Create app.py
Browse files
app.py
ADDED
|
@@ -0,0 +1,818 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
import requests
|
| 3 |
+
import re
|
| 4 |
+
import json
|
| 5 |
+
import csv
|
| 6 |
+
import io
|
| 7 |
+
from urllib.parse import urlparse, urljoin
|
| 8 |
+
import time
|
| 9 |
+
import random
|
| 10 |
+
|
| 11 |
+
# Mock contacts database (same as your API)
|
| 12 |
+
CONTACTS_DB = [
|
| 13 |
+
{
|
| 14 |
+
"id": 1,
|
| 15 |
+
"first_name": "Sarah",
|
| 16 |
+
"last_name": "Chen",
|
| 17 |
+
"email": "sarah.chen@techflowsolutions.com",
|
| 18 |
+
"phone": "+1-555-0101",
|
| 19 |
+
"job_title": "CTO",
|
| 20 |
+
"company": "TechFlow Solutions",
|
| 21 |
+
"website": "https://techflowsolutions.com"
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"id": 2,
|
| 25 |
+
"first_name": "Marcus",
|
| 26 |
+
"last_name": "Rodriguez",
|
| 27 |
+
"email": "m.rodriguez@techflowsolutions.com",
|
| 28 |
+
"phone": "+1-555-0102",
|
| 29 |
+
"job_title": "Senior Developer",
|
| 30 |
+
"company": "TechFlow Solutions",
|
| 31 |
+
"website": "https://techflowsolutions.com"
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"id": 3,
|
| 35 |
+
"first_name": "Emma",
|
| 36 |
+
"last_name": "Thompson",
|
| 37 |
+
"email": "emma@greenleafconsult.com",
|
| 38 |
+
"phone": "+1-555-0201",
|
| 39 |
+
"job_title": "Managing Partner",
|
| 40 |
+
"company": "GreenLeaf Consulting",
|
| 41 |
+
"website": "https://greenleafconsult.com"
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"id": 4,
|
| 45 |
+
"first_name": "David",
|
| 46 |
+
"last_name": "Park",
|
| 47 |
+
"email": "david.park@greenleafconsult.com",
|
| 48 |
+
"phone": "+1-555-0202",
|
| 49 |
+
"job_title": "Environmental Analyst",
|
| 50 |
+
"company": "GreenLeaf Consulting",
|
| 51 |
+
"website": "https://greenleafconsult.com"
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"id": 5,
|
| 55 |
+
"first_name": "Jessica",
|
| 56 |
+
"last_name": "Williams",
|
| 57 |
+
"email": "jessica@blueskymarketing.net",
|
| 58 |
+
"phone": "+1-555-0301",
|
| 59 |
+
"job_title": "Creative Director",
|
| 60 |
+
"company": "BlueSky Marketing",
|
| 61 |
+
"website": "https://blueskymarketing.net"
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"id": 6,
|
| 65 |
+
"first_name": "Ryan",
|
| 66 |
+
"last_name": "Mitchell",
|
| 67 |
+
"email": "ryan.mitchell@blueskymarketing.net",
|
| 68 |
+
"phone": "+1-555-0302",
|
| 69 |
+
"job_title": "Account Manager",
|
| 70 |
+
"company": "BlueSky Marketing",
|
| 71 |
+
"website": "https://blueskymarketing.net"
|
| 72 |
+
},
|
| 73 |
+
{
|
| 74 |
+
"id": 7,
|
| 75 |
+
"first_name": "Lisa",
|
| 76 |
+
"last_name": "Zhang",
|
| 77 |
+
"email": "l.zhang@quantumdynamics.org",
|
| 78 |
+
"phone": "+1-555-0401",
|
| 79 |
+
"job_title": "Research Director",
|
| 80 |
+
"company": "Quantum Dynamics Corp",
|
| 81 |
+
"website": "https://quantumdynamics.org"
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"id": 8,
|
| 85 |
+
"first_name": "Ahmed",
|
| 86 |
+
"last_name": "Hassan",
|
| 87 |
+
"email": "ahmed.hassan@quantumdynamics.org",
|
| 88 |
+
"phone": "+1-555-0402",
|
| 89 |
+
"job_title": "Quantum Engineer",
|
| 90 |
+
"company": "Quantum Dynamics Corp",
|
| 91 |
+
"website": "https://quantumdynamics.org"
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"id": 9,
|
| 95 |
+
"first_name": "Maria",
|
| 96 |
+
"last_name": "Gonzalez",
|
| 97 |
+
"email": "maria@stellarlogistics.biz",
|
| 98 |
+
"phone": "+1-555-0501",
|
| 99 |
+
"job_title": "Operations Manager",
|
| 100 |
+
"company": "Stellar Logistics",
|
| 101 |
+
"website": "https://stellarlogistics.biz"
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"id": 10,
|
| 105 |
+
"first_name": "James",
|
| 106 |
+
"last_name": "O'Connor",
|
| 107 |
+
"email": "james.oconnor@stellarlogistics.biz",
|
| 108 |
+
"phone": "+1-555-0502",
|
| 109 |
+
"job_title": "Fleet Coordinator",
|
| 110 |
+
"company": "Stellar Logistics",
|
| 111 |
+
"website": "https://stellarlogistics.biz"
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"id": 11,
|
| 115 |
+
"first_name": "Robert",
|
| 116 |
+
"last_name": "Kim",
|
| 117 |
+
"email": "robert.kim@nexusfinancial.pro",
