Skip to content

Transforms complex documents like PDFs and Office docs into LLM-ready markdown/JSON for your Agentic workflows.

refactor(analysis): index pdf_page.size directly in text content fill

jinzhenj271d01d95464
diff --git a/mineru/backend/analysis/pdf/text/content.py b/mineru/backend/analysis/pdf/text/content.py
index e3a6e605..7bb1087f 100644
--- a/mineru/backend/analysis/pdf/text/content.py
+++ b/mineru/backend/analysis/pdf/text/content.py
@@ -393,7 +393,7 @@ def _fill_window_block_content_and_lines(
):
page_pil_image = image_dict["img_pil"]
render_scale = float(image_dict["scale"])
- page_size = tuple(float(value) for value in pdf_page.size)
+ page_size = (float(pdf_page.size[0]), float(pdf_page.size[1]))
page_spans = _build_page_text_formula_spans(
page_inline_formula_list,
page_ocr_res_list,