Upload 59 files
Browse files- app/api/routes/analyze.py +23 -0
- app/api/routes/ingest.py +24 -0
app/api/routes/analyze.py
CHANGED
|
@@ -91,6 +91,21 @@ async def analyze_text(request: AnalyzeRequest):
|
|
| 91 |
|
| 92 |
db = next(get_db())
|
| 93 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 94 |
# Process entities
|
| 95 |
for entity in result.entities:
|
| 96 |
entity_data = ExtractedEntityResponse(
|
|
@@ -116,6 +131,9 @@ async def analyze_text(request: AnalyzeRequest):
|
|
| 116 |
if coords:
|
| 117 |
lat, lng = coords
|
| 118 |
|
|
|
|
|
|
|
|
|
|
| 119 |
# Create new entity
|
| 120 |
new_entity = Entity(
|
| 121 |
name=entity.name,
|
|
@@ -124,6 +142,7 @@ async def analyze_text(request: AnalyzeRequest):
|
|
| 124 |
source="llm_extraction",
|
| 125 |
latitude=lat,
|
| 126 |
longitude=lng,
|
|
|
|
| 127 |
properties={"role": entity.role, "aliases": entity.aliases}
|
| 128 |
)
|
| 129 |
db.add(new_entity)
|
|
@@ -166,10 +185,14 @@ async def analyze_text(request: AnalyzeRequest):
|
|
| 166 |
).first()
|
| 167 |
|
| 168 |
if not existing_rel:
|
|
|
|
|
|
|
|
|
|
| 169 |
new_rel = Relationship(
|
| 170 |
source_id=source_entity.id,
|
| 171 |
target_id=target_entity.id,
|
| 172 |
type=rel.relationship_type,
|
|
|
|
| 173 |
properties={"context": rel.context}
|
| 174 |
)
|
| 175 |
db.add(new_rel)
|
|
|
|
| 91 |
|
| 92 |
db = next(get_db())
|
| 93 |
|
| 94 |
+
# Helper function to parse date strings
|
| 95 |
+
def parse_date(date_str):
|
| 96 |
+
if not date_str:
|
| 97 |
+
return None
|
| 98 |
+
from datetime import datetime
|
| 99 |
+
try:
|
| 100 |
+
# Try YYYY-MM-DD format
|
| 101 |
+
return datetime.strptime(date_str[:10], "%Y-%m-%d")
|
| 102 |
+
except:
|
| 103 |
+
try:
|
| 104 |
+
# Try YYYY format
|
| 105 |
+
return datetime.strptime(date_str[:4], "%Y")
|
| 106 |
+
except:
|
| 107 |
+
return None
|
| 108 |
+
|
| 109 |
# Process entities
|
| 110 |
for entity in result.entities:
|
| 111 |
entity_data = ExtractedEntityResponse(
|
|
|
|
| 131 |
if coords:
|
| 132 |
lat, lng = coords
|
| 133 |
|
| 134 |
+
# Parse event_date if available
|
| 135 |
+
event_date = parse_date(getattr(entity, 'event_date', None))
|
| 136 |
+
|
| 137 |
# Create new entity
|
| 138 |
new_entity = Entity(
|
| 139 |
name=entity.name,
|
|
|
|
| 142 |
source="llm_extraction",
|
| 143 |
latitude=lat,
|
| 144 |
longitude=lng,
|
| 145 |
+
event_date=event_date,
|
| 146 |
properties={"role": entity.role, "aliases": entity.aliases}
|
| 147 |
)
|
| 148 |
db.add(new_entity)
|
|
|
|
| 185 |
).first()
|
| 186 |
|
| 187 |
if not existing_rel:
|
| 188 |
+
# Parse event_date if available
|
| 189 |
+
rel_event_date = parse_date(getattr(rel, 'event_date', None))
|
| 190 |
+
|
| 191 |
new_rel = Relationship(
|
| 192 |
source_id=source_entity.id,
|
| 193 |
target_id=target_entity.id,
|
| 194 |
type=rel.relationship_type,
|
| 195 |
+
event_date=rel_event_date,
|
| 196 |
properties={"context": rel.context}
|
| 197 |
)
|
| 198 |
db.add(new_rel)
|
app/api/routes/ingest.py
CHANGED
|
@@ -5,6 +5,7 @@ Endpoints para importar dados de fontes externas
|
|
| 5 |
from fastapi import APIRouter, Depends, HTTPException, BackgroundTasks
|
| 6 |
from sqlalchemy.orm import Session
|
| 7 |
from typing import Optional, List
|
|
|
|
| 8 |
import asyncio
|
| 9 |
|
| 10 |
from app.core.database import get_db
|
|
