atsfill 0.0.4.dev85__tar.gz → 0.0.4.dev86__tar.gz

{atsfill-0.0.4.dev85 → atsfill-0.0.4.dev86}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: atsfill
-Version: 0.0.4.dev85
+Version: 0.0.4.dev86
 Summary: Read, dry-run and submit job application forms on hiring systems: Ashby, Greenhouse, Teamtailor, Welcome to the Jungle.
 Keywords: job-applications,ats,ashby,greenhouse,teamtailor,ai-agents,cli
 Author: shbernal

{atsfill-0.0.4.dev85 → atsfill-0.0.4.dev86}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "atsfill"
-version = "0.0.4.dev85"
+version = "0.0.4.dev86"
 description = "Read, dry-run and submit job application forms on hiring systems: Ashby, Greenhouse, Teamtailor, Welcome to the Jungle."
 readme = "README.md"
 license = "MIT"

{atsfill-0.0.4.dev85 → atsfill-0.0.4.dev86}/pyproject.toml.orig

@@ -1,6 +1,6 @@
 [project]
 name = "atsfill"
-version = "0.0.4.dev85"
+version = "0.0.4.dev86"
 description = "Read, dry-run and submit job application forms on hiring systems: Ashby, Greenhouse, Teamtailor, Welcome to the Jungle."
 readme = "README.md"
 license = "MIT"

{atsfill-0.0.4.dev85 → atsfill-0.0.4.dev86}/src/atsfill/adapters/__init__.py

@@ -22,7 +22,7 @@ from atsfill.adapters.teamtailor import Teamtailor
 from atsfill.adapters.workable import Workable
 from atsfill.adapters.wttj import WelcomeToTheJungle
 from atsfill.adapters.zohorecruit import ZohoRecruit
-from atsfill.core.adapter import Adapter
+from atsfill.core.adapter import Adapter, Filtered
 
 
 def adapters() -> list[Adapter]:
@@ -53,3 +53,8 @@ def adapters() -> list[Adapter]:
         Eightfold(),
         Phenom(),  # last: it claims a URL by its path alone
     ]
+
+
+def filters() -> list[str]:
+    """The cosmetic filters of each adapter whose system's banner the blocker's lists miss."""
+    return [line for a in adapters() if isinstance(a, Filtered) for line in a.filters]

{atsfill-0.0.4.dev85 → atsfill-0.0.4.dev86}/src/atsfill/adapters/eightfold.py

@@ -32,25 +32,9 @@ QUESTIONS = "/api/application/v2/questions?pids={pid}&user_mode=logged_out_candi
 CSRF = re.compile(r'<meta name="_csrf" content="([^"]+)"')
 SUBMIT = "[data-test-id=submitApplicationButton]"
 AGREE = "#confirmUploadResume"  # the privacy dialog a resume upload opens
-COOKIES = [
-    "#onetrust-reject-all-handler",
-    "#truste-consent-required",
-    "#didomi-notice-disagree-button",
-]
-# An employer's own banner, declined by its button's name.
-BANNERS = ", ".join(
-    f"[{attribute}*={word} i]"
-    for word in ("cookie", "consent", "truste")
-    for attribute in ("id", "class")
-)
-DECLINE = re.compile(
-    r"^\s*((reject|decline|refuse|deny)( all)?( cookies)?|(required|necessary|essential) only)\s*$",
-    re.IGNORECASE,
-)
 PICKER = "[data-testid=picker-dialog]"
 MAX_DEPTH = 5  # the deepest chain of questions rendering one another that a label names"
 FORM_TIMEOUT_MS = 30_000
-COOKIE_TIMEOUT_MS = 5_000
 SHOWN_TIMEOUT_MS = 2_000
 OPTIONS_TIMEOUT_MS = 5_000
 UPLOAD_TIMEOUT_MS = 30_000
@@ -221,7 +205,6 @@ class Eightfold:
         if "status" in api:
             raise Unsupported(f"posting_gone: the questions API answered {api['status']}")
         self._controls = _form(json.loads(api["text"])["data"])
-        _decline_cookies(page)
         doms = page.evaluate(_DOM_IDS, [[c.field.id, c.ends] for c in self._controls])
         for control in self._controls:
             control.dom = doms.get(control.field.id)
@@ -239,7 +222,6 @@ class Eightfold:
     def fill(self, page: Page, answers: dict[str, Answer]) -> None:
         # the resume first: what its parse finds is written over the fields filled before it
         order = sorted(self._controls, key=lambda c: not c.resume)
-        _decline_cookies(page, 0)  # a banner shown since the read
         given: dict[str, Answer] = {}
         for control in order:
             if (answer := answers.get(control.field.id)) is None:
@@ -252,12 +234,7 @@ class Eightfold:
                 control.dom = _find(page, control)
                 if control.dom is None:
                     continue
-            try:
-                _fill(page, control, answer)
-            except browser.Covered:  # a banner that showed late
-                page.keyboard.press("Escape")
-                _decline_cookies(page)
-                _fill(page, control, answer)
+            _fill(page, control, answer)
             given[control.field.id] = answer
 
     def read_back(self, page: Page) -> dict[str, Answer]:
@@ -408,20 +385,6 @@ def _control(q: dict[str, Any], by_id: dict[str, dict[str, Any]]) -> _Control:
     return _Control(question, kind, ends, parents, q.get("type") == "resumeUpload")
 
