[mesa.git] / src / mapi / glapi / gen / gl_XML.py

#!/usr/bin/env python

# (C) Copyright IBM Corporation 2004, 2005
# All Rights Reserved.
#
# Permission is hereby granted, free of charge, to any person obtaining a
# copy of this software and associated documentation files (the "Software"),
# to deal in the Software without restriction, including without limitation
# on the rights to use, copy, modify, merge, publish, distribute, sub
# license, and/or sell copies of the Software, and to permit persons to whom
# the Software is furnished to do so, subject to the following conditions:
#
# The above copyright notice and this permission notice (including the next
# paragraph) shall be included in all copies or substantial portions of the
# Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.  IN NO EVENT SHALL
# IBM AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
# FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
# IN THE SOFTWARE.
#
# Authors:
#    Ian Romanick <idr@us.ibm.com>

from decimal import Decimal
import libxml2
import re, sys, string
import typeexpr


def parse_GL_API( file_name, factory = None ):
    doc = libxml2.readFile( file_name, None, libxml2.XML_PARSE_XINCLUDE + libxml2.XML_PARSE_NOBLANKS + libxml2.XML_PARSE_DTDVALID + libxml2.XML_PARSE_DTDATTR + libxml2.XML_PARSE_DTDLOAD + libxml2.XML_PARSE_NOENT )
    ret = doc.xincludeProcess()

    if not factory:
        factory = gl_item_factory()

    api = factory.create_item( "api", None, None )
    api.process_element( doc )

    # After the XML has been processed, we need to go back and assign
    # dispatch offsets to the functions that request that their offsets
    # be assigned by the scripts.  Typically this means all functions
    # that are not part of the ABI.

    for func in api.functionIterateByCategory():
        if func.assign_offset:
            func.offset = api.next_offset;
            api.next_offset += 1

    doc.freeDoc()

    return api


def is_attr_true( element, name ):
    """Read a name value from an element's attributes.

    The value read from the attribute list must be either 'true' or
    'false'.  If the value is 'false', zero will be returned.  If the
    value is 'true', non-zero will be returned.  An exception will be
    raised for any other value."""

    value = element.nsProp( name, None )
    if value == "true":
        return 1
    elif value == "false":
        return 0
    else:
        raise RuntimeError('Invalid value "%s" for boolean "%s".' % (value, name))


class gl_print_base(object):
    """Base class of all API pretty-printers.

    In the model-view-controller pattern, this is the view.  Any derived
    class will want to over-ride the printBody, printRealHader, and
    printRealFooter methods.  Some derived classes may want to over-ride
    printHeader and printFooter, or even Print (though this is unlikely).
    """

    def __init__(self):
        # Name of the script that is generating the output file.
        # Every derived class should set this to the name of its
        # source file.

        self.name = "a"


        # License on the *generated* source file.  This may differ
        # from the license on the script that is generating the file.
        # Every derived class should set this to some reasonable
        # value.
        #
        # See license.py for an example of a reasonable value.

        self.license = "The license for this file is unspecified."


        # The header_tag is the name of the C preprocessor define
        # used to prevent multiple inclusion.  Typically only
        # generated C header files need this to be set.  Setting it
        # causes code to be generated automatically in printHeader
        # and printFooter.

        self.header_tag = None


        # List of file-private defines that must be undefined at the
        # end of the file.  This can be used in header files to define
        # names for use in the file, then undefine them at the end of
        # the header file.

        self.undef_list = []
        return


    def Print(self, api):
        self.printHeader()
        self.printBody(api)
        self.printFooter()
        return


    def printHeader(self):
        """Print the header associated with all files and call the printRealHeader method."""

        print '/* DO NOT EDIT - This file generated automatically by %s script */' \
                % (self.name)
        print ''
        print '/*'
        print ' * ' + self.license.replace('\n', '\n * ')
        print ' */'
        print ''
        if self.header_tag:
            print '#if !defined( %s )' % (self.header_tag)
            print '#  define %s' % (self.header_tag)
            print ''
        self.printRealHeader();
        return


    def printFooter(self):
        """Print the header associated with all files and call the printRealFooter method."""

        self.printRealFooter()

        if self.undef_list:
            print ''
            for u in self.undef_list:
                print "#  undef %s" % (u)

        if self.header_tag:
            print ''
            print '#endif /* !defined( %s ) */' % (self.header_tag)


    def printRealHeader(self):
        """Print the "real" header for the created file.