@@ -17,6 +18,21 @@ from app.services.geocoding import geocode
|
|
| 17 |
router = APIRouter(prefix="/ingest", tags=["Data Ingestion"])
|
| 18 |
|
| 19 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 20 |
# ========== Wikipedia ==========
|
| 21 |
|
| 22 |
@router.get("/wikipedia/search")
|
|
@@ -96,6 +112,9 @@ async def import_from_wikipedia(
|
|
| 96 |
if coords:
|
| 97 |
lat, lng = coords
|
| 98 |
|
|
|
|
|
|
|
|
|
|
| 99 |
new_ent = Entity(
|
| 100 |
name=ext_entity.name,
|
| 101 |
type=ext_entity.type if ext_entity.type in ["person", "organization", "location", "event"] else "person",
|
|
@@ -103,6 +122,7 @@ async def import_from_wikipedia(
|
|
| 103 |
source="wikipedia_extraction",
|
| 104 |
latitude=lat,
|
| 105 |
longitude=lng,
|
|
|
|
| 106 |
properties={"role": ext_entity.role, "aliases": ext_entity.aliases, "extracted_from": title}
|
| 107 |
)
|
| 108 |
db.add(new_ent)
|
|
@@ -124,10 +144,14 @@ async def import_from_wikipedia(
|
|
| 124 |
).first()
|
| 125 |
|
| 126 |
if not existing_rel:
|
|
|
|
|
|
|
|
|
|
| 127 |
new_rel = Relationship(
|
| 128 |
source_id=source_ent.id,
|
| 129 |
target_id=target_ent.id,
|
| 130 |
type=rel.relationship_type,
|
|
|
|
| 131 |
properties={"context": rel.context, "extracted_from": title}
|
| 132 |
)
|
| 133 |
db.add(new_rel)
|
|
|
|
| 5 |
from fastapi import APIRouter, Depends, HTTPException, BackgroundTasks
|
| 6 |
from sqlalchemy.orm import Session
|
| 7 |
from typing import Optional, List
|
| 8 |
+
from datetime import datetime
|
| 9 |
import asyncio
|
| 10 |
|
| 11 |
from app.core.database import get_db
|
|
|
|
| 18 |
router = APIRouter(prefix="/ingest", tags=["Data Ingestion"])
|
| 19 |
|
| 20 |
|
| 21 |
+
def parse_event_date(date_str):
|
| 22 |
+
"""Parse date string to datetime object"""
|
| 23 |
+
if not date_str:
|
| 24 |
+
return None
|
| 25 |
+
try:
|
| 26 |
+
# Try YYYY-MM-DD format
|
| 27 |
+
return datetime.strptime(date_str[:10], "%Y-%m-%d")
|
| 28 |
+
except:
|
| 29 |
+
try:
|
| 30 |
+
# Try YYYY format
|
| 31 |
+
return datetime.strptime(date_str[:4], "%Y")
|
| 32 |
+
except:
|
| 33 |
+
return None
|
| 34 |
+
|
| 35 |
+
|
| 36 |
# ========== Wikipedia ==========
|
| 37 |
|
| 38 |
@router.get("/wikipedia/search")
|
|
|
|
| 112 |
if coords:
|
| 113 |
lat, lng = coords
|
| 114 |
|
| 115 |
+
# Parse event_date
|
| 116 |
+
event_date = parse_event_date(getattr(ext_entity, 'event_date', None))
|
| 117 |
+
|
| 118 |
new_ent = Entity(
|
| 119 |
name=ext_entity.name,
|
| 120 |
type=ext_entity.type if ext_entity.type in ["person", "organization", "location", "event"] else "person",
|
|
|
|
| 122 |
source="wikipedia_extraction",
|
| 123 |
latitude=lat,
|
| 124 |
longitude=lng,
|
| 125 |
+
event_date=event_date,
|
| 126 |
properties={"role": ext_entity.role, "aliases": ext_entity.aliases, "extracted_from": title}
|
| 127 |
)
|
| 128 |
db.add(new_ent)
|
|
|
|
| 144 |
).first()
|
| 145 |
|
| 146 |
if not existing_rel:
|
| 147 |
+
# Parse relationship event_date
|
| 148 |
+
rel_event_date = parse_event_date(getattr(rel, 'event_date', None))
|
| 149 |
+
|
| 150 |
new_rel = Relationship(
|
| 151 |
source_id=source_ent.id,
|
| 152 |
target_id=target_ent.id,
|
| 153 |
type=rel.relationship_type,
|
| 154 |
+
event_date=rel_event_date,
|
| 155 |
properties={"context": rel.context, "extracted_from": title}
|
| 156 |
)
|
| 157 |
db.add(new_rel)
|