Crossword Compiler: successful and non-successful scraping

[arelkin]

Two different sites featuring Crossword Compiler. One successful, one not.

I suspect it could have to do with the version of CCW being used. The successful scrape gathers from CCW with copyright 2021, while the unsuccessful scrape is from CCW with copyright 2015.

Error: https://www.washingtonexaminer.com/crossword-mind-games

Success: https://crosswordsbackwards.com/backward-crossword-puzzle-380/

[arelkin]

Actually an amendment to this, it may be just that one puzzle that has some sort of error. Another, earlier, puzzle from Washington Examiner was successfully scraped:

https://www.washingtonexaminer.com/crossword-empty-shelves

Here is a list of all puzzles: https://www.washingtonexaminer.com/search-result?q=CROSSWORD

[jpd236]

Yeah, this specific puzzle does something different with the JPZ that the parser didn't know how to handle:

      <word id="12" x="10-15" y="5" solution="[redacted]">
        <cells x="1-10" y="6"/>
      </word>

We need to support ranges in the cells tag as well. This should hopefully be rare (in years of downloading puzzles from various sources, I must not have come across a puzzle that did this).

[jpd236]

Should be fixed in the next release. Thanks for the report!

[arelkin]

Thanks for all you hard work on this plug-in.

[arelkin]

By the way, there are some unusual grids out there that make use of squares/lights that are larger than 1x1. Here are two examples:

https://www.xwordinfo.com/Crossword?date=4/4/2013 https://www.xwordinfo.com/Crossword?date=9/6/2012

[jpd236]

PUZ/JPZ doesn't support large squares - and the NYT applet (that we scrape from) doesn't either. Not much we can do there except match what the applet is doing.

[arelkin]

Very true, but the PDF could still be successfully created!

[jpd236]

I don't see a way to do this automatically, short of just pointing to the PDF provided by the NYT directly instead of generating our own. The problem is that the embedded puzzle data doesn't provide (to my knowledge) any signal that the squares are large, so we have no way of knowing that there's anything special going on in the scraper. It just circles them (or in some cases, does nothing at all) and puts a note that the printed version is different.

[arelkin]

I didn't mean to imply about scraping NYT puzzles. I understand they are behind a paywall.

I was only using those two examples because they also had unusual grid structure, just like what you discovered with the scrape error I initially submitted.

I was also suggesting that, if some grids show a scraping error for PUZ or JPZ, the Scraper could still offer just the PDF option.