Markdown Metadata

[Dalcyks]

I am trying to use Markdown-tagged TXT files (.MD) as input to Calibre. These I want to convert, ultimately to Kindle PW3, but I'll settle for ePub. The problem I'm having is that the metadata in the .MD doesn't seem to be reading into Calibre properly.

My desired outcome is to (1) populate the Calibre metadata fields with the values in the .MD, and (2) be able to include metadata values in the converted ePub. Neither of which seem to be working.

Here is my test-flow...

I created a TEST.MD file containing the following 9 lines (copied from the Metadata Extension doc):

Title: My Document
Summary: A brief description of my document.
Authors: Waylan Limberg
John Doe
Date: October 2, 2007
blank-value:
base_url: http://example.com
<blank line>
This is the first paragraph of the document.

Calibre Preference settings:

[Preferences - Input Options - TXT Input -]
Paragraph Style: auto
Formatting Style: markdown
Markdown - Meta: checked
all other options: unchecked
[Preferences - Common Options - Structure Detection -]
Insert Metadata as page at start of book: checked
[Preferences - Adding Books - The Add Process -]
Read Metadata from File Contents rather than file name: checked

Procedure:

1) restarted Calibre (to be sure settings "took")
2) 'Calibre - Add Books' for the TEST.MD file
3) I note: at this point, Calibre metadata fields contain all default values. They were not automatically changed by the 'Add Book' (would have been nice).
4) In Calibre, I set Author, Tags, and Publisher metadata values to the word 'test'.
5) 'Calibre - Edit Metadata' (clicking the MD format, then clicking the 'Set metadata for the book from the selected format' icon).
6) Calibre metadata values do not change to those in TEST.MD.
7) 'Calibre - Convert Books' (MD to EPUB).
8) I note: The in-file metadata lines are NOT included in the ePub as text/paragraphs (this implies that they were actually recognized as something special).
9) I note: NO 'Insert Metadata as page at start of book' page was created in the ePub (this should have worked regardless of the metadata values).
10) 'Calibre - Convert Books' (MD to MOBI).
11) Same results as the ePub, except a final blank page was included (as viewed with the Calibre viewer).

Any idea what I might be missing?

Calibre 2.70[64 bit] / Win7

[kovidgoyal]

calibre has no support for random markdown extensions. The help for the markdown extensions option shows you the list of extensions calibre supports.

[Dalcyks]

Originally Posted by kovidgoyal

calibre has no support for random markdown extensions. The help for the markdown extensions option shows you the list of extensions calibre supports.

Thank you for your response!

I see the the "Meta" extension listed in both the help link you mentioned and in the 'checkable' option box above it. It was from the help text for that extension that I copy-n-pasted the test file contents I used.

Are you saying that even if it's in that list, it may be considered a 'random extension' and thus actually unsupported?

[kovidgoyal]

Ah, I had no idea that extension was included in the markdown standard library.

I have no idea what that extension does, in fact, I dont see how it could possibly work, since all markdown extensions do is change the html generated from the input txt and html has no standard for metadata. Indeed, reading https://pythonhosted.org/Markdown/ex...meta_data.html the page says it does not do anything.

So if you want support for metadata in markdown, you'd have to modify the calibre txt input plugin to make use of the metadata read by that extension.

[kovidgoyal]

Another approach is to get that extension to output dublin core metadata into the generated html, which will be read by calibre.

[Dalcyks]

OK, that now makes sense! I was wondering why it did remove those lines from the conversion input stream, but it not affect the books metadata values. It is the input plugin being unaware of the extension. Thank you.

[kovidgoyal]

This should do the trick for the basic metadata fields:

https://github.com/kovidgoyal/calibr...1937b71a2be8c9

[Dalcyks]

Kovid, Thank you very much for that. I look forward to the next build! With this being the first time I've actually looked at Python code, I have a couple of curiosities:

1) It appears (new line 141 in processor.py) that the 5-star rating is kept internally as a value 0 to 10. Is that true?
2) Will this distinguish between the "date" and "pubdate" as separate metadata fields or are both labels synonymous with "pubdate" (lines 127,130 in processor.py).

Thanks again for the feedback and the fixes.

[kovidgoyal]

1) yes
2) Both are pubdate. The date field in calibre is timestamp