gentest.py

#!/usr/bin/env python3
#
# generate a tester program for the API
#
import sys
import os
import string
try:
    import libxml2
except:
    print("libxml2 python bindings not available, skipping testapi.c generation")
    sys.exit(0)

if len(sys.argv) > 1:
    srcPref = sys.argv[1] + '/'
else:
    srcPref = ''

#
# Modules we want to skip in API test
#
skipped_modules = [ "SAX", "xlink", "threads", "globals",
  "xmlmemory", "xmlversion", "xmlexports",
]

#
# defines for each module
#
modules_defines = {
    "HTMLparser": "LIBXML_HTML_ENABLED",
    "catalog": "LIBXML_CATALOG_ENABLED",
    "xmlreader": "LIBXML_READER_ENABLED",
    "relaxng": "LIBXML_SCHEMAS_ENABLED",
    "schemasInternals": "LIBXML_SCHEMAS_ENABLED",
    "xmlschemas": "LIBXML_SCHEMAS_ENABLED",
    "xmlschemastypes": "LIBXML_SCHEMAS_ENABLED",
    "xpath": "LIBXML_XPATH_ENABLED",
    "xpathInternals": "LIBXML_XPATH_ENABLED",
    "xinclude": "LIBXML_XINCLUDE_ENABLED",
    "xpointer": "LIBXML_XPTR_ENABLED",
    "xmlregexp" : "LIBXML_REGEXP_ENABLED",
    "xmlautomata" : "LIBXML_AUTOMATA_ENABLED",
    "xmlsave" : "LIBXML_OUTPUT_ENABLED",
    "xmlmodule" : "LIBXML_MODULES_ENABLED",
    "pattern" : "LIBXML_PATTERN_ENABLED",
    "schematron" : "LIBXML_SCHEMATRON_ENABLED",
}

#
# defines for specific functions
#
function_defines = {
    "htmlDefaultSAXHandlerInit": "LIBXML_HTML_ENABLED",
    "xmlSAX2EndElement" : "LIBXML_SAX1_ENABLED",
    "xmlSAX2StartElement" : "LIBXML_SAX1_ENABLED",
    "xmlSAXDefaultVersion" : "LIBXML_SAX1_ENABLED",
    "UTF8Toisolat1" : "LIBXML_OUTPUT_ENABLED",
    "xmlIOParseDTD": "LIBXML_VALID_ENABLED",
    "xmlParseDTD": "LIBXML_VALID_ENABLED",
    "xmlParseDoc": "LIBXML_SAX1_ENABLED",
    "xmlParseMemory": "LIBXML_SAX1_ENABLED",
    "xmlRecoverDoc": "LIBXML_SAX1_ENABLED",
    "xmlParseFile": "LIBXML_SAX1_ENABLED",
    "xmlRecoverFile": "LIBXML_SAX1_ENABLED",
    "xmlRecoverMemory": "LIBXML_SAX1_ENABLED",
    "xmlSAXParseFileWithData": "LIBXML_SAX1_ENABLED",
    "xmlSAXParseMemory": "LIBXML_SAX1_ENABLED",
    "xmlSAXUserParseMemory": "LIBXML_SAX1_ENABLED",
    "xmlSAXParseDoc": "LIBXML_SAX1_ENABLED",
    "xmlSAXParseDTD": "LIBXML_SAX1_ENABLED",
    "xmlSAXUserParseFile": "LIBXML_SAX1_ENABLED",
    "xmlParseEntity": "LIBXML_SAX1_ENABLED",
    "xmlParseExternalEntity": "LIBXML_SAX1_ENABLED",
    "xmlSAXParseMemoryWithData": "LIBXML_SAX1_ENABLED",
    "xmlParseBalancedChunkMemory": "LIBXML_SAX1_ENABLED",
    "xmlParseBalancedChunkMemoryRecover": "LIBXML_SAX1_ENABLED",
    "xmlSetupParserForBuffer": "LIBXML_SAX1_ENABLED",
    "xmlStopParser": "LIBXML_PUSH_ENABLED",
    "xmlAttrSerializeTxtContent": "LIBXML_OUTPUT_ENABLED",
    "xmlSAXParseFile": "LIBXML_SAX1_ENABLED",
    "xmlSAXParseEntity": "LIBXML_SAX1_ENABLED",
    "xmlNewTextChild": "LIBXML_TREE_ENABLED",
    "xmlNewDocRawNode": "LIBXML_TREE_ENABLED",
    "xmlNewProp": "LIBXML_TREE_ENABLED",
    "xmlReconciliateNs": "LIBXML_TREE_ENABLED",
    "xmlValidateNCName": "LIBXML_TREE_ENABLED",
    "xmlValidateNMToken": "LIBXML_TREE_ENABLED",
    "xmlValidateName": "LIBXML_TREE_ENABLED",
    "xmlNewChild": "LIBXML_TREE_ENABLED",
    "xmlValidateQName": "LIBXML_TREE_ENABLED",
    "xmlSprintfElementContent": "LIBXML_OUTPUT_ENABLED",
    "xmlValidGetPotentialChildren" : "LIBXML_VALID_ENABLED",
    "xmlValidGetValidElements" : "LIBXML_VALID_ENABLED",
    "xmlTextReaderPreservePattern" : "LIBXML_PATTERN_ENABLED",
}

