The application does the following steps:
- check if OAuth Token is present in the session, if not - redirect to Google web OAuth
- check if Gmail Label name is present in the session, if not - fetch all labels and choose one on
/labels - fetch un-read emails under a given Label using Token though GMail API (using a search query).
- (optional) fetch un-read emails the same way (if enabled by
-read, only supported by CLI now) - extract paper mentions from each read email, sort and aggregate by paper title, count frequency of the duplicates
- (optional) fetch&extract the read emails (enabled by
-read) - render read papers using the tempates in one of the supported formats (JSONL, Markdown, HTML)
- (optional) render read emails (in separate "Archive" section, enabled by
-read) - (optional) mark all un-read emails that were used to produce this report as read on gmail (enabled by
-mark, only supported by CLI now)
The application data model consits of:
(one) email Message
- Subject (plain-text, user local language, follows one of the number of templates)
- Body (MIME encoded parts)
that contains, or produces
(many) Papers
- Title, URL, Abstract
- Author (only displayed if enabled by
-author, on by default on server) - Refs[] (
[{ID, Title}, ...]all emails that are "origins of the citation" or "sources, refering to" this paper) - Freq (citation frequency: a total number of Messages reffering to this paper)
The reference information in a paper is utilized to produce links to the original email messages on Gmail e.g. for user to mark them as "read" in manually, if no -mark is provided (wich is only supported by CLI)
Statistics in the header reports the following:
- datetime when the report was generated
- total number of unread email Messages that were used to generate it
- total number of Papers (by unique paper titles) cited/references in these unread emails