524 lines
19 KiB
Python
524 lines
19 KiB
Python
#!/usr/bin/env python2
|
|
# Copyright 2017 The NXT Authors
|
|
#
|
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
# you may not use this file except in compliance with the License.
|
|
# You may obtain a copy of the License at
|
|
#
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
#
|
|
# Unless required by applicable law or agreed to in writing, software
|
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
# See the License for the specific language governing permissions and
|
|
# limitations under the License.
|
|
|
|
############################################################
|
|
# COMMON
|
|
############################################################
|
|
from collections import namedtuple
|
|
|
|
class Name:
|
|
def __init__(self, name, native=False):
|
|
self.native = native
|
|
if native:
|
|
self.chunks = [name]
|
|
else:
|
|
self.chunks = name.split(' ')
|
|
|
|
def CamelChunk(self, chunk):
|
|
return chunk[0].upper() + chunk[1:]
|
|
|
|
def canonical_case(self):
|
|
return (' '.join(self.chunks)).lower()
|
|
|
|
def concatcase(self):
|
|
return ''.join(self.chunks)
|
|
|
|
def camelCase(self):
|
|
return self.chunks[0] + ''.join([self.CamelChunk(chunk) for chunk in self.chunks[1:]])
|
|
|
|
def CamelCase(self):
|
|
return ''.join([self.CamelChunk(chunk) for chunk in self.chunks])
|
|
|
|
def SNAKE_CASE(self):
|
|
return '_'.join([chunk.upper() for chunk in self.chunks])
|
|
|
|
def snake_case(self):
|
|
return '_'.join(self.chunks)
|
|
|
|
class Type:
|
|
def __init__(self, name, record, native=False):
|
|
self.record = record
|
|
self.dict_name = name
|
|
self.name = Name(name, native=native)
|
|
self.category = record['category']
|
|
self.is_builder = self.name.canonical_case().endswith(" builder")
|
|
|
|
EnumValue = namedtuple('EnumValue', ['name', 'value'])
|
|
class EnumType(Type):
|
|
def __init__(self, name, record):
|
|
Type.__init__(self, name, record)
|
|
self.values = [EnumValue(Name(m['name']), m['value']) for m in self.record['values']]
|
|
|
|
BitmaskValue = namedtuple('BitmaskValue', ['name', 'value'])
|
|
class BitmaskType(Type):
|
|
def __init__(self, name, record):
|
|
Type.__init__(self, name, record)
|
|
self.values = [BitmaskValue(Name(m['name']), m['value']) for m in self.record['values']]
|
|
self.full_mask = 0
|
|
for value in self.values:
|
|
self.full_mask = self.full_mask | value.value
|
|
|
|
class NativeType(Type):
|
|
def __init__(self, name, record):
|
|
Type.__init__(self, name, record, native=True)
|
|
|
|
class NativelyDefined(Type):
|
|
def __init__(self, name, record):
|
|
Type.__init__(self, name, record)
|
|
|
|
class MethodArgument:
|
|
def __init__(self, name, typ, annotation):
|
|
self.name = name
|
|
self.type = typ
|
|
self.annotation = annotation
|
|
self.length = None
|
|
|
|
Method = namedtuple('Method', ['name', 'return_type', 'arguments'])
|
|
class ObjectType(Type):
|
|
def __init__(self, name, record):
|
|
Type.__init__(self, name, record)
|
|
self.methods = []
|
|
self.native_methods = []
|
|
self.built_type = None
|
|
|
|
class StructureMember:
|
|
def __init__(self, name, typ, annotation):
|
|
self.name = name
|
|
self.type = typ
|
|
self.annotation = annotation
|
|
self.length = None
|
|
|
|
class StructureType(Type):
|
|
def __init__(self, name, record):
|
|
Type.__init__(self, name, record)
|
|
self.extensible = record.get("extensible", False)
|
|
self.members = []
|
|
|
|
############################################################
|
|
# PARSE
|
|
############################################################
|
|
import json
|
|
|
|
def is_native_method(method):
|
|
return method.return_type.category == "natively defined" or \
|
|
any([arg.type.category == "natively defined" for arg in method.arguments])
|
|
|
|
def link_object(obj, types):
|
|
def make_method(record):
|
|
arguments = []
|
|
arguments_by_name = {}
|
|
for a in record.get('args', []):
|
|
arg = MethodArgument(Name(a['name']), types[a['type']], a.get('annotation', 'value'))
