Mẫu dữ liệu vòng đời nhân viên từ tuyển dụng đến nghỉ việc

WITH EmployeeData AS (
  SELECT
    papf.person_id,
    papf.person_number,
    paam.assignment_id,
    paam.effective_start_date,
    paam.effective_end_date,
    ppnf.full_name,
    haouf.name AS department_name,
    pjfl.name AS job_role,
    paam.recruitment_activity_id,
    paam.original_date_of_hire,
    pdp.reason_code AS termination_reason
  FROM
    per_all_people_f papf
    JOIN per_person_names_f ppnf ON papf.person_id = ppnf.person_id AND ppnf.name_type = 'GLOBAL'
    JOIN per_all_assignments_m paam ON papf.person_id = paam.person_id
    LEFT JOIN hr_all_organization_units_f_vl haouf ON paam.organization_id = haouf.organization_id
    LEFT JOIN per_jobs_f_vl pjfl ON paam.job_id = pjfl.job_id
    LEFT JOIN per_periods_of_service ppos ON paam.period_of_service_id = ppos.period_of_service_id
    LEFT JOIN per_termination_details_f ptdf ON ppos.period_of_service_id = ptdf.period_of_service_id
    LEFT JOIN per_reasons_vl pdp ON ptdf.termination_reason_id = pdp.reason_id
  WHERE
    paam.assignment_type IN ('E', 'C', 'N') AND paam.primary_flag = 'Y'
    AND SYSDATE BETWEEN papf.effective_start_date AND papf.effective_end_date
    AND SYSDATE BETWEEN ppnf.effective_start_date AND ppnf.effective_end_date
    AND SYSDATE BETWEEN paam.effective_start_date AND paam.effective_end_date
    AND SYSDATE BETWEEN haouf.effective_start_date(+) AND haouf.effective_end_date(+)
    AND SYSDATE BETWEEN pjfl.effective_start_date(+) AND pjfl.effective_end_date(+)
)
-- 1. Offer Accepted
SELECT
  ed.person_number AS "EmployeeId",
  'Offer Accepted' AS "ActivityName",
  CAST(offers.accepted_on AS TIMESTAMP) AS "EventTimestamp",
  NULL AS "EventEndTime",
  offers.last_updated_by AS "UserId",
  ed.department_name AS "DepartmentName",
  ed.job_role AS "JobRole",
  subs.source_type AS "RecruitmentSource",
  NULL AS "TerminationReason",
  ed.original_date_of_hire AS "HireDate"
FROM irc_offers offers
JOIN irc_submissions subs ON offers.submission_id = subs.submission_id
JOIN EmployeeData ed ON subs.person_id = ed.person_id
WHERE offers.offer_status = 'EXTENDED' AND offers.offer_resolution = 'ACCEPTED'
AND offers.accepted_on BETWEEN TO_DATE(:p_start_date, 'YYYY-MM-DD') AND TO_DATE(:p_end_date, 'YYYY-MM-DD')

UNION ALL

-- 2. Background Check Initiated & 3. Background Check Completed
SELECT
  ed.person_number AS "EmployeeId",
  CASE WHEN task.task_status = 'COMPLETED' THEN 'Background Check Completed' ELSE 'Background Check Initiated' END AS "ActivityName",
  CASE WHEN task.task_status = 'COMPLETED' THEN CAST(task.actual_completion_date AS TIMESTAMP) ELSE CAST(task.creation_date AS TIMESTAMP) END AS "EventTimestamp",
  CAST(task.actual_completion_date AS TIMESTAMP) AS "EventEndTime",
  task.last_updated_by AS "UserId",
  ed.department_name AS "DepartmentName",
  ed.job_role AS "JobRole",
  NULL AS "RecruitmentSource",
  NULL AS "TerminationReason",
  ed.original_date_of_hire AS "HireDate"
FROM per_allocated_tasks task
JOIN per_checklist_definitions_vl chk ON task.checklist_id = chk.checklist_id
JOIN EmployeeData ed ON task.person_id = ed.person_id
WHERE chk.category = 'ONBOARDING' AND chk.name LIKE '%Background Check%'
AND (task.creation_date BETWEEN TO_DATE(:p_start_date, 'YYYY-MM-DD') AND TO_DATE(:p_end_date, 'YYYY-MM-DD') OR task.actual_completion_date BETWEEN TO_DATE(:p_start_date, 'YYYY-MM-DD') AND TO_DATE(:p_end_date, 'YYYY-MM-DD'))