        In the base class, this function is empty.  All derived
        classes should over-ride this function."""
        return


    def printRealFooter(self):
        """Print the "real" footer for the created file.

        In the base class, this function is empty.  All derived
        classes should over-ride this function."""
        return


    def printPure(self):
        """Conditionally define `PURE' function attribute.

        Conditionally defines a preprocessor macro `PURE' that wraps
        GCC's `pure' function attribute.  The conditional code can be
        easilly adapted to other compilers that support a similar
        feature.

        The name is also added to the file's undef_list.
        """
        self.undef_list.append("PURE")
        print """#  if defined(__GNUC__) || (defined(__SUNPRO_C) && (__SUNPRO_C >= 0x590))
#    define PURE __attribute__((pure))
#  else
#    define PURE
#  endif"""
        return


    def printFastcall(self):
        """Conditionally define `FASTCALL' function attribute.

        Conditionally defines a preprocessor macro `FASTCALL' that
        wraps GCC's `fastcall' function attribute.  The conditional
        code can be easilly adapted to other compilers that support a
        similar feature.

        The name is also added to the file's undef_list.
        """

        self.undef_list.append("FASTCALL")
        print """#  if defined(__i386__) && defined(__GNUC__) && !defined(__CYGWIN__) && !defined(__MINGW32__)
#    define FASTCALL __attribute__((fastcall))
#  else
#    define FASTCALL
#  endif"""
        return


    def printVisibility(self, S, s):
        """Conditionally define visibility function attribute.

        Conditionally defines a preprocessor macro name S that wraps
        GCC's visibility function attribute.  The visibility used is
        the parameter s.  The conditional code can be easilly adapted
        to other compilers that support a similar feature.

        The name is also added to the file's undef_list.
        """

        self.undef_list.append(S)
        print """#  if (defined(__GNUC__) && !defined(__CYGWIN__) && !defined(__MINGW32__)) || (defined(__SUNPRO_C) && (__SUNPRO_C >= 0x590) && defined(__ELF__))
#    define %s  __attribute__((visibility("%s")))
#  else
#    define %s
#  endif""" % (S, s, S)
        return


    def printNoinline(self):
        """Conditionally define `NOINLINE' function attribute.

        Conditionally defines a preprocessor macro `NOINLINE' that
        wraps GCC's `noinline' function attribute.  The conditional
        code can be easilly adapted to other compilers that support a
        similar feature.

        The name is also added to the file's undef_list.
        """

        self.undef_list.append("NOINLINE")
        print """#  if defined(__GNUC__) || (defined(__SUNPRO_C) && (__SUNPRO_C >= 0x590))
#    define NOINLINE __attribute__((noinline))
#  else
#    define NOINLINE
#  endif"""
        return


def real_function_name(element):
    name = element.nsProp( "name", None )
    alias = element.nsProp( "alias", None )

    if alias:
        return alias
    else:
        return name


def real_category_name(c):
    if re.compile("[1-9][0-9]*[.][0-9]+").match(c):
        return "GL_VERSION_" + c.replace(".", "_")
    else:
        return c


def classify_category(name, number):
    """Based on the category name and number, select a numerical class for it.

    Categories are divided into four classes numbered 0 through 3.  The
    classes are:

            0. Core GL versions, sorted by version number.
            1. ARB extensions, sorted by extension number.
            2. Non-ARB extensions, sorted by extension number.
            3. Un-numbered extensions, sorted by extension name.
    """

    try:
        core_version = float(name)
    except Exception,e:
        core_version = 0.0

    if core_version > 0.0:
        cat_type = 0
        key = name
    elif name.startswith("GL_ARB_") or name.startswith("GLX_ARB_") or name.startswith("WGL_ARB_"):
        cat_type = 1
        key = int(number)
    else:
        if number != None:
            cat_type = 2
            key = int(number)
        else:
            cat_type = 3
            key = name


    return [cat_type, key]


def create_parameter_string(parameters, include_names):
    """Create a parameter string from a list of gl_parameters."""

