christopher
commited on
Commit
·
c708265
1
Parent(s):
e113735
removed async
Browse files- database/query_processor.py +14 -37
database/query_processor.py
CHANGED
@@ -4,8 +4,6 @@ import numpy as np
|
|
4 |
from models.LexRank import degree_centrality_scores
|
5 |
import logging
|
6 |
from datetime import datetime as dt
|
7 |
-
import asyncio
|
8 |
-
from concurrent.futures import ThreadPoolExecutor
|
9 |
|
10 |
logger = logging.getLogger(__name__)
|
11 |
|
@@ -15,7 +13,6 @@ class QueryProcessor:
|
|
15 |
self.summarization_model = summarization_model
|
16 |
self.nlp_model = nlp_model
|
17 |
self.db_service = db_service
|
18 |
-
self.executor = ThreadPoolExecutor(max_workers=4) # For CPU-bound tasks
|
19 |
logger.info("QueryProcessor initialized")
|
20 |
|
21 |
async def process(
|
@@ -26,33 +23,33 @@ class QueryProcessor:
|
|
26 |
end_date: Optional[str] = None
|
27 |
) -> Dict[str, Any]:
|
28 |
try:
|
29 |
-
# Date handling
|
30 |
start_dt = self._parse_date(start_date) if start_date else None
|
31 |
end_dt = self._parse_date(end_date) if end_date else None
|
32 |
|
33 |
-
#
|
34 |
-
query_embedding =
|
35 |
logger.debug(f"Generated embedding for query: {query[:50]}...")
|
36 |
|
37 |
-
# Entity extraction
|
38 |
-
entities =
|
39 |
logger.debug(f"Extracted entities: {entities}")
|
40 |
|
41 |
-
#
|
42 |
articles = await self._execute_semantic_search(
|
43 |
query_embedding,
|
44 |
start_dt,
|
45 |
end_dt,
|
46 |
topic,
|
47 |
-
[ent[0] for ent in entities]
|
48 |
)
|
49 |
|
50 |
if not articles:
|
51 |
logger.info("No articles found matching criteria")
|
52 |
return {"message": "No articles found", "articles": []}
|
53 |
|
54 |
-
#
|
55 |
-
summary_data =
|
56 |
|
57 |
return {
|
58 |
"summary": summary_data["summary"],
|
@@ -94,35 +91,14 @@ class QueryProcessor:
|
|
94 |
logger.error(f"Semantic search failed: {str(e)}")
|
95 |
raise
|
96 |
|
97 |
-
|
98 |
-
"""
|
99 |
-
loop = asyncio.get_running_loop()
|
100 |
-
return await loop.run_in_executor(
|
101 |
-
self.executor,
|
102 |
-
lambda: self.embedding_model.encode(text).tolist()
|
103 |
-
)
|
104 |
-
|
105 |
-
async def _async_generate_summary(self, articles: List[Dict[str, Any]]) -> Dict[str, Any]:
|
106 |
-
"""Run summary generation in thread pool"""
|
107 |
-
loop = asyncio.get_running_loop()
|
108 |
-
return await loop.run_in_executor(
|
109 |
-
self.executor,
|
110 |
-
lambda: self._sync_generate_summary(articles)
|
111 |
-
)
|
112 |
-
|
113 |
-
def _sync_generate_summary(self, articles: List[Dict[str, Any]]) -> Dict[str, Any]:
|
114 |
-
"""Synchronous version for thread pool execution"""
|
115 |
try:
|
116 |
# Extract and process content
|
117 |
sentences = []
|
118 |
for article in articles:
|
119 |
if content := article.get("content"):
|
120 |
-
sentences.extend(
|
121 |
-
asyncio.run_coroutine_threadsafe(
|
122 |
-
asyncio.to_thread(self.nlp_model.tokenize_sentences, content),
|
123 |
-
loop=asyncio.get_event_loop()
|
124 |
-
).result()
|
125 |
-
)
|
126 |
|
127 |
if not sentences:
|
128 |
logger.warning("No sentences available for summarization")
|
