From a7e8b072b72af1e23506bbf696ebfd929932d050 Mon Sep 17 00:00:00 2001
From: santhreal <64453045+santhreal@users.noreply.github.com>
Date: Fri, 17 Jul 2026 22:19:52 -0700
Subject: [PATCH] fix: keep valid JSON-LD when a sibling script is malformed
Under errors=ignore/log, one bad application/ld+json script dropped all
siblings. Honor the errors policy per script inside JsonLdExtractor.
---
extruct/_extruct.py | 2 +-
extruct/jsonld.py | 26 +++++++++++++++++++-------
tests/test_extruct.py | 21 +++++++++++++++++++++
tests/test_jsonld.py | 26 ++++++++++++++++++++++++++
4 files changed, 67 insertions(+), 8 deletions(-)
diff --git a/extruct/_extruct.py b/extruct/_extruct.py
index 51da7f49..d19c804a 100644
--- a/extruct/_extruct.py
+++ b/extruct/_extruct.py
@@ -99,7 +99,7 @@ def extract(
processors.append(
(
"json-ld",
- JsonLdExtractor().extract_items,
+ JsonLdExtractor(errors=errors).extract_items,
tree,
)
)
diff --git a/extruct/jsonld.py b/extruct/jsonld.py
index d25a4183..8a1d184e 100644
--- a/extruct/jsonld.py
+++ b/extruct/jsonld.py
@@ -4,6 +4,7 @@
"""
import json
+import logging
import re
import jstyleson
@@ -11,6 +12,7 @@
from extruct.utils import parse_html
+logger = logging.getLogger(__name__)
HTML_OR_JS_COMMENTLINE = re.compile(r"^\s*(//.*|)")
@@ -19,18 +21,28 @@ class JsonLdExtractor:
'descendant-or-self::script[@type="application/ld+json"]'
)
+ def __init__(self, errors="strict"):
+ self.errors = errors
+
def extract(self, htmlstring, base_url=None, encoding="UTF-8"):
tree = parse_html(htmlstring, encoding=encoding)
return self.extract_items(tree, base_url=base_url)
def extract_items(self, document, base_url=None):
- return [
- item
- for items in map(self._extract_items, self._xp_jsonld(document)) # type: ignore[arg-type]
- if items
- for item in items
- if item
- ]
+ items = []
+ for node in self._xp_jsonld(document): # type: ignore[arg-type]
+ try:
+ for item in self._extract_items(node) or ():
+ if item:
+ items.append(item)
+ except ValueError as e:
+ if self.errors == "strict":
+ raise
+ if self.errors == "log":
+ logger.exception(
+ "Failed to extract json-ld script, raises {}".format(e)
+ )
+ return items
def _extract_items(self, node):
script = node.xpath("string()").strip()
diff --git a/tests/test_extruct.py b/tests/test_extruct.py
index f7ef6fc3..9c0fe950 100644
--- a/tests/test_extruct.py
+++ b/tests/test_extruct.py
@@ -92,3 +92,24 @@ def test_errors(self):
# ignore exceptions
data = extruct.extract(body, errors="log")
assert data == {}
+
+ def test_errors_ignore_keeps_valid_jsonld_siblings(self):
+ body = (
+ ""
+ '"
+ ''
+ ""
+ )
+ expected = {"json-ld": [{"@type": "Person", "name": "Ada"}]}
+ data = extruct.extract(body, errors="ignore", syntaxes=["json-ld"])
+ self.assertEqual(data, expected)
+
+ with self.assertLogs("extruct.jsonld", level="ERROR") as cm:
+ data = extruct.extract(body, errors="log", syntaxes=["json-ld"])
+ self.assertEqual(data, expected)
+ self.assertTrue(any("json-ld script" in line for line in cm.output))
+
+ with self.assertRaises(ValueError):
+ extruct.extract(body, errors="strict", syntaxes=["json-ld"])
diff --git a/tests/test_jsonld.py b/tests/test_jsonld.py
index 178b3229..f31b998d 100644
--- a/tests/test_jsonld.py
+++ b/tests/test_jsonld.py
@@ -69,3 +69,29 @@ def test_empty_jsonld_script(self):
body = ''
data = jsonlde.extract(body)
self.assertEqual(data, [])
+
+ def test_malformed_sibling_raises_by_default(self):
+ jsonlde = JsonLdExtractor()
+ body = (
+ ""
+ '"
+ ''
+ ""
+ )
+ with self.assertRaises(ValueError):
+ jsonlde.extract(body)
+
+ def test_malformed_sibling_skipped_when_ignoring(self):
+ body = (
+ ""
+ '"
+ ''
+ ""
+ )
+ data = JsonLdExtractor(errors="ignore").extract(body)
+ self.assertEqual(data, [{"@type": "Person", "name": "Ada"}])
+