def __init__(self,
path_or_stream,
dest_dir=None,
log=None,
detect_cover=True,
notes_text=None,
notes_nopb=False,
nosupsub=False):
self.docx = DOCX(path_or_stream, log=log)
self.namespace = self.docx.namespace
self.ms_pat = re.compile(r'\s{2,}')
self.ws_pat = re.compile(r'[\n\r\t]')
self.log = self.docx.log
self.detect_cover = detect_cover
self.notes_text = notes_text or _('Notes')
self.notes_nopb = notes_nopb
self.nosupsub = nosupsub
self.dest_dir = dest_dir or os.getcwdu()
self.mi = self.docx.metadata
self.body = BODY()
self.theme = Theme(self.namespace)
self.settings = Settings(self.namespace)
self.tables = Tables(self.namespace)
self.fields = Fields(self.namespace)
self.styles = Styles(self.namespace, self.tables)
self.images = Images(self.namespace, self.log)
self.object_map = OrderedDict()
self.html = HTML(
HEAD(
META(charset='utf-8'),
TITLE(self.mi.title or _('Unknown')),
LINK(rel='stylesheet', type='text/css', href='docx.css'),
), self.body)
self.html.text = '\n\t'
self.html[0].text = '\n\t\t'
self.html[0].tail = '\n'
for child in self.html[0]:
child.tail = '\n\t\t'
self.html[0][-1].tail = '\n\t'
self.html[1].text = self.html[1].tail = '\n'
lang = html_lang(self.mi.language)
if lang:
self.html.set('lang', lang)
self.doc_lang = lang
else:
self.doc_lang = None
