from config import config
import model 
import os
from comment_tools import safeparsename
import urlparse
import md5

def shorten(uri):
        # Break apart the URI according to RFC 3986
    parts = urlparse.urlsplit(uri)
    # Create a list of the host name, the components of the path, and themd5 hash of the whole URI
    simple_name = [parts.scheme, parts.hostname]
    simple_name.extend(parts.path.split("/")[1:])
    # Return the list concatenated with dashes.
    return "-".join(simple_name)

outdir = config.get('dir', 'commentsbyauthor')
comments = config.get('dir', 'comments')

# Dictionary mapping commenter URI to a set of alias's they've used
# Also a list of comments as (entry, commentid) tuples
by_uri = {}
for entry in os.listdir(comments): 
    if os.path.isdir(os.path.join(comments, entry)):
        coll = model.comments(entry)
        if coll:
            for id in coll.id_list():
                member = coll.get(id)
                author = safeparsename(member.get('author', ''))
                if author['uri']:
                    uri = author['uri']
                    if uri not in by_uri:
                        by_uri[uri] = {"names": set(), "comments": []}
                    by_uri[uri]['names'].add(author['name'])
                    by_uri[uri]['comments'].append((entry, id))

for k, v in by_uri.iteritems():
    try:
        print shorten(k), k 
    except:
        pass