Search in folder (encoding)

PeterJones

@Mayson-Kword said in Search in folder (encoding):

So “Find in files” function actually search in ANSI encoding

I was vaguely remembering something like that discussed earlier, but I cannot find the relevant discussion and/or issue: sorry. (I thought I’d found it with previous links, but apparently I misread.)

guy038

Hello, @mayson-kword, @Peterjones and All,

I cannot reproduce your issue !

I pasted the string “ты знаешь о” phrase in a a new UTF-8 encoded file
In addition, I saved it, in the directory of my portable v7.9.2 installation, as ты знаешь.txt
As you can verify in the first picture, this file is not presently opened
Searching in any file , through all levels of the D:\@@\792 folder, does find the line of your file with some Russian characters !

Best regards

guy038

Alan Kilborn

@Mayson-Kword said in Search in folder (encoding):

Changes in theese options do nothing with search in folder.

This is curious as the way I believe it works (and has to work) is that N++ opens each file that matches the Find in File specification for Directory and Filters into a tab that isn’t shown to the user (if the file it is going to search is not already open).
That tab is opened the same way in all other respects except for visibility as a normal tab.
At least this is my knowledge about it – which is limited. :-)

Can you provide your Debug Info found on the ? menu?

Mayson Kword

@guy038, you can read my posts some more, search doesn’t work only if there is unknown data in file. This post. My apologies.

@Alan-Kilborn, of course.

Notepad++ v7.9.1 (64-bit)
Build time : Nov 2 2020 - 01:07:46
Path : D:\Programs\Notepad++\notepad++.exe
Admin mode : OFF
Local Conf mode : ON
OS Name : Windows 10 Pro (64-bit)
OS Version : 2004
OS Build : 19041.685
Current ANSI codepage : 1251
Plugins : AutoCodepage.dll DSpellCheck.dll JSMinNPP.dll MarkdownViewerPlusPlus.dll mimeTools.dll NppConverter.dll NppExec.dll NppExport.dll XMLTools.dll

Maybe I also should provide my example file?

Alan Kilborn

@Mayson-Kword said in Search in folder (encoding):

Maybe I also should provide my example file?

Yes, in general the more you can provide to help someone reproduce, the better!

AutoCodepage.dll

I wonder if this plugin is interacting in some way?

Mayson Kword

@Alan-Kilborn, here they are, 3 files. 01 is original, 02 is cut, 03 is minimal. Word for search is “Мисти” (correct but not working) or “РњРёСЃС‚Рё” (incorrect but working), here is my search result.

Also plugin AutoCodepage was my try to solve this problem, but it didn’t help. However, it works pretty nice when I open files, not search in them. The problem remains both with and without it.

Thank you for so much attention.

guy038

Hi, @mayson-kword, @Peterjones, @alan-kilborn and All,

First, thanks for your files that I could download without any problem

Now, file encodings are really a puzzle for everybody and it’s a bit difficult to do pertinent tests because :

From your Debug-Info, your current ANSI codepage is 1251
From my Debug-Info, my current ANSI codepage is 1252

For instance, when one opens the jackie_default_01.json file, which is an ANSI encoded file, each of the 8 occurrences of the Russian word Мисти, found in lines 13, 14, 21, 22, 35, 123, 127 and 128, are encoded as :

ÐœÐ¸ÑÑ‚Ð¸ in my configuration, as ANSI represents really the Windows-1252 encoding
РњРёСЃС‚Рё, with your configuration, as ANSI represents really the Windows-1251 encoding

