The error is triggered upon initial interaction with a doc after Zotero
restart or if new external citations (copied into the document) are
peresnt and `session.updateSession()` is called without a subsequent
`session.updateDocument()` call. `session.updateSession()` is called
without a subsequent `session.updateDocument()` call every time the
user cancels a citation insert.
More specifically, `session.updateSession()` is called every time a
citation dialog is invoked. It retrieves all citations and writes them
into a local `session.citationsByIndex` object. Moreover, it marks
each citation that hasn't seen before in a `session.newIndices` object.
`session.newIndices` is there to ensure that we load every new citation
into citeproc upon document update. This object is built by marking any
citation that does not appear in the previous invocation's list of
citations as new. However, if the document is never updated (because the
user cancels the insertion) then the new indices are not loaded
into citeproc. This commit fixes that, by excluding citeproc unloaded
items from the previous invocation's citation list.
- Use react-virtualized to render tags on demand, reducing the number
of DOM elements from potentially tens of thousands to <100. This
requires tags to be absolutely positioned, so sizing and
positioning need to be precomputed rather than relying on CSS.
- Avoid unnecessary refreshes, speed up tag retrieval, and optimize
sorting
- Debounce reflowing when resizing tag selector
Also:
- Scroll to top when changing collections
- Allow tags to take up full width of tag selector without truncation
Closes#1649Closes#281
Due to a typo in d0f7fd6df7, linked files were still being renamed even
with the pref off if metadata was found for the file. The test I added
was only for adding a file to an existing item, which didn't trigger
metadata retrieval.
This also adds a hook for stubbing the actual PDF recognition process so
we can test certain behaviors without making HTTP requests.
It seems like the Zotero.Utilities.debounce() on handleSearch() in
tagSelector.jsx was somehow causing the function to be run without being
triggered from the onSearch events, resulting in an extra render. I'm
not sure why that was happening, but it's fixed now that there's no
longer a debounce() there.
And take an optional second parameter in waitForTagSelector() to
indicate how many updates to wait for, since certain operations trigger
two updates, one from notify() and the other from onItemViewChanged().
The test incorrectly contained an item wait for an attachment that
doesn't get saved to the group library, but the test was passing because
the main item was being moved to the group twice. 94ccba45b somehow
fixed that, and since the behavior during the test is now correct, I'm
not going to worry about this unless we notice a problem.
This adds selectItems() to ZoteroPane and collectionTreeView and removes
the ancient, unused 'expand' argument to selectItem(), which didn't
really make sense there. It also includes a new
itemTreeView::ensureRowsAreVisible() that tries to scroll to an
appropriate place (or, better yet, not scroll at all) given the
specified rows and page size.
- Fix incorrect results for ANY search with multiple "Attachment
Content" conditions and no other conditions
- Dramatically speed up single-word searches by avoiding unnecessary
text scans (which probably addresses #1595)
- Clean up code
- Added icon-button UI code for the menubutton
- Upgrade to React 16 to allow non-standard attrs, such as `tooltiptext`
to support XUL tooltips
- Add i18n support for React UI elements
- Update tests for reactified tag selector
This is hard to do currently because the natural place to do it (and
where the previous seeAlso stuff was done) is translate_item.js, but
with async import translators that now only gets one item at a time,
whereas saving item relations requires all items to be saved. So this
would probably need to be done in the import code in translate.js.
It might also require undoing
https://github.com/zotero/zotero/pull/453 so that getResourceURI() works
on notes and figuring out another solution for the problem that was
trying to solve.
Invalid paths, including Windows UNC paths on other OSes, caused exports
to fail. Now they're ignored, which is what we do for other missing
attachment files.
Fixes#1622
This fixes direct and VPN-based retrieval of PDFs for Elsevier (e.g.,
ScienceDirect) items that have a DOI but no URL, since Elsevier resolves
DOIs through an intermediate page.
Delay requests to the same domain by 1 second, respect a Retry-After
header if present for 429 and 503, and delay for 10 seconds on 429 or
5xx otherwise.
Currently only .status and .getResponseHeader() (for getting 'Location')
are available in the returned object, but we could make the body
available if necessary.
If there's no translated PDF or the translated PDF fails and the item
has a DOI, check Zotero's Unpaywall mirror for possible sources and try
to download one of those.
