Add getter for worse public fragments

src/python/common/tool_arguments.py

@@ -86,3 +86,6 @@ class RunToolArgument(Enum):
                                         f'"{ColumnName.LANG.value}" column are: '
                                         f'{LanguageVersion.PYTHON_3.value}, {LanguageVersion.JAVA_8.value}, '
                                         f'{LanguageVersion.JAVA_11.value}, {LanguageVersion.KOTLIN.value}.')
+
+    DIFFS_FILE_PATH = ArgumentsInfo(None, 'diffs_file_path',
+                                    'Path to a file with serialized diffs that were founded by diffs_between_df.py')

src/python/evaluation/common/pandas_util.py

@@ -1,7 +1,7 @@
 import json
 import logging
 from pathlib import Path
-from typing import List, Set, Union
+from typing import Any, List, Set, Union
 
 import numpy as np
 import pandas as pd
@@ -21,6 +21,10 @@ def filter_df_by_language(df: pd.DataFrame, languages: Set[LanguageVersion],
     return df.loc[df[column].isin(set(map(lambda l: l.value, languages)))]
 
 
+def filter_df_by_condition(df: pd.DataFrame, column: str, value: Any) -> pd.DataFrame:
+    return df.loc[df[column] == value]
+
+
 def drop_duplicates(df: pd.DataFrame, column: str = ColumnName.CODE.value) -> pd.DataFrame:
     return df.drop_duplicates(column, keep='last')
 

src/python/evaluation/common/util.py

@@ -15,6 +15,7 @@ class ColumnName(Enum):
     ROW = 'row'
     OLD = 'old'
     NEW = 'new'
+    IS_PUBLIC = 'is_public'
 
 
 @unique

src/python/evaluation/inspectors/README.md

@@ -11,8 +11,10 @@ This module contains _preprocessing_ stage and _analysing_ stage.
 `Analysing` stage includes:
 - [diffs_between_df.py](diffs_between_df.py) allows finding a difference between 
   old and new grades and collect issues that were found in new data
-- [print_inspectors_statistics.py](print_inspectors_statistics.py) allows print statistics 
+- [print_inspectors_statistics.py](print_inspectors_statistics.py) allows printing statistics 
   that were found by [diffs_between_df.py](diffs_between_df.py)
+- [get_worse_public_examples.py](get_worse_public_examples.py) allows getting 
+  top N worse public examples from a dataset. The measure is to count unique new inspections.
 
 ___
 
@@ -200,4 +202,34 @@ ERROR_PRONE: 17 issues, 2363 fragments
 COMPLEXITY: 17 issues, 13928 fragments
 COHESION: 1 issues, 3826 fragments
 ______
-```
+```
+
+---
+
+### Get worse public examples
+
+[get_worse_public_examples.py](get_worse_public_examples.py) allows getting 
+  top N worse public examples from a dataset. The measure is to count unique new inspections.
+
+#### Usage
+
+Run the [get_worse_public_examples.py](get_worse_public_examples.py) with the arguments from command line.
+
+Required arguments:
+
+- `solutions_file_path` — path to xlsx-file or csv-file with graded code samples;
+- `diffs_file_path` — path to a `pickle` file, that was calculated by [diffs_between_df.py](diffs_between_df.py).
+
+Please, note that your `solutions_file_path` file with code fragments should consist of at least 2 obligatory columns:
+
+- `code`,
+- `traceback`,
+- `is_public`,
+- `id`.
+
+Optional arguments:
+Argument | Description
+--- | ---
+|**‑n**, **‑‑n**| The N worse fragments will be saved.|
+
+The resulting file will be stored in the same folder as the `solutions_file_path` input file.

