X-Git-Url: https://projects.mako.cc/source/harrypotter-wikipedia-cdsw/blobdiff_plain/7a786062b74eb0335b21b1bd2ae860ea349748f8..2e473702cd932f0a9f9c05545134c372089ef30b:/build_hpwp_dataset.py

diff --git a/build_hpwp_dataset.py b/build_hpwp_dataset.py
index 38c299d..9d381ae 100644
--- a/build_hpwp_dataset.py
+++ b/build_hpwp_dataset.py
@@ -1,6 +1,7 @@
 #!/usr/bin/env python
 # coding=utf-8
 
+import encoding_fix
 import requests
 
 # get_article_revisions is a function that takes an article title in
@@ -12,7 +13,7 @@ def get_article_revisions(title):
     # create a base url for the api and then a normal url which is initially
     # just a copy of it
     # The following line is what the requests call is doing, basically.
-    # "http://en.wikipedia.org/w/api.php/?action=query&titles={0}&prop=revisions&rvprop=flags|timestamp|user|size|ids&rvlimit=500&format=json".format(title)
+    # "http://en.wikipedia.org/w/api.php/?action=query&titles={0}&prop=revisions&rvprop=flags|timestamp|user|size|ids&rvlimit=500&format=json&continue=".format(title)
     wp_api_url = "http://en.wikipedia.org/w/api.php/"
 
     parameters = {'action' : 'query',
@@ -39,6 +40,7 @@ def get_article_revisions(title):
 
             # for every revision, first we do some cleaning up
             for rev in query_revisions:
+                #print(rev)
                 # let's continue/skip this revision if the user is hidden
                 if "userhidden" in rev:
                     continue
@@ -84,16 +86,18 @@ category = "Harry Potter"
 #
 # The following requests call basically does the same thing as this string:
 # "http://tools.wmflabs.org/catscan2/catscan2.php?depth=10&categories={0}&doit=1&format=json".format(category)
-url_catscan = "http://tools.wmflabs.org/catscan2/catscan2.php"
+url_catscan = "http://tools.wmflabs.org/catscan3/catscan2.php"
 
 parameters = {'depth' : 10,
               'categories' : category,
               'format' : 'json',
               'doit' : 1}
 
+# r = requests.get("http://tools.wmflabs.org/catscan2/catscan2.php?depth=10&categories=Harry Potter&doit=1&format=json"
+
 r = requests.get(url_catscan, params=parameters)
 articles_json = r.json()
-articles = articles_json["*"][0]["a"]["*"]
+articles = articles_json["*"][0]["*"]
 
 # open a file to write all the output
 output = open("hp_wiki.tsv", "w", encoding="utf-8")
@@ -104,6 +108,7 @@ for article in articles:
 
     # first grab the article's title
     title = article["a"]["title"]
+    print(title)
 
     # get the list of revisions from our function and then iterate through it,
     # printing it to our output file