pattern.c.svn-base

这是一个用于解析xml文件的类库。使用这个类库

/*
 * pattern.c: Implemetation of selectors for nodes
 *
 * Reference:
 *   http://www.w3.org/TR/2001/REC-xmlschema-1-20010502/
 *   to some extent 
 *   http://www.w3.org/TR/1999/REC-xml-19991116
 *
 * See Copyright for the status of this software.
 *
 * [e-mail]
 */

#define IN_LIBXML
#include "libxml.h"

#include <string.h>
#include <libxml/xmlmemory.h>
#include <libxml/tree.h>
#include <libxml/hash.h>
#include <libxml/dict.h>
#include <libxml/xmlerror.h>
#include <libxml/parserInternals.h>
#include <libxml/pattern.h>

#ifdef LIBXML_PATTERN_ENABLED

#define ERROR(a, b, c, d)
#define ERROR5(a, b, c, d, e)

/*
 * Types are private:
 */

typedef enum {
    XML_OP_END=0,
    XML_OP_ROOT,
    XML_OP_ELEM,
    XML_OP_CHILD,
    XML_OP_ATTR,
    XML_OP_PARENT,
    XML_OP_ANCESTOR,
    XML_OP_NS,
    XML_OP_ALL
} xmlPatOp;


typedef struct _xmlStepOp xmlStepOp;
typedef xmlStepOp *xmlStepOpPtr;
struct _xmlStepOp {
    xmlPatOp op;
    const xmlChar *value;
    const xmlChar *value2;
};

struct _xmlPattern {
    void *data;    		/* the associated template */
    struct _xmlPattern *next; /* siblings */
    const xmlChar *pattern;	/* the pattern */

    /* TODO fix the statically allocated size steps[] */
    int nbStep;
    int maxStep;
    xmlStepOp steps[10];        /* ops for computation */
};

typedef struct _xmlPatParserContext xmlPatParserContext;
typedef xmlPatParserContext *xmlPatParserContextPtr;
struct _xmlPatParserContext {
    const xmlChar *cur;			/* the current char being parsed */
    const xmlChar *base;		/* the full expression */
    int	           error;		/* error code */
    xmlDictPtr     dict;		/* the dictionnary if any */
    xmlPatternPtr  comp;		/* the result */
    xmlNodePtr     elem;		/* the current node if any */    
    const xmlChar **namespaces;		/* the namespaces definitions */
    int   nb_namespaces;		/* the number of namespaces */
};

/************************************************************************
 * 									*
 * 			Type functions 					*
 * 									*
 ************************************************************************/

/**
 * xmlNewPattern:
 *
 * Create a new XSLT Pattern
 *
 * Returns the newly allocated xmlPatternPtr or NULL in case of error
 */
static xmlPatternPtr
xmlNewPattern(void) {
    xmlPatternPtr cur;

    cur = (xmlPatternPtr) xmlMalloc(sizeof(xmlPattern));
    if (cur == NULL) {
	ERROR(NULL, NULL, NULL,
		"xmlNewPattern : malloc failed\n");
	return(NULL);
    }
    memset(cur, 0, sizeof(xmlPattern));
    cur->maxStep = 10;
    return(cur);
}

/**
 * xmlFreePattern:
 * @comp:  an XSLT comp
 *
 * Free up the memory allocated by @comp
 */
void
xmlFreePattern(xmlPatternPtr comp) {
    xmlStepOpPtr op;
    int i;

    if (comp == NULL)
	return;
    if (comp->pattern != NULL)
	xmlFree((xmlChar *)comp->pattern);
    for (i = 0;i < comp->nbStep;i++) {
	op = &comp->steps[i];
	if (op->value != NULL)
	    xmlFree((xmlChar *) op->value);
	if (op->value2 != NULL)
	    xmlFree((xmlChar *) op->value2);
    }
    memset(comp, -1, sizeof(xmlPattern));
    xmlFree(comp);
}

/**
 * xmlFreePatternList:
 * @comp:  an XSLT comp list
 *
 * Free up the memory allocated by all the elements of @comp
 */
void
xmlFreePatternList(xmlPatternPtr comp) {
    xmlPatternPtr cur;

    while (comp != NULL) {
	cur = comp;
	comp = comp->next;
	xmlFreePattern(cur);
    }
}

