Possible to get a word count in Calibre?

[Notjohn]

I have an epub consisting of 35 files. Is there any way to get a word count of the book in Calibre (or Calibre Editor)?

Never mind -- I found it! https://www.mobileread.com/forums/sho...d.php?t=242335

Thanks!

[jackie_w]

You could have a look at the Count Pages plugin which also does Wordcount and some other metrics.

[BetterRed]

@Notjohn - you can get word counts from the Sigil and Calibre book editors, they're in Tools->Reports. In Sigil they're in the "HTML" report, in Calibre they're in the "Words" report

The Count Pages PI needs a calibre library in which to store the results

BR

[eschwartz]

He knows -- he posted the same question in the Sigil forum precisely two minutes before he double-posted to this forum.

https://www.mobileread.com/forums/sho...d.php?t=270111

[BetterRed]

Originally Posted by eschwartz

He knows -- he posted the same question in the Sigil forum precisely two minutes before he double-posted to this forum.

https://www.mobileread.com/forums/sho...d.php?t=270111

<sigh>

Same book (English)

EPUB - Count Pages    = 20,563
EPUB - Calibre Editor = 21,405
EPUB - Sigil          = 21,382

RTF/DOCX - Word       = 20,751

IMO a 4.1% spread is quite a lot, it's a sample of one so it could be an outlier.

BTW that's the current official release of Count Pages

BR

[theducks]

Originally Posted by BetterRed

<sigh>

Same book (English)

EPUB - Count Pages    = 20,563
EPUB - Calibre Editor = 21,405
EPUB - Sigil          = 21,382

RTF/DOCX - Word       = 20,751

IMO a 4.1% spread is quite a lot, it's a sample of one so it could be an outlier.

BTW that's the current official release of Count Pages

BR

BR
What count method did you set in Count Pages?

[BetterRed]

Originally Posted by theducks

BR - What count method did you set in Count Pages?

Whatever this is



It was a 'new' book (one of today's intake), consequently the epub had copious quantities of crud. After delousing, the editor counts changed somewhat, so they must count crud (spans, superfluous styles, etc)

Deloused counts are in parentheses

Same book 

EPUB - Count Pages    = 20,563 (20,886)
EPUB - Calibre Editor = 21,405 (20,793)
EPUB - Sigil          = 21,382 (20,753)

RTF/DOCX - Word       = 20,751 (20,749)

TXT - Notepad++       =        (20,751)

So, if you're of the view that Word is the arbiter of good all things under the sun... then the winner is Sigil

You know my view on this - near enough, is good enough.

BR

[JSWolf]

Originally Posted by BetterRed

<sigh>

Same book (English)

EPUB - Count Pages    = 20,563
EPUB - Calibre Editor = 21,405
EPUB - Sigil          = 21,382

RTF/DOCX - Word       = 20,751

IMO a 4.1% spread is quite a lot, it's a sample of one so it could be an outlier.

BTW that's the current official release of Count Pages

BR

Are you using the new Count Pages beta that uses a more accurate routine to count words? Check the Count Pages thread. It's in there.

[BetterRed]

Originally Posted by JSWolf

Are you using the new Count Pages beta that uses a more accurate routine to count words? Check the Count Pages thread. It's in there.

Originally Posted by BetterRed

BTW that's the current official release of Count Pages

Which syllable of 'official release' don't you understand

I will look at the new version if and when its released. If it does not have the option of retaining the current algorithm, irrespective of the fact that according to you it is flawed, I shall not be installing it. For my purposes consistency trumps any puritan notion of accuracy.

BR

[davidfor]

Originally Posted by BetterRed

Whatever this is

Same book 

EPUB - Count Pages    = 20,563 (20,886)
EPUB - Calibre Editor = 21,405 (20,793)
EPUB - Sigil          = 21,382 (20,753)

RTF/DOCX - Word       = 20,751 (20,749)

TXT - Notepad++       =        (20,751)

That's even more fun. While looking at how Count Pages did the counting, I forgot to look at what it counted. They are each counting different things.

Based on a little experimenting and reading calibre code:

- calibre editor counts book text, text in the alt and title attributes of tags, text in the metadata and in the title tags of each internal file.
- Sigil counts the book text and the title tags.
- Count Pages just counts the book text.

The other reason for the difference is of course what they consider to be a word. Based on the some DLLs included with Sigil, I think it uses the same ICU method that the calibre editor uses.

