zlei6/ragflow
1
0
Fork 0
mirror of https://github.com/infiniflow/ragflow.git synced 2026-08-06 07:34:04 +08:00
Code Issues Packages Projects Releases Wiki Activity

refa: simplify RAPTOR tree clustering and configuration (#17614)

Browse Source
This commit is contained in:
buua436
2026-08-03 17:46:50 +08:00
committed by GitHub
parent e290a0d476
commit 3e7cfbe052
32 changed files with 318 additions and 1356 deletions
Download Patch File Download Diff File Expand all files Collapse all files

56
test/testcases/restful_api/test_datasets.py
View File

@@ -272,20 +272,16 @@ def test_dataset_update_chunk_method_contract(rest_client, clear_datasets, chunk
("raptor_true", {"raptor": {"use_raptor": True}}),
("raptor_false", {"raptor": {"use_raptor": False}}),
("raptor_prompt", {"raptor": {"prompt": "Who are you?"}}),
("raptor_max_token_min", {"raptor": {"max_token": 1}}),
("raptor_max_token_min", {"raptor": {"max_token": 512}}),
("raptor_max_token_mid", {"raptor": {"max_token": 1024}}),
("raptor_max_token_max", {"raptor": {"max_token": 2048}}),
("raptor_threshold_min", {"raptor": {"threshold": 0.0}}),
("raptor_threshold_mid", {"raptor": {"threshold": 0.5}}),
("raptor_threshold_max", {"raptor": {"threshold": 1.0}}),
("raptor_clustering_threshold_min", {"raptor": {"clustering_threshold": 0.0}}),
("raptor_clustering_threshold_mid", {"raptor": {"clustering_threshold": 0.5}}),
("raptor_clustering_threshold_max", {"raptor": {"clustering_threshold": 1.0}}),
("raptor_max_cluster_min", {"raptor": {"max_cluster": 1}}),
("raptor_max_cluster_mid", {"raptor": {"max_cluster": 512}}),
("raptor_max_cluster_max", {"raptor": {"max_cluster": 1024}}),
("raptor_random_seed_min", {"raptor": {"random_seed": 0}}),
("raptor_clustering_method_gmm", {"raptor": {"clustering_method": "gmm"}}),
("raptor_clustering_method_ahc", {"raptor": {"clustering_method": "ahc"}}),
("raptor_tree_builder_raptor", {"raptor": {"tree_builder": "raptor"}}),
("raptor_tree_builder_psi", {"raptor": {"tree_builder": "psi"}}),
],
ids=[
"auto_keywords_min",
@@ -329,17 +325,13 @@ def test_dataset_update_chunk_method_contract(rest_client, clear_datasets, chunk
"raptor_max_token_min",
"raptor_max_token_mid",
"raptor_max_token_max",
"raptor_threshold_min",
"raptor_threshold_mid",
"raptor_threshold_max",
"raptor_clustering_threshold_min",
"raptor_clustering_threshold_mid",
"raptor_clustering_threshold_max",
"raptor_max_cluster_min",
"raptor_max_cluster_mid",
"raptor_max_cluster_max",
"raptor_random_seed_min",
"raptor_clustering_method_gmm",
"raptor_clustering_method_ahc",
"raptor_tree_builder_raptor",
"raptor_tree_builder_psi",
],
)
def test_dataset_update_parser_config_valid_matrix_contract(rest_client, clear_datasets, name, parser_config):
@@ -1028,13 +1020,13 @@ def test_dataset_update_parser_config_invalid_contract(rest_client, clear_datase
({"raptor": {"use_raptor": "string"}}, "Input should be a valid boolean"),
({"raptor": {"prompt": ""}}, "String should have at least 1 character"),
({"raptor": {"prompt": " "}}, "String should have at least 1 character"),
({"raptor": {"max_token": 0}}, "Input should be greater than or equal to 1"),
({"raptor": {"max_token": 0}}, "Input should be greater than or equal to 512"),
({"raptor": {"max_token": 2049}}, "Input should be less than or equal to 2048"),
({"raptor": {"max_token": 3.14}}, "Input should be a valid integer"),
({"raptor": {"max_token": "string"}}, "Input should be a valid integer"),
({"raptor": {"threshold": -0.1}}, "Input should be greater than or equal to 0"),
({"raptor": {"threshold": 1.1}}, "Input should be less than or equal to 1"),
({"raptor": {"threshold": "string"}}, "Input should be a valid number"),
({"raptor": {"clustering_threshold": -0.1}}, "Input should be greater than or equal to 0"),
({"raptor": {"clustering_threshold": 1.1}}, "Input should be less than or equal to 1"),
({"raptor": {"clustering_threshold": "string"}}, "Input should be a valid number"),
({"raptor": {"max_cluster": 0}}, "Input should be greater than or equal to 1"),
({"raptor": {"max_cluster": 1025}}, "Input should be less than or equal to 1024"),
({"raptor": {"max_cluster": 3.14}}, "Input should be a valid integer"),
@@ -1042,10 +1034,6 @@ def test_dataset_update_parser_config_invalid_contract(rest_client, clear_datase
