UnicodeError running arch

[Kf6spf]

I just deployed the latest release (2.1.7 stable) and
decided to rebuild an archive. I'm getting the
following error trace caused by it hitting high-order
byte characters. These were not a problem when I first
built the archive from 2.1.6 so something may have changed.

Pickling archive state into
/usr/local/mailman/archives/private/basfa/pipermail.pck
Traceback (most recent call last):
  File "bin/arch", line 200, in ?
    main()
  File "bin/arch", line 188, in main
    archiver.processUnixMailbox(fp, start, end)
  File
"/usr/local/mailman/Mailman/Archiver/pipermail.py",
line 585, in processUnixMailbox
    self.add_article(a)
  File
"/usr/local/mailman/Mailman/Archiver/pipermail.py",
line 626, in add_article
    filename))
  File
"/usr/local/mailman/Mailman/Archiver/HyperArch.py",
line 1116, in write_article
    f.write(article.as_text())
  File
"/usr/local/mailman/Mailman/Archiver/HyperArch.py",
line 579, in as_text
    '\g<1>' + _(' at ') + '\g<2>', body)
  File "/usr/lib/python2.2/sre.py", line 143, in sub
    return _compile(pattern, 0).sub(repl, string, count)
UnicodeError: ASCII decoding error: ordinal not in
range(128)

[http://sourceforge.net/tracker/index.php?func=detail&aid=1395683&group_id=103&atid=100103]

[Mark Sapiro (msapiro)]

I also observed this problem. It will be properly fixed. In
the mean time, I have attached a patch - HyperArch.patch -
which will allow you to rebuild the full archive. The patch
is not a real fix for the problem however, as the problem
messages will not have email addresses obscured in the body
of the message in the monthly .txt files. The HTML archive
will be OK.

[Mark Sapiro (msapiro)]

I've developed a new patch which I've attached here in place
of the previous one. I think the new patch actually fixes
the problem. Please try it and report.

[Mark Sapiro (msapiro)]

The previous patch wasn't effective in all circumstances. I
have replaced it again.

[Mark Sapiro (msapiro)]

One other followup note on this. The problem can also occur
when archiving an individual post to a list. In this case,
the HTML file for the post will be written but not indexed;
the post will not be added to the .txt file; the error will
be logged in the 'error' log, and the message will be shunted.

[Tokio Kikuchi (tkikuchi)]

Here is a revised patch to incorporate non-ascii '_at_'
replacement.

[Mark Sapiro (msapiro)]

I have deleted my patch # 161995 to avoid confusion as patch
# 162200 is the correct one.

[Tokio Kikuchi (tkikuchi)]

Hi, I've just committed this patch in the CVS.

I'll be releasing 2.1.8 but, in the meantime, please use
this patch.

Sorry for the inconvenience.

[Tokio Kikuchi (tkikuchi)]

Re-open for visitors can find this.

[Tokio Kikuchi (tkikuchi)]

Closing now bcause this is included in a new patch.