@@ -131,11 +107,12 @@ class QueryProcessor:
|
|
131 |
"key_sentences": []
|
132 |
}
|
133 |
|
134 |
-
#
|
135 |
embeddings = self.embedding_model.encode(sentences)
|
136 |
similarity_matrix = np.inner(embeddings, embeddings)
|
137 |
centrality_scores = degree_centrality_scores(similarity_matrix, threshold=None)
|
138 |
|
|
|
139 |
top_indices = np.argsort(-centrality_scores)[:10]
|
140 |
key_sentences = [sentences[idx].strip() for idx in top_indices]
|
141 |
|
|
|
4 |
from models.LexRank import degree_centrality_scores
|
5 |
import logging
|
6 |
from datetime import datetime as dt
|
|
|
|
|
7 |
|
8 |
logger = logging.getLogger(__name__)
|
9 |
|
|
|
13 |
self.summarization_model = summarization_model
|
14 |
self.nlp_model = nlp_model
|
15 |
self.db_service = db_service
|
|
|
16 |
logger.info("QueryProcessor initialized")
|
17 |
|
18 |
async def process(
|
|
|
23 |
end_date: Optional[str] = None
|
24 |
) -> Dict[str, Any]:
|
25 |
try:
|
26 |
+
# Date handling
|
27 |
start_dt = self._parse_date(start_date) if start_date else None
|
28 |
end_dt = self._parse_date(end_date) if end_date else None
|
29 |
|
30 |
+
# Query processing
|
31 |
+
query_embedding = self.embedding_model.encode(query).tolist()
|
32 |
logger.debug(f"Generated embedding for query: {query[:50]}...")
|
33 |
|
34 |
+
# Entity extraction
|
35 |
+
entities = self.nlp_model.extract_entities(query)
|
36 |
logger.debug(f"Extracted entities: {entities}")
|
37 |
|
38 |
+
# Database search
|
39 |
articles = await self._execute_semantic_search(
|
40 |
query_embedding,
|
41 |
start_dt,
|
42 |
end_dt,
|
43 |
topic,
|
44 |
+
[ent[0] for ent in entities] # Just the entity texts
|
45 |
)
|
46 |
|
47 |
if not articles:
|
48 |
logger.info("No articles found matching criteria")
|
49 |
return {"message": "No articles found", "articles": []}
|
50 |
|
51 |
+
# Summary generation
|
52 |
+
summary_data = self._generate_summary(articles)
|
53 |
|
54 |
return {
|
55 |
"summary": summary_data["summary"],
|
|
|
91 |
logger.error(f"Semantic search failed: {str(e)}")
|
92 |
raise
|
93 |
|
94 |
+
def _generate_summary(self, articles: List[Dict[str, Any]]) -> Dict[str, Any]:
|
95 |
+
"""Generate summary from articles with fallback handling"""
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
96 |
try:
|
97 |
# Extract and process content
|
98 |
sentences = []
|
99 |
for article in articles:
|
100 |
if content := article.get("content"):
|
101 |
+
sentences.extend(self.nlp_model.tokenize_sentences(content))
|
|
|
|
|
|
|
|
|
|
|
102 |
|
103 |
if not sentences:
|
104 |
logger.warning("No sentences available for summarization")
|
|
|
107 |
"key_sentences": []
|
108 |
}
|
109 |
|
110 |
+
# Generate summary
|
111 |
embeddings = self.embedding_model.encode(sentences)
|
112 |
similarity_matrix = np.inner(embeddings, embeddings)
|
113 |
centrality_scores = degree_centrality_scores(similarity_matrix, threshold=None)
|
114 |
|
115 |
+
# Get top 10 most central sentences
|
116 |
top_indices = np.argsort(-centrality_scores)[:10]
|
117 |
key_sentences = [sentences[idx].strip() for idx in top_indices]
|
118 |
|