Refactoring

Code/generate_claimed_results.py

@@ -1,39 +1,43 @@
 import logging
+import argparse
 import pathlib
-import configparser
+import os
+
+from threading import active_count
+from multiprocessing import Pool
+from multiprocessing.pool import ThreadPool
+from random import shuffle
 from tabulate import tabulate
 from pathlib import Path
-
-from Code.S_run_aifeynman import run_aifeynman
+from functools import partial
 
 
+from S_run_aifeynman import run_aifeynman
+
+_CFG = {
+  "dataset_path"      : "../Feynman_without_units/",
+  "operations_file"   : "./14ops.txt",
+  "polynomial_degree" : 3,
+  "number_of_epochs"  : 500,
+  "bruteforce_time"   : 60,
+  "test_percentage"   : 0,
+}
 
 class RunAll:
     """
-    Run the solver on all the whole dataset
+    Run the solver on the whole dataset
     """
 
-    def __init__(self, *, cfg_path: Path):
-        logging.basicConfig(filename="output.log", level=logging.DEBUG)
-        self.config = configparser.ConfigParser()
-        self.config.read(cfg_path)
-        self.cfg = self.config["Default"]
-        self.print_results()
+    def __init__(self, *,  cfg=_CFG):
+        logging.basicConfig(filename="output_no_units_parallel.log", level=logging.DEBUG)
+        self.cfg = cfg
         self.results = {}
 
  
-        self.run_solver()
-
-    def log_results(self):
-        pass
-
     def print_results(self):
-        table = [
-            ["foo", 696000, 1989100000],
-            ["bar", 6371, 5973.6],
-            ["baz", 1737, 73.5],
-            ["qux", 3390, 641.85],
-        ]
+        table = []
+        for file, sol in self.results.items():
+            table.append(sol[-1])
         print(tabulate(
                 table,
                 headers=[
@@ -45,12 +49,23 @@ class RunAll:
             )
         )
     
-    def run_solver(self):
+    def run_solver(self, dirs=None):
+        if not dirs:
             path = Path(self.cfg["dataset_path"])
-        for child in path.iterdir():
+            dirs = list(path.iterdir())
+            shuffle(dirs) # Shuffle to sample a different file each time
+            
+        else:
+            path=Path(self.cfg["dataset_path"])
+            child = dirs
+            
+            
+#         for child in dirs:
+#         print(child)
+        print(f"Process PID: {os.getpid()} ---------------- Number of threads: {active_count()}" )
         self.results[str(child).split("/")[-1]] = run_aifeynman(
-                pathdir="/home/aziz/lambda_lab/AI-Feynman/example_data/",#str(path.resolve()) + "/",
-                filename="example2.txt",#str(child).split("/")[-1],
+            pathdir=str(path.resolve()) + "/",
+            filename=str(child).split("/")[-1],
             BF_try_time=int(self.cfg["bruteforce_time"]),
             BF_ops_file_type=Path(self.cfg["operations_file"]),
             polyfit_deg=int(self.cfg["polynomial_degree"]),
@@ -58,12 +73,43 @@ class RunAll:
             vars_name=[],
             test_percentage=int(self.cfg["test_percentage"]),
         )
+
         logging.info(self.results)
-            break
+        print("@"*120)
+        print("@"*120)
+
+        self.print_results()
+
+
+def get_files(dirs, chunks=5):
+    dirs = list(path.iterdir())
+    dirs = [file for file in dirs if not (str(file).endswith("test") or str(file).endswith("train"))]
+    for i in range(0, len(dirs), chunks):
+        yield dirs[i : i + chunks]
 
 if __name__ == "__main__":
-    cfg_path = pathlib.Path("/home/aziz/lambda_lab/AI-Feynman/configs.cfg")
-    if cfg_path.exists():
-        RunAll(cfg_path=cfg_path)
-    else:
-        print(f"No such a file {cfg_path}")
+
+    #cfg_path = pathlib.Path("/home/aziz/lambda_lab/AI-Feynman/configs.cfg")
+    #if cfg_path.exists():
+    #    RunAll(cfg_path=cfg_path)
+    #else:
+    #    print(f"No such a file {cfg_path}")
+
+    solver = RunAll().run_solver
+    path = Path(_CFG["dataset_path"])
+    #dirs = list(path.iterdir())
+    #chunked_dirs = list(get_files(dirs, chunks=24))
+# print(chunked_dirs[0], len(chunked_dirs[0]))
+#    for dd in chunked_dirs:
+ #       pool = Pool(len(dd))
+  #      print(dd, len(dd))
+    #    pool.map(print, dd)
+   #     pool.map(solver, dd)
+   # pool.close()
+   
+    parser = argparse.ArgumentParser(description='Solver')
+    parser.add_argument('--file', help='Enter file path')
+
+    args = parser.parse_args()
+    solver(args.file)
+