diff --git a/scraper/scrapers/scraperHelper.js b/scraper/scrapers/scraperHelper.js index d15378a..3598a7c 100644 --- a/scraper/scrapers/scraperHelper.js +++ b/scraper/scrapers/scraperHelper.js @@ -7,6 +7,7 @@ function sanitizePtName(name) { .replace(/(.*)\b(\d{3,4}P)\b(?!.*\d{3,4}[Pp])(.*)/, '$1$3 $2') // add resolution to the end if missing .replace(/^[\[{]?(?:ACESSE.*|WWW\.)?[A-Z]+\.(COM|NET|ORG|TO|TV|ME)\b\s*[-\]}]+[\s.]*/i, '') // replace watermarks .replace(/^(\d*(?:\.\d{1,2})?(?:[4A-Z-]{3,}|P)[-.]+)+/, '') // replace metadata prefixes + .replace(/^[\[{]?(?:ACESSE.*|WWW\.)?[A-Z]+\.(COM|NET|ORG|TO|TV|ME)\b\s*[-\]}]+[\s.]*/i, '') // replace watermarks2 .replace(/^(COM|NET|ORG|TO|TV|ME)\b\s*-+[\s.]*/, '') // replace dangling site endings .trim(); }