Spaces:
Sleeping
Sleeping
Update backend/app/api/ingest.py
Browse files- backend/app/api/ingest.py +107 -0
backend/app/api/ingest.py
CHANGED
|
@@ -32,6 +32,113 @@ logging.basicConfig(
|
|
| 32 |
|
| 33 |
logger = logging.getLogger(__name__)
|
| 34 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 35 |
# Setup
|
| 36 |
BASE_DIR = Path(__file__).parent.parent.parent.parent
|
| 37 |
STORAGE_PATH = Path(os.getenv('STORAGE_PATH', str(BASE_DIR / "data" / "docs")))
|
|
|
|
| 32 |
|
| 33 |
logger = logging.getLogger(__name__)
|
| 34 |
|
| 35 |
+
# ============================================
|
| 36 |
+
# FORCE DATABASE INITIALIZATION ON MODULE LOAD
|
| 37 |
+
# ============================================
|
| 38 |
+
|
| 39 |
+
def init_tables_now():
|
| 40 |
+
"""Force create tables on module import"""
|
| 41 |
+
try:
|
| 42 |
+
DB_PATH.parent.mkdir(parents=True, exist_ok=True)
|
| 43 |
+
|
| 44 |
+
with FileLock(str(LOCK_PATH), timeout=30):
|
| 45 |
+
conn = sqlite3.connect(str(DB_PATH))
|
| 46 |
+
cursor = conn.cursor()
|
| 47 |
+
|
| 48 |
+
# Check if table exists
|
| 49 |
+
cursor.execute("SELECT name FROM sqlite_master WHERE type='table' AND name='ingest_jobs'")
|
| 50 |
+
if cursor.fetchone():
|
| 51 |
+
conn.close()
|
| 52 |
+
logger.info("✅ Database tables already exist")
|
| 53 |
+
return
|
| 54 |
+
|
| 55 |
+
logger.warning("⚠️ Creating database tables...")
|
| 56 |
+
|
| 57 |
+
# Create all tables (abbreviated version shown)
|
| 58 |
+
cursor.execute("""
|
| 59 |
+
CREATE TABLE IF NOT EXISTS ingest_jobs (
|
| 60 |
+
job_id TEXT PRIMARY KEY,
|
| 61 |
+
doc_id TEXT,
|
| 62 |
+
filename TEXT NOT NULL,
|
| 63 |
+
status TEXT NOT NULL DEFAULT 'queued',
|
| 64 |
+
error_text TEXT,
|
| 65 |
+
created_at TEXT DEFAULT CURRENT_TIMESTAMP,
|
| 66 |
+
updated_at TEXT DEFAULT CURRENT_TIMESTAMP
|
| 67 |
+
)
|
| 68 |
+
""")
|
| 69 |
+
|
| 70 |
+
cursor.execute("""
|
| 71 |
+
CREATE TABLE IF NOT EXISTS documents (
|
| 72 |
+
doc_id TEXT PRIMARY KEY,
|
| 73 |
+
job_id TEXT NOT NULL,
|
| 74 |
+
path TEXT NOT NULL,
|
| 75 |
+
filename TEXT NOT NULL,
|
| 76 |
+
content_type TEXT NOT NULL,
|
| 77 |
+
uploaded_at TEXT DEFAULT CURRENT_TIMESTAMP
|
| 78 |
+
)
|
| 79 |
+
""")
|
| 80 |
+
|
| 81 |
+
cursor.execute("""
|
| 82 |
+
CREATE TABLE IF NOT EXISTS extractions (
|
| 83 |
+
doc_id TEXT PRIMARY KEY,
|
| 84 |
+
raw_text TEXT,
|
| 85 |
+
tables_json TEXT,
|
| 86 |
+
entities_json TEXT,
|
| 87 |
+
classification_json TEXT,
|
| 88 |
+
summary_text TEXT,
|
| 89 |
+
extracted_at TEXT DEFAULT CURRENT_TIMESTAMP
|
| 90 |
+
)
|
| 91 |
+
""")
|
| 92 |
+
|
| 93 |
+
cursor.execute("""
|
| 94 |
+
CREATE TABLE IF NOT EXISTS invoice_fields (
|
| 95 |
+
invoice_id INTEGER PRIMARY KEY AUTOINCREMENT,
|
| 96 |
+
doc_id TEXT NOT NULL,
|
| 97 |
+
cust_number TEXT,
|
| 98 |
+
posting_date TEXT,
|
| 99 |
+
total_open_amount REAL,
|
| 100 |
+
business_code TEXT,
|
| 101 |
+
cust_payment_terms TEXT,
|
| 102 |
+
confidence_map TEXT,
|
| 103 |
+
created_at TEXT DEFAULT CURRENT_TIMESTAMP
|
| 104 |
+
)
|
| 105 |
+
""")
|
| 106 |
+
|
| 107 |
+
cursor.execute("""
|
| 108 |
+
CREATE TABLE IF NOT EXISTS batch_jobs (
|
| 109 |
+
batch_id TEXT PRIMARY KEY,
|
| 110 |
+
total_files INTEGER,
|
| 111 |
+
message TEXT,
|
| 112 |
+
created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
|
| 113 |
+
)
|
| 114 |
+
""")
|
| 115 |
+
|
| 116 |
+
cursor.execute("""
|
| 117 |
+
CREATE TABLE IF NOT EXISTS batch_job_mapping (
|
| 118 |
+
batch_id TEXT,
|
| 119 |
+
job_id TEXT
|
| 120 |
+
)
|
| 121 |
+
""")
|
| 122 |
+
|
| 123 |
+
# Create indexes
|
| 124 |
+
cursor.execute("CREATE INDEX IF NOT EXISTS idx_ingest_jobs_status ON ingest_jobs(status)")
|
| 125 |
+
cursor.execute("CREATE INDEX IF NOT EXISTS idx_documents_job_id ON documents(job_id)")
|
| 126 |
+
cursor.execute("CREATE INDEX IF NOT EXISTS idx_invoice_fields_doc_id ON invoice_fields(doc_id)")
|
| 127 |
+
|
| 128 |
+
conn.commit()
|
| 129 |
+
conn.close()
|
| 130 |
+
|
| 131 |
+
logger.info("✅ Database tables created successfully!")
|
| 132 |
+
|
| 133 |
+
except Exception as e:
|
| 134 |
+
logger.error(f"❌ Failed to create tables: {e}")
|
| 135 |
+
import traceback
|
| 136 |
+
logger.error(traceback.format_exc())
|
| 137 |
+
|
| 138 |
+
# Run immediately on import
|
| 139 |
+
logger.info("🔍 Initializing database on module load...")
|
| 140 |
+
init_tables_now()
|
| 141 |
+
|
| 142 |
# Setup
|
| 143 |
BASE_DIR = Path(__file__).parent.parent.parent.parent
|
| 144 |
STORAGE_PATH = Path(os.getenv('STORAGE_PATH', str(BASE_DIR / "data" / "docs")))
|