 
-def _decline_cookies(page: Page, wait_ms: int = COOKIE_TIMEOUT_MS) -> None:
-    """Declines the employer's cookie banner, which covers the form's lower half."""
-    decline = page.locator(", ".join(COOKIES)).or_(
-        page.locator(BANNERS).get_by_role("button", name=DECLINE)
-    )
-    if wait_ms:  # a timeout of 0 waits for ever
-        with suppress(Timeout):
-            decline.first.wait_for(timeout=wait_ms)
-    for button in decline.all():
-        if button.is_visible():
-            button.click()
-            return
-
-
 def _find(page: Page, control: _Control) -> str | None:
     return page.evaluate(_DOM_IDS, [[control.field.id, control.ends]]).get(control.field.id)
 

{atsfill-0.0.4.dev85 → atsfill-0.0.4.dev86}/src/atsfill/adapters/hiringthing.py

@@ -46,7 +46,6 @@ PROPS = "[data-react-class='HiringThing.Components.ApplyButtonGroup']"
 APPLY = "button.job-apply-button"
 FORM = "#job-application-form"
 SUBMIT = f"{FORM} button[type=submit]"
-BANNER = ".cookie-banner"
 UPLOAD = "/client_api/add_file"
 FORM_TIMEOUT_MS = 30_000
 PAGE_TIMEOUT_MS = 30_000
@@ -164,7 +163,6 @@ class HiringThing:
             props.wait_for(state="attached", timeout=PAGE_TIMEOUT_MS)
         except Timeout:
             raise Unsupported("posting_gone: no application on the page") from None
-        _hide_banner(page)
         form = page.locator(FORM)
         if not form.count():
             page.locator(APPLY).filter(visible=True).first.click()
@@ -184,7 +182,6 @@ class HiringThing:
         ]
 
     def fill(self, page: Page, answers: dict[str, Answer]) -> None:
-        _hide_banner(page)
         form = page.locator(FORM)
         # the state last: its options are the chosen country's
         for field_id, kind in sorted(self._kinds.items(), key=lambda k: k[1] == "region"):
@@ -278,12 +275,6 @@ def _field(control: dict, questions: dict[str, str], options: list[str] | None)
     return Field(field_id, label, TYPES[kind], required, options, max_length)
 
 
-def _hide_banner(page: Page) -> None:
-    """Hides the cookie banner, which covers the form's bottom, without answering it."""
-    with suppress(Exception):
-        page.locator(BANNER).evaluate_all("es => es.forEach(e => e.style.display = 'none')")
-
-
 def _combobox(page: Page, field_id: str) -> Locator:
     return page.locator(f"{FORM} input[role=combobox][id='{field_id}']")
 

{atsfill-0.0.4.dev85 → atsfill-0.0.4.dev86}/src/atsfill/adapters/jazzhr.py

@@ -19,13 +19,11 @@ from atsfill.core.results import Answer, Field, FieldType, SideEffect
 HOST = re.compile(r"^[\w-]+\.applytojob\.com$")
 JOB = re.compile(r"^/apply/([A-Za-z0-9]{10})\b")
 FORM = "#form_submit_new_resume"
-COOKIES = "#tracking-consent-banner .reject"
 CAPTCHA = "#resumator-recaptcha iframe[title=reCAPTCHA]"
 CHALLENGE = "iframe[src*='recaptcha/api2/bframe']"
 RESUME_TYPES = [".pdf", ".doc", ".docx", ".odt", ".txt", ".rtf"]  # as the page states them
 YES_NO = ["Yes", "No"]
 FORM_TIMEOUT_MS = 20_000
-COOKIE_TIMEOUT_MS = 3_000
 CAPTCHA_TIMEOUT_MS = 20_000
 SUBMIT_TIMEOUT_MS = 90_000
 
@@ -141,7 +139,6 @@ class JazzHR:
             form.wait_for(state="attached", timeout=FORM_TIMEOUT_MS)
         except Timeout:
             raise Unsupported("posting_gone: no application form on the page") from None
-        _reject_cookies(page)
         fields, self._fields = [], {}
         for raw in form.evaluate(_READ):
             fields.append(_field(raw))
@@ -149,7 +146,6 @@ class JazzHR:
         return fields
 
     def fill(self, page: Page, answers: dict[str, Answer]) -> None:
-        _reject_cookies(page, wait=False)
         form = page.locator(FORM)
         for field_id, answer in answers.items():
             values = [answer] if isinstance(answer, str) else answer
@@ -247,13 +243,3 @@ def _field(raw: dict[str, Any]) -> Field:
     return Field(
         raw["id"], raw["label"], field_type, raw["required"], options, raw.get("maxLength"), accept
     )
-
-
-def _reject_cookies(page: Page, wait: bool = True) -> None:
-    """Rejects the tracking banner, which shows a moment after the page, over its foot."""
-    reject = page.locator(COOKIES)
-    with suppress(Timeout):
-        if wait:
-            reject.wait_for(timeout=COOKIE_TIMEOUT_MS)
-        if reject.is_visible():
-            reject.click()

{atsfill-0.0.4.dev85 → atsfill-0.0.4.dev86}/src/atsfill/adapters/oracle.py

@@ -34,7 +34,6 @@ LOADING = ".app-loading-spinner, .cx-select--loading, .apply-flow-block .skeleto
 OTP_PAGE = ".apply-flow-pin-code-required"
 CAPTCHA = "iframe[src*='hcaptcha.com']"
 MANUALLY = ".quick-apply-flow__manually--link:visible"  # past an offer to import a profile
-COOKIES = ".cookie-consent button:visible"
 ATTACHED = ".attachment-upload-button__download, .misc-attachment-list__item-name"
 STEP_TIMEOUT_MS = 30_000
 LIST_TIMEOUT_MS = 3_000
@@ -237,6 +236,7 @@ _READ_BACK = """(ids) => {
 class Oracle:
     name = "oracle"
     known_gaps = {"entries": Gap.RENDERING}
+    filters = ["##.cookie-consent-modal"]  # some tenants' banner, over the email step and the form
     side_effects = [
         SideEffect(
             "load",
@@ -501,16 +501,12 @@ class Oracle:
 
