@@ -3,7 +3,9 @@
 import warnings
 from collections.abc import Iterable
 from concurrent.futures import ThreadPoolExecutor
+from datetime import date
 from pathlib import Path
+from typing import get_args
 
 import pandas as pd
 from pydantic import ValidationError
@@ -27,8 +29,6 @@
 from jobspy.naukri import Naukri
 from jobspy.util import (
     configure_logging,
-    desired_order,
-    extract_salary,
 )
 from jobspy.ziprecruiter import ZipRecruiter
 
@@ -44,6 +44,17 @@
     Site.HELLOWORK: HelloWork,
 }
 
+column_types = {
+    name: {
+        str: "str",
+        float: "float64",
+        int: "int64",
+        bool: "bool",
+        date: "object",
+    }.get((get_args(field.annotation) or (field.annotation,))[0], "str")
+    for name, field in JobPost.model_fields.items()
+}
+
 
 def scrape_jobs(
     site_name: str | Site | Iterable[str | Site] | None = None,
@@ -110,55 +121,19 @@ def scrape_site(site: Site) -> tuple[str, list[JobPost]]:
             executor.map(scrape_site, dict.fromkeys(scraper_input.site_name))
         )
 
-    rows = []
-    for site, jobs in site_to_jobs_dict.items():
-        for job in jobs:
-            job_data = job.model_dump()
-            job_data["site"] = site
-            job_data["company"] = job_data["company_name"]
-            job_data["job_type"] = (
-                ", ".join(job_type.value for job_type in job_data["job_type"])
-                if job_data["job_type"]
-                else None
-            )
-            for name in ("emails", "skills"):
-                job_data[name] = ", ".join(job_data[name]) if job_data[name] else None
-            job_data["location"] = (
-                job.location.display_location() if job.location else None
-            )
-
-            job_data |= dict.fromkeys(
-                ("interval", "min_amount", "max_amount", "currency", "salary_source")
-            )
-            if job.compensation:
-                job_data |= job.compensation.model_dump(mode="json")
-                job_data["salary_source"] = "direct_data"
-            elif scraper_input.country_indeed == Country.USA:
-                (
-                    job_data["interval"],
-                    job_data["min_amount"],
-                    job_data["max_amount"],
-                    job_data["currency"],
-                ) = extract_salary(job_data["description"])
-                job_data["salary_source"] = "description"
-            if scraper_input.enforce_annual_salary and job_data["interval"] not in (
-                None,
-                "yearly",
-            ):
-                factor = {"hourly": 2080, "daily": 260, "weekly": 52, "monthly": 12}[
-                    job_data["interval"]
-                ]
-                for name in ("min_amount", "max_amount"):
-                    if job_data[name] is not None:
-                        job_data[name] = round(job_data[name] * factor, 2)
-                job_data["interval"] = "yearly"
-            if not (job_data["min_amount"] or job_data["max_amount"]):
-                job_data["salary_source"] = None
-            rows.append(job_data)
+    rows = [
+        (job.annualised() if scraper_input.enforce_annual_salary else job).model_copy(
+            update={"site": site}
+        )
+        for site, jobs in site_to_jobs_dict.items()
+        for job in jobs
+    ]
 
+    jobs_df = pd.DataFrame(
+        [job.model_dump() for job in rows], columns=list(JobPost.model_fields)
+    )
     if not rows:
-        return pd.DataFrame()
-    jobs_df = pd.DataFrame(rows, columns=desired_order)
+        return jobs_df.astype(column_types)
     return jobs_df.sort_values(
         by=["site", "date_posted"], ascending=[True, False]
     ).reset_index(drop=True)
@@ -29,7 +29,7 @@
     TLSRotating,
     check_status,
     create_session,
-    format_description,
+    description_fields,
     utc_date,
 )
 
@@ -180,7 +180,7 @@ def _fetch_hybrid_ids(self) -> set[str] | None:
     def _timestamp(job: Tag) -> int | None:
         posted = job.select_one("[data-automation-jobactivedate]")
         stamp = posted.get("data-automation-jobactivedate", "") if posted else ""
-        return int(stamp) if isinstance(stamp, str) and stamp.isdigit() else None
+        return int(stamp) if isinstance(stamp, str) and stamp.isdecimal() else None
 
     def _process_job(self, job: Tag) -> JobPost | None:
         link = job.select_one("h2 a[href]")
@@ -225,7 +225,7 @@ def _process_job(self, job: Tag) -> JobPost | None:
         job_post = JobPost(
             id=f"bayt-{job_id}",
             title=link.get_text(strip=True),
-            company_name=(company.get_text(strip=True) or None) if company else None,
+            company=(company.get_text(strip=True) or None) if company else None,
             company_url=(
                 urljoin(self.base_url, str(company_link["href"]))
                 if company_link
@@ -251,9 +251,9 @@ def _process_job(self, job: Tag) -> JobPost | None:
                 ),
                 None,
             ),
-            compensation=(
-                self._parse_salary(salary.get_text(" ", strip=True)) if salary else None
-            ),
+            **self._parse_salary(
+                salary.get_text(" ", strip=True) if salary else ""
+            ).model_dump(),
         )
         if self.scraper_input.fetch_description:
             details = self._fetch_details(job_id, job_url)
@@ -262,14 +262,14 @@ def _process_job(self, job: Tag) -> JobPost | None:
         return job_post
 
     @staticmethod
-    def _parse_salary(text: str) -> Compensation | None:
+    def _parse_salary(text: str) -> Compensation:
         """ "AED 29,380 - AED 33,053" / "$3,000 - $4,000"; monthly, as job pages say"""
         match = re.fullmatch(
             r"([A-Z]{3}|\$) ?([\d,]+(?:\.\d+)?) - (?:[A-Z]{3}|\$)? ?([\d,]+(?:\.\d+)?)",
             text,
         )
         if not match:
-            return None
+            return Compensation()
         currency, low, high = match.groups()
         return Compensation(
             interval=CompensationInterval.MONTHLY,
@@ -304,15 +304,14 @@ def row(name: str) -> list[str]:
             tag = soup.select_one(f'[data-automation-id="id_{name}"]')
             return [part.strip() for part in tag.get_text().split("·")] if tag else []
 
-        description, emails = format_description(
+        fields = description_fields(
             posting.get("description"), self.scraper_input.description_format
         )
         job_type = job_type_labels.get(next(iter(row("type_level_experience")), ""))
         company = row("company_employees_industry")
         openings = re.match(r"\d+", "".join(row("number_of_vacancies")))
         return {
-            "description": description,
-            "emails": emails,
+            **fields,
             "job_type": [job_type] if job_type else None,
             "is_remote": row("remote_working") == ["Remote"],
             "vacancy_count": int(openings[0]) if openings else None,
 
@@ -29,7 +29,7 @@
     RequestsRotating,
     check_status,
     create_session,
-    format_description,
+    description_fields,
     too_old,
 )
 
@@ -133,13 +133,13 @@ def _process_job(self, raw: dict[str, object]) -> JobPost | None:
         return JobPost(
             id=f"bd-{job_id}",
             title=job.jobTitle,
-            company_name=job.companyName,
+            company=job.companyName,
             location=Location(city=city or None, country="Bangladesh"),
             job_url=f"{self.base_url}/h/details/{job_id}",
             date_posted=posted.date(),
             job_type=self._parse_job_type(job.JobType),
             is_remote=is_remote,
-            compensation=self._parse_salary(job.Salary),
+            **self._parse_salary(job.Salary).model_dump(),
             company_logo=job.logoUrl or None,
             vacancy_count=job.Vacancies,
             experience_range=job.experience if job.experience != "NA" else None,
@@ -151,10 +151,10 @@ def _parse_job_type(text: str) -> list[JobType] | None:
         return [job_type_labels[t] for t in labels if t in job_type_labels] or None
 
