From eab19de080e2b2b1de93cdce7704c6535f2b2ced Mon Sep 17 00:00:00 2001 From: Jactus Date: Tue, 27 Apr 2021 16:56:07 +0800 Subject: [PATCH 1/7] Support start exp with given exp & recorder id --- qlib/workflow/__init__.py | 18 +++++++++++++++--- qlib/workflow/exp.py | 8 +++++--- qlib/workflow/expm.py | 12 ++++++++++-- 3 files changed, 30 insertions(+), 8 deletions(-) diff --git a/qlib/workflow/__init__.py b/qlib/workflow/__init__.py index a03665626..7cb1cf5cb 100644 --- a/qlib/workflow/__init__.py +++ b/qlib/workflow/__init__.py @@ -23,7 +23,9 @@ class QlibRecorder: @contextmanager def start( self, + experiment_id: Optional[Text] = None, experiment_name: Optional[Text] = None, + recorder_id: Optional[Text] = None, recorder_name: Optional[Text] = None, uri: Optional[Text] = None, resume: bool = False, @@ -45,8 +47,12 @@ class QlibRecorder: Parameters ---------- + experiment_id : str + id of the experiment one wants to start. experiment_name : str name of the experiment one wants to start. + recorder_id : str + id of the recorder under the experiment one wants to start. recorder_name : str name of the recorder under the experiment one wants to start. uri : str @@ -57,7 +63,7 @@ class QlibRecorder: resume : bool whether to resume the specific recorder with given name under the given experiment. """ - run = self.start_exp(experiment_name, recorder_name, uri, resume) + run = self.start_exp(experiment_id, experiment_name, recorder_id, recorder_name, uri, resume) try: yield run except Exception as e: @@ -65,7 +71,9 @@ class QlibRecorder: raise e self.end_exp(Recorder.STATUS_FI) - def start_exp(self, experiment_name=None, recorder_name=None, uri=None, resume=False): + def start_exp( + self, experiment_id=None, experiment_name=None, recorder_id=None, recorder_name=None, uri=None, resume=False + ): """ Lower level method for starting an experiment. When use this method, one should end the experiment manually and the status of the recorder may not be handled properly. Here is the example code: @@ -79,8 +87,12 @@ class QlibRecorder: Parameters ---------- + experiment_id : str + id of the experiment one wants to start. experiment_name : str the name of the experiment to be started + recorder_id : str + id of the recorder under the experiment one wants to start. recorder_name : str name of the recorder under the experiment one wants to start. uri : str @@ -93,7 +105,7 @@ class QlibRecorder: ------- An experiment instance being started. """ - return self.exp_manager.start_exp(experiment_name, recorder_name, uri, resume) + return self.exp_manager.start_exp(experiment_id, experiment_name, recorder_id, recorder_name, uri, resume) def end_exp(self, recorder_status=Recorder.STATUS_FI): """ diff --git a/qlib/workflow/exp.py b/qlib/workflow/exp.py index 7b3d1f507..0a7e0a5a9 100644 --- a/qlib/workflow/exp.py +++ b/qlib/workflow/exp.py @@ -39,12 +39,14 @@ class Experiment: output["recorders"] = list(recorders.keys()) return output - def start(self, recorder_name=None, resume=False): + def start(self, recorder_id=None, recorder_name=None, resume=False): """ Start the experiment and set it to be active. This method will also start a new recorder. Parameters ---------- + recorder_id : str + the id of the recorder to be created. recorder_name : str the name of the recorder to be created. resume : bool @@ -238,14 +240,14 @@ class MLflowExperiment(Experiment): def __repr__(self): return "{name}(id={id}, info={info})".format(name=self.__class__.__name__, id=self.id, info=self.info) - def start(self, recorder_name=None, resume=False): + def start(self, recorder_id=None, recorder_name=None, resume=False): logger.info(f"Experiment {self.id} starts running ...") # Get or create recorder if recorder_name is None: recorder_name = self._default_rec_name # resume the recorder if resume: - recorder, _ = self._get_or_create_rec(recorder_name=recorder_name) + recorder, _ = self._get_or_create_rec(recorder_id=recorder_id, recorder_name=recorder_name) # create a new recorder else: recorder = self.create_recorder(recorder_name) diff --git a/qlib/workflow/expm.py b/qlib/workflow/expm.py index 590790c9e..5549bb9bf 100644 --- a/qlib/workflow/expm.py +++ b/qlib/workflow/expm.py @@ -33,7 +33,9 @@ class ExpManager: def start_exp( self, + experiment_id: Optional[Text] = None, experiment_name: Optional[Text] = None, + recorder_id: Optional[Text] = None, recorder_name: Optional[Text] = None, uri: Optional[Text] = None, resume: bool = False, @@ -45,8 +47,12 @@ class ExpManager: Parameters ---------- + experiment_id : str + id of the active experiment. experiment_name : str name of the