|
|
arguments.append(arg)
|
|
arguments_by_name[arg.name.canonical_case()] = arg
|
|
|
|
for (arg, a) in zip(arguments, record.get('args', [])):
|
|
if arg.annotation != 'value':
|
|
if not 'length' in a:
|
|
if arg.type.category == 'structure':
|
|
arg.length = "constant"
|
|
arg.constant_length = 1
|
|
else:
|
|
assert(false)
|
|
elif a['length'] == 'strlen':
|
|
arg.length = 'strlen'
|
|
else:
|
|
arg.length = arguments_by_name[a['length']]
|
|
|
|
return Method(Name(record['name']), types[record.get('returns', 'void')], arguments)
|
|
|
|
methods = [make_method(m) for m in obj.record.get('methods', [])]
|
|
obj.methods = [method for method in methods if not is_native_method(method)]
|
|
obj.native_methods = [method for method in methods if is_native_method(method)]
|
|
|
|
# Compute the built object type for builders
|
|
if obj.is_builder:
|
|
for method in obj.methods:
|
|
if method.name.canonical_case() == "get result":
|
|
obj.built_type = method.return_type
|
|
break
|
|
assert(obj.built_type != None)
|
|
|
|
def link_structure(struct, types):
|
|
def make_member(m):
|
|
return StructureMember(Name(m['name']), types[m['type']], m.get('annotation', 'value'))
|
|
|
|
members = []
|
|
members_by_name = {}
|
|
for m in struct.record['members']:
|
|
member = make_member(m)
|
|
members.append(member)
|
|
members_by_name[member.name.canonical_case()] = member
|
|
struct.members = members
|
|
|
|
for (member, m) in zip(members, struct.record['members']):
|
|
# TODO(kainino@chromium.org): More robust pointer/length handling?
|
|
if 'length' in m:
|
|
member.length = members_by_name[m['length']]
|
|
|
|
def parse_json(json):
|
|
category_to_parser = {
|
|
'bitmask': BitmaskType,
|
|
'enum': EnumType,
|
|
'native': NativeType,
|
|
'natively defined': NativelyDefined,
|
|
'object': ObjectType,
|
|
'structure': StructureType,
|
|
}
|
|
|
|
types = {}
|
|
|
|
by_category = {}
|
|
for name in category_to_parser.keys():
|
|
by_category[name] = []
|
|
|
|
for (name, record) in json.items():
|
|
if name[0] == '_':
|
|
continue
|
|
category = record['category']
|
|
parsed = category_to_parser[category](name, record)
|
|
by_category[category].append(parsed)
|
|
types[name] = parsed
|
|
|
|
for obj in by_category['object']:
|
|
link_object(obj, types)
|
|
|
|
for struct in by_category['structure']:
|
|
link_structure(struct, types)
|
|
|
|
for category in by_category.keys():
|
|
by_category[category] = sorted(by_category[category], key=lambda typ: typ.name.canonical_case())
|
|
|
|
return {
|
|
'types': types,
|
|
'by_category': by_category
|
|
}
|
|
|
|
#############################################################
|
|
# OUTPUT
|
|
#############################################################
|
|
import re, os, sys
|
|
from collections import OrderedDict
|
|
|
|
try:
|
|
import jinja2
|
|
except ImportError:
|
|
# Try using Chromium's Jinja2
|
|
dir, _ = os.path.split(os.path.realpath(__file__))
|
|
third_party_dir = os.path.normpath(dir + (os.path.sep + os.path.pardir) * 2)
|
|
sys.path.insert(1, third_party_dir)
|
|
import jinja2
|
|
|
|
# A custom Jinja2 template loader that removes the extra indentation
|
|
# of the template blocks so that the output is correctly indented
|
|
class PreprocessingLoader(jinja2.BaseLoader):
|
|
def __init__(self, path):
|
|
self.path = path
|
|
|
|
def get_source(self, environment, template):
|
|
path = os.path.join(self.path, template)
|
|
if not os.path.exists(path):
|
|
raise jinja2.TemplateNotFound(template)
|
|
mtime = os.path.getmtime(path)
|
|
with open(path) as f:
|
|
source = self.preprocess(f.read())
|
|
return source, path, lambda: mtime == os.path.getmtime(path)
|
|
|
|
blockstart = re.compile('{%-?\s*(if|for|block)[^}]*%}')
|
|
blockend = re.compile('{%-?\s*end(if|for|block)[^}]*%}')
|
|
|
|
def preprocess(self, source):
|
|
lines = source.split('\n')
|
|
|
|
# Compute the current indentation level of the template blocks and remove their indentation
|
|
result = []
|
|
indentation_level = 0
|
|
|
|
for line in lines:
