no href attribute", NULL, NULL);
	return(cur->next);
    }

    /* do the URI base composition, load and find the root */
    base = xmlNodeGetBase(cur->doc, cur);
    URI = xmlBuildURI(href, base);
    doc = xmlReadFile((const char *) URI, NULL, SCHEMATRON_PARSE_OPTIONS);
    if (doc == NULL) {
	xmlSchematronPErr(ctxt, cur,
		      XML_SCHEMAP_FAILED_LOAD,
		      "could not load include '%s'.\n",
		      URI, NULL);
	goto done;
    }
    ret = xmlDocGetRootElement(doc);
    if (ret == NULL) {
	xmlSchematronPErr(ctxt, cur,
		      XML_SCHEMAP_FAILED_LOAD,
		      "could not find root from include '%s'.\n",
		      URI, NULL);
	goto done;
    }

    /* Success, push the include for rollback on exit */
    xmlSchematronPushInclude(ctxt, doc, cur);

done:
    if (ret == NULL) {
        if (doc != NULL)
	    xmlFreeDoc(doc);
    }
    xmlFree(href);
    if (base != NULL)
        xmlFree(base);
    if (URI != NULL)
        xmlFree(URI);
    return(ret);
}
#endif

/**
 * xmlSchematronParse:
 * @ctxt:  a schema validation context
 *
 * parse a schema definition resource and build an internal
 * XML Shema struture which can be used to validate instances.
 *
 * Returns the internal XML Schematron structure built from the resource or
 *         NULL in case of error
 */
xmlSchematronPtr
xmlSchematronParse(xmlSchematronParserCtxtPtr ctxt)
{
    xmlSchematronPtr ret = NULL;
    xmlDocPtr doc;
    xmlNodePtr root, cur;
    int preserve = 0;

    if (ctxt == NULL)
        return (NULL);

    ctxt->nberrors = 0;

    /*
     * First step is to parse the input document into an DOM/Infoset
     */
    if (ctxt->URL != NULL) {
        doc = xmlReadFile((const char *) ctxt->URL, NULL,
	                  SCHEMATRON_PARSE_OPTIONS);
        if (doc == NULL) {
	    xmlSchematronPErr(ctxt, NULL,
			  XML_SCHEMAP_FAILED_LOAD,
                          "xmlSchematronParse: could not load '%s'.\n",
                          ctxt->URL, NULL);
            return (NULL);
        }
	ctxt->preserve = 0;
    } else if (ctxt->buffer != NULL) {
        doc = xmlReadMemory(ctxt->buffer, ctxt->size, NULL, NULL,
	                    SCHEMATRON_PARSE_OPTIONS);
        if (doc == NULL) {
	    xmlSchematronPErr(ctxt, NULL,
			  XML_SCHEMAP_FAILED_PARSE,
                          "xmlSchematronParse: could not parse.\n",
                          NULL, NULL);
            return (NULL);
        }
        doc->URL = xmlStrdup(BAD_CAST "in_memory_buffer");
        ctxt->URL = xmlDictLookup(ctxt->dict, BAD_CAST "in_memory_buffer", -1);
	ctxt->preserve = 0;
    } else if (ctxt->doc != NULL) {
        doc = ctxt->doc;
	preserve = 1;
	ctxt->preserve = 1;
    } else {
	xmlSchematronPErr(ctxt, NULL,
		      XML_SCHEMAP_NOTHING_TO_PARSE,
		      "xmlSchematronParse: could not parse.\n",
		      NULL, NULL);
        return (NULL);
    }

    /*
     * Then extract the root and Schematron parse it
     */
    root = xmlDocGetRootElement(doc);
    if (root == NULL) {
	xmlSchematronPErr(ctxt, (xmlNodePtr) doc,
		      XML_SCHEMAP_NOROOT,
		      "The schema has no document element.\n", NULL, NULL);
	if (!preserve) {
	    xmlFreeDoc(doc);
	}
        return (NULL);
    }

    if (!IS_SCHEMATRON(root, "schema")) {
	xmlSchematronPErr(ctxt, root,
	    XML_SCHEMAP_NOROOT,
	    "The XML document '%s' is not a XML schematron document",
	    ctxt->URL, NULL);
	goto exit;
    }
    ret = xmlSchematronNewSchematron(ctxt);
    if (ret == NULL)
        goto exit;
    ctxt->schema = ret;