    list = []
    for p in parameters:
        if p.is_padding:
            continue

        if include_names:
            list.append( p.string() )
        else:
            list.append( p.type_string() )

    if len(list) == 0: list = ["void"]

    return string.join(list, ", ")


class gl_item(object):
    def __init__(self, element, context):
        self.context = context
        self.name = element.nsProp( "name", None )
        self.category = real_category_name( element.parent.nsProp( "name", None ) )
        return


class gl_type( gl_item ):
    def __init__(self, element, context):
        gl_item.__init__(self, element, context)
        self.size = int( element.nsProp( "size", None ), 0 )

        te = typeexpr.type_expression( None )
        tn = typeexpr.type_node()
        tn.size = int( element.nsProp( "size", None ), 0 )
        tn.integer = not is_attr_true( element, "float" )
        tn.unsigned = is_attr_true( element, "unsigned" )
        tn.pointer = is_attr_true( element, "pointer" )
        tn.name = "GL" + self.name
        te.set_base_type_node( tn )

        self.type_expr = te
        return


    def get_type_expression(self):
        return self.type_expr


class gl_enum( gl_item ):
    def __init__(self, element, context):
        gl_item.__init__(self, element, context)
        self.value = int( element.nsProp( "value", None ), 0 )

        temp = element.nsProp( "count", None )
        if not temp or temp == "?":
            self.default_count = -1
        else:
            try:
                c = int(temp)
            except Exception,e:
                raise RuntimeError('Invalid count value "%s" for enum "%s" in function "%s" when an integer was expected.' % (temp, self.name, n))

            self.default_count = c

        return


    def priority(self):
        """Calculate a 'priority' for this enum name.

        When an enum is looked up by number, there may be many
        possible names, but only one is the 'prefered' name.  The
        priority is used to select which name is the 'best'.

        Highest precedence is given to core GL name.  ARB extension
        names have the next highest, followed by EXT extension names.
        Vendor extension names are the lowest.
        """

        if self.name.endswith( "_BIT" ):
            bias = 1
        else:
            bias = 0

        if self.category.startswith( "GL_VERSION_" ):
            priority = 0
        elif self.category.startswith( "GL_ARB_" ):
            priority = 2
        elif self.category.startswith( "GL_EXT_" ):
            priority = 4
        else:
            priority = 6

        return priority + bias



class gl_parameter(object):
    def __init__(self, element, context):
        self.name = element.nsProp( "name", None )

        ts = element.nsProp( "type", None )
        self.type_expr = typeexpr.type_expression( ts, context )

        temp = element.nsProp( "variable_param", None )
        if temp:
            self.count_parameter_list = temp.split( ' ' )
        else:
            self.count_parameter_list = []

        # The count tag can be either a numeric string or the name of
        # a variable.  If it is the name of a variable, the int(c)
        # statement will throw an exception, and the except block will
        # take over.

        c = element.nsProp( "count", None )
        try: 
            count = int(c)
            self.count = count
            self.counter = None
        except Exception,e:
            count = 1
            self.count = 0
            self.counter = c

        self.count_scale = int(element.nsProp( "count_scale", None ))

        elements = (count * self.count_scale)
        if elements == 1:
            elements = 0

        #if ts == "GLdouble":
        #       print '/* stack size -> %s = %u (before)*/' % (self.name, self.type_expr.get_stack_size())
        #       print '/* # elements = %u */' % (elements)
        self.type_expr.set_elements( elements )
        #if ts == "GLdouble":
        #       print '/* stack size -> %s = %u (after) */' % (self.name, self.type_expr.get_stack_size())

        self.is_client_only = is_attr_true( element, 'client_only' )
        self.is_counter     = is_attr_true( element, 'counter' )
        self.is_output      = is_attr_true( element, 'output' )