|
|
# The capture in the regex adds one element per block start or end so we divide by two
|
|
# there is also an extra line chunk corresponding to the line end, so we substract it.
|
|
numends = (len(self.blockend.split(line)) - 1) // 2
|
|
indentation_level -= numends
|
|
|
|
line = self.remove_indentation(line, indentation_level)
|
|
|
|
# Manually perform the lstrip_blocks jinja2 env options as it available starting from 2.7
|
|
# and Travis only has Jinja 2.6
|
|
if line.lstrip().startswith('{%'):
|
|
line = line.lstrip()
|
|
|
|
result.append(line)
|
|
|
|
numstarts = (len(self.blockstart.split(line)) - 1) // 2
|
|
indentation_level += numstarts
|
|
|
|
return '\n'.join(result) + '\n'
|
|
|
|
def remove_indentation(self, line, n):
|
|
for _ in range(n):
|
|
if line.startswith(' '):
|
|
line = line[4:]
|
|
elif line.startswith('\t'):
|
|
line = line[1:]
|
|
else:
|
|
assert(line.strip() == '')
|
|
return line
|
|
|
|
FileRender = namedtuple('FileRender', ['template', 'output', 'params_dicts'])
|
|
|
|
def do_renders(renders, template_dir, output_dir):
|
|
env = jinja2.Environment(loader=PreprocessingLoader(template_dir), trim_blocks=True, line_comment_prefix='//*')
|
|
for render in renders:
|
|
params = {}
|
|
for param_dict in render.params_dicts:
|
|
params.update(param_dict)
|
|
output = env.get_template(render.template).render(**params)
|
|
|
|
output_file = output_dir + os.path.sep + render.output
|
|
directory = os.path.dirname(output_file)
|
|
if not os.path.exists(directory):
|
|
os.makedirs(directory)
|
|
|
|
content = ""
|
|
try:
|
|
with open(output_file, 'r') as outfile:
|
|
content = outfile.read()
|
|
except:
|
|
pass
|
|
|
|
if output != content:
|
|
with open(output_file, 'w') as outfile:
|
|
outfile.write(output)
|
|
|
|
#############################################################
|
|
# MAIN SOMETHING WHATEVER
|
|
#############################################################
|
|
import argparse, sys
|
|
|
|
def as_varName(*names):
|
|
return names[0].camelCase() + ''.join([name.CamelCase() for name in names[1:]])
|
|
|
|
def as_cType(name):
|
|
if name.native:
|
|
return name.concatcase()
|
|
else:
|
|
return 'nxt' + name.CamelCase()
|
|
|
|
def as_cppType(name):
|
|
if name.native:
|
|
return name.concatcase()
|
|
else:
|
|
return name.CamelCase()
|
|
|
|
def decorate(name, typ, arg):
|
|
if arg.annotation == 'value':
|
|
return typ + ' ' + name
|
|
elif arg.annotation == '*':
|
|
return typ + '* ' + name
|
|
elif arg.annotation == 'const*':
|
|
return typ + ' const * ' + name
|
|
else:
|
|
assert(False)
|
|
|
|
def annotated(typ, arg):
|
|
name = as_varName(arg.name)
|
|
return decorate(name, typ, arg)
|
|
|
|
def as_cEnum(type_name, value_name):
|
|
assert(not type_name.native and not value_name.native)
|
|
return 'NXT' + '_' + type_name.SNAKE_CASE() + '_' + value_name.SNAKE_CASE()
|
|
|
|
def as_cppEnum(value_name):
|
|
assert(not value_name.native)
|
|
if value_name.concatcase()[0].isdigit():
|
|
return "e" + value_name.CamelCase()
|
|
return value_name.CamelCase()
|
|
|
|
def as_cMethod(type_name, method_name):
|
|
assert(not type_name.native and not method_name.native)
|
|
return 'nxt' + type_name.CamelCase() + method_name.CamelCase()
|
|
|
|
def as_MethodSuffix(type_name, method_name):
|
|
assert(not type_name.native and not method_name.native)
|
|
return type_name.CamelCase() + method_name.CamelCase()
|
|
|
|
def as_cProc(type_name, method_name):
|
|
assert(not type_name.native and not method_name.native)
|
|
