Update tracking code extractor.

This commit is contained in:
AccentuSoft
2022-04-01 20:43:42 +03:00
parent 4193be31c5
commit 5cf2a442a1

View File

@@ -28,6 +28,7 @@ class JSCodeExtractor:
gRegex = re.compile(r'\bG-[A-Z0-9]{1,15}\b', re.IGNORECASE) gRegex = re.compile(r'\bG-[A-Z0-9]{1,15}\b', re.IGNORECASE)
qualtricsRegex = re.compile(r'\bQ_ZID=[a-zA-Z_0-9]*\b', re.IGNORECASE) qualtricsRegex = re.compile(r'\bQ_ZID=[a-zA-Z_0-9]*\b', re.IGNORECASE)
pingdomRegex = re.compile(r'\bpa-[a-fA-F0-9]{24}.js\b$', re.IGNORECASE) pingdomRegex = re.compile(r'\bpa-[a-fA-F0-9]{24}.js\b$', re.IGNORECASE)
mPulseRegex = re.compile(r'\b[a-zA-Z_0-9]{5}(-[a-zA-Z_0-9]{5}){4}\b', re.IGNORECASE)
def GetTrackingCodes(pageUid, requestUrl) -> None: def GetTrackingCodes(pageUid, requestUrl) -> None:
if requestUrl not in requestUrlsParsed: if requestUrl not in requestUrlsParsed:
@@ -62,6 +63,11 @@ class JSCodeExtractor:
'Entity Type': 'Phrase'}, 'Entity Type': 'Phrase'},
{pageUid: {'Resolution': 'Pingdom Tracking Code', {pageUid: {'Resolution': 'Pingdom Tracking Code',
'Notes': ''}}]) 'Notes': ''}}])
for mCode in mPulseRegex.findall(requestUrl):
returnResults.append([{'Phrase': mCode,
'Entity Type': 'Phrase'},
{pageUid: {'Resolution': 'mPulse Tracking Code',
'Notes': ''}}])
with sync_playwright() as p: with sync_playwright() as p:
browser = p.firefox.launch() browser = p.firefox.launch()