     @staticmethod
-    def _parse_salary(text: str) -> Compensation | None:
+    def _parse_salary(text: str) -> Compensation:
         match = re.fullmatch(r"Tk\. (\d+)(?: - (\d+))? \(Monthly\)", text)
         if not match:
-            return None
+            return Compensation()
         low, high = match.groups()
         return Compensation(
             interval=CompensationInterval.MONTHLY,
@@ -175,7 +175,7 @@ def _fetch_details(self, job_id: str) -> dict[str, object]:
                 self.log.warning(f"job {job_id}: {result.message}")
                 return {}
             details = result.data[0]
-            description, emails = format_description(
+            fields = description_fields(
                 "".join(
                     f"<h4>{heading}</h4>{getattr(details, field)}"
                     for heading, field in description_sections
@@ -199,8 +199,7 @@ def _fetch_details(self, job_id: str) -> dict[str, object]:
             if host == "bdjobs.com" or host.endswith(".bdjobs.com"):
                 apply_url = None
             return {
-                "description": description,
-                "emails": emails,
+                **fields,
                 "skills": skills.split(", ") if skills else None,
                 "job_url_direct": apply_url,
                 "company_url": company_url,
 
@@ -35,7 +35,7 @@
     TLSRotating,
     check_status,
     create_session,
-    format_description,
+    description_fields,
     get_enum_from_job_type,
 )
 
@@ -200,7 +200,7 @@ def _process_job(self, raw: dict[str, object]) -> JobPost | None:
         return JobPost(
             id=f"gd-{job_id}",
             title=job.jobTitleText,
-            company_name=header.employerNameFromSearch,
+            company=header.employerNameFromSearch,
             company_url=(
                 f"{self.base_url}/Overview/W-EI_IE{company_id}.htm"
                 if company_id
@@ -213,7 +213,7 @@ def _process_job(self, raw: dict[str, object]) -> JobPost | None:
             job_type=job_types or None,
             is_remote="WORK_FROM_HOME" in (header.remoteWorkTypes or [])
             or header.locationName == "Remote",
-            compensation=self._parse_salary(header),
+            **self._parse_salary(header).model_dump(),
             listing_type=header.adOrderSponsorshipLevel.lower(),
             company_rating=header.rating or None,
             company_industry=company.industryName,
@@ -225,10 +225,10 @@ def _process_job(self, raw: dict[str, object]) -> JobPost | None:
         )
 
     @staticmethod
-    def _parse_salary(header: Header) -> Compensation | None:
+    def _parse_salary(header: Header) -> Compensation:
         pay = header.payPeriodAdjustedPay
         if not pay:
-            return None
+            return Compensation()
         return Compensation(
             interval=pay_intervals.get(header.payPeriod or ""),
             min_amount=round(pay.p10, 2),
@@ -264,14 +264,10 @@ def _fetch_details(self, job_ids: list[str]) -> dict[str, dict[str, object]]:
                 check_status(response)
                 for alias, view in response.json()[0]["data"].items():
                     if view:
-                        description, emails = format_description(
+                        details[f"gd-{alias[1:]}"] = description_fields(
                             view["job"]["description"],
                             self.scraper_input.description_format,
                         )
-                        details[f"gd-{alias[1:]}"] = {
-                            "description": description,
-                            "emails": emails,
-                        }
         except Exception as e:
             self.log.warning(f"job details: {e}")
         return details
@@ -199,7 +199,7 @@ def _parse_job(self, job_info: list[Any]) -> JobPost | None:
         job_post = JobPost(
             id=f"go-{job_info[28]}",
             title=title,
-            company_name=company_name,
+            company=company_name,
             location=Location(
                 city=city, state=state, country=country[0] if country else None
             ),
 
