From d32b677652e0c6306daad2914b11ed853019863f Mon Sep 17 00:00:00 2001
From: Luke Lee <durden2.0@gmail.com>
Date: Mon, 27 Jun 2016 12:46:46 +0200
Subject: [PATCH] Fix bug with passing unicode data to urllib.pathname2url

- This can result in a KeyError for some unicode strings because pathname2url
  is meant to be passed bytes, not unicode.
- This fixes an exception in the logs we've seen related to requests to this
  URL in production.  Note this isn't a valid article URL however it shouldn't
  result in an exception, just an error message about not finding the guide.
    - '/microsoft-net/%E2%80%A6'
---
 pskb_website/remote.py | 24 ++++++++++++++++++++++--
 1 file changed, 22 insertions(+), 2 deletions(-)

diff --git a/pskb_website/remote.py b/pskb_website/remote.py
index 535344d..e9f8ba7 100644
--- a/pskb_website/remote.py
+++ b/pskb_website/remote.py
@@ -269,8 +269,18 @@ def read_file_from_github(path, branch=u'master', rendered_text=True,
         # would be wrong.  However, those URLs have been the same for years so
         # seems like a safe enough bet at this point.
         owner, repo, file_path = split_full_file_path(path)
-        url = u'https://github.com/%s/%s/blob/%s/%s' % (owner, repo, branch,
-                                                        urllib.pathname2url(file_path))
+
+        # Cannot pass unicode data to pathname2url or it can raise KeyError.
+        # Must only pass URL-safe bytes. So, something like u'\u2026' will
+        # raise a # KeyError but if we encode it to bytes, '%E2%80%A6', things
+        # work correctly.
+        # http://stackoverflow.com/questions/15115588/urllib-quote-throws-keyerror
+
+        url = u'https://github.com/%s/%s/blob/%s/%s' % (
+                owner,
+                repo,
+                branch,
+                urllib.pathname2url(file_path.encode('utf-8')))
 
         details = file_details(path, branch, None, None, url, text)
     else:
@@ -522,6 +532,16 @@ def contents_url_from_path(path):
     """
 
     owner, repo, file_path = split_full_file_path(path)
+
+    # Cannot pass unicode data to pathname2url or it can raise KeyError. Must
+    # only pass URL-safe bytes. So, something like u'\u2026' will raise a
+    # KeyError but if we encode it to bytes, '%E2%80%A6', things work
+    # correctly.
+    # http://stackoverflow.com/questions/15115588/urllib-quote-throws-keyerror
+    owner = owner.encode('utf-8')
+    repo = repo.encode('utf-8')
+    file_path = file_path.encode('utf-8')
+
     return urllib.pathname2url('repos/%s/%s/contents/%s' % (owner, repo,
                                                             file_path))