Refer to the table, below, and the following links :

https://en.wikipedia.org/wiki/Windows-1251

https://en.wikipedia.org/wiki/Windows-1252

https://www.unicode.org/charts/PDF/U0400.pdf

•--------------•-------------•-------------•-------------•-------------•
|      М       |     и       |    с        |    т        |     и       |
•--------------•-------------•-------------•-------------•-------------•
|  CAPITAL EM  |   SMALL I   |   SMALL ES  |  SMALL TE   |   SMALL I   |    CYRILLIC letters
•--------------•-------------•-------------•-------------•-------------•
|    U+041C    |   U+0438    |   U+0441    |   U+0442    |   U+0438    |    UNICODE code-points
•--------------•-------------•-------------•-------------•-------------•
|   D0   9C    |   D0   B8   |   D1   81   |   D1   82   |   D0   B8   |    BYTE values of the characters with an UTF-8 encoding
•--------------•-------------•-------------•-------------•-------------•
|   Р    њ     |   Р    ё    |   С    Ѓ    |   С    ‚    |   Р    ё    |    Characters displayed, AFTER "Encoding > Character sets > Cyrillic > Windows-1251"
•--------------•-------------•-------------•-------------•-------------•
|   Ð    œ     |   Ð    ¸    |   Ñ    HOP  |   Ñ    ‚    |   Ð    ¸    |    Characters displayed, AFTER "Encoding > Character sets > Cyrillic > Windows-1252"
•--------------•-------------•-------------•-------------•-------------•

It’s important to understand that, when you 're using the ANSI, UTF-8, …Character sets option, your file contents do not change at all . Notepad++ just re-interprets the file bytes as it would represent this new encoding !

So I suppose that if you want to change the UTF-8 encoding of a file to your current ANSI encoding ( so Windows-1251 ) you should :

Run the option Encoding > Convert to ANSI
Save the modifications

As you can see the Russian word Мисти is still displayed ( but with the ANSI encoding ) and the search of the string Мисти should work as expected !

Refer the table, below, and the link :

https://en.wikipedia.org/wiki/Windows-1251

•--------------•-------------•-------------•-------------•-------------•
|      М       |     и       |    с        |    т        |     и       |
•--------------•-------------•-------------•-------------•-------------•
|  CAPITAL EM  |   SMALL I   |   SMALL ES  |  SMALL TE   |   SMALL I   |    CYRILLIC letters
•--------------•-------------•-------------•-------------•-------------•
|    U+041C    |   U+0438    |   U+0441    |   U+0442    |   U+0438    |    UNICODE code-points
•--------------•-------------•-------------•-------------•-------------•
|   D0   9C    |   D0   B8   |   D1   81   |   D1   82   |   D0   B8   |    BYTE values with an UTF-8 encoding
•--------------•-------------•-------------•-------------•-------------•
|      CC      |     E8      |     F1      |     F2      |    E8       |    BYTE values, of the word "Мисти", AFTER "Encoding > Convert to ANSI"
•--------------•-------------•-------------•-------------•-------------•

This time, as you have used a Encoding > Convert ... option, the file did change and the present byte values of the characters are replaced with the byte values of these characters in this new encoding !

Best Regards,

guy038

guy038

Hi @mayson-kword and All,

Sorry, in my previous post I said :

ÐœÐ¸ÑÑ‚Ð¸ in your configuration as ANSI represents really the Windows-1251 encoding

РњРёСЃС‚Рё, with my configuration as ANSI represents really the Windows-1252 encoding

In fact, the correct phrasing is the opposite :

ÐœÐ¸ÑÑ‚Ð¸ in my configuration as ANSI represents really the Windows-1252 encoding
РњРёСЃС‚Рё, with your configuration as ANSI represents really the Windows-1251 encoding

Note that I also modified my previous post to be exact !

BR

guy038

Mayson Kword

@guy038, converting file to ANSI works for my search, but it’s not a solution.

As I said in my first post, I have a lot of files - 3222 (in all sub-folders) at this moment. I can’t open them all and convert to ANSI, then edit, then convert back to UTF8 (because I need them in UTF8).
Even if I had few files converting them to UTF8 and back to ANSI is not a solution because the result is not equal to original file. See the image. I need no such structure changes because theese files will be put in a specific archive and used in game. Ah, it’s all very fragile.

Some users can fix this problem, of course. Convert file to ANSI and back to UTF8 / remove problematic symbols / open all files in current session. But I can’t do such thing because converting and such editing changes my files too much, also there are too many files to open them all.

