atsfill 0.0.4.dev61__tar.gz → 0.0.4.dev63__tar.gz

{atsfill-0.0.4.dev61 → atsfill-0.0.4.dev63}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: atsfill
-Version: 0.0.4.dev61
+Version: 0.0.4.dev63
 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.dev61 → atsfill-0.0.4.dev63}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "atsfill"
-version = "0.0.4.dev61"
+version = "0.0.4.dev63"
 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.dev61 → atsfill-0.0.4.dev63}/pyproject.toml.orig

@@ -1,6 +1,6 @@
 [project]
 name = "atsfill"
-version = "0.0.4.dev61"
+version = "0.0.4.dev63"
 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.dev61 → atsfill-0.0.4.dev63}/src/atsfill/adapters/greenhouse.py

@@ -5,6 +5,7 @@ Nothing but an attached file reaches Greenhouse before submit, so the read-back
 input values, the shown value of each select, the phone's dial code, the attached file's name.
 """
 
+import itertools
 import json
 import re
 import time
@@ -21,6 +22,10 @@ from atsfill.core.results import Answer, Field, FieldType, SideEffect
 HOSTS = re.compile(r"^(job-boards|boards)(\.eu)?\.greenhouse\.io$")
 FORM = "#application-form"
 API = "https://boards-api.greenhouse.io/v1/boards/{board}/jobs/{job}?questions=true"
+# An education row's selects (`school--0`, …) and the endpoint listing their options a hundred
+# per page: the opened menu shows only the first page.
+EDUCATION = re.compile(r"^(school|degree|discipline)--\d+$")
+EDUCATION_API = "https://boards-api.greenhouse.io/v1/boards/{board}/education/{kind}s?page={page}"
 EMBED = "https://job-boards.greenhouse.io/embed/job_app"
 EMBED_TIMEOUT_MS = 15_000
 OPTIONS_TIMEOUT_MS = 3_000
@@ -130,6 +135,7 @@ class Greenhouse:
         self._fields: dict[str, Field] = {}
         self._kinds: dict[str, str] = {}  # how the page renders each field
         self._embed: str | None = None  # the job_app URL an employer page embeds
+        self._education_options: dict[str, list[str] | None] = {}  # by school, degree, discipline
 
     def detect(self, url: str, page: Page | None) -> bool:
         """A board URL, or an employer page whose `#grnhse_iframe` embeds a job application."""
@@ -162,11 +168,7 @@ class Greenhouse:
         if found is None:
             return None
         try:
-            request = urllib.request.Request(
-                API.format(board=found[0], job=found[1]), headers={"User-Agent": "atsfill"}
-            )
-            with urllib.request.urlopen(request, timeout=HTTP_TIMEOUT_S) as response:
-                return _api_fields(json.load(response))
+            return _api_fields(_get_json(API.format(board=found[0], job=found[1])))
         except OSError, ValueError, KeyError, TypeError:
             return None
 
@@ -236,7 +238,7 @@ class Greenhouse:
             case "checkboxes":
                 field_type = FieldType.MULTI_CHOICE
             case "select" | "multi":
-                options = self._options(page, raw["id"])
+                options = self._education(page, raw["id"]) or self._options(page, raw["id"])
                 if options is None:
                     field_type = FieldType.TYPEAHEAD
                 else:
@@ -256,6 +258,17 @@ class Greenhouse:
             raw.get("accept"),
         )
 
+    def _education(self, page: Page, field_id: str) -> list[str] | None:
+        """An education select's options, every page of them, read once per kind; None for
+        another field or when the endpoint does not answer."""
+        found = _board_job(page.url)
+        if not (found and (match := EDUCATION.match(field_id))):
+            return None
+        kind = match[1]
+        if kind not in self._education_options:
+            self._education_options[kind] = _education_options(found[0], kind)
+        return self._education_options[kind]
+
     def _options(self, page: Page, field_id: str) -> list[str] | None:
         """A select's options, from opening it; None for one that suggests as you type."""
         box = page.locator(_by_id(field_id))
@@ -302,6 +315,26 @@ def _board_job(url: str) -> tuple[str, str] | None:
     return None
 
