Sublime MQL rules: paypal

type.inbound
and length(attachments) == 0
and sender.email.domain.root_domain in (
  "paypal.com",
  "paypal.com.mx",
  "paypal.com.br",
  "paypal.com.ar",
  "paypal.co.uk"
)
and (
  any(filter(html.xpath(body.html,
                        // all the tables, which don't have descendant tables, after the first image and before a hr with a footerDivider class that appears after the current table.
                        // using //text()[normalize-space()] allows us to split each table up by line breaks, so each line can be inspected uniquely
                        '//img[@alt="PayPal"]/following::table[not(descendant::table) and count(following::hr[@class="footerDivider"]) > 0]//text()[normalize-space()]'
             ).nodes,
             (
               // icontains a phone number
               (
                 regex.icontains(strings.replace_confusables(.display_text),
                                 '\b\+?([ilo0-9]{1}.)?\(?[ilo0-9]{3}?\)?.[ilo0-9]{3}.?[ilo0-9]{4}\b'
                 )
                 or regex.icontains(strings.replace_confusables(.display_text),
                                    '\+[ilo0-9]{1,3}[ilo0-9]{10}'
                 )
                 // +12028001238
                 or regex.icontains(strings.replace_confusables(.display_text),
                                    '[ilo0-9]{3}\.[ilo0-9]{3}\.[ilo0-9]{4}'
                 )
                 // 202-800-1238
                 or regex.icontains(strings.replace_confusables(.display_text),
                                    '[ilo0-9]{3}-[ilo0-9]{3}-[ilo0-9]{4}'
                 )
                 // (202) 800-1238
                 // (202) -800 -1238
                 // (202) 800 1238
                 // (202)-800-1238
                 or regex.icontains(strings.replace_confusables(.display_text),
                                    '\([ilo0-9]{3}\)[\s-]+[ilo0-9]{3}[\s-]+[ilo0-9]{4}'
                 )
                 // (202)-800-1238
                 or regex.icontains(strings.replace_confusables(.display_text),
                                    '\([ilo0-9]{3}\)-[ilo0-9]{3}-[ilo0-9]{4}'
                 )
                 or ( // 8123456789
                   regex.icontains(strings.replace_confusables(.display_text),
                                   '8[ilo0-9]{9}'
                   )
                   and regex.icontains(strings.replace_confusables(.display_text
                                       ),
                                       '\+[1l]'
                   )
                 )
               )
             )
             // filter out elements which contain non actor controlled details
             // this often wording from paypal templates that might contain phone numbers
             // but are not elements that are actor controlled.