active experiment. + recorder_id : str + id of the recorder to be started. recorder_name : str name of the recorder to be started. uri : str @@ -298,7 +304,9 @@ class MLflowExpManager(ExpManager): def start_exp( self, + experiment_id: Optional[Text] = None, experiment_name: Optional[Text] = None, + recorder_id: Optional[Text] = None, recorder_name: Optional[Text] = None, uri: Optional[Text] = None, resume: bool = False, @@ -308,11 +316,11 @@ class MLflowExpManager(ExpManager): # Create experiment if experiment_name is None: experiment_name = self._default_exp_name - experiment, _ = self._get_or_create_exp(experiment_name=experiment_name) + experiment, _ = self._get_or_create_exp(experiment_id=experiment_id, experiment_name=experiment_name) # Set up active experiment self.active_experiment = experiment # Start the experiment - self.active_experiment.start(recorder_name, resume) + self.active_experiment.start(recorder_id, recorder_name, resume) return self.active_experiment From f58c61a2e0c313074729da6715d30d58e1503e69 Mon Sep 17 00:00:00 2001 From: Jactus Date: Thu, 29 Apr 2021 16:54:51 +0800 Subject: [PATCH 2/7] Fix logger pickling error --- qlib/log.py | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/qlib/log.py b/qlib/log.py index 5888b3841..1d604e0c0 100644 --- a/qlib/log.py +++ b/qlib/log.py @@ -17,6 +17,7 @@ class MetaLogger(type): wrapper_dict = logging.Logger.__dict__.copy() wrapper_dict.update(dict) wrapper_dict["__doc__"] = logging.Logger.__doc__ + del wrapper_dict["__reduce__"] # make Logger object can be pickled return type.__new__(cls, name, bases, wrapper_dict) @@ -29,6 +30,15 @@ class QlibLogger(metaclass=MetaLogger): self.module_name = module_name self.level = 0 + def __getstate__(self): + return vars(self) + + def __setstate__(self, state): + vars(self).update(state) + + def __reduce__(self): + return (QlibLogger, (self.module_name,)) + @property def logger(self): logger = logging.getLogger(self.module_name) From ca92cb980ca9a49d9c41f98e5f2c2c6941a8a1ae Mon Sep 17 00:00:00 2001 From: Jactus Date: Thu, 29 Apr 2021 22:40:52 +0800 Subject: [PATCH 3/7] Update meta logger --- qlib/log.py | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) diff --git a/qlib/log.py b/qlib/log.py index 1d604e0c0..19331f5d5 100644 --- a/qlib/log.py +++ b/qlib/log.py @@ -15,10 +15,11 @@ from .config import C class MetaLogger(type): def __new__(cls, name, bases, dict): wrapper_dict = logging.Logger.__dict__.copy() - wrapper_dict.update(dict) - wrapper_dict["__doc__"] = logging.Logger.__doc__ - del wrapper_dict["__reduce__"] # make Logger object can be pickled - return type.__new__(cls, name, bases, wrapper_dict) + for key in wrapper_dict: + if key not in dict and key != "__reduce__": + dict[key] = wrapper_dict[key] + dict["__doc__"] = logging.Logger.__doc__ + return type.__new__(cls, name, bases, dict) class QlibLogger(metaclass=MetaLogger): From 51b649ec395f4a80e96dd88b51ebdd8d2a192db2 Mon Sep 17 00:00:00 2001 From: Jactus Date: Fri, 30 Apr 2021 13:13:05 +0800 Subject: [PATCH 4/7] Update QlibLogger --- qlib/log.py | 9 +++------ 1 file changed, 3 insertions(+), 6 deletions(-) diff --git a/qlib/log.py b/qlib/log.py index 19331f5d5..d095d571a 100644 --- a/qlib/log.py +++ b/qlib/log.py @@ -31,12 +31,6 @@ class QlibLogger(metaclass=MetaLogger): self.module_name = module_name self.level = 0 - def __getstate__(self): - return vars(self) - - def __setstate__(self, state): - vars(self).update(state) - def __reduce__(self): return (QlibLogger, (self.module_name,)) @@ -50,6 +44,9 @@ class QlibLogger(metaclass=MetaLogger): self.level = level def __getattr__(self, name): + # During unpickling, python will call __getattr__. Use this line to avoid maximum recursion error. + if name in {"__setstate__"}: + raise AttributeError return self.logger.__getattribute__(name) From 694ae3402766e582a6c067de807a997f1a9719c4 Mon Sep 17 00:00:00 2001 From: Jactus Date: Fri, 30 Apr 2021 13:27:19 +0800 Subject: [PATCH 5/7] Update api --- qlib/workflow/__init__.py | 21 ++++++++++++++++++--- qlib/workflow/exp.py | 4 ++-- qlib/workflow/expm.py | 4 +++- 3 files changed, 23 insertions(+), 6 deletions(-) diff --git a/qlib/workflow/__init__.py b/qlib/workflow/__init__.py index 7cb1cf5cb..8135bab60 100644 --- a/qlib/workflow/__init__.py +++ b/qlib/workflow/__init__.py @@ -23,6 +23,7 @@ class QlibRecorder: @contextmanager def start( self, + *, experiment_id: Optional[Text] = None, experiment_name: Optional[Text] = None, recorder_id: Optional[Text] = None, @@ -63,7 +64,14 @@ class QlibRecorder: resume : bool whether to resume the specific recorder with given name under the given experiment. """ - run = self.start_exp(experiment_id, experiment_name, recorder_id, recorder_name, uri, resume) + run = self.start_exp( + experiment_id=experiment_id, + experiment_name=experiment_name, + recorder_id=recorder_id, + recorder_name=recorder_name, + uri=uri, + resume=resume, + ) try: yield run except Exception as e: @@ -72,7 +80,7 @@ class QlibRecorder: self.end_exp(Recorder.STATUS_FI) def start_exp( - self, experiment_id=None, experiment_name=None, recorder_id=None, recorder_name=None, uri=None, resume=False + self, *, experiment_id=None, experiment_name=None, recorder_id=None, recorder_name=None, uri=None, resume=False ): """ Lower level method for starting an experiment. When use this method, one should end the experiment manually @@ -105,7 +113,14 @@ class QlibRecorder: ------- An experiment instance being started. """ - return self.exp_manager.start_exp(experiment_id, experiment_name, recorder_id, recorder_name, uri, resume) + return self.exp_manager.start_exp( + experiment_id=experiment_id, + experiment_name=experiment_name, + recorder_id=recorder_id, + recorder_name=recorder_name, + uri=uri, + resume=resume, + ) def end_exp(self, recorder_status=Recorder.STATUS_FI): """ diff --git a/qlib/workflow/exp.py b/qlib/workflow/exp.py index 0a7e0a5a9..467c7c3f4 100644 --- a/qlib/workflow/exp.py +++ b/qlib/workflow/exp.py @@ -39,7 +39,7 @@ class Experiment: output["recorders"] = list(recorders.keys()) return output - def start(self, recorder_id=None, recorder_name=None, resume=False): + def start(self, *, recorder_id=None, recorder_name=None, resume=False): """ Start the experiment and set it to be active. This method will also start a new recorder. @@ -240,7 +240,7 @@ class MLflowExperiment(Experiment): def __repr__(self): return "{name}(id={id}, info={info})".format(name=self.__class__.__name__, id=self.id, info=self.info) - def start(self, recorder_id=None, recorder_name=None, resume=False): + def start(self, *, recorder_id=None, recorder_name=None, resume=False): logger.info(f"Experiment {self.id} starts running ...") # Get or create recorder if recorder_name is None: diff --git a/qlib/workflow/expm.py b/qlib/workflow/expm.py index 5549bb9bf..04cc3bcb7 100644 --- a/qlib/workflow/expm.py +++ b/qlib/workflow/expm.py @@ -33,6 +33,7 @@ class ExpManager: def start_exp( self, + *, experiment_id: Optional[Text] = None, experiment_name: Optional[Text] = None, recorder_id: Optional[Text] = None, @@ -304,6 +305,7 @@ class MLflowExpManager(ExpManager): def start_exp( self, + *, experiment_id: Optional[Text] = None, experiment_name: Optional[Text] = None, recorder_id: Optional[Text] = None, @@ -320,7 +322,7 @@ class MLflowExpManager(ExpManager): # Set up active experiment self.active_experiment = experiment # Start the experiment - self.active_experiment.start(recorder_id, recorder_name, resume) + self.active_experiment.start(recorder_id=recorder_id, recorder_name=recorder_name, resume=resume) return self.active_experiment From 5eb9dfff166b79cdd2e00bc0ff7430f266db46b0 Mon Sep 17 00:00:00 2001 From: Jactus Date: Fri, 30 Apr 2021 15:28:37 +0800 Subject: [PATCH 6/7] Remove redundant --- qlib/log.py | 4 ---- 1 file changed, 4 deletions(-) diff --git a/qlib/log.py b/qlib/log.py index d095d571a..e714bc15a 100644 --- a/qlib/log.py +++ b/qlib/log.py @@ -18,7 +18,6 @@ class MetaLogger(type): for key in wrapper_dict: if key not in dict and key != "__reduce__": dict[key] = wrapper_dict[key] - dict["__doc__"] = logging.Logger.__doc__ return type.__new__(cls, name, bases, dict) @@ -31,9 +30,6 @@ class QlibLogger(metaclass=MetaLogger): self.module_name = module_name self.level = 0 - def __reduce__(self): - return (QlibLogger, (self.module_name,)) - @property def logger(self): logger = logging.getLogger(self.module_name) From 5bc2b96346605404faa571e76ee7c37755514b0c Mon Sep 17 00:00:00 2001 From: you-n-g Date: Mon, 3 May 2021 12:34:08 +0800 Subject: [PATCH 7/7] Update data.rst --- docs/component/data.rst | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/docs/component/data.rst b/docs/component/data.rst index 26f44a076..3cee803e6 100644 --- a/docs/component/data.rst +++ b/docs/component/data.rst @@ -182,6 +182,11 @@ The `trade unit` defines the unit number of stocks can be used in a trade, and t qlib.init(provider_uri='~/.qlib/qlib_data/us_data', region=REG_US) +.. note:: + + PRs for new data source are highly welcome! Users could commit the code to crawl data as a PR like `the examples here `_. And then we will use the code to create data cache on our server which other users could use directly. + + Data API ========================