fix+feat: adds the resulting percentage of similarity by source.

- Simplifies getting paths from the configuration file;
- Adds the resulting percentage of similarity by source.

Refs: #190

Makefile

@@ -1,4 +1,4 @@
-UTIL_VERSION            := 0.4.7
+UTIL_VERSION            := 0.4.8
 UTIL_NAME               := codeplag
 PWD                     := $(shell pwd)
 

locales/codeplag.pot

@@ -6,16 +6,16 @@
 #, fuzzy
 msgid ""
 msgstr ""
-"Project-Id-Version: codeplag 0.4.6\n"
+"Project-Id-Version: codeplag 0.4.8\n"
 "Report-Msgid-Bugs-To: EMAIL@ADDRESS\n"
-"POT-Creation-Date: 2024-06-02 20:54+0300\n"
+"POT-Creation-Date: 2024-06-30 12:31+0300\n"
 "PO-Revision-Date: YEAR-MO-DA HO:MI+ZONE\n"
 "Last-Translator: Artyom Semidolin\n"
 "Language-Team: LANGUAGE <LL@li.org>\n"
 "MIME-Version: 1.0\n"
 "Content-Type: text/plain; charset=utf-8\n"
 "Content-Transfer-Encoding: 8bit\n"
-"Generated-By: Babel 2.14.0\n"
+"Generated-By: Babel 2.15.0\n"
 
 #: src/codeplag/codeplagcli.py:44
 msgid "You cannot specify the same value multiple times. You provided '{values}'."
@@ -262,7 +262,7 @@ msgid "Part of the second program"
 msgstr ""
 
 #: src/templates/general.templ:95 src/templates/general.templ:126
-#: src/templates/sources.templ:61
+#: src/templates/sources.templ:63
 msgid "Similarity"
 msgstr ""
 
@@ -278,19 +278,23 @@ msgstr ""
 msgid "The path to the file being checked"
 msgstr ""
 
-#: src/templates/sources.templ:56
-msgid "Similar parts of the programs"
+#: src/templates/sources.templ:49
+msgid "Total maximum compliance, %"
 msgstr ""
 
 #: src/templates/sources.templ:58
+msgid "Similar parts of the programs"
+msgstr ""
+
+#: src/templates/sources.templ:60
 msgid "The name of a similar file"
 msgstr ""
 
-#: src/templates/sources.templ:59
+#: src/templates/sources.templ:61
 msgid "Part of the file that is being checked"
 msgstr ""
 
-#: src/templates/sources.templ:60
+#: src/templates/sources.templ:62
 msgid "Part of a similar file"
 msgstr ""
 

locales/translations/en/LC_MESSAGES/codeplag.po

@@ -5,7 +5,7 @@
 #
 msgid ""
 msgstr ""
-"Project-Id-Version: codeplag 0.4.5\n"
+"Project-Id-Version: codeplag 0.4.8\n"
 "Report-Msgid-Bugs-To: EMAIL@ADDRESS\n"
 "POT-Creation-Date: 2024-05-21 09:28+0300\n"
 "PO-Revision-Date: 2024-05-16 19:15+0300\n"
@@ -16,7 +16,7 @@ msgstr ""
 "MIME-Version: 1.0\n"
 "Content-Type: text/plain; charset=utf-8\n"
 "Content-Transfer-Encoding: 8bit\n"
-"Generated-By: Babel 2.14.0\n"
+"Generated-By: Babel 2.15.0\n"
 
 #: src/codeplag/codeplagcli.py:44
 msgid "You cannot specify the same value multiple times. You provided '{values}'."
@@ -279,7 +279,7 @@ msgid "Part of the second program"
 msgstr "Part of the second program"
 
 #: src/templates/general.templ:95 src/templates/general.templ:126
-#: src/templates/sources.templ:61
+#: src/templates/sources.templ:63
 msgid "Similarity"
 msgstr "Similarity"
 
@@ -295,19 +295,22 @@ msgstr "Verification results"
 msgid "The path to the file being checked"
 msgstr "The path to the file being checked"
 
-#: src/templates/sources.templ:56
+#: src/templates/sources.templ:49
+msgid "Total maximum compliance, %"
+msgstr "Total maximum compliance, %"
+
+#: src/templates/sources.templ:58
 msgid "Similar parts of the programs"
 msgstr "Similar parts of the programs"
 
-#: src/templates/sources.templ:58
+#: src/templates/sources.templ:60
 msgid "The name of a similar file"
 msgstr "The name of a similar file"
 
-#: src/templates/sources.templ:59
+#: src/templates/sources.templ:61
 msgid "Part of the file that is being checked"
 msgstr "Part of the file that is being checked"
 
