Yesterday, 05:56 AM
|
#1 |
|
Connoisseur
 
Posts: 67
Karma: 10
Join Date: Dec 2016
Location: France
Device: Kindle PaperWhite
|
Detect red underline spelling mistakes with a regex?
Hi gents,
Calibre editor and epub format How can I detect spelling mistakes with a regex? Calibre highlights these errors with a red underline... I would like to detect these visual cues and avoid comparing with a dictionary. My regex skills are not such an up to level. The work has already been done by Calibre. The idea is to jump from contentious points to contentious points with a quick read and make a manual correction. Thank you for your help. |
|
|
|
Yesterday, 06:11 AM
|
#2 | |
|
Well trained by Cats
 
Posts: 31,560
Karma: 62543878
Join Date: Aug 2009
Location: The Central Coast of California
Device: Kobo Libra2,Kobo Aura2v1, K4NT(Fixed: New Bat.), Galaxy Tab A
|
Quote:
|
|
|
|
|
| Advert | |
|
|
|
Yesterday, 07:16 AM
|
#3 | |
|
Connoisseur
Posts: 67
Karma: 10
Join Date: Dec 2016
Location: France
Device: Kindle PaperWhite
|
Quote:
Spell Check does the job. Many of the errors are in compound words separated by a hyphen or words broken by a line break. I am looking to reduce the column reading in Spell Check. Quick detection of red waves seems to me to be a way to skip a lot of pages and save time. Since this red wave exists, it must be possible to detect it through programming, that is the point of my question, and visual verification will be very quick by skipping thirty pages without errors. Spell Check will do a fine tuning job on the other errors. I haven't looked into the options you suggest, but I will do my homework. Best regards. |
|
|
|
|
|
Yesterday, 07:17 AM
|
#4 |
|
Groupie
Posts: 185
Karma: 1537710
Join Date: Jul 2021
Device: N/A
|
... and a double-click on the word in the list brings you to the first occurrence of the culprit, in the text
Edit : Useless reply not covering your needs, I was writing this while you were posting, and I don't know how to delete my msg.... Last edited by lomkiri; Yesterday at 07:21 AM. |
|
|
|
|
Yesterday, 08:38 AM
|
#5 | |
|
Groupie
Posts: 185
Karma: 1537710
Join Date: Jul 2021
Device: N/A
|
Quote:
--------------- I propose you this work-around : for each page, pass this regex-function, it will say you if there is some errors in the page and print the number and the list of them (number may be bigger than the number of words in the list, since it may have several errors for a word). Then you have a base if you want to adapt it (number of occurrences for each error, or scanning the whole book at once and give this list with the name of the file and the list for this page, in a debug box or in a text file, etc.) -------------- The search is scanning the whole page and replace it with itself, so you'll always get the message "replacement of 1 occurrence", but if there is errors, you'll get the list in a debug box. The the regex inside the function is scanning each word (skipping the html code) and asking if it is in the dictionary. It lowered all words before asking, if you don't want that, change the value of the variable "lowered" in the function. If there is no debug-box, it means there is no error detected The language in your opf file must be right, so the correct dictionary is selected. Put the cursor at the beginning of the file Code:
search: <body[^>]*>(.+)</body>
function-regex:
def replace(match, number, file_name, metadata, dictionaries, data, functions, *args, **kwargs):
"""
Count the number of errors using dictionary(), for one file.
