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.

/*
 * schematron.c : implementation of the Schematron schema validity checking
 *
 * See Copyright for the status of this software.
 *
 * Daniel Veillard <[e-mail]>
 */

/*
 * TODO:
 * + double check the semantic, especially
 *        - multiple rules applying in a single pattern/node
 *        - the semantic of libxml2 patterns vs. XSLT production referenced
 *          by the spec.
 * + export of results in SVRL
 * + full parsing and coverage of the spec, conformance of the input to the
 *   spec
 * + divergences between the draft and the ISO proposed standard :-(
 * + hook and test include
 * + try and compare with the XSLT version
 */

#define IN_LIBXML
#include "libxml.h"

#ifdef LIBXML_SCHEMATRON_ENABLED

#include <string.h>
#include <libxml/parser.h>
#include <libxml/tree.h>
#include <libxml/uri.h>
#include <libxml/xpath.h>
#include <libxml/xpathInternals.h>
#include <libxml/pattern.h>
#include <libxml/schematron.h>

#define SCHEMATRON_PARSE_OPTIONS XML_PARSE_NOENT

#define SCT_OLD_NS BAD_CAST "http://www.ascc.net/xml/schematron"

#define XML_SCHEMATRON_NS BAD_CAST "http://purl.oclc.org/dsdl/schematron"


static const xmlChar *xmlSchematronNs = XML_SCHEMATRON_NS;
static const xmlChar *xmlOldSchematronNs = SCT_OLD_NS;

#define IS_SCHEMATRON(node, elem)					\
   ((node != NULL) && (node->type == XML_ELEMENT_NODE ) &&		\
    (node->ns != NULL) &&						\
    (xmlStrEqual(node->name, (const xmlChar *) elem)) &&		\
    ((xmlStrEqual(node->ns->href, xmlSchematronNs)) ||			\
     (xmlStrEqual(node->ns->href, xmlOldSchematronNs))))

#define NEXT_SCHEMATRON(node)						\
   while (node != NULL) {						\
       if ((node->type == XML_ELEMENT_NODE ) && (node->ns != NULL) && 	\
           ((xmlStrEqual(node->ns->href, xmlSchematronNs)) ||		\
	    (xmlStrEqual(node->ns->href, xmlOldSchematronNs))))		\
	   break;							\
       node = node->next;						\
   }

/**
 * TODO:
 *
 * macro to flag unimplemented blocks
 */
#define TODO 								\
    xmlGenericError(xmlGenericErrorContext,				\
	    "Unimplemented block at %s:%d\n",				\
            __FILE__, __LINE__);

typedef enum {
    XML_SCHEMATRON_ASSERT=1,
    XML_SCHEMATRON_REPORT=2
} xmlSchematronTestType;

/**
 * _xmlSchematronTest:
 *
 * A Schematrons test, either an assert or a report
 */
typedef struct _xmlSchematronTest xmlSchematronTest;
typedef xmlSchematronTest *xmlSchematronTestPtr;
struct _xmlSchematronTest {
    xmlSchematronTestPtr next;	/* the next test in the list */
    xmlSchematronTestType type;	/* the test type */
    xmlNodePtr node;		/* the node in the tree */
    xmlChar *test;		/* the expression to test */
    xmlXPathCompExprPtr comp;	/* the compiled expression */
    xmlChar *report;		/* the message to report */
};

/**
 * _xmlSchematronRule:
 *
 * A Schematrons rule
 */
typedef struct _xmlSchematronRule xmlSchematronRule;
typedef xmlSchematronRule *xmlSchematronRulePtr;
struct _xmlSchematronRule {
    xmlSchematronRulePtr next;	/* the next rule in the list */
    xmlSchematronRulePtr patnext;/* the next rule in the pattern list */
    xmlNodePtr node;		/* the node in the tree */
    xmlChar *context;		/* the context evaluation rule */
    xmlSchematronTestPtr tests;	/* the list of tests */
    xmlPatternPtr pattern;	/* the compiled pattern associated */
    xmlChar *report;		/* the message to report */
};

/**
 * _xmlSchematronPattern:
 *
 * A Schematrons pattern
 */
typedef struct _xmlSchematronPattern xmlSchematronPattern;
typedef xmlSchematronPattern *xmlSchematronPatternPtr;
struct _xmlSchematronPattern {
    xmlSchematronPatternPtr next;/* the next pattern in the list */
    xmlSchematronRulePtr rules;	/* the list of rules */
    xmlChar *name;		/* the name of the pattern */
};

