Decode retrieved files as UTF-8 even on Python 2

[weather.git] / weather.py

diff --git a/weather.py b/weather.py
index 78fd79c..fa94727 100644 (file)
--- a/weather.py
+++ b/weather.py
@@ -222,8 +222,7 @@ def get_uri(
         dcache_fd.close()
     else:
         try:
-            if pyversion("3"): data = urlopen(uri).read().decode("utf-8")
-            else: data = urlopen(uri).read()
+            data = urlopen(uri).read().decode("utf-8")
         except URLError:
             if ignore_fail: return ""
             else:
@@ -238,6 +237,9 @@ def get_uri(
                     )
                 sys.stderr.write(message)
                 sys.exit(1)
+        # Some data sources are HTML with the plain text wrapped in pre tags
+        if "<pre>" in data:
+            data = data[data.find("<pre>")+5:data.find("</pre>")]
         if cache_data:
             try:
                 import codecs