 
 def _click(page: Page, target: Locator) -> None:
-    """Clicks, once more after what came over the target: the app's loading spinner, the
-    cookie banner (declined), or a header fixed over the top of the page."""
+    """Clicks, once more after what came over the target: the app's loading spinner, or a
+    header fixed over the top of the page."""
     try:
         browser.click(target)
-    except browser.Covered as e:
-        if "cookie-consent" in str(e):
-            buttons = page.locator(COOKIES)
-            decline = buttons.filter(has_text=re.compile("decline|reject|necessary", re.I))
-            browser.click((decline if decline.count() else buttons).first)
-        elif "loading" in str(e):
+    except browser.Overlaid as e:
+        if "loading" in str(e):
             _settle(page)
         else:  # a header fixed over the top of the page
             target.evaluate("e => e.scrollIntoView({block: 'center'})")

{atsfill-0.0.4.dev85 → atsfill-0.0.4.dev86}/src/atsfill/adapters/paylocity.py

@@ -31,7 +31,6 @@ SUBMIT = "[data-automation-id=btnSubmit]"
 STEP = "#appDetailDiv .type-bold.float-right"  # "Step 2 of 5"
 FORCED_RESUME = ".modal.in #btn-forceResume"
 FORM_TIMEOUT_MS = 30_000
-COOKIE_TIMEOUT_MS = 5_000
 STEP_TIMEOUT_MS = 20_000
 UPLOAD_TIMEOUT_MS = 30_000
 SUBMIT_TIMEOUT_MS = 90_000
@@ -354,7 +353,6 @@ class Paylocity:
     def fill(self, page: Page, answers: dict[str, Answer]) -> None:
         self._values = {}
         page.locator("form, #app").first.wait_for(timeout=FORM_TIMEOUT_MS)
-        _dismiss_cookies(page)
         if "resume" in answers:  # first: a parsed resume fills fields the answers then override
             _upload_resume(page, _one(answers["resume"]))
         while True:
@@ -524,13 +522,6 @@ def _present(page: Page, control: Control) -> bool:
             return target.count() == 1 and target.is_visible()
 
 
-def _dismiss_cookies(page: Page) -> None:
-    close = page.locator("#onetrust-banner-sdk .onetrust-close-btn-handler")
-    with suppress(Timeout):
-        close.first.wait_for(state="visible", timeout=COOKIE_TIMEOUT_MS)
-        close.first.click()
-
-
 def _upload_resume(page: Page, path: str) -> None:
     forced = page.locator(FORCED_RESUME)
     if forced.count():

{atsfill-0.0.4.dev85 → atsfill-0.0.4.dev86}/src/atsfill/adapters/phenom.py

@@ -23,7 +23,7 @@ from urllib.parse import parse_qs, urlsplit
 
 from atsfill.core import browser, matching
 from atsfill.core.adapter import Gap, NotTaken, Unsupported, known_gap
-from atsfill.core.browser import HTTP_TIMEOUT_S, BrowserError, Covered, Locator, Page, Timeout
+from atsfill.core.browser import HTTP_TIMEOUT_S, BrowserError, Locator, Overlaid, Page, Timeout
 from atsfill.core.results import Answer, Field, FieldType, SideEffect
 
 JOB = re.compile(r"^(/[a-z]{2,8}/[a-z]{2})/(job|\w*apply)(?:/|$)")
@@ -141,22 +141,6 @@ _DIAL = """form => {
   return (text.match(/\\(\\+(\\d+)/) || (el.value || '').match(/_(\\d+)$/) || [])[1] || null;
 }"""
 
-# Declines, else accepts, the cookie banner: it sits over the form's controls.
-_COOKIES = """() => {
-  const banners = [...document.querySelectorAll('[ph-module=gdpr], [id*=cookie i],'
-    + ' [class*=cookie i], [id*=truste i], [id*=onetrust i], [class*=consent i], .cc-window,'
-    + ' [role=dialog][aria-label*=cookie i]')];
-  const buttons = banners.flatMap(b => [...b.querySelectorAll('button, a[role=button], a')])
-    .filter(b => b.offsetWidth || b.offsetHeight);
-  const pick = buttons.find(b => /^\\W*(deny|decline|reject)/i.test(b.innerText.trim()))
-    || buttons.find(b => /^\\W*(allow|accept|agree|i consent|ok|got it|close|dismiss)/i
-      .test(b.innerText.trim() || b.getAttribute('aria-label') || ''));
-  if (pick) { pick.click(); return true; }
-  for (const b of banners)  // no button to answer it: out of the way, unanswered
-    if (getComputedStyle(b).position === 'fixed') b.style.display = 'none';
-  return false;
-}"""
-
 # Picks a dropdown's option by its text, as React sees a change, even to the value it holds.
 _SELECT = """(el, text) => {
   const option = [...el.options].find(o => o.text.trim() === text);
@@ -1128,11 +1112,9 @@ def _next(page: Page, number: int) -> None:
 
 
 def _click(page: Page, target: Locator) -> None:
-    """Clicks once the page's loading overlay is gone, past the cookie banner, which can show
-    at any time and covers the form."""
+    """Clicks once the page's loading overlay is gone."""
     with suppress(Timeout):
         page.locator(LOADING).first.wait_for(state="hidden", timeout=STEP_TIMEOUT_MS)
-    page.evaluate(_COOKIES)
     browser.click(target)
 
