"Special characters" in Search Results window (encoding issues?)

Alan Kilborn

Maybe this opens a can of worms, but I must ask.

The lead-up:

I recorded a macro for a Find All in Current Document search.
I edited shortcuts.xml to get to the macro’s text.
I changed some text in the macro to use a “special” UTF-8 character in the search term.
The search works; that isn’t the issue.

The issue is that when I run the macro, what appears in the Search Results window is strange for this character:

My “special character” is the a-with-the-hat then the black-boxed “APC”, in case that is not obvious. :-)

BTW, the character displays nicely if used in an editor window.

So, in thinking about the “why” of this happening, I start to wonder:

Is the search results window set up correctly for UTF-8? Or, alternatively for users of other encodings, is the search results window set up for the same encoding as the editor window that a search is initiated from?
What if a user has several files open in tabs that are various different encoding types, and the user runs a Find All in Opened Documents search? All of the results get put into the same result window, so how are the different encodings accounted for in the same window simultaneously?

Any and all input is appreciated.
We seem to be talking about “encoding” a lot lately.

Alan Kilborn

The usual group of very-opinionated posters have remained silent on this. :-)
Maybe it is a dumb/bad question?

Ekopalypse

@Alan-Kilborn said in "Special characters" in Search Results window (encoding issues?):

I know you have access to the find result window, did you try to read the current encoding?
As for the many documents with multiple encodings - your knowledge of
C++ is better than mine - if you haven’t figured it out, I won’t figure it out :-)

Alan Kilborn

@Ekopalypse said in "Special characters" in Search Results window (encoding issues?):

I know you have access to the find result window, did you try to read the current encoding?

I did not, but that is a good suggestion.
I will do that and report back.

your knowledge of

C++ is better than mine

Perhaps only slightly?
Wait, wait, Mr. Ctypes!

Truly, though, I thought I was missing something really obvious – how could no one have thought about N++ doing a find-in-files for perhaps many different encodings of the candidate files, and writing content output to a single window that, being a Scintilla buffer window, must have a single encoding specified for it?

Ekopalypse

@Alan-Kilborn

ctypes

:-D - with a “speakable” language like python - no problema

What I hope npp does is

read the content from the document and convert it to utf16
search for the string in utf16 as well
display the search result in whatever encoding is needs to do

but it looks like this isn’t the case.

Alan Kilborn

I realized I could have added more info to my original post:

The character I manually added to my macro was this one:
⟯
which looks like a right-parens but is actually this:
https://graphemica.com/⟯

When I put it into shortcuts.xml I just pasted it as a single character, but now I notice that it looks like this if I open the xml:

â¯

This all makes sense as the xml is UTF-8; I just wanted to add the info to the thread.

guy038

Hi, @alan-kilborn,

Many thanks for pointing us to https://graphemica.com ! Really, this site gives a ton of information on each character ;-))

Just for information :

https://graphemica.com                     =>  SEARCH of a SPECIFIC Unicode character

https://graphemica.com/characters          =>  List of Unicode CHARACTERS by NAME

https://graphemica.com/categories          =>  List of Unicode GENERAL_CATOGORIES

https://graphemica.com/blocks              =>  List of Unicode BLOCKS

https://graphemica.com/scripts             =>  List of Unicode SCRIPTS

https://graphemica.com/unicode/characters  =>  List of Unicode CHARACTERS per PAGE of 256 chars


INPUT Examples :

Char       =>  Character ITSELF


&10180
           =>  CHOICE between character U+10180 and character U+27C4  =  U + Hex(10180)
#10180


10180
           =>  Character U+10180 ( GREEK FIVE OBOLS SIGN )
&#x10180


&#10180    =>  Character U+27C4 = U + Hex(10180)  ( OPEN SUPERSET )

Cheers,

guy038

Alan Kilborn

@guy038 said in "Special characters" in Search Results window (encoding issues?):

Many thanks for pointing us to https://graphemica.com ! Really, this site gives a ton of information on each character

It was a site previously unknown to me.
Glad you enjoyed it.
BTW, I found it because I googled E29FAF after finding that Notepad++ had converted my entry of the UTF-8 character to these codes – my quickest way to explain the “more info” I wanted to provide in my previous post.

Alan Kilborn

@Ekopalypse said in "Special characters" in Search Results window (encoding issues?):

I know you have access to the find result window, did you try to read the current encoding?

My investigation down this path:

# Get the code page used to interpret the bytes of the document as characters.
get int GetCodePage=2137(,)

Calling that function on the Find result window results in a 65001 being returned.

Looking in scintilla.h I see:

#define SC_CP_UTF8 65001

So it does appear that the encoding for the Find result window is UTF-8.

