Script for making cut-down version of pyyaml that is a single-file yaml loader

make_yaml_decoder_module.py

#!/usr/bin/env python

"""Command-line utility to combine the necessary modules to make a single-file yaml_decoder.py"""

import shutil
import os
from os.path import abspath, dirname, exists, join

include_modules = 'loader reader parser scanner composer constructor resolver nodes events tokens error'.split()
# this should be placed in the pyyaml source tree
root_dir = dirname(abspath(__file__))
src_dir = join(root_dir, 'lib', 'yaml')
dist_dir = join(root_dir, 'dist')
if not exists(dist_dir):
    os.makedirs(dist_dir)
output_file = join(dist_dir, 'yaml_decoder.py')

header = '''#!/usr/bin/env python

"""yaml_decoder is a stripped-down bundle of pyyaml that offers only decoding/loading (not encoding/dumping) in a single pure-python module"""

# See https://gist.github.com/davidfraser/3d4b19ad41d7a7d0409eb92cd5907016 for updated versions

'''

with open(output_file, 'w') as out_f:
    out_f.write(header)
    with open(join(root_dir, 'LICENSE')) as in_f:
        for line in in_f:
            out_f.write(('# %s' % line).rstrip() + '\n')
    # include the first part of the file, before the functions
    with open(join(src_dir, '__init__.py')) as in_f:
        for line in in_f:
            if line.startswith('from '):
                module_name = line.split()[1]
                if module_name in include_modules:
                    continue
            elif line.startswith('def '):
                break
            else:
                out_f.write(line)
    out_f.write('\n\n')
    for module in reversed(include_modules):
        filename = '%s.py' % module
        out_f.write('\n\n# including %s\n\n' % filename)
        with open(join(src_dir, filename)) as in_f:
            await_closebracket = False
            for line in in_f:
                if line.startswith('from '):
                    module_name = line.split()[1]
                    if module_name in include_modules:
                        continue
                elif line.startswith('__all__'):
                    if '[' in line and ']' not in line:
                        await_closebracket = True
                elif await_closebracket and ']' in line:
                    await_closebracket = False
                elif not await_closebracket:
                    out_f.write(line.replace('pureyaml.grammar.', 'pureyaml_grammar_'))
    # include the load functions
    out_f.write('\n\n# load functions from __init__.py\n\n')
    with open(join(src_dir, '__init__.py')) as in_f:
        in_load, post_load = False, False
        for line in in_f:
            if line.startswith('def '):
                in_load = True
            if line.startswith('def emit'):
                post_load = True
            if in_load and not post_load:
                out_f.write(line)