logger added, bugs to fix

conf.yaml

@@ -99,4 +99,4 @@ main_conf: &main_conf
       - atc_ne
       - doc_feat
 
-exec: *main_conf
+exec: *debug_conf

quacc.log

@@ -0,0 +1,26 @@
+dataset rcv1_CCAT
+28/10/23 00:45:46| INFO: dataset rcv1_CCAT
+Dataset sample 0.50 of dataset rcv1_CCAT started
+Dataset sample 0.50 of dataset rcv1_CCAT failed. Exception: Queue objects should only be shared between processes through inheritance
+28/10/23 00:45:50| INFO: Dataset sample 0.50 of dataset rcv1_CCAT started
+Dataset sample 0.50 of dataset rcv1_CCAT failed. Exception: Queue objects should only be shared between processes through inheritance
+28/10/23 00:45:51| ERROR: Dataset sample 0.50 of dataset rcv1_CCAT failed. Exception: Queue objects should only be shared between processes through inheritance
+Dataset sample 0.50 of dataset rcv1_CCAT failed. Exception: Queue objects should only be shared between processes through inheritance
+28/10/23 00:45:52| ERROR: Dataset sample 0.50 of dataset rcv1_CCAT failed. Exception: Queue objects should only be shared between processes through inheritance
+Dataset sample 0.50 of dataset rcv1_CCAT failed. Exception: Queue objects should only be shared between processes through inheritance
+28/10/23 00:45:52| ERROR: Dataset sample 0.50 of dataset rcv1_CCAT failed. Exception: Queue objects should only be shared between processes through inheritance
+Dataset sample 0.50 of dataset rcv1_CCAT finished [took 1.8041s
+28/10/23 00:45:52| ERROR: Dataset sample 0.50 of dataset rcv1_CCAT failed. Exception: Queue objects should only be shared between processes through inheritance
+28/10/23 00:45:52| INFO: Dataset sample 0.50 of dataset rcv1_CCAT finished [took 1.8041s
+Configuration rcv1_CCAT_1prevs failed. Exception: too many indices for array: array is 1-dimensional, but 2 were indexed
+28/10/23 00:45:52| ERROR: Configuration rcv1_CCAT_1prevs failed. Exception: too many indices for array: array is 1-dimensional, but 2 were indexed
+dataset rcv1_CCAT
+28/10/23 00:47:52| INFO: dataset rcv1_CCAT
+Dataset sample 0.50 of dataset rcv1_CCAT started
+Dataset sample 0.50 of dataset rcv1_CCAT failed. Exception: Queue objects should only be shared between processes through inheritance
+28/10/23 00:47:56| INFO: Dataset sample 0.50 of dataset rcv1_CCAT started
+Dataset sample 0.50 of dataset rcv1_CCAT failed. Exception: Queue objects should only be shared between processes through inheritance
+28/10/23 00:47:57| ERROR: Dataset sample 0.50 of dataset rcv1_CCAT failed. Exception: Queue objects should only be shared between processes through inheritance
+Dataset sample 0.50 of dataset rcv1_CCAT failed. Exception: Queue objects should only be shared between processes through inheritance
+Dataset sample 0.50 of dataset rcv1_CCAT failed. Exception: Queue objects should only be shared between processes through inheritance
+Dataset sample 0.50 of dataset rcv1_CCAT finished [took 1.7186s

quacc/evaluation/comp.py

@@ -1,9 +1,7 @@
-import logging as log
 import multiprocessing
 import time
 from typing import List
 
-import numpy as np
 import pandas as pd
 import quapy as qp
 from quapy.protocol import APP
@@ -13,6 +11,7 @@ from quacc.dataset import Dataset
 from quacc.environment import env
 from quacc.evaluation import baseline, method
 from quacc.evaluation.report import CompReport, DatasetReport, EvaluationReport
+from quacc.logger import Logger, SubLogger
 
 qp.environ["SAMPLE_SIZE"] = env.SAMPLE_SIZE
 
@@ -41,8 +40,11 @@ class CompEstimator:
 CE = CompEstimator
 
 
-def fit_and_estimate(_estimate, train, validation, test, _env=None):
+def fit_and_estimate(_estimate, train, validation, test, _env=None, q=None):
     _env = env if _env is None else _env
+    SubLogger.setup(q)
+    log = SubLogger.logger()
+
     model = LogisticRegression()
 
     model.fit(*train.Xy)
@@ -76,16 +78,22 @@ def fit_and_estimate(_estimate, train, validation, test, _env=None):
 def evaluate_comparison(
     dataset: Dataset, estimators=["OUR_BIN_SLD", "OUR_MUL_SLD"]
 ) -> EvaluationReport:
+    log = Logger.logger()
     # with multiprocessing.Pool(1) as pool:
     with multiprocessing.Pool(len(estimators)) as pool:
         dr = DatasetReport(dataset.name)
         log.info(f"dataset {dataset.name}")
         for d in dataset():
-            log.info(f"train prev.: {np.around(d.train_prev, decimals=2)}")
+            log.info(
+                f"Dataset sample {d.train_prev[1]:.2f} of dataset {dataset.name} started"
+            )
             tstart = time.time()
             tasks = [(estim, d.train, d.validation, d.test) for estim in CE[estimators]]
             results = [
-                pool.apply_async(fit_and_estimate, t, {"_env": env}) for t in tasks
+                pool.apply_async(
+                    fit_and_estimate, t, {"_env": env, "q": Logger.queue()}
+                )
+                for t in tasks
             ]
 
