Spaces:
Sleeping
Sleeping
File size: 5,535 Bytes
b7d0804 | 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 |
from datetime import datetime, timezone
from typing import Dict, Any, List
from app.storage.processed_storage import (
read_processed_chunks,
read_processed_metadata
)
from app.schemas.graph_schema import (
DocumentGraph,
GraphEntity,
GraphRelation
)
from app.graph.entity_extractor import extract_entities_from_text
from app.graph.relation_extractor import extract_relations_from_text
from app.graph.graph_storage import save_document_graph
def get_value(obj, key: str, default=None):
if isinstance(obj, dict):
return obj.get(key, default)
return getattr(obj, key, default)
def add_unique(existing_list: List, value):
if value is None:
return
if value not in existing_list:
existing_list.append(value)
def build_document_graph(document_id: str) -> Dict[str, Any]:
chunks = read_processed_chunks(document_id)
if chunks is None:
return {
"status": "failed",
"message": "No processed chunks found for this document. Upload and process the document first.",
"document_id": document_id
}
metadata = read_processed_metadata(document_id) or {}
source_file_name = None
if isinstance(metadata, dict):
source_file_name = metadata.get("source_file_name") or metadata.get("filename")
entity_map: Dict[str, GraphEntity] = {}
relation_map: Dict[str, GraphRelation] = {}
for chunk in chunks:
content = (
get_value(chunk, "content")
or get_value(chunk, "text")
or ""
)
if not content:
continue
chunk_id = get_value(chunk, "chunk_id", "")
page_number = get_value(chunk, "page_number", None)
extracted_entities = extract_entities_from_text(content)
for item in extracted_entities:
entity_id = item["entity_id"]
if entity_id not in entity_map:
entity_map[entity_id] = GraphEntity(
entity_id=entity_id,
name=item["name"],
entity_type=item["entity_type"],
mention_count=0
)
entity = entity_map[entity_id]
entity.mention_count += content.lower().count(item["name"].lower())
add_unique(entity.chunk_ids, chunk_id)
add_unique(entity.pages, page_number)
if len(entity.evidence) < 5:
entity.evidence.append(
{
"chunk_id": chunk_id,
"page_number": page_number,
"text_preview": content[:250]
}
)
extracted_relations = extract_relations_from_text(
text=content,
entities=extracted_entities
)
for item in extracted_relations:
rel_id = item["relation_id"]
if rel_id not in relation_map:
relation_map[rel_id] = GraphRelation(
relation_id=rel_id,
source_entity_id=item["source_entity_id"],
target_entity_id=item["target_entity_id"],
source_name=item["source_name"],
target_name=item["target_name"],
relation_type=item["relation_type"],
weight=0
)
relation = relation_map[rel_id]
relation.weight += 1
add_unique(relation.chunk_ids, chunk_id)
add_unique(relation.pages, page_number)
if len(relation.evidence) < 5:
relation.evidence.append(
{
"chunk_id": chunk_id,
"page_number": page_number,
"sentence": item["evidence_sentence"]
}
)
entities = sorted(
entity_map.values(),
key=lambda entity: entity.mention_count,
reverse=True
)
relations = sorted(
relation_map.values(),
key=lambda relation: relation.weight,
reverse=True
)
graph = DocumentGraph(
document_id=document_id,
source_file_name=source_file_name,
total_entities=len(entities),
total_relations=len(relations),
entities=entities,
relations=relations,
build_metadata={
"builder": "rule_based_entity_relation_extractor",
"created_at": datetime.now(timezone.utc).isoformat(),
"chunk_count": len(chunks),
"note": "This is the graph foundation layer before adding a dedicated graph database."
}
)
save_document_graph(graph)
return {
"status": "success",
"message": "Document graph built successfully.",
"document_id": document_id,
"total_entities": graph.total_entities,
"total_relations": graph.total_relations,
"top_entities": [
{
"entity_id": entity.entity_id,
"name": entity.name,
"type": entity.entity_type,
"mention_count": entity.mention_count
}
for entity in entities[:15]
],
"top_relations": [
{
"source": relation.source_name,
"relation": relation.relation_type,
"target": relation.target_name,
"weight": relation.weight
}
for relation in relations[:15]
]
}
|