             // main customer service number
             and not strings.icontains(.display_text, '888-221-1161')
             // credit services number
             and not strings.icontains(.display_text, '844-896-4937')
             // pay in 4 number
             and not strings.icontains(.display_text, '(800) 504-7534')
             // debt collection for pay later
             and not strings.icontains(.display_text, '877-589-5171')
             // debt collections for PayPal
             and not strings.icontains(.display_text, '866-380-1798')
             // card activation
             and not strings.icontains(.display_text, '800-314-8298')
             // often the transcation id looks like a phone number and matches the regex
             and not regex.icontains(.display_text, "Transaction (date|ID)\n")
             // this segment can include phone numbers, but the wording is not actor controlled and shows up "under the fold" in the templates
             and not strings.istarts_with(.display_text,
                                          "If you have any questions about this payment, you can"
             )
      ),
      strings.icontains(.display_text, "you did not")
      or strings.icontains(.display_text, "Critical Alert")
      or strings.icontains(.display_text, "is not for")
      or strings.icontains(.display_text, "done by you")
      or regex.icontains(.display_text, "don['’]t recognize")
      or regex.icontains(.display_text, "didn['’]t (?:ma[kd]e|place) this")
      or regex.icontains(.display_text, "[wh]as.*placed by you")
      or strings.icontains(.display_text, "issue with")
      or regex.icontains(.display_text, "was.*made by you")
      or (
        strings.icontains(.display_text, "Fraud")
        and strings.icontains(.display_text, "Alert")
      )
      or strings.icontains(.display_text, "fraudulent")
      or strings.icontains(.display_text, "using your PayPal")
      or strings.icontains(.display_text, "subscription")
      or strings.icontains(.display_text, "antivirus")
      or strings.icontains(.display_text, "support")
      or strings.icontains(.display_text, "sincerely apologize")
      or strings.icontains(.display_text, "receipt")
      // pull in common wording from transaction types from paypal
      // this wording should be part of the template and not actor controlled
      // but is generally prepended or appended with actor controlled elements
      // such as using the business name to deliver callback details
      // when a phone number is present
      or strings.icontains(.display_text, "sent you an invoice")
      or strings.icontains(.display_text, "a money request")
      or strings.icontains(.display_text, "invited you as a developer")
      or strings.icontains(.display_text, "Purchase")
      or strings.icontains(.display_text, "Market*Value")
      or strings.icontains(.display_text, "BTC")
      or strings.icontains(.display_text, "Etherium (ETH)")
      or strings.icontains(.display_text, "get in touch with our")
      or strings.icontains(.display_text, "quickly inform")
      or strings.icontains(.display_text, "quickly reach ")
      or strings.icontains(.display_text, "detected unusual transactions")
      or strings.icontains(.display_text, "without your authorization")
      or strings.icontains(.display_text, "unauthorized activitiy")
      or strings.icontains(.display_text, "unauthorized transaction")
      or strings.icontains(.display_text, "cancel")
      or strings.icontains(.display_text, "renew")
      or strings.icontains(.display_text, "refund")
      or regex.icontains(.display_text, 'help.{0,3}desk')
      or strings.icontains(.display_text, " your funds")
      or strings.icontains(.display_text, " your checking")
      or strings.icontains(.display_text, " your saving")
      or strings.icontains(.display_text, "secure your account")
      or strings.icontains(.display_text, "recover your")
      or strings.icontains(.display_text, "unusual activity")
      or strings.icontains(.display_text, "suspicious transaction")
      or strings.icontains(.display_text, "transaction history")
      or strings.icontains(.display_text, "please ignore this")
      or strings.icontains(.display_text, "will be approved")
      or strings.icontains(.display_text, "report activity")
      or strings.icontains(.display_text, "paypal billing team")

      // Unicode confusables words obfuscated in note
      or regex.icontains(.display_text,
                         '\+𝟭|𝗽𝗮𝘆𝗺𝗲𝗻𝘁|𝗛𝗲𝗹𝗽 𝗗𝗲𝘀𝗸|𝗿𝗲𝗳𝘂𝗻𝗱|𝗮𝗻𝘁𝗶𝘃𝗶𝗿𝘂𝘀|𝗰𝗮𝗹𝗹|𝗰𝗮𝗻𝗰𝗲𝗹|𝗰𝗼𝗻𝘁𝗮𝗰𝘁|cᴀʟʟ'
      )
      or strings.icontains(.display_text, "kindly")
      or regex.icontains(strings.replace_confusables(.display_text),
                         '(?:call|cᴀʟʟ|reach|contact|get in touch|\binform\b|let us know|coɴᴛᴀcᴛ)'
      )
  )
  // message contains phone number regex and dollar ammounts in the subject
  or (
    regex.icontains(strings.replace_confusables(subject.base),
                    '\+?(?:[ilo0-9]{1}.)?\(?[ilo0-9]{3}?\)?.[ilo0-9]{3}.?[ilo0-9]{4}',
                    '\+?(?:[ilo0-9]{1,2})?\s?\(?\d{3}\)?[\s\.\-⋅]{0,5}[ilo0-9]{3}[\s\.\-⋅]{0,5}[ilo0-9]{4}'
    )
    and (
      regex.icontains(subject.base, '(?:USD|\$)\s?\d')
      or regex.icontains(subject.base, '\d+\.\d{2}\s?(?:USD|usd)')
    )
  )
)
// Handle emails for class action settlement refunds. Not official PayPal communications but a common-enough thing where a small refund is sent from a court-appointed settlement administrator
and not (
  strings.icontains(body.html.display_text, "settlement")
  and (
    strings.icontains(body.html.display_text, "court")
    or strings.icontains(body.html.display_text, "FTC")
    or strings.icontains(body.html.display_text, "administrator")
  )
)