Spaces:
Running
Running
File size: 6,625 Bytes
09801ca | 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 | # Graph builder module
import networkx as nx
import re
import networkx as nx
import re
from config.settings import Settings
from graph.schema import detect_schema
from utils.paths import STORAGE_BASE
def detect_currency_from_value(value: str) -> str:
"""Detect currency from a value string like '$100' or '₹1000'."""
if not isinstance(value, str):
return 'USD'
value = value.strip()
# Check multi-char symbols first
if value.startswith('A$') or 'AUD' in value.upper():
return 'AUD'
if value.startswith('C$') or 'CAD' in value.upper():
return 'CAD'
if value.startswith('S$') or 'SGD' in value.upper():
return 'SGD'
if value.startswith('CHF') or 'CHF' in value.upper():
return 'CHF'
# Single char symbols
if '€' in value:
return 'EUR'
if '£' in value:
return 'GBP'
if '₹' in value:
return 'INR'
if '$' in value and not any(x in value for x in ['A$', 'C$', 'S$']):
return 'USD'
if 'Â¥' in value:
return 'JPY' if '.' not in value else 'CNY'
return 'USD'
class GraphBuilder:
@staticmethod
def build(company_id: str, tables: list, default_currency: str = 'USD', source_files: list = None):
"""
Build knowledge graph from tabular data with multi-currency and source tracking.
Args:
company_id: Company identifier
tables: List of pandas DataFrames
default_currency: Default currency if none detected
source_files: List of source file names (one per table)
"""
G = nx.Graph()
currencies_detected = {} # Track currency counts
sources_tracked = {} # Track records per source file
for table_idx, df in enumerate(tables):
# Get source file name for this table
source_file = source_files[table_idx] if source_files and table_idx < len(source_files) else f"file_{table_idx}"
sources_tracked[source_file] = sources_tracked.get(source_file, 0)
schema = detect_schema(df.columns)
print(f"📊 Detected schema for {source_file}: {schema}")
print(f"📊 DataFrame columns: {list(df.columns)}")
print(f"📊 DataFrame shape: {df.shape}")
for idx, row in df.iterrows():
# Create invoice node with source file tracking
inv_id = f"invoice:{row[schema['invoice']]}_{source_file}" if "invoice" in schema else f"invoice:row_{idx}_{source_file}"
# Get amount and detect currency from the value
amount_raw = row.get(schema.get("amount"), 0.0)
currency = default_currency
try:
if isinstance(amount_raw, str):
# Detect currency from the string value
currency = detect_currency_from_value(amount_raw)
# Strip currency symbols and commas
amount = re.sub(r'[₹$€£¥,\s]', '', amount_raw)
amount = float(amount)
else:
amount = float(amount_raw)
except:
amount = 0.0
# Track currency frequency
currencies_detected[currency] = currencies_detected.get(currency, 0) + 1
sources_tracked[source_file] = sources_tracked.get(source_file, 0) + 1
# Store amount, currency AND source_file in the invoice node
G.add_node(inv_id, type="invoice", label=inv_id, amount=amount, currency=currency, source_file=source_file)
# Connect to customer (Fallback if missing)
if "customer" in schema:
val = str(row[schema["customer"]])
node_id = f"customer:{val}"
if not G.has_node(node_id):
G.add_node(node_id, type="customer", label=val)
G.add_edge(inv_id, node_id, relation="has_customer")
else:
# Fallback for missing customer
node_id = "customer:Unknown"
if not G.has_node(node_id):
G.add_node(node_id, type="customer", label="Unknown Customer")
G.add_edge(inv_id, node_id, relation="has_customer")
# Connect to product (Fallback if missing)
if "product" in schema:
val = str(row[schema["product"]])
node_id = f"product:{val}"
if not G.has_node(node_id):
G.add_node(node_id, type="product", label=val)
G.add_edge(inv_id, node_id, relation="has_product")
else:
# Fallback for missing product
node_id = "product:General"
if not G.has_node(node_id):
G.add_node(node_id, type="product", label="General Item")
G.add_edge(inv_id, node_id, relation="has_product")
# Connect to date (Optional)
if "date" in schema:
val = str(row[schema["date"]])
node_id = f"date:{val}"
if not G.has_node(node_id):
G.add_node(node_id, type="date", label=val)
G.add_edge(inv_id, node_id, relation="has_date")
print(f"📊 Built graph with {G.number_of_nodes()} nodes and {G.number_of_edges()} edges")
print(f"💰 Currencies detected: {currencies_detected}")
# Save graph using pickle format to USER-SPECIFIC directory
import pickle
# Primary: User-specific graph directory (Consolidated storage)
user_graph_dir = STORAGE_BASE / company_id / "graph"
user_graph_dir.mkdir(parents=True, exist_ok=True)
user_graph_path = user_graph_dir / f"{company_id}.gpickle"
with open(user_graph_path, 'wb') as f:
pickle.dump(G, f)
print(f"✅ Graph saved to user directory: {user_graph_path}")
# Also save to Settings.GRAPH_DIR for backward compatibility
output_path = Settings.GRAPH_DIR / f"{company_id}.gpickle"
output_path.parent.mkdir(parents=True, exist_ok=True)
with open(output_path, 'wb') as f:
pickle.dump(G, f)
print(f"✅ Graph also saved to: {output_path}")
return G
|