        # Pixel data has special parameters.

        self.width      = element.nsProp('img_width',  None)
        self.height     = element.nsProp('img_height', None)
        self.depth      = element.nsProp('img_depth',  None)
        self.extent     = element.nsProp('img_extent', None)

        self.img_xoff   = element.nsProp('img_xoff',   None)
        self.img_yoff   = element.nsProp('img_yoff',   None)
        self.img_zoff   = element.nsProp('img_zoff',   None)
        self.img_woff   = element.nsProp('img_woff',   None)

        self.img_format = element.nsProp('img_format', None)
        self.img_type   = element.nsProp('img_type',   None)
        self.img_target = element.nsProp('img_target', None)

        self.img_pad_dimensions = is_attr_true( element, 'img_pad_dimensions' )
        self.img_null_flag      = is_attr_true( element, 'img_null_flag' )
        self.img_send_null      = is_attr_true( element, 'img_send_null' )

        self.is_padding = is_attr_true( element, 'padding' )
        return


    def compatible(self, other):
        return 1


    def is_array(self):
        return self.is_pointer()


    def is_pointer(self):
        return self.type_expr.is_pointer()


    def is_image(self):
        if self.width:
            return 1
        else:
            return 0


    def is_variable_length(self):
        return len(self.count_parameter_list) or self.counter


    def is_64_bit(self):
        count = self.type_expr.get_element_count()
        if count:
            if (self.size() / count) == 8:
                return 1
        else:
            if self.size() == 8:
                return 1

        return 0


    def string(self):
        return self.type_expr.original_string + " " + self.name


    def type_string(self):
        return self.type_expr.original_string


    def get_base_type_string(self):
        return self.type_expr.get_base_name()


    def get_dimensions(self):
        if not self.width:
            return [ 0, "0", "0", "0", "0" ]

        dim = 1
        w = self.width
        h = "1"
        d = "1"
        e = "1"

        if self.height:
            dim = 2
            h = self.height

        if self.depth:
            dim = 3
            d = self.depth

        if self.extent:
            dim = 4
            e = self.extent

        return [ dim, w, h, d, e ]


    def get_stack_size(self):
        return self.type_expr.get_stack_size()


    def size(self):
        if self.is_image():
            return 0
        else:
            return self.type_expr.get_element_size()


    def get_element_count(self):
        c = self.type_expr.get_element_count()
        if c == 0:
            return 1

        return c


    def size_string(self, use_parens = 1):
        s = self.size()
        if self.counter or self.count_parameter_list:
            list = [ "compsize" ]

            if self.counter and self.count_parameter_list:
                list.append( self.counter )
            elif self.counter:
                list = [ self.counter ]

            if s > 1:
                list.append( str(s) )

            if len(list) > 1 and use_parens :
                return "(%s)" % (string.join(list, " * "))
            else:
                return string.join(list, " * ")

        elif self.is_image():
            return "compsize"
        else:
            return str(s)


    def format_string(self):
        if self.type_expr.original_string == "GLenum":
            return "0x%x"
        else:
            return self.type_expr.format_string()


# Regular expression used to parse "mesa_name" attributes.  A
# mesa_name attribute describes how to adjust a GL function name
# suffix to obtain the name of the function in Mesa that implements
# the functionality.  The attribute string consists of a part preceded
# by a "-", indicating the suffix to remove, and a part preceded by a
# "+" indicating the suffix to add.  Either part is optional.
#
# For example:
#
#   <function name="EnableIndexedEXT" mesa_name="-EXT">...</function>
#   <function name="IsProgramNV" mesa_name="-NV+ARB">...</function>
#
# means that EnableIndexedEXT is implemented by a Mesa function called
# _mesa_EnableIndexed, and IsProgramNV is implemented by a Mesa function
# called _mesa_IsProgramARB.
#
# Note: the prefix "_mesa_" is handled separately, by the "exec"
# attribute.
name_modification_regexp = re.compile(
    r'^(-(?P<minus>[a-zA-Z0-9_]+))?(\+(?P<plus>[a-zA-Z0-9_]+))?$')


# Interpret a "mesa_name" attribute (see above) to determine the
# appropriate suffix for the Mesa function implementing a piece of GL
# functionality, and return the properly suffixed name.
def interpret_name_modification(name, mod):
    m = name_modification_regexp.match(mod)
    if m is None:
        raise Exception('Unintelligible mesa_name property: {0!r}'.format(mod))
    new_name = name
    if m.group('minus'):
        if not new_name.endswith(m.group('minus')):
            raise Exception(
                'Cannot subtract suffix {0!r} from function {1}'.format(
                    m.group('minus'), name))
        new_name = new_name[:-len(m.group('minus'))]
    if m.group('plus'):
        new_name += m.group('plus')
    return new_name


class gl_function( gl_item ):
    def __init__(self, element, context):
        self.context = context
        self.name = None

        self.entry_points = []
        self.return_type = "void"
        self.parameters = []
        self.offset = -1
        self.initialized = 0
        self.images = []
        self.exec_flavor = 'mesa'
        self.desktop = True
        self.deprecated = None
        self.mesa_name = None

