generator.py

libxml,在UNIX/LINUX下非常重要的一个库,为XML相关应用提供方便.目前上载的是最新版本,若要取得最新版本,请参考里面的readme.

#!/usr/bin/python -u
#
# generate python wrappers from the XML API description
#

functions = {}
enums = {} # { enumType: { enumConstant: enumValue } }

import os
import sys
import string

if __name__ == "__main__":
    # launched as a script
    srcPref = os.path.dirname(sys.argv[0])
else:
    # imported
    srcPref = os.path.dirname(__file__)

#######################################################################
#
#  That part if purely the API acquisition phase from the
#  XML API description
#
#######################################################################
import os
import xmllib
try:
    import sgmlop
except ImportError:
    sgmlop = None # accelerator not available

debug = 0

if sgmlop:
    class FastParser:
        """sgmlop based XML parser.  this is typically 15x faster
           than SlowParser..."""

        def __init__(self, target):

            # setup callbacks
            self.finish_starttag = target.start
            self.finish_endtag = target.end
            self.handle_data = target.data

            # activate parser
            self.parser = sgmlop.XMLParser()
            self.parser.register(self)
            self.feed = self.parser.feed
            self.entity = {
                "amp": "&", "gt": ">", "lt": "<",
                "apos": "'", "quot": '"'
                }

        def close(self):
            try:
                self.parser.close()
            finally:
                self.parser = self.feed = None # nuke circular reference

        def handle_entityref(self, entity):
            # <string> entity
            try:
                self.handle_data(self.entity[entity])
            except KeyError:
                self.handle_data("&%s;" % entity)

else:
    FastParser = None


class SlowParser(xmllib.XMLParser):
    """slow but safe standard parser, based on the XML parser in
       Python's standard library."""

    def __init__(self, target):
        self.unknown_starttag = target.start
        self.handle_data = target.data
        self.unknown_endtag = target.end
        xmllib.XMLParser.__init__(self)

def getparser(target = None):
    # get the fastest available parser, and attach it to an
    # unmarshalling object.  return both objects.
    if target is None:
        target = docParser()
    if FastParser:
        return FastParser(target), target
    return SlowParser(target), target

class docParser:
    def __init__(self):
        self._methodname = None
        self._data = []
        self.in_function = 0

    def close(self):
        if debug:
            print "close"

    def getmethodname(self):
        return self._methodname

    def data(self, text):
        if debug:
            print "data %s" % text
        self._data.append(text)

    def start(self, tag, attrs):
        if debug:
            print "start %s, %s" % (tag, attrs)
        if tag == 'function':
            self._data = []
            self.in_function = 1
            self.function = None
            self.function_args = []
            self.function_descr = None
            self.function_return = None
            self.function_file = None
            if attrs.has_key('name'):
                self.function = attrs['name']
            if attrs.has_key('file'):
                self.function_file = attrs['file']
        elif tag == 'info':
            self._data = []
        elif tag == 'arg':
            if self.in_function == 1:
                self.function_arg_name = None
                self.function_arg_type = None
                self.function_arg_info = None
                if attrs.has_key('name'):
                    self.function_arg_name = attrs['name']
                if attrs.has_key('type'):
                    self.function_arg_type = attrs['type']
                if attrs.has_key('info'):
                    self.function_arg_info = attrs['info']
        elif tag == 'return':
            if self.in_function == 1:
                self.function_return_type = None
                self.function_return_info = None
                self.function_return_field = None
                if attrs.has_key('type'):
                    self.function_return_type = attrs['type']
                if attrs.has_key('info'):
                    self.function_return_info = attrs['info']
                if attrs.has_key('field'):
                    self.function_return_field = attrs['field']
        elif tag == 'enum':
            enum(attrs['type'],attrs['name'],attrs['value'])

    def end(self, tag):
        if debug:
            print "end %s" % tag
        if tag == 'function':
            if self.function != None:
                function(self.function, self.function_descr,
                         self.function_return, self.function_args,
                         self.function_file)
                self.in_function = 0
        elif tag == 'arg':
            if self.in_function == 1:
                self.function_args.append([self.function_arg_name,
                                           self.function_arg_type,
                                           self.function_arg_info])
        elif tag == 'return':
            if self.in_function == 1:
                self.function_return = [self.function_return_type,
                                        self.function_return_info,
                                        self.function_return_field]
        elif tag == 'info':
            str = ''
            for c in self._data:
                str = str + c
            if self.in_function == 1:
                self.function_descr = str
                
                
def function(name, desc, ret, args, file):
    functions[name] = (desc, ret, args, file)

def enum(type, name, value):
    if not enums.has_key(type):
        enums[type] = {}
    enums[type][name] = value