 
@@ -1309,10 +1291,9 @@ def _fill(page: Page, raw: dict[str, Any], answer: Answer) -> None:
                 try:
                     _click(page, item.first if item.count() else items.first)
                     break
-                except BrowserError, Covered:  # rendered again under the click
+                except BrowserError, Overlaid:  # rendered again under the click
                     continue
         case "radio":  # its label wraps it or names it by `for`
-            page.evaluate(_COOKIES)
             if not form.evaluate(_RADIO, [raw["id"], _one(answer)]):
                 raise NotTaken(f"{raw['id']} has no option {_one(answer)!r}")
         case "checkbox":

{atsfill-0.0.4.dev85 → atsfill-0.0.4.dev86}/src/atsfill/adapters/pinpoint.py

@@ -34,8 +34,6 @@ FORM_TIMEOUT_MS = 30_000
 HYDRATE_TIMEOUT_MS = 30_000
 OPTION_TIMEOUT_MS = 5_000
 SHOWN_TIMEOUT_MS = 2_000
-COOKIE_TIMEOUT_MS = 5_000
-ONETRUST = "#onetrust-consent-sdk"
 US_STATE = ".react-select:has(input[type=hidden][name$='[town]'])"  # the town's place in the US
 UPLOAD_TIMEOUT_MS = 30_000
 SUBMIT_TIMEOUT_MS = 90_000
@@ -145,7 +143,6 @@ class Pinpoint:
         return [f for f, _ in listed]
 
     def fill(self, page: Page, answers: dict[str, Answer]) -> None:
-        _decline_cookies(page)
         form = page.locator(FORM)
         for f in self._fields:
             if f.id not in answers:
@@ -236,16 +233,6 @@ def _hydrated(page: Page) -> None:
         )
 
 
-def _decline_cookies(page: Page) -> None:
-    """Declines an employer's OneTrust banner, whose backdrop covers the form."""
-    if not page.locator(ONETRUST).count():
-        return
-    reject = page.locator(f"{ONETRUST} #onetrust-reject-all-handler")
-    with suppress(Timeout):
-        reject.wait_for(state="visible", timeout=COOKIE_TIMEOUT_MS)
-        reject.click()
-
-
 def _shown(root: Locator) -> bool:
     """Whether a component shows a control, waiting a moment for one an earlier answer opens."""
     try:

{atsfill-0.0.4.dev85 → atsfill-0.0.4.dev86}/src/atsfill/adapters/rippling.py

@@ -39,7 +39,6 @@ SUBMIT_TIMEOUT_MS = 90_000
 DIAL_CODES = (3, 2, 1)  # the lengths a dial code can have, longest tried first
 FILE_TYPES = [".doc", ".docx", ".pdf"]
 OTHER = "Other"
-COOKIES = "#transcend-consent-manager"
 TESTIDS = {"pronouns": "pronouns_strategy"}  # a field whose element is not named by its id
 
 # A standard question's type, by its `fieldType`; the location is a place typeahead.
@@ -160,6 +159,7 @@ _SHOWN = """(root, [id, nth, testids]) => {
 class Rippling:
     name = "rippling"
     known_gaps: dict[str, Gap] = {}
+    filters = ["ats.rippling.com###transcend-consent-manager"]  # Transcend's banner
     side_effects = [
         SideEffect("upload", "hiring_system", "an attached file is uploaded, a résumé parsed"),
         SideEffect("submit", "employer", "the application reaches the employer"),
@@ -199,7 +199,6 @@ class Rippling:
             page.locator("[data-testid=Apply]").wait_for(timeout=FORM_TIMEOUT_MS)
         except Timeout:
             raise Unsupported("posting_gone: no application form on the page") from None
-        _hide_cookies(page)
         data = page.locator("#__NEXT_DATA__").text_content() or "{}"
         out = []
         for field in _form(_api(json.loads(data))):
@@ -340,13 +339,6 @@ def _custom(field_id: str, q: dict[str, Any]) -> list[Field]:
     return out
 
 
-def _hide_cookies(page: Page) -> None:
-    """Hides the cookie banner, whose layer takes the clicks on the lower part of the page: it
-    renders in a closed shadow root, out of reach of a click, and shows a moment after the
-    form. Nothing is answered to it."""
-    page.add_style_tag(content=f"{COOKIES} {{display: none !important}}")
-
-
 def _click(target: Locator) -> None:
     """Clicks the target scrolled to the middle of the window, clear of the Apply bar fixed
     at its bottom."""

{atsfill-0.0.4.dev85 → atsfill-0.0.4.dev86}/src/atsfill/adapters/smartrecruiters.py

@@ -27,7 +27,6 @@ CONTAINER = f"{QUESTIONS} [data-test=question-container]"
 NEXT = "[data-test=footer-next] button"
 SUBMIT = "[data-test=footer-submit] button"
 FORM_TIMEOUT_MS = 30_000
-COOKIE_TIMEOUT_MS = 5_000
 OPTION_TIMEOUT_MS = 10_000
 UPLOAD_TIMEOUT_MS = 30_000
 SUBMIT_TIMEOUT_MS = 90_000
@@ -147,7 +146,6 @@ class SmartRecruiters:
             page.locator(PROFILE[0][1]).wait_for(timeout=FORM_TIMEOUT_MS)
         except Timeout:
             raise Unsupported("posting_gone: no application form on the page") from None
-        _decline_cookies(page)
         context = page.evaluate(_CONTEXT) or {}
         screening = context.get("screeningConfiguration") or {}
         self._questions = {q["id"]: q for q in screening.get("questions") or []}
@@ -170,7 +168,6 @@ class SmartRecruiters:
         return fields
 
     def fill(self, page: Page, answers: dict[str, Answer]) -> None:
-        _decline_cookies(page)
         for field_id, selector, _ in PROFILE:
             if field_id in answers:
                 _fill_profile(page, field_id, selector, answers[field_id])
@@ -315,13 +312,6 @@ def _plain(text: str) -> str:
     return re.sub(r"\s+", " ", text.replace("*", "")).strip()
 
 
-def _decline_cookies(page: Page) -> None:
-    reject = page.locator("#onetrust-reject-all-handler")
-    with suppress(Timeout):
-        reject.wait_for(state="visible", timeout=COOKIE_TIMEOUT_MS)
-        reject.click()
-
-
 def _profile_field(page: Page, field_id: str, selector: str, host: str) -> Field | None:
     element = page.locator(host).first
     if not page.locator(selector).count() or not element.count():

{atsfill-0.0.4.dev85 → atsfill-0.0.4.dev86}/src/atsfill/adapters/teamtailor.py

@@ -221,7 +221,6 @@ class Teamtailor:
     def fill(self, page: Page, answers: dict[str, Answer]) -> None:
         form = page.locator(FORM)
         self._videos = {}
-        _decline_cookies(page)
         for field_id in [i for i in answers if i in self._qualifying]:
             self._fill(form, field_id, answers[field_id])
         if form.locator(LOCKED).count():
@@ -252,7 +251,7 @@ class Teamtailor:
         return {k: v for k, v in values.items() if v not in (None, "", [])}
 
     def submit(self, page: Page) -> str | None:
-        _click(page, page.locator(f"{FORM} [type=submit]"))
+        browser.click(page.locator(f"{FORM} [type=submit]"))
         page.wait_for_url(re.compile(r"/applications/[^/]+/thanks"), timeout=SUBMIT_TIMEOUT_MS)
         return page.url
 
@@ -266,17 +265,17 @@ class Teamtailor:
             case FieldType.CHOICE, str() if host.locator("input[type=checkbox]").count():
                 box = host.locator("input[type=checkbox]")
                 if box.is_checked() != (answer == "Yes"):
-                    _click(page, page.locator(f"label[for='{box.get_attribute('id')}']"))
+                    browser.click(page.locator(f"label[for='{box.get_attribute('id')}']"))
             case FieldType.CHOICE, str() if host.locator("input[type=range]").count():
                 host.locator("input[type=range]").fill(answer)
             case FieldType.TYPEAHEAD, str():
-                _suggest(page, field, host, answer)
+                _suggest(field, host, answer)
             case FieldType.CHOICE | FieldType.MULTI_CHOICE, _:
                 for value in [answer] if isinstance(answer, str) else answer:
-                    _pick(page, host, value)
+                    _pick(host, value)
                 trigger = host.locator("button[aria-haspopup][aria-expanded=true]")
                 if trigger.count():  # a multiple choice menu stays open over the next fields
-                    _click(page, trigger.first)
+                    browser.click(trigger.first)
             case FieldType.TEXT, str() if host.locator("input[type=number]").count():
                 if re.fullmatch(r"-?\d+(\.\d+)?", answer):  # else left empty: a mismatch
                     host.locator("input[type=number]").first.fill(answer)
@@ -374,23 +373,7 @@ _VIDEO = """url => fetch(url, {headers: {Accept: "application/json"}})
   .then(r => r.ok ? r.json() : {}).then(v => v.file_file_name ?? null)"""
 
 
-def _decline_cookies(page: Page) -> None:
-    """Closes the cookie banner, a modal dialog that leaves the form inert while it is open."""
-    decline = page.locator("[data-action*='cookies--alert#disableAll']").first
-    if decline.is_visible():
-        decline.click()
-
-
-def _click(page: Page, target: Locator) -> None:
-    """Clicks, first declining the cookie banner when it covers the target: it comes back."""
-    try:
-        browser.click(target)
-    except browser.Covered:
-        _decline_cookies(page)
-        browser.click(target)
-
-
-def _suggest(page: Page, field: Field, host: Locator, answer: str) -> None:
+def _suggest(field: Field, host: Locator, answer: str) -> None:
     """Types the address key by key, which loads the suggestions, and picks the first that
     matches it: the form keeps an address only once a suggestion is picked."""
     box = host.locator("input[role=combobox]")
@@ -400,20 +383,20 @@ def _suggest(page: Page, field: Field, host: Locator, answer: str) -> None:
     options.first.wait_for()
     for i in range(options.count()):
         if matching.matches(field, answer, options.nth(i).get_attribute("data-description") or ""):
-            _click(page, options.nth(i))
+            browser.click(options.nth(i))
             return
     raise NotTaken(f"no suggestion for {answer!r} in {field.id}")
 
 
-def _pick(page: Page, question: Locator, value: str) -> None:
+def _pick(question: Locator, value: str) -> None:
     """Checks the option whose text is `value`, a radio or checkbox label or a dropdown item."""
     label = question.locator("label:visible").filter(
         has_text=re.compile(rf"^\s*{re.escape(value)}\s*$")
     )
     if label.count():
-        _click(page, label.first)
+        browser.click(label.first)
         return
     trigger = question.locator("button[aria-haspopup]").first
     if trigger.get_attribute("aria-expanded") != "true":
-        _click(page, trigger)
-    _click(page, question.locator(f"[role^=menuitem][data-search-text='{value}']").first)
+        browser.click(trigger)
+    browser.click(question.locator(f"[role^=menuitem][data-search-text='{value}']").first)