#
# Some functions really need to be skipped for the tests.
#
skipped_functions = [
# block on I/O
"xmlFdRead", "xmlReadFd", "xmlCtxtReadFd",
"htmlFdRead", "htmlReadFd", "htmlCtxtReadFd",
"xmlReaderNewFd", "xmlReaderForFd",
"xmlIORead", "xmlReadIO", "xmlCtxtReadIO",
"htmlIORead", "htmlReadIO", "htmlCtxtReadIO",
"xmlReaderNewIO", "xmlBufferDump", "xmlNanoFTPConnect",
"xmlNanoFTPConnectTo", "xmlNanoHTTPMethod", "xmlNanoHTTPMethodRedir",
# Complex I/O APIs
"xmlCreateIOParserCtxt", "xmlParserInputBufferCreateIO",
"xmlRegisterInputCallbacks", "xmlReaderForIO",
"xmlOutputBufferCreateIO", "xmlRegisterOutputCallbacks",
"xmlSaveToIO", "xmlIOHTTPOpenW",
# library state cleanup, generate false leak information and other
# troubles, heavillyb tested otherwise.
"xmlCleanupParser", "xmlRelaxNGCleanupTypes", "xmlSetListDoc",
"xmlSetTreeDoc", "xmlUnlinkNode",
# hard to avoid leaks in the tests
"xmlStrcat", "xmlStrncat", "xmlCatalogAddLocal", "xmlNewTextWriterDoc",
"xmlXPathNewValueTree", "xmlXPathWrapString",
# unimplemented
"xmlTextReaderReadInnerXml", "xmlTextReaderReadOuterXml",
"xmlTextReaderReadString",
# destructor
"xmlListDelete", "xmlOutputBufferClose", "xmlNanoFTPClose", "xmlNanoHTTPClose",
# deprecated
"xmlCatalogGetPublic", "xmlCatalogGetSystem", "xmlEncodeEntities",
"xmlNewGlobalNs", "xmlHandleEntity", "xmlNamespaceParseNCName",
"xmlNamespaceParseNSDef", "xmlNamespaceParseQName",
"xmlParseNamespace", "xmlParseQuotedString", "xmlParserHandleReference",
"xmlScanName",
"xmlDecodeEntities", 
# allocators
"xmlMemFree",
# verbosity
"xmlCatalogSetDebug", "xmlShellPrintXPathError", "xmlShellPrintNode",
# Internal functions, no user space should really call them
"xmlParseAttribute", "xmlParseAttributeListDecl", "xmlParseName",
"xmlParseNmtoken", "xmlParseEntityValue", "xmlParseAttValue",
"xmlParseSystemLiteral", "xmlParsePubidLiteral", "xmlParseCharData",
"xmlParseExternalID", "xmlParseComment", "xmlParsePITarget", "xmlParsePI",
"xmlParseNotationDecl", "xmlParseEntityDecl", "xmlParseDefaultDecl",
"xmlParseNotationType", "xmlParseEnumerationType", "xmlParseEnumeratedType",
"xmlParseAttributeType", "xmlParseAttributeListDecl",
"xmlParseElementMixedContentDecl", "xmlParseElementChildrenContentDecl",
"xmlParseElementContentDecl", "xmlParseElementDecl", "xmlParseMarkupDecl",
"xmlParseCharRef", "xmlParseEntityRef", "xmlParseReference",
"xmlParsePEReference", "xmlParseDocTypeDecl", "xmlParseAttribute",
"xmlParseStartTag", "xmlParseEndTag", "xmlParseCDSect", "xmlParseContent",
"xmlParseElement", "xmlParseVersionNum", "xmlParseVersionInfo",
"xmlParseEncName", "xmlParseEncodingDecl", "xmlParseSDDecl",
"xmlParseXMLDecl", "xmlParseTextDecl", "xmlParseMisc",
"xmlParseExternalSubset", "xmlParserHandlePEReference",
"xmlSkipBlankChars",
# Legacy
"xmlCleanupPredefinedEntities", "xmlInitializePredefinedEntities",
"xmlSetFeature", "xmlGetFeature", "xmlGetFeaturesList",
# location sets
"xmlXPtrLocationSetAdd",
"xmlXPtrLocationSetCreate",
"xmlXPtrLocationSetDel",
"xmlXPtrLocationSetMerge",
"xmlXPtrLocationSetRemove",
"xmlXPtrWrapLocationSet",
]