|
| 118 |
+
"phone": "+1-555-0601",
|
| 119 |
+
"job_title": "Senior Advisor",
|
| 120 |
+
"company": "Nexus Financial",
|
| 121 |
+
"website": "https://nexusfinancial.pro"
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"id": 12,
|
| 125 |
+
"first_name": "Catherine",
|
| 126 |
+
"last_name": "Lee",
|
| 127 |
+
"email": "catherine@nexusfinancial.pro",
|
| 128 |
+
"phone": "+1-555-0602",
|
| 129 |
+
"job_title": "Investment Analyst",
|
| 130 |
+
"company": "Nexus Financial",
|
| 131 |
+
"website": "https://nexusfinancial.pro"
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"id": 13,
|
| 135 |
+
"first_name": "Michael",
|
| 136 |
+
"last_name": "Johnson",
|
| 137 |
+
"email": "m.johnson@horizonhealth.care",
|
| 138 |
+
"phone": "+1-555-0701",
|
| 139 |
+
"job_title": "Chief Medical Officer",
|
| 140 |
+
"company": "Horizon Health Systems",
|
| 141 |
+
"website": "https://horizonhealth.care"
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"id": 14,
|
| 145 |
+
"first_name": "Jennifer",
|
| 146 |
+
"last_name": "Adams",
|
| 147 |
+
"email": "jennifer.adams@horizonhealth.care",
|
| 148 |
+
"phone": "+1-555-0702",
|
| 149 |
+
"job_title": "Head Nurse",
|
| 150 |
+
"company": "Horizon Health Systems",
|
| 151 |
+
"website": "https://horizonhealth.care"
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"id": 15,
|
| 155 |
+
"first_name": "Tony",
|
| 156 |
+
"last_name": "Ricci",
|
| 157 |
+
"email": "tony.ricci@phoenixmfg.com",
|
| 158 |
+
"phone": "+1-555-0801",
|
| 159 |
+
"job_title": "Plant Manager",
|
| 160 |
+
"company": "Phoenix Manufacturing",
|
| 161 |
+
"website": "https://phoenixmfg.com"
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"id": 16,
|
| 165 |
+
"first_name": "Linda",
|
| 166 |
+
"last_name": "Martinez",
|
| 167 |
+
"email": "linda.martinez@phoenixmfg.com",
|
| 168 |
+
"phone": "+1-555-0802",
|
| 169 |
+
"job_title": "Quality Control Supervisor",
|
| 170 |
+
"company": "Phoenix Manufacturing",
|
| 171 |
+
"website": "https://phoenixmfg.com"
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"id": 17,
|
| 175 |
+
"first_name": "Patricia",
|
| 176 |
+
"last_name": "White",
|
| 177 |
+
"email": "patricia.white@alpineeducation.edu",
|
| 178 |
+
"phone": "+1-555-0901",
|
| 179 |
+
"job_title": "Director of Programs",
|
| 180 |
+
"company": "Alpine Education Group",
|
| 181 |
+
"website": "https://alpineeducation.edu"
|
| 182 |
+
},
|
| 183 |
+
{
|
| 184 |
+
"id": 18,
|
| 185 |
+
"first_name": "Kevin",
|
| 186 |
+
"last_name": "Brown",
|
| 187 |
+
"email": "kevin.brown@alpineeducation.edu",
|
| 188 |
+
"phone": "+1-555-0902",
|
| 189 |
+
"job_title": "Curriculum Specialist",
|
| 190 |
+
"company": "Alpine Education Group",
|
| 191 |
+
"website": "https://alpineeducation.edu"
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"id": 19,
|
| 195 |
+
"first_name": "Sophia",
|
| 196 |
+
"last_name": "Taylor",
|
| 197 |
+
"email": "sophia@crimsoncreative.studio",
|
| 198 |
+
"phone": "+1-555-1001",
|
| 199 |
+
"job_title": "Art Director",
|
| 200 |
+
"company": "Crimson Creative Studio",
|
| 201 |
+
"website": "https://crimsoncreative.studio"
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"id": 20,
|
| 205 |
+
"first_name": "Alex",
|
| 206 |
+
"last_name": "Cooper",
|
| 207 |
+
"email": "alex.cooper@crimsoncreative.studio",
|
| 208 |
+
"phone": "+1-555-1002",
|
| 209 |
+
"job_title": "Graphic Designer",
|
| 210 |
+
"company": "Crimson Creative Studio",
|
| 211 |
+
"website": "https://crimsoncreative.studio"
|
| 212 |
+
}
|
| 213 |
+
]
|
| 214 |
+
|
| 215 |
+
def extract_domain(url):
|
| 216 |
+
"""Extract domain from URL"""
|
| 217 |
+
try:
|
| 218 |
+
if not url.startswith(('http://', 'https://')):
|
| 219 |
+
url = 'https://' + url
|
| 220 |
+
parsed = urlparse(url)
|
| 221 |
+
domain = parsed.netloc.lower()
|
| 222 |
+
# Remove www. if present
|
| 223 |
+
if domain.startswith('www.'):
|
| 224 |
+
domain = domain[4:]
|
| 225 |
+
return domain
|
| 226 |
+
except:
|
| 227 |
+
return ""
|
| 228 |
+
|
| 229 |
+
def find_contacts_by_website(website_url):
|
| 230 |
+
"""Find contacts that match the website domain"""
|
| 231 |
+
