from hachoir.parser import QueryParser
from hachoir.regex import PatternMatching


# XXX hachoir.regex uses str but the rest of hachoir uses bytes,
# which means we have to convert bytes to str using latin1 encoding
# (the closest "raw bytes" encoding) in order for matching to work.
class HachoirPatternMatching(PatternMatching):

    def __init__(self, categories=None, parser_ids=None):
        PatternMatching.__init__(self)

        # Load parser list
        tags = []
        if categories:
            tags += [("category", cat) for cat in categories]
        if parser_ids:
            tags += [("id", parser_id) for parser_id in parser_ids]
        if tags:
            tags += [None]
        parser_list = QueryParser(tags)

        # Create string patterns
        for parser in parser_list:
            for (magic, offset) in parser.getParserTags().get("magic", ()):
                self.addString(magic.decode('latin1'), (offset, parser))

        # Create regex patterns
        for parser in parser_list:
            for (regex, offset) in parser.getParserTags().get("magic_regex", ()):
                self.addRegex(regex.decode('latin1'), (offset, parser))
        self.commit()

    def search(self, data):
        for start, stop, item in PatternMatching.search(self, data.decode('latin1')):
            yield (item.user[1], start * 8 - item.user[0])