From e7d9b60737c8ffd21d3be5651d84bdf07a11b3d8 Mon Sep 17 00:00:00 2001
From: benoit74 <benoit74@users.noreply.github.com>
Date: Fri, 16 Oct 2026 11:14:02 +0000
Subject: [PATCH] Do not silently discard errors raised while finishing a ZIM

Creator.finish() caught and dropped every RuntimeError raised by
libzim's finishZimCreation(), including genuine finalization failures,
so a scraper could exit successfully (code 0) with no final ZIM file
and no error logged. Track whether finish() already ran instead, so a
second call stays a no-op but a real error on the first call now
propagates.
---
 CHANGELOG.md                     |  1 +
 src/zimscraperlib/zim/creator.py | 11 +++++++----
 tests/zim/test_zim_creator.py    | 18 ++++++++++++++++++
 3 files changed, 26 insertions(+), 4 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index b260efc1..c1edec26 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -19,6 +19,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Keep URL-encoded `#` characters of the path or querystring when rewriting document URIs, instead of dropping everything after them (#341)
 - Rewrite the `poster` attribute (e.g. of `<video>`) when rewriting HTML, like `src` and `href` (#339)
 - Do not collapse consecutive slashes `//` into a single `/` when normalizing URLs into ZIM paths, in both Python and JS rewriting (#340)
+- `Creator.finish()` no longer silently discards errors raised by libzim while finalizing the ZIM, so a failure during finalization is no longer reported as a successful run (openzim/warc2zim#462)
 
 ## [5.4.1] - 2026-07-31
 
diff --git a/src/zimscraperlib/zim/creator.py b/src/zimscraperlib/zim/creator.py
index e1f010d4..c6dfd5b7 100644
--- a/src/zimscraperlib/zim/creator.py
+++ b/src/zimscraperlib/zim/creator.py
@@ -117,6 +117,7 @@ def __init__(
         self.__indexing_configured = False
         self.__indexing_value: bool = False
         self.can_finish = True
+        self._finish_called = False
 
         self.set_mainpath(main_path)
 
@@ -475,10 +476,12 @@ def finish(
         """Triggers finalization of ZIM creation and create final ZIM file."""
         if not getattr(self, "can_finish", False):
             return
-        try:
-            super().__exit__(None, None, None)
-        except RuntimeError:
-            pass
+        # finishing an already-finished creator makes libzim raise a RuntimeError;
+        # make this a no-op instead of calling into libzim again
+        if getattr(self, "_finish_called", False):
+            return
+        self._finish_called = True
+        super().__exit__(None, None, None)
 
     def __enter__(self):
         self.start()
diff --git a/tests/zim/test_zim_creator.py b/tests/zim/test_zim_creator.py
index 279c0f54..e16e558d 100644
--- a/tests/zim/test_zim_creator.py
+++ b/tests/zim/test_zim_creator.py
@@ -12,6 +12,7 @@
 from typing import Any, NamedTuple
 from unittest.mock import call, patch
 
+import libzim.writer  # pyright: ignore[reportMissingModuleSource]
 import pytest
 from libzim.writer import Compression  # pyright: ignore[reportMissingModuleSource]
 
@@ -268,6 +269,23 @@ def test_cannot_finish(tmp_path: pathlib.Path):
     creator.finish()
 
 
+def test_finish_error_propagates(tmp_path: pathlib.Path):
+    fpath = tmp_path / "test.zim"
+    creator = Creator(fpath, "welcome").config_dev_metadata()
+    creator.start()
+    creator.add_item(StaticItem(path="welcome", content="hello"))
+
+    with (
+        patch.object(
+            libzim.writer.Creator,
+            "__exit__",
+            side_effect=RuntimeError("boom"),
+        ),
+        pytest.raises(RuntimeError, match="boom"),
+    ):
+        creator.finish()
+
+
 def test_sourcefile_removal(tmp_path: pathlib.Path, html_file: pathlib.Path):
     fpath = tmp_path / "test.zim"
     with Creator(fpath, "").config_dev_metadata() as creator: