Merge branch library_api_tag_ci into master

Title: api tagging for pipeline/train/evaluate 
        Link: https://code.alibaba-inc.com/Ali-MaaS/MaaS-lib/codereview/10588387

modelscope/hub/utils/utils.py

@@ -5,6 +5,8 @@ import os
 from datetime import datetime
 from typing import Optional
 
+import requests
+
 from modelscope.hub.constants import (DEFAULT_MODELSCOPE_DOMAIN,
                                       DEFAULT_MODELSCOPE_GROUP,
                                       MODEL_ID_SEPARATOR, MODELSCOPE_SDK_DEBUG,
@@ -85,3 +87,16 @@ def file_integrity_validation(file_path, expected_sha256):
         msg = 'File %s integrity check failed, the download may be incomplete, please try again.' % file_path
         logger.error(msg)
         raise FileIntegrityError(msg)
+
+
+def create_library_statistics(method: str, name: str, cn_name: Optional[str]):
+    try:
+        from modelscope.hub.api import ModelScopeConfig
+        path = f'{get_endpoint()}/api/v1/statistics/library'
+        headers = {'user-agent': ModelScopeConfig.get_user_agent()}
+        params = {'Method': method, 'Name': name, 'CnName': cn_name}
+        r = requests.post(path, params=params, headers=headers)
+        r.raise_for_status()
+    except Exception:
+        pass
+    return

modelscope/models/base/base_model.py

@@ -131,6 +131,8 @@ class Model(ABC):
 
         if not hasattr(model, 'cfg'):
             model.cfg = cfg
+
+        model.name = model_name_or_path
         return model
 
     def save_pretrained(self,

modelscope/pipelines/base.py

@@ -10,6 +10,7 @@ from typing import Any, Dict, Generator, List, Mapping, Union
 
 import numpy as np
 
+from modelscope.hub.utils.utils import create_library_statistics
 from modelscope.models.base import Model
 from modelscope.msdatasets import MsDataset
 from modelscope.outputs import TASK_OUTPUTS
@@ -151,7 +152,9 @@ class Pipeline(ABC):
                  **kwargs) -> Union[Dict[str, Any], Generator]:
         # model provider should leave it as it is
         # modelscope library developer will handle this function
-
+        for single_model in self.models:
+            if hasattr(single_model, 'name'):
+                create_library_statistics('pipeline', single_model.name, None)
         # place model to cpu or gpu
         if (self.model or (self.has_multiple_models and self.models[0])):
             if not self._model_prepare:

modelscope/trainers/trainer.py

@@ -15,6 +15,7 @@ from torch.utils.data.dataloader import default_collate
 from torch.utils.data.distributed import DistributedSampler
 
 from modelscope.hub.snapshot_download import snapshot_download
+from modelscope.hub.utils.utils import create_library_statistics
 from modelscope.metainfo import Trainers
 from modelscope.metrics import build_metric, task_default_metrics
 from modelscope.models.base import Model, TorchModel
@@ -436,6 +437,8 @@ class EpochBasedTrainer(BaseTrainer):
 
     def train(self, checkpoint_path=None, *args, **kwargs):
         self._mode = ModeKeys.TRAIN
+        if hasattr(self.model, 'name'):
+            create_library_statistics('train', self.model.name, None)
 
         if self.train_dataset is None:
             self.train_dataloader = self.get_train_dataloader()
@@ -456,6 +459,8 @@ class EpochBasedTrainer(BaseTrainer):
         self.train_loop(self.train_dataloader)
 
     def evaluate(self, checkpoint_path=None):
+        if hasattr(self.model, 'name'):
+            create_library_statistics('evaluate', self.model.name, None)
         if checkpoint_path is not None and os.path.isfile(checkpoint_path):
             from modelscope.trainers.hooks import CheckpointHook
             CheckpointHook.load_checkpoint(checkpoint_path, self)