target_domain = extract_domain(website_url)
|
| 232 |
+
if not target_domain:
|
| 233 |
+
return []
|
| 234 |
+
|
| 235 |
+
print(f"Searching for domain: {target_domain}") # Debug info
|
| 236 |
+
|
| 237 |
+
matching_contacts = []
|
| 238 |
+
for contact in CONTACTS_DB:
|
| 239 |
+
contact_domain = extract_domain(contact['website'])
|
| 240 |
+
print(f"Comparing with: {contact_domain}") # Debug info
|
| 241 |
+
|
| 242 |
+
# Exact domain match or subdomain match
|
| 243 |
+
if target_domain == contact_domain or target_domain in contact_domain or contact_domain in target_domain:
|
| 244 |
+
matching_contacts.append(contact)
|
| 245 |
+
|
| 246 |
+
print(f"Found {len(matching_contacts)} matching contacts") # Debug info
|
| 247 |
+
return matching_contacts
|
| 248 |
+
|
| 249 |
+
def simulate_website_scraping(url):
|
| 250 |
+
"""Simulate scraping a website and finding contact information"""
|
| 251 |
+
# Add some delay to simulate real scraping
|
| 252 |
+
time.sleep(random.uniform(0.5, 1))
|
| 253 |
+
|
| 254 |
+
# Find matching contacts from our database
|
| 255 |
+
contacts = find_contacts_by_website(url)
|
| 256 |
+
|
| 257 |
+
return contacts
|
| 258 |
+
|
| 259 |
+
def is_valid_url(url):
|
| 260 |
+
"""Check if a string looks like a valid URL"""
|
| 261 |
+
url = url.strip()
|
| 262 |
+
if not url:
|
| 263 |
+
return False
|
| 264 |
+
|
| 265 |
+
# Check if it contains common URL patterns
|
| 266 |
+
url_patterns = [
|
| 267 |
+
r'^\w+\.\w+', # domain.com
|
| 268 |
+
r'^\w+\.\w+\.\w+', # subdomain.domain.com
|
| 269 |
+
r'^https?://', # starts with http/https
|
| 270 |
+
r'www\.', # contains www
|
| 271 |
+
]
|
| 272 |
+
|
| 273 |
+
for pattern in url_patterns:
|
| 274 |
+
if re.search(pattern, url.lower()):
|
| 275 |
+
return True
|
| 276 |
+
|
| 277 |
+
return False
|
| 278 |
+
|
| 279 |
+
def parse_csv_file(file_obj):
|
| 280 |
+
"""Parse CSV file and extract website URLs from column H (or auto-detect)"""
|
| 281 |
+
websites = []
|
| 282 |
+
debug_info = []
|
| 283 |
+
|
| 284 |
+
try:
|
| 285 |
+
# Handle different file input types
|
| 286 |
+
if hasattr(file_obj, 'name'):
|
| 287 |
+
# This is a file path (NamedString from Gradio)
|
| 288 |
+
with open(file_obj.name, 'r', encoding='utf-8') as f:
|
| 289 |
+
content = f.read()
|
| 290 |
+
elif isinstance(file_obj, str):
|
| 291 |
+
# This is already a string
|
| 292 |
+
content = file_obj
|
| 293 |
+
else:
|
| 294 |
+
# This might be bytes
|
| 295 |
+
content = file_obj.decode('utf-8')
|
| 296 |
+
|
| 297 |
+
# Parse CSV
|
| 298 |
+
csv_reader = csv.reader(io.StringIO(content))
|
| 299 |
+
rows = list(csv_reader)
|
| 300 |
+
|
| 301 |
+
if not rows:
|
| 302 |
+
debug_info.append("CSV file is empty")
|
| 303 |
+
return [], debug_info
|
| 304 |
+
|
| 305 |
+
debug_info.append(f"Total rows in CSV: {len(rows)}")
|
| 306 |
+
|
| 307 |
+
# Analyze the first few rows to understand structure
|
| 308 |
+
if len(rows) > 0:
|
| 309 |
+
debug_info.append(f"First row has {len(rows[0])} columns: {rows[0]}")
|
| 310 |
+
if len(rows) > 1:
|
| 311 |
+
debug_info.append(f"Second row has {len(rows[1])} columns: {rows[1]}")
|
| 312 |
+
|
| 313 |
+
# Check if first row looks like headers
|
| 314 |
+
first_row = rows[0]
|
| 315 |
+
has_headers = any(col.lower() in ['website', 'url', 'domain', 'site', 'web'] for col in first_row)
|
| 316 |
+
|
| 317 |
+
# Try to find website column
|
| 318 |
+
website_column_index = None
|
| 319 |
+
|
| 320 |
+
# First, try column H (index 7)
|
| 321 |
+
if len(first_row) >= 8:
|
| 322 |
+
website_column_index = 7
|
| 323 |
+
debug_info.append(f"Using column H (index 7) as specified")
|
| 324 |
+
else:
|
| 325 |
+
# Auto-detect website column
|
| 326 |
+
for i, col in enumerate(first_row):
|
| 327 |
+
if col.lower() in ['website', 'url', 'domain', 'site', 'web']:
|
| 328 |
+
website_column_index = i
|
| 329 |
+
debug_info.append(f"Auto-detected website column at index {i}: '{col}'")
|
| 330 |
+
break
|
| 331 |
+
|
| 332 |
+
if website_column_index is None:
|
| 333 |
+
# If no obvious column found, scan all columns for URLs
|
| 334 |
+
debug_info.append("No obvious website column found, scanning all columns for URLs...")