-#: src/templates/sources.templ:60
+#: src/templates/sources.templ:62
 msgid "Part of a similar file"
 msgstr "Part of a similar file"
-

locales/translations/ru/LC_MESSAGES/codeplag.po

@@ -5,7 +5,7 @@
 #
 msgid ""
 msgstr ""
-"Project-Id-Version: codeplag 0.4.5\n"
+"Project-Id-Version: codeplag 0.4.8\n"
 "Report-Msgid-Bugs-To: EMAIL@ADDRESS\n"
 "POT-Creation-Date: 2024-05-21 09:28+0300\n"
 "PO-Revision-Date: 2024-05-11 12:05+0300\n"
@@ -17,7 +17,7 @@ msgstr ""
 "MIME-Version: 1.0\n"
 "Content-Type: text/plain; charset=utf-8\n"
 "Content-Transfer-Encoding: 8bit\n"
-"Generated-By: Babel 2.14.0\n"
+"Generated-By: Babel 2.15.0\n"
 
 #: src/codeplag/codeplagcli.py:44
 msgid "You cannot specify the same value multiple times. You provided '{values}'."
@@ -261,15 +261,15 @@ msgstr "Коэффициент Жаккара"
 
 #: src/templates/general.templ:71
 msgid "Similarity of operators"
-msgstr "Схожеть операторов"
+msgstr "Схожесть операторов"
 
 #: src/templates/general.templ:75
 msgid "Similarity of keywords"
-msgstr "Схожеть ключевых слов"
+msgstr "Схожесть ключевых слов"
 
 #: src/templates/general.templ:79
 msgid "Similarity of literals"
-msgstr "Схожеть литералов"
+msgstr "Схожеcть литералов"
 
 #: src/templates/general.templ:82
 msgid "Weighted average by fast metrics"
@@ -292,7 +292,7 @@ msgid "Part of the second program"
 msgstr "Часть второй программы"
 
 #: src/templates/general.templ:95 src/templates/general.templ:126
-#: src/templates/sources.templ:61
+#: src/templates/sources.templ:63
 msgid "Similarity"
 msgstr "Схожесть"
 
@@ -308,19 +308,22 @@ msgstr "Результаты проверки"
 msgid "The path to the file being checked"
 msgstr "Путь до проверяемого файла"
 
-#: src/templates/sources.templ:56
+#: src/templates/sources.templ:49
+msgid "Total maximum compliance, %"
+msgstr "Суммарное максимальное соответствие, %"
+
+#: src/templates/sources.templ:58
 msgid "Similar parts of the programs"
 msgstr "Схожие части программ"
 
-#: src/templates/sources.templ:58
+#: src/templates/sources.templ:60
 msgid "The name of a similar file"
 msgstr "Имя схожего файла"
 
-#: src/templates/sources.templ:59
+#: src/templates/sources.templ:61
 msgid "Part of the file that is being checked"
 msgstr "Часть проверяемого файла"
 
-#: src/templates/sources.templ:60
+#: src/templates/sources.templ:62
 msgid "Part of a similar file"
 msgstr "Часть схожего файла"
-

src/codeplag/config.py

@@ -1,8 +1,6 @@
 import json
 from pathlib import Path
-from typing import Any, ForwardRef, Literal, Mapping, overload
-
-from typing_extensions import NotRequired
+from typing import Any, Literal, Mapping, overload
 
 from codeplag.consts import (
     CONFIG_PATH,
@@ -59,19 +57,13 @@ def read_settings_conf() -> Settings:
         )
         return DefaultSettingsConfig
 
-    for key, key_type in Settings.__annotations__.items():
+    for key in Settings.__annotations__:
         if key not in loaded_settings_config:
             if key in DefaultSettingsConfig:
                 loaded_settings_config[key] = DefaultSettingsConfig[key]
             continue
 
-        if key_type in [
-            Path,
-            NotRequired[Path],  # type: ignore
-            # Hook for proper work of the Cythonized version.
-            ForwardRef("Path"),
-            ForwardRef("NotRequired[Path]"),
-        ]:
+        if key in ["environment", "reports"]:
             loaded_settings_config[key] = Path(loaded_settings_config[key])
 
     return Settings(

src/codeplag/handlers/report.py

@@ -183,50 +183,63 @@ def _get_parsed_line(
 class Elements(TypedDict):
     cnt_elements: int
     same_parts: SameFuncs
+    max_funcs_same_percentages: dict[str, float]
 
 
 SamePartsOfAll = dict[str, dict[str, Elements]]
 