#
# These functions have side effects on the global state
# and hence generate errors on memory allocation tests
#
skipped_memcheck = [ "xmlLoadCatalog", "xmlAddEncodingAlias",
   "xmlSchemaInitTypes", "xmlNanoFTPProxy", "xmlNanoFTPScanProxy",
   "xmlNanoHTTPScanProxy", "xmlResetLastError", "xmlCatalogConvert",
   "xmlCatalogRemove", "xmlLoadCatalogs", "xmlCleanupCharEncodingHandlers",
   "xmlInitCharEncodingHandlers", "xmlCatalogCleanup",
   "xmlSchemaGetBuiltInType",
   "htmlParseFile", "htmlCtxtReadFile", # loads the catalogs
   "xmlTextReaderSchemaValidate", "xmlSchemaCleanupTypes", # initialize the schemas type system
   "xmlCatalogResolve", "xmlIOParseDTD" # loads the catalogs
]

#
# Extra code needed for some test cases
#
extra_pre_call = {
   "xmlSAXUserParseFile": """
#ifdef LIBXML_SAX1_ENABLED
        if (sax == (xmlSAXHandlerPtr)&xmlDefaultSAXHandler) user_data = NULL;
#endif
""",
   "xmlSAXUserParseMemory": """
#ifdef LIBXML_SAX1_ENABLED
        if (sax == (xmlSAXHandlerPtr)&xmlDefaultSAXHandler) user_data = NULL;
#endif
""",
   "xmlParseBalancedChunkMemory": """
#ifdef LIBXML_SAX1_ENABLED
        if (sax == (xmlSAXHandlerPtr)&xmlDefaultSAXHandler) user_data = NULL;
#endif
""",
   "xmlParseBalancedChunkMemoryRecover": """
#ifdef LIBXML_SAX1_ENABLED
        if (sax == (xmlSAXHandlerPtr)&xmlDefaultSAXHandler) user_data = NULL;
#endif
""",
   "xmlParserInputBufferCreateFd":
       "if (fd >= 0) fd = -1;",
   "xmlSAXDefaultVersion": """
        {
            int original_version = xmlSAXDefaultVersion(2);
""",
}
extra_post_call = {
   "xmlAddChild": 
       "if (ret_val == NULL) { xmlFreeNode(cur) ; cur = NULL ; }",
   "xmlAddEntity":
       "if (ret_val != NULL) { xmlFreeNode(ret_val) ; ret_val = NULL; }",
   "xmlAddChildList": 
       "if (ret_val == NULL) { xmlFreeNodeList(cur) ; cur = NULL ; }",
   "xmlAddSibling":
       "if (ret_val == NULL) { xmlFreeNode(elem) ; elem = NULL ; }",
   "xmlAddNextSibling":
       "if (ret_val == NULL) { xmlFreeNode(elem) ; elem = NULL ; }",
   "xmlAddPrevSibling": 
       "if (ret_val == NULL) { xmlFreeNode(elem) ; elem = NULL ; }",
   "xmlDocSetRootElement": 
       "if (doc == NULL) { xmlFreeNode(root) ; root = NULL ; }",
   "xmlReplaceNode": 
       """if (cur != NULL) {
              xmlUnlinkNode(cur);
              xmlFreeNode(cur) ; cur = NULL ; }
          if (old != NULL) {
              xmlUnlinkNode(old);
              xmlFreeNode(old) ; old = NULL ; }
\t  ret_val = NULL;""",
   "xmlTextMerge": 
       """if (ret_val == NULL) {
              xmlUnlinkNode(second);
              xmlFreeNode(second) ; second = NULL ;
              ret_val = first; }""",
   "xmlBuildQName": 
       """if ((ret_val != NULL) && (ret_val != ncname) &&
              (ret_val != prefix) && (ret_val != memory))
              xmlFree(ret_val);
\t  ret_val = NULL;""",
   "xmlNewDocElementContent":
       """xmlFreeDocElementContent(doc, ret_val); ret_val = NULL;""",
   "xmlDictReference": "xmlDictFree(dict);",
   # Functions which deallocates one of their parameters
   "xmlXPathConvertBoolean": """val = NULL;""",
   "xmlXPathConvertNumber": """val = NULL;""",
   "xmlXPathConvertString": """val = NULL;""",
   "xmlSaveFileTo": """buf = NULL;""",
   "xmlSaveFormatFileTo": """buf = NULL;""",
   "xmlIOParseDTD": "input = NULL;",
   "xmlRemoveProp": "cur = NULL;",
   "xmlNewNs": "if ((node == NULL) && (ret_val != NULL)) xmlFreeNs(ret_val);",
   "xmlCopyNamespace": "if (ret_val != NULL) xmlFreeNs(ret_val);",
   "xmlCopyNamespaceList": "if (ret_val != NULL) xmlFreeNsList(ret_val);",
   "xmlNewTextWriter": "if (ret_val != NULL) out = NULL;",
   "xmlNewTextWriterPushParser": "if (ctxt != NULL) {xmlFreeDoc(ctxt->myDoc); ctxt->myDoc = NULL;} if (ret_val != NULL) ctxt = NULL;",
   "xmlNewIOInputStream": "if (ret_val != NULL) input = NULL;",
   "htmlParseChunk": "if (ctxt != NULL) {xmlFreeDoc(ctxt->myDoc); ctxt->myDoc = NULL;}",
   "htmlParseDocument": "if (ctxt != NULL) {xmlFreeDoc(ctxt->myDoc); ctxt->myDoc = NULL;}",
   "xmlParseDocument": "if (ctxt != NULL) {xmlFreeDoc(ctxt->myDoc); ctxt->myDoc = NULL;}",
   "xmlParseChunk": "if (ctxt != NULL) {xmlFreeDoc(ctxt->myDoc); ctxt->myDoc = NULL;}",
   "xmlParseExtParsedEnt": "if (ctxt != NULL) {xmlFreeDoc(ctxt->myDoc); ctxt->myDoc = NULL;}",
   "xmlDOMWrapAdoptNode": "if ((node != NULL) && (node->parent == NULL)) {xmlUnlinkNode(node);xmlFreeNode(node);node = NULL;}",
   "xmlSAXDefaultVersion": """
            (void)xmlSAXDefaultVersion(original_version);
        }
""",
}

