search does not work for books whose name contains an underscore followed by 4 digits

[mangtronix]

Search does not return any results, or even "no results" for this book.

http://www.archive.org/stream/TheZenithYearbook1993HighPointUniversity/THE_ZENITH_1993

[mangtronix (mang)]

Search also does not work for this book: http://www.archive.org/stream/ANestableTaperedColumnConceptForLargeSpaceStructures/19760022270_1976022270#search/shuttle

The OCR txt and xml files both have "Shuttle" in them.

[mangtronix (mang)]

For the NASA book it looks like the problem is the regex for matching thing page value in the search results.

NASA book:
<PARAM name="PAGE" value="19760022270_1976022270_0000.djvu"/>

Other book:
<PARAM name="PAGE" value="armageddonafter00couruoft_0001.djvu"/>

Regex code:
            var re = new RegExp (/_(\d{4})/);
            var reMatch = re.exec(pages[i].getAttribute('file'));

[mangtronix (mang)]

Fixed for both books by tightening up the regex.

[mangtronix (mang)]

Please check that search works as expected for the following books:

1. http://www-testflip.archive.org/stream/ANestableTaperedColumnConceptForLargeSpaceStructures/19760022270_1976022270

2. http://www-testflip.archive.org/stream/TheZenithYearbook1993HighPointUniversity/THE_ZENITH_1993

3. A random selection of books from www-testflip.archive.org

There shouldn't be any cross-browser differences possible from this fix, so testing two browser/OS combinations should be sufficient.

Just so you know, the OCR text is available from the "All files: HTTP" link and the file names are identifier/bookname_djvu.xml and identifier/bookname_djvu.txt. These are also available via our /download/ links, e.g.:
http://www.archive.org/download/ANestableTaperedColumnConceptForLargeSpaceStructures/19760022270_1976022270_djvu.txt
http://www.archive.org/download/ANestableTaperedColumnConceptForLargeSpaceStructures/19760022270_1976022270_djvu.xml

[mangtronix (mang)]

Added bug report on searching for numbers not working: https://bugs.edge.launchpad.net/bookreader/+bug/540433

[Winnie (winnie-archive)]

regressed search for alpha characters (words). fixed.

[mangtronix (mang)]

Awesome you rock! :)