UNION ALL

-- 4. Onboarding Journey Initiated
SELECT
  ed.person_number AS "EmployeeId",
  'Onboarding Journey Initiated' AS "ActivityName",
  CAST(journey.creation_date AS TIMESTAMP) AS "EventTimestamp",
  NULL AS "EventEndTime",
  journey.created_by AS "UserId",
  ed.department_name AS "DepartmentName",
  ed.job_role AS "JobRole",
  NULL AS "RecruitmentSource",
  NULL AS "TerminationReason",
  ed.original_date_of_hire AS "HireDate"
FROM per_allocated_checklists journey
JOIN per_checklist_definitions_vl chk ON journey.checklist_id = chk.checklist_id
JOIN EmployeeData ed ON journey.person_id = ed.person_id
WHERE chk.category IN ('ONBOARDING', '[Your Onboarding Journey Category]')
AND journey.creation_date BETWEEN TO_DATE(:p_start_date, 'YYYY-MM-DD') AND TO_DATE(:p_end_date, 'YYYY-MM-DD')

UNION ALL

-- 5. New Hire Data Submitted & 6. I-9 Verification Completed
SELECT
  ed.person_number AS "EmployeeId",
  CASE WHEN chk.name LIKE '%I-9%' THEN 'I-9 Verification Completed' ELSE 'New Hire Data Submitted' END AS "ActivityName",
  CAST(task.actual_completion_date AS TIMESTAMP) AS "EventTimestamp",
  NULL AS "EventEndTime",
  task.last_updated_by AS "UserId",
  ed.department_name AS "DepartmentName",
  ed.job_role AS "JobRole",
  NULL AS "RecruitmentSource",
  NULL AS "TerminationReason",
  ed.original_date_of_hire AS "HireDate"
FROM per_allocated_tasks task
JOIN per_checklist_definitions_vl chk ON task.checklist_id = chk.checklist_id
JOIN EmployeeData ed ON task.person_id = ed.person_id
WHERE chk.category = 'ONBOARDING' AND task.task_status = 'COMPLETED'
AND (chk.name LIKE '%Personal Data%' OR chk.name LIKE '%I-9%')
AND task.actual_completion_date BETWEEN TO_DATE(:p_start_date, 'YYYY-MM-DD') AND TO_DATE(:p_end_date, 'YYYY-MM-DD')

UNION ALL

-- 7. Employee Onboarded
SELECT
  ed.person_number AS "EmployeeId",
  'Employee Onboarded' AS "ActivityName",
  CAST(pao.actual_date AS TIMESTAMP) AS "EventTimestamp",
  NULL AS "EventEndTime",
  pao.last_updated_by AS "UserId",
  ed.department_name AS "DepartmentName",
  ed.job_role AS "JobRole",
  NULL AS "RecruitmentSource",
  NULL AS "TerminationReason",
  ed.original_date_of_hire AS "HireDate"
FROM per_action_occurrences pao
JOIN per_actions_vl pav ON pao.action_id = pav.action_id
JOIN EmployeeData ed ON pao.person_id = ed.person_id
WHERE pav.action_code = 'HIRE'
AND pao.actual_date BETWEEN TO_DATE(:p_start_date, 'YYYY-MM-DD') AND TO_DATE(:p_end_date, 'YYYY-MM-DD')

UNION ALL

-- 8. Payroll Setup Completed
SELECT 
  ed.person_number AS "EmployeeId",
  'Payroll Setup Completed' AS "ActivityName",
  CAST(pra.effective_date AS TIMESTAMP) AS "EventTimestamp",
  NULL AS "EventEndTime",
  pra.created_by AS "UserId",
  ed.department_name AS "DepartmentName",
  ed.job_role AS "JobRole",
  NULL AS "RecruitmentSource",
  NULL AS "TerminationReason",
  ed.original_date_of_hire AS "HireDate"
FROM pay_payroll_rel_actions pra
JOIN pay_payroll_assignments ppa ON pra.payroll_assignment_id = ppa.payroll_assignment_id
JOIN EmployeeData ed ON ppa.person_id = ed.person_id
WHERE pra.action_status = 'C' -- Completed
AND pra.action_type = 'H' -- Hire or Rehire
AND pra.effective_date BETWEEN TO_DATE(:p_start_date, 'YYYY-MM-DD') AND TO_DATE(:p_end_date, 'YYYY-MM-DD')

UNION ALL

-- 9. Training Assigned
SELECT
  ed.person_number AS "EmployeeId",
  'Training Assigned' AS "ActivityName",
  CAST(we.creation_date AS TIMESTAMP) AS "EventTimestamp",
  CAST(we.completed_date AS TIMESTAMP) AS "EventEndTime",
  we.created_by AS "UserId",
  ed.department_name AS "DepartmentName",
  ed.job_role AS "JobRole",
  NULL AS "RecruitmentSource",
  NULL AS "TerminationReason",
  ed.original_date_of_hire AS "HireDate"
FROM wlf_enrollments we
JOIN EmployeeData ed ON we.person_id = ed.person_id
WHERE we.creation_date BETWEEN TO_DATE(:p_start_date, 'YYYY-MM-DD') AND TO_DATE(:p_end_date, 'YYYY-MM-DD')

