File size: 10,238 Bytes
d712cef | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 | import pandas as pd
import sqlite3
import requests
import json
import os
import re
import asyncio
from pathlib import Path
# Import the necessary components from your email pipeline script.
# Ensure "emailspecific" matches the actual filename of your email script (without .py).
import sys
# Add the directory containing email_specific.py to the system path
sys.path.append(os.path.join(os.environ.get('WORKSPACE_ROOT', '.'), 'src_2/ColdEmail/'))
from email_specific import run_batch, write_excel, DEEP_RES_PDF_DIR, BATCH_SIZE, EXCEL_OUTPUT
# Outreach data pipeline β no email validation or website scraping needed
# --- CONFIGURATION ---
LLM_URL = "https://unscotched-devon-interpapillary.ngrok-free.dev/generate"
DB_PATH = os.path.join(os.environ.get('WORKSPACE_ROOT', '.'), 'Database/outreach_data/excel_data.db')
os.makedirs(os.path.dirname(DB_PATH), exist_ok=True)
# --- 1. DATABASE SETUP ---
def init_db():
conn = sqlite3.connect(DB_PATH)
cursor = conn.cursor()
cursor.execute('''
CREATE TABLE IF NOT EXISTS outreach_companies (
id INTEGER PRIMARY KEY AUTOINCREMENT,
company_name TEXT,
website TEXT,
address TEXT,
total_shipments TEXT,
top_suppliers TEXT,
hs_codes TEXT,
company_description TEXT,
key_executives TEXT,
deep_research_pdf TEXT,
email TEXT,
status TEXT DEFAULT 'under_review',
imported_at DATETIME DEFAULT CURRENT_TIMESTAMP
)
''')
conn.commit()
conn.close()
# --- 2. LLM INTERACTION ---
def call_llm(system_prompt, user_query):
payload = {
"system_prompt": system_prompt,
"query": user_query,
"max_new_tokens": 5000 # Keep it short, we only need a JSON object
}
response = requests.post(LLM_URL, json=payload)
if response.status_code == 200:
return response.json()["response"]
raise Exception(f"LLM Error: {response.text}")
# --- 3. DYNAMIC COLUMN MAPPING ---
def get_column_mapping(raw_columns):
print(f"π§ Asking LLM to map these columns: {raw_columns}")
system_prompt = (
"You are a precise data engineering assistant. You map raw spreadsheet columns to a target schema. "
"Output ONLY a valid JSON object. No explanations, no markdown formatting."
)
user_query = f"""
Target Schema exact keys required:
1. "Company Name"
2. "Website"
3. "Address"
4. "Total Shipments"
5. "Top Suppliers"
6. "HS Codes"
7. "Company Description"
8. "Key Executives"
9. "Deep Research PDF"
10. "Email"
Raw column names found in the file: {raw_columns}
Match the exact raw column names from the list above to the target schema.
CRITICAL RULE: The values in your JSON must be EXACT STRINGS chosen ONLY from the 'Raw column names found in the file' list. DO NOT invent or guess column names.
If a target concept does not exist in the raw columns, map it to null.
Format example: {{"Company Name": "Company Name", "Website": "Website", "Email": null}}
"""
json_raw = call_llm(system_prompt, user_query)
# Clean the LLM output (Robust JSON parsing)
json_raw = re.sub(r"```json|```", "", json_raw).strip()
match = re.search(r'\{.*\}', json_raw, re.DOTALL)
if match:
json_raw = match.group(0)
try:
mapping = json.loads(json_raw)
print(f"β
LLM Mapping decided: {json.dumps(mapping, indent=2)}")
return mapping
except Exception as e:
print(f"β Failed to parse LLM mapping JSON. Raw output: {json_raw}")
return None
# --- 4. MAIN PROCESSING LOGIC ---
def process_excel(excel_path):
print(f"\nπ Processing Excel: {excel_path}")
# Step A: Load the file β supports .xlsx, .xls and .numbers
try:
ext = Path(excel_path).suffix.lower()
if ext == '.numbers':
import numbers_parser
doc = numbers_parser.Document(str(excel_path))
sheet = doc.sheets[0]
table = sheet.tables[0]
rows = list(table.iter_rows())
headers = [cell.value for cell in rows[0]]
data = [[cell.value for cell in row] for row in rows[1:]]
df = pd.DataFrame(data, columns=headers)
print(f" β Loaded .numbers file: {len(df)} rows, {len(headers)} columns")
else:
df = pd.read_excel(excel_path)
except Exception as e:
print(f"β Could not read file: {e}")
return
# Convert pandas NaN values to None (so SQLite accepts them)
df = df.where(pd.notnull(df), None)
raw_columns = df.columns.tolist()
# Step B: Get the intelligent mapping from the LLM
mapping = get_column_mapping(raw_columns)
