start to find duplicates

duplicates/__init__.py

@@ -1,3 +1,3 @@
-from .dupes import listfiles
+from .dupes import listfiles, finddupes
 
-__ALL__ = ['listfiles']
+__ALL__ = ['listfiles', 'finddupes']

duplicates/dupes.py

@@ -1,6 +1,6 @@
 from pathlib import Path
 
-def listfiles(in_path: Path) -> dict:
+def listfiles(in_path: Path) -> dict[int, set]:
     filedict = dict()
     for root, dirs, files in in_path.walk():
         for file in files:
@@ -10,4 +10,11 @@ def listfiles(in_path: Path) -> dict:
                 filedict[size].add(filepath)
             else:
                 filedict[size] = {filepath}
-    return filedict
+    return filedict
+
+def finddupes(filesbysize: dict[int, set]) -> set[frozenset]:
+    dupes = {
+        frozenset(filepath for filepath in files)
+        for size, files in filesbysize.items() if len(files) > 1
+    }
+    return dupes

test/majorver/__init__.py

@@ -1,9 +1,22 @@
+from pytest import fixture
 from .. import *
 from pathlib import Path
 
 dir1 = Path('test/majorver/data/dir1')
 dir2 = Path('test/majorver/data/dir2')
 
+@fixture
+def testfiles(tmp_path) -> Path:
+    dir1.copy(tmp_path)
+    dir2.copy(tmp_path)
+    return tmp_path
+
+@fixture
+def duplicatedir1(testfiles) -> Path:
+    tmp_path = testfiles
+    dir1.copy(testfiles / 'alt')
+    return tmp_path    
+
 def _copy(self: Path, target: Path) -> None:
     from shutil import copytree
     if self.is_dir():

test/majorver/test_filecompare.py

@@ -0,0 +1,8 @@
+from pytest import fixture
+from . import listfiles, finddupes, Path, duplicatedir1, testfiles
+
+def test_fileisduplicate(duplicatedir1):
+    testfiles = duplicatedir1
+    filesdict = listfiles(testfiles)
+    duplicatefiles = finddupes(filesdict)
+    assert duplicatefiles == {frozenset((testfiles / 'dir1' / 'FileA.txt', testfiles / 'alt' / 'dir1' / 'FileA.txt'))}

test/majorver/test_filedetails.py

@@ -1,19 +1,6 @@
-from pytest import fixture
 from recurtools import flatten
 
-from . import listfiles, Path, dir1, dir2
-
-@fixture
-def testfiles(tmp_path):
-    dir1.copy(tmp_path)
-    dir2.copy(tmp_path)
-    return tmp_path
-
-@fixture
-def duplicatedir1(testfiles):
-    tmp_path = testfiles
-    dir1.copy(testfiles / 'alt')
-    return tmp_path    
+from . import listfiles, Path, testfiles, duplicatedir1
 
 def test_fileslisted(testfiles):
     filesdict = listfiles(testfiles)