Branch#1

README.md

@@ -0,0 +1,33 @@
+# CNE340 Winter2024 - Job Hunter Project
+
+## Objectives
+
+   * Using python to complete the following tasks:
+     - Connect to MySql on WAMP Server
+     - Create a database
+     - Grab new jobs from a website, parses JSON code and inserts the data into a list of dictionaries do not need to edit
+     - Import to database
+     - Perform Query commands
+
+## Requirements:
+
+   * Project Fork
+   * GitHub Commits
+   * CREATE table
+   * Query table for already-existing job entries
+   * Add new job function
+   * Notify user of new jobs
+   * Meets specific design specifications listed below (covered in depth in the lecture recording)
+       - Run program once and leave running
+       - Upon load grabs configuration
+       - Load desired job keywords
+       - Connect to DB
+       - Check once per 4 hours for job
+       - Adds jobs to the DB
+       - Avoids duplicates
+       - Notify user of matches
+       - Extra Credit: if it deletes jobs over 14 days old
+
+## Special Thanks
+
+Thank you **Ixius Procopios** for the initial code and all the guidance for this project.

jobhunter.py

@@ -1,8 +1,12 @@
+# CNE 340 Fall 2024
+# Student name: Van Vuong
+# Project: Job Hunter
+
+import datetime as dt
 import mysql.connector
 import time
 import json
 import requests
-from datetime import date
 import html2text
 
 
@@ -19,12 +23,14 @@ def create_tables(cursor):
     # Creates table
     # Must set Title to CHARSET utf8 unicode Source: http://mysql.rjweb.org/doc.php/charcoll.
     # Python is in latin-1 and error (Incorrect string value: '\xE2\x80\xAFAbi...') will occur if Description is not in unicode format due to the json data
-    cursor.execute('''CREATE TABLE IF NOT EXISTS jobs (id INT PRIMARY KEY auto_increment, Job_id varchar(50) , 
-    company varchar (300), Created_at DATE, url varchar(30000), Title LONGBLOB, Description LONGBLOB ); ''')
+    cursor.execute('''CREATE TABLE IF NOT EXISTS jobs (id INT PRIMARY KEY auto_increment, Job_id varchar(50), 
+    company varchar(300), Created_at DATE, url TEXT, Title LONGBLOB, Description LONGBLOB );''')
+    return
 
 
 # Query the database.
 # You should not need to edit anything in this function
+
 def query_sql(cursor, query):
     cursor.execute(query)
     return cursor
@@ -33,32 +39,38 @@ def query_sql(cursor, query):
 # Add a new job
 def add_new_job(cursor, jobdetails):
     # extract all required columns
+    job_id = jobdetails['id']
+    Company = jobdetails['company_name']
+    URL = jobdetails['url']
+    title = jobdetails['title']
     description = html2text.html2text(jobdetails['description'])
+    # print(description)
     date = jobdetails['publication_date'][0:10]
-    query = cursor.execute("INSERT INTO jobs( Description, Created_at " ") "
-               "VALUES(%s,%s)", (  description, date))
-     # %s is what is needed for Mysqlconnector as SQLite3 uses ? the Mysqlconnector uses %s
+    # print(date)
+    query = cursor.execute("INSERT INTO jobs (Job_id, company,  url, Title, Description, Created_at)"
+                           "VALUES(%s,%s,%s,%s,%s,%s)", (job_id, Company, URL, title, description, date))
+    # %s is what is needed for Mysqlconnector as SQLite3 uses ? the Mysqlconnector uses %s
     return query_sql(cursor, query)
 
 
 # Check if new job
 def check_if_job_exists(cursor, jobdetails):
     ##Add your code here
-    query = "UPDATE"
+    query = "SELECT * FROM jobs WHERE Job_id = \"%s\" " % jobdetails['id']
     return query_sql(cursor, query)
 
+
 # Deletes job
 def delete_job(cursor, jobdetails):
     ##Add your code here
-    query = "UPDATE"
+    query = "DELETE FROM jobs WHERE Job_id = \"%s\" " % jobdetails['id']
     return query_sql(cursor, query)
 
 
 # Grab new jobs from a website, Parses JSON code and inserts the data into a list of dictionaries do not need to edit
 def fetch_new_jobs():
     query = requests.get("https://remotive.io/api/remote-jobs")
     datas = json.loads(query.text)
-
     return datas
 
 
@@ -73,17 +85,27 @@ def jobhunt(cursor):
 
 def add_or_delete_job(jobpage, cursor):
     # Add your code here to parse the job page
-    for jobdetails in jobpage['jobs']:  # EXTRACTS EACH JOB FROM THE JOB LIST. It errored out until I specified jobs. This is because it needs to look at the jobs dictionary from the API. https://careerkarma.com/blog/python-typeerror-int-object-is-not-iterable/
+    for jobdetails in jobpage['jobs']:
+        # EXTRACTS EACH JOB FROM THE JOB LIST. It errored out until I specified jobs. This is because it needs to look at the jobs dictionary from the API. https://careerkarma.com/blog/python-typeerror-int-object-is-not-iterable/
         # Add in your code here to check if the job already exists in the DB
         check_if_job_exists(cursor, jobdetails)
-        is_job_found = len(
-        cursor.fetchall()) > 0  # https://stackoverflow.com/questions/2511679/python-number-of-rows-affected-by-cursor-executeselect
+        is_job_found = len(cursor.fetchall()) > 0  # https://stackoverflow.com/questions/2511679/python-number-of-rows-affected-by-cursor-executeselect
         if is_job_found:
+            current_date = dt.datetime.now()
+            post_date = dt.datetime.strptime(jobdetails['publication_date'], "%Y-%m-%dT%H:%M:%S")
 
+            # Delete if jobs over 14 days old
+            if (current_date.day - post_date.day) > 14:
+                print("Sorry, Job is not available!")
+                delete_job(cursor, jobdetails)
         else:
             # INSERT JOB
             # Add in your code here to notify the user of a new posting. This code will notify the new user
 
+            print(f"Jobs found match for you. Title: " + jobdetails['title'] + ". Company: " + jobdetails[
+                'company_name'] + ", posted on " + jobdetails['publication_date'] + ", Job ID: " + str(
+                jobdetails['id']))
+            add_new_job(cursor, jobdetails)
 
 
 # Setup portion of the program. Take arguments and set up the script
@@ -104,4 +126,3 @@ def main():
 # If you want to test if script works change time.sleep() to 10 seconds and delete your table in MySQL
 if __name__ == '__main__':
     main()
-

test.py

@@ -0,0 +1,8 @@
+import datetime as dt
+
+current_date = dt.datetime.now()
+post_date = dt.datetime.strptime("2024-02-07T00:51:00", "%Y-%m-%dT%H:%M:%S")
+
+
+print(current_date.day)
+print(post_date.day)