modules = []

def is_skipped_module(name):
    for mod in skipped_modules:
        if mod == name:
            return 1
    return 0

def is_skipped_function(name):
    for fun in skipped_functions:
        if fun == name:
            return 1
    # Do not test destructors
    if name.find('Free') != -1:
        return 1
    return 0

def is_skipped_memcheck(name):
    for fun in skipped_memcheck:
        if fun == name:
            return 1
    return 0

missing_types = {}
def add_missing_type(name, func):
    try:
        list = missing_types[name]
        list.append(func)
    except:
        missing_types[name] = [func]

generated_param_types = []
def add_generated_param_type(name):
    generated_param_types.append(name)

generated_return_types = []
def add_generated_return_type(name):
    generated_return_types.append(name)

missing_functions = {}
missing_functions_nr = 0
def add_missing_functions(name, module):
    global missing_functions_nr

    missing_functions_nr = missing_functions_nr + 1
    try:
        list = missing_functions[module]
        list.append(name)
    except:
        missing_functions[module] = [name]

#
# Provide the type generators and destructors for the parameters
#

def type_convert(str, name, info, module, function, pos):
#    res = str.replace("    ", " ")
#    res = str.replace("   ", " ")
#    res = str.replace("  ", " ")
    res = str.replace(" *", "_ptr")
#    res = str.replace("*", "_ptr")
    res = res.replace(" ", "_")
    if res == 'const_char_ptr':
        if name.find("file") != -1 or \
           name.find("uri") != -1 or \
           name.find("URI") != -1 or \
           info.find("filename") != -1 or \
           info.find("URI") != -1 or \
           info.find("URL") != -1:
            if function.find("Save") != -1 or \
               function.find("Create") != -1 or \
               function.find("Write") != -1 or \
               function.find("Fetch") != -1:
                return('fileoutput')
            return('filepath')
    if res == 'void_ptr':
        if module == 'nanoftp' and name == 'ctx':
            return('xmlNanoFTPCtxtPtr')
        if function == 'xmlNanoFTPNewCtxt' or \
           function == 'xmlNanoFTPConnectTo' or \
           function == 'xmlNanoFTPOpen':
            return('xmlNanoFTPCtxtPtr')
        if module == 'nanohttp' and name == 'ctx':
            return('xmlNanoHTTPCtxtPtr')
        if function == 'xmlNanoHTTPMethod' or \
           function == 'xmlNanoHTTPMethodRedir' or \
           function == 'xmlNanoHTTPOpen' or \
           function == 'xmlNanoHTTPOpenRedir':
            return('xmlNanoHTTPCtxtPtr');
        if function == 'xmlIOHTTPOpen':
            return('xmlNanoHTTPCtxtPtr')
        if name.find("data") != -1:
            return('userdata')
        if name.find("user") != -1:
            return('userdata')
    if res == 'xmlDoc_ptr':
        res = 'xmlDocPtr'
    if res == 'xmlNode_ptr':
        res = 'xmlNodePtr'
    if res == 'xmlDict_ptr':
        res = 'xmlDictPtr'
    if res == 'xmlNodePtr' and pos != 0:
        if (function == 'xmlAddChild' and pos == 2) or \
           (function == 'xmlAddChildList' and pos == 2) or \
           (function == 'xmlAddNextSibling' and pos == 2) or \
           (function == 'xmlAddSibling' and pos == 2) or \
           (function == 'xmlDocSetRootElement' and pos == 2) or \
