runtest.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.

/*
 * runtest.c: C program to run libxml2 regression tests without
 *            requiring make or Python, and reducing platform dependancies
 *            to a strict minimum.
 *
 * To compile on Unixes:
 * cc -o runtest `xml2-config --cflags` runtest.c `xml2-config --libs` -lpthread
 *
 * See Copyright for the status of this software.
 *
 * [e-mail]
 */

#ifdef HAVE_CONFIG_H
#include "libxml.h"
#else
#include <stdio.h>
#endif

#if !defined(_WIN32) || defined(__CYGWIN__)
#include <unistd.h>
#endif
#include <string.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>

#include <libxml/parser.h>
#include <libxml/tree.h>
#include <libxml/uri.h>

#ifdef LIBXML_OUTPUT_ENABLED
#ifdef LIBXML_READER_ENABLED
#include <libxml/xmlreader.h>
#endif

#ifdef LIBXML_XINCLUDE_ENABLED
#include <libxml/xinclude.h>
#endif

#ifdef LIBXML_XPATH_ENABLED
#include <libxml/xpath.h>
#include <libxml/xpathInternals.h>
#ifdef LIBXML_XPTR_ENABLED
#include <libxml/xpointer.h>
#endif
#endif

#ifdef LIBXML_SCHEMAS_ENABLED
#include <libxml/relaxng.h>
#include <libxml/xmlschemas.h>
#include <libxml/xmlschemastypes.h>
#endif

#ifdef LIBXML_PATTERN_ENABLED
#include <libxml/pattern.h>
#endif

#ifdef LIBXML_C14N_ENABLED
#include <libxml/c14n.h>
#endif

#ifdef LIBXML_HTML_ENABLED
#include <libxml/HTMLparser.h>
#include <libxml/HTMLtree.h>

/*
 * pseudo flag for the unification of HTML and XML tests
 */
#define XML_PARSE_HTML 1 << 24
#endif

#if defined(LIBXML_THREAD_ENABLED) && defined(LIBXML_CATALOG_ENABLED)
#include <libxml/globals.h>
#include <libxml/threads.h>
#include <libxml/parser.h>
#include <libxml/catalog.h>
#include <string.h>
#endif

/*
 * O_BINARY is just for Windows compatibility - if it isn't defined
 * on this system, avoid any compilation error
 */
#ifdef	O_BINARY
#define RD_FLAGS	O_RDONLY | O_BINARY
#else
#define	RD_FLAGS	O_RDONLY
#endif

typedef int (*functest) (const char *filename, const char *result,
                         const char *error, int options);

typedef struct testDesc testDesc;
typedef testDesc *testDescPtr;
struct testDesc {
    const char *desc; /* descripton of the test */
    functest    func; /* function implementing the test */
    const char *in;   /* glob to path for input files */
    const char *out;  /* output directory */
    const char *suffix;/* suffix for output files */
    const char *err;  /* suffix for error output files */
    int     options;  /* parser options for the test */
};

static int checkTestFile(const char *filename);

#if defined(_WIN32) && !defined(__CYGWIN__)

#include <windows.h>
#include <io.h>

typedef struct
{
      size_t gl_pathc;    /* Count of paths matched so far  */
      char **gl_pathv;    /* List of matched pathnames.  */
      size_t gl_offs;     /* Slots to reserve in 'gl_pathv'.  */
} glob_t;

#define GLOB_DOOFFS 0
static int glob(const char *pattern, int flags,
                int errfunc(const char *epath, int eerrno),
                glob_t *pglob) {
    glob_t *ret;
    WIN32_FIND_DATA FindFileData;
    HANDLE hFind;
    unsigned int nb_paths = 0;
    char directory[500];
    int len;

    if ((pattern == NULL) || (pglob == NULL)) return(-1);
    
    strncpy(directory, pattern, 499);
    for (len = strlen(directory);len >= 0;len--) {
        if (directory[len] == '/') {
	    len++;
	    directory[len] = 0;
	    break;
	}
    }
    if (len <= 0)
        len = 0;

