feat(pacer): Updates logic to create error message.

[ERosendo]

While debugging another issue in the recap-fetch API, I noticed that our code was incorrectly identifying certain sealed documents on PACER. This led to a misleading error message being added to the fetch queue:

Unable to download PDF. PDF not served as binary data and unable to find iframe src attribute. URL: https://ecf.caed.uscourts.gov/doc1/03319001890, caseid: 300601, magic_num: None. court_id='caed'

To address this issue, I've added a new phrase to the sealed document check logic and refactored the if statement

Here's another example of a sealed document: https://ecf.vawd.uscourts.gov/doc1/19111488805

[ERosendo]

@mlissner I looked at other queues with the same error message to see if there was a common cause. I found these messages that we're not handling properly:

1. This document is SEALED (https://ecf.ncmd.uscourts.gov/doc1/13313991712)
2. No matter of public record has been filed (https://ecf.cacd.uscourts.gov/doc1/031112493036)
3. Cannot locate the case with caseid 602254

The first message is similar to the one we're fixing in this PR, but not exactly the same. I'll update the PR to handle it too.

For the second message, I can also add logic to handle that case in this PR. But what kind of error message should we show in this case?

We already have checks for the third message:

juriscraper/juriscraper/pacer/reports.py

Lines 260 to 267 in 14c85d0

	if b"Cannot locate the case with caseid" in r.content:
	# This document is from a different docket, but is included in
	# this docket. Probably a criminal case with the doppelganger
	# bug. Try again, but do so without the pacer_case_id.
	# This should work, but will omit the blue header on the PDFs.
	r, url = self._query_pdf_download(
	None, pacer_doc_id, pacer_magic_num, got_receipt="1"
	)

However, we don't currently check if the same error persists after a second attempt to retrieve the PDF. What kind of error message would be most helpful in this situation?

[mlissner]

I have no idea what "No matter of public record has been filed" means. I guess it means it's sealed? We could just pass the same message through to the user?

For the third error, I guess passing through the original message makes the most sense too?

[johnhawkinson]

I have no idea what "No matter of public record has been filed" means. I guess it means it's sealed? We could just pass the same message through to the user?

This language is intended to avoid the implication that there is a sealed document behind the link. Maybe it's a deleted document. Maybe it's a document number that has not been used. Maybe it's a document that has been restricted without being sealed (err?). And, of course, maybe it's an FRCP 5.2(c) document (usually "You do not have permission to view this document.")

[mlissner]

Great, in that case, feels like we should just use that error message too.

[ERosendo]

@mlissner Ready for review.

[mlissner]

OK, now we surely have all the error messages. :)