             results_got = []
@@ -96,14 +104,14 @@ def evaluate_comparison(
                         results_got.append(r)
                 except Exception as e:
                     log.error(
-                        f"Dataset sample {d.train[1]:.2f} of dataset {dataset.name} failed. Exception: {e}"
+                        f"Dataset sample {d.train_prev[1]:.2f} of dataset {dataset.name} failed. Exception: {e}"
                     )
 
             tend = time.time()
             times = {r["name"]: r["time"] for r in results_got}
             times["tot"] = tend - tstart
             log.info(
-                f"Dataset sample {d.train[1]:.2f} of dataset {dataset.name} finished [took {times['tot']:.4f}s"
+                f"Dataset sample {d.train_prev[1]:.2f} of dataset {dataset.name} finished [took {times['tot']:.4f}s"
             )
             dr += CompReport(
                 [r["result"] for r in results_got],

quacc/logger.py

@@ -0,0 +1,107 @@
+import logging
+import logging.handlers
+import multiprocessing
+import threading
+
+
+class Logger:
+    __logger_file = "quacc.log"
+    __logger_name = "queue_logger"
+    __queue = None
+    __thread = None
+    __setup = False
+
+    @classmethod
+    def __logger_listener(cls, q):
+        while True:
+            record = q.get()
+            if record is None:
+                break
+            root = logging.getLogger()
+            root.handle(record)
+
+    @classmethod
+    def setup(cls):
+        if cls.__setup:
+            return
+
+        # setup root
+        root = logging.getLogger()
+        rh = logging.FileHandler(cls.__logger_file, mode="a")
+        root.addHandler(rh)
+
+        # setup logger
+        if cls.__queue is None:
+            cls.__queue = multiprocessing.Queue()
+
+        logger = logging.getLogger(cls.__logger_name)
+        logger.setLevel(logging.DEBUG)
+        qh = logging.handlers.QueueHandler(cls.__queue)
+        qh.setLevel(logging.DEBUG)
+        qh.setFormatter(
+            logging.Formatter(
+                fmt="%(asctime)s| %(levelname)s: %(message)s",
+                datefmt="%d/%m/%y %H:%M:%S",
+            )
+        )
+        logger.addHandler(qh)
+
+        # start listener
+        cls.__thread = threading.Thread(
+            target=cls.__logger_listener,
+            args=(cls.__queue,),
+        )
+        cls.__thread.start()
+
+        cls.__setup = True
+
+    @classmethod
+    def queue(cls):
+        if not cls.__setup:
+            cls.setup()
+
+        return cls.__queue
+
+    @classmethod
+    def logger(cls):
+        if not cls.__setup:
+            cls.setup()
+
+        return logging.getLogger(cls.__logger_name)
+
+    @classmethod
+    def join_listener(cls):
+        if cls.__setup and cls.__thread is not None:
+            cls.__thread.join()
+
+
+class SubLogger:
+    __queue = None
+    __setup = False
+
+    @classmethod
+    def setup(cls, q):
+        if cls.__setup:
+            return
+
+        cls.__queue = q
+
+        # setup root
+        root = logging.getLogger()
+        rh = logging.handlers.QueueHandler(q)
+        rh.setFormatter(
+            logging.Formatter(
+                fmt="%(asctime)s| %(levelname)s: %(message)s",
+                datefmt="%d/%m/%y %H:%M:%S",
+            )
+        )
+        root.addHandler(rh)
+
+        cls.__setup = True
+
+    @classmethod
+    def logger(cls):
+        if not cls.__setup:
+            return None
+
+        return logging.getLogger()

quacc/main.py

@@ -1,12 +1,14 @@
-import logging as log
 import traceback
 from sys import platform
 
 import quacc.evaluation.comp as comp
 from quacc.dataset import Dataset
 from quacc.environment import env
+from quacc.logger import Logger
 from quacc.utils import create_dataser_dir
 
+log = Logger.logger()
+
 
 def toast():
     if platform == "win32":
@@ -46,14 +48,9 @@ def estimate_comparison():
 
 
 def main():
-    log.basicConfig(
-        filename="quacc.log",
-        filemode="a",
-        format="%(asctime)s| %(levelname)s: %(message)s",
-        datefmt="%d/%m/%y %H:%M:%S",
-    )
     estimate_comparison()
     toast()
+    Logger.join_listener()
 
 
 if __name__ == "__main__":