It looks like Python 2.3 will have some pretty cool stuff.

This post on comp.lang.python describes how to use the dbiRaw object you get back when working with binary (BLOB or IMAGE, depending on your back end) fields. And, for some reason, I seem to be getting back two bytes for every byte inserted. This is annoying.

I wrote a simple python module for handling MS Word files. If you've been following along, you know that I've been wrestling with python and Word a bit recently. It turns out that it's easy enough to do, you just have to put together the right information. (Note: requires win32 extensions, Word, and FPropSet.)

Based on some code from this article, here's how to extract the "Title" property from a Word document in python.

  import win32com.client
  app = win32com.client.Dispatch('Word.Application')
  doc = app.Documents.Add(self.filename)
  self.title = doc.BuiltinDocumentProperties("Title")
  doc.Close(0)
  app.Quit()

Hmm. There has got to be a faster way of getting that info. I'm pretty sure that Explorer isn't launching Word every time it wants to display the properties for a document.

There is: FPropSet. This is about two magnitudes of order faster than the previous solution.

    fps = win32com.client.Dispatch('FilePropertySet.FilePropertySet')
    fps.Pathname = file
    print fps.BuiltInFileProperties().Item("Title")

After spending an hour trying to figure out what misunderstanding of COM was causing my problem, I realized that the file I was trying to test against DID NOT EXIST (typo). I was calling everything correctly, but the information I was sending was wrong... note that setting fps.Pathname to a non-existent file will cause an exception.

This is a python script that basically implements a crude sort of upstreaming (but it's not scanning constantly, only when you run the script). I wrote this because I got tired of fighting with other software that didn't quite integrate the way I wanted it to. This is a small piece in a set of loosely joined tools that I use to manage my website(s). The script is available under an MIT style license.

This is pretty simple. All I did was to convert the MS KB article from Visual Basic to Python. Note that this won't actually create the .mdb file. That's a little trickier -- I can't find a Python binding for the DLL call and I don't have the need for it at the moment. See init_dsn.py for details.

Check out this code sample for a quick and dirty system-tray demo using wxPython. (Lifted, rearranged, and simplified from the wxPython demo.)

Note that you'll need an icon file for this to work, or you'll have to add back the images.py code from the demo.

Note: this assumes you have python, win32all, and MS Word all installed properly.

Running Word from python is incredibly easy.

import win32com.client
# Start the application.
app = win32com.client.Dispatch('Word.Application')
# Open a document.
doc = app.Documents.Add('c:\path\to\document\test.doc')
# Or:
app.ChangeFileOpenDirectory('c:\path\to\document\')
doc = app.Documents.Add('test.doc')
# To save as HTML (filtered; i.e. without most of the MS cruft)
# Note: I don't know if the type-number is portable??
doc.SaveAs('test.html', 10) ## 10 == HTML-Filtered
doc.Close(0) ## 0 == don't save changes?
app.Quit()

This will all run "invisibly". If you want to display the app while doing all of this, set app.Visible = 1 somewhere near the top. If you simply want to fetch the text from the document, use doc.Content.Text. It will return a Unicode string and it's worth noting that those "smart quotes" have no representation in ASCII so you might have to be careful decoding it.