|
| 335 |
+
for row_idx, row in enumerate(rows[1:] if has_headers else rows, start=1):
|
| 336 |
+
for col_idx, cell in enumerate(row):
|
| 337 |
+
if is_valid_url(cell):
|
| 338 |
+
website_column_index = col_idx
|
| 339 |
+
debug_info.append(f"Found URLs in column {col_idx} (row {row_idx}): '{cell}'")
|
| 340 |
+
break
|
| 341 |
+
if website_column_index is not None:
|
| 342 |
+
break
|
| 343 |
+
|
| 344 |
+
if website_column_index is None:
|
| 345 |
+
debug_info.append("ERROR: Could not find any column with website URLs")
|
| 346 |
+
return [], debug_info
|
| 347 |
+
|
| 348 |
+
# Extract websites from the identified column
|
| 349 |
+
start_row = 1 if has_headers else 0
|
| 350 |
+
for row_idx, row in enumerate(rows[start_row:], start=start_row + 1):
|
| 351 |
+
if len(row) > website_column_index:
|
| 352 |
+
website_url = row[website_column_index].strip()
|
| 353 |
+
if website_url and is_valid_url(website_url):
|
| 354 |
+
websites.append(website_url)
|
| 355 |
+
debug_info.append(f"Found website in row {row_idx}: {website_url}")
|
| 356 |
+
elif website_url:
|
| 357 |
+
debug_info.append(f"Row {row_idx}: '{website_url}' doesn't look like a valid URL")
|
| 358 |
+
else:
|
| 359 |
+
debug_info.append(f"Row {row_idx}: Empty cell in website column")
|
| 360 |
+
else:
|
| 361 |
+
debug_info.append(f"Row {row_idx}: Has only {len(row)} columns, need at least {website_column_index + 1}")
|
| 362 |
+
|
| 363 |
+
debug_info.append(f"Total websites extracted: {len(websites)}")
|
| 364 |
+
if websites:
|
| 365 |
+
debug_info.append(f"Sample websites: {websites[:5]}")
|
| 366 |
+
|
| 367 |
+
return websites, debug_info
|
| 368 |
+
|
| 369 |
+
except Exception as e:
|
| 370 |
+
debug_info.append(f"Error parsing CSV: {e}")
|
| 371 |
+
return [], debug_info
|
| 372 |
+
|
| 373 |
+
def search_csv_websites(csv_file, max_results=10):
|
| 374 |
+
"""Search for contacts from websites listed in CSV file"""
|
| 375 |
+
if csv_file is None:
|
| 376 |
+
return "Please upload a CSV file", ""
|
| 377 |
+
|
| 378 |
+
try:
|
| 379 |
+
# Parse CSV file
|
| 380 |
+
websites, debug_info = parse_csv_file(csv_file)
|
| 381 |
+
|
| 382 |
+
debug_text = "\n".join(debug_info)
|
| 383 |
+
|
| 384 |
+
if not websites:
|
| 385 |
+
error_msg = "No websites found in the CSV file.\n\n"
|
| 386 |
+
error_msg += "DEBUG INFORMATION:\n" + debug_text + "\n\n"
|
| 387 |
+
error_msg += "TROUBLESHOOTING:\n"
|
| 388 |
+
error_msg += "1. Ensure your CSV has website URLs in column H (8th column)\n"
|
| 389 |
+
error_msg += "2. Or have a column header named 'website', 'url', 'domain', etc.\n"
|
| 390 |
+
error_msg += "3. Check that URLs are properly formatted (e.g., example.com or https://example.com)\n"
|
| 391 |
+
error_msg += "4. Verify the CSV file is not corrupted\n"
|
| 392 |
+
return error_msg, ""
|
| 393 |
+
|
| 394 |
+
all_contacts = []
|
| 395 |
+
processed_websites = []
|
| 396 |
+
|
| 397 |
+
# Search each website
|
| 398 |
+
for website in websites[:20]: # Limit to first 20 websites
|
| 399 |
+
print(f"Processing website: {website}")
|
| 400 |
+
contacts = simulate_website_scraping(website)
|
| 401 |
+
if contacts:
|
| 402 |
+
all_contacts.extend(contacts)
|
| 403 |
+
processed_websites.append(website)
|
| 404 |
+
print(f"Found {len(contacts)} contacts for {website}")
|
| 405 |
+
else:
|
| 406 |
+
print(f"No contacts found for {website}")
|
| 407 |
+
|
| 408 |
+
# Remove duplicates based on email
|
| 409 |
+
unique_contacts = []
|
| 410 |
+
seen_emails = set()
|
| 411 |
+
for contact in all_contacts:
|
| 412 |
+
if contact['email'] not in seen_emails:
|
| 413 |
+
unique_contacts.append(contact)
|
| 414 |
+
seen_emails.add(contact['email'])
|
| 415 |
+
|
| 416 |
+
# Limit results
|
| 417 |
+
unique_contacts = unique_contacts[:max_results]
|
| 418 |
+
|
| 419 |
+
if not unique_contacts:
|
| 420 |
+
result_msg = f"No contacts found for the {len(websites)} websites from the CSV file.\n\n"
|
| 421 |
+
result_msg += "DEBUG INFORMATION:\n" + debug_text + "\n\n"
|
| 422 |
+
result_msg += f"Websites processed: {', '.join(websites[:10])}\n"
|
| 423 |
+
result_msg += "This might be because the websites are not in our sample database."
