Importing Ebooks, regexp confusion

[ZaMotH]

Hi all,
I'm trying to import fields from formated files to import to calibre.

here is the regexp I'm trying to make work :
((?P<series>.+) # (?P<series_index>.+) - |)(?P<title>.+)( \[(?P<author>.+)\]|)

If it belong to a series, then my filename will start by :
- "**Serie name** # **Serie number ** - "

There might not be a serie name, so after this, comes the title :
- **title**

And afterwards, I add authors, but not always :
- "[**author**]"

Here is a sample of one of the titles I'm willing to parse :
Serie Name # 01 - Title name [author1 & Author2].epub

This Works : ((?P<series>.+) # (?P<series_index>.+) - |)(?P<title>.+)
This Works if there is a title : (?P<title>.+) \[(?P<author>.+)\]

But this : ((?P<series>.+) # (?P<series_index>.+) - |)(?P<title>.+)( \[(?P<author>.+)\]|)
Always gives up the following error : Attribute editor :'Nonetype' object has no attribute replace

Can anyone help ?

Regards

[ZaMotH]

I tired again and got some good things ... withou the author

Here are all my possible filename list :
1) Series # 00.zip
2) Series # 00 - Title.zip
3) Series # 00 - Title [author].zip
4) Series # 00 [Author].zip
5) Title.zip
6) Title [Author].zip

Without the author, I found out that this would work perfectly for 1, 2 and 5
(?P<series>.+) # (?P<series_index>\d+)( - (?P<title>.+)|)

For case 1, series and series_index are filled, AND title = Series # 00
For case 2, series, index and title are filled
For case 3 title is filled


But everytime I try to add the " [Author]", I get messed up all around.
I know the parsing should go like ( \[(?P<author>.+)\[)
But I can not make it conditional
added a "?" at the end, but after reading some other forums, I do not fully understant the "?" meaning
Tried the trick (( \[(?P<author>.+)\[)|), but that throws me the error "Attribute editor :'Nonetype' object has no attribute replace"

Can anyone help ?

[chaley]

@kovid: an AttributeError exception is not caught in the try block in ebooks/metadata/meta.py lines 149-165.

This used to work. Was there an update somewhere that changed referencing None to an AttributeError?

[kovidgoyal]

Originally Posted by chaley

@kovid: an AttributeError exception is not caught in the try block in ebooks/metadata/meta.py lines 149-165.

This used to work. Was there an update somewhere that changed referencing None to an AttributeError?

I'm going to need more context. Are you saying that match.group('authors') is raising an AttributeError?

[chaley]

Originally Posted by kovidgoyal

I'm going to need more context. Are you saying that match.group('authors') is raising an AttributeError?

No. It returns None, which is passed to authors_to_string, which raises an attribute error when it attempts to do None.replace(...)

Specifically, if (?P<author>... is in the pattern, and if that part of the pattern matches nothing, then match.group... returns None.

What I don't understand is "what changed?"

[kovidgoyal]

I dont recall anything changing that could affect this, but who knows...

I've committed a fix.

[chaley]

@ZaMotH:

Try

((?P<series>.+) # (?P<series_index>.+) -\s*)?(?P<title>[^\[]*)(\s*\[(?P<author>.+)\])?

If there is no author, this will still throw the exception until the next calibre release or until you update if you are running from source.

[ZaMotH]

Thank you for this quick answer, and quick fix.

Acutally, I'm using Calibre Portable ... I prefere portable applications. I will wait until the next release I guess.

Thanks to both of you.