Unexpected C++ exception in tesseract::addAvailableLanguages

[jobermayr]

Current Behavior

$ LANG=en ls /usr/share/tessdata
ls: cannot access '/usr/share/tessdata': No such file or directory

#8 0x00007ffff7aad57b in std::filesystem::__cxx11::recursive_directory_iterator::recursive_directory_iterator
(this=0x7fffffffc2b0, __p=filesystem::path "/usr/share/tessdata/" = {...}, __options=(std::filesystem::directory_options::follow_directory_symlink | std::filesystem::directory_options::skip_permission_denied))
at /usr/include/c++/14/bits/fs_dir.h:514
#9 tesseract::addAvailableLanguages (datadir="/usr/share/tessdata/", langs=langs@entry=0x7fffffffc620) at src/api/baseapi.cpp:152
#10 0x00007ffff7aae2d0 in tesseract::TessBaseAPI::GetAvailableLanguagesAsVector (this=this@entry=0x7fffdc0021d8, langs=langs@entry=0x7fffffffc620) at src/api/baseapi.cpp:399

Expected Behavior

It doesn't crash.

Suggested Fix

https://github.com/tesseract-ocr/tesseract/blob/main/src/api/baseapi.cpp#L149:
Add check here whether path/file exists and return if not.

tesseract -v

No response

Operating System

No response

Other Operating System

No response

uname -a

No response

Compiler

No response

CPU

No response

Virtualization / Containers

No response

Other Information

No response

[stweil]

There is also an exception if /usr/share/tessdata is a file. Just try touch /usr/share/tessdata and run the test again.

[stweil]

@jobermayr, do you really get a crash? Or is it a C++ exception (which is not a crash)?

[zdenop]

@stweil: I get a crash on Windows (built with VS2017)

[jobermayr]

$ crow
terminate called after throwing an instance of 'std::filesystem::__cxx11::filesystem_error'
  what():  filesystem error: recursive directory iterator cannot open directory: No such file or directory [/usr/share/tessdata/]
Abgebrochen (Speicherabzug geschrieben)

(gdb) bt
#0  0x00007ffff5a9a25c in __pthread_kill_implementation () at /lib64/libc.so.6
#1  0x00007ffff5a414b6 in raise () at /lib64/libc.so.6
#2  0x00007ffff5a2891a in abort () at /lib64/libc.so.6
#3  0x00007ffff5eadc4d in ??? () at /lib64/libstdc++.so.6
#4  0x00007ffff5ebf28c in ??? () at /lib64/libstdc++.so.6
#5  0x00007ffff5ead7f5 in std::terminate() () at /lib64/libstdc++.so.6
#6  0x00007ffff5ebf518 in __cxa_throw () at /lib64/libstdc++.so.6
#7  0x00007ffff5eb4947 in ??? () at /lib64/libstdc++.so.6
#8  0x00007ffff7aad57b in std::filesystem::__cxx11::recursive_directory_iterator::recursive_directory_iterator
    (this=0x7fffffffc290, __p=filesystem::path "/usr/share/tessdata/" = {...}, __options=(std::filesystem::directory_options::follow_directory_symlink | std::filesystem::directory_options::skip_permission_denied))
    at /usr/include/c++/14/bits/fs_dir.h:514
#9  tesseract::addAvailableLanguages (datadir="/usr/share/tessdata/", langs=langs@entry=0x7fffffffc600) at src/api/baseapi.cpp:152
#10 0x00007ffff7aae2d0 in tesseract::TessBaseAPI::GetAvailableLanguagesAsVector (this=this@entry=0x7fffdc0020c8, langs=langs@entry=0x7fffffffc600) at src/api/baseapi.cpp:399
#11 0x00005555555ac9f0 in Ocr::availableLanguages (this=0x7fffdc0020a0) at /usr/src/debug/crow-translate-v3.1.0/src/ocr/ocr.cpp:42
#12 SettingsDialog::SettingsDialog (this=<optimized out>, parent=<optimized out>) at /usr/src/debug/crow-translate-v3.1.0/src/settings/settingsdialog.cpp:64
#13 MainWindow::openSettings (this=0x7fffffffd6c0) at /usr/src/debug/crow-translate-v3.1.0/src/mainwindow.cpp:293
#14 0x00007ffff6535672 in ??? () at /lib64/libQt5Core.so.5
#15 0x00007ffff7299b22 in QAbstractButton::clicked(bool) () at /lib64/libQt5Widgets.so.5
#16 0x00007ffff7299dea in ??? () at /lib64/libQt5Widgets.so.5
#17 0x00007ffff729b665 in ??? () at /lib64/libQt5Widgets.so.5
#18 0x00007ffff729b894 in QAbstractButton::mouseReleaseEvent(QMouseEvent*) () at /lib64/libQt5Widgets.so.5
#19 0x00007ffff7395cda in QToolButton::mouseReleaseEvent(QMouseEvent*) () at /lib64/libQt5Widgets.so.5
#20 0x00007ffff71e6b58 in QWidget::event(QEvent*) () at /lib64/libQt5Widgets.so.5
#21 0x00007ffff71a522e in QApplicationPrivate::notify_helper(QObject*, QEvent*) () at /lib64/libQt5Widgets.so.5
#22 0x00007ffff71ad4ba in QApplication::notify(QObject*, QEvent*) () at /lib64/libQt5Widgets.so.5
#23 0x00007ffff64fc188 in QCoreApplication::notifyInternal2(QObject*, QEvent*) () at /lib64/libQt5Core.so.5
#24 0x00007ffff71ab50e in QApplicationPrivate::sendMouseEvent(QWidget*, QMouseEvent*, QWidget*, QWidget*, QWidget**, QPointer<QWidget>&, bool, bool) () at /lib64/libQt5Widgets.so.5
#25 0x00007ffff72003b6 in ??? () at /lib64/libQt5Widgets.so.5
#26 0x00007ffff7203a3f in ??? () at /lib64/libQt5Widgets.so.5
#27 0x00007ffff71a522e in QApplicationPrivate::notify_helper(QObject*, QEvent*) () at /lib64/libQt5Widgets.so.5
#28 0x00007ffff64fc188 in QCoreApplication::notifyInternal2(QObject*, QEvent*) () at /lib64/libQt5Core.so.5
#29 0x00007ffff6982833 in QGuiApplicationPrivate::processMouseEvent(QWindowSystemInterfacePrivate::MouseEvent*) () at /lib64/libQt5Gui.so.5
#30 0x00007ffff69538cc in QWindowSystemInterface::sendWindowSystemEvents(QFlags<QEventLoop::ProcessEventsFlag>) () at /lib64/libQt5Gui.so.5
#31 0x00007ffff555f010 in ??? () at /lib64/libQt5WaylandClient.so.5
#32 0x00007ffff4510eb8 in ??? () at /lib64/libglib-2.0.so.0
#33 0x00007ffff4512ca8 in ??? () at /lib64/libglib-2.0.so.0
#34 0x00007ffff45134bc in g_main_context_iteration () at /lib64/libglib-2.0.so.0
#35 0x00007ffff6556f79 in QEventDispatcherGlib::processEvents(QFlags<QEventLoop::ProcessEventsFlag>) () at /lib64/libQt5Core.so.5
#36 0x00007ffff64fab82 in QEventLoop::exec(QFlags<QEventLoop::ProcessEventsFlag>) () at /lib64/libQt5Core.so.5
#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