Use "replace all", for "current file", with "dot all"
search string is <body[^>]*>(.+)</body>
"""
import regex
lowered = False # True or False
words = regex.findall(r"(?:<[^>]+>)(*SKIP)(*FAIL)|\b\w+\b", match[0])
nberr = 0
errors = set()
for word in words:
word = word.lower() if lowered else word
if not dictionaries.recognized(word):
nberr += 1
errors.add(word)
if nberr:
print(f'"Lower words before check" is {str(lowered)}')
print(f"{nberr} error(s) in this page, {len(words)} words")
print("\n", errors)
return match[0]
 Edit: This function obsolete and superset by its next version, see my next post. Last edited by lomkiri; Today at 07:28 AM. Reason: This version is obsoloete, see my next post |
|
|
|
|
| Advert | |
|
|
|
Yesterday, 11:38 AM
|
#6 |
|
Well trained by Cats
Posts: 31,560
Karma: 62543878
Join Date: Aug 2009
Location: The Central Coast of California
Device: Kobo Libra2,Kobo Aura2v1, K4NT(Fixed: New Bat.), Galaxy Tab A
|
And don't forget: F8 jumps to the next red squiggle in the editor
|
|
|
|
|
Yesterday, 12:05 PM
|
#7 |
|
Groupie
Posts: 185
Karma: 1537710
Join Date: Jul 2021
Device: N/A
|
I made the same function, but for all text files at once (or any number of files : current file, or selected files, etc.).
Same search string : <body[^>]*>(.+)</body> Put the cursor at the top of a file, preferably at the top of the first one to be scanned. Code:
def replace(match, number, file_name, metadata, dictionaries, data, functions, *args, **kwargs):
"""
Count the number of errors using dictionary(), for one file.
Use "replace all" with "dot all"
search string is <body[^>]*>(.+)</body>
"""
import regex
lowered = False # True or False
# First passage
if not data:
replace.call_after_last_match = True
data["regex"] = regex.compile(r"(?:<[^>]+>)(*SKIP)(*FAIL)|\b\w+\b")
data["files"]= {}
data["total_err"] = 0
# Last passage
if not match:
if data["total_err"]:
print(f'"Lower words before check" is {str(lowered)}')
print(f"{len(data['files'])} files scanned, {data['total_err']} errors in it")
print("==============================")
for el in data["files"]:
res = data["files"][el]
print("\n", f"file {el}: , {res[0]} error(s), {res[1]} words")
if res[0]:
print(res[2])
return
# Normal passage
nberr = 0
errors = set()
words = data["regex"].findall(match[0])
for word in words:
word = word.lower() if lowered else word
if not dictionaries.recognized(word):
nberr += 1
errors.add(word)
data["files"].setdefault(file_name, (nberr, len(words), errors))
data["total_err"] += nberr
return match[0]
Last edited by lomkiri; Yesterday at 09:11 PM. Reason: Fixed bug, can be used on any number of files |
|
|
|
|
Today, 07:26 AM
|
#8 | |
|
Groupie
Posts: 185
Karma: 1537710
Join Date: Jul 2021
Device: N/A
|
Quote:
Anyway, my regex-function gives another view on the error list, and in a book with lots of files and lots of proper names, it may indicates more quickly which files must be inspected. And it gives also the ability to inspect only one or few files instead of the whole epub. Anyway, it was funny to do it ;-) An example of the output: Spoiler:
(I could have sorted the list on alphabetic order; well, it's easy to add it) Last edited by lomkiri; Today at 07:32 AM. |
|
|
|
|
|
Today, 11:21 AM
|
#9 | |
|
Well trained by Cats
Posts: 31,560
Karma: 62543878
Join Date: Aug 2009
Location: The Central Coast of California
Device: Kobo Libra2,Kobo Aura2v1, K4NT(Fixed: New Bat.), Galaxy Tab A
|
Quote:
|
|
|
|
|
 |
«
Previous Thread
|
Next Thread
»
| Thread Tools | Search this Thread |
|
Similar Threads
|
||||
| Thread | Thread Starter | Forum | Replies | Last Post |
| Can Calibre correct all spelling mistakes at once? | UzmanKasap | Editor | 4 | 12-04-2021 09:27 PM |
| Any way to change the "red wavy underline"? | martyger | Sigil | 8 | 08-05-2015 12:24 PM |
| Useless without underline | Quetzalcoatlus | EPUBReader | 0 | 04-12-2014 02:30 AM |
| Underline | magavi | Devices | 5 | 06-21-2013 02:08 PM |
| another regex puzzle - detect capitalised phrases | cybmole | Sigil | 6 | 02-24-2012 10:04 AM |
All times are GMT -4. The time now is 12:16 PM.