Compare commits

..

2 Commits

Author SHA1 Message Date
dgtlmoon
cd83704a5e oops 2026-03-15 16:56:05 +01:00
dgtlmoon
6333f72578 Content Fetchers / Browsers - Improvements for pluggable extra fetchers/browsers. 2026-03-15 16:33:06 +01:00
7 changed files with 13 additions and 106 deletions

View File

@@ -2,7 +2,7 @@
# Read more https://github.com/dgtlmoon/changedetection.io/wiki
# Semver means never use .01, or 00. Should be .1.
__version__ = '0.54.6'
__version__ = '0.54.5'
from changedetectionio.strtobool import strtobool
from json.decoder import JSONDecodeError

View File

@@ -154,8 +154,9 @@
</span>
</div>
<div class="pure-control-group">
<br>
{{ _('Tip:') }} <a href="{{ url_for('settings.settings_page')}}#proxies">{{ _('Connect using Bright Data proxies, find out more here.') }}</a>
<br>
{{ _('Tip:') }} <a href="https://github.com/dgtlmoon/changedetection.io/wiki/Proxy-configuration#brightdata-proxy-support">{{ _('Connect using Bright Data and Oxylabs Proxies, find out more here.') }}</a>
</div>
</div>
@@ -351,7 +352,7 @@ nav
</div>
</div>
<p><strong>{{ _('Tip') }}</strong>: {{ _('"Residential" and "Mobile" proxy type can be more successful than "Data Center" for blocked websites.') }}</p>
<p><strong>{{ _('Tip') }}</strong>: {{ _('"Residential" and "Mobile" proxy type can be more successfull than "Data Center" for blocked websites.') }}</p>
<div class="pure-control-group" id="extra-proxies-setting">
{{ render_fieldlist_with_inline_errors(form.requests.form.extra_proxies) }}

View File

@@ -487,25 +487,13 @@ def extract_json_as_string(content, json_filter, ensure_is_ldjson_info_type=None
except json.JSONDecodeError as e:
logger.warning(f"Error processing JSON {content[:20]}...{str(e)})")
else:
# Check for JSONP wrapper: someCallback({...}) or some.namespace({...})
# Server may claim application/json but actually return JSONP
jsonp_match = re.match(r'^\w[\w.]*\s*\((.+)\)\s*;?\s*$', content.lstrip("\ufeff").strip(), re.DOTALL)
if jsonp_match:
try:
inner = jsonp_match.group(1).strip()
logger.warning(f"Content looks like JSONP, attempting to extract inner JSON for filter '{json_filter}'")
stripped_text_from_html = _parse_json(json.loads(inner), json_filter)
except json.JSONDecodeError as e:
logger.warning(f"Error processing JSONP inner content {content[:20]}...{str(e)})")
if not stripped_text_from_html:
# Probably something else, go fish inside for it
try:
stripped_text_from_html = extract_json_blob_from_html(content=content,
ensure_is_ldjson_info_type=ensure_is_ldjson_info_type,
json_filter=json_filter)
except json.JSONDecodeError as e:
logger.warning(f"Error processing JSON while extracting JSON from HTML blob {content[:20]}...{str(e)})")
# Probably something else, go fish inside for it
try:
stripped_text_from_html = extract_json_blob_from_html(content=content,
ensure_is_ldjson_info_type=ensure_is_ldjson_info_type,
json_filter=json_filter )
except json.JSONDecodeError as e:
logger.warning(f"Error processing JSON while extracting JSON from HTML blob {content[:20]}...{str(e)})")
if not stripped_text_from_html:
# Re 265 - Just return an empty string when filter not found

View File

@@ -100,13 +100,7 @@ class guess_stream_type():
if any(s in http_content_header for s in RSS_XML_CONTENT_TYPES):
self.is_rss = True
elif any(s in http_content_header for s in JSON_CONTENT_TYPES):
# JSONP detection: server claims application/json but content is actually JSONP (e.g. cb({...}))
# A JSONP response starts with an identifier followed by '(' - not valid JSON
if re.match(r'^\w[\w.]*\s*\(', test_content):
logger.warning(f"Content-Type header claims JSON but content looks like JSONP (starts with identifier+parenthesis) - treating as plaintext")
self.is_plaintext = True
else:
self.is_json = True
self.is_json = True
elif 'pdf' in magic_content_header:
self.is_pdf = True
# magic will call a rss document 'xml'

