- Added pdfinfo support to determine number of PDF pages (requires custom binary)
- Automatic downloading and installation of customized Xpdf tools from zotero.org
Other changes:
- Various API Zotero.Fulltext API changes
- Prevent buggy automatic prefpane sizing
- New 'refresh' Notifier action to just reselect the currently selected item without sending a 'modify' trigger
- Zotero.File.putContents(file, str) -- can only handle ASCII text
- Zotero.isLinux property
National Bureau of Economic Research (nber.org)
NASA ADS (adsabs.harvard.edu)
SPIE Digital Library (spiedigitallibrary.aip.org)
ERIC (eric.ed.gov)
RePEc (ideas.repec.org)
Engineering Village (engineeringvillage2.org)
International Herald Tribune (iht.com)
updated translators:
InnoPAC: adds u texas austin compatibility and uc san francisco compatibility
Open WorldCat (Web): adds saved bibliography compatibility
the function is actually called copyCitationToClipboard, because it generates a single citation of multiple items if multiple items are passed to it. if this isn't the desired behavior, it's easy to change.
for some reason, passing the DOM window, rather than the URL, when creating the sandbox fixed this issue. i'm not entirely sure why this is, but it seems like the javascript on the washingtonpost.com site might be modifying document.domain in such a way that this is necessary.
- does not add shortTitle field to imported items (only items scraped via web)
- fixes issues with ibid in Chicago Manual of Style (Note with Bibliography)
- fixes issues with repeat parenthetical citations
I don't know this code at all, but here's a quick shot at fixing this problem. John, let me know if this is sufficient or if more needs to be done.
Simon, feel free to do something a bit more elegant (like clear missing items from the cache and the like)...
- The (empty) created collection is now deleted after import failure
- Sped up general DB performance dramatically (especially on XP and Linux) by using a dummy statement to keep the SQLite shared cache active between transactions -- this seems to have only a moderate effect on OS X, which was already much faster, perhaps because of its own disk cache (though running XP and Linux under Parallels with limited RAM probably didn't help)
- Added some additional sanity checks to Date.strToDate() to prevent errors in the metadata pane and on import with dates like "1998-99" -- strToDate() was treating 99 as the month, which caused an error in the stricter DB trigger. "99" is now left as an unparsed part, though "2006-08" (where the second part is <= 31) will still get parsed as August 2006. Can't do much about that, though.
- Fixed a bug when rolling back nested transactions that could temporarily leave things somewhat funky