    
    ret = pglob;
    memset(ret, 0, sizeof(glob_t));
    
    hFind = FindFirstFileA(pattern, &FindFileData);
    if (hFind == INVALID_HANDLE_VALUE) 
        return(0);
    nb_paths = 20;
    ret->gl_pathv = (char **) malloc(nb_paths * sizeof(char *));
    if (ret->gl_pathv == NULL) {
	FindClose(hFind);
        return(-1);
    }
    strncpy(directory + len, FindFileData.cFileName, 499 - len);
    ret->gl_pathv[ret->gl_pathc] = strdup(directory);
    if (ret->gl_pathv[ret->gl_pathc] == NULL)
        goto done;
    ret->gl_pathc++;
    while(FindNextFileA(hFind, &FindFileData)) {
        if (FindFileData.cFileName[0] == '.')
	    continue;
        if (ret->gl_pathc + 2 > nb_paths) {
            char **tmp = realloc(ret->gl_pathv, nb_paths * 2 * sizeof(char *));
            if (tmp == NULL)
                break;
            ret->gl_pathv = tmp;
            nb_paths *= 2;
	}
	strncpy(directory + len, FindFileData.cFileName, 499 - len);
	ret->gl_pathv[ret->gl_pathc] = strdup(directory);
        if (ret->gl_pathv[ret->gl_pathc] == NULL)
            break;
        ret->gl_pathc++;
    }
    ret->gl_pathv[ret->gl_pathc] = NULL;

done:
    FindClose(hFind);
    return(0);
}
 


static void globfree(glob_t *pglob) {
    unsigned int i;
    if (pglob == NULL)
        return;
    
    for (i = 0;i < pglob->gl_pathc;i++) {
         if (pglob->gl_pathv[i] != NULL)
             free(pglob->gl_pathv[i]);
    }
}
#define vsnprintf _vsnprintf
#define snprintf _snprintf
#else
#include <glob.h>
#endif

/************************************************************************
 *									*
 *		Libxml2 specific routines				*
 *									*
 ************************************************************************/

static int nb_tests = 0;
static int nb_errors = 0;
static int nb_leaks = 0;
static int extraMemoryFromResolver = 0;

static int
fatalError(void) {
    fprintf(stderr, "Exitting tests on fatal error\n");
    exit(1);
}

/*
 * We need to trap calls to the resolver to not account memory for the catalog
 * which is shared to the current running test. We also don't want to have
 * network downloads modifying tests.
 */
static xmlParserInputPtr 
testExternalEntityLoader(const char *URL, const char *ID,
			 xmlParserCtxtPtr ctxt) {
    xmlParserInputPtr ret;

    if (checkTestFile(URL)) {
	ret = xmlNoNetExternalEntityLoader(URL, ID, ctxt);
    } else {
	int memused = xmlMemUsed();
	ret = xmlNoNetExternalEntityLoader(URL, ID, ctxt);
	extraMemoryFromResolver += xmlMemUsed() - memused;
    }
      
    return(ret);
}

/*
 * Trapping the error messages at the generic level to grab the equivalent of
 * stderr messages on CLI tools.
 */
static char testErrors[32769];
static int testErrorsSize = 0;

static void XMLCDECL
testErrorHandler(void *ctx  ATTRIBUTE_UNUSED, const char *msg, ...) {
    va_list args;
    int res;

    if (testErrorsSize >= 32768)
        return;
    va_start(args, msg);
    res = vsnprintf(&testErrors[testErrorsSize],
                    32768 - testErrorsSize,
		    msg, args);
    va_end(args);
    if (testErrorsSize + res >= 32768) {
        /* buffer is full */
	testErrorsSize = 32768;
	testErrors[testErrorsSize] = 0;
    } else {
        testErrorsSize += res;
    }
    testErrors[testErrorsSize] = 0;
}

static void XMLCDECL
channel(void *ctx  ATTRIBUTE_UNUSED, const char *msg, ...) {
    va_list args;
    int res;

    if (testErrorsSize >= 32768)
        return;
    va_start(args, msg);
    res = vsnprintf(&testErrors[testErrorsSize],
                    32768 - testErrorsSize,
		    msg, args);
    va_end(args);
    if (testErrorsSize + res >= 32768) {
        /* buffer is full */
	testErrorsSize = 32768;
	testErrors[testErrorsSize] = 0;
    } else {
        testErrorsSize += res;
    }
    testErrors[testErrorsSize] = 0;
}

