add jbovlaste xml export download script

The readme is also amended to use this script rather than wget, since
the introduction of the captcha on jbovlaste has broken the old way.

Closes #4

download.py

@@ -0,0 +1,28 @@
+#!/usr/bin/env python
+
+import requests
+
+CREDENTIALS = dict(
+        username='YOUR JBOVLASTE USERNAME',
+        password='YOUR JBOVLASTE PASSWORD',
+)
+
+# jbovlaste definitions language to download
+LANG = 'en'
+
+if __name__ == '__main__':
+    s = requests.Session()
+    r = s.post(
+            'http://jbovlaste.lojban.org/login.html',
+            data=CREDENTIALS,
+    )
+    r.raise_for_status()
+
+    r = s.get(
+            'http://jbovlaste.lojban.org/export/xml-export.html?lang=' + LANG,
+    )
+
+    r.raise_for_status()
+
+    with open('vlasisku/data/jbovlaste.xml', 'wb') as f:
+        f.write(r.text.encode('utf-8'))

readme.yml

@@ -13,15 +13,17 @@ Recommended for developers:
 Installation for developers: |
     git clone https://github.com/lojban/vlasisku
     cd vlasisku
-    wget 'http://jbovlaste.lojban.org/export/xml-export.html?lang=en' \
-         -O vlasisku/data/jbovlaste.xml
 
     # Isolated Python environment
     mkvirtualenv --no-site-packages vlasisku
 
     # Installs to that environment
     pip install -r requirements.txt
 
+    # Modify download.py to use your jbovlaste credentials, then use it
+    # to download a jbovlaste XML dump. (It takes a while to run.)
+    ./download.py
+
     # This takes some 20 seconds the first time
     # and must be done whenever the jbovlaste export is changed
     ./manage.py runserver

requirements.txt

@@ -6,3 +6,4 @@ simplejson
 pystemmer
 Twisted
 jellyfish
+requests