#######################################################################
#
#  Some filtering rukes to drop functions/types which should not
#  be exposed as-is on the Python interface
#
#######################################################################

skipped_modules = {
    'xmlmemory': None,
    'DOCBparser': None,
    'SAX': None,
    'hash': None,
    'list': None,
    'threads': None,
#    'xpointer': None,
}
skipped_types = {
    'int *': "usually a return type",
    'xmlSAXHandlerPtr': "not the proper interface for SAX",
    'htmlSAXHandlerPtr': "not the proper interface for SAX",
    'xmlRMutexPtr': "thread specific, skipped",
    'xmlMutexPtr': "thread specific, skipped",
    'xmlGlobalStatePtr': "thread specific, skipped",
    'xmlListPtr': "internal representation not suitable for python",
    'xmlBufferPtr': "internal representation not suitable for python",
    'FILE *': None,
}

#######################################################################
#
#  Table of remapping to/from the python type or class to the C
#  counterpart.
#
#######################################################################

py_types = {
    'void': (None, None, None, None),
    'int':  ('i', None, "int", "int"),
    'long':  ('i', None, "int", "int"),
    'double':  ('d', None, "double", "double"),
    'unsigned int':  ('i', None, "int", "int"),
    'xmlChar':  ('c', None, "int", "int"),
    'unsigned char *':  ('z', None, "charPtr", "char *"),
    'char *':  ('z', None, "charPtr", "char *"),
    'const char *':  ('z', None, "charPtrConst", "const char *"),
    'xmlChar *':  ('z', None, "xmlCharPtr", "xmlChar *"),
    'const xmlChar *':  ('z', None, "xmlCharPtrConst", "const xmlChar *"),
    'xmlNodePtr':  ('O', "xmlNode", "xmlNodePtr", "xmlNodePtr"),
    'const xmlNodePtr':  ('O', "xmlNode", "xmlNodePtr", "xmlNodePtr"),
    'xmlNode *':  ('O', "xmlNode", "xmlNodePtr", "xmlNodePtr"),
    'const xmlNode *':  ('O', "xmlNode", "xmlNodePtr", "xmlNodePtr"),
    'xmlDtdPtr':  ('O', "xmlNode", "xmlNodePtr", "xmlNodePtr"),
    'const xmlDtdPtr':  ('O', "xmlNode", "xmlNodePtr", "xmlNodePtr"),
    'xmlDtd *':  ('O', "xmlNode", "xmlNodePtr", "xmlNodePtr"),
    'const xmlDtd *':  ('O', "xmlNode", "xmlNodePtr", "xmlNodePtr"),
    'xmlAttrPtr':  ('O', "xmlNode", "xmlNodePtr", "xmlNodePtr"),
    'const xmlAttrPtr':  ('O', "xmlNode", "xmlNodePtr", "xmlNodePtr"),
    'xmlAttr *':  ('O', "xmlNode", "xmlNodePtr", "xmlNodePtr"),
    'const xmlAttr *':  ('O', "xmlNode", "xmlNodePtr", "xmlNodePtr"),
    'xmlEntityPtr':  ('O', "xmlNode", "xmlNodePtr", "xmlNodePtr"),
    'const xmlEntityPtr':  ('O', "xmlNode", "xmlNodePtr", "xmlNodePtr"),
    'xmlEntity *':  ('O', "xmlNode", "xmlNodePtr", "xmlNodePtr"),
    'const xmlEntity *':  ('O', "xmlNode", "xmlNodePtr", "xmlNodePtr"),
    'xmlElementPtr':  ('O', "xmlElement", "xmlElementPtr", "xmlElementPtr"),
    'const xmlElementPtr':  ('O', "xmlElement", "xmlElementPtr", "xmlElementPtr"),
    'xmlElement *':  ('O', "xmlElement", "xmlElementPtr", "xmlElementPtr"),
    'const xmlElement *':  ('O', "xmlElement", "xmlElementPtr", "xmlElementPtr"),
    'xmlAttributePtr':  ('O', "xmlAttribute", "xmlAttributePtr", "xmlAttributePtr"),
    'const xmlAttributePtr':  ('O', "xmlAttribute", "xmlAttributePtr", "xmlAttributePtr"),
    'xmlAttribute *':  ('O', "xmlAttribute", "xmlAttributePtr", "xmlAttributePtr"),
    'const xmlAttribute *':  ('O', "xmlAttribute", "xmlAttributePtr", "xmlAttributePtr"),
    'xmlNsPtr':  ('O', "xmlNode", "xmlNsPtr", "xmlNsPtr"),