Unlike with "Add Item by Identifier" and "Find Available PDF" in the
item context menu, this does not try the DOI/URL page, since it would
result in more data leakage and most of the time you'd be saving from
the DOI page already. We could consider offering it as an option, but
for it to be useful, you'd have to have an institutional subscription,
be on-campus or connected via VPN (for now), and be saving from
somewhere other than the main page.
A new connector endpoint, sessionProgress, takes the place of
attachmentProgress. Unlike attachmentProgress, sessionProgress can show
new attachments that have been added to the save, and with a little more
work should also be able to show when a parent item has been recognized
for a directly saved PDF.
This also adds support for custom PDF resolvers, available to all PDF
retrieval methods. I'll document those separately.
Closes#1542
- Add the ability to extract a PDF URL from a given webpage using the
translation framework
- Add the ability to get open-access PDFs from landing pages from
Unpaywall data in addition to direct PDF URLs
- Use the above functionality to improve PDF retrieval for "Add Item by
Identifier"
- Add "Find Available PDFs" option to the item context menu to retrieve
PDFs for existing items from the DOI or URL page or using Unpaywall
data. The option appears for single items with a DOI or URL and no PDF,
and it always appears when selecting multiple top-level items (but
skips ineligible items).
PDF extraction from DOI/URL pages will currently only work with
unauthenticated access (i.e., on-campus or VPN, but not via a web-based
proxy).
Supersedes and closes#948
Use `getResource` in Zotero.Date.init (this turns it into a
synchronous function). Zotero.File.getResource makes it easier
to load local files on platforms that do not support the
`resource://` URLs.
When the associated-files pref is enabled, Add Item by Identifier uses a
Zotero Unpaywall mirror to find available open-access PDFs. No details
about the contents of searches are logged.
The progress percentage is based on the most recent transaction
(or undeterminate if this is the first session transaction)
Fix undefined function call error
Move an item and its attachments to another library. Attachments are
removed as necessary if linked files or all files aren't supported in
the target library.
Add newly added attachments to a queue, start processing it after five
seconds have passed since the last attachment was added, and process
another every half second after that unless another is added.
This queue won't survive a restart, so the queue should really be in the
DB, but this should avoid problems when adding multiple attachments at
once.
Addresses #1284
Applies to dragging to the collections pane or the items pane, adding
via New Item menu, or saving via the connector server
If the renaming pref is enabled, the PDF is renamed after recognition.
Can be disabled in the preferences
Closes#917
Do cleanup on 'unload' rather than 'close' (which is limited to a click
on the close button and doesn't get called for win.close()) and clear
the queue after each test.
Automatic renaming is now done for dragging of an external file onto an
item (as added in 7cb95f41) as well as dragging as a linked file,
dragging a non-native (not text or HTML) URL, "Attach Link to File…",
"Attach Stored Copy of File…", and "Retrieve Metadata for PDF". It only
applies if a single file is being added and if the parent item has no
non-HTML attachments. By default, the renaming only applies to PDFs, but
this can be changed with the renameAttachmentFiles.automatic.fileTypes
hidden pref.
A new General pref, "Automatically rename attachment files using parent
metadata", controls whether the renaming happens.
Files saved via web translators are renamed regardless of this pref,
because they would often be gibberish otherwise.
Closes#113
Check file-editing access for the group from the API before offering to
reset, update the filesEditable setting properly, and restart the sync
automatically after resetting.
If a transaction took over 30 seconds and another transaction timed out
waiting for it, the second transaction would reset the notifier queue,
but if the first transaction then tried to queue an event, it would fail
with this error and roll back. (It would be nice to figure out why
transactions are taking over 30 seconds, though.)
- Updates /saveItems and /saveSnapshot to take a sessionID
- Provides a list of editable collections in the current library
- Adds an /updateSession method that takes a sessionID and updates the
collection and tags of any items saved via that operation (and changes
the currently selected collection)
Cross-library changes are not yet supported
This is a simplified version of the fix from #872. Unlike the proposal
in #36, this doesn't require all child items to be selected, since in a
search some children might be grayed out. If the child of an unselected
parent item is included, the drag isn't allowed.
Closes#36
- Moves a bunch of citation related processing from Integration.Session
- Replaces missing item handling with a function instead of exception
- Solves some really confusing flow issues in _processFields
Restores the "Restore to Zotero Server" functionality, now using the
API:
1. Get all remote keys and send `DELETE` for any that don't exist
locally.