View File

@@ -199,31 +199,8 @@ def handle_watch_update(socketio, **kwargs):
logger.error(f"Socket.IO error in handle_watch_update: {str(e)}")
def _suppress_werkzeug_ws_abrupt_disconnect_noise():
"""Patch BaseWSGIServer.log to suppress the AssertionError traceback that fires when
a browser closes a WebSocket connection mid-handshake (e.g. closing a tab).
The exception is caught inside run_wsgi and routed to self.server.log() — it never
propagates out, so wrapping run_wsgi doesn't help. Patching the log method is the
only reliable intercept point. The error is cosmetic: Socket.IO already handles the
disconnect correctly via its own disconnect handler and timeout logic."""
try:
from werkzeug.serving import BaseWSGIServer
_original_log = BaseWSGIServer.log
def _filtered_log(self, type, message, *args):
if type == 'error' and 'write() before start_response' in message:
return
_original_log(self, type, message, *args)
BaseWSGIServer.log = _filtered_log
except Exception:
pass
def init_socketio(app, datastore):
"""Initialize SocketIO with the main Flask app"""
_suppress_werkzeug_ws_abrupt_disconnect_noise()
import platform
import sys

View File

@@ -116,14 +116,6 @@ $(document).ready(function () {
$('#realtime-conn-error').show();
});
// Tell the server we're leaving cleanly so it can release the connection
// immediately rather than waiting for a timeout.
// Note: this only fires for voluntary closes (tab/window close, navigation away).
// Hard kills, crashes and network drops will still timeout normally on the server.
window.addEventListener('beforeunload', function () {
socket.disconnect();
});
socket.on('queue_size', function (data) {
console.log(`${data.event_timestamp} - Queue size update: ${data.q_length}`);
if(queueSizePagerInfoText) {

View File

@@ -16,51 +16,6 @@ except ModuleNotFoundError:
def test_jsonp_treated_as_plaintext():
from ..processors.magic import guess_stream_type
# JSONP content (server wrongly claims application/json) should be detected as plaintext
# Callback names are arbitrary identifiers, not always 'cb'
jsonp_content = 'jQuery123456({ "version": "8.0.41", "url": "https://example.com/app.apk" })'
result = guess_stream_type(http_content_header="application/json", content=jsonp_content)
assert result.is_json is False
assert result.is_plaintext is True
# Variation with dotted callback name e.g. jQuery.cb(...)
jsonp_dotted = 'some.callback({ "version": "1.0" })'
result = guess_stream_type(http_content_header="application/json", content=jsonp_dotted)
assert result.is_json is False
assert result.is_plaintext is True
# Real JSON should still be detected as JSON
json_content = '{ "version": "8.0.41", "url": "https://example.com/app.apk" }'
result = guess_stream_type(http_content_header="application/json", content=json_content)
assert result.is_json is True
assert result.is_plaintext is False
def test_jsonp_json_filter_extraction():
from .. import html_tools
# Tough case: dotted namespace callback, trailing semicolon, deeply nested content with arrays
jsonp_content = 'weixin.update.callback({"platforms": {"android": {"variants": [{"arch": "arm64", "versionName": "8.0.68", "url": "https://example.com/app-arm64.apk"}, {"arch": "arm32", "versionName": "8.0.41", "url": "https://example.com/app-arm32.apk"}]}}});'
# Deep nested jsonpath filter into array element
text = html_tools.extract_json_as_string(jsonp_content, "json:$.platforms.android.variants[0].versionName")
assert text == '"8.0.68"'
# Filter that selects the second array element
text = html_tools.extract_json_as_string(jsonp_content, "json:$.platforms.android.variants[1].arch")
assert text == '"arm32"'
if jq_support:
text = html_tools.extract_json_as_string(jsonp_content, "jq:.platforms.android.variants[0].versionName")
assert text == '"8.0.68"'
text = html_tools.extract_json_as_string(jsonp_content, "jqraw:.platforms.android.variants[1].url")
assert text == "https://example.com/app-arm32.apk"
def test_unittest_inline_html_extract():
# So lets pretend that the JSON we want is inside some HTML
content="""