Active_prgs/Redgrave/vCardSearchAndDownload.py

"""

vCardSearchAndDownload

Created by:
Emanuel Borges
12.30.2024

This program uses chromedriver to first search for a person on a public company people search and then, if found, download the vCard for that person.

"""

import os
import time
from selenium import webdriver
from selenium.webdriver.common.by import By
from selenium.webdriver.common.keys import Keys
from urllib.parse import urljoin
import requests


#BASE_URL = "https://perkinscoie.com/people-search"
BASE_URL = "https://www.klgates.com/people"
#BASE_URL = "https://www.hoganlovells.com/en/our-people"

VCARD_DIR = "vcards"
os.makedirs(VCARD_DIR, exist_ok = True)


driver = webdriver.Chrome()

version = '0.2.0'


def search_and_download_vcards(person_name):
    try:
        driver.get(BASE_URL)

        time.sleep(3)


        search_box = driver.find_element(By.ID, "searchbox")
        #search_box = driver.find_element(By.ID, "edit-keyword")
        #search_box = driver.find_element(By.ID, "name3")
        search_box.clear()
        #search_box.send_keys(person_name)
        search_box.send_keys(f'"{person_name}"')
        search_box.send_keys(Keys.RETURN)

        time.sleep(3)


        results = driver.find_elements(By.CSS_SELECTOR, "a[href*='vcard']")
        
        if not results:
            print(f"No vCards found for '{person_name}'.")
            return


        for link in results:
            vcard_url = link.get_attribute("href")
            download_vcard(vcard_url,person_name)

    except Exception as e:
        print(f"Error during search or download: {e}")
##    finally:
##        driver.quit()


def download_vcard(vcard_url, person_name):
    try:
        response = requests.get(vcard_url)
        response.raise_for_status()

        #filename = os.path.basename(vcard_url)
        filename = f"{person_name}.vcf"
        filepath = os.path.join(VCARD_DIR,filename)


        with open(filepath,"wb") as file:
            file.write(response.content)
        print(f"Downloaded vCard: {filepath}")
    except Exception as e:
        print(f"Error downloading vCard from {vcard_url}: {e}")


if __name__ == '__main__':
    #inputFilePath = r"C:\Users\eborges\OneDrive - Redgrave LLP\Documents\Cases\Amazon\MAL-NamesToFind\KLGATES.txt"
    inputFilePath = r"C:\Users\eborges\OneDrive - Redgrave LLP\Documents\Cases\Amazon\MAL-NamesToFind\KLGATES_SUB3.txt"
    #inputFilePath = r"C:\Users\eborges\OneDrive - Redgrave LLP\Documents\Cases\Amazon\MAL-NamesToFind\TEST.txt"


    ##  This can be a list of any value to search on but for now I need it to be unique, which is why I'm searching only on email addresses
    listOfEmailAddresses = set()


    contents = open(inputFilePath).readlines()
    for line in contents:
        line = line.replace("\n","")
        uniqueRowNumb,emailAddr = line.split("|")
        person_to_search = emailAddr
        search_and_download_vcards(person_to_search)


    #person_to_search = "Gaia Bacchi"
    #person_to_search = "tim.weston@klgates.com"
    #person_to_search = "BPeters@perkinscoie.com"
    #person_to_search = "Katie.McMullan@hoganlovells.com"
    #search_and_download_vcards(person_to_search)

