A few typos corrected, and generic formatting (PEP 8) #10
Conversation
utils module includes: safe_decode function (and factory make_safe_decoder); add_info/get_info, to store and access cumulated information; helpers for tarball download The WriterSection uses add_info to tell the transmogrifier about the ExportContext.
The ValueError raised by lxml.etree is not very helpful; in my case, the problem was a vertical tab (0x0b, one of string.whitespace). This change makes the inappropriate value visible, at least to observers of a foreground session.
if safe_decode is given, use it for *all* strings, including unicode; it might serve additional purposes, e.g. removal of vertical tabs etc. (which can cause lxml to crash)
This enables sections to specify a whitelist of properties they are interested in. By default, the Helper class skips the i18n_domain property. It is now enabled to use a whitelist *or* a blacklist of property names; unless given a whitelist, 'i18n_domain' will still be skipped. A method _skip_this is generated by the factory function make_skipfunc.
When created with verbose=True, the skip function will write a line to stdout whenever called.
... which was never meant for the master branch
Two changes: - Generally feed the xml text to the safe_decode function which will try utf-8 first, then latin-1. Another safe_decode function could be generated with other supported encodings, and it could be configurable, but this was sufficient to make my export succeed. - If this is not sufficient, and an ExpatError still occurs, try to be as helpful as possible printing information to stdout.
Enable the calling code to inform the user about the location of the written directory
| try: | ||
| doc = minidom.parseString(xml) | ||
| except ExpatError as e: | ||
| print '-->' * 26 |
There was a problem hiding this comment.
Don't use print. You wil only see it, if instance runs in foreground and it hurts if running doctests. See http://blog.redturtle.it/2010/02/15/do-not-use-print.-again-plone_log-is-your-friend
There was a problem hiding this comment.
Ok, print is not the ultimate solution, and it could be improved. It won't hurt doctests in this case because it is done for ExpatErrors only (unless there are doctests for such errors).
I had a very hard time to find out where the error occurred which made my export fail; and I re-raise the exception. The default output of ExpatErrors is not sufficiently helpful. Having useful information (which enabled me to tweak my export to make it work) in foreground sessions only is better than having it not at all.
|
Oops, right. Github apparently continuously updates my pull request and added commits which I didn't intend to include (this applies to My original pull request included just what the title says; perhaps we can start with the commits up to July 22nd. I'm pretty sure they don't break anything; IIRC, the travis tests failed before as well. |
|
@tobiasherp friendly reminder that this PR is still open :) |
This doesn't contain any functional changes; I corrected a few typos (
pathes->patches), and I did some generic format corrections.Before using
pep8ify -wn, I removed the trailing spaces (which would have been removed bypep8ifyas well) to make the process more transparent.Pep8ify has currently a bug (it inserts a space between a comma and the closing
)of a tuple, so I corrected this in another step.