IMSLP Forums

In trying to parse the wiki code embedded in a page, I'm trying to understand and to generalize the various coding options that I see.

In http://imslp.org/wiki/Symphony_No.9,_Op ... udwig_van), The FILES section begins with the tag However, Hindemith's works have this:
.

Are there any other such imslp tags that I should be aware of?

As I'm really in learning mode for MediaWiki, I'm not sure if these tags are links to template files, or what....

The latter template is for files hosted on other servers, ex. the US server (since Hindemith is not public domain in either Canada or the EU).

The Externalimslpfile template is now deprecated for uploads to IMSLP·US, there is a new fte template

#fte:server-us

specially for these.

The Hindemith works should gradually be transitioned to using the new template, which is less fiddly to use...

Cheers, Philip

There are only three FTE template types for file entries:

#fte:imslpfile
#fte:imslpaudio, and
#fte:server-us

As PML said, the Externalimslpfile template is deprecated. Its variable names are very strange, so I would advise against trying to parse it.

I think there are two things we can do for the iPad app:
1. ignore fte:server-us (these files are files PD only in the US), or
2. do a mass convert of Externalimslpfile into server-us and implement server-us; conversion maybe via a bot (the conversion is actually very simple, see below). Note that because the US server is not associated with the IMSLP site proper, US server files are not catalogued in the IMSLP index system. The download URL is also different.

For some background on what happened with server-us/externalimslpfile, and how to convert them (note that the bot must be logged in and send the correct login cookies):
viewtopic.php?f=7&t=4716&p=24425#p24455

Note that, while most of the variable names are the same, server-us has |File Path= and |File Size= instead of |File Name=, and |Date Submitted= format is different.

imslp wrote:There are only three FTE template types for file entries:

#fte:imslpfile
#fte:imslpaudio, and
#fte:server-us

OK, so just to be clear, I should write 2 parsers (or have 2 sets of variables in one parser, whichever is easier). One for #fte:imslpfile and one for #fte:server-us? (for PDF files, audio files will come later)

Well, most of the variables are the same between the two except for the ones I mentioned above. I'd just put them in one parser, but that's your call of course.

I'm noticing a number of (what I suppose to be) incorrect formatting. Here's an example:
Note the 2 "||"s before File Name 1. This only occurs for File Name 1, not File Name 2, etc...

Is this a new format, or is this wrong?

As far as I know, this was the result of some kind of glitch in the file uploader that occurred from late 2008 to early 2009. There are probably a few thousand pages (at most, I'd guess) that still have this extra pipe character. It is not new, and can be ignored wherever it appears (it doesn't affect the displaying of the template at all).

KGill wrote: It is not new, and can be ignored wherever it appears (it doesn't affect the displaying of the template at all).

It's good to know that it's not new. Unfortunately, it's screwing up my importer!

I will log those entries and fix them as the logfile reports them...

The best regex for matching the entries is probably (with // as the delimiter):

/\|\s*File Name\s*=\s*([^|]*)/

After which you strip the whitespace from the end of the captured string. This would be the most natural way of matching MediaWiki template variables using a single regex, however, it does not take into account nested templates and links, which can sometimes occur and include a pipe sign, breaking the above regex.

If you want to make a bullet-proof MediaWiki template parser, I would suggest looking at the Ruby-based parser in the Ruby bot I sent you a while back. That is an almost completely bulletproof parser.

(Of course, these are all just suggestions.)

Great, thanks!

I will look into the parser again, and see about porting it to Objective-C.