 
+def _get_json(url: str) -> Any:
+    request = urllib.request.Request(url, headers={"User-Agent": "atsfill"})
+    with urllib.request.urlopen(request, timeout=HTTP_TIMEOUT_S) as response:
+        return json.load(response)
+
+
+def _education_options(board: str, kind: str) -> list[str] | None:
+    """Every option the board's education endpoint lists for `kind`, page after page."""
+    options: list[str] = []
+    try:
+        for number in itertools.count(1):
+            body = _get_json(EDUCATION_API.format(board=board, kind=kind, page=number))
+            options += [item["text"] for item in body["items"]]
+            if not body["items"] or len(options) >= body["meta"]["total_count"]:
+                return options or None
+    except OSError, ValueError, KeyError, TypeError:
+        return None
+    return None
+
+
 def _api_fields(job: dict[str, Any]) -> list[Field] | None:
     if any(job.get(k) for k in UNMAPPED) or any(
         c.get("requires_consent") or c.get("requires_processing_consent")
@@ -336,17 +369,23 @@ def _exact(value: str) -> re.Pattern[str]:
 
 
 def _select(page: Page, field_id: str, value: str, suggested: bool) -> None:
-    """Types into a react-select and picks the option: hovered then Enter, which every select
-    takes. A suggestion list takes its first suggestion. No option leaves the select empty."""
+    """Types into a react-select and picks the option: focused with the arrow keys then Enter,
+    which every select takes (a hover does not move the focus once typed). A suggestion list
+    takes its first suggestion. No option leaves the select empty."""
     box = page.locator(_by_id(field_id))
     box.click()
     box.press_sequentially(value, delay=KEY_DELAY_MS)
-    menu = page.locator(f"{_by_id(f'react-select-{field_id}-listbox')} [role=option]")
+    listbox = _by_id(f"react-select-{field_id}-listbox")
+    menu = page.locator(f"{listbox} [role=option]")
     option = menu.first if suggested else menu.filter(has_text=_exact(value)).first
     try:
         option.wait_for(timeout=ANSWER_TIMEOUT_MS)
     except Timeout:
         box.press("Escape")
         return
-    option.hover()
+    focused = page.locator(f"{listbox} .select__option--is-focused")
+    for _ in range(menu.count()):
+        if suggested or any(_exact(value).match(t) for t in focused.all_inner_texts()):
+            break
+        box.press("ArrowDown")
     box.press("Enter")

{atsfill-0.0.4.dev61 → atsfill-0.0.4.dev63}/src/atsfill/adapters/manatal.py

@@ -34,6 +34,9 @@ LISTS: dict[str, tuple[str, Callable[[dict[str, Any]], str]]] = {
     "industries": ("clients/{client}/industries/", lambda n: n["name"]),
 }
 ENTRIES = ("educations", "experiences")  # a sub-form of entries
+# A salary's amount comes with a currency and a frequency select, each with an id of its own.
+SALARIES = {"expected_salary": "expected", "current_salary": "current"}
+FREQUENCIES = ["Hourly", "Daily", "Weekly", "Monthly", "Yearly"]
 OTHER = "Other"  # the radio or checkbox list option that opens a text box once picked
 SHORT = {"full_name", "email", "phone_number", "city", "address", "latest_degree",
          "latest_university", "current_company", "current_department", "current_position",
@@ -43,8 +46,8 @@ LOAD_TIMEOUT_MS = 30_000
 SUBMIT_TIMEOUT_MS = 120_000
 
 # The answer each input holds, by field id, read as the page reads them on Apply.
-_READ = """fields => Object.fromEntries(fields.map(([id, kind]) => {
-  const el = document.getElementById('field' + id);
+_READ = """fields => Object.fromEntries(fields.map(([id, kind, dom]) => {
+  const el = document.getElementById(dom);
   const named = [...document.querySelectorAll(`input[name="${id}"]`)];
   const texts = s => [...s.selectedOptions].filter(o => o.value).map(o => o.text.trim());
   switch (kind) {
@@ -107,6 +110,24 @@ def _kind(entry: dict[str, Any]) -> str:
     return "text"
 
