ae2cc9119b
I've started to look into how we can fix issue 7580, but quickly became worried that fixing it could easily mean that we'd trade one fixed PDF file for a multitude of broken ones. Hence I started going through the history of the code that choose the fallback encoding, and noticed that it has been changed a number of times over the years to deal with various cases of weirdness/errors in non-embedded fonts. To my relief it turned out that almost all the PRs, please see a possibly incomplete [list here], that changed this code actually included `eq` test-cases. However, in one case it appears that a PR missed to add a test-case. Furthermore since the fallback encoding may also be the only source for creating a `toUnicode` map, changing the encoding could possibly regress only the text-selection despite a PDF file still rendering correctly. Therefore, this PR adds one new `eq` test, and also a number of additional `text` tests for PDF files already present in the test-suite. Note that it's obviously possible that there's a certain overlap between the added tests, but I'd be *a whole lot* more concerned with causing regressions. |
||
|---|---|---|
| .. | ||
| chromium | ||
| features | ||
| font | ||
| pdfs | ||
| resources | ||
| stats | ||
| ttx | ||
| unit | ||
| .gitignore | ||
| annotation_layer_test.css | ||
| downloadutils.js | ||
| driver.js | ||
| test_manifest.json | ||
| test_slave.html | ||
| test.js | ||
| testutils.js | ||
| text_layer_test.css | ||
| webbrowser.js | ||
| webserver.js | ||