more steps towards an generator based data loading API

Author: thomasWeise

README.md

@@ -352,8 +352,8 @@ with temp_dir() as td:  # create temporary directory `td`
                            make_random_sampling],  # provide RS run creator
                    n_runs=5)  # we will execute 5 runs per setup
 
-    from_logs(  # parse all log files and print end results
-        td, lambda er: print(f"{er.algorithm} on {er.instance}: {er.best_f}"))
+    for er in from_logs(td):  # parse all log files and print end results
+        print(f"{er.algorithm} on {er.instance}: {er.best_f}")
 # The temp directory is deleted as soon as we leave the `with` block.
 ```
 
@@ -682,8 +682,8 @@ with temp_dir() as td:  # create temporary directory `td`
                    instances=problems,  # define the problem instances
                    setups=[make_execution],  # creator for our algorithm
                    n_runs=5)  # we will execute 5 runs per setup
-    from_logs(  # parse all log files and print end results
-        td, lambda er: print(f"{er.algorithm} on {er.instance}: {er.best_f}"))
+    for er in from_logs(td):  # parse all log files and print end results
+        print(f"{er.algorithm} on {er.instance}: {er.best_f}")
 # The temp directory is deleted as soon as we leave the `with` block.
 ```
 
@@ -1564,12 +1564,9 @@ with temp_dir() as td:
         max_fes=10000,  # we grant 10000 FEs per run
         n_runs=4)  # perform 4 runs per algorithm * instance combination
 