src/python/evaluation/inspectors/get_worse_public_examples.py

@@ -0,0 +1,68 @@
+import argparse
+from pathlib import Path
+from typing import Dict, List
+
+import pandas as pd
+from src.python.common.tool_arguments import RunToolArgument
+from src.python.evaluation.common.csv_util import write_dataframe_to_csv
+from src.python.evaluation.common.pandas_util import filter_df_by_condition, get_solutions_df_by_file_path
+from src.python.evaluation.common.util import ColumnName, EvaluationArgument
+from src.python.review.common.file_system import deserialize_data_from_file, Extension, get_parent_folder
+from src.python.review.inspectors.issue import BaseIssue
+
+
+def configure_arguments(parser: argparse.ArgumentParser) -> None:
+    parser.add_argument(RunToolArgument.SOLUTIONS_FILE_PATH.value.long_name,
+                        type=lambda value: Path(value).absolute(),
+                        help=RunToolArgument.SOLUTIONS_FILE_PATH.value.description)
+
+    parser.add_argument(RunToolArgument.DIFFS_FILE_PATH.value.long_name,
+                        type=lambda value: Path(value).absolute(),
+                        help=RunToolArgument.DIFFS_FILE_PATH.value.description)
+
+    parser.add_argument('-n', '--n',
+                        help='The N worse fragments will be saved',
+                        type=int,
+                        default=10)
+
+
+def __get_new_inspections(fragment_id_to_issues: Dict[int, List[BaseIssue]], fragment_id: int) -> str:
+    return ','.join(set(map(lambda i: i.origin_class, fragment_id_to_issues.get(fragment_id, []))))
+
+
+def __get_public_fragments(solutions_df: pd.DataFrame, diffs_dict: dict) -> pd.DataFrame:
+    # Keep only public solutions
+    public_fragments = filter_df_by_condition(solutions_df, ColumnName.IS_PUBLIC.value, 'YES')
+    count_inspections_column = 'count_inspections'
+    new_inspections_column = 'new_inspections'
+
+    # Get only new inspections and count them
+    fragment_id_to_issues = diffs_dict[EvaluationArgument.TRACEBACK.value]
+    public_fragments[new_inspections_column] = public_fragments.apply(
+        lambda row: __get_new_inspections(fragment_id_to_issues, row[ColumnName.ID.value]), axis=1)
+    public_fragments[count_inspections_column] = public_fragments.apply(
+        lambda row: len(row[new_inspections_column].split(',')), axis=1)
+
+    public_fragments = public_fragments.sort_values(count_inspections_column, ascending=False)
+    # Keep only public columns
+    return public_fragments[[ColumnName.CODE.value, EvaluationArgument.TRACEBACK.value, new_inspections_column]]
+
+
+# TODO: add readme
+def main() -> None:
+    parser = argparse.ArgumentParser()
+    configure_arguments(parser)
+    args = parser.parse_args()
+
+    solutions_file_path = args.solutions_file_path
+    solutions_df = get_solutions_df_by_file_path(solutions_file_path)
+    diffs = deserialize_data_from_file(args.diffs_file_path)
+
+    public_fragments = __get_public_fragments(solutions_df, diffs)
+
+    output_path = get_parent_folder(Path(solutions_file_path)) / f'worse_fragments{Extension.CSV.value}'
+    write_dataframe_to_csv(output_path, public_fragments.head(args.n))
+
+
+if __name__ == '__main__':
+    main()

src/python/evaluation/inspectors/print_inspectors_statistics.py

@@ -3,16 +3,17 @@
 from pathlib import Path
 from typing import Dict
 
+from src.python.common.tool_arguments import RunToolArgument
 from src.python.evaluation.common.util import ColumnName, EvaluationArgument
 from src.python.evaluation.inspectors.common.statistics import IssuesStatistics
 from src.python.review.common.file_system import deserialize_data_from_file
 from src.python.review.inspectors.issue import ShortIssue
 
 
 def configure_arguments(parser: argparse.ArgumentParser) -> None:
-    parser.add_argument('diffs_file_path',
+    parser.add_argument(RunToolArgument.DIFFS_FILE_PATH.value.long_name,
                         type=lambda value: Path(value).absolute(),
-                        help='Path to a file with serialized diffs that were founded by diffs_between_df.py')
+                        help=RunToolArgument.DIFFS_FILE_PATH.value.description)
 
     parser.add_argument('--categorize',
                         help='If True, statistics will be categorized by several categories.',