 
+def _get_resulting_same_percentages(
+    same_parts_of_all: SamePartsOfAll,
+) -> dict[str, float]:
+    resulting_same_percentages: dict[str, float] = {}
+    for first_path, same_works in same_parts_of_all.items():
+        max_funcs_same_percentages = {}
+        for second_work in same_works.values():
+            for function, same_percentage in second_work[
+                "max_funcs_same_percentages"
+            ].items():
+                if same_percentage <= max_funcs_same_percentages.get(function, 0):
+                    continue
+                max_funcs_same_percentages[function] = same_percentage
+        if not (cnt_functions := len(max_funcs_same_percentages)):
+            continue
+        resulting_percentage = round(
+            sum(max_funcs_same_percentages.values()) / cnt_functions,
+            2,
+        )
+        resulting_same_percentages[first_path] = resulting_percentage
+    return resulting_same_percentages
+
+
 def _search_sources(
     df: pd.DataFrame, threshold: int = DEFAULT_THRESHOLD
 ) -> SamePartsOfAll:
     same_parts_of_all: SamePartsOfAll = defaultdict(lambda: {})
     for line, _, same_parts_of_second, same_parts_of_first in _get_parsed_line(
         df, threshold, include_funcs_less_threshold=False
     ):
-        same_parts_of_all[line.first_path][line.second_path] = Elements(
-            cnt_elements=0, same_parts=deepcopy(same_parts_of_second)
-        )
-        same_parts_of_all[line.second_path][line.first_path] = Elements(
-            cnt_elements=0, same_parts=deepcopy(same_parts_of_first)
-        )
-        for function, same_functions in same_parts_of_second.items():
-            cnt_same_functions = len(same_functions)
-            if cnt_same_functions == 0:
-                same_parts_of_all[line.first_path][line.second_path]["same_parts"].pop(
-                    function
+        for first_path, second_path, same_parts in (
+            (line.first_path, line.second_path, same_parts_of_second),
+            (line.second_path, line.first_path, same_parts_of_first),
+        ):
+            element = same_parts_of_all[first_path][second_path] = Elements(
+                cnt_elements=0,
+                same_parts=deepcopy(same_parts),
+                max_funcs_same_percentages={},
+            )
+            for function, same_functions in same_parts.items():
+                if (cnt_same_functions := len(same_functions)) == 0:
+                    element["same_parts"].pop(function)
+                    continue
+                element["max_funcs_same_percentages"][function] = max(
+                    same_function.percent for same_function in same_functions
                 )
-                continue
-            same_parts_of_all[line.first_path][line.second_path][
-                "cnt_elements"
-            ] += cnt_same_functions
-        for function, same_functions in same_parts_of_first.items():
-            cnt_same_functions = len(same_functions)
-            if cnt_same_functions == 0:
-                same_parts_of_all[line.second_path][line.first_path]["same_parts"].pop(
-                    function
-                )
-                continue
-            same_parts_of_all[line.second_path][line.first_path][
-                "cnt_elements"
-            ] += cnt_same_functions
-        if same_parts_of_all[line.first_path][line.second_path]["cnt_elements"] == 0:
-            del same_parts_of_all[line.first_path][line.second_path]
-        if same_parts_of_all[line.second_path][line.first_path]["cnt_elements"] == 0:
-            del same_parts_of_all[line.second_path][line.first_path]
-    same_parts_of_all = {k: v for k, v in same_parts_of_all.items() if v}
-    return same_parts_of_all
+                element["cnt_elements"] += cnt_same_functions
+            if element["cnt_elements"] == 0:
+                del same_parts_of_all[first_path][second_path]
+
+    return {k: v for k, v in same_parts_of_all.items() if v}
 
 
 def _create_report(
@@ -259,12 +272,14 @@ def _create_sources_report(
     language: Language = DEFAULT_LANGUAGE,
 ) -> None:
     data = _search_sources(read_df(df_path), threshold)
+    same_percentages = _get_resulting_same_percentages(data)
     template = environment.from_string(SOURCES_TEMPLATE_PATH.read_text())
     if save_path.is_dir():
         save_path = save_path / DEFAULT_SOURCES_REPORT_NAME
     save_path.write_text(
         template.render(
             data=data,
+            same_percentages=same_percentages,
             language=language,
             enumerate=enumerate,
             Path=Path,

src/templates/sources.templ

@@ -46,9 +46,11 @@
                 <caption>{{ _("General information") }}</caption>
                 <tr class="table__header">
                     <th>{{ _("The path to the file being checked") }}</th>
+                    <th>{{ _("Total maximum compliance, %") }}</th>
                 </tr>
                 <tr class="table__row">
                     <td>{{ work_path }}</td>
+                    <th>{{ same_percentages[work_path] }}</th>
                 </tr>
             </table>
             <div align="center">