source: nagslang/yamlish.py@ 103:adf3cd83bf7a

'''
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, and strings.
'''

import re


def dump(data, file_object):
    file_object.write('\n'.join(_dump(data, 0)))


def _dump(obj, indent):
    if isinstance(obj, list):
        for item in obj:
            yield '%s- %s' % (' ' * indent,
                              '\n'.join(_dump(item, indent + 2)).strip())
    elif isinstance(obj, dict):
        for k, v in obj.iteritems():
            if isinstance(v, basestring):
                yield '%s%s: %s' % (' ' * indent, k, v)
            else:
                extra = 0 if isinstance(v, list) else 2
                yield '%s%s:\n%s' % (' ' * indent, k,
                                     '\n'.join(_dump(v, indent + extra)))
    elif isinstance(obj, basestring):
        yield '%s%s' % (' ' * indent, obj)


spaces_re = re.compile(r'^(\s*)(.*)')
list_re = re.compile(r'^(-\s+)(.*)')
dict_re = re.compile(r'^([^-:]+):\s?(.*)')


def load(file_object):
    # Stack of (indent level, container object)
    stack = [(0, {})]
    indent = lambda: stack[-1][0]
    obj = lambda: stack[-1][1]

    # When a dict's value is a nested block, remember the key
    parent_key = None

    for line in file_object:
        spaces, line = spaces_re.match(line).groups()

        while len(spaces) < indent():
            stack.pop()

        lm = list_re.match(line)
        dm = dict_re.match(line)
        if len(spaces) == indent():
            if lm:
                assert isinstance(obj(), list) or parent_key
                if isinstance(obj(), dict) and parent_key:
                    # Starting a list in a dict
                    stack.append((indent(), []))
                    stack[-2][1][parent_key] = obj()
                    parent_key = None
            elif dm and isinstance(obj(), list):
                # Left an embedded list
                stack.pop()

        if len(spaces) > indent():
            assert parent_key
            if dm:
                # Nested dict
                stack.append((len(spaces), {}))
            elif lm:
                # Over-indented list in a dict
                stack.append((len(spaces), []))
            stack[-2][1][parent_key] = obj()
            parent_key = None

        while lm and lm.group(2).startswith('- '):
            # Nested lists
            prefix, line = lm.groups()
            stack.append((indent() + len(prefix), []))
            stack[-2][1].append(obj())
            lm = list_re.match(line)

        if lm:
            prefix, line = lm.groups()
            dm = dict_re.match(line)
            if dm:
                stack.append((indent() + len(prefix), {}))
                stack[-2][1].append(obj())
            else:
                obj().append(line)

        if dm:
            key, value = dm.groups()
            if value:
                obj()[key] = value
            else:
                parent_key = key

    return stack[0][1]