        # self.entry_point_api_map[name][api] is a decimal value
        # indicating the earliest version of the given API in which
        # each entry point exists.  Every entry point is included in
        # the first level of the map; the second level of the map only
        # lists APIs which contain the entry point in at least one
        # version.  For example,
        # self.entry_point_api_map['ClipPlanex'] == { 'es1':
        # Decimal('1.1') }.
        self.entry_point_api_map = {}

        # self.api_map[api] is a decimal value indicating the earliest
        # version of the given API in which ANY alias for the function
        # exists.  The map only lists APIs which contain the function
        # in at least one version.  For example, for the ClipPlanex
        # function, self.entry_point_api_map == { 'es1':
        # Decimal('1.1') }.
        self.api_map = {}

        self.assign_offset = 0

        self.static_entry_points = []

        # Track the parameter string (for the function prototype)
        # for each entry-point.  This is done because some functions
        # change their prototype slightly when promoted from extension
        # to ARB extension to core.  glTexImage3DEXT and glTexImage3D
        # are good examples of this.  Scripts that need to generate
        # code for these differing aliases need to real prototype
        # for each entry-point.  Otherwise, they may generate code
        # that won't compile.

        self.entry_point_parameters = {}

        self.process_element( element )

        return


    def process_element(self, element):
        name = element.nsProp( "name", None )
        alias = element.nsProp( "alias", None )

        if is_attr_true(element, "static_dispatch"):
            self.static_entry_points.append(name)

        self.entry_points.append( name )

        self.entry_point_api_map[name] = {}
        for api in ('es1', 'es2'):
            version_str = element.nsProp(api, None)
            assert version_str is not None
            if version_str != 'none':
                version_decimal = Decimal(version_str)
                self.entry_point_api_map[name][api] = version_decimal
                if api not in self.api_map or \
                        version_decimal < self.api_map[api]:
                    self.api_map[api] = version_decimal

        exec_flavor = element.nsProp('exec', None)
        if exec_flavor:
            self.exec_flavor = exec_flavor

        deprecated = element.nsProp('deprecated', None)
        if deprecated != 'none':
            self.deprecated = Decimal(deprecated)

        if not is_attr_true(element, 'desktop'):
            self.desktop = False

        if alias:
            true_name = alias
        else:
            true_name = name

            # Only try to set the offset and mesa_name when a
            # non-alias entry-point is being processed.

            offset = element.nsProp( "offset", None )
            if offset:
                try:
                    o = int( offset )
                    self.offset = o
                except Exception, e:
                    self.offset = -1
                    if offset == "assign":
                        self.assign_offset = 1

            mesa_name = element.nsProp('mesa_name', None)
            if mesa_name is None:
                self.mesa_name = name
            else:
                self.mesa_name = interpret_name_modification(name, mesa_name)


        if not self.name:
            self.name = true_name
        elif self.name != true_name:
            raise RuntimeError("Function true name redefined.  Was %s, now %s." % (self.name, true_name))


        # There are two possible cases.  The first time an entry-point
        # with data is seen, self.initialized will be 0.  On that
        # pass, we just fill in the data.  The next time an
        # entry-point with data is seen, self.initialized will be 1.
        # On that pass we have to make that the new values match the
        # valuse from the previous entry-point.

        parameters = []
        return_type = "void"
        child = element.children
        while child:
            if child.type == "element":
                if child.name == "return":
                    return_type = child.nsProp( "type", None )
                elif child.name == "param":
                    param = self.context.factory.create_item( "parameter", child, self.context)
                    parameters.append( param )

            child = child.next


        if self.initialized:
            if self.return_type != return_type:
                raise RuntimeError( "Return type changed in %s.  Was %s, now %s." % (name, self.return_type, return_type))

            if len(parameters) != len(self.parameters):
                raise RuntimeError( "Parameter count mismatch in %s.  Was %d, now %d." % (name, len(self.parameters), len(parameters)))

