Fix crash when tessdata directory doesn't exist

[kdt523]

Solution
Added two-layer protection in addAvailableLanguages():

1. Preventive check: Verify directory exists before attempting iteration
2. Exception handling: Catch filesystem errors during traversal

Changes

• Add std::filesystem::exists() and std::filesystem::is_directory() checks
• Wrap recursive_directory_iterator in try-catch block
• Function now returns gracefully with empty language list instead of crashing

Testing

• Verified fix handles missing directories without crashing
• Tested edge cases (empty paths, permission denied scenarios)

Impact

• After: Graceful degradation - returns empty language list and continues running

This fix improves robustness for users with incomplete Tesseract installations.

Contributing to Hacktoberfest 2025

[stweil]

How did you get the crash?

[stweil]

I get a exception which is handled:

% tesseract --tessdata-dir /missing  --list-langs
exception: filesystem error: in recursive_directory_iterator: No such file or directory ["/missing/"]

% tesseract --tessdata-dir file  --list-langs 
exception: filesystem error: in recursive_directory_iterator: Not a directory ["file/"]

Isn't it better to get such an error message instead of silently failing?

[stweil]

I think this code block is not needed. The iteration will raise an exception for both bases, and this exception is handled.

[stweil]

Suggested change

	// Check if directory exists before attempting to iterate
	if (!std::filesystem::exists(datadir) || !std::filesystem::is_directory(datadir)) {
	return;
	}

[kdt523]

Done Removed the existence check as suggested. The code now uses only the try-catch approach to handle filesystem errors, which is cleaner and still prevents the crash

[stweil]

"Permission denied" was already silently handled.

[zdenop]

@kdt523 : please check if #4372 solve your problem.

[kdt523]

How did you get the crash?

The crash was reported in a GitHub issue with a complete stack trace. I didn't personally reproduce it, but analyzed the code to understand the root cause:

The crash happens when:

User installs tesseract-ocr package without language data packages
The /usr/share/tessdata/ directory doesn't exist

[stweil]

How did you get the crash?

The crash was reported in a GitHub issue with a complete stack trace.

Are you referring to issue #4364?

[kdt523]

How did you get the crash?

The crash was reported in a GitHub issue with a complete stack trace.

Are you referring to issue #4364?
yes

[kdt523]

i made required changes could you please approve it

[egorpugin]

Wait.

What is the program that showed crash in #4364?
If it is not tesseract, that program is responsible for C++ exception handling.

From #4364

#37 0x00007ffff65032be in QCoreApplication::exec() () at /lib64/libQt5Core.so.5
#38 0x00005555555eb33c in launchGui(int, char**) [clone .constprop.0] (argv=<optimized out>, argc=<optimized out>) at /usr/src/debug/crow-translate-v3.1.0/src/main.cpp:77
#39 0x00007ffff5a2a2ae in __libc_start_call_main () at /lib64/libc.so.6
#40 0x00007ffff5a2a379 in __libc_start_main_impl () at /lib64/libc.so.6
#41 0x00005555555926c5 in _start () at ../sysdeps/x86_64/start.S:115

/usr/src/debug/crow-translate-v3.1.0/src/main.cpp:77

It is not libtesseract issue, you should handle C++ exceptions.

And it is known that Qt does not use C++ exceptions, so it might be something new for a programmer, but tess calls must be wrapped with try..catch in that user program.

[egorpugin]

In C++ with exceptions we consider every line "correct" without any double checks or paranoid checks.
Every previous line, if it is errorred, will throw exception and we won't get anything bad on the current line.