{atsfill-0.0.4.dev85 → atsfill-0.0.4.dev86}/src/atsfill/adapters/workable.py

@@ -24,10 +24,8 @@ JOB = re.compile(r"^(?:/[^/]+)?/j/([0-9A-F]{10})\b")
 API = "https://apply.workable.com/api/v1/jobs/{code}/form"
 APPLIED = re.compile(r"/api/v1/jobs/[0-9A-F]{10}/apply$")
 FORM = "form[data-ui=application-form]"
-COOKIES = "[data-ui=cookie-consent-decline]"
 UPLOADED = "[data-action=delete]"  # an uploading file shows its name and an abort button
 FORM_TIMEOUT_MS = 30_000
-COOKIE_TIMEOUT_MS = 5_000
 OPTIONS_TIMEOUT_MS = 5_000
 CHECK_TIMEOUT_MS = 3_000
 UPLOAD_TIMEOUT_MS = 30_000
@@ -184,7 +182,6 @@ class Workable:
             form.wait_for(timeout=FORM_TIMEOUT_MS)
         except Timeout:
             raise Unsupported("posting_gone: no application form on the page") from None
-        _decline_cookies(page)
         fields, self._kinds, self._formats = [], {}, {}
         for raw in form.evaluate(_READ):
             fields.append(self._field(page, raw))
@@ -194,7 +191,6 @@ class Workable:
         return fields
 
     def fill(self, page: Page, answers: dict[str, Answer]) -> None:
-        _decline_cookies(page, wait=False)
         # attaching the resume empties the photo, so it goes first
         for field_id, answer in sorted(answers.items(), key=lambda a: a[0] != "resume"):
             values = [answer] if isinstance(answer, str) else answer
@@ -340,17 +336,6 @@ def _code(url: str) -> str | None:
     return match[1] if match else None
 
 
-def _decline_cookies(page: Page, wait: bool = True) -> None:
-    """Declines the cookie banner, a modal dialog that leaves the form inert while it is open;
-    it shows a moment after the page."""
-    decline = page.locator(COOKIES)
-    with suppress(Timeout):
-        if wait:
-            decline.wait_for(timeout=COOKIE_TIMEOUT_MS)
-        if decline.is_visible():
-            decline.click()
-
-
 def _exact(value: str) -> re.Pattern[str]:
     return re.compile(rf"^\s*{re.escape(value)}\s*$")
 

{atsfill-0.0.4.dev85 → atsfill-0.0.4.dev86}/src/atsfill/adapters/wttj.py

@@ -106,7 +106,7 @@ class WelcomeToTheJungle:
         if not form.is_visible():  # an easyapply link opens the form by itself
             try:
                 browser.click(apply)
-            except browser.Covered:
+            except browser.Overlaid:
                 if not form.is_visible():
                     raise
         try:

{atsfill-0.0.4.dev85 → atsfill-0.0.4.dev86}/src/atsfill/adapters/zohorecruit.py

@@ -28,12 +28,9 @@ POSTING = "https://{host}/recruit/v2/public/Job_Openings/{job}"
 API = POSTING + "/apply/{form}?pagename={page}&source=CareerSite"
 WEBFORM = re.compile(r'"webform_id":"(\d+)"')
 APPLY = "#detail-page-applybtn button"
-COOKIES = "button.cookie-decline-btn"
 FORM = "rec-form-component[webformid]"
 SUBMIT = "#cw-submit-btn button"
 FORM_TIMEOUT_MS = 30_000
-COOKIE_TIMEOUT_MS = 5_000
-COOKIE_ATTEMPTS = 3
 SHOWN_TIMEOUT_MS = 2_000
 TAG_TIMEOUT_MS = 2_000
 TAG_ATTEMPTS = 4
@@ -131,6 +128,7 @@ class ZohoRecruit:
     name = "zohorecruit"
     known_gaps: dict[str, Gap] = {}
     side_effects = [SideEffect("submit", "employer", "the application reaches the employer")]
+    filters = ["##career-cookie-consent"]  # its banner, on Zoho's domains and employers' own
 
     def __init__(self) -> None:
         self._controls: list[_Control] = []  # in the form's order
@@ -168,7 +166,7 @@ class ZohoRecruit:
         html = page.evaluate(_FETCH, page.url)
         if "status" in html or (form := WEBFORM.search(html["text"])) is None:
             raise Unsupported("posting_gone: the job page names no application form")
-        _open(page, apply)
+        browser.click(apply)
         page.locator(SUBMIT).wait_for(timeout=FORM_TIMEOUT_MS)
         api = page.evaluate(_FETCH, _api(page.url, form[1]))
         if "status" in api:
@@ -260,23 +258,6 @@ def _get(url: str) -> str:
         return response.read().decode()
 
 
-def _open(page: Page, apply) -> None:
-    """Clicks "I'm interested", declining the cookie banner first: its layer takes every
-    click while it shows, and it shows a moment after the page, at times after a while."""
-    decline = page.locator(COOKIES)
-    for attempt in range(COOKIE_ATTEMPTS):
-        with suppress(Timeout):
-            decline.wait_for(timeout=COOKIE_TIMEOUT_MS if attempt == 0 else 0)
-            decline.click()
-        try:
-            browser.click(apply)
-            return
-        except browser.Covered:
-            if attempt == COOKIE_ATTEMPTS - 1:
-                raise
-            decline.wait_for(timeout=COOKIE_TIMEOUT_MS)
-
-
 def _form(api: dict[str, Any]) -> tuple[list[_Control], bool]:
     """Each field of the form, in its order, and whether it asks for a CAPTCHA. A repeatable
     section is read as its first entry; the pre-screening questions are read when the form

{atsfill-0.0.4.dev85 → atsfill-0.0.4.dev86}/src/atsfill/core/adapter.py

@@ -84,6 +84,14 @@ class Adapter(Protocol):
         ...
 
