# Changelog of Reed.co.uk Jobs Scraper — UK Salaries & Full Descriptions (`teodor_banea/reed-jobs-scraper`) Actor

- **URL**: https://apify.com/teodor_banea/reed-jobs-scraper/changelog.md
- **Full Actor documentation**: https://apify.com/teodor_banea/reed-jobs-scraper.md

## Changelog

### 0.1.9 (2026-10-22)

- Removed the only-new-since-last-run option and its named key-value store; the actor no longer leaves any storage
  in your account. Use the date filter (*Posted within*) for scheduled runs.

### 0.1.8 (2026-10-20)

- Only new: runs through Reed's date window (last 3 days while the last full listing is at most 2 days old, last
  week while at most 6 days old) never move the full-check time; once it is 7 days old the next run lists every job
  again. A job published days after it was created is caught within about a week.

### 0.1.7 (2026-10-20)

- Only new: delivered jobs added while the memory was being saved are no longer forgotten (about 4% of jobs came
  back in the next run of big searches without job details).
- Only new: after a run that checked every match of a search, later runs of it (with "Posted within" on "Any
  time") only list jobs created since that check, through Reed's own date filter — frequent schedules of big
  searches stay cheap. Big searches without such a check get a tip to set "Posted within".
- Salaries in other currencies (EUR, USD, AUD, AED) are labelled with their own currency and symbol; the yearly
  figures stay Reed's, in GBP.
- Salary texts as the job page shows them: "£95 - £105 per day", extras in the page's order ("OTE, inc benefits,
  pro-rata, negotiable").
- Times of jobs from job feeds (written without fractions of a second) are read as UK time; all output stays UTC.
- `locationText` is the location the job page shows; `city` stays the postal town.
- Template leftovers ("{{}}", a stray "undefined", broken characters) are removed from texts.
- Problems are written to ERRORS at most every few seconds (one write per problem was slower and costlier, and a
  late write could drop a problem).
- Faster end of capped runs that re-read result pages; clearer final status (jobs Reed moved between pages, jobs in
  the sector Reed cannot filter, expired jobs counted once).

### 0.1.6 (2026-10-20)

- First version: Reed.co.uk jobs in three modes — search (keywords + location + filters), job URLs (job pages or
  bare job ids) and search URLs (Reed search pages with any filters).
- Filters applied by Reed's own search: location radius, salary range (Reed's yearly figure), job type, full / part
  time, remote / hybrid / on-site, posting date, specialism, agency or employer, salary shown, training courses
  hidden (default), graduate, visa sponsorship, Easy Apply.
- 60 fields per row: full description (plain text and clean HTML), normalised salary (amounts, period, Reed's yearly
  figures, negotiable / competitive / commission-only / OTE / pro-rata / benefits flags), town, county, region,
  nation and postcode district, occupation taxonomy, sector and specialism, skills, contract and hours, remote
  option, employer or agency with logo and Reed page, recruiter type, apply method, posting / update / expiry dates.
- Searches above Reed's 10,000-result limit are split by specialism (and by job type when a specialism is still
  too large).
- Keyword results that Reed reorders between page loads are caught by re-reading the result pages while jobs are
  missing (at most three readings).
- A job whose page cannot be read (or that expires during the run) is still delivered from its search data
  (`detailsScraped: false`) and listed in ERRORS; expired job URLs and invalid URLs are listed in ERRORS.
- Times are UTC. Jobs imported from one job feed carry UK local time (up to an hour "in the future"); such times
  are converted from UK time.
- "Only jobs not delivered by earlier runs": per-search memory in the named store `reed-jobs-seen`; every result
  page is still read, so capped runs never hide jobs from later runs.