Bugs reported by OSS-Fuzz

[stweil]

There are currently several open issues which were detected by OSS-Fuzz and which I could not fix in time: https://bugs.chromium.org/p/oss-fuzz/issues/list?q=Proj%3Dtesseract-ocr. One of them is classified as a security bug.

[jmechnich]

The link from above is missing a j, this one works (and just as a reference for building: the instructions are here).

[stweil]

Currently still open:

https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=32142 (bug security)
https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=32246 (bug)

Both should be fixed now, so let's wait for the next report from OSS-Fuzz.

[stweil]

OSS-Fuzz issue 32246 is verified as fixed.

Commit 91b2b4f now also fixes OSS-Fuzz issue 32142.

[stweil]

Commit 6d0595b fixed another OSS-Fuzz issue (33220).

[amitdo]

https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=33309&q=Proj%3Dtesseract-ocr

[stweil]

That one should be fixed for 5.0.0. The testcase shows a memory leak with 124 unreleased EDGEPT objects.

2021-11-02 It looks like the objects get lost in make_edgept.

[stweil]

https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=39185 is also a bug which has to be fixed.

It might be caused by a very small line_spacing = 2.08616257e-07 which gives an even smaller disp_quant_factor_ = 9.7788870334625244e-09. That causes an integer overflow in IntCastRounded(max_dist / disp_quant_factor_), and the resulting value is used to allocate memory.

[amitdo]

According to the report, the regression happened in one of these commits:

0401b94...4daf781

[stweil]

Commit 0401b94 was not better: it crashes early with a container overflow when running the same fuzzing test case. That was fixed in commit 91b2b4f.

[amitdo]

Probably this commit is the source of the regression: 57c90ee