tumblrnotesと協力してリブログツリーを出力するそれ: https://gist.github.com/u-e-d/5083450

rebloggraph.py

import tumblrnotes as tn
import igraph
import re
import sys
import os

def notes(url):
    def prittyfy(li):
        for r in li.findall("blockquote"):
            li.remove(r)
        return li.text_content().strip()
    
    posts = tn.notes(*tn.fromURL(url))
    ret = []
    for li in posts:
        ret.append(prittyfy(li))
    return ret

def reblogs(txt):
    note = re.findall("(\S+) reblogged this from (\S+)",txt)
    if len(note) != 1:
        return None
    else:
        return note[0]

def posted(txt):
    note = re.findall("(\S+) posted this", txt)
    if len(note) != 1:
        return None
    else:
        return note[0]

def seeds(elist):
    ret = set()
    for (l,r) in elist:
        ret.add(l)
        ret.add(r)
    return ret

def normalize(elist, seed=None):
    nodes  = seeds(elist)
    tables = dict()
    indexes= list()
    index  = 0
    ret    = list()
        
    if seed is not None:
        tables[seed] = index
        indexes.append(seed)
        index += 1
    
    for x, y in elist:
        if x not in tables:
            tables[x] = index
            indexes.append(x)
            index += 1
        if y not in tables:
            tables[y] = index
            indexes.append(y)
            index += 1
        ret.append((tables[x], tables[y]))
        
    return (indexes, ret)

def mkgraph(elist):
    names, nodes = normalize(elist)

    g = igraph.Graph(0)
    g.add_vertices(len(nodes)+1)
    g.vs["name"] = names
    g.vs["Label"] = names
    g.add_edges(nodes)
    return g

def main(url):
    note = notes(url)
    seed = posted(note[-1])
    rblg = filter(lambda x: x is not None, map(reblogs, note))
    g = mkgraph(rblg)
    
    g.write_graphml(open("%s.graphml"%seed, "w",1))

if __name__ == '__main__': 
    if len(sys.argv) != 2:
        print "useage: tublrnotes.py url"
    else:
        main(sys.argv[1])