Subversion Repositories SmartDukaan

Last modification

• Rev 37420 2026-08-26 16:23:17
• Author: amit

• Log message:

  Oppo warranty check: read the JSON response instead of scraping the date
  
  Oppo has recorded no activation dates since 23-Aug. With logging restored
  (r37410) the cause is visible: the captcha IS solved -- 10 successes per tick --
  and it then fails on the very next line with
  
  no such element: //*[contains(text(),'UTC+5.5') or contains(text(),'Non-Activate')]
  
  Their result is now label/value pairs, and the value is rendered from TWO text
  nodes: dayjs(regDate).format('DD/MM/YYYY') plus a '(UTC+x)' suffix computed from
  the browser's own timezone. XPath contains(text(),...) tests only the first node
  -- the date -- so it never matched. 'Non-Activate' is gone too: an inactive
  device now reads 'The system will update the date within 7 days of device
  activation.' Verified against a replica of their DOM: old locator 0 matches, and
  the value element yields '23/08/2026 (UTC+5.5)'.
  
  Rather than chase their markup again, take the value from the source: the page
  calls /oppo-api/basic/v1/getDeviceInfo and the response carries regDate as epoch
  millis -- no date format, no locale, no timezone suffix to parse. A small hook
  records that response as the page receives it; nothing extra is requested and no
  captcha behaviour changes.
  
  Also switches the outcome to always record the imei. findElement THREW when the
  element was missing, which skipped the fallback entirely, so nothing was written
  to dateMap and the row was re-queued on every run indefinitely -- 'Could not
  capture date' appears 0 times in the log because it was unreachable.