From 0e6429b6ab416de8f3ca0b3d49a65ba8c44df4a3 Mon Sep 17 00:00:00 2001
From: benoit74 <benoit74@users.noreply.github.com>
Date: Tue, 26 May 2026 15:46:09 +0000
Subject: [PATCH] Fix `format_for` to not modify input BytesIO position

---
 CHANGELOG.md                       |  1 +
 src/zimscraperlib/image/probing.py |  6 ++++++
 tests/image/test_image.py          | 19 +++++++++++++++++++
 3 files changed, 26 insertions(+)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index e41d8ca..f91b56d 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -22,6 +22,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Improve contribution setup instructions: use `hatch shell`, clarify commands must be run from local clone root (#153)
 - Bring coverage back to 100% (#293)
 - Fix `Creator.config_indexing` docstring to reflect that only the full-text index is toggled; title indexing is always performed by libzim (#294)
+- Fix `format_for` to not modify input BytesIO position (#296)
 
 ### Changed
 
diff --git a/src/zimscraperlib/image/probing.py b/src/zimscraperlib/image/probing.py
index 5f40874..6a8d263 100644
--- a/src/zimscraperlib/image/probing.py
+++ b/src/zimscraperlib/image/probing.py
@@ -68,6 +68,9 @@ def format_for(
 ) -> str | None:
     """Pillow format of a given filename, either Pillow-detected or from suffix"""
     if not from_suffix:
+        original_position = 0
+        if isinstance(src, io.BytesIO):
+            original_position = src.tell()
         try:
             with PIL.Image.open(src) as img:
                 return img.format
@@ -85,6 +88,9 @@ def format_for(
                 return "SVG"
             else:  # pragma: no cover
                 raise
+        finally:
+            if isinstance(src, io.BytesIO):
+                src.seek(original_position)
 
     if not isinstance(src, pathlib.Path):
         raise ValueError(
diff --git a/tests/image/test_image.py b/tests/image/test_image.py
index ed925cf..3b1174e 100644
--- a/tests/image/test_image.py
+++ b/tests/image/test_image.py
@@ -1058,6 +1058,25 @@ def test_format_for_cannot_use_suffix_with_byte_array():
         assert format_for(src=io.BytesIO(), from_suffix=True)
 
 
+@pytest.mark.parametrize(
+    "seek_to_zero",
+    [
+        (True),
+        (False),
+    ],
+)
+def test_format_for_does_not_alter_buffer(
+    svg_image: pathlib.Path, *, seek_to_zero: bool
+):
+    svg_bytes = io.BytesIO()
+    svg_bytes.write(svg_image.read_bytes())
+    if seek_to_zero:
+        svg_bytes.seek(0)
+    original_position = svg_bytes.tell()
+    assert format_for(svg_bytes, from_suffix=False) == "SVG"
+    assert svg_bytes.tell() == original_position
+
+
 def test_is_valid_image(
     png_image: pathlib.Path,
     png_image2: pathlib.Path,