issues Search Results · language:Dune language:Python language:JavaScript language:C++ language:Java language:HTML
Filter by
69.1M results
Hi,
I recently discovered OrderFlow and would be interested in contributing to the project.
The idea behind the project is interesting, and I believe it has the potential to become a useful learning ...
Summary
When a conversation is exported as ShareGPT or CSV via Prompt Storage and re-imported, messages that contain thinking
blocks or tool call blocks do not render correctly. Instead of the styled ...
Bug (security — content-triggered arbitrary file deletion)
PDFReader/PDFWriter accept a parsed-document .json file as a documented input. Image elements carry a
metadata.file_path, which is therefore ...
bug
Problem
The eng-team director currently derives what I manage from whatever survives in session records and its own transcript.
On 2026-06-10 this resurrected the pre-0.12.0 topology: it replayed three ...
Bug
The native-schema to_markdown never splits paragraphs and leaks literal \r into output, because pdfium emits \r\n line
endings.
get_text_bounded() returns text with \r\n (verified: raw repr ... line ...
bug
Summary
A Metal vertex shader that reads camera.invViewMatrix[3].xyz from a constant-buffer argument translates to SPIR-V
assembly with unresolved type/member diagnostics. The shader assembles, but the ...
Bug
Encrypted / password-protected PDFs raise a raw backend exception from every engine, with no datagrunt-level error and
no way to supply a password.
An AES-256 encrypted PDF: PDFReader(f, engine=X).to_dicts() ...
bug
Bug
Markdown writers emit extracted body text verbatim, so document text that happens to start with a markdown metacharacter
(#, -, *, , 1., etc.) is reinterpreted as markdown structure.
A PDF containing ...
bug
Bug
Page classification is inverted: pure-text pages are labeled mixed, and text_only is effectively unreachable.
A simple text-only PDF (body_text elements, no images/tables) → classification == mixed ...
bug
Bug (violates the 100%-extraction contract)
When OCR fails (e.g. the tesseract binary is missing), the entire page is discarded — including text objects and images
that were already successfully extracted. ...
bug

Learn how you can use GitHub Issues to plan and track your work.
Save views for sprints, backlogs, teams, or releases. Rank, sort, and filter issues to suit the occasion. The possibilities are endless.Learn more about GitHub IssuesProTip! Restrict your search to the title by using the in:title qualifier.