/**
 * xmlNewPatParserContext:
 * @pattern:  the pattern context
 * @dict:  the inherited dictionnary or NULL
 * @namespaces: the prefix definitions, array of [URI, prefix] or NULL
 *
 * Create a new XML pattern parser context
 *
 * Returns the newly allocated xmlPatParserContextPtr or NULL in case of error
 */
static xmlPatParserContextPtr
xmlNewPatParserContext(const xmlChar *pattern, xmlDictPtr dict,
                       const xmlChar **namespaces) {
    xmlPatParserContextPtr cur;

    if (pattern == NULL)
        return(NULL);

    cur = (xmlPatParserContextPtr) xmlMalloc(sizeof(xmlPatParserContext));
    if (cur == NULL) {
	ERROR(NULL, NULL, NULL,
		"xmlNewPatParserContext : malloc failed\n");
	return(NULL);
    }
    memset(cur, 0, sizeof(xmlPatParserContext));
    cur->dict = dict;
    cur->cur = pattern;
    cur->base = pattern;
    if (namespaces != NULL) {
        int i;
	for (i = 0;namespaces[2 * i] != NULL;i++);
        cur->nb_namespaces = i;
    } else {
        cur->nb_namespaces = 0;
    }
    cur->namespaces = namespaces;
    return(cur);
}

/**
 * xmlFreePatParserContext:
 * @ctxt:  an XSLT parser context
 *
 * Free up the memory allocated by @ctxt
 */
static void
xmlFreePatParserContext(xmlPatParserContextPtr ctxt) {
    if (ctxt == NULL)
	return;
    memset(ctxt, -1, sizeof(xmlPatParserContext));
    xmlFree(ctxt);
}

/**
 * xmlPatternAdd:
 * @comp:  the compiled match expression
 * @op:  an op
 * @value:  the first value
 * @value2:  the second value
 *
 * Add an step to an XSLT Compiled Match
 *
 * Returns -1 in case of failure, 0 otherwise.
 */
static int
xmlPatternAdd(xmlPatParserContextPtr ctxt ATTRIBUTE_UNUSED,
                xmlPatternPtr comp,
                xmlPatOp op, xmlChar * value, xmlChar * value2)
{
    if (comp->nbStep >= 10) {
        ERROR(ctxt, NULL, NULL,
                         "xmlPatternAdd: overflow\n");
        return (-1);
    }
    comp->steps[comp->nbStep].op = op;
    comp->steps[comp->nbStep].value = value;
    comp->steps[comp->nbStep].value2 = value2;
    comp->nbStep++;
    return (0);
}

#if 0
/**
 * xsltSwapTopPattern:
 * @comp:  the compiled match expression
 *
 * reverse the two top steps.
 */
static void
xsltSwapTopPattern(xmlPatternPtr comp) {
    int i;
    int j = comp->nbStep - 1;

    if (j > 0) {
	register const xmlChar *tmp;
	register xmlPatOp op;
	i = j - 1;
	tmp = comp->steps[i].value;
	comp->steps[i].value = comp->steps[j].value;
	comp->steps[j].value = tmp;
	tmp = comp->steps[i].value2;
	comp->steps[i].value2 = comp->steps[j].value2;
	comp->steps[j].value2 = tmp;
	op = comp->steps[i].op;
	comp->steps[i].op = comp->steps[j].op;
	comp->steps[j].op = op;
    }
}
#endif

/**
 * xmlReversePattern:
 * @comp:  the compiled match expression
 *
 * reverse all the stack of expressions
 */
static void
xmlReversePattern(xmlPatternPtr comp) {
    int i = 0;
    int j = comp->nbStep - 1;

    while (j > i) {
	register const xmlChar *tmp;
	register xmlPatOp op;
	tmp = comp->steps[i].value;
	comp->steps[i].value = comp->steps[j].value;
	comp->steps[j].value = tmp;
	tmp = comp->steps[i].value2;
	comp->steps[i].value2 = comp->steps[j].value2;
	comp->steps[j].value2 = tmp;
	op = comp->steps[i].op;
	comp->steps[i].op = comp->steps[j].op;
	comp->steps[j].op = op;
	j--;
	i++;
    }
    comp->steps[comp->nbStep++].op = XML_OP_END;
}

/************************************************************************
 * 									*
 * 		The interpreter for the precompiled patterns		*
 * 									*
 ************************************************************************/

/**
 * xmlPatMatch:
 * @comp: the precompiled pattern
 * @node: a node
 *
 * Test wether the node matches the pattern
 *
 * Returns 1 if it matches, 0 if it doesn't and -1 in case of failure
 */
static int
xmlPatMatch(xmlPatternPtr comp, xmlNodePtr node) {
    int i;
    xmlStepOpPtr step;

    if ((comp == NULL) || (node == NULL)) return(-1);
    for (i = 0;i < comp->nbStep;i++) {
	step = &comp->steps[i];
	switch (step->op) {
            case XML_OP_END:
		return(1);
            case XML_OP_ROOT:
		if ((node->type == XML_DOCUMENT_NODE) ||
#ifdef LIBXML_DOCB_ENABLED
		    (node->type == XML_DOCB_DOCUMENT_NODE) ||
#endif
		    (node->type == XML_HTML_DOCUMENT_NODE))
		    continue;
		return(0);
            case XML_OP_ELEM:
		if (node->type != XML_ELEMENT_NODE)
		    return(0);
		if (step->value == NULL)
		    continue;
		if (step->value[0] != node->name[0])
		    return(0);
		if (!xmlStrEqual(step->value, node->name))
		    return(0);

