Ciphr iRecruit Jobs API.

# The first host label is the tenant id; the listings path is fixed.
tenant = "refugeecouncil"
board_url = f"https://{tenant}.ciphr-irecruit.com/Applicants/vacancy"

from curl_cffi import requests  # pip install curl_cffi

headers = {
    "Accept": "text/html,application/xhtml+xml",
    "Accept-Language": "en-GB,en;q=0.9",
}

resp = requests.get(board_url, headers=headers, impersonate="chrome131", timeout=20)
resp.raise_for_status()  # a plain HTTP client would 500 here instead
html = resp.text

import re
from bs4 import BeautifulSoup
from urllib.parse import urljoin

soup = BeautifulSoup(html, "html.parser")

# Authoritative total — a missing or disagreeing count means an incomplete snapshot.
match = re.search(r"([\d,]+)\s+vacanc(?:y|ies)\s+found", soup.get_text(" "), re.I)
advertised = int(match.group(1).replace(",", "")) if match else None

table = soup.select_one("table.table")
rows = table.select("tbody > tr") if table else []
headers_row = [th.get_text(" ", strip=True) for th in table.select("thead th")] if table else []

def cell(cells, *labels):
    wanted = {label.lower() for label in labels}
    for i, header in enumerate(headers_row):
        if i < len(cells) and header.lower() in wanted:
            return cells[i].get_text(" ", strip=True)
    return ""

jobs = []
for row in rows:
    cells = row.select("td")
    anchor = cells[0].select_one("a[href]") if cells else None
    if not anchor:
        continue
    listing_url = urljoin(board_url, anchor["href"])
    id_match = re.search(r"/Applicants/vacancy/(\d+)", listing_url, re.I)
    if not id_match:
        continue
    jobs.append({
        "external_id": id_match.group(1),   # numeric segment only — never the slug
        "title": anchor.get_text(" ", strip=True),
        "listing_url": listing_url,
        "location": cell(cells, "Location"),
        "salary_text": cell(cells, "Salary"),
        "contract_type": cell(cells, "Type", "Contract Type", "Vacancy Type"),
        "job_category": cell(cells, "Job Category", "Category"),
        "closing_text": cell(cells, "Application Deadline", "Closing Date"),
    })

print(f"Parsed {len(jobs)} of {advertised} advertised vacancies")

for job in jobs:
    resp = requests.get(job["listing_url"], headers=headers, impersonate="chrome131", timeout=20)
    if resp.status_code in (404, 410):
        continue  # vacancy was removed
    resp.raise_for_status()
    detail = BeautifulSoup(resp.text, "html.parser")

    root = detail.select_one("#VacancyDetails") or detail.body
    hidden = root.select_one("#txtJobId")
    if not hidden or hidden.get("value") != job["external_id"]:
        continue  # route id and native id must agree

    heading = root.select_one("h1")
    reference = heading.select_one("small") if heading else None
    fields = {}
    for term in root.select("dl > dt"):
        value = term.find_next_sibling("dd")
        if value is not None:
            fields[term.get_text(" ", strip=True)] = value

    summary = fields.get("Job Summary")
    apply_anchor = root.select_one("a[href*='/Applicants/vacancy/apply/']")

    job.update({
        "reference": reference.get_text(strip=True) if reference else "",
        "description": summary.get_text(" ", strip=True) if summary else "",
        "hours_text": fields["Hours per week"].get_text(" ", strip=True) if "Hours per week" in fields else "",
        "apply_url": urljoin(job["listing_url"], apply_anchor["href"]) if apply_anchor else None,
    })

EMPTY_MARKERS = re.compile(r"no vacancies|there are no current vacancies|0 vacancies", re.I)

def snapshot_is_complete(soup, rows, advertised):
    if not rows:
        # A genuinely empty board says so; silence usually means a blocked fetch.
        return advertised == 0 or EMPTY_MARKERS.search(soup.get_text(" ")) is not None
    return advertised is not None and advertised == len(rows)

if not snapshot_is_complete(soup, rows, advertised):
    raise RuntimeError("Ciphr snapshot incomplete — re-fetch before trusting the results")