initial commit

.gitignore

@@ -0,0 +1 @@
+*.pyc

delicious_import.py

@@ -0,0 +1,51 @@
+#!/usr/bin/env python
+# encoding: utf-8
+"""
+delicious_import.py
+
+Created by Hilary Mason on 2010-11-28.
+Copyright (c) 2010 Hilary Mason. All rights reserved.
+"""
+
+import sys, os
+import urllib
+from xml.dom import minidom
+from optparse import OptionParser
+
+from gitmark import gitMark
+
+class delicious_import(object):
+    def __init__(self, username, password=''):
+		# API URL: https://user:[email protected]/v1/posts/all
+        url = "https://%s:%[email protected]/v1/posts/all" % (username, password)
+        h = urllib.urlopen(url)
+        content = h.read()
+        h.close()
+
+        x = minidom.parseString(content)
+
+        # sample post: <post href="http://www.pixelbeat.org/cmdline.html" hash="e3ac1d1e4403d077ee7e65f62a55c406" description="Linux Commands - A practical reference" tag="linux tutorial reference" time="2010-11-29T01:07:35Z" extended="" meta="c79362665abb0303d577b6b9aa341599" />
+        for post in x.getElementsByTagName("post"):
+            url = post.getAttribute('href')
+            desc = post.getAttribute('description')
+            tags = ",".join([t for t in post.getAttribute('tag').split()])
+            timestamp = post.getAttribute('time')
+
+            options = {}
+            options['tags'] = tags
+            options['push'] = False
+            options['msg'] = desc
+
+            args = [url]
+
+            g = gitMark(options, args)
+
+
+if __name__ == '__main__':
+    try:
+        (username, password) = sys.argv[1:]
+    except ValueError:
+        print "Usage: python delicious_import.py username password"
+
+    d = delicious_import(username, password)
+

gitmark.py

@@ -0,0 +1,105 @@
+#!/usr/bin/env python
+# encoding: utf-8
+"""
+gitmark.py
+
+Created by Hilary Mason on 2010-09-24.
+Copyright (c) 2010 Hilary Mason. All rights reserved.
+"""
+
+import sys, os
+import urllib
+import re
+import hashlib
+import csv
+import subprocess
+from optparse import OptionParser
+
+from settings import *
+
+class gitMark(object):
+
+    def __init__(self, options, args):
+        modified = [] # track files we need to add - a hack, because it will add files that are already tracked by git
+
+        url = args[0].strip('/')
+        content = self.getContent(url)
+        title = self.parseTitle(content)
+        content_filename = self.generateHash(url)
+
+        modified.append(self.saveContent(content_filename, content))
+        tags = ['all']
+        tags.extend(options['tags'].split(','))
+        for tag in tags:
+            t = tag.strip()
+            modified.append(self.saveTagData(t, url, title, content_filename))
+
+        self.gitAdd(modified)
+
+        commit_msg = options['msg']
+        if not commit_msg:
+            commit_msg = 'adding %s' % url
+
+        self.gitCommit(commit_msg)
+
+        if options['push']:
+            self.gitPush()
+
+    def gitAdd(self, files):
+        for f in files:
+            cmd = 'git add %s' % f
+            pipe = subprocess.Popen(cmd, shell=True)
+            pipe.wait()
+
+    def gitCommit(self, msg):
+        pipe = subprocess.Popen("git commit -m '%s'" % msg, shell=True)
+        pipe.wait()
+
+    def gitPush(self):
+        pipe = subprocess.Popen("git push origin master", shell=True)
+        pipe.wait()
+
+    def saveContent(self, filename, content):
+        f = open('%s%s' % (CONTENT_PATH, filename), 'w')
+        f.write(content)
+        f.close()
+        return '%s%s' % (CONTENT_PATH, filename)
+
+    def saveTagData(self, tag, url, title, content_filename):
+        tag_writer = csv.writer(open('%s%s' % (TAG_PATH, tag), 'a'))
+        tag_writer.writerow([url, title, content_filename])
+        return '%s%s' % (TAG_PATH, tag)
+
+    def parseTitle(self, content):
+        re_htmltitle = re.compile(".*<title>(.*)</title>.*")
+        t = re_htmltitle.search(content)
+        try:
+            title = t.group(1)
+        except AttributeError:
+            title = '[No Title]'
+
+        return title
+
+    def generateHash(self, text):
+        m = hashlib.md5()
+        m.update(text)
+        return m.hexdigest()
+
+    def getContent(self, url):
+        h = urllib.urlopen(url)
+        content = h.read()
+        h.close()
+        return content
+
+
+
+if __name__ == '__main__':
+    parser = OptionParser("usage: %prog [options]")
+    parser.add_option("-p", "--push", dest="push", action="store_false", default=True, help="don't push to origin.")
+    parser.add_option("-t", "--tags", dest="tags", action="store", default='notag', help="comma seperated list of tags")
+    parser.add_option("-m", "--message", dest="msg", action="store", default=None, help="specify a commit message (default is 'adding [url]')")
+    (options, args) = parser.parse_args()
+
+    opts = {'push': options.push, 'tags': options.tags, 'msg': options.msg}
+
+    g = gitMark(opts, args)

settings.py

@@ -0,0 +1,3 @@
+TAG_PATH = 'tags/'
+
+CONTENT_PATH = 'content/'