@@ -33,7 +33,7 @@
     TLSRotating,
     check_status,
     create_session,
-    format_description,
+    description_fields,
     utc_date,
 )
 
@@ -166,7 +166,7 @@ def _process_job(self, card: Tag) -> JobPost | None:
                 for tag in contract.parent.find_all("div")
                 if (compensation := self._parse_salary(tag.get_text(strip=True)))
             ),
-            None,
+            Compensation(),
         )
         logo = card.select_one('img[src*="/img/entreprises/"]')
         location = card.select_one("div[data-cy=localisationCard]")
@@ -176,15 +176,15 @@ def _process_job(self, card: Tag) -> JobPost | None:
         job_post = JobPost(
             id=f"hw-{job_id}",
             title=title,
-            company_name="".join(company) or None,
+            company="".join(company) or None,
             location=self._parse_location(location.get_text(strip=True)),
             job_url=f"{self.base_url}{link['href']}",
             date_posted=(
                 utc_date(time.time() - hours * 3600) if hours is not None else None
             ),
-            job_type=job_types,
+            job_type=job_types or None,
             is_remote="Télétravail complet" in label,
-            compensation=salary,
+            **salary.model_dump(),
             company_logo=str(logo["src"]) if logo else None,
         )
         if self.scraper_input.fetch_description:
@@ -252,10 +252,7 @@ def _fetch_details(self, job_id: str, job_url: str) -> dict[str, object]:
                     for button in section.find_all("button"):
                         button.decompose()
                 html = str(section) if section else None
-            description, emails = format_description(
-                html, self.scraper_input.description_format
-            )
-            details: dict[str, object] = {"description": description, "emails": emails}
+            details = description_fields(html, self.scraper_input.description_format)
             if posting.datePosted:
                 posted = datetime.fromisoformat(posting.datePosted.rstrip("Z"))
                 posted = posted.replace(tzinfo=ZoneInfo(paris_time))
 
@@ -31,7 +31,7 @@
     RequestsRotating,
     check_status,
     create_session,
-    format_description,
+    description_fields,
     utc_date,
 )
 
@@ -162,9 +162,7 @@ def attribute(key: str, match: str = "") -> str:
 
     def _process_job(self, raw: dict[str, object]) -> JobPost:
         job = Job.model_validate(raw)
-        description, emails = format_description(
-            job.html, self.scraper_input.description_format
-        )
+        fields = description_fields(job.html, self.scraper_input.description_format)
 
         employer = job.employer or Employer()
         company_page = employer.relativeCompanyPageUrl
@@ -176,8 +174,7 @@ def _process_job(self, raw: dict[str, object]) -> JobPost:
         return JobPost(
             id=f"in-{job.key}",
             title=job.title,
-            description=description,
-            company_name=employer.name or job.sourceEmployerName,
+            company=employer.name or job.sourceEmployerName,
             company_url=f"{self.base_url}{company_page}" if company_page else None,
             company_url_direct=employer.corporateWebsite,
             location=Location(
@@ -186,11 +183,10 @@ def _process_job(self, raw: dict[str, object]) -> JobPost:
                 country=country_code and countries.get(country_code, country_code),
             ),
             job_type=get_job_type(job.attributes, self.scraper_input.country_indeed),
-            compensation=get_compensation(job.compensation),
+            **get_compensation(job.compensation).model_dump(),
             date_posted=utc_date(job.dateOnIndeed / 1000),
             job_url=f"{self.base_url}/viewjob?jk={job.key}",
             job_url_direct=apply_url,
-            emails=emails,
             is_remote=is_job_remote(job.attributes),
             company_addresses=employer.addresses[0] if employer.addresses else None,
             company_industry=(
@@ -204,4 +200,4 @@ def _process_job(self, raw: dict[str, object]) -> JobPost:
             company_logo=employer.squareLogoUrl,
             company_rating=employer.value if employer.count else None,
             company_reviews_count=employer.count or None,
-        )
+        ).model_copy(update=fields)