    /*
     * scan the schema elements
     */
    cur = root->children;
    NEXT_SCHEMATRON(cur);
    if (IS_SCHEMATRON(cur, "title")) {
        xmlChar *title = xmlNodeGetContent(cur);
	if (title != NULL) {
	    ret->title = xmlDictLookup(ret->dict, title, -1);
	    xmlFree(title);
	}
	cur = cur->next;
	NEXT_SCHEMATRON(cur);
    }
    while (IS_SCHEMATRON(cur, "ns")) {
        xmlChar *prefix = xmlGetNoNsProp(cur, BAD_CAST "prefix");
        xmlChar *uri = xmlGetNoNsProp(cur, BAD_CAST "uri");
	if ((uri == NULL) || (uri[0] == 0)) {
	    xmlSchematronPErr(ctxt, cur,
		XML_SCHEMAP_NOROOT,
		"ns element has no uri", NULL, NULL);
	}
	if ((prefix == NULL) || (prefix[0] == 0)) {
	    xmlSchematronPErr(ctxt, cur,
		XML_SCHEMAP_NOROOT,
		"ns element has no prefix", NULL, NULL);
	}
	if ((prefix) && (uri)) {
	    xmlXPathRegisterNs(ctxt->xctxt, prefix, uri);
	    xmlSchematronAddNamespace(ctxt, prefix, uri);
	    ret->nbNs++;
	}
	if (uri)
	    xmlFree(uri);
	if (prefix)
	    xmlFree(prefix);
	cur = cur->next;
	NEXT_SCHEMATRON(cur);
    }
    while (cur != NULL) {
	if (IS_SCHEMATRON(cur, "pattern")) {
	    xmlSchematronParsePattern(ctxt, cur);
	    ret->nbPattern++;
	} else {
	    xmlSchematronPErr(ctxt, cur,
		XML_SCHEMAP_NOROOT,
		"Expecting a pattern element instead of %s", cur->name, NULL);
	}
	cur = cur->next;
	NEXT_SCHEMATRON(cur);
    }
    if (ret->nbPattern == 0) {
	xmlSchematronPErr(ctxt, root,
	    XML_SCHEMAP_NOROOT,
	    "The schematron document '%s' has no pattern",
	    ctxt->URL, NULL);
	goto exit;
    }
    /* the original document must be kept for reporting */
    ret->doc = doc;
    preserve = 1;

exit:
    if (!preserve) {
	xmlFreeDoc(doc);
    }
    if (ret != NULL) {
	if (ctxt->nberrors != 0) {
	    xmlSchematronFree(ret);
	    ret = NULL;
	} else {
	    ret->namespaces = ctxt->namespaces;
	    ret->nbNamespaces = ctxt->nbNamespaces;
	    ctxt->namespaces = NULL;
	}
    }
    return (ret);
}

/************************************************************************
 *									*
 *		Schematrontron Reports handler				*
 *									*
 ************************************************************************/

static xmlNodePtr
xmlSchematronGetNode(xmlSchematronValidCtxtPtr ctxt,
                     xmlNodePtr cur, const xmlChar *xpath) {
    xmlNodePtr node = NULL;
    xmlXPathObjectPtr ret;

    if ((ctxt == NULL) || (cur == NULL) || (xpath == NULL))
        return(NULL);

    ctxt->xctxt->doc = cur->doc;
    ctxt->xctxt->node = cur;
    ret = xmlXPathEval(xpath, ctxt->xctxt);
    if (ret == NULL)
        return(NULL);

    if ((ret->type == XPATH_NODESET) &&
        (ret->nodesetval != NULL) && (ret->nodesetval->nodeNr > 0))
	node = ret->nodesetval->nodeTab[0];

    xmlXPathFreeObject(ret);
    return(node);
}

/**
 * xmlSchematronReportOutput:
 * @ctxt: the validation context
 * @cur: the current node tested
 * @msg: the message output
 *
 * Output part of the report to whatever channel the user selected
 */
static void
xmlSchematronReportOutput(xmlSchematronValidCtxtPtr ctxt ATTRIBUTE_UNUSED,
                          xmlNodePtr cur ATTRIBUTE_UNUSED,
                          const char *msg) {
    /* TODO */
    fprintf(stderr, "%s", msg);
}

/**
 * xmlSchematronFormatReport:
 * @ctxt:  the validation context
 * @test: the test node
 * @cur: the current node tested
 *
 * Build the string being reported to the user.
 *
 * Returns a report string or NULL in case of error. The string needs
 *         to be deallocated by teh caller
 */
static xmlChar *
xmlSchematronFormatReport(xmlSchematronValidCtxtPtr ctxt, 
			  xmlNodePtr test, xmlNodePtr cur) {
    xmlChar *ret = NULL;
    xmlNodePtr child, node;

    if ((test == NULL) || (cur == NULL))
        return(ret);

    child = test->children;
    while (child != NULL) {
        if ((child->type == XML_TEXT_NODE) ||
	    (child->type == XML_CDATA_SECTION_NODE))
	    ret = xmlStrcat(ret, child->content);
	else if (IS_SCHEMATRON(child, "name")) {
	    xmlChar *path;

	    path = xmlGetNoNsProp(child, BAD_CAST "path");

            node = cur;
	    if (path != NULL) {