Sync changes

app.py

@@ -31,26 +31,73 @@ def process(query_type, corpus_desc, engine_desc, query, maxnum, request: gr.Req
 
 def count(corpus_desc, engine_desc, query, request: gr.Request):
     result = process('count', corpus_desc, engine_desc, query, None, request)
-    return result.get('latency', ''), result.get('tokenized', ''), result.get('count', '')
+    latency = '' if 'latency' not in result else f'{result["latency"]:.3f}'
+    tokenized = '' if 'tokenized' not in result else result['tokenized']
+    if 'error' in result:
+        count = result['error']
+    else:
+        count = f'{result["count"]:,}'
+    return latency, tokenized, count
+
 def prob(corpus_desc, engine_desc, query, request: gr.Request):
     result = process('prob', corpus_desc, engine_desc, query, None, request)
-    return result.get('latency', ''), result.get('tokenized', ''), result.get('probability', '')
+    latency = '' if 'latency' not in result else f'{result["latency"]:.3f}'
+    tokenized = '' if 'tokenized' not in result else result['tokenized']
+    if 'error' in result:
+        prob = result['error']
+    else:
+        prob = f'{result["prob"]:.4f} ({result["cont_cnt"]:,} / {result["prompt_cnt"]:,})'
+    return latency, tokenized, prob
+
 def ntd(corpus_desc, engine_desc, query, request: gr.Request):
     result = process('ntd', corpus_desc, engine_desc, query, None, request)
-    return result.get('latency', ''), result.get('tokenized', ''), result.get('distribution', '')
+    latency = '' if 'latency' not in result else f'{result["latency"]:.3f}'
+    tokenized = '' if 'tokenized' not in result else result['tokenized']
+    if 'error' in result:
+        ntd = result['error']
+    else:
+        ntd = result['ntd']
+    return latency, tokenized, ntd
+
 def infgram_prob(corpus_desc, engine_desc, query, request: gr.Request):
     result = process('infgram_prob', corpus_desc, engine_desc, query, None, request)
-    return result.get('latency', ''), result.get('tokenized', ''), result.get('longest_suffix', ''), result.get('probability', '')
+    latency = '' if 'latency' not in result else f'{result["latency"]:.3f}'
+    tokenized = '' if 'tokenized' not in result else result['tokenized']
+    if 'error' in result:
+        longest_suffix = ''
+        prob = result['error']
+    else:
+        longest_suffix = result['longest_suffix']
+        prob = f'{result["prob"]:.4f} ({result["cont_cnt"]:,} / {result["prompt_cnt"]:,})'
+    return latency, tokenized, longest_suffix, prob
+
 def infgram_ntd(corpus_desc, engine_desc, query, request: gr.Request):
     result = process('infgram_ntd', corpus_desc, engine_desc, query, None, request)
-    return result.get('latency', ''), result.get('tokenized', ''), result.get('longest_suffix', ''), result.get('distribution', '')
+    latency = '' if 'latency' not in result else f'{result["latency"]:.3f}'
+    tokenized = '' if 'tokenized' not in result else result['tokenized']
+    if 'error' in result:
+        longest_suffix = ''
+        ntd = result['error']
+    else:
+        longest_suffix = result['longest_suffix']
+        ntd = result['ntd']
+    return latency, tokenized, longest_suffix, ntd
+
 def search_docs(corpus_desc, engine_desc, query, maxnum, request: gr.Request):
     result = process('search_docs', corpus_desc, engine_desc, query, maxnum, request)
-    outputs = result.get('outputs', [])
-    outputs = outputs[:maxnum]
-    while len(outputs) < 10:
-        outputs.append([])
-    return result.get('latency', ''), result.get('tokenized', ''), result.get('message', ''), outputs[0], outputs[1], outputs[2], outputs[3], outputs[4], outputs[5], outputs[6], outputs[7], outputs[8], outputs[9]
+    latency = '' if 'latency' not in result else f'{result["latency"]:.3f}'
+    tokenized = '' if 'tokenized' not in result else result['tokenized']
+    if 'error' in result:
+        message = result['error']
+        docs = [[] for _ in range(10)]
+    else:
+        message = result['message']
+        docs = result['docs']
+    docs = docs[:maxnum]
+    while len(docs) < 10:
+        docs.append([])
+    return latency, tokenized, message, docs[0], docs[1], docs[2], docs[3], docs[4], docs[5], docs[6], docs[7], docs[8], docs[9]
+
 def analyze_document(corpus_desc, engine_desc, query, request: gr.Request):
     result = process('analyze_document', corpus_desc, engine_desc, query, None, request)
     return result.get('latency', ''), result.get('html', '')

constants.py

@@ -18,7 +18,7 @@ ENGINES = list(ENGINE_BY_DESC.values())
 MAX_QUERY_CHARS = int(os.environ.get('MAX_QUERY_CHARS', 1000))
 MAX_INPUT_DOC_TOKENS = int(os.environ.get('MAX_INPUT_DOC_TOKENS', 1000))
 MAX_OUTPUT_DOC_TOKENS = int(os.environ.get('MAX_OUTPUT_DOC_TOKENS', 5000))
-MAX_OUTPUT_NUM_DOCS = int(os.environ.get('MAX_OUTPUT_NUM_DOCS', 10))
+MAX_OUTPUT_NUM_DOCS = int(os.environ.get('MAX_OUTPUT_NUM_DOCS', 10)) # This number is also hard-coded in app.py
 MAX_CNT_FOR_NTD = int(os.environ.get('MAX_CNT_FOR_NTD', 1000))
 MAX_CLAUSE_FREQ = int(os.environ.get('MAX_CLAUSE_FREQ', 10000))
 MAX_CLAUSE_FREQ_FAST = int(os.environ.get('MAX_CLAUSE_FREQ_FAST', 1000000))
@@ -34,7 +34,6 @@ DEFAULT_CONCURRENCY_LIMIT = os.environ.get('DEFAULT_CONCURRENCY_LIMIT', 10)
 MAX_SIZE = os.environ.get('MAX_SIZE', 100)
 MAX_THREADS = os.environ.get('MAX_THREADS', 40)
 DEBUG = (os.environ.get('DEBUG', 'False') != 'False')
-MIN_QUERY_INTERVAL_SECONDS = int(os.environ.get('MIN_QUERY_INTERVAL_SECONDS', 5))
 
 # C++ engine
 CPP_PORT = int(os.environ.get('CPP_PORT', 3786))