|
| 424 |
+
return result_msg, ""
|
| 425 |
+
|
| 426 |
+
# Format results
|
| 427 |
+
results_text = f"CONTACT DISCOVERY REPORT\n"
|
| 428 |
+
results_text += f"CSV Processing Details:\n"
|
| 429 |
+
results_text += f"Total Websites in CSV: {len(websites)}\n"
|
| 430 |
+
results_text += f"Websites Processed: {len(processed_websites)}\n"
|
| 431 |
+
results_text += f"Websites with Contacts: {len(processed_websites)}\n"
|
| 432 |
+
results_text += f"Unique Contacts Found: {len(unique_contacts)}\n"
|
| 433 |
+
results_text += f"Processed Websites: {', '.join(processed_websites)}\n"
|
| 434 |
+
results_text += f"{'='*60}\n\n"
|
| 435 |
+
|
| 436 |
+
# Add debug info
|
| 437 |
+
results_text += "DEBUG INFORMATION:\n" + debug_text + "\n\n"
|
| 438 |
+
results_text += f"{'='*60}\n\n"
|
| 439 |
+
|
| 440 |
+
for i, contact in enumerate(unique_contacts, 1):
|
| 441 |
+
results_text += f"CONTACT #{i}\n"
|
| 442 |
+
results_text += f"Name: {contact['first_name']} {contact['last_name']}\n"
|
| 443 |
+
results_text += f"Position: {contact['job_title']}\n"
|
| 444 |
+
results_text += f"Email: {contact['email']}\n"
|
| 445 |
+
results_text += f"Phone: {contact['phone']}\n"
|
| 446 |
+
results_text += f"Company: {contact['company']}\n"
|
| 447 |
+
results_text += f"Website: {contact['website']}\n\n"
|
| 448 |
+
|
| 449 |
+
# Create CSV output
|
| 450 |
+
csv_output = "First Name,Last Name,Job Title,Email,Phone,Company,Website\n"
|
| 451 |
+
for contact in unique_contacts:
|
| 452 |
+
csv_output += f"{contact['first_name']},{contact['last_name']},{contact['job_title']},{contact['email']},{contact['phone']},{contact['company']},{contact['website']}\n"
|
| 453 |
+
|
| 454 |
+
return results_text, csv_output
|
| 455 |
+
|
| 456 |
+
except Exception as e:
|
| 457 |
+
return f"Error processing CSV file: {str(e)}", ""
|
| 458 |
+
|
| 459 |
+
def search_website_contacts(website_url, max_results=10):
|
| 460 |
+
"""Main function to search for contacts on a website"""
|
| 461 |
+
if not website_url:
|
| 462 |
+
return "Please enter a website URL", ""
|
| 463 |
+
|
| 464 |
+
# Clean up URL
|
| 465 |
+
if not website_url.startswith(('http://', 'https://')):
|
| 466 |
+
website_url = 'https://' + website_url
|
| 467 |
+
|
| 468 |
+
try:
|
| 469 |
+
# Simulate finding contacts
|
| 470 |
+
contacts = simulate_website_scraping(website_url)
|
| 471 |
+
|
| 472 |
+
if not contacts:
|
| 473 |
+
return f"No contacts found on {website_url}. \n\nThis website is not in our contact database. Try one of the sample websites listed below, or the website might not have publicly available contact information.", ""
|
| 474 |
+
|
| 475 |
+
# Limit results
|
| 476 |
+
contacts = contacts[:max_results]
|
| 477 |
+
|
| 478 |
+
# Format results
|
| 479 |
+
results_text = f"CONTACT INTELLIGENCE REPORT\n"
|
| 480 |
+
results_text += f"Website: {website_url}\n"
|
| 481 |
+
results_text += f"Contacts Found: {len(contacts)}\n"
|
| 482 |
+
results_text += f"{'='*60}\n\n"
|
| 483 |
+
|
| 484 |
+
for i, contact in enumerate(contacts, 1):
|
| 485 |
+
results_text += f"CONTACT #{i}\n"
|
| 486 |
+
results_text += f"First Name: {contact['first_name']}\n"
|
| 487 |
+
results_text += f"Last Name: {contact['last_name']}\n"
|
| 488 |
+
results_text += f"Position: {contact['job_title']}\n"
|
| 489 |
+
results_text += f"Email: {contact['email']}\n"
|
| 490 |
+
results_text += f"Phone: {contact['phone']}\n"
|
| 491 |
+
results_text += f"Company: {contact['company']}\n\n"
|
| 492 |
+
|
| 493 |
+
# Create a simple table format for the second output
|
| 494 |
+
table_text = "First Name,Last Name,Job Title,Email,Phone,Company\n"
|
| 495 |
+
for contact in contacts:
|
| 496 |
+
table_text += f"{contact['first_name']},{contact['last_name']},{contact['job_title']},{contact['email']},{contact['phone']},{contact['company']}\n"
|
| 497 |
+
|
| 498 |
+
return results_text, table_text
|
| 499 |
+
|
| 500 |
+
except Exception as e:
|
| 501 |
+
return f"Error searching website: {str(e)}", ""
