pyzotero

Create a Zotero instance with your library ID, library type ('user' or 'group'), and API key from zotero.org/settings/keys. Then call methods like zot.items() or zot.top() to retrieve references, using zot.everything() to fetch all results.

Call zot.add_parameters(format='bibtex') before a read call such as zot.top(limit=50). The result is a bibtexparser BibDatabase object whose entries can be written to a .bib file with bibtexparser.dump().

Yes, local mode connects to a running Zotero desktop app without an API key by passing local=True when creating the Zotero instance. Local mode is read-only, so write operations like creating or updating items are not supported.

Iterate items with zot.collection_items(), fetch each item's children with zot.children(), filter for attachments with contentType 'application/pdf', and save each file using zot.dump() with the attachment key and output path.

TooManyRequests means the Zotero API rate limit was hit; retry with exponential backoff. PreConditionFailed indicates a version conflict from optimistic locking, so re-fetch the item and apply your changes to the fresh copy before updating.

Read calls return at most 100 items per request, and write batches are limited to 50 items per call. Saved search results cannot be retrieved via the API, only their metadata, and attachment upload methods are still in beta.