return 'nxt' + 'Proc' + type_name.CamelCase() + method_name.CamelCase()
|
|
|
|
def as_backendType(typ):
|
|
if typ.category == 'object':
|
|
return typ.name.CamelCase() + '*'
|
|
else:
|
|
return as_cType(typ.name)
|
|
|
|
def cpp_native_methods(types, typ):
|
|
methods = typ.methods + typ.native_methods
|
|
|
|
if typ.is_builder:
|
|
methods.append(Method(Name('set error callback'), types['void'], [
|
|
MethodArgument(Name('callback'), types['builder error callback'], 'value'),
|
|
MethodArgument(Name('userdata1'), types['callback userdata'], 'value'),
|
|
MethodArgument(Name('userdata2'), types['callback userdata'], 'value'),
|
|
]))
|
|
|
|
return methods
|
|
|
|
def c_native_methods(types, typ):
|
|
return cpp_native_methods(types, typ) + [
|
|
Method(Name('reference'), types['void'], []),
|
|
Method(Name('release'), types['void'], []),
|
|
]
|
|
|
|
def js_native_methods(types, typ):
|
|
return cpp_native_methods(types, typ)
|
|
|
|
def debug(text):
|
|
print(text)
|
|
|
|
def main():
|
|
targets = ['nxt', 'nxtcpp', 'mock_nxt', 'opengl', 'metal', 'd3d12', 'null', 'wire', 'blink']
|
|
|
|
parser = argparse.ArgumentParser(
|
|
description = 'Generates code for various target for NXT.',
|
|
formatter_class = argparse.ArgumentDefaultsHelpFormatter
|
|
)
|
|
parser.add_argument('json', metavar='NXT_JSON', nargs=1, type=str, help ='The NXT JSON definition to use.')
|
|
parser.add_argument('-t', '--template-dir', default='templates', type=str, help='Directory with template files.')
|
|
parser.add_argument('-o', '--output-dir', default=None, type=str, help='Output directory for the generated source files.')
|
|
parser.add_argument('-T', '--targets', default=None, type=str, help='Comma-separated subset of targets to output. Available targets: ' + ', '.join(targets))
|
|
parser.add_argument('--print-dependencies', action='store_true', help='Prints a space separated list of file dependencies, used for CMake integration')
|
|
parser.add_argument('--print-outputs', action='store_true', help='Prints a space separated list of file outputs, used for CMake integration')
|
|
parser.add_argument('--gn', action='store_true', help='Make the printing of dependencies by GN friendly')
|
|
|
|
args = parser.parse_args()
|
|
|
|
if args.targets != None:
|
|
targets = args.targets.split(',')
|
|
|
|
with open(args.json[0]) as f:
|
|
loaded_json = json.loads(f.read())
|
|
|
|
api_params = parse_json(loaded_json)
|
|
|
|
base_params = {
|
|
'enumerate': enumerate,
|
|
'format': format,
|
|
'len': len,
|
|
'debug': debug,
|
|
|
|
'Name': lambda name: Name(name),
|
|
|
|
'as_annotated_cType': lambda arg: annotated(as_cType(arg.type.name), arg),
|
|
'as_annotated_cppType': lambda arg: annotated(as_cppType(arg.type.name), arg),
|
|
'as_cEnum': as_cEnum,
|
|
'as_cppEnum': as_cppEnum,
|
|
'as_cMethod': as_cMethod,
|
|
'as_MethodSuffix': as_MethodSuffix,
|
|
'as_cProc': as_cProc,
|
|
'as_cType': as_cType,
|
|
'as_cppType': as_cppType,
|
|
'as_varName': as_varName,
|
|
'decorate': decorate,
|
|
}
|
|
|
|
renders = []
|
|
|
|
c_params = {'native_methods': lambda typ: c_native_methods(api_params['types'], typ)}
|
|
|
|
if 'nxt' in targets:
|
|
renders.append(FileRender('api.h', 'nxt/nxt.h', [base_params, api_params, c_params]))
|
|
renders.append(FileRender('api.c', 'nxt/nxt.c', [base_params, api_params, c_params]))
|
|
|
|
if 'nxtcpp' in targets:
|
|
additional_params = {'native_methods': lambda typ: cpp_native_methods(api_params['types'], typ)}
|
|