|
| 502 |
+
|
| 503 |
+
def get_all_available_websites():
|
| 504 |
+
"""Get list of all available websites from the database"""
|
| 505 |
+
websites = list(set([contact['website'] for contact in CONTACTS_DB]))
|
| 506 |
+
return "\n".join(sorted(websites))
|
| 507 |
+
|
| 508 |
+
# Custom CSS
|
| 509 |
+
custom_css = """
|
| 510 |
+
.gradio-container {
|
| 511 |
+
background: linear-gradient(135deg, #f8f9fa 0%, #e9ecef 100%);
|
| 512 |
+
font-family: 'Segoe UI', Tahoma, Geneva, Verdana, sans-serif;
|
| 513 |
+
}
|
| 514 |
+
|
| 515 |
+
.main-header {
|
| 516 |
+
background: linear-gradient(135deg, #1e3a8a 0%, #3b82f6 50%, #1e40af 100%);
|
| 517 |
+
color: white;
|
| 518 |
+
padding: 40px 20px;
|
| 519 |
+
text-align: center;
|
| 520 |
+
border-radius: 15px;
|
| 521 |
+
margin-bottom: 30px;
|
| 522 |
+
box-shadow: 0 10px 30px rgba(30, 58, 138, 0.3);
|
| 523 |
+
}
|
| 524 |
+
|
| 525 |
+
.main-header h1 {
|
| 526 |
+
font-size: 2.5em;
|
| 527 |
+
margin: 0;
|
| 528 |
+
font-weight: 700;
|
| 529 |
+
text-shadow: 2px 2px 4px rgba(0,0,0,0.3);
|
| 530 |
+
}
|
| 531 |
+
|
| 532 |
+
.main-header p {
|
| 533 |
+
font-size: 1.2em;
|
| 534 |
+
margin: 15px 0 0 0;
|
| 535 |
+
opacity: 0.9;
|
| 536 |
+
}
|
| 537 |
+
|
| 538 |
+
.corporate-card {
|
| 539 |
+
background: white;
|
| 540 |
+
border: 1px solid #d1d5db;
|
| 541 |
+
border-radius: 12px;
|
| 542 |
+
padding: 25px;
|
| 543 |
+
margin: 15px 0;
|
| 544 |
+
box-shadow: 0 4px 15px rgba(0,0,0,0.1);
|
| 545 |
+
border-left: 5px solid #1e40af;
|
| 546 |
+
}
|
| 547 |
+
|
| 548 |
+
.tips-section {
|
| 549 |
+
background: linear-gradient(135deg, #f1f5f9 0%, #e2e8f0 100%);
|
| 550 |
+
border: 2px solid #cbd5e1;
|
| 551 |
+
border-radius: 15px;
|
| 552 |
+
padding: 20px;
|
| 553 |
+
margin: 10px 0;
|
| 554 |
+
}
|
| 555 |
+
|
| 556 |
+
.tips-section h3 {
|
| 557 |
+
color: #1e40af;
|
| 558 |
+
margin-top: 0;
|
| 559 |
+
font-weight: 600;
|
| 560 |
+
}
|
| 561 |
+
|
| 562 |
+
.primary-btn {
|
| 563 |
+
background: linear-gradient(135deg, #1e40af 0%, #3b82f6 100%);
|
| 564 |
+
color: white;
|
| 565 |
+
border: none;
|
| 566 |
+
border-radius: 8px;
|
| 567 |
+
padding: 15px 30px;
|
| 568 |
+
font-weight: 600;
|
| 569 |
+
font-size: 16px;
|
| 570 |
+
transition: all 0.3s ease;
|
| 571 |
+
}
|
| 572 |
+
|
| 573 |
+
.primary-btn:hover {
|
| 574 |
+
background: linear-gradient(135deg, #1e3a8a 0%, #2563eb 100%);
|
| 575 |
+
transform: translateY(-2px);
|
| 576 |
+
box-shadow: 0 6px 20px rgba(30, 64, 175, 0.4);
|
| 577 |
+
}
|
| 578 |
+
|
| 579 |
+
.secondary-btn {
|
| 580 |
+
background: white;
|
| 581 |
+
color: #374151;
|
| 582 |
+
border: 2px solid #d1d5db;
|
| 583 |
+
border-radius: 6px;
|
| 584 |
+
padding: 8px 16px;
|
| 585 |
+
font-weight: 500;
|
| 586 |
+
transition: all 0.3s ease;
|
| 587 |
+
}
|
| 588 |
+
|
| 589 |
+
.secondary-btn:hover {
|
| 590 |
+
border-color: #1e40af;
|
| 591 |
+
color: #1e40af;
|
| 592 |
+
background: #f8fafc;
|
| 593 |
+
}
|
| 594 |
+
|
| 595 |
+
.custom-input {
|
| 596 |
+
border: 2px solid #d1d5db;
|
| 597 |
+
border-radius: 8px;
|
| 598 |
+
padding: 12px;
|
| 599 |
+
font-size: 16px;
|
| 600 |
+
transition: border-color 0.3s ease;
|
| 601 |
+
}
|
| 602 |
+
|
| 603 |
+
.custom-input:focus {
|
| 604 |
+
border-color: #3b82f6;
|
| 605 |
+
box-shadow: 0 0 0 3px rgba(59, 130, 246, 0.1);
|
| 606 |
+
}
|
| 607 |
+
|
| 608 |
+
.results-container {
|
| 609 |
+
background: white;
|
| 610 |
+
border: 1px solid #e5e7eb;
|
| 611 |
+
border-radius: 10px;
|
| 612 |
+
padding: 20px;
|
| 613 |
+
margin: 15px 0;
|
| 614 |
+
box-shadow: 0 2px 10px rgba(0,0,0,0.05);
|
| 615 |
+
}
|
| 616 |
+
|
| 617 |
+
.section-header {
|
| 618 |
+
background: linear-gradient(135deg, #64748b 0%, #475569 100%);
|
| 619 |
+
color: white;
|
| 620 |
+
padding: 15px 20px;
|
| 621 |
+
border-radius: 10px;
|
| 622 |
+
margin: 20px 0 15px 0;
|
| 623 |
+
font-weight: 600;
|
| 624 |
+
}
|
| 625 |
+
"""
|
| 626 |
+
|
| 627 |
+
# Create Gradio interface
|
| 628 |
+
