Plugin Development

[AxaRu]

Yes. This good way.
I think if you adds to API class like cursor you will get more interactivity in plugins.

[KevinH]

Not straightforward, as we use copy on write file interface in a separate process to protect Sigil from plugin errors. In addition, internally everything Qt is encoded as utf-16 with no carriage returns, but the files themselves utf-8 encoded and carriage returns are included but only on Windows. And finally, each xhtml parser treats DOM building and especially how whitespace and text is handled differently making cfi and path approaches incorrect as well.

So even something simple such as a cursor position is in a Qt based QPlainTextEdit internal to Sigil is hard to translate to a utf-8 encoded file with platform specific line endings, tab expansions, etc, being parsed by a different xhtml parser.

So using a marker such as an inserted comment, class or id, is the most reliable way to handle cursor/text marking in xhtml that works cross platform across file formats.

Furthermore, there is very little simple text manipulation that can not be handled internal to Sigil using Regular Expressions, clips, Saved Searches, etc.

Originally Posted by AxaRu

Yes. This good way.
I think if you adds to API class like cursor you will get more interactivity in plugins.

[DiapDealer]

Cross-platform/line-ending issues aside, how would one even go about translating a C++ QTextCursor object (or any reliable representation thereof) to the plugin framework's independent python process in a way that could be used? I'm certainly at a loss.

[The_book]

Is there any plugin API to get "current file" and 'tabbed file' like it in find/replace?

[KevinH]

There is an interface to get the list of files that are selected in BookBrowser.
So if you only want the current file or a specific subset of files, make sure they are properly selected in BookBrowser before launching your plugin. The interface call is here:

https://github.com/Sigil-Ebook/Sigil...tainer.py#L313


Alternatively your plugin can use either Tk or Qt to allow the user to select any set of files they desire.

[The_book]

Can I use exec in pulgin file? It seems changes nothing.
If I can't, how can I run some code from user's input?Create python file from user's input and then import it?

[DiapDealer]

Originally Posted by The_book

Can I use exec in pulgin file? It seems changes nothing.
If I can't, how can I run some code from user's input?Create python file from user's input and then import it?

Plugins are written in Python. If you want user input, you need to create a gui using Python's tkinter or Python for Qt to collect that user input.

You cannot use the plugin framework to collect user input on the command-line.

[The_book]

Originally Posted by DiapDealer

Plugins are written in Python. If you want user input, you need to create a gui using Python's tkinter or Python for Qt to collect that user input.

I know this. And when I write replay to you, I know what make me wrong. I run exec in function, and exec changes global. I need get my change from global

[DiapDealer]

Upcoming Sigil Qt6 migration and GUI plugin development changes as a result.

See this sticky thread for details.

[philm]

Hello and Happy New Year!

I'd need to count selected words. I didn't find a way to do it with existing functions and plugins.

I downloaded and studied testme3 but can't find how to access selected text.

Would you please point me in the right direction to access text selection from preview and work from there?

Or if you have a complete idea how to count words in selection, please let me know

[Doitsu]

Originally Posted by philm

I'd need to count selected words. I didn't find a way to do it with existing functions and plugins.

Unfortunately, there are no API functions for detecting the cursor position or selected text.
As a workaround, you could copy selected text to the clipboard, read the clipboard contents with Python and then count the words.

[KevinH]

You can restrict searches to marked text in Sigil and use Find to count anything you want. If do not want counts of specific words you can use regular expressions to count all words in marked text.

[philm]

Thank you very much for your replies!

Actually, I need to know where the 280th -ish word is, so I thought to select some text approximately, and have a word count (\w+ ) but you open up possibilities.

Originally Posted by KevinH

You can restrict searches to marked text in Sigil and use Find to count anything you want. If do not want counts of specific words you can use regular expressions to count all words in marked text.

When I select text in preview and search > mark selected text it only unmarks, as stated in the info bar: "Text selection unmarked". I'm using Ubuntu 22.04

When I select text in source xhtml, CTRL+SHIFT+M does mark text but searching for "\w+ " results in all tags content to be counted as words, like "a " in `<a href...` or elsewhere in tags (style, etc).

I need to count rendered words, not everything from code.
Actually I need to spot every 280th rendered word. Sorry for not being clearer in my first draft.
Any idea welcome!

[Doitsu]

Originally Posted by philm

I need to count rendered words, not everything from code.

Tools > Reports > HTML Files > All words

Originally Posted by philm

Actually I need to spot every 280th rendered word.

You could use BeautifulSoup as a text filter:

Spoiler:

#!/usr/bin/env python
# -*- coding: utf-8 -*-
import sys, os
from sigil_bs4 import BeautifulSoup

# main routine
def run(bk):

    # iterate over all html files
    for html_id, href in bk.text_iter():

        # read orignal html code from file
        html = bk.readfile(html_id)
        name = os.path.basename(href)

        # use BeautifulSoup to remove the tags
        soup = BeautifulSoup(html, 'html.parser')
        text = soup.get_text()

        # split text 
        words = text.split(' ')
        
        # get the 280th word
        if len(words) >= 280:
            print(name, '280:', words[279])

    return 0

def main():
    print('I reached main when I should not have\n')
    return -1

if __name__ == "__main__":
    sys.exit(main())

[philm]

Thank you again! Unfortunately, I don't know how to add Beautiful Soup to make a Sigil plugin, and reports doesn't count in a selection (spot every 280th).

chatGPT threw me something that partially works, except it also count words in links but it doesn't happen often in my texts: `(\b\w+\b)(?=(?:[^<>]*<[^<>]*>)*[^<>]*$)`

PS: Well, after adding some small improvements (space, parenthesis and punctuation added) it gets pretty decent: `((|\()\b\w+\b( |:|,|\)))(?=(?:[^<>]*<[^<>]*>)*[^<>]*$)` Hmm, nope, too limiting.

FYI: It misses the word " a " since it's alone and still counts some content in html tags.