/**
 * _xmlSchematron:
 *
 * A Schematrons definition
 */
struct _xmlSchematron {
    const xmlChar *name;	/* schema name */
    int preserve;		/* was the document passed by the user */
    xmlDocPtr doc;		/* pointer to the parsed document */
    int flags;			/* specific to this schematron */

    void *_private;		/* unused by the library */
    xmlDictPtr dict;		/* the dictionnary used internally */

    const xmlChar *title;	/* the title if any */

    int nbNs;			/* the number of namespaces */

    int nbPattern;		/* the number of patterns */
    xmlSchematronPatternPtr patterns;/* the patterns found */
    xmlSchematronRulePtr rules;	/* the rules gathered */
    int nbNamespaces;		/* number of namespaces in the array */
    int maxNamespaces;		/* size of the array */
    const xmlChar **namespaces;	/* the array of namespaces */
};

/**
 * xmlSchematronValidCtxt:
 *
 * A Schematrons validation context
 */
struct _xmlSchematronValidCtxt {
    int type;
    int flags;			/* an or of xmlSchematronValidOptions */

    xmlDictPtr dict;
    int nberrors;
    int err;

    xmlSchematronPtr schema;
    xmlXPathContextPtr xctxt;

    FILE *outputFile;		/* if using XML_SCHEMATRON_OUT_FILE */
    xmlBufferPtr outputBuffer;	/* if using XML_SCHEMATRON_OUT_BUFFER */
    xmlOutputWriteCallback iowrite; /* if using XML_SCHEMATRON_OUT_IO */
    xmlOutputCloseCallback  ioclose;
    void *ioctx;
};

struct _xmlSchematronParserCtxt {
    int type;
    const xmlChar *URL;
    xmlDocPtr doc;
    int preserve;               /* Whether the doc should be freed  */
    const char *buffer;
    int size;

    xmlDictPtr dict;            /* dictionnary for interned string names */

    int nberrors;
    int err;
    xmlXPathContextPtr xctxt;	/* the XPath context used for compilation */
    xmlSchematronPtr schema;

    int nbNamespaces;		/* number of namespaces in the array */
    int maxNamespaces;		/* size of the array */
    const xmlChar **namespaces;	/* the array of namespaces */

    int nbIncludes;		/* number of includes in the array */
    int maxIncludes;		/* size of the array */
    xmlNodePtr *includes;	/* the array of includes */

    /* error rreporting data */
    void *userData;                      /* user specific data block */
    xmlSchematronValidityErrorFunc error;/* the callback in case of errors */
    xmlSchematronValidityWarningFunc warning;/* callback in case of warning */
    xmlStructuredErrorFunc serror;       /* the structured function */

};

#define XML_STRON_CTXT_PARSER 1
#define XML_STRON_CTXT_VALIDATOR 2

/************************************************************************
 *									*
 *			Error reporting					*
 *									*
 ************************************************************************/

/**
 * xmlSchematronPErrMemory:
 * @node: a context node
 * @extra:  extra informations
 *
 * Handle an out of memory condition
 */
static void
xmlSchematronPErrMemory(xmlSchematronParserCtxtPtr ctxt,
                        const char *extra, xmlNodePtr node)
{
    if (ctxt != NULL)
        ctxt->nberrors++;
    __xmlSimpleError(XML_FROM_SCHEMASP, XML_ERR_NO_MEMORY, node, NULL,
                     extra);
}

/**
 * xmlSchematronPErr:
 * @ctxt: the parsing context
 * @node: the context node
 * @error: the error code
 * @msg: the error message
 * @str1: extra data
 * @str2: extra data
 * 
 * Handle a parser error
 */
static void
xmlSchematronPErr(xmlSchematronParserCtxtPtr ctxt, xmlNodePtr node, int error,
              const char *msg, const xmlChar * str1, const xmlChar * str2)
{
    xmlGenericErrorFunc channel = NULL;
    xmlStructuredErrorFunc schannel = NULL;
    void *data = NULL;

    if (ctxt != NULL) {
        ctxt->nberrors++;
        channel = ctxt->error;
        data = ctxt->userData;
	schannel = ctxt->serror;
    }
    __xmlRaiseError(schannel, channel, data, ctxt, node, XML_FROM_SCHEMASP,
                    error, XML_ERR_ERROR, NULL, 0,
                    (const char *) str1, (const char *) str2, NULL, 0, 0,
                    msg, str1, str2);
}

/**
 * xmlSchematronVTypeErrMemory:
 * @node: a context node
 * @extra:  extra informations
 *
 * Handle an out of memory condition
 */
static void
xmlSchematronVErrMemory(xmlSchematronValidCtxtPtr ctxt,
                        const char *extra, xmlNodePtr node)
{
    if (ctxt != NULL) {
        ctxt->nberrors++;
        ctxt->err = XML_SCHEMAV_INTERNAL;
    }
    __xmlSimpleError(XML_FROM_SCHEMASV, XML_ERR_NO_MEMORY, node, NULL,
                     extra);
}

