"""
This module stores the deler configuration, and specified in the
xml configuration files.  See examples/html-wdc.xml for an example
with documenting comments.
"""

from html import HtmlAccountedString, HtmlReduction
import os
from xml.etree import ElementTree


class Config(object):

    def __init__(self, config, accounting, gml_mode, paragraph_mode, segmenter):
        super(Config, self).__init__()
        if os.path.exists(config):
            self.root = ElementTree.parse(config).getroot()

            # JR - removed for simplicity until other input possible
            # get the input format (only html in this version)
            # input_format = self.root.attrib['input']
            # if input_format is None or input_format == 'html':
            #     self.reduction_type = HtmlReduction
            #     self.account_type = HtmlAccountedString
            # else:
            #     raise ValueError('unrecognised input format ' + input_format)

            self.reduction_type = HtmlReduction
            self.account_type = HtmlAccountedString

            # check segmenter path references an executable
            if not os.path.exists(segmenter):
                raise ValueError(segmenter + ' does not exist')
            elif not os.access(segmenter, os.X_OK):
                raise ValueError(segmenter + ' is not executable')
            self.segmenter = segmenter

            self.accounting = accounting
            self.gml_mode = gml_mode
            self.paragraph_mode = paragraph_mode

            self.purge = _ElementSet(self.root, 'purge')
            self.map = _MapDict(self.root, 'map')
            self.mask = _ElementSet(self.root, 'mask')
            self.newline = _ElementSet(self.root, 'newline')
            self.non_closing = _ElementSet(self.root, 'non-closing')
            self.paragraph_like = _ElementSet(self.root, 'paragraph-like')
            self.unsegmentable = _ElementSet(self.root, 'unsegmentable')

            if self.gml_mode:
                for name, gml in self.map.items():
                    if gml is None:
                        raise ValueError('gml-mode activated but not all map elements have a gml equivalent')

        else:
            raise ValueError(config + ' does not exist')


class _ElementSet(set):
    """ A set of child elements with the specified name """
    def __init__(self, root, name):
        super(_ElementSet, self).__init__()
        for element in root.findall(name):
            for child in element.findall('element'):
                self.add(child.attrib['name'])

class _MapDict(dict):
    """ A dictionary of maps between element names and gml """
    def __init__(self, root, name):
        super(_MapDict, self).__init__()
        for element in root.findall(name):
            for child in element.findall('element'):
                name = child.attrib['name']
                gml = child.attrib['gml'] if 'gml' in child.attrib else None
                self[name] = gml