 
+def _salary(entry: dict[str, Any]) -> list[tuple[str, str]]:
+    """A salary's currency and frequency selects: the field id each is listed as, and its own id."""
+    prefix = SALARIES.get(entry["client_field_slug"])
+    if prefix is None or entry["field_type"] != "candidate_field":
+        return []
+    return [(f"{entry['id']}.{part}", f"{prefix}_{part}") for part in ("currency", "frequency")]
+
+
+def _inputs(entries: list[dict[str, Any]]) -> list[tuple[str, str, str]]:
+    """Each input the page reads on Apply: its field id, its kind and its element id."""
+    inputs = []
+    for entry in entries:
+        if (kind := _kind(entry)) != "entries":
+            inputs.append((str(entry["id"]), kind, f"field{entry['id']}"))
+            inputs += [(fid, "select", dom) for fid, dom in _salary(entry)]
+    return inputs
+
+
 def _field(entry: dict[str, Any], options: list[str] | None) -> Field:
     """The field an entry of the endpoint lists, a sub-form of entries aside."""
     kind, slug = _kind(entry), entry["client_field_slug"]
@@ -182,7 +203,10 @@ class Manatal:
                 return None
             entries = _get(f"jobs/{job.group(1)}/application-form/")
             terms = bool(TERMS_REQUIRED.search(html))
-            return self._fields(entries, terms, lambda e: _options(e, client))
+            salary = any(_salary(e) for e in entries)
+            currencies = [c["name"] for c in _get("currencies/")] if salary else []
+            salaries = lambda dom: currencies if dom.endswith("currency") else FREQUENCIES  # noqa: E731
+            return self._fields(entries, terms, lambda e: _options(e, client), salaries)
         except OSError, ValueError, KeyError, TypeError:
             return None
 
@@ -199,19 +223,31 @@ class Manatal:
             raise Unsupported(f"posting_gone: the form answered {response.status}")
         self._entries = response.json()
         page.wait_for_selector("#submit-id-submit", timeout=LOAD_TIMEOUT_MS)
-        # The lists of a multiple select come in their own requests, once the form is drawn.
+        # The lists of a multiple select and the currencies come in their own requests, once the
+        # form is drawn.
         page.wait_for_function(
             "() => [...document.querySelectorAll('select.multiselect')]"
-            ".every(s => s.options.length && s.classList.contains('select2-hidden-accessible'))",
+            ".every(s => s.options.length && s.classList.contains('select2-hidden-accessible'))"
+            " && [...document.querySelectorAll('#expected_currency, #current_currency')]"
+            ".every(s => s.options.length)",
             timeout=LOAD_TIMEOUT_MS,
         )
         terms = page.locator(f'input[name="{TERMS}"]').count() > 0
-        return self._fields(self._entries, terms, lambda e: self._shown(page, e))
+        return self._fields(
+            self._entries,
+            terms,
+            lambda e: self._shown(page, e),
+            lambda dom: page.locator(f"#{dom} option").evaluate_all(
+                "os => os.map(o => o.text.trim())"
+            ),
+        )
 
     def fill(self, page: Page, answers: dict[str, Answer]) -> None:
-        kinds = {str(e["id"]): _kind(e) for e in self._entries} | {TERMS: "terms"}
+        inputs = _inputs(self._entries)
+        kinds = {fid: kind for fid, kind, _ in inputs} | {TERMS: "terms"}
+        doms = {fid: dom for fid, _, dom in inputs}
         for field_id, answer in answers.items():
-            box = page.locator(f"#field{field_id}")
+            box = page.locator(f"#{doms.get(field_id, f'field{field_id}')}")
             named = page.locator(f'input[name="{field_id}"]')
             match kinds.get(field_id), answer:
                 case "text", str():
@@ -247,10 +283,10 @@ class Manatal:
                     )
 
     def read_back(self, page: Page) -> dict[str, Answer]:
-        kinds = [[str(e["id"]), _kind(e)] for e in self._entries if _kind(e) != "entries"]
+        inputs = [list(i) for i in _inputs(self._entries)]
         if page.locator(f'input[name="{TERMS}"]').count():
-            kinds.append([TERMS, "terms"])
-        held = page.evaluate(_READ, kinds)
+            inputs.append([TERMS, "terms", ""])
+        held = page.evaluate(_READ, inputs)
         return {k: v for k, v in held.items() if v not in (None, [])}
 
     def submit(self, page: Page) -> str | None:
@@ -271,6 +307,7 @@ class Manatal:
         entries: list[dict[str, Any]],
         terms: bool,
         options: Callable[[dict[str, Any]], list[str] | None],
+        salaries: Callable[[str], list[str]],
     ) -> list[Field]:
         fields = []
         for entry in entries:
@@ -278,6 +315,9 @@ class Manatal:
                 fields.append(self._entries_field(entry))
                 continue
             fields.append(field := _field(entry, options(entry)))
+            for fid, dom in _salary(entry):
+                label = f"{entry['label']}: {fid.rsplit('.', 1)[1]}"
+                fields.append(Field(fid, label, FieldType.CHOICE, field.required, salaries(dom)))
             if _kind(entry) in ("radio", "checkbox") and OTHER in (field.options or []):
                 fields.append(self._other_field(field))
         if terms: