pywb/utils.py

import re

LINK_SPLIT = re.compile(',\s*(?=[<])')
LINK_SEG_SPLIT = re.compile(';\s*')
LINK_URL = re.compile('<(.*)>')
LINK_PROP = re.compile('([\w]+)="([^"]+)')


#=================================================================
class MementoUtils(object):
    @staticmethod
    def parse_links(link_header, def_name='timemap'):
        links = LINK_SPLIT.split(link_header)
        results = {}
        mementos = []

        for link in links:
            props = LINK_SEG_SPLIT.split(link)
            m = LINK_URL.match(props[0])
            if not m:
                raise Exception('Invalid Link Url: ' + props[0])

            result = dict(url=m.group(1))
            key = ''
            is_mem = False

            for prop in props[1:]:
                m = LINK_PROP.match(prop)
                if not m:
                    raise Exception('Invalid prop ' + prop)

                name = m.group(1)
                value = m.group(2)

                if name == 'rel':
                    if 'memento' in value:
                        is_mem = True
                        result[name] = value
                    elif value == 'self':
                        key = def_name
                    else:
                        key = value
                else:
                    result[name] = value

            if key:
                results[key] = result
            elif is_mem:
                mementos.append(result)

        results['mementos'] = mementos
        return results