           (function == 'xmlReplaceNode' and pos == 2) or \
           (function == 'xmlTextMerge') or \
           (function == 'xmlAddPrevSibling' and pos == 2):
            return('xmlNodePtr_in');
    if res == 'const xmlBufferPtr':
        res = 'xmlBufferPtr'
    if res == 'xmlChar_ptr' and name == 'name' and \
       function.find("EatName") != -1:
        return('eaten_name')
    if res == 'void_ptr*':
        res = 'void_ptr_ptr'
    if res == 'char_ptr*':
        res = 'char_ptr_ptr'
    if res == 'xmlChar_ptr*':
        res = 'xmlChar_ptr_ptr'
    if res == 'const_xmlChar_ptr*':
        res = 'const_xmlChar_ptr_ptr'
    if res == 'const_char_ptr*':
        res = 'const_char_ptr_ptr'
    if res == 'FILE_ptr' and module == 'debugXML':
        res = 'debug_FILE_ptr';
    if res == 'int' and name == 'options':
        if module == 'parser' or module == 'xmlreader':
            res = 'parseroptions'

    return res

known_param_types = []

def is_known_param_type(name):
    for type in known_param_types:
        if type == name:
            return 1
    return name[-3:] == 'Ptr' or name[-4:] == '_ptr'

def generate_param_type(name, rtype):
    global test
    for type in known_param_types:
        if type == name:
            return
    for type in generated_param_types:
        if type == name:
            return

    if name[-3:] == 'Ptr' or name[-4:] == '_ptr':
        define = 0
        if module in modules_defines:
            test.write("#ifdef %s\n" % (modules_defines[module]))
            define = 1
        test.write("""
#define gen_nb_%s 1
#define gen_%s(no, nr) NULL
#define des_%s(no, val, nr)
""" % (name, name, name))
        if define == 1:
            test.write("#endif\n\n")
        add_generated_param_type(name)

#
# Provide the type destructors for the return values
#

known_return_types = []

def is_known_return_type(name):
    for type in known_return_types:
        if type == name:
            return 1
    return 0

#
# Copy the beginning of the C test program result
#

try:
    input = open("testapi.c", "r")
except:
    input = open(srcPref + "testapi.c", "r")
test = open('testapi.c.new', 'w')

def compare_and_save():
    global test

    test.close()
    try:
        input = open("testapi.c", "r").read()
    except:
        input = ''
    test = open('testapi.c.new', "r").read()
    if input != test:
        try:
            os.system("rm testapi.c; mv testapi.c.new testapi.c")
        except:
            os.system("mv testapi.c.new testapi.c")
        print("Updated testapi.c")
    else:
        print("Generated testapi.c is identical")

line = input.readline()
while line != "":
    if line == "/* CUT HERE: everything below that line is generated */\n":
        break;
    if line[0:15] == "#define gen_nb_":
        type = line[15:].split()[0]
        known_param_types.append(type)
    if line[0:19] == "static void desret_":
        type = line[19:].split('(')[0]
        known_return_types.append(type)
    test.write(line)
    line = input.readline()
input.close()

if line == "":
    print("Could not find the CUT marker in testapi.c skipping generation")
    test.close()
    sys.exit(0)

print("Scanned testapi.c: found %d parameters types and %d return types\n" % (
      len(known_param_types), len(known_return_types)))
test.write("/* CUT HERE: everything below that line is generated */\n")


