feature: added language analyzer

Author: n0rfas

.github/workflows/code-check-dev.yml

@@ -44,4 +44,4 @@ jobs:
       - run: pipx install "poetry~=2.0.0"
       - run: poetry install --with dev
       - run: poetry run ruff check --select I .
-      - run: poetry run pytest --cov=git_analytics --cov-report=term-missing --cov-fail-under=40
+      - run: poetry run pytest --cov=git_analytics --cov-report=term-missing --cov-fail-under=45

README.md

@@ -49,7 +49,7 @@ poetry run git-analytics
 
 ```bash
 poetry run pytest
-poetry run pytest --cov=git_analytics --cov-report=term-missing --cov-fail-under=40
+poetry run pytest --cov=git_analytics --cov-report=term-missing --cov-fail-under=45
 ```
 
 ### Type Checking

git_analytics/analyzers/__init__.py

@@ -2,12 +2,14 @@
 from .commit_type import CommitTypeAnalyzer
 from .commits_summary import CommitsSummaryAnalyzer
 from .historical_statistics import HistoricalStatisticsAnalyzer
+from .language_statistics import LanguageAnalyzer
 from .lines_statistics import LinesAnalyzer
 
 __all__ = [
     "AuthorsStatisticsAnalyzer",
     "CommitTypeAnalyzer",
     "CommitsSummaryAnalyzer",
     "HistoricalStatisticsAnalyzer",
+    "LanguageAnalyzer",
     "LinesAnalyzer",
 ]

git_analytics/analyzers/commit_type.py

@@ -24,8 +24,8 @@ class CommitType(Enum):
 @dataclass
 class Result(AnalyticsResult):
     timeseries: Dict[date, Dict[CommitType, int]]
-    total_counter: Counter
-    author_total_counter: Dict[str, Counter]
+    total_counter: Dict[CommitType, int]
+    author_total_counter: Dict[str, Dict[CommitType, int]]
 
 
 TYPE_COMMIT_LIST: tuple = tuple(ct.value for ct in CommitType)
@@ -57,6 +57,6 @@ def process(self, commit: AnalyticsCommit) -> None:
     def result(self) -> Result:
         return Result(
             timeseries={dt: dict(counter) for dt, counter in self._by_date.items()},
-            total_counter=self._total_counter,
-            author_total_counter=self._author_total_counter,
+            author_total_counter={author: dict(counter) for author, counter in self._author_total_counter.items()},
+            total_counter=dict(self._total_counter),
         )

git_analytics/analyzers/language_statistics.py

@@ -0,0 +1,53 @@
+from collections import defaultdict
+from dataclasses import dataclass
+from typing import Dict
+
+from git_analytics.entities import AnalyticsCommit, AnalyticsResult
+from git_analytics.interfaces import CommitAnalyzer
+
+
+@dataclass
+class FileExtensionChangeStats:
+    insertions: int = 0
+    deletions: int = 0
+
+
+@dataclass
+class Result(AnalyticsResult):
+    files_extensions_total: Dict[str, FileExtensionChangeStats]
+    files_extensions_by_author: Dict[str, Dict[str, FileExtensionChangeStats]]
+
+
+def _get_file_extension(file_path: str) -> str:
+    filename = file_path.split("/")[-1]
+    filename_parts = filename.split(".")
+    if len(filename_parts) == 1 or filename_parts[0] == "":
+        return "no_extension"
+    return filename_parts[-1].lower()
+
+
+class LanguageAnalyzer(CommitAnalyzer):
+    name = "language_statistics"
+
+    def __init__(self) -> None:
+        self._total: Dict[str, FileExtensionChangeStats] = defaultdict(FileExtensionChangeStats)
+        self._by_author: Dict[str, Dict[str, FileExtensionChangeStats]] = defaultdict(
+            lambda: defaultdict(FileExtensionChangeStats)
+        )
+
+    def process(self, commit: AnalyticsCommit) -> None:
+        for changed_file in commit.files:
+            file_extension = _get_file_extension(changed_file)
+            self._total[file_extension].insertions += commit.files[changed_file].insertions
+            self._total[file_extension].deletions += commit.files[changed_file].deletions
+            self._by_author[commit.commit_author][file_extension].insertions += commit.files[changed_file].insertions
+            self._by_author[commit.commit_author][file_extension].deletions += commit.files[changed_file].deletions
+
+    def result(self) -> Result:
+        return Result(
+            files_extensions_total=dict(sorted(self._total.items(), key=lambda item: item[1].insertions, reverse=True)),
+            files_extensions_by_author={
+                author: dict(sorted(counter.items(), key=lambda item: item[1].insertions, reverse=True))
+                for author, counter in self._by_author.items()
+            },
+        )

git_analytics/entities.py

@@ -5,6 +5,12 @@
 from typing import Any, Dict
 
 
+@dataclass
+class FileChangeStats:
+    insertions: int
+    deletions: int
+
+
 @dataclass
 class AnalyticsCommit:
     sha: str
@@ -14,6 +20,7 @@ class AnalyticsCommit:
     lines_deletions: int
     files_changed: int
     message: str
+    files: Dict[str, FileChangeStats]
 
 
 @dataclass

git_analytics/main.py

@@ -7,6 +7,7 @@
     CommitsSummaryAnalyzer,
     CommitTypeAnalyzer,
     HistoricalStatisticsAnalyzer,
+    LanguageAnalyzer,
     LinesAnalyzer,
 )
 from git_analytics.engine import CommitAnalyticsEngine
@@ -20,6 +21,7 @@ def make_analyzers():
         CommitsSummaryAnalyzer(),
         CommitTypeAnalyzer(),
         HistoricalStatisticsAnalyzer(),
+        LanguageAnalyzer(),
         LinesAnalyzer(),
     ]
 

git_analytics/sources/git_commit_adapter.py

@@ -3,7 +3,7 @@
 from git import Repo
 from git.objects import Commit
 
-from git_analytics.entities import AnalyticsCommit
+from git_analytics.entities import AnalyticsCommit, FileChangeStats
 from git_analytics.interfaces import CommitSource
 
 
@@ -25,4 +25,8 @@ def git_commit_to_analytics_commit(commit: Commit) -> AnalyticsCommit:
             lines_deletions=commit.stats.total["deletions"],
             files_changed=commit.stats.total["files"],
             message=str(commit.summary).strip(),
+            files={
+                str(file): FileChangeStats(insertions=value.get("insertions", 0), deletions=value.get("deletions", 0))
+                for file, value in commit.stats.files.items()
+            },
         )

git_analytics/static/js/git-analytics.js

@@ -64,7 +64,7 @@ async function loadAndRender(type, value, label) {
     buildWeekByAuthorChart(stats.historical_statistics.day_of_week);
     buildDayOfMonthByAuthorChart(stats.historical_statistics.day_of_month);
     buildLinesChart(stats.lines_statistics.items);
-    buildCommitTypeChart(stats.commit_type.items);
+    buildCommitTypeChart(stats.commit_type.timeseries);
     buildAuthorsTable(stats.authors_statistics.authors);
 
   } catch (err) {