2. Upload all local objects in full (non-patch) mode using only library
version so that the remotes are overwritten.
3. Reset file sync history, causing all files to be uploaded (or, more
likely, reassociated with existing remote files).
Since these are treated as regular updates on the server, they'll sync
down to other clients normally. Unsynced changes by other clients might
still trigger conflicts.
This and Reset File Sync History can also now be run on group libraries,
with a library selector in the Reset pane (which I forgot to do with
React).
The full sync option is now removed from the Reset pane, since there
wasn't ever really a reason to run it manually.
We should be able to reimplement Restore from Online Library (#1386)
using the inverse of this approach.
Closes#914
A XUL one for the current use in Advanced Search and an HTML one for
future uses. Sets the value to libraryID and adds data attributes for
editable/filesEditable on the HTML one.
In advance of #1356
We're not properly handling DOIs in parentheses or brackets (which would
require non-regex logic), so those tests are skipped for now.
Use Atom namespace when getting fields, and use `<updated>` date before
`<published>`. (The dates are also available on the nsIFeedContainer
(`feedEntry`), but we're getting them directly from the fields for some
reason.)
- Return `undefined` instead of throwing an error trying to access
`libraryTypeID` on a Zotero.Feed -- this fixes a test failure with
the latest Chai, which annoyingly runs inspect() on an object passed
to .include() regardless of whether the test succeeds
- Make some deprecated properties non-enumerable to avoid unnecessary
logging when the object is dumped
When an item is created, an active quick search is cleared, but that's
now an async operation. We weren't waiting for that, which meant that
new items weren't selected and depending on a race condition could even
show the welcome pane despite there being items in the library.
- Move identifier detection to `Zotero.Utilities.Internal.extractIdentifiers()`
so that it can be used for things other than Add Item by Identifier
(e.g., translation-server)
- Add a `Zotero.Translate.Search::setIdentifier()` function that takes an
identifier object produced by `extractIdentifiers()` (`{ DOI: "10/..." }`),
converts that to the search format expected by translators, and calls setSearch()
processDocuments() now uses an XHR 'document' request, wrapped to
provide a 'location' property, and uses promises for a simpler call
signature (though the old one will continue to work, for existing
translators). 'done' and 'exception' can now be handled via promises,
and in the translator sandbox an optional noCompleteOnError argument
instructs it not to automatically cancel the translation process with an
error (e.g., for supplementary materials).
Since we do need a hidden browser in some situations (e.g., for saving
snapshots), the old hidden-browser-based processDocuments() is still
available as Zotero.HTTP.loadDocuments().
This hopefully also fixes various problems with document property access
in translation-server.
Return a 500 for read-only libraries for all save modes. Read-only views
within editable libraries will save to the library root.
Addresses #185, RIS/BibTeX interception to read-only view behaves
differently from save button
The items will still match full-text word searches, but they won't match
phrase searches (because those require cache files for non-text
attachments) and the full-text won't sync to other computers, so they
should really be reindexed.
This should be tested, but we run tests in Firefox, and this doesn't
exist in Firefox... Easiest option is probably to add the submenu to the
Firefox menus for the purposes of testing.
If this works out I think we'll want to use this approach for
all data layer changes.
Previously, an unsaved change on an object would update its state
immediately, which was fine for synchronous code but breaks down if a
save involves multiple asynchronous calls, because modifying state after
the relevant data has been saved to the DB but before the `_changed`
object has been cleared would mean that new changes would be lost. Now,
changes are written to _changedData, and a get for the data first checks
_changedData before checking the state property (e.g., _tags) directly.
The changedData property is cleared as it's written, and once the object
is saved, the reload updates the state property with the new data.
If a standalone attachment existed in a collection and then was added to
a parent (e.g., via Create Parent Item), and attachment metadata was
also changed at the same time (e.g., due to file syncing), the
'collection item must be top level' trigger could throw on another
syncing computer. To work around this, remove collections first, then
make changes to the parentItemID columns, and then add new collections.
I think it might be worth having a tag management window that lets you
view tags as a grid, sort by column (e.g., type), select ranges, delete,
consolidate, etc., but until then, this fulfills a popular request.
And use them in new importTextAttachment() and importHTMLAttachment()
test support functions. These can be used to avoid needing a hidden
browser for determining the character set of the imported text
documents.