#
# Open the input API description
#
doc = libxml2.readFile(srcPref + 'doc/libxml2-api.xml', None, 0)
if doc == None:
    print("Failed to load doc/libxml2-api.xml")
    sys.exit(1)
ctxt = doc.xpathNewContext()

#
# Generate a list of all function parameters and select only
# those used in the api tests
#
argtypes = {}
args = ctxt.xpathEval("/api/symbols/function/arg")
for arg in args:
    mod = arg.xpathEval('string(../@file)')
    func = arg.xpathEval('string(../@name)')
    if (mod not in skipped_modules) and (func not in skipped_functions):
        type = arg.xpathEval('string(@type)')
        if type not in argtypes:
            argtypes[type] = func

# similarly for return types
rettypes = {}
rets = ctxt.xpathEval("/api/symbols/function/return")
for ret in rets:
    mod = ret.xpathEval('string(../@file)')
    func = ret.xpathEval('string(../@name)')
    if (mod not in skipped_modules) and (func not in skipped_functions):
        type = ret.xpathEval('string(@type)')
        if type not in rettypes:
            rettypes[type] = func

#
# Generate constructors and return type handling for all enums
# which are used as function parameters
#
enums = ctxt.xpathEval("/api/symbols/typedef[@type='enum']")
for enum in enums:
    module = enum.xpathEval('string(@file)')
    name = enum.xpathEval('string(@name)')
    #
    # Skip any enums which are not in our filtered lists
    #
    if (name == None) or ((name not in argtypes) and (name not in rettypes)):
        continue;
    define = 0

    if (name in argtypes) and is_known_param_type(name) == 0:
        values = ctxt.xpathEval("/api/symbols/enum[@type='%s']" % name)
        i = 0
        vals = []
        for value in values:
            vname = value.xpathEval('string(@name)')
            if vname == None:
                continue;
            i = i + 1
            if i >= 5:
                break;
            vals.append(vname)
        if vals == []:
            print("Didn't find any value for enum %s" % (name))
            continue
        if module in modules_defines:
            test.write("#ifdef %s\n" % (modules_defines[module]))
            define = 1
        test.write("#define gen_nb_%s %d\n" % (name, len(vals)))
        test.write("""static %s gen_%s(int no, int nr ATTRIBUTE_UNUSED) {\n""" %
                   (name, name))
        i = 1
        for value in vals:
            test.write("    if (no == %d) return(%s);\n" % (i, value))
            i = i + 1
        test.write("""    return(0);
}

static void des_%s(int no ATTRIBUTE_UNUSED, %s val ATTRIBUTE_UNUSED, int nr ATTRIBUTE_UNUSED) {
}

""" % (name, name));
        known_param_types.append(name)

    if (is_known_return_type(name) == 0) and (name in rettypes):
        if define == 0 and (module in modules_defines):
            test.write("#ifdef %s\n" % (modules_defines[module]))
            define = 1
        test.write("""static void desret_%s(%s val ATTRIBUTE_UNUSED) {
}

""" % (name, name))
        known_return_types.append(name)
    if define == 1:
        test.write("#endif\n\n")

#
# Load the interfaces
# 
headers = ctxt.xpathEval("/api/files/file")
for file in headers:
    name = file.xpathEval('string(@name)')
    if (name == None) or (name == ''):
        continue

    #
    # Some module may be skipped because they don't really consists
    # of user callable APIs
    #
    if is_skipped_module(name):
        continue

    #
    # do not test deprecated APIs
    #
    desc = file.xpathEval('string(description)')
    if desc.find('DEPRECATED') != -1:
        print("Skipping deprecated interface %s" % name)
        continue;

    test.write("#include <libxml/%s.h>\n" % name)
    modules.append(name)
        
#
# Generate the callers signatures
# 
for module in modules:
    test.write("static int test_%s(void);\n" % module);

#
# Generate the top caller
# 

test.write("""
/**
 * testlibxml2:
 *
 * Main entry point of the tester for the full libxml2 module,
 * it calls all the tester entry point for each module.
 *
 * Returns the number of error found
 */
static int
testlibxml2(void)
{
    int test_ret = 0;

