Stable OCR snapshots
Archived versions of the OCR editor. Saved .chaya files use versioned JSON formats.
OCR experiments
-
Split chunks — unfinished
Previews horizontal or vertical chunk splits in the OCR editor. Done only reports the selected split; it does not change the document. Uses the 2024.03 file format.
Bookchop
Mark and label regions manually or with AI assistance. Both interfaces save interchangeable
.chaya ZIP packages containing the PDF and annotations. These differ from the OCR editor's JSON files.
-
Mark and read
Draw, move, resize, and label regions, then switch to Read to view the selected snippets.
-
Marking workspace
A simplified interface focused on marking regions.
Early Bookchop prototypes
To try the original workflow, use the fit-to-width annotator to label PDF regions and export JSON. Open the snippet viewer with the same PDF and that JSON.
-
Annotator: fit to width
Fits pages to the available width and exports regions in PDF coordinates.
retry-v2-pdfcoord-v3.html -
Snippet viewer
Displays labeled snippets from a PDF and annotations saved in PDF coordinates.
retry-reader.html -
PDF coordinates, second revision
Uses a fixed render scale, with a fix for resetting the viewport cache.
retry-v2-pdfcoord-v2.html -
PDF coordinates, first revision
Historical snapshot with a viewport-cache reset bug; use a later revision to annotate.
retry-v2-pdfcoord.html -
Original annotator: canvas coordinates
Exports canvas pixel coordinates. Its JSON format cannot be used with the snippet viewer.
retry.html
Bookchop and the early prototypes load browser libraries from CDNs and need an internet connection.