		/* Namespace test */
		if (node->ns == NULL) {
		    if (step->value2 != NULL)
			return(0);
		} else if (node->ns->href != NULL) {
		    if (step->value2 == NULL)
			return(0);
		    if (!xmlStrEqual(step->value2, node->ns->href))
			return(0);
		}
		continue;
            case XML_OP_CHILD: {
		xmlNodePtr lst;

		if ((node->type != XML_ELEMENT_NODE) &&
		    (node->type != XML_DOCUMENT_NODE) &&
#ifdef LIBXML_DOCB_ENABLED
		    (node->type != XML_DOCB_DOCUMENT_NODE) &&
#endif
		    (node->type != XML_HTML_DOCUMENT_NODE))
		    return(0);

		lst = node->children;

		if (step->value != NULL) {
		    while (lst != NULL) {
			if ((lst->type == XML_ELEMENT_NODE) &&
			    (step->value[0] == lst->name[0]) &&
			    (xmlStrEqual(step->value, lst->name)))
			    break;
			lst = lst->next;
		    }
		    if (lst != NULL)
			continue;
		}
		return(0);
	    }
            case XML_OP_ATTR:
		if (node->type != XML_ATTRIBUTE_NODE)
		    return(0);
		if (step->value != NULL) {
		    if (step->value[0] != node->name[0])
			return(0);
		    if (!xmlStrEqual(step->value, node->name))
			return(0);
		}
		/* Namespace test */
		if (node->ns == NULL) {
		    if (step->value2 != NULL)
			return(0);
		} else if (step->value2 != NULL) {
		    if (!xmlStrEqual(step->value2, node->ns->href))
			return(0);
		}
		continue;
            case XML_OP_PARENT:
		if ((node->type == XML_DOCUMENT_NODE) ||
		    (node->type == XML_HTML_DOCUMENT_NODE) ||
#ifdef LIBXML_DOCB_ENABLED
		    (node->type == XML_DOCB_DOCUMENT_NODE) ||
#endif
		    (node->type == XML_NAMESPACE_DECL))
		    return(0);
		node = node->parent;
		if (node == NULL)
		    return(0);
		if (step->value == NULL)
		    continue;
		if (step->value[0] != node->name[0])
		    return(0);
		if (!xmlStrEqual(step->value, node->name))
		    return(0);
		/* Namespace test */
		if (node->ns == NULL) {
		    if (step->value2 != NULL)
			return(0);
		} else if (node->ns->href != NULL) {
		    if (step->value2 == NULL)
			return(0);
		    if (!xmlStrEqual(step->value2, node->ns->href))
			return(0);
		}
		continue;
            case XML_OP_ANCESTOR:
		/* TODO: implement coalescing of ANCESTOR/NODE ops */
		if (step->value == NULL) {
		    i++;
		    step = &comp->steps[i];
		    if (step->op == XML_OP_ROOT)
			return(1);
		    if (step->op != XML_OP_ELEM)
			return(0);
		    if (step->value == NULL)
			return(-1);
		}
		if (node == NULL)
		    return(0);
		if ((node->type == XML_DOCUMENT_NODE) ||
		    (node->type == XML_HTML_DOCUMENT_NODE) ||
#ifdef LIBXML_DOCB_ENABLED
		    (node->type == XML_DOCB_DOCUMENT_NODE) ||
#endif
		    (node->type == XML_NAMESPACE_DECL))
		    return(0);
		node = node->parent;
		while (node != NULL) {
		    if (node == NULL)
			return(0);
		    if ((node->type == XML_ELEMENT_NODE) &&
			(step->value[0] == node->name[0]) &&
			(xmlStrEqual(step->value, node->name))) {
			/* Namespace test */
			if (node->ns == NULL) {
			    if (step->value2 == NULL)
				break;
			} else if (node->ns->href != NULL) {
			    if ((step->value2 != NULL) &&
			        (xmlStrEqual(step->value2, node->ns->href)))
				break;
			}
		    }
		    node = node->parent;
		}
		if (node == NULL)
		    return(0);
		continue;
            case XML_OP_NS:
		if (node->type != XML_ELEMENT_NODE)
		    return(0);
		if (node->ns == NULL) {
		    if (step->value != NULL)
			return(0);
		} else if (node->ns->href != NULL) {
		    if (step->value == NULL)