""")

for module in modules:
    test.write("    test_ret += test_%s();\n" % module)

test.write("""
    printf("Total: %d functions, %d tests, %d errors\\n",
           function_tests, call_tests, test_ret);
    return(test_ret);
}

""")

#
# How to handle a function
# 
nb_tests = 0

def generate_test(module, node):
    global test
    global nb_tests
    nb_cond = 0
    no_gen = 0

    name = node.xpathEval('string(@name)')
    if is_skipped_function(name):
        return

    #
    # check we know how to handle the args and return values
    # and store the information for the generation
    #
    try:
        args = node.xpathEval("arg")
    except:
        args = []
    t_args = []
    n = 0
    for arg in args:
        n = n + 1
        rtype = arg.xpathEval("string(@type)")
        if rtype == 'void':
            break;
        info = arg.xpathEval("string(@info)")
        nam = arg.xpathEval("string(@name)")
        type = type_convert(rtype, nam, info, module, name, n)
        if is_known_param_type(type) == 0:
            add_missing_type(type, name);
            no_gen = 1
        t_args.append((nam, type, rtype, info))
    
    try:
        rets = node.xpathEval("return")
    except:
        rets = []
    t_ret = None
    for ret in rets:
        rtype = ret.xpathEval("string(@type)")
        info = ret.xpathEval("string(@info)")
        type = type_convert(rtype, 'return', info, module, name, 0)
        if rtype == 'void':
            break
        if is_known_return_type(type) == 0:
            add_missing_type(type, name);
            no_gen = 1
        t_ret = (type, rtype, info)
        break

    if no_gen == 0:
        for t_arg in t_args:
            (nam, type, rtype, info) = t_arg
            generate_param_type(type, rtype)

    test.write("""
static int
test_%s(void) {
    int test_ret = 0;

""" % (name))

    if no_gen == 1:
        add_missing_functions(name, module)
        test.write("""
    /* missing type support */
    return(test_ret);
}

""")
        return

    try:
        conds = node.xpathEval("cond")
        for cond in conds:
            test.write("#if %s\n" % (cond.get_content()))
            nb_cond = nb_cond + 1
    except:
        pass

    define = 0
    if name in function_defines:
        test.write("#ifdef %s\n" % (function_defines[name]))
        define = 1
    
    # Declare the memory usage counter
    no_mem = is_skipped_memcheck(name)
    if no_mem == 0:
        test.write("    int mem_base;\n");

    # Declare the return value
    if t_ret != None:
        test.write("    %s ret_val;\n" % (t_ret[1]))

    # Declare the arguments
    for arg in t_args:
        (nam, type, rtype, info) = arg;
        # add declaration
        test.write("    %s %s; /* %s */\n" % (rtype, nam, info))
        test.write("    int n_%s;\n" % (nam))
    test.write("\n")

    # Cascade loop on of each argument list of values
    for arg in t_args:
        (nam, type, rtype, info) = arg;
        #
        test.write("    for (n_%s = 0;n_%s < gen_nb_%s;n_%s++) {\n" % (
                   nam, nam, type, nam))
    
    # log the memory usage
    if no_mem == 0:
        test.write("        mem_base = xmlMemBlocks();\n");

    # prepare the call
    i = 0;
    for arg in t_args:
        (nam, type, rtype, info) = arg;
        #
        test.write("        %s = gen_%s(n_%s, %d);\n" % (nam, type, nam, i))
        i = i + 1;

    # add checks to avoid out-of-bounds array access
    i = 0;
    for arg in t_args:
        (nam, type, rtype, info) = arg;
        # assume that "size", "len", and "start" parameters apply to either
        # the nearest preceding or following char pointer
        if type == "int" and (nam == "size" or nam == "len" or nam == "start"):
            for j in (list(range(i - 1, -1, -1)) + list(range(i + 1, len(t_args)))):
                (bnam, btype) = t_args[j][:2]
                if btype == "const_char_ptr" or btype == "const_xmlChar_ptr":
                    test.write(
                        "        if ((%s != NULL) &&\n"
                        "            (%s > xmlStrlen(BAD_CAST %s)))\n"
                        "            %s = 0;\n"
                        % (bnam, nam, bnam, nam))
                    break
        i = i + 1;

