diff --git a/job_scout/_ap.py b/job_scout/_ap.py deleted file mode 100644 index ab6ea31..0000000 --- a/job_scout/_ap.py +++ /dev/null @@ -1,20 +0,0 @@ -import io -from playwright.sync_api import sync_playwright -out = io.open('_ap_out.txt', 'w', encoding='utf-8') -with sync_playwright() as p: - b = p.chromium.launch() - pg = b.new_page() - calls = [] - pg.on('response', lambda r: calls.append((r.request.method, r.url, r.status, - r.headers.get('content-type', ''), r.request.post_data))) - pg.goto('https://jobs.apple.com/en-us/search?location=switzerland-CHE', - wait_until='domcontentloaded', timeout=60000) - pg.wait_for_timeout(10000) - for m, u, st, ct, pd in calls: - if 'jobs.apple.com' in u and ('api' in u or 'search' in u or 'graphql' in u): - out.write('%s %s [%s] %s\n' % (m, u[:170], st, ct[:40])) - if pd: out.write(' POST %s\n' % pd[:400]) - out.write('---- page text ----\n') - out.write(pg.inner_text('body')[:1200]) - b.close() -out.close() diff --git a/job_scout/_fix_probe.py b/job_scout/_fix_probe.py deleted file mode 100644 index 16208ce..0000000 --- a/job_scout/_fix_probe.py +++ /dev/null @@ -1,31 +0,0 @@ -import io, json -from playwright.sync_api import sync_playwright -SITES = { - 'roche': 'https://careers.roche.com/global/en/search-results?keywords=&locationsearch=Switzerland', - 'meta': 'https://www.metacareers.com/jobs?offices[0]=Zurich%2C%20Switzerland', - 'apple': 'https://jobs.apple.com/en-us/search?location=switzerland-CHE', -} -out = io.open('_fix_out.txt', 'w', encoding='utf-8') -with sync_playwright() as p: - b = p.chromium.launch() - for k, u in SITES.items(): - pg = b.new_page(); calls = [] - def on_resp(r, calls=calls): - ct = r.headers.get('content-type', '') - if 'json' in ct and r.request.method in ('GET', 'POST'): - calls.append((r.request.method, r.url, r.request.post_data)) - pg.on('response', on_resp) - try: - pg.goto(u, wait_until='domcontentloaded', timeout=60000) - pg.wait_for_timeout(9000) - pg.mouse.wheel(0, 5000); pg.wait_for_timeout(4000) - out.write('== %s\n' % k) - for m, url, pd in calls: - if any(w in url.lower() for w in ('search', 'job', 'graphql', 'widget', 'role')): - out.write(' %s %s\n' % (m, url[:190])) - if pd: out.write(' POST %s\n' % pd[:600]) - except Exception as e: - out.write('== %s ERR %s\n' % (k, str(e)[:150])) - pg.close() - b.close() -out.close() diff --git a/job_scout/_mt.py b/job_scout/_mt.py deleted file mode 100644 index 8216c0b..0000000 --- a/job_scout/_mt.py +++ /dev/null @@ -1,21 +0,0 @@ -import io -from playwright.sync_api import sync_playwright -out = io.open('_mt_out.txt', 'w', encoding='utf-8') -with sync_playwright() as p: - b = p.chromium.launch() - pg = b.new_page() - pg.goto('https://www.metacareers.com/jobs?offices[0]=Zurich%2C%20Switzerland', - wait_until='domcontentloaded', timeout=60000) - pg.wait_for_timeout(9000) - for _ in range(6): - pg.mouse.wheel(0, 5000); pg.wait_for_timeout(1200) - n = pg.locator("a[href*='/jobs/']").count() - n2 = pg.locator("a[href*='/profile/job_details/']").count() - out.write('a[/jobs/]=%d a[/profile/job_details/]=%d\n' % (n, n2)) - for i in range(min(n, 30)): - el = pg.locator("a[href*='/jobs/']").nth(i) - out.write(' %s | %s\n' % ((el.inner_text() or '').replace('\n', ' / ')[:90], - (el.get_attribute('href') or '')[:60])) - out.write('---TEXT---\n' + pg.inner_text('body')[:900]) - b.close() -out.close()