def parse(self, input, partial=False): if not isinstance(input, str): input = ''.join(input) if not partial: input = fix_line_end(input) parser = zim.parser.Parser( Rule(LINK, url_re.r, process=self.parse_url) # FIXME need .r attribute because url_re is a Re object ) builder = ParseTreeBuilder(partial=partial) builder.start(FORMATTEDTEXT) parser(builder, input) builder.end(FORMATTEDTEXT) return builder.get_parsetree()
def parse(self, input, partial=False): if not isinstance(input, basestring): input = ''.join(input) if not partial: input = fix_line_end(input) parser = zim.parser.Parser( Rule(LINK, url_re.r, process=self.parse_url) # FIXME need .r atribute because url_re is a Re object ) builder = ParseTreeBuilder(partial=partial) builder.start(FORMATTEDTEXT) parser(builder, input) builder.end(FORMATTEDTEXT) return builder.get_parsetree()
def parse(self, input, partial=False, file_input=False):
if not isinstance(input, str):
input = ''.join(input)
input = input.replace(
'\u2029', ' '
) # Unicode PARAGRAPH SEPARATOR, causes conflict between "splitlines()" and regex "\n" matching - see issues #1760
if not partial:
input = fix_line_end(input)
meta, version = None, False
if file_input:
input, meta = parse_header_lines(input)
version = meta.get('Wiki-Format')
# Support backward compatibility - see history notes WIKI_FORMAT_VERSION
version = version or self.version
if version == 'zim 0.6':
mywikiparser = wikiparser
elif version in ('zim 0.4', 'zim 0.5'):
mywikiparser = WikiParser(backward_url_parsing=True)
else:
mywikiparser = WikiParser(backward_indented_blocks=True,
backward_url_parsing=True)
builder = ParseTreeBuilder(partial=partial)
mywikiparser(builder, input)
parsetree = builder.get_parsetree()
if meta is not None:
for k, v in list(meta.items()):
# Skip headers that are only interesting for the parser
#
# Also remove "Modification-Date" here because it causes conflicts
# when merging branches with version control, use mtime from filesystem
# If we see this header, remove it because it will not be updated.
if k not in ('Content-Type', 'Wiki-Format',
'Modification-Date'):
parsetree.meta[k] = v
return parsetree