UNION ALL

-- 10. Performance Review Conducted
SELECT
  ed.person_number AS "EmployeeId",
  'Performance Review Conducted' AS "ActivityName",
  CAST(he.evaluation_complete_date AS TIMESTAMP) AS "EventTimestamp",
  NULL AS "EventEndTime",
  he.last_updated_by AS "UserId",
  ed.department_name AS "DepartmentName",
  ed.job_role AS "JobRole",
  NULL AS "RecruitmentSource",
  NULL AS "TerminationReason",
  ed.original_date_of_hire AS "HireDate"
FROM hra_evaluations he
JOIN EmployeeData ed ON he.person_id = ed.person_id
WHERE he.evaluation_status = 'COMPLETED'
AND he.evaluation_complete_date BETWEEN TO_DATE(:p_start_date, 'YYYY-MM-DD') AND TO_DATE(:p_end_date, 'YYYY-MM-DD')

UNION ALL

-- 11. Role Change Initiated & 12. Promotion Approved
SELECT 
  ed.person_number AS "EmployeeId",
  CASE WHEN pao.action_reason = 'PROMOTION' THEN 'Promotion Approved' ELSE 'Role Change Initiated' END AS "ActivityName",
  CAST(pao.actual_date AS TIMESTAMP) AS "EventTimestamp",
  NULL AS "EventEndTime",
  pao.last_updated_by AS "UserId",
  ed.department_name AS "DepartmentName",
  ed.job_role AS "JobRole",
  NULL AS "RecruitmentSource",
  NULL AS "TerminationReason",
  ed.original_date_of_hire AS "HireDate"
FROM per_action_occurrences pao
JOIN per_actions_vl pav ON pao.action_id = pav.action_id
JOIN EmployeeData ed ON pao.person_id = ed.person_id
WHERE pav.action_code IN ('PROMOTION', 'TRANSFER', 'ASG_CHANGE')
AND pao.status = 'A' -- Approved
AND pao.actual_date BETWEEN TO_DATE(:p_start_date, 'YYYY-MM-DD') AND TO_DATE(:p_end_date, 'YYYY-MM-DD')

UNION ALL

-- 13. Termination Initiated
SELECT
  ed.person_number AS "EmployeeId",
  'Termination Initiated' AS "ActivityName",
  CAST(ptdf.creation_date AS TIMESTAMP) AS "EventTimestamp",
  CAST(ptdf.actual_termination_date AS TIMESTAMP) AS "EventEndTime",
  ptdf.created_by AS "UserId",
  ed.department_name AS "DepartmentName",
  ed.job_role AS "JobRole",
  NULL AS "RecruitmentSource",
  ed.termination_reason AS "TerminationReason",
  ed.original_date_of_hire AS "HireDate"
FROM per_termination_details_f ptdf
JOIN per_periods_of_service ppos ON ptdf.period_of_service_id = ppos.period_of_service_id
JOIN EmployeeData ed ON ppos.person_id = ed.person_id
WHERE ptdf.creation_date BETWEEN TO_DATE(:p_start_date, 'YYYY-MM-DD') AND TO_DATE(:p_end_date, 'YYYY-MM-DD')

UNION ALL

-- 14. Exit Interview Conducted & 15. Offboarding Tasks Completed
SELECT
  ed.person_number AS "EmployeeId",
  CASE WHEN chk.name LIKE '%Exit Interview%' THEN 'Exit Interview Conducted' ELSE 'Offboarding Tasks Completed' END AS "ActivityName",
  CAST(task.actual_completion_date AS TIMESTAMP) AS "EventTimestamp",
  NULL AS "EventEndTime",
  task.last_updated_by AS "UserId",
  ed.department_name AS "DepartmentName",
  ed.job_role AS "JobRole",
  NULL AS "RecruitmentSource",
  ed.termination_reason AS "TerminationReason",
  ed.original_date_of_hire AS "HireDate"
FROM per_allocated_tasks task
JOIN per_checklist_definitions_vl chk ON task.checklist_id = chk.checklist_id
JOIN EmployeeData ed ON task.person_id = ed.person_id
WHERE chk.category = 'OFFBOARDING' AND task.task_status = 'COMPLETED'
AND task.actual_completion_date BETWEEN TO_DATE(:p_start_date, 'YYYY-MM-DD') AND TO_DATE(:p_end_date, 'YYYY-MM-DD')

/*
This is a configuration plan for Oracle BI Cloud Connector (BICC).
It is not a single executable SQL script. Instead, it details which PVOs
and columns to select within the BICC user interface to generate the raw
data files needed for the Hire to Retire event log.
The post-processing script mentioned in the 'howto' section will use these
files to construct the final event log by mapping columns and assigning
the correct ActivityName.
*/

-- PVO 1: Offer Data
-- Purpose: Capture Offer Accepted events.
PVO_NAME: HcmTopModelAnalyticsGlobalAM.RecruitingAM.OfferPVO
COLUMNS:
  - PersonId (as EmployeeId)
  - OfferAcceptedDate (as EventTimestamp)
  - [RecruitmentSource] (as RecruitmentSource)
  - [DepartmentName] (as DepartmentName)
  - [JobRole] (as JobRole)
EVENT_LOG_MAPPING:
  - ActivityName: 'Offer Accepted'
  - Filter: OfferAcceptedDate IS NOT NULL AND OfferStatus = 'ORA_ACCEPTED'

