From 1721393b971453bce9625e5654e997f5e37a22ca Mon Sep 17 00:00:00 2001
From: Hanmiao Li <894876246@qq.com>
Date: Mon, 8 Jun 2026 15:27:14 +0800
Subject: [PATCH] fix(pdf): use extract_text_by_pages to avoid hang on full-PDF
 reads

`pdf_extract::extract_text` uses an internal codepath that can hang on
certain PDF cross-reference tables or font encodings. The per-page
`extract_text_by_pages` path does not trigger this hang and produces
identical output when joined.

When `pages` is not specified, route through `extract_text_by_pages`
and join all pages instead of calling `extract_text`.

Fixes #2641.
---
 crates/tui/src/tools/file.rs | 18 ++++++++++++------
 1 file changed, 12 insertions(+), 6 deletions(-)

diff --git a/crates/tui/src/tools/file.rs b/crates/tui/src/tools/file.rs
index 671f1366..084c05c8 100644
--- a/crates/tui/src/tools/file.rs
+++ b/crates/tui/src/tools/file.rs
@@ -361,12 +361,18 @@ fn read_pdf_via_pdf_extract(
             }
         }
     } else {
-        pdf_extract::extract_text(path).map_err(|e| {
-            ToolError::execution_failed(format!(
-                "pdf-extract failed on {}: {e} (set `prefer_external_pdftotext = true` in settings.toml to retry via pdftotext)",
-                path.display()
-            ))
-        })?
+        // Call extract_text_by_pages even when the caller wants every page:
+        // extract_text uses an internal codepath that can hang on certain PDF
+        // cross-reference tables or font encodings (#2641). The per-page path
+        // avoids that hang and produces identical output when joined.
+        pdf_extract::extract_text_by_pages(path)
+            .map(|pages| pages.join("\n"))
+            .map_err(|e| {
+                ToolError::execution_failed(format!(
+                    "pdf-extract failed on {}: {e} (set `prefer_external_pdftotext = true` in settings.toml to retry via pdftotext)",
+                    path.display()
+                ))
+            })?
     };
     Ok(ToolResult::success(clean_pdf_text(&text)))
 }