 
+@runtime_checkable
+class Filtered(Protocol):
+    """An adapter whose system shows a cookie banner the blocker's lists miss: its own
+    uBlock Origin cosmetic filters hide it (ADR 0004)."""
+
+    filters: list[str]
+
+
 @runtime_checkable
 class NeedsAccount(Protocol):
     """An adapter whose system needs a signed-in session, one per user and account scope

{atsfill-0.0.4.dev85 → atsfill-0.0.4.dev86}/src/atsfill/core/browser.py

@@ -5,6 +5,8 @@ import fcntl
 import json
 import locale
 import platform
+import re
+import shutil
 import subprocess
 import sys
 import threading
@@ -13,8 +15,11 @@ from collections.abc import Generator
 from concurrent.futures import Future
 from contextlib import contextmanager
 from pathlib import Path
+from urllib.parse import urljoin
+from urllib.request import Request as HttpRequest
+from urllib.request import urlopen
 
-from camoufox.addons import DefaultAddons, maybe_download_addons
+from camoufox.addons import ADDONS_DIR, DefaultAddons, get_addon_path, maybe_download_addons
 from camoufox.exceptions import CamoufoxNotInstalled
 from camoufox.fingerprints import Screen, generate_fingerprint
 from camoufox.pkgman import installed_verstr
@@ -27,7 +32,7 @@ from playwright.sync_api import TimeoutError as Timeout
 __all__ = [
     "BUILD",
     "BrowserError",
-    "Covered",
+    "Overlaid",
     "Locator",
     "OutOfTime",
     "Page",
@@ -50,14 +55,22 @@ BUDGET = 180.0  # seconds per command, lock wait included
 LOCK_POLL_S = 0.1
 HTTP_TIMEOUT_S = 20  # per request to a hiring system's API
 BUILD = "156.0.1-beta.36"  # the Camoufox browser build, pinned so results compare across machines
+BLOCKER = ADDONS_DIR / "UBO-annoyances"  # Camoufox's uBlock Origin, its annoyance lists on
+BLOCKER_LISTS = "annoyances"  # the group of uBlock Origin lists: cookie banners, pop-ups, widgets
+# The adapters' filters for the banners the lists miss: a list of the blocker's, on like them.
+FILTERS = "assets/atsfill/filters.txt"
+INCLUDE = re.compile(r"^!#include (\S+)$", re.M)
+# A page holding a banner the lists hide: until they do, the blocker is still starting.
+PROBE = ("https://blocker.invalid/", '<div id="cookie-banner">probe</div>')
+BLOCKER_TIMEOUT_MS = 30_000  # its lists take seconds to compile on each launch, more under load
 
 
 class OutOfTime(Exception):
     """The command's time budget ran out."""
 
 
-class Covered(Exception):
-    """Another element is on top of the click target; the message describes it."""
+class Overlaid(Exception):
+    """An overlay is on top of the click target; the message describes it."""
 
 
 def version() -> str:
@@ -77,9 +90,63 @@ def install() -> int:
         print(f"the Camoufox build {BUILD} is not installed", file=sys.stderr)
         return 1
     maybe_download_addons(list(DefaultAddons))
+    try:
+        _build_blocker()
+    except OSError as e:
+        print(f"the annoyance lists could not be fetched: {e}", file=sys.stderr)
+        return 1
     return 0
 
 
+def _build_blocker() -> None:
+    """Copies Camoufox's uBlock Origin to `BLOCKER` with only its annoyance lists on, read
+    from disk: fetched by each launch, they would hold its first page for seconds, and each
+    list on is compiled by each launch before the blocker starts."""
+    ADDONS_DIR.mkdir(parents=True, exist_ok=True)
+    with (ADDONS_DIR / f"{BLOCKER.name}.lock").open("w") as f:
+        fcntl.flock(f, fcntl.LOCK_EX)
+        build = BLOCKER.with_name(f"{BLOCKER.name}.build")
+        shutil.rmtree(build, ignore_errors=True)
+        shutil.copytree(get_addon_path(DefaultAddons.UBO.name), build)
+        assets = build / "assets" / "assets.json"
+        entries = json.loads(assets.read_text())
+        for key, entry in entries.items():
+            if entry.get("group") != BLOCKER_LISTS:
+                if entry.get("content") == "filters":  # ads and trackers: compiled for nothing
+                    entry["off"] = True
+                continue
+            urls = entry["contentURL"]
+            urls = [urls] if isinstance(urls, str) else urls
+            local = f"assets/{BLOCKER_LISTS}/{key}.txt"
+            (build / local).parent.mkdir(exist_ok=True)
+            (build / local).write_text(_filter_list([*urls, *entry.get("cdnURLs", [])]))
+            entry.pop("off", None)  # on from the start, the way a fresh profile reads it
+            entry["contentURL"] = [local, *urls]
+        entries["atsfill"] = {"content": "filters", "group": "custom", "title": "atsfill",
+                              "contentURL": [FILTERS]}  # fmt: skip
+        (build / FILTERS).parent.mkdir()
+        (build / FILTERS).touch()
+        assets.write_text(json.dumps(entries))
+        shutil.rmtree(BLOCKER, ignore_errors=True)
+        build.rename(BLOCKER)
+
+
+def _filter_list(urls: list[str]) -> str:
+    """A filter list from the first of its sources that answers, its includes inlined."""
+    for url in (u for u in urls if u.startswith("https://")):
+        try:
+            return INCLUDE.sub(lambda m, base=url: _fetch(urljoin(base, m[1])), _fetch(url))
+        except OSError:
+            continue
+    raise OSError(f"no source answered for {urls[0]}")
+
+
+def _fetch(url: str) -> str:
+    request = HttpRequest(url, headers={"User-Agent": "Mozilla/5.0"})  # some refuse urllib's
+    with urlopen(request, timeout=HTTP_TIMEOUT_S) as response:
+        return response.read().decode()
+
+
 def identity(data_dir: Path) -> dict:
     """The device identity every launch reuses, generated for the host OS on the first one."""
     path = data_dir / "identity.json"
@@ -117,17 +184,53 @@ def lock(data_dir: Path, deadline: float) -> Generator[None]:
 