-- PVO 2: Onboarding and Offboarding Tasks
-- Purpose: Capture multiple task-based events from checklists.
PVO_NAME: HcmTopModelAnalyticsGlobalAM.OnboardingAM.ChecklistTaskPVO
COLUMNS:
  - PersonId (as EmployeeId)
  - TaskName
  - TaskCreationDate
  - TaskCompletionDate
  - TaskPerformerPersonId (as UserId)
EVENT_LOG_MAPPING:
  - WHEN TaskName LIKE 'Background Check%' THEN ActivityName: 'Background Check Initiated', EventTimestamp: TaskCreationDate
  - WHEN TaskName LIKE 'Background Check%' AND TaskStatus = 'ORA_COMPLETED' THEN ActivityName: 'Background Check Completed', EventTimestamp: TaskCompletionDate
  - WHEN TaskName LIKE '%Personal Data%' AND TaskStatus = 'ORA_COMPLETED' THEN ActivityName: 'New Hire Data Submitted', EventTimestamp: TaskCompletionDate
  - WHEN TaskName LIKE '%I-9 Verification%' AND TaskStatus = 'ORA_COMPLETED' THEN ActivityName: 'I-9 Verification Completed', EventTimestamp: TaskCompletionDate
  - WHEN TaskName LIKE '%Payroll Setup%' AND TaskStatus = 'ORA_COMPLETED' THEN ActivityName: 'Payroll Setup Completed', EventTimestamp: TaskCompletionDate
  - WHEN TaskName LIKE '%Exit Interview%' AND TaskStatus = 'ORA_COMPLETED' THEN ActivityName: 'Exit Interview Conducted', EventTimestamp: TaskCompletionDate
  - WHEN [Category] = 'OFFBOARDING' AND TaskStatus = 'ORA_COMPLETED' THEN ActivityName: 'Offboarding Tasks Completed', EventTimestamp: TaskCompletionDate

-- PVO 3: Onboarding Journey Initiation
-- Purpose: Capture the start of the formal onboarding process.
PVO_NAME: HcmTopModelAnalyticsGlobalAM.OnboardingAM.JourneyPVO
COLUMNS:
  - PersonId (as EmployeeId)
  - CreationDate (as EventTimestamp)
  - AssignorPersonId (as UserId)
  - JourneyStatus
EVENT_LOG_MAPPING:
  - ActivityName: 'Onboarding Journey Initiated'
  - Filter: JourneyStatus IS NOT NULL

-- PVO 4: Core HR Work Relationship
-- Purpose: Capture hire and termination events.
PVO_NAME: HcmTopModelAnalyticsGlobalAM.CoreAM.WorkRelationshipPVO
COLUMNS:
  - PersonId (as EmployeeId)
  - DateStart (as HireDate, and as EventTimestamp for 'Employee Onboarded')
  - TerminationDate (as EventTimestamp for 'Termination Initiated')
  - [TerminationReason] (as TerminationReason)
EVENT_LOG_MAPPING:
  - ActivityName: 'Employee Onboarded', EventTimestamp: DateStart, Filter: DateStart IS NOT NULL
  - ActivityName: 'Termination Initiated', EventTimestamp: TerminationDate, Filter: TerminationDate IS NOT NULL

-- PVO 5: Learning Assignments
-- Purpose: Capture training assignment events.
PVO_NAME: HcmTopModelAnalyticsGlobalAM.LearningManagementAM.LearningAssignmentPVO
COLUMNS:
  - PersonId (as EmployeeId)
  - AssignmentCreationDate (as EventTimestamp)
  - AssignerId (as UserId)
  - LearningItemTitle
EVENT_LOG_MAPPING:
  - ActivityName: 'Training Assigned'

-- PVO 6: Performance Reviews
-- Purpose: Capture performance review completion.
PVO_NAME: HcmTopModelAnalyticsGlobalAM.TalentManagementAM.PerformanceDocumentPVO
COLUMNS:
  - WorkerPersonId (as EmployeeId)
  - CompletionDate (as EventTimestamp)
  - ManagerPersonId (as UserId)
  - DocumentStatus
EVENT_LOG_MAPPING:
  - ActivityName: 'Performance Review Conducted'
  - Filter: DocumentStatus = 'ORA_COMPLETED' AND CompletionDate IS NOT NULL

-- PVO 7: HR Actions
-- Purpose: Capture role changes and promotions.
PVO_NAME: HcmTopModelAnalyticsGlobalAM.CoreAM.ActionPVO
COLUMNS:
  - PersonId (as EmployeeId)
  - ActionCode
  - ActionName
  - ActionReasonCode
  - EffectiveDate (as EventTimestamp)
EVENT_LOG_MAPPING:
  - WHEN ActionCode IN ('PROMOTION', 'ORA_PROMOTION') THEN ActivityName: 'Promotion Approved', EventTimestamp: EffectiveDate
  - WHEN ActionCode IN ('TRANSFER', 'ORA_TRANSFER', 'ROLE_CHANGE') THEN ActivityName: 'Role Change Initiated', EventTimestamp: EffectiveDate

import requests
import pandas as pd
import os
from datetime import datetime

class=class="hl-string">"hl-comment"># --- Configuration Settings --- #
class=class="hl-string">"hl-comment"># Replace with your Oracle HCM Cloud instance URL
HCM_BASE_URL = class="hl-string">"https://[your_instance_name].fa.em2.oraclecloud.com"
class=class="hl-string">"hl-comment"># Replace with your integration user credentials
USERNAME = class="hl-string">"your_api_username"
PASSWORD = class="hl-string">"your_api_password"

class=class="hl-string">"hl-comment"># Set the date range for extraction (YYYY-MM-DD)
START_DATE = class="hl-string">"2023-01-01"
END_DATE = class="hl-string">"2023-12-31"

class=class="hl-string">"hl-comment"># --- End of Configuration --- #

class=class="hl-string">"hl-comment"># Global list to hold all event data
event_log = []

def get_paginated_data(endpoint, params={}):
    class="hl-string">"""Helper function to handle pagination for Oracle HCM REST API calls."""
    all_items = []
    url = fclass="hl-string">"{HCM_BASE_URL}{endpoint}"
    params.setdefault(class="hl-string">'limit', 100)
    offset = 0
    
    while True:
        params[class="hl-string">'offset'] = offset
        print(fclass="hl-string">"Fetching from {url} with offset {offset}...")
        try:
            response = requests.get(url, auth=(USERNAME, PASSWORD), params=params)
            response.raise_for_status() class=class="hl-string">"hl-comment"># Raise an exception for bad status codes
            data = response.json()
            items = data.get(class="hl-string">'items', [])
            all_items.extend(items)
            
            if not data.get(class="hl-string">'hasMore', False):
                break
            offset += len(items)
        except requests.exceptions.RequestException as e:
            print(fclass="hl-string">"Error fetching data from {endpoint}: {e}")
            break
            
    print(fclass="hl-string">"Fetched a total of {len(all_items)} items from {endpoint}.")
    return all_items

def extract_recruiting_and_onboarding_events():
    class="hl-string">"""Extracts events from Recruiting and Onboarding modules."""
    print(class="hl-string">"\n--- Starting Recruiting and Onboarding Extraction ---")
    class=class="hl-string">"hl-comment"># Fetch Offer Data
    offer_params = {
        class="hl-string">'q': fclass="hl-string">"AcceptedDate between '{START_DATE}' and '{END_DATE}'",
        class="hl-string">'fields': class="hl-string">'OfferId,AcceptedDate,CandidateNumber,assignments'
    }
    offers = get_paginated_data(class="hl-string">'/hcmRestApi/resources/11.13.18.05/recruitingCECandidateOffers', offer_params)
    for offer in offers:
        if offer.get(class="hl-string">'AcceptedDate'):
            event_log.append({
                class="hl-string">'EmployeeId': offer.get(class="hl-string">'CandidateNumber'), class=class="hl-string">"hl-comment"># Using CandidateNumber as early identifier
                class="hl-string">'ActivityName': class="hl-string">'Offer Accepted',
                class="hl-string">'EventTimestamp': offer[class="hl-string">'AcceptedDate'],
                class="hl-string">'EventEndTime': None,
                class="hl-string">'UserId': None, 
                class="hl-string">'DepartmentName': offer.get(class="hl-string">'assignments', [{}])[0].get(class="hl-string">'DepartmentName'),
                class="hl-string">'JobRole': offer.get(class="hl-string">'assignments', [{}])[0].get(class="hl-string">'JobName'),
                class="hl-string">'RecruitmentSource': offer.get(class="hl-string">'assignments', [{}])[0].get(class="hl-string">'RecruitingSource'),
                class="hl-string">'TerminationReason': None,
                class="hl-string">'HireDate': None
            })

    class=class="hl-string">"hl-comment"># Fetch Onboarding/Offboarding Journey Data
    journey_params = {
        class="hl-string">'q': fclass="hl-string">"CreationDate between '{START_DATE}' and '{END_DATE}'",
        class="hl-string">'fields': class="hl-string">'PersonNumber,JourneyType,CreationDate,CreatedBy,tasks;fields=TaskName,TaskStatus,ActualCompletionDate,LastUpdatedBy'
    }
    journeys = get_paginated_data(class="hl-string">'/hcmRestApi/resources/11.13.18.05/onboardingJourneys', journey_params)
    for journey in journeys:
        person_id = journey.get(class="hl-string">'PersonNumber')
        journey_type = journey.get(class="hl-string">'JourneyType')
        
        activity_name = class="hl-string">'Onboarding Journey Initiated' if journey_type == class="hl-string">'ONBOARD' else class="hl-string">'Offboarding Journey Initiated'
        
        event_log.append({
            class="hl-string">'EmployeeId': person_id,
            class="hl-string">'ActivityName': activity_name,
            class="hl-string">'EventTimestamp': journey[class="hl-string">'CreationDate'],
            class="hl-string">'EventEndTime': None,
            class="hl-string">'UserId': journey.get(class="hl-string">'CreatedBy'),
            class="hl-string">'DepartmentName': None,
            class="hl-string">'JobRole': None,
            class="hl-string">'RecruitmentSource': None,
            class="hl-string">'TerminationReason': None,
            class="hl-string">'HireDate': None
        })
        
        class=class="hl-string">"hl-comment"># Process tasks within the journey
        for task in journey.get(class="hl-string">'tasks', []):
            task_activity = None
            task_name_lower = task.get(class="hl-string">'TaskName', class="hl-string">'').lower()
            if class="hl-string">'background check' in task_name_lower:
                task_activity = class="hl-string">'Background Check Initiated' if task[class="hl-string">'TaskStatus'] != class="hl-string">'COMPLETED' else class="hl-string">'Background Check Completed'
            elif class="hl-string">'personal data' in task_name_lower and task[class="hl-string">'TaskStatus'] == class="hl-string">'COMPLETED':
                task_activity = class="hl-string">'New Hire Data Submitted'
            elif class="hl-string">'i-9' in task_name_lower and task[class="hl-string">'TaskStatus'] == class="hl-string">'COMPLETED':
                task_activity = class="hl-string">'I-9 Verification Completed'
            elif class="hl-string">'exit interview' in task_name_lower and task[class="hl-string">'TaskStatus'] == class="hl-string">'COMPLETED':
                task_activity = class="hl-string">'Exit Interview Conducted'
            
            if task_activity and task.get(class="hl-string">'ActualCompletionDate'):
                event_log.append({
                    class="hl-string">'EmployeeId': person_id,
                    class="hl-string">'ActivityName': task_activity,
                    class="hl-string">'EventTimestamp': task[class="hl-string">'ActualCompletionDate'],
                    class="hl-string">'EventEndTime': task[class="hl-string">'ActualCompletionDate'],
                    class="hl-string">'UserId': task.get(class="hl-string">'LastUpdatedBy'),
                    class="hl-string">'DepartmentName': None,
                    class="hl-string">'JobRole': None,
                    class="hl-string">'RecruitmentSource': None,
                    class="hl-string">'TerminationReason': None,
                    class="hl-string">'HireDate': None
                })
        
        class=class="hl-string">"hl-comment"># Offboarding Tasks Completed (approximated by last task completion)
        if journey_type == class="hl-string">'OFFBOARD':
            completed_tasks = [t for t in journey.get(class="hl-string">'tasks', []) if t.get(class="hl-string">'ActualCompletionDate')]
            if completed_tasks and all(t[class="hl-string">'TaskStatus'] == class="hl-string">'COMPLETED' for t in journey.get(class="hl-string">'tasks', [])):
                last_completion_date = max(t[class="hl-string">'ActualCompletionDate'] for t in completed_tasks)
                event_log.append({
                    class="hl-string">'EmployeeId': person_id,
                    class="hl-string">'ActivityName': class="hl-string">'Offboarding Tasks Completed',
                    class="hl-string">'EventTimestamp': last_completion_date,
                    class="hl-string">'EventEndTime': last_completion_date,
                    class="hl-string">'UserId': None,
                    class="hl-string">'DepartmentName': None, class="hl-string">'JobRole': None, class="hl-string">'RecruitmentSource': None, class="hl-string">'TerminationReason': None, class="hl-string">'HireDate': None
                })

def extract_worker_and_lifecycle_events():
    class="hl-string">"""Extracts core worker data and major lifecycle events like hire, promotion, termination."""
    print(class="hl-string">"\n--- Starting Worker and Lifecycle Extraction ---")
    worker_params = {
        class="hl-string">'q': fclass="hl-string">"(workRelationships.assignments.CreationDate between '{START_DATE}' and '{END_DATE}') or (workRelationships.TerminationDate between '{START_DATE}' and '{END_DATE}')",
        class="hl-string">'onlyData': class="hl-string">'true',
        class="hl-string">'fields': class="hl-string">'PersonNumber,workRelationships;fields=assignments;fields=CreationDate,LastUpdateDate,ActionCode,JobId,DepartmentId,BusinessUnitId,JobName,DepartmentName,EffectiveStartDate,assignmentsDFF,HireDate,TerminationDate,TerminationReason'
    }
    workers = get_paginated_data(class="hl-string">'/hcmRestApi/resources/11.13.18.05/workers', worker_params)
    for worker in workers:
        person_id = worker.get(class="hl-string">'PersonNumber')
        for rel in worker.get(class="hl-string">'workRelationships', []):
            hire_date = rel.get(class="hl-string">'HireDate')
            for assign in rel.get(class="hl-string">'assignments', []):
                if hire_date and hire_date.startswith(START_DATE[:4]): class=class="hl-string">"hl-comment"># Simple check for new hires
                    event_log.append({
                        class="hl-string">'EmployeeId': person_id, class="hl-string">'ActivityName': class="hl-string">'Employee Onboarded', class="hl-string">'EventTimestamp': hire_date,
                        class="hl-string">'EventEndTime': None, class="hl-string">'UserId': None, class="hl-string">'DepartmentName': assign.get(class="hl-string">'DepartmentName'),
                        class="hl-string">'JobRole': assign.get(class="hl-string">'JobName'), class="hl-string">'RecruitmentSource': None, class="hl-string">'TerminationReason': None, class="hl-string">'HireDate': hire_date
                    })
                
                action = assign.get(class="hl-string">'ActionCode')
                if action in (class="hl-string">'PROMOTION', class="hl-string">'ASG_CHANGE_PROMOTION'):
                    event_log.append({
                        class="hl-string">'EmployeeId': person_id, class="hl-string">'ActivityName': class="hl-string">'Promotion Approved', class="hl-string">'EventTimestamp': assign.get(class="hl-string">'EffectiveStartDate'),
                        class="hl-string">'EventEndTime': None, class="hl-string">'UserId': None, class="hl-string">'DepartmentName': assign.get(class="hl-string">'DepartmentName'), 
                        class="hl-string">'JobRole': assign.get(class="hl-string">'JobName'), class="hl-string">'RecruitmentSource': None, class="hl-string">'TerminationReason': None, class="hl-string">'HireDate': hire_date
                    })
                elif action in (class="hl-string">'TRANSFER', class="hl-string">'GLB_TRANSFER', class="hl-string">'POS_CHANGE', class="hl-string">'ASG_CHANGE'):
                    event_log.append({
                        class="hl-string">'EmployeeId': person_id, class="hl-string">'ActivityName': class="hl-string">'Role Change Initiated', class="hl-string">'EventTimestamp': assign.get(class="hl-string">'EffectiveStartDate'),
                        class="hl-string">'EventEndTime': None, class="hl-string">'UserId': None, class="hl-string">'DepartmentName': assign.get(class="hl-string">'DepartmentName'),
                        class="hl-string">'JobRole': assign.get(class="hl-string">'JobName'), class="hl-string">'RecruitmentSource': None, class="hl-string">'TerminationReason': None, class="hl-string">'HireDate': hire_date
                    })
            
            if rel.get(class="hl-string">'TerminationDate'):
                event_log.append({
                    class="hl-string">'EmployeeId': person_id, class="hl-string">'ActivityName': class="hl-string">'Termination Initiated', class="hl-string">'EventTimestamp': rel[class="hl-string">'TerminationDate'],
                    class="hl-string">'EventEndTime': None, class="hl-string">'UserId': None, class="hl-string">'DepartmentName': rel.get(class="hl-string">'assignments', [{}])[0].get(class="hl-string">'DepartmentName'),
                    class="hl-string">'JobRole': rel.get(class="hl-string">'assignments', [{}])[0].get(class="hl-string">'JobName'), class="hl-string">'RecruitmentSource': None,
                    class="hl-string">'TerminationReason': rel.get(class="hl-string">'TerminationReason'), class="hl-string">'HireDate': hire_date
                })

def extract_other_hcm_events():
    class="hl-string">"""Extracts events from other modules like Performance, Learning, and Payroll."""
    print(class="hl-string">"\n--- Starting Ancillary HCM Event Extraction ---")
    class=class="hl-string">"hl-comment"># Performance Reviews
    perf_params = {class="hl-string">'q': fclass="hl-string">"ReviewPeriodEndDate between '{START_DATE}' and '{END_DATE}' and DocumentStatus='COMPLETED'", class="hl-string">'fields': class="hl-string">'PersonNumber,OverallRating,ReviewPeriodEndDate,ManagerPersonNumber'}
    reviews = get_paginated_data(class="hl-string">'/hcmRestApi/resources/11.13.18.05/performanceDocuments', perf_params)
    for review in reviews:
        event_log.append({
            class="hl-string">'EmployeeId': review.get(class="hl-string">'PersonNumber'), class="hl-string">'ActivityName': class="hl-string">'Performance Review Conducted', class="hl-string">'EventTimestamp': review.get(class="hl-string">'ReviewPeriodEndDate'),
            class="hl-string">'EventEndTime': review.get(class="hl-string">'ReviewPeriodEndDate'), class="hl-string">'UserId': review.get(class="hl-string">'ManagerPersonNumber'),
            class="hl-string">'DepartmentName': None, class="hl-string">'JobRole': None, class="hl-string">'RecruitmentSource': None, class="hl-string">'TerminationReason': None, class="hl-string">'HireDate': None
        })

    class=class="hl-string">"hl-comment"># Learning Assignments
    learn_params = {class="hl-string">'q': fclass="hl-string">"AssignedOn between '{START_DATE}' and '{END_DATE}'", class="hl-string">'fields': class="hl-string">'LearnerPersonNumber,AssignedOn,LearningItemName'}
    learnings = get_paginated_data(class="hl-string">'/learningRestApi/resources/latest/learnerLearningRecords', learn_params)
    for item in learnings:
        event_log.append({
            class="hl-string">'EmployeeId': item.get(class="hl-string">'LearnerPersonNumber'), class="hl-string">'ActivityName': class="hl-string">'Training Assigned', class="hl-string">'EventTimestamp': item.get(class="hl-string">'AssignedOn'),
            class="hl-string">'EventEndTime': None, class="hl-string">'UserId': None,
            class="hl-string">'DepartmentName': None, class="hl-string">'JobRole': None, class="hl-string">'RecruitmentSource': None, class="hl-string">'TerminationReason': None, class="hl-string">'HireDate': None
        })

    class=class="hl-string">"hl-comment"># Payroll Setup (Approximation)
    payroll_params = {class="hl-string">'q': fclass="hl-string">"CreationDate between '{START_DATE}' and '{END_DATE}'", class="hl-string">'fields': class="hl-string">'PersonNumber,CreationDate'}
    pay_records = get_paginated_data(class="hl-string">'/hcmRestApi/resources/11.13.18.05/personalPaymentMethods', payroll_params)
    for record in pay_records:
        event_log.append({
            class="hl-string">'EmployeeId': record.get(class="hl-string">'PersonNumber'), class="hl-string">'ActivityName': class="hl-string">'Payroll Setup Completed', class="hl-string">'EventTimestamp': record.get(class="hl-string">'CreationDate'),
            class="hl-string">'EventEndTime': None, class="hl-string">'UserId': None,
            class="hl-string">'DepartmentName': None, class="hl-string">'JobRole': None, class="hl-string">'RecruitmentSource': None, class="hl-string">'TerminationReason': None, class="hl-string">'HireDate': None
        })

if __name__ == class="hl-string">"__main__":
    print(class="hl-string">"Starting Hire to Retire data extraction from Oracle HCM Cloud...")
    
    class=class="hl-string">"hl-comment"># Execute extraction functions
    extract_recruiting_and_onboarding_events()
    extract_worker_and_lifecycle_events()
    extract_other_hcm_events()
    
    if not event_log:
        print(class="hl-string">"No events found for the given criteria. Exiting.")
    else:
        class=class="hl-string">"hl-comment"># Create DataFrame and perform final transformations
        df = pd.DataFrame(event_log)
        
        class=class="hl-string">"hl-comment"># Convert timestamps to consistent format
        df[class="hl-string">'EventTimestamp'] = pd.to_datetime(df[class="hl-string">'EventTimestamp'], errors=class="hl-string">'coerce').dt.strftime(class="hl-string">'%Y-%m-%d %H:%M:%S')
        df[class="hl-string">'EventEndTime'] = pd.to_datetime(df[class="hl-string">'EventEndTime'], errors=class="hl-string">'coerce').dt.strftime(class="hl-string">'%Y-%m-%d %H:%M:%S')
        df.dropna(subset=[class="hl-string">'EmployeeId', class="hl-string">'ActivityName', class="hl-string">'EventTimestamp'], inplace=True)
        
        class=class="hl-string">"hl-comment"># Fill missing employee attributes by forward/backward filling within each employee's history
        df = df.sort_values(by=[class="hl-string">'EmployeeId', class="hl-string">'EventTimestamp'])
        df[class="hl-string">'DepartmentName'] = df.groupby(class="hl-string">'EmployeeId')[class="hl-string">'DepartmentName'].fillna(method=class="hl-string">'ffill').fillna(method=class="hl-string">'bfill')
        df[class="hl-string">'JobRole'] = df.groupby(class="hl-string">'EmployeeId')[class="hl-string">'JobRole'].fillna(method=class="hl-string">'ffill').fillna(method=class="hl-string">'bfill')
        df[class="hl-string">'HireDate'] = df.groupby(class="hl-string">'EmployeeId')[class="hl-string">'HireDate'].fillna(method=class="hl-string">'ffill').fillna(method=class="hl-string">'bfill')
        
        class=class="hl-string">"hl-comment"># Define final column order
        final_columns = [
            class="hl-string">'EmployeeId', class="hl-string">'ActivityName', class="hl-string">'EventTimestamp', class="hl-string">'EventEndTime', class="hl-string">'UserId',
            class="hl-string">'DepartmentName', class="hl-string">'JobRole', class="hl-string">'RecruitmentSource', class="hl-string">'TerminationReason', class="hl-string">'HireDate'
        ]
        df = df[final_columns]

        class=class="hl-string">"hl-comment"># Save to CSV
        output_filename = class="hl-string">'hire_to_retire_event_log.csv'
        df.to_csv(output_filename, index=False)
        print(fclass="hl-string">"\nExtraction complete. Event log saved to '{output_filename}'. Total events: {len(df)}")