Wordcount reports extra words

[jtmasaki]

A document containing "this doesn't seem right" is reported as having 5 words, it looks like pyroom splits on an apostrophe.

Also, it looks like trailing space at the end of a document is counted as an additional word (this is rather minor, however).

Gedit looks like it has the same bug with the apostrophes, though until just now I couldn't figure out the whitespace bug.

It looks like the bug is due to your using gtk's text widget to advance words. I'm not sure where to report the bug upstream, however, and it might be worth working around.

I tried to modify the wordcount code to just use whitespace a while back, but it kind of just blew up in my face (I don't know quite enough python to be helpful, I am afraid).

[Nathan Haines (nhaines)]

This should definitely be reported upstream, and filing the bug against gedit is probably a good start.

I wanted to thank you for investigating this bug as much as you did--while patches are always nice, they are by no means necessarily the most important part of a bug report, and the clear explanation as well as the research you did is extremely valuable.

[jtmasaki (jtmasaki)]

Glad I could help.

I'll see if I can find the faulty function again (it's been a few months since I tried to come up with a patch) and try to report it upstream properly (a bug against the library, that is).

[Rafael Gattringer (rafael.gattringer)]

OpenOffice Writer 3.1 gets it right (4 words).

[Lionel Dricot (ploum-deactivatedaccount)]

This is a lot less trivial than you should expect. Indeed, it depends of the language.

In French, apostrophes are between words and, thus, you cannot have an uniform way of counting.

I'm also curious about Engilsh because : "I'm" is obviously two words. "doesn't" could be counted as two words too "does not".

[jtmasaki (jtmasaki)]

In English, to the best of my knowledge, both "I'm" and "doesn't" are counted as single words.

I am actually not aware of a situation where words are separated by something other than white-space. (Though, one could possibly make the case of hyphenated words counting as multiple words.)

I believe GTK's text widget changes behavior based on the language selected. So at least a fix to the English word breaking algorithm should be possible.

[Rafael Gattringer (rafael.gattringer)]

This bug still exists in Ubuntu 10.10 Maverick Meerkat Alpha 2 + updates: version 2.30.3-0ubuntu2 (64bit).