/************************************************************************
 *									*
 *		Parsing and compilation of the Schematrontrons		*
 *									*
 ************************************************************************/

/**
 * xmlSchematronAddTest:
 * @ctxt: the schema parsing context
 * @type:  the type of test
 * @rule:  the parent rule
 * @node:  the node hosting the test
 * @test: the associated test
 * @report: the associated report string
 *
 * Add a test to a schematron
 *
 * Returns the new pointer or NULL in case of error
 */
static xmlSchematronTestPtr
xmlSchematronAddTest(xmlSchematronParserCtxtPtr ctxt,
                     xmlSchematronTestType type,
                     xmlSchematronRulePtr rule,
                     xmlNodePtr node, xmlChar *test, xmlChar *report)
{
    xmlSchematronTestPtr ret;
    xmlXPathCompExprPtr comp;

    if ((ctxt == NULL) || (rule == NULL) || (node == NULL) ||
        (test == NULL))
        return(NULL);

    /*
     * try first to compile the test expression
     */
    comp = xmlXPathCtxtCompile(ctxt->xctxt, test);
    if (comp == NULL) {
	xmlSchematronPErr(ctxt, node,
	    XML_SCHEMAP_NOROOT,
	    "Failed to compile test expression %s",
	    test, NULL);
	return(NULL);
    }

    ret = (xmlSchematronTestPtr) xmlMalloc(sizeof(xmlSchematronTest));
    if (ret == NULL) {
        xmlSchematronPErrMemory(ctxt, "allocating schema test", node);
        return (NULL);
    }
    memset(ret, 0, sizeof(xmlSchematronTest));
    ret->type = type;
    ret->node = node;
    ret->test = test;
    ret->comp = comp;
    ret->report = report;
    ret->next = NULL;
    if (rule->tests == NULL) {
	rule->tests = ret;
    } else {
        xmlSchematronTestPtr prev = rule->tests;

	while (prev->next != NULL)
	     prev = prev->next;
        prev->next = ret;
    }
    return (ret);
}

/**
 * xmlSchematronFreeTests:
 * @tests:  a list of tests
 *
 * Free a list of tests.
 */
static void
xmlSchematronFreeTests(xmlSchematronTestPtr tests) {
    xmlSchematronTestPtr next;

    while (tests != NULL) {
        next = tests->next;
	if (tests->test != NULL)
	    xmlFree(tests->test);
	if (tests->comp != NULL)
	    xmlXPathFreeCompExpr(tests->comp);
	if (tests->report != NULL)
	    xmlFree(tests->report);
	xmlFree(tests);
	tests = next;
    }
}

/**
 * xmlSchematronAddRule:
 * @ctxt: the schema parsing context
 * @schema:  a schema structure
 * @node:  the node hosting the rule
 * @context: the associated context string
 * @report: the associated report string
 *
 * Add a rule to a schematron
 *
 * Returns the new pointer or NULL in case of error
 */
static xmlSchematronRulePtr
xmlSchematronAddRule(xmlSchematronParserCtxtPtr ctxt, xmlSchematronPtr schema,
                     xmlSchematronPatternPtr pat, xmlNodePtr node,
		     xmlChar *context, xmlChar *report)
{
    xmlSchematronRulePtr ret;
    xmlPatternPtr pattern;

    if ((ctxt == NULL) || (schema == NULL) || (node == NULL) ||
        (context == NULL))
        return(NULL);

    /*
     * Try first to compile the pattern
     */
    pattern = xmlPatterncompile(context, ctxt->dict, XML_PATTERN_XPATH,
                                ctxt->namespaces);
    if (pattern == NULL) {
	xmlSchematronPErr(ctxt, node,
	    XML_SCHEMAP_NOROOT,
	    "Failed to compile context expression %s",
	    context, NULL);
    }

    ret = (xmlSchematronRulePtr) xmlMalloc(sizeof(xmlSchematronRule));
    if (ret == NULL) {
        xmlSchematronPErrMemory(ctxt, "allocating schema rule", node);
        return (NULL);
    }
    memset(ret, 0, sizeof(xmlSchematronRule));
    ret->node = node;
    ret->context = context;
    ret->pattern = pattern;
    ret->report = report;
    ret->next = NULL;
    if (schema->rules == NULL) {
	schema->rules = ret;
    } else {
        xmlSchematronRulePtr prev = schema->rules;

	while (prev->next != NULL)
	     prev = prev->next;
        prev->next = ret;
    }
    ret->patnext = NULL;
    if (pat->rules == NULL) {
	pat->rules = ret;
    } else {
        xmlSchematronRulePtr prev = pat->rules;

	while (prev->patnext != NULL)
	     prev = prev->patnext;