/**
 * xmlParserPrintFileContext:
 * @input:  an xmlParserInputPtr input
 * 
 * Displays current context within the input content for error tracking
 */

static void
xmlParserPrintFileContextInternal(xmlParserInputPtr input , 
		xmlGenericErrorFunc chanl, void *data ) {
    const xmlChar *cur, *base;
    unsigned int n, col;	/* GCC warns if signed, because compared with sizeof() */
    xmlChar  content[81]; /* space for 80 chars + line terminator */
    xmlChar *ctnt;

    if (input == NULL) return;
    cur = input->cur;
    base = input->base;
    /* skip backwards over any end-of-lines */
    while ((cur > base) && ((*(cur) == '\n') || (*(cur) == '\r'))) {
	cur--;
    }
    n = 0;
    /* search backwards for beginning-of-line (to max buff size) */
    while ((n++ < (sizeof(content)-1)) && (cur > base) && 
    	   (*(cur) != '\n') && (*(cur) != '\r'))
        cur--;
    if ((*(cur) == '\n') || (*(cur) == '\r')) cur++;
    /* calculate the error position in terms of the current position */
    col = input->cur - cur;
    /* search forward for end-of-line (to max buff size) */
    n = 0;
    ctnt = content;
    /* copy selected text to our buffer */
    while ((*cur != 0) && (*(cur) != '\n') && 
    	   (*(cur) != '\r') && (n < sizeof(content)-1)) {
		*ctnt++ = *cur++;
	n++;
    }
    *ctnt = 0;
    /* print out the selected text */
    chanl(data ,"%s\n", content);
    /* create blank line with problem pointer */
    n = 0;
    ctnt = content;
    /* (leave buffer space for pointer + line terminator) */
    while ((n<col) && (n++ < sizeof(content)-2) && (*ctnt != 0)) {
	if (*(ctnt) != '\t')
	    *(ctnt) = ' ';
	ctnt++;
    }
    *ctnt++ = '^';
    *ctnt = 0;
    chanl(data ,"%s\n", content);
}

static void
testStructuredErrorHandler(void *ctx  ATTRIBUTE_UNUSED, xmlErrorPtr err) {
    char *file = NULL;
    int line = 0;
    int code = -1;
    int domain;
    void *data = NULL;
    const char *str;
    const xmlChar *name = NULL;
    xmlNodePtr node;
    xmlErrorLevel level;
    xmlParserInputPtr input = NULL;
    xmlParserInputPtr cur = NULL;
    xmlParserCtxtPtr ctxt = NULL;

    if (err == NULL)
        return;

    file = err->file;
    line = err->line;
    code = err->code;
    domain = err->domain;
    level = err->level;
    node = err->node;
    if ((domain == XML_FROM_PARSER) || (domain == XML_FROM_HTML) ||
        (domain == XML_FROM_DTD) || (domain == XML_FROM_NAMESPACE) ||
	(domain == XML_FROM_IO) || (domain == XML_FROM_VALID)) {
	ctxt = err->ctxt;
    }
    str = err->message;

    if (code == XML_ERR_OK)
        return;

    if ((node != NULL) && (node->type == XML_ELEMENT_NODE))
        name = node->name;