Why should I touch not installed file/dir?

$ zypper se -i tesseract
Loading repository data...
Reading installed packages...

S  | Name                              | Summary                                     | Type
---+-----------------------------------+---------------------------------------------+--------
i  | libtesseract5                     | Open Source OCR Engine                      | package
i  | libtesseract5-debuginfo           | Debug information for package libtesseract5 | package
i  | libtesseract5-x86-64-v3           | Open Source OCR Engine                      | package
i+ | libtesseract5-x86-64-v3-debuginfo | Debug information for package libtesseract5 | package
i  | tesseract-ocr-debugsource         | Debug sources for package tesseract-ocr     | package

$ rpm -ql libtesseract5
/usr/lib64/libtesseract.so.5
/usr/lib64/libtesseract.so.5.0.5
/usr/share/licenses/libtesseract5
/usr/share/licenses/libtesseract5/LICENSE

 $ rpm -ql libtesseract5-x86-64-v3
/usr/lib64/glibc-hwcaps/x86-64-v3/libtesseract.so.5
/usr/lib64/glibc-hwcaps/x86-64-v3/libtesseract.so.5.0.5

There are also tesseract-data and some tesseract-ocr* but they thankfully aren't hard dependencies.

Rather, libtesseract has to react correctly and not leading to the crash if they aren't installed (correct error handling!).

[stweil]

$ crow
terminate called after throwing an instance of 'std::filesystem::__cxx11::filesystem_error'
what(): filesystem error: recursive directory iterator cannot open directory: No such file or directory [/usr/share/tessdata/]
Abgebrochen (Speicherabzug geschrieben)

So this is a C++ exception, not a crash. I'll update the issue title.

[stweil]

Why should I touch not installed file/dir?

There are two different issues with similar results:

1. There is an exception if there is neither a directory nor a file for tessdata.
2. There is an exception if there is a file instead of a directory for tessdata.

The touch command is one way to test the 2nd case.

[stweil]

Current results on macOS:

% touch /tmp/file
% tesseract --tessdata-dir /tmp/file --list-langs
exception: filesystem error: in recursive_directory_iterator: Not a directory ["/tmp/file/"]
% ~/src/github/tesseract-ocr/tesseract/bin/debug/clang,sanitize/tesseract --tessdata-dir /tmp/missed --list-langs
exception: filesystem error: in recursive_directory_iterator: No such file or directory ["/tmp/missed/"]

Current results on Linux:

$ touch /tmp/file
$ tesseract --tessdata-dir /tmp/file --list-langs
exception: filesystem error: recursive directory iterator cannot open directory: No such file or directory [/tmp/file/]
$ tesseract --tessdata-dir /tmp/missed --list-langs
exception: filesystem error: recursive directory iterator cannot open directory: Not a directory [/tmp/missed/]

These results are fine if we accept an exception for an invalid tessdata-dir.