From 01066f0b6eed5c43552ca58f7292e283db66eece Mon Sep 17 00:00:00 2001 From: Christoph Auer Date: Fri, 18 Jul 2025 16:48:35 +0200 Subject: [PATCH] Unload doc backend Signed-off-by: Christoph Auer --- docling/pipeline/threaded_standard_pdf_pipeline.py | 3 +++ tests/test_threaded_pipeline.py | 2 +- 2 files changed, 4 insertions(+), 1 deletion(-) diff --git a/docling/pipeline/threaded_standard_pdf_pipeline.py b/docling/pipeline/threaded_standard_pdf_pipeline.py index d3ea2f82..6c734bd9 100644 --- a/docling/pipeline/threaded_standard_pdf_pipeline.py +++ b/docling/pipeline/threaded_standard_pdf_pipeline.py @@ -527,6 +527,9 @@ class ThreadedStandardPdfPipeline(BasePipeline): for st in ctx.stages: st.stop() ctx.output_queue.close() + + if conv_res.input._backend is not None: + conv_res.input._backend.unload() return conv_res # -------------------------------------------------------------- feed_pages diff --git a/tests/test_threaded_pipeline.py b/tests/test_threaded_pipeline.py index b431c508..bc6fb4e5 100644 --- a/tests/test_threaded_pipeline.py +++ b/tests/test_threaded_pipeline.py @@ -18,7 +18,7 @@ from docling.pipeline.threaded_standard_pdf_pipeline import ThreadedStandardPdfP def test_threaded_pipeline_multiple_documents(): """Test threaded pipeline with multiple documents and compare with standard pipeline""" - test_files = [ + test_files = [str(f) for f in Path("tests/data/pdf").rglob("*.pdf")] or [ "tests/data/pdf/2203.01017v2.pdf", "tests/data/pdf/2206.01062.pdf", "tests/data/pdf/2305.03393v1.pdf",