Merge lp:~jameinel/hydrazine/bug_612641_encoding into lp:hydrazine

[John A Meinel]

Encode to ascii if we can't print the comment normally.

An attempt to avoid encoding failures crashing feed-pqm completely.

Ideally we'd do something based on sys.stdout.encoding or whatever. But since we are just doing "print", I just did the easy hack.

[Martin Packman (gz)]

I think this if fine as a stop-script-throwing-so-I-can-work hack. As a proper fix, how about using the bzr ui object rather than print as I had to init it anyway to get password promting working earlier?

[Martin Pool (mbp)]

On 4 October 2010 19:39, Martin [gz] <email address hidden> wrote:
> I think this if fine as a stop-script-throwing-so-I-can-work hack. As a proper fix, how about using the bzr ui object rather than print as I had to init it anyway to get password promting working earlier?

mm, it would probably be better to write everything to a stream with
(get_terminal_encoding(), 'replace').

--
Martin

[Martin Packman (gz)]

I was thinking the ui object would do that step for us, like here:

    def prompt(self, prompt, **kwargs):
        """Emit prompt on the CLI.

        :param kwargs: Dictionary of arguments to insert into the prompt,
            to allow UIs to reformat the prompt.
        """
        if kwargs:
            # See <https://launchpad.net/bugs/365891>
            prompt = prompt % kwargs
        prompt = prompt.encode(osutils.get_terminal_encoding(), 'replace')
        self.clear_term()
        self.stderr.write(prompt)

But it seems most of the methods aren't actually that careful with unicode vs. str and could do with sorting out.

=== modified file 'feed-pqm'
--- feed-pqm	2010-06-28 19:13:18 +0000
+++ feed-pqm	2010-10-01 21:08:46 +0000
@@ -102,7 +102,12 @@
     print "Recent comments:"
     comments = list(mp.all_comments) #XXX: See bug lp:583761
     for comment in comments[-1:]:
-        print "%s: %s" % (comment.author.name, comment.message_body)
+        val = "%s: %s" % (comment.author.name, comment.message_body)
+        try:
+            print val
+        except UnicodeError:
+            val = val.encode('ascii', 'replace')
+            print val
 
 
 def set_message(mp):