[API] ALTO renderer does not work if no input name is set.

[sunoru]

Environment

• Tesseract Version: 5.1.0
• Commit Number: tag 5.1.0
• Platform: Linux 5.16.18-200.fc35.x86_64 #1 SMP PREEMPT Mon Mar 28 14:10:07 UTC 2022 x86_64 x86_64 x86_64 GNU/Linux

Current Behavior:

There would raise a segmentation fault if I use the following APIs (in pseudo code):

api = TessbaseAPICreate()
TessBaseAPIInit3(API, nullptr, nullptr)
renderer = TessAltoRendererCreate(output_file)
TessResultRendererBeginDocument(renderer, "")
TessBaseAPISetImage2(api, image)

Expected Behavior:

No segmentation fault should be raised. The file name would be left empty in the output.

Suggested Fix:

The segmentation fault is caused by strlen(nullptr), so maybe we should check if input_file_ is nullptr before doing AppendString (here)? It could also be fixed if the API user calls TessBaseAPISetInputName every time before TessResultRendererBeginDocument, but I still think this should be counted as a bug.

BTW, it worked for tesseract v4. #2700 is where a related breaking change was introduced.

[sunoru]

Oh, actually an empty input_file_ would also cause a segmentation fault, because api->GetInputName() returns nullptr for it. It can be easily fixed by checking input for AppendString.