    driver.quit()
Revision:	874
Committed:	Tue Dec 31 01:27:14 2024 UTC (14 months, 3 weeks ago) by nino.borges
Content type:	text/x-python
File size:	3301 byte(s)
Log Message:	This version reads from a file with many email addresses and worked perfectly for all of the KLGATES addresses.
#	User	Rev	Content
1	nino.borges	873	"""
2
3			vCardSearchAndDownload
4
5			Created by:
6			Emanuel Borges
7			12.30.2024
8
9			This program uses chromedriver to first search for a person on a public company people search and then, if found, download the vCard for that person.
10
11			"""
12
13			import os
14			import time
15			from selenium import webdriver
16			from selenium.webdriver.common.by import By
17			from selenium.webdriver.common.keys import Keys
18			from urllib.parse import urljoin
19			import requests
20
21
22	nino.borges	874	#BASE_URL = "https://perkinscoie.com/people-search"
23	nino.borges	873	BASE_URL = "https://www.klgates.com/people"
24	nino.borges	874	#BASE_URL = "https://www.hoganlovells.com/en/our-people"
25	nino.borges	873
26			VCARD_DIR = "vcards"
27			os.makedirs(VCARD_DIR, exist_ok = True)
28
29
30			driver = webdriver.Chrome()
31
32	nino.borges	874	version = '0.2.0'
33	nino.borges	873
34
35			def search_and_download_vcards(person_name):
36			try:
37			driver.get(BASE_URL)
38
39			time.sleep(3)
40
41
42			search_box = driver.find_element(By.ID, "searchbox")
43	nino.borges	874	#search_box = driver.find_element(By.ID, "edit-keyword")
44			#search_box = driver.find_element(By.ID, "name3")
45	nino.borges	873	search_box.clear()
46	nino.borges	874	#search_box.send_keys(person_name)
47			search_box.send_keys(f'"{person_name}"')
48	nino.borges	873	search_box.send_keys(Keys.RETURN)
49
50			time.sleep(3)
51
52
53			results = driver.find_elements(By.CSS_SELECTOR, "a[href*='vcard']")
54	nino.borges	874
55	nino.borges	873	if not results:
56			print(f"No vCards found for '{person_name}'.")
57			return
58
59
60			for link in results:
61			vcard_url = link.get_attribute("href")
62			download_vcard(vcard_url,person_name)
63
64			except Exception as e:
65			print(f"Error during search or download: {e}")
66	nino.borges	874	## finally:
67			## driver.quit()
68	nino.borges	873
69
70
71			def download_vcard(vcard_url, person_name):
72			try:
73			response = requests.get(vcard_url)
74			response.raise_for_status()
75
76			#filename = os.path.basename(vcard_url)
77			filename = f"{person_name}.vcf"
78			filepath = os.path.join(VCARD_DIR,filename)
79
80
81			with open(filepath,"wb") as file:
82			file.write(response.content)
83			print(f"Downloaded vCard: {filepath}")
84			except Exception as e:
85			print(f"Error downloading vCard from {vcard_url}: {e}")
86
87
88			if __name__ == '__main__':
89	nino.borges	874	#inputFilePath = r"C:\Users\eborges\OneDrive - Redgrave LLP\Documents\Cases\Amazon\MAL-NamesToFind\KLGATES.txt"
90			inputFilePath = r"C:\Users\eborges\OneDrive - Redgrave LLP\Documents\Cases\Amazon\MAL-NamesToFind\KLGATES_SUB3.txt"
91			#inputFilePath = r"C:\Users\eborges\OneDrive - Redgrave LLP\Documents\Cases\Amazon\MAL-NamesToFind\TEST.txt"
92	nino.borges	873
93	nino.borges	874
94
95			## This can be a list of any value to search on but for now I need it to be unique, which is why I'm searching only on email addresses
96			listOfEmailAddresses = set()
97
98
99			contents = open(inputFilePath).readlines()
100			for line in contents:
101			line = line.replace("\n","")
102			uniqueRowNumb,emailAddr = line.split("\|")
103			person_to_search = emailAddr
104			search_and_download_vcards(person_to_search)
105
106
107	nino.borges	873
108	nino.borges	874
109	nino.borges	873	#person_to_search = "Gaia Bacchi"
110			#person_to_search = "tim.weston@klgates.com"
111	nino.borges	874	#person_to_search = "BPeters@perkinscoie.com"
112			#person_to_search = "Katie.McMullan@hoganlovells.com"
113			#search_and_download_vcards(person_to_search)
114
115			driver.quit()