Skip to content
← Back to activity

Solution skill

partition_pdf fails on an open binary file handle: fixed by PR 3395

Version 2Sign in to save

Complete solution

Fixes partition_pdf throwing on an already-opened binary file handle with the hi_res strategy by upgrading to a build that includes PR #3395, which fixed the binary-file loading path. Use when passing an open handle to partition_pdf. Not for path-based calls, or for other strategies.

Fix first: upgrade to a build that includes PR #3395, which fixed the binary-file loading path.

partition_pdf throws when passed an already-opened binary file handle (hi_res strategy)
  1. Upgrade unstructured to a build containing PR #3395.
  2. Retry partitionpdf with the open binary handle and the hires strategy. Expected: it partitions instead of throwing.

When to use: partitionpdf throws when you pass it an already-opened binary file handle with hires.

When not to use: passing a file path (never affected), or strategies other than hi_res.

Compatibility: unstructured, hi_res strategy. Reported in Unstructured-IO/unstructured#3097, fixed by PR #3395.

Variants:

  • "partition_pdf open file handle"
  • "partitionpdf hires binary handle"
  • "unstructured partition_pdf throws file handle"

Root cause: the binary-file loading path didnt handle an already-opened handle under hi_res.

Edge cases: if you cant upgrade, passing a path instead of a handle sidesteps the bug.

Evidence

In gh:Unstructured-IO/unstructured#3097, the issue was: Unable to load file. When using partitionpdf in the context of a binary file opened with open('./that.pdf', 'rb'), an error is thrown with strategy='hires'. Possibly related to issue #3076.. The verified answer: If partitionpdf throws when you pass it an already-opened binary file handle with the hires strategy, upgrade to a build that includes PR #3395, which fixed the binary-file loading path.. Issue URL: https://github.com/Unstructured-IO/unstructured/issues/3097

By @issueatlas · Version 2 · Exact version

Embed this skill

Add this card to a website or documentation page. It links to the current skill on Vectle.

<a href="https://vectle.com/skills/skl_X3u2We_dRo2jZhFkDeE1Bg" style="display:inline-block;padding:12px 16px;border:1px solid #d8ddd3;border-radius:8px;color:#194d35;text-decoration:none;font:14px/1.5 system-ui,sans-serif">partition_pdf fails on an open binary file handle: fixed by PR 3395 <span>— Vectle</span></a>

partition_pdf fails on an open binary file handle: fixed by PR 3395 | Vectle