 @contextmanager
 def _launch(data_dir: Path, headless: bool) -> Generator[Browser]:
+    if not (BLOCKER / FILTERS).exists():  # installed before the blocker, or its filters, were
+        maybe_download_addons(list(DefaultAddons))
+        _build_blocker()
+    _write_filters()
     with Camoufox(
         headless=headless,
+        addons=[str(BLOCKER)],
+        exclude_addons=list(DefaultAddons),
         fingerprint=identity(data_dir),
         os=HOST_OS,
         locale=machine_locale(),
         i_know_what_im_doing=True,
     ) as browser:
         assert isinstance(browser, Browser)  # not a persistent context
+        _wait_for_blocker(browser)
         yield browser
 
 
+def _write_filters() -> None:
+    """Writes the adapters' filters into the blocker when they changed, in one rename so that
+    a launch beside it reads either version whole."""
+    from atsfill.adapters import filters  # adapters import this module
+
+    path = BLOCKER / FILTERS
+    text = "".join(f"{line}\n" for line in filters())
+    if path.read_text() != text:
+        staged = path.with_suffix(f".{threading.get_native_id()}")
+        staged.write_text(text)
+        staged.replace(path)
+
+
+def _wait_for_blocker(browser: Browser) -> None:
+    """Returns once the blocker hides banners: until then a page shows them, and a fresh
+    browser starts on the posting itself."""
+    context = browser.new_context()
+    url, html = PROBE
+    context.route(url, lambda route: route.fulfill(content_type="text/html", body=html))
+    page = context.new_page()
+    page.goto(url)
+    try:
+        page.locator("#cookie-banner").wait_for(state="hidden", timeout=BLOCKER_TIMEOUT_MS)
+    except Timeout:
+        raise BrowserError("the blocker hid no banner: run `atsfill install-browser`") from None
+    finally:
+        context.close()
+
+
 @contextmanager
 def open_page(
     data_dir: Path, session: Path | None = None, seconds: float = BUDGET
@@ -198,7 +301,7 @@ def session_request(path: Path) -> Generator[Request]:
 # What sits on top of a point, null when the element (or the shadow host holding it) is what
 # a click there would reach. The point defaults to where Playwright clicks: the centre of the
 # first box, which for wrapped inline text is its first line, not the bounding box's centre.
-_COVER = """(el, point) => {
+_OVERLAY = """(el, point) => {
     const r = [...el.getClientRects()].find(b => b.width && b.height)
         ?? el.getBoundingClientRect();
     const [x, y] = point ?? [r.left + r.width / 2, r.top + r.height / 2];
@@ -214,12 +317,12 @@ _COVER = """(el, point) => {
 }"""
 
 
-def covered_by(target: Locator) -> str | None:
-    """What covers the target, in its own document and then at its point in each frame up."""
+def overlay_on(target: Locator) -> str | None:
+    """The overlay on the target, in its own document and then at its point in each frame up."""
     target.scroll_into_view_if_needed()
     handle = target.element_handle()
-    if cover := handle.evaluate(_COVER, None):
-        return cover
+    if overlay := handle.evaluate(_OVERLAY, None):
+        return overlay
     box = target.bounding_box()  # in the top page's viewport
     frame = handle.owner_frame()
     if box is None or frame is None:
@@ -228,14 +331,14 @@ def covered_by(target: Locator) -> str | None:
     while (parent := frame.parent_frame) is not None:
         origin = parent.frame_element().bounding_box() if parent.parent_frame else None
         point = [x - origin["x"], y - origin["y"]] if origin else [x, y]
-        if cover := frame.frame_element().evaluate(_COVER, point):
-            return cover
+        if overlay := frame.frame_element().evaluate(_OVERLAY, point):
+            return overlay
         frame = parent
     return None
 
 
 def click(target: Locator) -> None:
-    """Clicks the target, raising `Covered` when something else is on top of it."""
-    if cover := covered_by(target):
-        raise Covered(cover)
+    """Clicks the target, raising `Overlaid` when an overlay is on top of it."""
+    if overlay := overlay_on(target):
+        raise Overlaid(overlay)
     target.click()

{atsfill-0.0.4.dev85 → atsfill-0.0.4.dev86}/src/atsfill/core/commands.py

@@ -17,7 +17,7 @@ from atsfill.core.adapter import (
     SignsIn,
     Unsupported,
 )
-from atsfill.core.browser import Covered, OutOfTime, Page, open_page
+from atsfill.core.browser import OutOfTime, Overlaid, Page, open_page
 from atsfill.core.results import (
     AdapterStatus,
     Answer,
@@ -188,8 +188,8 @@ def _stops(result: Result) -> Generator[None]:
         result.status, result.reason = Status.REFUSED, str(e)
     except Blocked as e:
         result.status, result.reason = Status.BLOCKED, str(e)
-    except Covered as e:
-        result.status, result.reason = Status.BLOCKED, f"covered: {e}"
+    except Overlaid as e:
+        result.status, result.reason = Status.BLOCKED, f"overlaid: {e}"
 
 
 def _in_budget(url: str, command: Callable[[list[str]], Result]) -> Result: