From 92cd221dde796c2e93f606d68a8d1a1b98de31c8 Mon Sep 17 00:00:00 2001
From: Jason <jason.d.baker@gmail.com>
Date: Fri, 30 May 2025 10:38:04 -0700
Subject: [PATCH] new script

---
 extract_v2.py        | 87 ++++++++++++++++++++++++++++++++++++++++++++
 job-bot.py           |  1 +
 search_criteria.yaml | 58 +++++++++++++++++++++++++++++
 3 files changed, 146 insertions(+)
 create mode 100644 extract_v2.py
 create mode 100644 search_criteria.yaml

diff --git a/extract_v2.py b/extract_v2.py
new file mode 100644
index 0000000..efddbba
--- /dev/null
+++ b/extract_v2.py
@@ -0,0 +1,87 @@
+from airflow import DAG
+from airflow.decorators import task
+from airflow.models import Variable
+from datetime import datetime, date, timedelta
+from pymongo import MongoClient
+
+from jobspy import scrape_jobs
+
+import pandas as pd
+import yaml
+import graypy
+import logging
+
+DEBUG = False
+
+default_args = {
+    'owner': 'airflow',
+    'depends_on_past': False,
+    'email_on_failure': False,
+    'email_on_retry': False,
+    'retries': 1,
+    'retry_delay': timedelta(minutes=5),
+}
+
+# Configure logging
+logger = logging.getLogger('JobBot')
+logger.setLevel(logging.INFO)
+
+# Configure graypy handler for GELF UDP
+graylog_handler = graypy.GELFUDPHandler('graylog.localdomain', 12201)
+graylog_handler.include_logger_name = True
+logger.addHandler(graylog_handler)
+
+def load_search_config(config_path):
+    with open(config_path, 'r') as file:
+        return yaml.safe_load(file)
+
+def process_jobs(search_params):
+    print(f"Scraping jobs with parameters: {search_params['name']}")
+    jobs = scrape_jobs(**search_params['params'])
+    return jobs
+
+def date_to_datetime(d):
+    if isinstance(d, date) and not isinstance(d, datetime):
+        return datetime.combine(d, datetime.min.time())
+    return d
+
+with DAG(
+    'job_bot_api_dag',
+    default_args=default_args,
+    description='A DAG to fetch data from job-bot API and process it',
+    schedule='*/10 * * * *',  # Every 10 minutes
+    start_date=datetime.now() - timedelta(days=1),  # Changed to today-1 day
+    catchup=False,
+    max_active_runs=1,
+    tags=['job-bot', 'api'],
+) as dag:
+
+    @task()
+    def fetch_jobs():
+        # Load configuration
+        config = load_search_config('search_criteria.yaml')
+        
+        jobs = pd.DataFrame()
+        
+        # Process each search configuration
+        for search in config['searches']:
+            try:
+                _jobs = process_jobs(search)
+                if len(_jobs) > 0:
+                    # Apply filters from search configuration if they exist
+                    if 'filter' in search:
+                        filter_field = search['filter']['field']
+                        exclude_list = search['filter']['exclude']
+                        _jobs = _jobs[~_jobs[filter_field].str.contains('|'.join(exclude_list), case=False, na=False)]
+                    jobs = pd.concat([jobs, _jobs])
+            except Exception as e:
+                print(f"Error processing search '{search['name']}': {str(e)}")
+                continue
+        # Basic stats
+        print(f"Found {len(jobs)} jobs")
+
+    # TaskFlow dependencies
+    api_results = fetch_jobs()
+
+    if __name__ == "__main__":
+        dag.test()
\ No newline at end of file
diff --git a/job-bot.py b/job-bot.py
index 91001a9..d5981c1 100644
--- a/job-bot.py
+++ b/job-bot.py
@@ -38,6 +38,7 @@ with DAG(
     schedule='*/10 * * * *',  # Every 10 minutes
     start_date=datetime.now() - timedelta(days=1),  # Changed to today-1 day
     catchup=False,
+    max_active_runs=1,
     tags=['job-bot', 'api'],
 ) as dag:
 
diff --git a/search_criteria.yaml b/search_criteria.yaml
new file mode 100644
index 0000000..c2b020e
--- /dev/null
+++ b/search_criteria.yaml
@@ -0,0 +1,58 @@
+# ├── Indeed limitations:
+# |    Only one from this list can be used in a search:
+# |    - hours_old
+# |    - job_type & is_remote
+# |    - easy_apply
+# │
+# └── LinkedIn limitations:
+# |    Only one from this list can be used in a search:
+# |    - hours_old
+# |    - easy_apply
+
+searches:
+  - name: indeed_engineering_manager
+    params:
+      site_name: ["indeed"]
+      search_term: '"engineering manager"'
+      location: "Mill Valley, CA"
+      distance: 17
+      results_wanted: 10
+      country_indeed: 'USA'
+
+  - name: indeed_engineering_manager_remote
+    params:
+      site_name: ["indeed"]
+      search_term: '"engineering manager"'
+      is_remote: true
+      results_wanted: 10
+      country_indeed: 'USA'
+
+  - name: google_engineering_manager
+    params:
+      site_name: ["google"]
+      google_search_term: 'senior engineering manager'
+      location: "San Francisco, CA"
+      results_wanted: 10
+      hours_old: 24
+    filter:
+      field: "job_url"
+      exclude:  ["https://www.linkedin.com", "https://www.glassdoor.com", "https://www.indeed.com"]
+
+  - name: linkedin_engineering_manager
+    params:
+      site_name: ["linkedin"]
+      search_term: 'senior engineering manager'
+      linkedin_fetch_description: true
+      location: "Mill Valley, CA"
+      distance: 17
+      results_wanted: 10
+      hours_old: 24
+
+  - name: linkedin_engineering_manager_remote
+    params:
+      site_name: ["linkedin"]
+      search_term: 'senior engineering manager'
+      linkedin_fetch_description: true
+      is_remote: true
+      results_wanted: 10
+      hours_old: 24      
\ No newline at end of file