[BetterRed]

Originally Posted by davidfor

That's even more fun. While looking at how Count Pages did the counting, I forgot to look at what it counted. They are each counting different things.

@davidfor - did you find any counting of angels on pins

The contentious issue is usually hyphenated words - one word or many, and if many what to do about 'not in dictionary' parts, and the ignoring of SHY.

Calibre's spell checker also checks the spelling in metadata. So if your book is written in EN-US and you're using an EN-US dictionary, but the Comments contain reviews from the LRB and The Torygraph you're likely to get misspelt words emanating from Comments. Would anyone markup the some of the Comments as EN-GB - assuming one can.

IMO, by default, only the substantial sections of the work should be counted and spell checked, not the scaffolding that glues it together or the marketing blurb.

I should point out that the 'book' in the counts in my earlier posts had no front or back matter, index, bibliography, or notes - it was an essay.

I just did some similar counts on a real book (Bankers New Clothes, Admati and Hellwig) which has lots of notes, has some tables and graphs, and a bibliography, and a long index. Sigil reports 163,081, Calibre Editor reports 163,959, and 'current official' Count Pages reports 173,007. How does that accord with your research.

BR

[davidfor]

Originally Posted by BetterRed

@davidfor - did you find any counting of angels on pins

The contentious issue is usually hyphenated words - one word or many, and if many what to do about 'not in dictionary' parts, and the ignoring of SHY.

To my mind, the point of the hyphen is to join thing into one word. Hence, it should always be counted as one. When I was experimenting, the problem of what was a hyphen was as much as a problem as anything else.

Calibre's spell checker also checks the spelling in metadata. So if your book is written in EN-US and you're using an EN-US dictionary, but the Comments contain reviews from the LRB and The Torygraph you're likely to get misspelt words emanating from Comments. Would anyone markup the some of the Comments as EN-GB - assuming one can.

IMO, by default, only the substantial sections of the work should be counted and spell checked, not the scaffolding that glues it together or the marketing blurb.

I want to check the spelling on the comments in the metadata, but I don't want them counted as part of the word count. Same goes for the ToC in an epub.

I should point out that the 'book' in the counts in my earlier posts had no front or back matter, index, bibliography, or notes - it was an essay.

Yes, and apparently there is a way to mark this in epub3. But, I've never seen a book that did it.

I just did some similar counts on a real book (Bankers New Clothes, Admati and Hellwig) which has lots of notes, has some tables and graphs, and a bibliography, and a long index. Sigil reports 163,081, Calibre Editor reports 163,959, and 'current official' Count Pages reports 173,007. How does that accord with your research.

The calibre editor and Sigil difference will be the metadata and any alt attributes of images. An extra 880 words is probably right there. But, 10,000 extra with Count Pages is surprising. They are usually a lot closer. I'm curious, so I've just grabbed the preview from Kobo to test. That has up to chapter 2, so hopefully it will have enough to see what the difference are.

[BetterRed]

Joining two words with a hyphen is done for different reasons, from Hyphens | Punctuation Rules

Rule 5. Never hesitate to add a hyphen if it solves a possible problem. [The following is an example of a well-advised hyphen]:

Confusing: Springfield has little town charm.
With hyphen: Springfield has little-town charm.

Without the hyphen, the sentence seems to say that Springfield is a dreary place. With the hyphen, little-town becomes a compound adjective, making the writer's intention clear: Springfield is a charming small town.

To my mind both those sentences have 5 words. If one was being paid on word count, why should the writer who is savvy enough to insert the 'well-advised' hyphen be penalised; or to put it the other way why should the writer without the wit to do so not be punished - 5 lashes afore the foremast, or a keel hauling at least.

I would count thirty-five, and self-obsessed as one word.

BR

[theducks]

BR
Didn't Word count (paid) used to be really Character count (not whitespace char) divided by N?

That was before we had markup.
So wouldn't converting to plain Text (removes all tags) and subtracting all the 'Spaces',Tabs, LF/CR get you 'just the letters'?

[davidfor]

BR: At the start of that article, it says:

Hyphens' main purpose is to glue words together.

Whenever I glue two things together, it is to make a single thing. Hence, the hyphen glues two words together to be one word.

And the devil in me wants to mention that your example is for a "compound adjective". Doesn't that mean the non-hyphenated version should be counted as a single word? Yeah, I'm stretching, but, what the hell

Anyway, the big problem is that without a very complete dictionary and correctly handling the grammar, there is no way to decide between the two. For simplicity, you have to decide that a hyphen either a word delimiter or part of the word.