source: nagslang/yamlish.py@ 120:e6e7a471146a

'''
Serializer and dumper for a simple, YAMLish format (actually a YAML subset).
The top level object is a dict.
lists and dicts can contain:
 * lists, dicts,
 * single line strings,
 * ints, floats,
 * True, False, and None
'''

import re


def dump(data, file_object):
    file_object.write(dump_s(data))


def dump_s(data):
    return Dumper().dump(data)


def load(file_object):
    yaml = file_object.read()
    return load_s(yaml)


def load_s(yaml):
    return Parser().parse(yaml.strip())


class Dumper(object):
    def dump(self, data):
        return '\n'.join(self._dump(data))

    def _dump(self, data, indent=0):
        for type_ in (list, dict, basestring, int, float):
            if isinstance(data, type_):
                f = getattr(self, '_dump_%s' % type_.__name__)
                return f(data, indent)
        if data in (True, False, None):
            return self._dump_literal(data, indent)
        raise NotImplementedError()

    def _dump_list(self, data, indent):
        output = []
        for item in data:
            dumped = self._dump(item, indent + 2)
            dumped[0] = '%s- %s' % (' ' * indent, dumped[0][indent + 2:])
            output += dumped
        return output

    def _dump_dict(self, data, indent):
        output = []
        for k, v in data.iteritems():
            output.append('%s%s:' % (' ' * indent, k))
            if isinstance(v, dict):
                output += self._dump(v, indent + 2)
            elif isinstance(v, list):
                output += self._dump(v, indent)
            else:
                value = self._dump(v)
                assert len(value) == 1
                output[-1] += ' %s' % value[0]
        return output

    def _dump_basestring(self, data, indent):
        if data in ('true', 'false', 'null'):
            data = "'%s'" % data
        return [' ' * indent + data]

    def _dump_int(self, data, indent):
        return ['%s%i' % (' ' * indent, data)]

    def _dump_float(self, data, indent):
        return ['%s%f' % (' ' * indent, data)]

    def _dump_literal(self, data, indent):
        string = {
            True: 'true',
            False: 'false',
            None: 'null',
        }[data]
        return [' ' * indent + string]


class Parser(object):
    _spaces_re = re.compile(r'^(\s*)(.*)')
    _list_re = re.compile(r'^(-\s+)(.*)')
    _dict_re = re.compile(r'^([^-:]+):\s?(.*)')

    def __init__(self):
        # Stack of (indent level, container object)
        self._stack = [(0, {})]
        # When a dict's value is a nested block, remember the key
        self._parent_key = None

    @property
    def _indent(self):
        return self._stack[-1][0]

    @property
    def _container(self):
        return self._stack[-1][1]

    @property
    def _in_list(self):
        return isinstance(self._container, list)

    @property
    def _in_dict(self):
        return isinstance(self._container, dict)

    def _push(self, container, indent=None):
        in_list = self._in_list
        assert in_list or self._parent_key

        if indent is None:
            indent = self._indent
        self._stack.append((indent, container()))
        if in_list:
            self._stack[-2][1].append(self._container)
        else:
            self._stack[-2][1][self._parent_key] = self._container
            self._parent_key = None

    def parse(self, yaml):
        for line in yaml.splitlines():
            spaces, line = self._spaces_re.match(line).groups()

            while len(spaces) < self._indent:
                self._stack.pop()

            lm = self._list_re.match(line)
            dm = self._dict_re.match(line)
            if len(spaces) == self._indent:
                if lm and self._in_dict:
                    # Starting a list in a dict
                    self._push(list)
                elif dm and self._in_list:
                    # Left an embedded list
                    self._stack.pop()

            if len(spaces) > self._indent:
                assert self._parent_key
                if dm:
                    # Nested dict
                    self._push(dict, len(spaces))
                elif lm:
                    # Over-indented list in a dict
                    self._push(list, len(spaces))

            indent = self._indent
            while lm and lm.group(2).startswith('- '):
                # Nested lists
                prefix, line = lm.groups()
                indent += len(prefix)
                self._push(list, indent)
                lm = self._list_re.match(line)
            del indent

            if lm:
                prefix, line = lm.groups()
                dm = self._dict_re.match(line)
                if dm:
                    self._push(dict, self._indent + len(prefix))
                else:
                    assert self._in_list
                    self._container.append(self._parse_value(line))

            if dm:
                key, value = dm.groups()
                assert self._in_dict
                if value:
                    self._container[key] = self._parse_value(value)
                else:
                    self._parent_key = key

        return self._stack[0][1]

    def _parse_value(self, value):
        if value.startswith("'") and value.endswith("'"):
            return value[1:-1]
        if value == 'true':
            return True
        if value == 'false':
            return False
        if value == 'null':
            return None
        for type_ in (int, float):
            try:
                return type_(value)
            except ValueError:
                pass
        return value