Sorry Francois, I should have tested this first before sending it to you. I got the conditional construction wrong (don't know what I was thinking)---now you get a nice error message if there is no pdf in the directory. Best, Mark

On 25 Nov 2006, at 11:32, Mark Eli Kalderon wrote:

You need to generate a pdf to use the enclosed command. Hope this helps, Mark

<Word Count.tmCommand>

On 25 Nov 2006, at 11:10, Francois wrote:

...

Hello All,

I am writing a document un LaTeX and I would like to make a word count. I search online manuals but did not find any word count function... does-it exist ?

Thanks

Francois

---

_ For new threads USE THIS: textmate@lists.macromates.com (threading gets destroyed and the universe will collapse if you don't) http://lists.macromates.com/mailman/listinfo/textmate

---

For new threads USE THIS: textmate@lists.macromates.com (threading gets destroyed and the universe will collapse if you don't) http://lists.macromates.com/mailman/listinfo/textmate

On Nov 29, 2006, at 6:29 PM, Michael Sheets wrote:

...

I am writing a document un LaTeX and I would like to make a word count. I search online manuals but did not find any word count function... does-it exist ?

Text Bundle -> Statistics for Selection (Word Count)

Next time your looking for a feature open up the bundle items window and type it in, you might just be surprised ;) (⌃⌘T)

I assume he only wants to count *typeset* words, not latex commands.

Probably the easiest thing to do is make a command that runs detex on the document (which strips the commands out) and pipe that through the word count.

On 11/29/06, Mark Eli Kalderon eli@markelikalderon.com wrote:

On 29 Nov 2006, at 23:52, Kevin Ballard wrote:

...

On Nov 29, 2006, at 6:29 PM, Michael Sheets wrote:

...

...

I am writing a document un LaTeX and I would like to make a word count. I search online manuals but did not find any word count function... does-it exist ?

Text Bundle -> Statistics for Selection (Word Count)

Charilaos Skiadas wrote:

On Nov 29, 2006, at 7:33 PM, Tim Lahey wrote:

...

Probably the easiest thing to do is make a command that runs detex on the document (which strips the commands out) and pipe that through the word count.

That assumes one is not using commands that automatically generate text. I.e. what happens if I do:

\newcommand{\this}{Hi there! }

and then the document contains:

\this\this\this

I would guess detex will not count those 6 words, will it?

So really the only thing we can do is actually typeset the document? This seems problematical to me, especially if we want general solutions that we can use with languages other than TeX.

On 30 Nov 2006, at 01:08, Tim Lahey wrote:

No,

I just checked and you're right. Plus, some things are still there that one wouldn't want. I did however find,

http://catdvi.sourceforge.net/

which is a DVI to plaintext translator. I don't know how well it works though.

Cheers,

Tim.

Well since most users are generating PDFs, it makes sense to use ps2ascii, instead.

Best, Mark

On 30 Nov 2006, at 01:12, Paul McCann wrote:

Haris wrote...

...

\newcommand{\this}{Hi there! } and then the document contains: \this\this\this

...

I would guess detex will not count those 6 words, will it?

No: but the alternative (typesetting and counting using something like ps2ascii) is unworkably slow on large documents, so it's going to depend on what sort of accuracy you're seeking (and what each method considers to be a "word"). It'd be nice to find something fast-ish and not-too-inaccurate, but I'm not holding my breath!

There is a limit to how fast an accurate utility could be if it is to accommodate the kind of procedural aspects of LaTeX syntax that Haris was referring to. Best, Mark

Mark Eli Kalderon wrote:

There is a limit to how fast an accurate utility could be if it is to accommodate the kind of procedural aspects of LaTeX syntax that Haris was referring to.

Indeed: but all that work has already been done, as the command is operating on the pdf file. So the real question becomes: why is ps2ascii (aka ghostscript) so slow? (Just checked on my work machine, a 2GHz intel iMac with 2G of memory, and it's still about 30 seconds on a 250 page, 1.2MB pdf file.)

I guess it's just an irreducibly difficult procedure... Moral of this story? Don't count words very often!

Cheers, Paul

Hi again,

If you look at the pdf/ps source file, it is filled with special commands and things. I suppose if you could export the pdf file to a txt file, then you could count the words there with ease.

Believe me, I've had a good look inside pdf files. Not somewhere I'd like to live! Re exporting: nothing obvious rears its head up here, and Preview only allows you to export to graphical formats from pdf.

You are better off with the small error from counting words in the latex source instead.

Agreed: that's definitely what I would do if faced with such a task, so a simple command with . Thankfully I've rarely encountered a task in which the exact number of words is going to be crucial, but I imagine some essays/projects might have strict limits. Still, if you submit a pdf file the assessor is going to have to find a way to determine the word count!

Just for completeness' sake I tried Adobe's online pdf conversion tool on the same file I mentioned earlier (250 page pdf).

http://www.adobe.com/products/acrobat/access_onlinetools.html

It's been literally 5 minutes now and the thing is still spinning its wheels ("Please wait while the requested PDF file is being converted"), so I don't think that's going to come charging in to rescue the situation!

Cheers, Paul

I rewrote the command with pdftotext, an open source text extraction utility that is included with xpdf.[1] It is *significantly* faster than the command using ps2ascii. However, it returns a different word count from the previous command. I compared the text generated by both ps2ascii and pdftotext. Neither will be the basis of an accurate word count. ps2ascii includes page numbers and inserts whitespace between the two parts of hyphenated words. pdftotext tends to subtitute whitespace for ligatures. There are other anomalies with each as well. Based on a small amount of experimentation with some of my documents, it seems that pdftotext gives a slightly better estimate. So the situation is this: with detex you get a low estimate and with pdf text extraction you get a high and slightly more accurate estimate. Disappointing, but perhaps there is a better way. All the best, Mark

[1] Mac OS X binary for xpdf (xpdf-tools-3.dmg) can be found here: http://users.phg-online.de/tk/MOSXS/