if not mapping:
return
# Step C: Extract and save to Database
conn = sqlite3.connect(DB_PATH)
try:
import openpyxl
wb = openpyxl.load_workbook(excel_path)
ws = wb.active
# Find the 'Status' column index
status_col = None
for col_idx in range(1, ws.max_column + 1):
if ws.cell(row=1, column=col_idx).value == 'Status':
status_col = col_idx
break
except Exception as e:
print(f"β οΈ Could not load workbook for status updates: {e}")
wb = None
cursor = conn.cursor()
records_saved = 0
for index, row in df.iterrows():
# 1. Skip if already marked as Completed in the spreadsheet
if 'Status' in df.columns and row['Status'] == 'Completed':
continue
# Resolve each column via the LLM mapping
def get(key):
col = mapping.get(key)
if col and col in df.columns:
val = row[col]
if val is None or (isinstance(val, float) and pd.isna(val)):
return None
s = str(val).strip()
return s if s and s.lower() != 'nan' else None
return None
company_name = get("Company Name")
website = get("Website")
address = get("Address")
total_shipments = get("Total Shipments")
top_suppliers = get("Top Suppliers")
hs_codes = get("HS Codes")
company_description = get("Company Description")
key_executives = get("Key Executives")
deep_research_pdf = get("Deep Research PDF")
email = get("Email") or "not updated"
# Skip completely empty rows
if not any([company_name, website, address]):
continue
# 2. Skip if company already exists in the database (duplicate check)
cursor.execute('SELECT id FROM outreach_companies WHERE company_name = ? AND website = ?', (company_name, website))
if cursor.fetchone():
print(f" βοΈ Skipping (already in DB): {company_name}")
# Ensure Excel is marked completed if it's already in DB but not marked
if wb and status_col:
ws.cell(row=index + 2, column=status_col, value='Completed')
continue
try:
cursor.execute('''
INSERT INTO outreach_companies
(company_name, website, address, total_shipments, top_suppliers,
hs_codes, company_description, key_executives, deep_research_pdf, email)
VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
''', (company_name, website, address, total_shipments, top_suppliers,
hs_codes, company_description, key_executives, deep_research_pdf, email))
# Update status in the workbook (index + 2 because pandas is 0-based and Excel is 1-based + header)
if wb and status_col:
ws.cell(row=index + 2, column=status_col, value='Completed')
records_saved += 1
print(f" β
Saved: {company_name} | email: {email}")
except Exception as e:
print(f" β Error saving {company_name}: {e}")
conn.commit()
conn.close()
# Step D: Save the Excel file back with updated status
if wb:
try:
wb.save(excel_path)
print(f"π Excel file updated with 'Completed' status: {excel_path}")
except Exception as e:
print(f"β Could not save updated Excel: {e}")
print(f"π Successfully saved {records_saved} records to the database.")
if __name__ == "__main__":
# Ensure DB is initialized before anything else
init_db()
print(f"\n{'β'*65}")
print(f" TEXBASE Cold Email Pipeline β Batch of {BATCH_SIZE}")
print(f"{'β'*65}")
print(f" π WORKSPACE_ROOT: {os.environ.get('WORKSPACE_ROOT', '.')}")
print(f" π DB_PATH (Source): {DB_PATH}")
all_states = asyncio.run(run_batch(BATCH_SIZE))
# Filter out empty states (queue was empty for that slot)
valid_states = [s for s in all_states if s.get("company_id")]
if valid_states:
write_excel(valid_states, EXCEL_OUTPUT)
print(f"\n{'β'*65}")
print(" BATCH SUMMARY")
print(f"{'β'*65}")
for s in valid_states:
print(f" β’ {s.get('company_name', '?')} β "
f"result: {s.get('result', s.get('error', ''))}")
print(f"\n Excel output β {EXCEL_OUTPUT}")
else:
print("\n No new companies processed by the generation pipeline.")
# =========================================================
# PHASE 2: IMPORT EXCEL DATA TO DB
# =========================================================
# We process the Excel file regardless of Phase 1 results (supports manual edits)
excel_path_to_process = Path(EXCEL_OUTPUT)
if os.path.exists(excel_path_to_process):
print(f"\n π₯ Triggering Excel Import for: {excel_path_to_process}")
process_excel(excel_path_to_process)
else:
print(f"β Excel file not found for import: {excel_path_to_process}") |