Better HTTP User-Agent strings (rot13 the library, say who we are)

[ypp-sc-tools.main.git] / yarrg / yppedia-ocean-scraper

diff --git a/yarrg/yppedia-ocean-scraper b/yarrg/yppedia-ocean-scraper
index 476c1cd9b381bb0d5d5ec1979cc4928bff9fa9f4..68efa38f6781980c63308a55b55b4334195873d5 100755 (executable)
--- a/yarrg/yppedia-ocean-scraper
+++ b/yarrg/yppedia-ocean-scraper
@@ -37,8 +37,8 @@ signal.signal(signal.SIGINT, signal.SIG_DFL)
 import sys
 import os
 import urllib
-import urllib2
 import re as regexp
+import subprocess
 from optparse import OptionParser
 from BeautifulSoup import BeautifulSoup
 
@@ -59,6 +59,20 @@ def fix_stdout():
 fix_stdout()
 
 
+# User agent:
+class YarrgURLopener(urllib.FancyURLopener):
+       base_version= urllib.URLopener().version
+       proc= subprocess.Popen(
+               ["./database-info-fetch", "useragentstringmap",
+                base_version, "manual islands/topology fetch"],
+               shell=False,
+               stderr=None,
+               stdout=subprocess.PIPE,
+               )
+       version = proc.communicate()[0].rstrip('\n');
+       assert(proc.returncode is not None and proc.returncode == 0)
+urllib._urlopener = YarrgURLopener()
+
 ocean = None
 soup = None
 opts = None
@@ -77,7 +91,7 @@ def fetch():
        url = ('http://yppedia.puzzlepirates.com/' +
                        (url_base % urllib.quote(ocean,'')))
        debug('fetching',url)
-       dataf = urllib2.urlopen(url)
+       dataf = urllib.urlopen(url)
        debug('fetched',dataf)
        soup = BeautifulSoup(dataf)