Integration test coverage

[gunnsth]

Package e2etest implements integration tests with a small test corpus (externally provided). It currently has two tests:

• Passthrough, which loads a PDF with unidoc, and writes back out and optionally validates with ghostscript ensuring that no errors are added by unidoc.
• Allobjects which combs through all streams in the PDF and decodes, ensuring that there are no errors.

In the future we will add more integration test coverage and validation that the output is as expected. The current integration tests only ensure that there are no errors in loading, and outputing.

Added to Jenkinsfile to run in automatic builds and test coverage measurements. The test coverage is now obtained with

go test -coverprofile=coverage.out -covermode=atomic -coverpkg=./... ./...

which measures the test coverage of all packages on all other packages within the projects, i.e. includes unit and integration tests.

Running the e2etests integration tests requires setting environment variables:

• UNIDOC_GS_BIN_PATH should point to the executable binary for ghostscript
• UNIDOC_E2E_FORCE_TESTS set to "1" if the tests are required to run
• UNIDOC_PASSTHROUGH_TESTDATA should point to the folder containing test corpus files for passthrough test
• UNIDOC_ALLOBJECTS_TESTDATA should point to the folder containing test corpus files for allobjects test.

[codecov]

Codecov Report

Merging #329 into v3 will increase coverage by 18.2%.
The diff coverage is 70%.

@@            Coverage Diff             @@
##               v3     #329      +/-   ##
==========================================
+ Coverage   38.54%   56.75%   +18.2%     
==========================================
  Files         122      144      +22     
  Lines       23167    25098    +1931     
==========================================
+ Hits         8930    14244    +5314     
+ Misses      12956    10488    -2468     
+ Partials     1281      366     -915

Impacted Files	Coverage Δ
pdf/internal/e2etest/validate.go	70% <70%> (ø)
pdf/contentstream/draw/bezier_curve.go	91.01% <0%> (ø)
pdf/core/security/crypt/filter_aesv2.go	31.81% <0%> (ø)
pdf/contentstream/draw/utils.go	100% <0%> (ø)
pdf/core/security/crypt/filters.go	45.45% <0%> (ø)
pdf/annotator/field_appearance.go	0% <0%> (ø)
pdf/annotator/line.go	0% <0%> (ø)
common/license/util.go	25% <0%> (ø)
pdf/internal/testutils/testutils.go	69.07% <0%> (ø)
... and 116 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 1dd3de8...18bacd0. Read the comment docs.

[adrg]

Remove empty line for consistency.

[adrg]

The package does work as intended when setting all the environment variables. The code looks great to me. I've added a couple of minor changes and suggestions.

The only other suggestion would be to maybe add a function similar to the readPDF function in unicli (https://github.com/unidoc/unicli/blob/master/pdf/utils.go#L18) so that we don't repeat code when reading a PDF file (read, decrypt, get number of pages etc...). This is not really related to this PR because this function would be added in another package in unidoc but would improve the code here as well. Just something to take into consideration for the future.

[gunnsth]

@adrg I have addressed all review comments, please take another look.

[adrg]

@gunnsth All the review points have been addressed. Tested again after pulling the new commits. Everything works as intended. 👍