            for j in range(0, len(parameters)):
                p1 = parameters[j]
                p2 = self.parameters[j]
                if not p1.compatible( p2 ):
                    raise RuntimeError( 'Parameter type mismatch in %s.  "%s" was "%s", now "%s".' % (name, p2.name, p2.type_expr.original_string, p1.type_expr.original_string))


        if true_name == name or not self.initialized:
            self.return_type = return_type
            self.parameters = parameters

            for param in self.parameters:
                if param.is_image():
                    self.images.append( param )

        if element.children:
            self.initialized = 1
            self.entry_point_parameters[name] = parameters
        else:
            self.entry_point_parameters[name] = []

        return

    def filter_entry_points(self, entry_point_list):
        """Filter out entry points not in entry_point_list."""
        if not self.initialized:
            raise RuntimeError('%s is not initialized yet' % self.name)

        entry_points = []
        for ent in self.entry_points:
            if ent not in entry_point_list:
                if ent in self.static_entry_points:
                    self.static_entry_points.remove(ent)
                self.entry_point_parameters.pop(ent)
            else:
                entry_points.append(ent)

        if not entry_points:
            raise RuntimeError('%s has no entry point after filtering' % self.name)

        self.entry_points = entry_points
        if self.name not in entry_points:
            # use the first remaining entry point
            self.name = entry_points[0]
            self.parameters = self.entry_point_parameters[entry_points[0]]

    def get_images(self):
        """Return potentially empty list of input images."""
        return self.images


    def parameterIterator(self, name = None):
        if name is not None:
            return self.entry_point_parameters[name].__iter__();
        else:
            return self.parameters.__iter__();


    def get_parameter_string(self, entrypoint = None):
        if entrypoint:
            params = self.entry_point_parameters[ entrypoint ]
        else:
            params = self.parameters

        return create_parameter_string( params, 1 )

    def get_called_parameter_string(self):
        p_string = ""
        comma = ""

        for p in self.parameterIterator():
            p_string = p_string + comma + p.name
            comma = ", "

        return p_string


    def is_abi(self):
        return (self.offset >= 0 and not self.assign_offset)

    def is_static_entry_point(self, name):
        return name in self.static_entry_points

    def dispatch_name(self):
        if self.name in self.static_entry_points:
            return self.name
        else:
            return "_dispatch_stub_%u" % (self.offset)

    def static_name(self, name):
        if name in self.static_entry_points:
            return name
        else:
            return "_dispatch_stub_%u" % (self.offset)

    def entry_points_for_api_version(self, api, version = None):
        """Return a list of the entry point names for this function
        which are supported in the given API (and optionally, version).

        Use the decimal.Decimal type to precisely express non-integer
        versions.
        """
        result = []
        for entry_point, api_to_ver in self.entry_point_api_map.iteritems():
            if api not in api_to_ver:
                continue
            if version is not None and version < api_to_ver[api]:
                continue
            result.append(entry_point)
        return result


class gl_item_factory(object):
    """Factory to create objects derived from gl_item."""

    def create_item(self, item_name, element, context):
        if item_name == "function":
            return gl_function(element, context)
        if item_name == "type":
            return gl_type(element, context)
        elif item_name == "enum":
            return gl_enum(element, context)
        elif item_name == "parameter":
            return gl_parameter(element, context)
        elif item_name == "api":
            return gl_api(self)
        else:
            return None


class gl_api(object):
    def __init__(self, factory):
        self.functions_by_name = {}
        self.enums_by_name = {}
        self.types_by_name = {}

        self.category_dict = {}
        self.categories = [{}, {}, {}, {}]

        self.factory = factory

        self.next_offset = 0

        typeexpr.create_initial_types()
        return

    def filter_functions(self, entry_point_list):
        """Filter out entry points not in entry_point_list."""
        functions_by_name = {}
        for func in self.functions_by_name.itervalues():
            entry_points = [ent for ent in func.entry_points if ent in entry_point_list]
            if entry_points:
                func.filter_entry_points(entry_points)
                functions_by_name[func.name] = func

        self.functions_by_name = functions_by_name

    def filter_functions_by_api(self, api, version = None):
        """Filter out entry points not in the given API (or
        optionally, not in the given version of the given API).
        """
        functions_by_name = {}
        for func in self.functions_by_name.itervalues():
            entry_points = func.entry_points_for_api_version(api, version)
            if entry_points:
                func.filter_entry_points(entry_points)
                functions_by_name[func.name] = func