with gr.Blocks(css=custom_css, title="Contact Discovery Platform", theme=gr.themes.Base()) as app:
|
| 629 |
+
gr.HTML("""
|
| 630 |
+
<div class="main-header">
|
| 631 |
+
<h1>Contact Discovery Platform</h1>
|
| 632 |
+
<p>Professional Contact Discovery & Lead Generation Tool</p>
|
| 633 |
+
<p style="font-size: 0.95em; opacity: 0.8;">Advanced website analysis for contact intelligence gathering</p>
|
| 634 |
+
</div>
|
| 635 |
+
""")
|
| 636 |
+
|
| 637 |
+
with gr.Tabs():
|
| 638 |
+
# Single Website Search Tab
|
| 639 |
+
with gr.TabItem("Single Website Search"):
|
| 640 |
+
with gr.Row():
|
| 641 |
+
with gr.Column(scale=2):
|
| 642 |
+
gr.HTML('<div class="section-header">Search Parameters</div>')
|
| 643 |
+
|
| 644 |
+
website_input = gr.Textbox(
|
| 645 |
+
label="Target Website URL",
|
| 646 |
+
placeholder="Enter company website (e.g., techflowsolutions.com)",
|
| 647 |
+
value="",
|
| 648 |
+
elem_classes=["custom-input"]
|
| 649 |
+
)
|
| 650 |
+
|
| 651 |
+
with gr.Row():
|
| 652 |
+
max_results = gr.Slider(
|
| 653 |
+
minimum=1,
|
| 654 |
+
maximum=20,
|
| 655 |
+
value=8,
|
| 656 |
+
step=1,
|
| 657 |
+
label="Maximum Results",
|
| 658 |
+
elem_classes=["custom-input"]
|
| 659 |
+
)
|
| 660 |
+
|
| 661 |
+
search_btn = gr.Button(
|
| 662 |
+
"Execute Search",
|
| 663 |
+
variant="primary",
|
| 664 |
+
size="lg",
|
| 665 |
+
elem_classes=["primary-btn"]
|
| 666 |
+
)
|
| 667 |
+
|
| 668 |
+
gr.HTML('<div class="section-header">Search Results</div>')
|
| 669 |
+
|
| 670 |
+
with gr.Row():
|
| 671 |
+
results_display = gr.Textbox(
|
| 672 |
+
label="Contact Intelligence Report",
|
| 673 |
+
lines=18,
|
| 674 |
+
max_lines=35,
|
| 675 |
+
show_copy_button=True,
|
| 676 |
+
elem_classes=["results-container"]
|
| 677 |
+
)
|
| 678 |
+
|
| 679 |
+
csv_output = gr.Textbox(
|
| 680 |
+
label="Export Data (CSV Format)",
|
| 681 |
+
lines=18,
|
| 682 |
+
max_lines=35,
|
| 683 |
+
show_copy_button=True,
|
| 684 |
+
elem_classes=["results-container"]
|
| 685 |
+
)
|
| 686 |
+
|
| 687 |
+
# CSV Upload Tab
|
| 688 |
+
with gr.TabItem("CSV Bulk Search"):
|
| 689 |
+
with gr.Row():
|
| 690 |
+
with gr.Column(scale=2):
|
| 691 |
+
gr.HTML('<div class="section-header">CSV Upload</div>')
|
| 692 |
+
|
| 693 |
+
csv_file = gr.File(
|
| 694 |
+
label="Upload CSV File",
|
| 695 |
+
file_types=[".csv"],
|
| 696 |
+
elem_classes=["custom-input"]
|
| 697 |
+
)
|
| 698 |
+
|
| 699 |
+
gr.HTML("""
|
| 700 |
+
<div style="background: #f8fafc; padding: 15px; border-radius: 8px; border-left: 4px solid #1e40af; margin: 10px 0;">
|
| 701 |
+
<strong>CSV Format - Multiple Options:</strong><br>
|
| 702 |
+
<strong>Option 1:</strong> Website URLs in <strong>Column H (8th column)</strong><br>
|
| 703 |
+
<strong>Option 2:</strong> Column header named 'website', 'url', 'domain', 'site', or 'web'<br>
|
| 704 |
+
<strong>Option 3:</strong> System will auto-detect columns with valid URLs<br><br>
|
| 705 |
+
<strong>Examples:</strong> techflowsolutions.com, https://example.com, www.company.com<br>
|
| 706 |
+
<strong>Note:</strong> The system will show detailed debugging information about your CSV structure
|
| 707 |
+
</div>
|
| 708 |
+
""")
|
| 709 |
+
|
| 710 |
+
with gr.Row():
|
| 711 |
+
csv_max_results = gr.Slider(
|
| 712 |
+
minimum=1,
|
| 713 |
+
maximum=50,
|
| 714 |
+
value=20,
|
| 715 |
+
step=1,
|
| 716 |
+
label="Maximum Results",
|
| 717 |
+
elem_classes=["custom-input"]
|
| 718 |
+
)
|
| 719 |
+
|
| 720 |
+
csv_search_btn = gr.Button(
|
| 721 |
+
"Process CSV",
|
| 722 |
+
variant="primary",
|
| 723 |
+
size="lg",
|
| 724 |
+
elem_classes=["primary-btn"]
|
| 725 |
+
)
|
| 726 |
+
|
| 727 |
+
gr.HTML('<div class="section-header">CSV Results</div>')
|
| 728 |
+
|
| 729 |
+
with gr.Row():
|
| 730 |
+
csv_results_display = gr.Textbox(
|
| 731 |
+
label="CSV Processing Report",
|
| 732 |
+
lines=18,
|
| 733 |
+
max_lines=35,
|
| 734 |
+
show_copy_button=True,
|
| 735 |
+
elem_classes=["results-container"]
|