    'const xmlNsPtr':  ('O', "xmlNode", "xmlNsPtr", "xmlNsPtr"),
    'xmlNs *':  ('O', "xmlNode", "xmlNsPtr", "xmlNsPtr"),
    'const xmlNs *':  ('O', "xmlNode", "xmlNsPtr", "xmlNsPtr"),
    'xmlDocPtr':  ('O', "xmlNode", "xmlDocPtr", "xmlDocPtr"),
    'const xmlDocPtr':  ('O', "xmlNode", "xmlDocPtr", "xmlDocPtr"),
    'xmlDoc *':  ('O', "xmlNode", "xmlDocPtr", "xmlDocPtr"),
    'const xmlDoc *':  ('O', "xmlNode", "xmlDocPtr", "xmlDocPtr"),
    'htmlDocPtr':  ('O', "xmlNode", "xmlDocPtr", "xmlDocPtr"),
    'const htmlDocPtr':  ('O', "xmlNode", "xmlDocPtr", "xmlDocPtr"),
    'htmlDoc *':  ('O', "xmlNode", "xmlDocPtr", "xmlDocPtr"),
    'const htmlDoc *':  ('O', "xmlNode", "xmlDocPtr", "xmlDocPtr"),
    'htmlNodePtr':  ('O', "xmlNode", "xmlNodePtr", "xmlNodePtr"),
    'const htmlNodePtr':  ('O', "xmlNode", "xmlNodePtr", "xmlNodePtr"),
    'htmlNode *':  ('O', "xmlNode", "xmlNodePtr", "xmlNodePtr"),
    'const htmlNode *':  ('O', "xmlNode", "xmlNodePtr", "xmlNodePtr"),
    'xmlXPathContextPtr':  ('O', "xmlXPathContext", "xmlXPathContextPtr", "xmlXPathContextPtr"),
    'xmlXPathContext *':  ('O', "xpathContext", "xmlXPathContextPtr", "xmlXPathContextPtr"),
    'xmlXPathParserContextPtr':  ('O', "xmlXPathParserContext", "xmlXPathParserContextPtr", "xmlXPathParserContextPtr"),
    'xmlParserCtxtPtr': ('O', "parserCtxt", "xmlParserCtxtPtr", "xmlParserCtxtPtr"),
    'xmlParserCtxt *': ('O', "parserCtxt", "xmlParserCtxtPtr", "xmlParserCtxtPtr"),
    'htmlParserCtxtPtr': ('O', "parserCtxt", "xmlParserCtxtPtr", "xmlParserCtxtPtr"),
    'htmlParserCtxt *': ('O', "parserCtxt", "xmlParserCtxtPtr", "xmlParserCtxtPtr"),
    'xmlValidCtxtPtr': ('O', "ValidCtxt", "xmlValidCtxtPtr", "xmlValidCtxtPtr"),
    'xmlCatalogPtr': ('O', "catalog", "xmlCatalogPtr", "xmlCatalogPtr"),
    'FILE *': ('O', "File", "FILEPtr", "FILE *"),
    'xmlURIPtr': ('O', "URI", "xmlURIPtr", "xmlURIPtr"),
    'xmlErrorPtr': ('O', "Error", "xmlErrorPtr", "xmlErrorPtr"),
    'xmlOutputBufferPtr': ('O', "outputBuffer", "xmlOutputBufferPtr", "xmlOutputBufferPtr"),
    'xmlParserInputBufferPtr': ('O', "inputBuffer", "xmlParserInputBufferPtr", "xmlParserInputBufferPtr"),
    'xmlRegexpPtr': ('O', "xmlReg", "xmlRegexpPtr", "xmlRegexpPtr"),
    'xmlTextReaderLocatorPtr': ('O', "xmlTextReaderLocator", "xmlTextReaderLocatorPtr", "xmlTextReaderLocatorPtr"),
    'xmlTextReaderPtr': ('O', "xmlTextReader", "xmlTextReaderPtr", "xmlTextReaderPtr"),
    'xmlRelaxNGPtr': ('O', "relaxNgSchema", "xmlRelaxNGPtr", "xmlRelaxNGPtr"),
    'xmlRelaxNGParserCtxtPtr': ('O', "relaxNgParserCtxt", "xmlRelaxNGParserCtxtPtr", "xmlRelaxNGParserCtxtPtr"),
    'xmlRelaxNGValidCtxtPtr': ('O', "relaxNgValidCtxt", "xmlRelaxNGValidCtxtPtr", "xmlRelaxNGValidCtxtPtr"),
    'xmlSchemaPtr': ('O', "Schema", "xmlSchemaPtr", "xmlSchemaPtr"),
    'xmlSchemaParserCtxtPtr': ('O', "SchemaParserCtxt", "xmlSchemaParserCtxtPtr", "xmlSchemaParserCtxtPtr"),
    'xmlSchemaValidCtxtPtr': ('O', "SchemaValidCtxt", "xmlSchemaValidCtxtPtr", "xmlSchemaValidCtxtPtr"),
}

py_return_types = {
    'xmlXPathObjectPtr':  ('O', "foo", "xmlXPathObjectPtr", "xmlXPathObjectPtr"),
}

unknown_types = {}

foreign_encoding_args = (
    'htmlCreateMemoryParserCtxt',
    'htmlCtxtReadMemory',
    'htmlParseChunk',
    'htmlReadMemory',
    'xmlCreateMemoryParserCtxt',
    'xmlCtxtReadMemory',
    'xmlCtxtResetPush',
    'xmlParseChunk',
    'xmlParseMemory',
    'xmlReadMemory',
    'xmlRecoverMemory',
)

#######################################################################
#