        self.functions_by_name = functions_by_name

    def process_element(self, doc):
        element = doc.children
        while element.type != "element" or element.name != "OpenGLAPI":
            element = element.next

        if element:
            self.process_OpenGLAPI(element)
        return


    def process_OpenGLAPI(self, element):
        child = element.children
        while child:
            if child.type == "element":
                if child.name == "category":
                    self.process_category( child )
                elif child.name == "OpenGLAPI":
                    self.process_OpenGLAPI( child )

            child = child.next

        return


    def process_category(self, cat):
        cat_name = cat.nsProp( "name", None )
        cat_number = cat.nsProp( "number", None )

        [cat_type, key] = classify_category(cat_name, cat_number)
        self.categories[cat_type][key] = [cat_name, cat_number]

        child = cat.children
        while child:
            if child.type == "element":
                if child.name == "function":
                    func_name = real_function_name( child )

                    temp_name = child.nsProp( "name", None )
                    self.category_dict[ temp_name ] = [cat_name, cat_number]

                    if self.functions_by_name.has_key( func_name ):
                        func = self.functions_by_name[ func_name ]
                        func.process_element( child )
                    else:
                        func = self.factory.create_item( "function", child, self )
                        self.functions_by_name[ func_name ] = func

                    if func.offset >= self.next_offset:
                        self.next_offset = func.offset + 1


                elif child.name == "enum":
                    enum = self.factory.create_item( "enum", child, self )
                    self.enums_by_name[ enum.name ] = enum
                elif child.name == "type":
                    t = self.factory.create_item( "type", child, self )
                    self.types_by_name[ "GL" + t.name ] = t


            child = child.next

        return


    def functionIterateByCategory(self, cat = None):
        """Iterate over functions by category.

        If cat is None, all known functions are iterated in category
        order.  See classify_category for details of the ordering.
        Within a category, functions are sorted by name.  If cat is
        not None, then only functions in that category are iterated.
        """
        lists = [{}, {}, {}, {}]

        for func in self.functionIterateAll():
            [cat_name, cat_number] = self.category_dict[func.name]

            if (cat == None) or (cat == cat_name):
                [func_cat_type, key] = classify_category(cat_name, cat_number)

                if not lists[func_cat_type].has_key(key):
                    lists[func_cat_type][key] = {}

                lists[func_cat_type][key][func.name] = func


        functions = []
        for func_cat_type in range(0,4):
            keys = lists[func_cat_type].keys()
            keys.sort()

            for key in keys:
                names = lists[func_cat_type][key].keys()
                names.sort()

                for name in names:
                    functions.append(lists[func_cat_type][key][name])

        return functions.__iter__()


    def functionIterateByOffset(self):
        max_offset = -1
        for func in self.functions_by_name.itervalues():
            if func.offset > max_offset:
                max_offset = func.offset


        temp = [None for i in range(0, max_offset + 1)]
        for func in self.functions_by_name.itervalues():
            if func.offset != -1:
                temp[ func.offset ] = func


        list = []
        for i in range(0, max_offset + 1):
            if temp[i]:
                list.append(temp[i])

        return list.__iter__();


    def functionIterateAll(self):
        return self.functions_by_name.itervalues()


    def enumIterateByName(self):
        keys = self.enums_by_name.keys()
        keys.sort()

        list = []
        for enum in keys:
            list.append( self.enums_by_name[ enum ] )

        return list.__iter__()


    def categoryIterate(self):
        """Iterate over categories.

        Iterate over all known categories in the order specified by
        classify_category.  Each iterated value is a tuple of the
        name and number (which may be None) of the category.
        """

        list = []
        for cat_type in range(0,4):
            keys = self.categories[cat_type].keys()
            keys.sort()

            for key in keys:
                list.append(self.categories[cat_type][key])

        return list.__iter__()


    def get_category_for_name( self, name ):
        if self.category_dict.has_key(name):
            return self.category_dict[name]
        else:
            return ["<unknown category>", None]


    def typeIterate(self):
        return self.types_by_name.itervalues()


    def find_type( self, type_name ):
        if type_name in self.types_by_name:
            return self.types_by_name[ type_name ].type_expr
        else:
            print "Unable to find base type matching \"%s\"." % (type_name)
            return None