1 files changed, 45 insertions, 0 deletions
diff --git a/grumpy/vdb/pypi.py b/grumpy/vdb/pypi.py
new file mode 100644
index 0000000..2e27059
--- /dev/null
+++ b/grumpy/vdb/pypi.py
@@ -0,0 +1,45 @@
+from lxml.html import fromstring
+import urllib
+
+class PyPi(object):
+    # TODO: Shall we use some kind of yaml or json based format instead?
+    # Or move this data into DB?
+    pkgs = {
+        # Package in Portage            :  Package in PyPi
+        'dev-python/flask'              : 'Flask',
+        'dev-python/flask-openid'       : 'Flask-OpenID',
+        'dev-python/flask-sqlalchemy'   : 'Flask-SQLAlchemy',
+        'dev-python/mako'               : 'Mako',
+        'dev-python/sphinx'             : 'Sphinx',
+        'dev-python/virtualenv'         : 'virtualenv',
+        'dev-python/vobject'            : 'vobject',
+        'dev-python/werkzeug'           : 'Werkzeug',
+        'dev-python/xlwt'               : 'xlwt',
+        'dev-python/yolk'               : 'yolk',
+        'dev-python/yolk-portage'       : 'yolk-portage',
+    }
+
+    # Url for fetching version information
+    url = 'http://pypi.python.org/pypi?:action=index'
+
+    def __init__(self):
+        pass
+
+    def fetch_and_parse_all(self):
+        """Download and parse package version information."""
+
+        items = {}
+        #f = urllib.urlopen(self.url)
+        f = open('utils/pypi.html')
+        if f:
+            data = fromstring(f.read()).cssselect('table.list')[0]
+            for row in data.getchildren():
+                val = row[0].getchildren()
+                if len(val) > 0:
+                    pkg = val[0].text.encode('utf-8').split('\xc2\xa0')
+                    if len(pkg) == 2:
+                        if pkg[0] not in items.keys():
+                            items[pkg[0]] = []
+                        items[pkg[0]].append(pkg[1])
+        f.close()
+        return items