Word frequency

[GBS]

Hello,

Every day I use this program it gets more into the purchase direction, I feel like there is no escape

Is it possible to have a word frequency or at least word count? Or does it have already and I didn't found?
Some journals require me to limit the words (including spaces) in abstract, witch is easy when I produce my own documents but sometimes I get to review from others.
By word frequency I mean a list of all the words (witch could include spaces like for short phrases) present in the PDF together with the its occurrences (number of times they show up). It seems rather simple algorithm but I couldn't find any pdf reader that does so.

[Patrick-Tracker Supp]

Hello GBS,

Thank you for the post. I am afraid this is not something we offer, however you can search a specific word to find how many times it occurs within the document. You can activate the advanced search via [Shift]+[Ctrl]+[F] or under Edit> Search.

I hope this helps!

[Joxon]

For a simple word count you can use the JaveScript Console: Ctrl+J. Enter:

Code: Select all

var cnt=0;
for (var p = 0; p < this.numPages; p++) cnt += getPageNumWords(p);
app.alert("There are " + cnt + " words in this file.");

Click Run.

[Bhikkhu Pesala]

For a simple word count you can use the JaveScript Console: Ctrl+J. Enter:

That's worth knowing. Can it be made to work for the selected text?

[Will - Tracker Supp]

Hi Bhikkhu,

From what I can see, there isn't any way to get the selected text via JS. One potential workaround is to use the Highlighter Tool to highlight a block of text and get the word count of that instead. I was able to find this article, which should provide a good starting point:
http://asserttrue.blogspot.co.uk/2010/0 ... -lack.html#

Cheers,

[Will - Tracker Supp]

Bhikkhu - thinking about it, can you provide a description of why you'd need that to work with selected text? If I know the exact use-case, I might be able to adapt the JS to do as you need and provide it as a toolbar button & the regular script.

I emphasize the word may because, as I believe you have seen in other topics, our official policy is that we don't provide users with JavaScripts, the onus is on them to write them. We make exceptions if the script is useful to a large number of users and is quick/simple to write. I'm also not hugely proficient in JS, so it may be beyond my ability to write.

Cheers,

[Bhikkhu Pesala]

I did not have any specific use case in mind, but I think it would be more useful than counting the words in the entire document.

Alfred on CommunityPlus suggested this improvement, which I will pass on: The Error Icon should be replaced with an information icon.

Code: Select all

var cnt=0;
for (var p = 0; p < this.numPages; p++) cnt += getPageNumWords(p);
app.alert("There are " + cnt + " words in this file.",3);

[yogi108]

Hi,

As Javascript has its limitations, we can use https://autohotkey.com/ as external tool (Software License: GNU General Public License)
Just got this to run, if someone knows a better regex etc:

1.) Install the autohotkey software
2.) make one file (e.g. clipboard.ahk) with contents:

Code: Select all

!c:: ;define ALT-c as Hotkey
#NoEnv  ; Recommended for performance and compatibility with future AutoHotkey releases.
#Warn  ; Enable warnings to assist with detecting common errors.
SendMode Input  ; Recommended for new scripts due to its superior speed and reliability.
SetWorkingDir %A_ScriptDir%  ; Ensures a consistent starting directory.
clipboard = ; Empty the clipboard
ClipWait ; wait for clipboard change
if ErrorLevel
{
    MsgBox, The attempt to copy text onto the clipboard failed.
    return
}
MsgBox, clipboard = %clipboard%

Array := StrSplit(clipboard, A_Space, ".")  
max:=Array.MaxIndex()
MsgBox, Words counted=%max%
return

3.) Open one app (e.g. PDF-XChange Editor)
4.) press the hotkey ALT-c (hopefully not used on your side - changeable), will clear the keyboard and waits for selected text to be copied with Cntrl-c
5.) select text
6.) press ctrl-c and see what happens

Just tested with word and PDF XChange, not bad
kind regards,

Edit: Here the executable compiled for testing:

clipboard.zip

Standalone for counting words

(385.27 KiB) Downloaded 101 times

[David.P]

Because the "word frequency" feature request of @GBS is partially related, here's a link to a thread with a similar request, about auto-highlighting of important keywords for easier rand faster reading:

Suggestion: Revolutionary Killer Feature for Power Reading

[GBS]

So it seems like there is quite some interest in this !

I've never thought about the "selected" part, I had in mind the whole file. I could use that simply by selecting all (Ctrl + A). But still not a word frequency as I see.

Wow David that's a lot more complex, because how would the program decide whats "important"?
Anyway, I really like the bottom box with colored words and their count.

[Tracker Supp-Stefan]

Hello GBS,

As mentioned above - JS can give you a word count, but for frequency and more complicated text manipulation and analysis it will be better if you use a proper text Editor. We offer a feature to export the PDF file to e.g. a Word format - so you can then use all the tools offered there to analyze the text as needed.
Given the complexity of PDF files, and their varied uses - I can't promise you if or when we might add a similar feature in the Editor, so for now it's best to export the text and analyze it in another tool.

As for David's suggestion - it is indeed a lot more complicated, and I again can't make any promises as to if or when such functionality might be available in our products.

Regards,
Stefan

[David.P]

Hi @all,

Wow David that's a lot more complex, because how would the program decide whats "important"?

It's not that complex, and I already had it implemented in the Firefox browser (see discussion here). For a start, you simply use the words that are used most in any given article or text, and exclude common words like and, is, etc. by using a simple (static) list of stop words. The tool works very good this (simple) way.

I.e., what you do is, you make a simple list of all words in a given article, subtract all stop words from the list, sort the remaining words by frequency, and then use like the 12 most prevalent words for a normal Search operation in PDF-XChange Editor.

That's about all this feature would need to be implemented

Cheers
David

[Will - Tracker Supp]

Hi David,

That does sound fairly complex and things are often more complex programmatically than they would at first appear. We'll obviously consider the feature, but it's likely a fairly low priority consideration at the moment.

Thanks,

[David.P]

That's understood Will. Thank you. The discussed feature of course would be more like a freestyle/luxury feature for advanced text analysis.

If someone wants to try a "predecessor" of that feature manually, here's a great online text analysis tool:
Textalyser

This tool gave gave me the following keywords list (with number of occurrences) from my attached sample e-book "Emma" (saved as *.txt from PDF-XChange Editor):

harriet 491
knightley 385
woodhouse 306
fairfax 237
churchill 221
hartfield 152
highbury 122
pleasure 115
morning 107
feelings 98

If I use these terms:
harriet knightley woodhouse fairfax churchill hartfield highbury pleasure morning feelings
...for an Advanced Search in PDF-XChange Editor ("Find text with any of these words"), then I already get a very nice colored highlighted overview about what this book actually is about and where those keywords are concentrated/distributed:


Anyway, this is of course a wide field. Strangely however, there still seems to be no software available whatsoever that can do what my original feature request describes (and partly delivers).

Cheers
David

[Will - Tracker Supp]

Thanks for the details, they've been noted and we'll put it in for consideration