renders.append(FileRender('apicpp.h', 'nxt/nxtcpp.h', [base_params, api_params, additional_params]))
|
|
renders.append(FileRender('apicpp.cpp', 'nxt/nxtcpp.cpp', [base_params, api_params, additional_params]))
|
|
renders.append(FileRender('apicpp_traits.h', 'nxt/nxtcpp_traits.h', [base_params, api_params, additional_params]))
|
|
|
|
if 'mock_nxt' in targets:
|
|
renders.append(FileRender('mock_api.h', 'mock/mock_nxt.h', [base_params, api_params, c_params]))
|
|
renders.append(FileRender('mock_api.cpp', 'mock/mock_nxt.cpp', [base_params, api_params, c_params]))
|
|
|
|
base_backend_params = [
|
|
base_params,
|
|
api_params,
|
|
c_params,
|
|
{
|
|
'as_backendType': lambda typ: as_backendType(typ), # TODO as_backendType and friends take a Type and not a Name :(
|
|
'as_annotated_backendType': lambda arg: annotated(as_backendType(arg.type), arg)
|
|
}
|
|
]
|
|
|
|
for backend in ['d3d12', 'metal', 'null', 'opengl', 'vulkan']:
|
|
if not backend in targets:
|
|
continue
|
|
|
|
extension = 'cpp'
|
|
if backend == 'metal':
|
|
extension = 'mm'
|
|
|
|
backend_params = {
|
|
'namespace': backend,
|
|
}
|
|
renders.append(FileRender('backend/ProcTable.cpp', 'backend/' + backend + '/ProcTable.' + extension, base_backend_params + [backend_params]))
|
|
|
|
if 'backend_utils' in targets:
|
|
renders.append(FileRender('backend/ValidationUtils.h', 'backend/ValidationUtils_autogen.h', base_backend_params))
|
|
renders.append(FileRender('backend/ValidationUtils.cpp', 'backend/ValidationUtils_autogen.cpp', base_backend_params))
|
|
|
|
if 'wire' in targets:
|
|
renders.append(FileRender('wire/WireCmd.h', 'wire/WireCmd_autogen.h', base_backend_params))
|
|
renders.append(FileRender('wire/WireCmd.cpp', 'wire/WireCmd_autogen.cpp', base_backend_params))
|
|
renders.append(FileRender('wire/WireClient.cpp', 'wire/WireClient.cpp', base_backend_params))
|
|
renders.append(FileRender('wire/WireServer.cpp', 'wire/WireServer.cpp', base_backend_params))
|
|
|
|
if 'blink' in targets:
|
|
js_params = {'native_methods': lambda typ: js_native_methods(api_params['types'], typ)}
|
|
renders.append(FileRender('autogen.gni', 'autogen.gni', [base_params, api_params, js_params]))
|
|
renders.append(FileRender('Objects.cpp', 'NXT.cpp', [base_params, api_params, js_params]))
|
|
renders.append(FileRender('Forward.h', 'Forward.h', [base_params, api_params, js_params]))
|
|
|
|
for typ in api_params['by_category']['object']:
|
|
file_prefix = 'NXT' + typ.name.CamelCase()
|
|
params = [base_params, api_params, js_params, {'type': typ}]
|
|
|
|
renders.append(FileRender('Object.h', file_prefix + '.h', params))
|
|
renders.append(FileRender('Object.idl', file_prefix + '.idl', params))
|
|
|
|
output_separator = '\n' if args.gn else ';'
|
|
if args.print_dependencies:
|
|
dependencies = set(
|
|
[os.path.abspath(args.template_dir + os.path.sep + render.template) for render in renders] +
|
|
[os.path.abspath(args.json[0])] +
|
|
[os.path.realpath(__file__)]
|
|
)
|
|
dependencies = [dependency.replace('\\', '/') for dependency in dependencies]
|
|
sys.stdout.write(output_separator.join(dependencies))
|
|
return 0
|
|
|
|
if args.print_outputs:
|
|
outputs = set(
|
|
[os.path.abspath(args.output_dir + os.path.sep + render.output) for render in renders]
|
|
)
|
|
outputs = [output.replace('\\', '/') for output in outputs]
|
|
sys.stdout.write(output_separator.join(outputs))
|
|
return 0
|
|
|
|
do_renders(renders, args.template_dir, args.output_dir)
|
|
|
|
if __name__ == '__main__':
|
|
sys.exit(main())
|