Upgrade StorageReader links hundreds of superscript/bracketed digits to two footnotes in a textbook with no citatio
I'm reading a 700-page math textbook PDF in the Zotero reader. The PDF has no embedded link annotations. My other PDF editor/viewers show nothing clickable. Yet every superscript or subscript 1 and 2 in equations (e.g. the 2 in x²), and every bracketed (1) or (2), is highlighted and turned into a clickable link that jumps elsewhere in the book.
Cause: the book has exactly two footnotes, numbered 1. and 2. Zotero's citation detection appears to treat them as the reference list, so every 1 and 2 across the entire book (hundreds of them, often hundreds of pages away) becomes a "citation" to those footnotes.
Other users have noticed this behaviour:
https://forums.zotero.org/discussion/121044/zotero-pdf-reader-incorrectly-interprets-superscript-numbers-in-equations-for-citations
https://forums.zotero.org/discussion/120973/erroneous-citation-preview-popup-with-equation-containing-1
https://forums.zotero.org/discussion/116479/disable-the-pop-up-window-when-hovering-over-citation-links
https://forums.zotero.org/discussion/comment/509828
*Request*
In lieu of having the ability to turn off this behaviour for a pdf object, two possible logic guards would have stopped this:
1. Minimum reference-list size. The extractor accepts any numbered group as a bibliography, even without a heading. Here, this is exactly two footnotes which are not even on the same page. A document with only a couple of numbered entries has no numeric citation style and perhaps shouldn't enter numeric matching at all.
2. Mention-to-reference ratio check. The matcher already counts how many distinct references the mentions resolve to (src/core/module/reference/matcher/number.js). Here that's 2, for 800+ mentions.
These would also fix the equation false positives in #121044 and #120973.
Cause: the book has exactly two footnotes, numbered 1. and 2. Zotero's citation detection appears to treat them as the reference list, so every 1 and 2 across the entire book (hundreds of them, often hundreds of pages away) becomes a "citation" to those footnotes.
Other users have noticed this behaviour:
https://forums.zotero.org/discussion/121044/zotero-pdf-reader-incorrectly-interprets-superscript-numbers-in-equations-for-citations
https://forums.zotero.org/discussion/120973/erroneous-citation-preview-popup-with-equation-containing-1
https://forums.zotero.org/discussion/116479/disable-the-pop-up-window-when-hovering-over-citation-links
https://forums.zotero.org/discussion/comment/509828
*Request*
In lieu of having the ability to turn off this behaviour for a pdf object, two possible logic guards would have stopped this:
1. Minimum reference-list size. The extractor accepts any numbered group as a bibliography, even without a heading. Here, this is exactly two footnotes which are not even on the same page. A document with only a couple of numbered entries has no numeric citation style and perhaps shouldn't enter numeric matching at all.
2. Mention-to-reference ratio check. The matcher already counts how many distinct references the mentions resolve to (src/core/module/reference/matcher/number.js). Here that's 2, for 800+ mentions.
These would also fix the equation false positives in #121044 and #120973.