Auto-close any open BMC/BDC when ending text (bug 1898053)

[calixteman]

No description provided.

[calixteman]

/botio test

[moz-tools-bot]

From: Bot.io (Linux m4)

---

Received

Command cmd_test from @calixteman received. Current queue size: 0

Live output at: http://54.241.84.105:8877/d43ceb4eba5d4d0/output.txt

[moz-tools-bot]

From: Bot.io (Windows)

---

Received

Command cmd_test from @calixteman received. Current queue size: 0

Live output at: http://54.193.163.58:8877/a19b0d1062ef276/output.txt

[moz-tools-bot]

From: Bot.io (Linux m4)

---

Failed

Full output at http://54.241.84.105:8877/d43ceb4eba5d4d0/output.txt

Total script time: 27.95 mins

• Unit tests: Passed
• Integration Tests: FAILED
• Regression tests: FAILED

  different ref/snapshot: 24
  different first/second rendering: 2

Image differences available at: http://54.241.84.105:8877/d43ceb4eba5d4d0/reftest-analyzer.html#web=eq.log

[moz-tools-bot]

From: Bot.io (Windows)

---

Failed

Full output at http://54.193.163.58:8877/a19b0d1062ef276/output.txt

Total script time: 43.38 mins

• Unit tests: Passed
• Integration Tests: FAILED
• Regression tests: FAILED

  different ref/snapshot: 8

Image differences available at: http://54.193.163.58:8877/a19b0d1062ef276/reftest-analyzer.html#web=eq.log

[Snuffleupagus]

You need to protect against this becoming negative, note

pdf.js/src/core/evaluator.js

Lines 3426 to 3434 in b7b8e5e

	case OPS.endMarkedContent:
	flushTextContentItem();
	if (includeMarkedContent) {
	if (markedContentData.level === 0) {
	// Handle unbalanced beginMarkedContent/endMarkedContent
	// operators (fixes issue15629.pdf).
	break;
	}
	markedContentData.level--;

[Snuffleupagus]

This behaves subtly different than the code in getTextContent, which keeps the existing level-state when parsing /Form XObjects.

Perhaps that's not actually necessary to do, but my point here is that getOperatorList and getTextContent should handle markedContent-levels in /Form XObjects the same way (whatever that way is).

[timvandermeij]

Moreover, it looks like the FreeText Editor FreeText accessibility must check that the parent structTree id is correct integration test fails on this PR since it fails on both Linux and Windows and I haven't seen this one fail before (so while it could still be an intermittent failure it seems unlikely). We should check if that is indeed caused by this change or if it's a new intermittent.

[Snuffleupagus]

Having looked at a few real-world PDF documents recently it seems to be fairly common (e.g. with /ActualText entries) for BDC/EMC-blocks to be nested within each other, hence I unfortunately cannot imagine that just closing any open ones on end-text is actually correct in the general case.

[calixteman]

Yep, I think I've a fix for it, I'll try to do that this week.