({"raptor": {"random_seed": -1}}, "Input should be greater than or equal to 0"),
({"raptor": {"random_seed": 3.14}}, "Input should be a valid integer"),
({"raptor": {"random_seed": "string"}}, "Input should be a valid integer"),
({"raptor": {"clustering_method": "unknown"}}, "Input should be 'gmm' or 'ahc'"),
({"raptor": {"clustering_method": None}}, "Input should be 'gmm' or 'ahc'"),
({"raptor": {"tree_builder": "ahc"}}, "Input should be 'raptor' or 'psi'"),
({"raptor": {"tree_builder": None}}, "Input should be 'raptor' or 'psi'"),
({"delimiter": "a" * 65536}, "Parser config exceeds size limit (max 65,535 characters)"),
]
for parser_config, expected_message in invalid_cases:
@@ -1375,12 +1363,12 @@ def test_dataset_create_concurrent_contract(rest_client, clear_datasets):
("raptor_true", {"raptor": {"use_raptor": True}}),
("raptor_false", {"raptor": {"use_raptor": False}}),
("raptor_prompt", {"raptor": {"prompt": "Who are you?"}}),
("raptor_max_token_min", {"raptor": {"max_token": 1}}),
("raptor_max_token_min", {"raptor": {"max_token": 512}}),
("raptor_max_token_mid", {"raptor": {"max_token": 1024}}),
("raptor_max_token_max", {"raptor": {"max_token": 2048}}),
("raptor_threshold_min", {"raptor": {"threshold": 0.0}}),
("raptor_threshold_mid", {"raptor": {"threshold": 0.5}}),
("raptor_threshold_max", {"raptor": {"threshold": 1.0}}),
("raptor_clustering_threshold_min", {"raptor": {"clustering_threshold": 0.0}}),
("raptor_clustering_threshold_mid", {"raptor": {"clustering_threshold": 0.5}}),
("raptor_clustering_threshold_max", {"raptor": {"clustering_threshold": 1.0}}),
("raptor_max_cluster_min", {"raptor": {"max_cluster": 1}}),
("raptor_max_cluster_mid", {"raptor": {"max_cluster": 512}}),
("raptor_max_cluster_max", {"raptor": {"max_cluster": 1024}}),
@@ -1432,9 +1420,9 @@ def test_dataset_create_concurrent_contract(rest_client, clear_datasets):
"raptor_max_token_min",
"raptor_max_token_mid",
"raptor_max_token_max",
"raptor_threshold_min",
"raptor_threshold_mid",
"raptor_threshold_max",
"raptor_clustering_threshold_min",
"raptor_clustering_threshold_mid",
"raptor_clustering_threshold_max",
"raptor_max_cluster_min",
"raptor_max_cluster_mid",
"raptor_max_cluster_max",
@@ -1757,13 +1745,13 @@ def test_dataset_create_parser_config_invalid_contract(rest_client, clear_datase
("raptor_type_invalid", {"raptor": {"use_raptor": "string"}}, "Input should be a valid boolean"),
("raptor_prompt_empty", {"raptor": {"prompt": ""}}, "String should have at least 1 character"),
("raptor_prompt_space", {"raptor": {"prompt": " "}}, "String should have at least 1 character"),
("raptor_max_token_min_limit", {"raptor": {"max_token": 0}}, "Input should be greater than or equal to 1"),
("raptor_max_token_min_limit", {"raptor": {"max_token": 0}}, "Input should be greater than or equal to 512"),
("raptor_max_token_max_limit", {"raptor": {"max_token": 2049}}, "Input should be less than or equal to 2048"),
("raptor_max_token_float_not_allowed", {"raptor": {"max_token": 3.14}}, "Input should be a valid integer"),
("raptor_max_token_type_invalid", {"raptor": {"max_token": "string"}}, "Input should be a valid integer"),
("raptor_threshold_min_limit", {"raptor": {"threshold": -0.1}}, "Input should be greater than or equal to 0"),
("raptor_threshold_max_limit", {"raptor": {"threshold": 1.1}}, "Input should be less than or equal to 1"),
("raptor_threshold_type_invalid", {"raptor": {"threshold": "string"}}, "Input should be a valid number"),
("raptor_clustering_threshold_min_limit", {"raptor": {"clustering_threshold": -0.1}}, "Input should be greater than or equal to 0"),
("raptor_clustering_threshold_max_limit", {"raptor": {"clustering_threshold": 1.1}}, "Input should be less than or equal to 1"),
("raptor_clustering_threshold_type_invalid", {"raptor": {"clustering_threshold": "string"}}, "Input should be a valid number"),
("raptor_max_cluster_min_limit", {"raptor": {"max_cluster": 0}}, "Input should be greater than or equal to 1"),
("raptor_max_cluster_max_limit", {"raptor": {"max_cluster": 1025}}, "Input should be less than or equal to 1024"),
("raptor_max_cluster_float_not_allowed", {"raptor": {"max_cluster": 3.14}}, "Input should be a valid integer"),