Import task fixes

[geekman]

I'm trying to migrate my existing Wordpress blog and have encountered a few issues which I'm addressing here.

• Raw HTML such as the <!--more--> comment and <table> tags were not passed through in the Pandoc conversion process into the Markdown format. This is fixed by passing --parse-raw to pandoc, although I'm unsure if there'll be any side effects for other formats
• The filter header is always written out to the Markdown front matter for all imported entries. This causes the markdown+codehilite+... filter specified in global conf.py to be ignored. I would think it makes sense to not specify the filter for each entry since you would likely be converting it into a common format used by all entries anyway.
• The draft header is not written out to the front matter, causing the draft status to be lost.
• There's a bug that causes an Exception when --keep-links is specified during import. In some cases, the Wordpress permalink may be http://blog.wordpress.com/?p=100 which causes a None to be assigned to the entry's permalink (since m.path == '/'). This in turn causes the Entry to not assign a permalink and hence retain permalink = None and dirname(entry.permalink)[1:] to throw an Exception. The correct fix in this case is to not assign entry['permalink'] at all and let the Entry create a permalink based on the safeslug from the entry title.

[posativ]

I'm not sure about the filter header... especially when you do not reconverse HTML (probably because html2text fails and pandoc is hard to install due to Haskell/GHC).

[posativ]

Merged with a note to include -a "filter: html" when importing without reconversion.

[geekman]

I'm converting the HTML in Wordpress to Markdown. The problem is, having filter: markdown in every post overrides the filter settings in conf.py, which has +codehilite+.... I was spending quite a bit of time debugging this issue, wondering why my code wasn't getting highlighted.