From 2a054276cdf5f80f724ab5d7de8215792da5dc5f Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?P=C5=91cze=20Bence?= <bence@pocze.ch>
Date: Thu, 14 Sep 2023 21:26:46 +0200
Subject: [PATCH] make data structure for shifted lines cleaner

---
 pydiffchecker/cli.py                |  8 +++--
 pydiffchecker/line_shift_checker.py | 48 ++++++++++++++++++++---------
 2 files changed, 39 insertions(+), 17 deletions(-)

diff --git a/pydiffchecker/cli.py b/pydiffchecker/cli.py
index 1cbb0f1..436d31f 100644
--- a/pydiffchecker/cli.py
+++ b/pydiffchecker/cli.py
@@ -1,5 +1,4 @@
 import argparse
-import json
 from .version import get_version
 from .line_shift_checker import LineShiftChecker
 
@@ -17,6 +16,9 @@ def main():
     args = parser.parse_args()
 
     line_shift_checker = LineShiftChecker(args.revision_since, args.revision_until)
-    out = line_shift_checker.get_shifted_lines()
+    all_shifted_lines = line_shift_checker.get_all_shifted_lines()
 
-    print(json.dumps(out, indent=4))
+    for src_path, shifted_lines in all_shifted_lines.items():
+        print(f'* {src_path}->{shifted_lines.dst_path}:')
+        for src_line_index, dst_line_index in shifted_lines:
+            print(f'    {src_line_index}->{dst_line_index}')
diff --git a/pydiffchecker/line_shift_checker.py b/pydiffchecker/line_shift_checker.py
index 9109f6c..e2d514c 100644
--- a/pydiffchecker/line_shift_checker.py
+++ b/pydiffchecker/line_shift_checker.py
@@ -1,8 +1,32 @@
 import re
-from typing import List, Dict
+from typing import List, Dict, Sized, Iterable
 from .helper import subprocess_readlines
 
 
+class ShiftedLines(Sized, Iterable):
+    def __init__(self, src_path: str, dst_path: str) -> None:
+        self.src_path = src_path
+        self.dst_path = dst_path
+        self.__lines: 'Dict[int, int | None]' = {}
+
+    def __setitem__(self, src_line_index, dst_line_index) -> None:
+        self.__lines[src_line_index] = dst_line_index
+
+    def __contains__(self, src_line_index) -> bool:
+        return src_line_index in self.__lines and self.__lines[src_line_index] is not None
+
+    def __getitem__(self, src_line_index) -> 'int | None':
+        if src_line_index not in self.__lines:
+            return None
+        return self.__lines[src_line_index]
+
+    def __iter__(self):
+        return iter(self.__lines.items())
+
+    def __len__(self) -> int:
+        return len(self.__lines)
+
+
 class LineShiftChecker:
     DIFF_BLOCK_REGEX = r'@@ -(\d+),(\d+) \+(\d+),(\d+) @@'
 
@@ -10,13 +34,9 @@ class LineShiftChecker:
         self.revision_since = revision_since
         self.revision_until = revision_until
 
-    def get_shifted_lines(self) -> Dict[str, Dict]:
-        shifted_lines = {}
-
-        for file_info in self.__get_changed_files():
-            shifted_lines[file_info['src']] = self.__get_shifted_lines_in_file(file_info)
-
-        return shifted_lines
+    def get_all_shifted_lines(self) -> Dict[str, ShiftedLines]:
+        return {file_info['src']: self.__get_shifted_lines_in_file(file_info)
+                for file_info in self.__get_changed_files()}
 
     def __get_changed_files(self) -> List[Dict]:
         process_output = subprocess_readlines(['git', 'diff', '--name-status', '--diff-filter=MR',
@@ -32,12 +52,12 @@ class LineShiftChecker:
 
         return file_list
 
-    def __get_shifted_lines_in_file(self, file_info) -> 'Dict[str, str | None]':
+    def __get_shifted_lines_in_file(self, file_info) -> ShiftedLines:
         process_output = subprocess_readlines(['git', 'diff',
                                                self.revision_since, self.revision_until, '--',
                                                file_info['src'], file_info['dst']])
 
-        shifted_lines: 'Dict[str, str | None]' = {}
+        shifted_lines = ShiftedLines(file_info['src'], file_info['dst'])
         src_line_index = 1
         dst_line_index = 1
         diff_started = False
@@ -49,7 +69,7 @@ class LineShiftChecker:
 
                 # fill shifted lines between 2 diff blocks
                 for i in range(0, diff_block_src_start - src_line_index):
-                    shifted_lines[f'{file_info["src"]}:{src_line_index+i}'] = f'{file_info["dst"]}:{dst_line_index+i}'
+                    shifted_lines[src_line_index+i] = dst_line_index+i
 
                 src_line_index = diff_block_src_start
                 dst_line_index = diff_block_dst_start
@@ -60,19 +80,19 @@ class LineShiftChecker:
                 continue
 
             if line.startswith(' '):
-                shifted_lines[f'{file_info["src"]}:{src_line_index}'] = f'{file_info["dst"]}:{dst_line_index}'
+                shifted_lines[src_line_index] = dst_line_index
                 src_line_index += 1
                 dst_line_index += 1
             elif line.startswith('+'):
                 dst_line_index += 1
             elif line.startswith('-'):
-                shifted_lines[f'{file_info["src"]}:{src_line_index}'] = None
+                shifted_lines[src_line_index] = None
                 src_line_index += 1
 
         # fill shifted lines until end of file
         lines_in_source_file = self.__count_lines_in_source_file(file_info['src'])
         for i in range(0, lines_in_source_file - src_line_index + 1):
-            shifted_lines[f'{file_info["src"]}:{src_line_index+i}'] = f'{file_info["dst"]}:{dst_line_index+i}'
+            shifted_lines[src_line_index+i] = dst_line_index+i
 
         assert lines_in_source_file == len(shifted_lines)