Detect brightcove tracker - JSCodeExtractor.py

This commit is contained in:
AccentuSoft
2022-04-26 17:01:35 +03:00
parent 624aee7013
commit 7c7288d949

View File

@@ -45,6 +45,7 @@ class JSCodeExtractor:
pushlyRegex = re.compile(r'cdn.p-n.io/.*domain_key=[\w%]*')
akamaiRegex = re.compile(r'/akam/.*a=[\w%=]*')
demdexRegex = re.compile(r'dpm\.demdex\.net/id\?.*d_orgid=[^&]*')
brightcoveRegex = re.compile(r'metrics\.brightcove\.com/.*/tracker\?.*&account=[^&]*')
def GetTrackingCodes(pageUid, requestUrl) -> None:
if requestUrl not in requestUrlsParsed:
@@ -162,6 +163,11 @@ class JSCodeExtractor:
'Entity Type': 'Phrase'},
{pageUid: {'Resolution': 'DemDex (Adobe) Website ID',
'Notes': ''}}])
for bCode in brightcoveRegex.findall(requestUrl):
returnResults.append([{'Phrase': bCode.split('account=')[1],
'Entity Type': 'Phrase'},
{pageUid: {'Resolution': 'BrightCove Website ID',
'Notes': ''}}])
with sync_playwright() as p:
browser = p.firefox.launch()