#318 percent-encoded '?' in URLs problem

[Janusz]

This URL with %3F gives
Received http status 404 while retrieving https://djvu.szukajwslownikach.uw.edu.pl/IMPACT_GT_1/00426979.djvu%3Fdjvuopts&page=1&zoom=width&showposition=0.460,0.523&highlight=357,1207,1457,442.
Unexpected End Of File.
This URL opens correctly.
Tested with Debian buster and the git version.

[Dr Bill C Riemers]

Why would you expect the URL to work? The purpose of the % encoding characters is to prevent the characters from being interpreted with special meaning.

If you create a file on your server named "IMPACT_GT_1/00426979.djvu?djvuopts" how would you specify you wanted that file, we ignored the escape intended by using %3F instead of ? in the URL.

Some code tries to do what you mean, not what you say. Meaning you may well have some URL parsers that see the & and decide the question mark must be literal. But if you review the RFC's you'll find & has special meaning on the server, not in the client. So the argument should be passed to open a filename of: "IMPACT_GT_1/00426979.djvu%3Fdjvuopts&page=1&zoom=width&showposition=0..."

e.g. The full string.

The other component of a URL you see frequently absused with do as I mean not what it litterally say is anchors. You'll have websites that use anchors as cgi-bin arguments, even though an anchor is intended just to tag a location in the page for the web browser.

[Janusz]

Thank you very much for your quick reaction.
To be on the safe side:
The first link in my posting is
https://djvu.szukajwslownikach.uw.edu.pl/IMPACT_GT_1/00426979.djvu%3Fdjvuopts&page=1&zoom=width&showposition=0.460,0.523&highlight=357,1207,1457,442.
The second one is
https://djvu.szukajwslownikach.uw.edu.pl/IMPACT_GT_1/00426979.djvu?djvuopts&page=1&zoom=width&showposition=0.460,0.523&highlight=357,1207,1457,442
I still don't understand why djview reports an unexpected end of file.

[Janusz]

In a sense a nonexisting file has an unexpected end of file, but I would prefer a more explicit error message :-)
Anyway please close the ticket.

[Leon Bottou]

Bill is enterely right. Escaping the question mark means that you're looking for a document named '00426979.djvu?djvuopts'. Then djview reports a sequence of errors. First a 404 http error because the url goes nowhere. Then an end-of-file error because the decoder tries to decode a zero length document. Finally the abort message summarizes the findings:

djview: Received http status 404 while retrieving https://djvu.szukajwslownikach.uw.edu.pl/IMPACT_GT_1/00426979.djvu0.000000djvuopts&page=1&zoom=width&showposition=0.460,0.523&highlight=357,1207,1457,442.
djview: Unexpected End Of File.
djview: Cannot open URL 'https://djvu.szukajwslownikach.uw.edu.pl/IMPACT_GT_1/00426979.djvu0.000000djvuopts&page=1&zoom=width&showposition=0.460,0.523&highlight=357,1207,1457,442'.

[Leon Bottou]

Closing. This is not a bug but the normal behavior. URLs are confusing things.