Which makes me wonder even more why my UTF-8 character doesn’t show up correctly there.

Ekopalypse

@Alan-Kilborn

The other usual suspects would be the font and technology (GDI/DirectX).
By the way, which font do you use?
None of my installed fonts show this symbol in the editor.

Alan Kilborn

@Ekopalypse said in "Special characters" in Search Results window (encoding issues?):

The other usual suspects would be the font and technology (GDI/DirectX).
By the way, which font do you use?

I have direct-write enabled.
I play around with different fonts (can’t seem to find the best–for me); “Consolas” is the one currently in favor for me.

None of my installed fonts show this symbol in the editor.

You mean this actual character?:

⟯

Ekopalypse

@Alan-Kilborn said in "Special characters" in Search Results window (encoding issues?):

You mean this actual character?:

I guess so, even my browser refuses to display it

But more I think about it, the less I’m convinced that it is a font or technology issue as you do get a representation - just a different one.

Alan Kilborn

@Ekopalypse said in "Special characters" in Search Results window (encoding issues?):

What I hope npp does is

read the content from the document and convert it to utf16

search for the string in utf16 as well

display the search result in whatever encoding is needs to do

When I noticed the following in the Scintilla documentation, it reminded me of Eko’s points above:

SCI_ENCODEDFROMUTF8(const char *utf8, char *encoded) → position
SCI_ENCODEDFROMUTF8 converts a UTF-8 string into the document’s encoding which is useful for taking the results of a find dialog, for example, and receiving a string of bytes that can be searched for in the document.

I’m not sure what I’m saying by pointing this out; perhaps just noticing a somewhat common theme? :-)

Ekopalypse

@Alan-Kilborn

since the Windows API uses utf16 for internal storage of strings, this does not sound efficient.
I wonder why the scintilla devs thought that this might be a good idea.
Hmm … most likely because the library is used on different platforms,
but then why not have a compile time variable to convert it to utf16 on Windows and utf8 on Linux … ??

Alan Kilborn

@Ekopalypse

In hindsight, I probably shouldn’t have mentioned SCI_ENCODEDFROMUTF8 as I did a search of the Notepad++ source code, and it isn’t used there.

Alan Kilborn

@Alan-Kilborn said in "Special characters" in Search Results window (encoding issues?):

When I put it into shortcuts.xml I just pasted it as a single character, but now I notice that it looks like this if I open the xml:
â¯

So just providing some more data on this, after I experimented with it a bit more:

I see that if I record a macro using the multibyte unicode character (discussed much further up in the thread), instead of “hand editing” shortcuts.xml after-the-fact, when I run the macro I DO see the correct character appearing in the Find result window:

AND…

if I later look at the saved xml, I see this for that character in the macro:

⟯

Which does indeed make sense.

So perhaps the error was mine and it comes down to directly inserting the unicode character into the XML instead of inserting its &#x.... code.

Alan Kilborn

More thoughts:

From my immediately preceding post, it follows that anytime a “special” character is used in a N++ configuration xml file, the “html syntax”, example &#27EF should be used, rather than inserting the character “directly”, for example via a paste.

However, I notice in english_customizable.xml the following, which does not follow this idea:

<Item id="1721" name="▲"/>
<Item id="1723" name="▼ Find Next"/>

But yet these items display correctly…
Okay, different usage from the above; these appear on buttons in the UI, the earlier discussion is some text in the Find result window…

But in general, I would be interest to know why these don’t require any special “treatment” in the xml.
Or what is the “rule”?
Always use the “html syntax” seems the “safest”.

Ideas? Or is the topic too “meh” for anyone to care? :-)

PeterJones

@Alan-Kilborn said in "Special characters" in Search Results window (encoding issues?):

Always use the “html syntax” seems the “safest”.
Ideas? Or is the topic too “meh” for anyone to care? :-)

I don’t know enough about XML to know the default encoding, but my guess is Windows-1251.
If you added

<?xml version="1.0" encoding="UTF-8" ?>

at the beginning of the shortcuts.xml (and reloaded Notepad++), does that allow you to hand-insert the https://graphemica.com/⟯ or other special character into the macro XML directly?

Alan Kilborn

@PeterJones said in "Special characters" in Search Results window (encoding issues?):

If you added

First, I’m a bit surprised that after adding that line manually, that Notepad++ allows it to remain (after recording a new macro, forcing N++ to rewrite shortcuts.xml).

Second, it was a good idea, but sadly, after trying it, I get the same result as earlier, specifically, “garbage” characters in the Find result window text.

Third, thanks for the interest, @PeterJones