Look at this post once again. The difference between case 2 and case 3 is only one “NUL” symbol - and it makes N++ not be able to find anything I want. I can provide more files to compare. Search works wierd if file contains wrong data - Pic 1 and Pic 2.

Mayson Kword

@Mayson-Kword said in Search in folder (encoding):

Even if I had few files converting them to UTF8 and back to ANSI is not a solution

Sorry for mistake.
I mean “Even if I had few files converting them to ANSI and back to UTF8 is not a solution”.
Can’t edit post after 3 minutes.

gstavi

What did you mean by:

Also plugin AutoCodepage was my try to solve this problem, but it didn’t help. However, it works pretty nice when I open files

Do you have problems detecting encoding in general? If so there is no reason to expect search to work for non-ascii strings.
Try saving as UTF8 with BOM. The BOM makes detection of UTF8 files explicit.

Mayson Kword

@gstavi said in Search in folder (encoding):

Do you have problems detecting encoding in general?

Yes and no, it only happens with some rare files if I don’t use “autodetect character encoding” feature. Sometimes even with it on. In that case N++ open them using ANSI, so I can’t read anything and I need to select encoding manually. AutoCodepage with my own rules just do this work automatically instead of me.

However there is an idea in your words. After some investigations I can suggest that N++ can’t search as expected in file using “Find in files” if file is not currently opened and if N++ can’t autodetect its encoding. Am I right?

Also using UTF8 with BOM solves this problem in a best way, I think. No data loss at least. Is there any way to automatically convert thousands of files from UTF8 w/o BOM to UTF8 with BOM?

gstavi

Adding BOM is not conversion it is just adding 3 bytes (for utf-8 bom) at the beginning of the file.
https://stackoverflow.com/questions/3127436/adding-bom-to-utf-8-files

The 2 risks are:

Adding BOM into already “BOMed” file.
Adding UTF-8 BOM into file with another encoding (e.g. UCS-2).

I never needed to use AutoCodepage and don’t know what hook from Notepad++ it uses to apply its functionality but maybe either its author or Notepad++ main developers can “fix” the problem by ensuring that it is called during search as well.

Mayson Kword

Ok, using files with BOM solves my problem because I can easily check files for BOM and change them if needed using Python scripting. However, N++ still doesn’t search properly in closed file if fails to autodetect its encoding.

Thank you all for your participation, that’s all.

guy038

Hi @mayson-kword and All,

I was able to download your txt.zip archive and correctly extract your two files example file 01.txt and example file 02.txt

With these two files opened in N++ v7.9.2, I got two occurrences of the string они должны searching in all opened tabs of the current session. So, I could not, again reproduce the issue !

After a while, I realized that, for a correct search, you need to tick the option Settings > Preferences > MISC. > Autodetect character encoding

If this option is not checked, the string они должны is found in the file example file 02.txt only, which does not contains an ending \x00, whatever the files are opened or not in current N++ session

When this option is enabled, and the two files opened in current session, a click on the Find All in All Opened Documents, of the Find dialog, produces :

When this option is enabled, and your two files not opened, a click on the Find All button, of the Find in Files dialog, produces :

Now, about the possibility of changing an UTF-8 to an UTF-8-BOM encoded file, this is very easy, from within Notepad++

Open the Find in Files dialog ( Ctrl + Shift + F )
SEARCH \A
REPLACE \x{FEFF}
Select the Regular expression search mode
Select the folder containing all your UTF-8 files which have to be modified
Select the appropriate files filter
Click on the Replace All button and confirm the dialog

Voilà !

If you want, first, to test this technique :

Open an UTF-8 file ( In the status bar, you should see UTF-8 ( and not UTF-8-BOM )
Open the Replace dialog ( Ctrl + H )
Tick the Wrap around version
Select the Regular expression search mode
Click, ONCE only, on the Replace All button ( Do not click, previously, on the Find Next button or any other button ! )

=> Message Replace All: 1 occurrence was replaced in entire file

Save, immediately, the modifications with Ctrl + S ( Just note that it still mentions the UTF-8 encoding, in the status bar )
Close this file ( Ctrl + W )
Re-open the file ( Ctrl + Shift + T )

=> In the status bar, we can verify, this time, the indication UTF-8-BOM, which proves that we are dealing, now, with a real UTF-8 file with a BOM.