    /*
     * Maintain the compatibility with the legacy error handling
     */
    if (ctxt != NULL) {
        input = ctxt->input;
        if ((input != NULL) && (input->filename == NULL) &&
            (ctxt->inputNr > 1)) {
            cur = input;
            input = ctxt->inputTab[ctxt->inputNr - 2];
        }
        if (input != NULL) {
            if (input->filename)
                channel(data, "%s:%d: ", input->filename, input->line);
            else if ((line != 0) && (domain == XML_FROM_PARSER))
                channel(data, "Entity: line %d: ", input->line);
        }
    } else {
        if (file != NULL)
            channel(data, "%s:%d: ", file, line);
        else if ((line != 0) && (domain == XML_FROM_PARSER))
            channel(data, "Entity: line %d: ", line);
    }
    if (name != NULL) {
        channel(data, "element %s: ", name);
    }
    if (code == XML_ERR_OK)
        return;
    switch (domain) {
        case XML_FROM_PARSER:
            channel(data, "parser ");
            break;
        case XML_FROM_NAMESPACE:
            channel(data, "namespace ");
            break;
        case XML_FROM_DTD:
        case XML_FROM_VALID:
            channel(data, "validity ");
            break;
        case XML_FROM_HTML:
            channel(data, "HTML parser ");
            break;
        case XML_FROM_MEMORY:
            channel(data, "memory ");
            break;
        case XML_FROM_OUTPUT:
            channel(data, "output ");
            break;
        case XML_FROM_IO:
            channel(data, "I/O ");
            break;
        case XML_FROM_XINCLUDE:
            channel(data, "XInclude ");
            break;
        case XML_FROM_XPATH:
            channel(data, "XPath ");
            break;
        case XML_FROM_XPOINTER:
            channel(data, "parser ");
            break;
        case XML_FROM_REGEXP:
            channel(data, "regexp ");
            break;
        case XML_FROM_MODULE:
            channel(data, "module ");
            break;
        case XML_FROM_SCHEMASV:
            channel(data, "Schemas validity ");
            break;
        case XML_FROM_SCHEMASP:
            channel(data, "Schemas parser ");
            break;
        case XML_FROM_RELAXNGP:
            channel(data, "Relax-NG parser ");
            break;
        case XML_FROM_RELAXNGV:
            channel(data, "Relax-NG validity ");
            break;
        case XML_FROM_CATALOG:
            channel(data, "Catalog ");
            break;
        case XML_FROM_C14N:
            channel(data, "C14N ");
            break;
        case XML_FROM_XSLT:
            channel(data, "XSLT ");
            break;
        default:
            break;
    }
    if (code == XML_ERR_OK)
        return;
    switch (level) {
        case XML_ERR_NONE:
            channel(data, ": ");
            break;
        case XML_ERR_WARNING:
            channel(data, "warning : ");
            break;
        case XML_ERR_ERROR:
            channel(data, "error : ");
            break;
        case XML_ERR_FATAL:
            channel(data, "error : ");
            break;
    }
    if (code == XML_ERR_OK)
        return;
    if (str != NULL) {
        int len;
	len = xmlStrlen((const xmlChar *)str);
	if ((len > 0) && (str[len - 1] != '\n'))
	    channel(data, "%s\n", str);
	else
	    channel(data, "%s", str);
    } else {
        channel(data, "%s\n", "out of memory error");
    }
    if (code == XML_ERR_OK)
        return;

    if (ctxt != NULL) {
        xmlParserPrintFileContextInternal(input, channel, data);
        if (cur != NULL) {
            if (cur->filename)
                channel(data, "%s:%d: \n", cur->filename, cur->line);
            else if ((line != 0) && (domain == XML_FROM_PARSER))
                channel(data, "Entity: line %d: \n", cur->line);
            xmlParserPrintFileContextInternal(cur, channel, data);
        }
    }
    if ((domain == XML_FROM_XPATH) && (err->str1 != NULL) &&
        (err->int1 < 100) &&
	(err->int1 < xmlStrlen((const xmlChar *)err->str1))) {
	xmlChar buf[150];
	int i;

	channel(data, "%s\n", err->str1);
	for (i=0;i < err->int1;i++)
	     buf[i] = ' ';
	buf[i++] = '^';
	buf[i] = 0;
	channel(data, "%s\n", buf);
    }
}