-    end_results = []  # this list will receive the end results records
-    from_logs(td, end_results.append)  # get results from log files
-
+    data = list(from_logs(td))  # load end results
     er_csv = to_csv(  # store end results to csv file (returns path)
-        end_results,  # the list of end results to store
-        td.resolve_inside("end_results.txt"))  # path to the file to generate
+        data, td.resolve_inside("end_results.txt"))  # path to output file
     print(er_csv.read_all_str())  # read generated file as string and print it
 # When leaving "while", the temp directory will be deleted
 ```
@@ -1700,8 +1697,7 @@ with temp_dir() as td:
         max_fes=10000,  # we grant 10000 FEs per run
         n_runs=4)  # perform 4 runs per algorithm * instance combination
 
-    end_results = []  # this list will receive the end results records
-    from_logs(td, end_results.append)  # get results from log files
+    end_results = list(from_logs(td))  # get results from log files
 
     end_stats = []  # the list to receive the statistics records
     # compute end result statistics for all algorithm+instance combinations

examples/ecdf_plot.py

@@ -128,11 +128,8 @@ def make_random_walk(problem) -> Execution:
                    n_runs=31)  # we will execute 31 runs per setup
     # Once we arrived here, the experiment with 2*1*31 = 62 runs has completed.
 
-    data = []  # we will load the data into this list
-    from_logs(path=td,  # the result directory
-              consumer=data.append,  # put the data into data
-              time_unit="FEs",  # time is in FEs (as opposed to "ms")
-              f_name="plainF")  # use raw, unscaled objective values
+    data = list(from_logs(  # load the data
+        path=td, time_unit="FEs", f_name="plainF"))
     ecdfs = []  # we will load the ECDFs into this list
     # The below function uses the goal objective values from the log files to
     # compute the ECDF functions. It groups all runs of one algorithm together

examples/end_results_jssp.py

@@ -22,11 +22,8 @@
         max_fes=10000,  # we grant 10000 FEs per run
         n_runs=4)  # perform 4 runs per algorithm * instance combination
 
-    end_results = []  # this list will receive the end results records
-    from_logs(td, end_results.append)  # get results from log files
-
+    data = list(from_logs(td))  # load end results
     er_csv = to_csv(  # store end results to csv file (returns path)
-        end_results,  # the list of end results to store
-        td.resolve_inside("end_results.txt"))  # path to the file to generate
+        data, td.resolve_inside("end_results.txt"))  # path to output file
     print(er_csv.read_all_str())  # read generated file as string and print it
 # When leaving "while", the temp directory will be deleted

examples/end_results_plot.py

@@ -106,8 +106,7 @@ def make_hill_climber(problem: Instance) -> Execution:
                    n_runs=31)  # we will execute 31 runs per setup
     # Once we arrived here, the experiment with 2*3*31=186 runs has completed.
 
-    data = []  # we will load the data into this list
-    from_logs(td, data.append)  # load all end results
+    data = list(from_logs(td))  # we will load the data into this list
 
     # Plot the end results in a scaled fashion: All objective values are
     # divided by the lower bound. Therefore, 1 is optimal.

examples/end_results_table.py

@@ -152,8 +152,7 @@ def make_random_sampling(problem: Instance) -> Execution:
                    n_runs=7)  # we will execute 31 runs per setup
     # Once we arrived here, the experiment with 3*3*7=63 runs has completed.
 
-    data = []  # we will load the data into this list
-    from_logs(td, data.append)  # load all end results
+    data = list(from_logs(td))  # we will load the data into this list
 
     file = tabulate_end_results(data, dir_name=td)  # create the table
     print(f"\nnow presenting markdown data from file {file!r}.\n")

examples/end_results_tests.py

@@ -87,9 +87,7 @@ def rls(problem, op1) -> Execution:
                 lambda p: rls(p, Op1MoverNflip(p.n, 2)),  # flip bits at p=2/n
                 ], n_runs=21)  # conduct 21 independent runs per setup
 
-    # load all the end results
-    end_results = []
-    from_logs(td, end_results.append)
+    end_results = list(from_logs(td))  # we will load the data into this list
 
     # create a markdown table with statistical test results
     file = tabulate_result_tests(end_results, dir_name=td)

examples/end_results_with_limits_plot.py

@@ -70,14 +70,10 @@ def make_rls(problem: OneMax) -> Execution:
 
 # We now collect the end results at different stages, namely at 16 FEs,
 # 32 FEs, 64 FEs, and after the full rumtime = 128 FEs.
-    data_16 = []  # results after 16 FEs
-    from_logs(td, data_16.append, max_fes=16)  # load end results
-    data_32 = []  # results after 32 FEs
-    from_logs(td, data_32.append, max_fes=32)  # load end results
-    data_64 = []  # results after 64 FEs
-    from_logs(td, data_64.append, max_fes=64)  # load end results
-    data_128 = []  # results after 128 FEs
-    from_logs(td, data_128.append, max_fes=128)  # load end results
+    data_16 = list(from_logs(td, max_fes=16))  # results after 16 FEs
+    data_32 = list(from_logs(td, max_fes=32))  # results after 16 FEs
+    data_64 = list(from_logs(td, max_fes=64))  # results after 16 FEs
+    data_128 = list(from_logs(td, max_fes=128))  # results after 16 FEs
     items = [[16, data_16], [32, data_32], [64, data_64], [128, data_128]]
 
 # We create a multi-figure, i.e., one figure with multiple charts inside.

examples/end_statistics_jssp.py

@@ -23,8 +23,7 @@
         max_fes=10000,  # we grant 10000 FEs per run
         n_runs=4)  # perform 4 runs per algorithm * instance combination
 
-    end_results = []  # this list will receive the end results records
-    from_logs(td, end_results.append)  # get results from log files
+    end_results = list(from_logs(td))  # get results from log files
 
     end_stats = []  # the list to receive the statistics records
     # compute end result statistics for all algorithm+instance combinations

examples/end_statistics_over_feature_plot.py

@@ -80,8 +80,7 @@ def make_rls(problem) -> Execution:
                    n_runs=7)  # we will execute 7 runs per setup
     # Once we arrived here, the 20*7 = 140 runs have completed.
 
-    end_results = []  # we will load the raw data into this list
-    from_logs(td, end_results.append)
+    end_results = list(from_logs(td))  # load results
 
     end_stats = []  # the end statistics go into this list
     from_end_results(end_results, end_stats.append)

examples/end_statistics_over_param_plot.py

@@ -93,8 +93,7 @@ def make_rls(problem, m: int) -> Execution:
                    n_runs=11)  # we will execute 11 runs per setup
     # Once we arrived here, the 11*11*2 = 242 runs have completed.
 
-    end_results = []  # we will load the raw data into this list
-    from_logs(td, end_results.append)
+    end_results = list(from_logs(td))  # load results
 
     end_stats = []  # the end statistics go into this list
     from_end_results(end_results, end_stats.append)

examples/ert_plot.py

@@ -110,11 +110,8 @@ def make_random_walk(problem) -> Execution:
                    n_runs=21)  # we will execute 71 runs per setup
     # Once we arrived here, the experiment with 2*1*31 = 62 runs has completed.
 
-    data = []  # we will load the data into this list
-    from_logs(path=td,  # the result directory
-              consumer=data.append,  # put the data into data
-              time_unit="FEs",  # time is in FEs (as opposed to "ms")
-              f_name="plainF")  # use raw, unscaled objective values
+    data = list(from_logs(  # load the data
+        path=td, time_unit="FEs", f_name="plainF"))
     ert = []  # we will load the ERT into this list
     # The below function groups all runs of one algorithm and instance
     # together and then computes the ERT.

examples/ertecdf_plot.py

@@ -71,11 +71,8 @@ def make_rls(problem) -> Execution:
                    n_runs=21)  # we will execute 21 runs per setup
     # Once we arrived here, the experiment with 11*21 = 231 runs has completed.
 
-    data = []  # we will load the data into this list
-    from_logs(path=td,  # the result directory
-              consumer=data.append,  # put the data into data
-              time_unit="FEs",  # time is in FEs (as opposed to "ms")
-              f_name="plainF")  # use raw, unscaled objective values
+    data = list(from_logs(  # load the data
+        path=td, time_unit="FEs", f_name="plainF"))
     ertecdf = []  # we will load the ERT-ECDFs into this list
     # The below function uses the goal objective values from the log files to
     # compute the ERT-ECDF functions. It groups all runs of one algorithm

examples/progress_plot.py

@@ -120,11 +120,10 @@ def make_random_walk(problem) -> Execution:
                    n_runs=5)  # we will execute 5 runs per setup
     # Once we arrived here, the experiment with 2*2*5 = 20 runs has completed.
 
-    data = []  # we will load the data into this list
-    pr_from_logs(path=td,  # the result directory
-                 consumer=data.append,  # put the data into data
-                 time_unit="FEs",  # time is in FEs (as opposed to "ms")
-                 f_name="plainF")  # use raw, unscaled objective values
+    data = list(pr_from_logs(  # we will load the data into this list
+        path=td,  # the result directory
+        time_unit="FEs",  # time is in FEs (as opposed to "ms")
+        f_name="plainF"))  # use raw, unscaled objective values
 
     # The first plot will contain every single one of the 20 runs.
     # The system will choose different styles for different algorithms