    # do the call, and clanup the result
    if name in extra_pre_call:
        test.write("        %s\n"% (extra_pre_call[name]))
    if t_ret != None:
        test.write("\n        ret_val = %s(" % (name))
        need = 0
        for arg in t_args:
            (nam, type, rtype, info) = arg
            if need:
                test.write(", ")
            else:
                need = 1
            test.write("%s" % nam);
        test.write(");\n")
        if name in extra_post_call:
            test.write("        %s\n"% (extra_post_call[name]))
        test.write("        desret_%s(ret_val);\n" % t_ret[0])
    else:
        test.write("\n        %s(" % (name));
        need = 0;
        for arg in t_args:
            (nam, type, rtype, info) = arg;
            if need:
                test.write(", ")
            else:
                need = 1
            test.write("%s" % nam)
        test.write(");\n")
        if name in extra_post_call:
            test.write("        %s\n"% (extra_post_call[name]))

    test.write("        call_tests++;\n");

    # Free the arguments
    i = 0;
    for arg in t_args:
        (nam, type, rtype, info) = arg;
        # This is a hack to prevent generating a destructor for the
        # 'input' argument in xmlTextReaderSetup.  There should be
        # a better, more generic way to do this!
        if info.find('destroy') == -1:
            test.write("        des_%s(n_%s, " % (type, nam))
            test.write("%s, %d);\n" % (nam, i))
        i = i + 1;

    test.write("        xmlResetLastError();\n");
    # Check the memory usage
    if no_mem == 0:
        test.write("""        if (mem_base != xmlMemBlocks()) {
            printf("Leak of %%d blocks found in %s",
\t           xmlMemBlocks() - mem_base);
\t    test_ret++;
""" % (name));
        for arg in t_args:
            (nam, type, rtype, info) = arg;
            test.write("""            printf(" %%d", n_%s);\n""" % (nam))
        test.write("""            printf("\\n");\n""")
        test.write("        }\n")

    for arg in t_args:
        test.write("    }\n")

    test.write("    function_tests++;\n")
    #
    # end of conditional
    #
    while nb_cond > 0:
        test.write("#endif\n")
        nb_cond = nb_cond -1
    if define == 1:
        test.write("#endif\n")

    nb_tests = nb_tests + 1;

    test.write("""
    return(test_ret);
}

""")
    
#
# Generate all module callers
#
for module in modules:
    # gather all the functions exported by that module
    try:
        functions = ctxt.xpathEval("/api/symbols/function[@file='%s']" % (module))
    except:
        print("Failed to gather functions from module %s" % (module))
        continue;

    # iterate over all functions in the module generating the test
    i = 0
    nb_tests_old = nb_tests
    for function in functions:
        i = i + 1
        generate_test(module, function);

    # header
    test.write("""static int
test_%s(void) {
    int test_ret = 0;

    if (quiet == 0) printf("Testing %s : %d of %d functions ...\\n");
""" % (module, module, nb_tests - nb_tests_old, i))

    # iterate over all functions in the module generating the call
    for function in functions:
        name = function.xpathEval('string(@name)')
        if is_skipped_function(name):
            continue
        test.write("    test_ret += test_%s();\n" % (name))

    # footer
    test.write("""
    if (test_ret != 0)
\tprintf("Module %s: %%d errors\\n", test_ret);
    return(test_ret);
}
""" % (module))

#
# Generate direct module caller
#
test.write("""static int
test_module(const char *module) {
""");
for module in modules:
    test.write("""    if (!strcmp(module, "%s")) return(test_%s());\n""" % (
        module, module))
test.write("""    return(0);
}
""");

print("Generated test for %d modules and %d functions" %(len(modules), nb_tests))

compare_and_save()

missing_list = []
for missing in missing_types.keys():
    if missing == 'va_list' or missing == '...':
        continue;

    n = len(missing_types[missing])
    missing_list.append((n, missing))

missing_list.sort(key=lambda a: a[0])
print("Missing support for %d functions and %d types see missing.lst" % (missing_functions_nr, len(missing_list)))
lst = open("missing.lst", "w")
lst.write("Missing support for %d types" % (len(missing_list)))
lst.write("\n")
for miss in missing_list:
    lst.write("%s: %d :" % (miss[1], miss[0]))
    i = 0
    for n in missing_types[miss[1]]:
        i = i + 1
        if i > 5:
            lst.write(" ...")
            break
        lst.write(" %s" % (n))
    lst.write("\n")
lst.write("\n")
lst.write("\n")
lst.write("Missing support per module");
for module in missing_functions.keys():
    lst.write("module %s:\n   %s\n" % (module, missing_functions[module]))

lst.close()