| 736 |
+
)
|
| 737 |
+
|
| 738 |
+
csv_export_output = gr.Textbox(
|
| 739 |
+
label="Export Data (CSV Format)",
|
| 740 |
+
lines=18,
|
| 741 |
+
max_lines=35,
|
| 742 |
+
show_copy_button=True,
|
| 743 |
+
elem_classes=["results-container"]
|
| 744 |
+
)
|
| 745 |
+
|
| 746 |
+
# Sample websites section
|
| 747 |
+
with gr.Accordion("Sample Websites Database", open=False):
|
| 748 |
+
gr.HTML('<div style="background: #f8fafc; padding: 15px; border-radius: 8px; border-left: 4px solid #1e40af;">')
|
| 749 |
+
sample_websites = gr.Textbox(
|
| 750 |
+
label="Available Websites in Database",
|
| 751 |
+
value=get_all_available_websites(),
|
| 752 |
+
lines=8,
|
| 753 |
+
interactive=False,
|
| 754 |
+
elem_classes=["custom-input"]
|
| 755 |
+
)
|
| 756 |
+
gr.HTML('</div>')
|
| 757 |
+
|
| 758 |
+
# Quick search buttons
|
| 759 |
+
gr.HTML('<div class="section-header">Quick Access Sample Websites</div>')
|
| 760 |
+
|
| 761 |
+
with gr.Row():
|
| 762 |
+
quick_btn1 = gr.Button("TechFlow Solutions", size="sm", elem_classes=["secondary-btn"])
|
| 763 |
+
quick_btn2 = gr.Button("GreenLeaf Consulting", size="sm", elem_classes=["secondary-btn"])
|
| 764 |
+
quick_btn3 = gr.Button("BlueSky Marketing", size="sm", elem_classes=["secondary-btn"])
|
| 765 |
+
quick_btn4 = gr.Button("Quantum Dynamics", size="sm", elem_classes=["secondary-btn"])
|
| 766 |
+
|
| 767 |
+
with gr.Row():
|
| 768 |
+
quick_btn5 = gr.Button("Stellar Logistics", size="sm", elem_classes=["secondary-btn"])
|
| 769 |
+
quick_btn6 = gr.Button("Nexus Financial", size="sm", elem_classes=["secondary-btn"])
|
| 770 |
+
quick_btn7 = gr.Button("Horizon Health", size="sm", elem_classes=["secondary-btn"])
|
| 771 |
+
quick_btn8 = gr.Button("Phoenix Manufacturing", size="sm", elem_classes=["secondary-btn"])
|
| 772 |
+
|
| 773 |
+
# Event handlers
|
| 774 |
+
search_btn.click(
|
| 775 |
+
fn=search_website_contacts,
|
| 776 |
+
inputs=[website_input, max_results],
|
| 777 |
+
outputs=[results_display, csv_output]
|
| 778 |
+
)
|
| 779 |
+
|
| 780 |
+
csv_search_btn.click(
|
| 781 |
+
fn=search_csv_websites,
|
| 782 |
+
inputs=[csv_file, csv_max_results],
|
| 783 |
+
outputs=[csv_results_display, csv_export_output]
|
| 784 |
+
)
|
| 785 |
+
|
| 786 |
+
# Quick search button handlers
|
| 787 |
+
quick_btn1.click(lambda: "techflowsolutions.com", outputs=website_input)
|
| 788 |
+
quick_btn2.click(lambda: "greenleafconsult.com", outputs=website_input)
|
| 789 |
+
quick_btn3.click(lambda: "blueskymarketing.net", outputs=website_input)
|
| 790 |
+
quick_btn4.click(lambda: "quantumdynamics.org", outputs=website_input)
|
| 791 |
+
quick_btn5.click(lambda: "stellarlogistics.biz", outputs=website_input)
|
| 792 |
+
quick_btn6.click(lambda: "nexusfinancial.pro", outputs=website_input)
|
| 793 |
+
quick_btn7.click(lambda: "horizonhealth.care", outputs=website_input)
|
| 794 |
+
quick_btn8.click(lambda: "phoenixmfg.com", outputs=website_input)
|
| 795 |
+
|
| 796 |
+
# Examples
|
| 797 |
+
gr.Examples(
|
| 798 |
+
examples=[
|
| 799 |
+
["techflowsolutions.com", 5],
|
| 800 |
+
["greenleafconsult.com", 3],
|
| 801 |
+
["blueskymarketing.net", 4],
|
| 802 |
+
["quantumdynamics.org", 6]
|
| 803 |
+
],
|
| 804 |
+
inputs=[website_input, max_results],
|
| 805 |
+
label="Sample Searches"
|
| 806 |
+
)
|
| 807 |
+
|
| 808 |
+
# Footer
|
| 809 |
+
gr.HTML("""
|
| 810 |
+
<div style="text-align: center; padding: 30px 20px; background: linear-gradient(135deg, #64748b 0%, #475569 100%); color: white; border-radius: 15px; margin-top: 30px;">
|
| 811 |
+
<h3 style="margin: 0 0 10px 0;">Contact Intelligence Platform</h3>
|
| 812 |
+
<p style="margin: 0; opacity: 0.9;">Professional-grade contact discovery and lead generation technology</p>
|
| 813 |
+
<p style="margin: 10px 0 0 0; font-size: 0.9em; opacity: 0.7;">Powered by advanced web intelligence algorithms</p>
|
| 814 |
+
</div>
|
| 815 |
+
""")
|
| 816 |
+
|
| 817 |
+
if __name__ == "__main__":
|
| 818 |
+
app.launch()
|