Best Regards

guy038

Alan Kilborn

@guy038 said in Search in folder (encoding):

about the possibility of changing an UTF-8 to an UTF-8-BOM encoded file, this is very easy, from within Notepad++

One thing to note about this is that it is a one-way operation.
You can’t go the other way (UTF-8-BOM —> UTF-8) using a similar technique (a N++ replacement).
@guy038, am I right about it?

guy038

Hi, @alan-kilborn,

Totally exact, Alan. Indeed, the BOM structure is not considered as part of file contents by Notepad++ !

Note also, that I found out a bug, relative to the \A assertion, while testing my method to add the BOM with a regex S/R

Let me some minutes to expose the problem and you tell me if I must create an issue for such a behaviour !

BR

guy038

guy038

Hi, @alan-kilborn and All,

Sorry for the wait, I need to eat a little bit !

Alan, I think that you already spoke about a similar behaviour, but I cannot remember the exact post

Just follow all these steps to see the issue !

Open a new tab ( Ctrl + N )
Type the three letters bar, only
Save this new tab as Test.txt
Open the Replace dialog ( Ctrl + H )
SEARCH \A
REPLACE foo
Tick on the Wrap around option
Select the Regular expression search mode
Click on the Replace All button

=> As expected, the file contents are changed into the string foobar !

Now :

Undo the modifications ( Ctrl + Z )
Re-open the Replace dialog ( Ctrl + H )
SEARCH \A ( Verify that text is indeed \A )
REPLACE foo
Tick on the Wrap around option
Select the Regular expression search mode
Click, first, on the Find Next button ( Important )

=> The classical call tip appears, saying zero length match

Now, click on the Replace All button

=> This time, no replacement occurs, even of you click, again, on the Replace All button

Even if you switch to an other tab and switch back to the Test.txt file

=> The same regex S/R, as above, with, only, a click on the Replace All button does not work anymore ! And you always get the message Replace All: 0 occurrences were replaced in entire file

In order to get the expected behaviour, you must :
Close this file ( Ctrl + W )
Re-open the Test.txt file ( Ctrl + Shift + T )

or

Close and re-start Notepad++, of course !

After these operations, a click on the Replace All button is, again, functional and do add the string foo, right before the string bar !

Here is my debug -info :

Notepad++ v7.9.2   (32-bit)
Build time : Dec 31 2020 - 03:58:36
Path : D:\@@\792\notepad++.exe
Admin mode : OFF
Local Conf mode : ON
OS Name : Microsoft Windows XP (32-bit) 
OS Build : 2600.0
Current ANSI codepage : 1252
Plugins : DSpellCheck.dll mimeTools.dll NppConverter.dll NppExport.dll

Best Regards

guy038

Alan Kilborn

@guy038

I can reproduce that ; nice instructions!
You should open a real issue on that.
It’s a weird one.
I thought it might have something to do with the call-tip being active, but no, it doesn’t seem to have any bearing on it.

Allan, I think that you already spoke about a similar behaviour, but I cannot remember the exact post

I don’t remember this at all, but at my advanced age…
I’m sure you could find the post, point me to it, and it would be like a stranger had written it. :-)

Mayson Kword

You cannot reproduce the issue because N++ autodetect encoding properly for most of files if using feature “Autodetect character encoding”. But there are some files that N++ cannot interpret right even with this option on. My file “jackie_default_01.json” is one of them. Also 3221 more files are one of them.

In case if you didn’t see my words, I can repeat myself from this post. N++ doesn’t search properly in closed file if fails to autodetect its encoding. Maybe I should repeat this once more in a different way? Conditions to reproduce the issue are:

File is not opened.
N++ cannot autodetect its encoding.

You said “I could not, again reproduce the issue” just because you don’t meet the second condition.