Sort list of RECORD entries

Without sorting, the 'installed' hash had entries in random order
that caused output to differ for every run.

See https://reproducible-builds.org/ for why this matters.

Sorting all entries to make testing easier.

news/5525.feature

@@ -0,0 +1 @@
+pip now ensures that the RECORD file is sorted when installing from a wheel file.

src/pip/_internal/wheel.py

@@ -500,16 +500,19 @@ def _get_script_text(entry):
         with open_for_csv(temp_record, 'w+') as record_out:
             reader = csv.reader(record_in)
             writer = csv.writer(record_out)
+            outrows = []
             for row in reader:
                 row[0] = installed.pop(row[0], row[0])
                 if row[0] in changed:
                     row[1], row[2] = rehash(row[0])
-                writer.writerow(row)
+                outrows.append(tuple(row))
             for f in generated:
                 digest, length = rehash(f)
-                writer.writerow((normpath(f, lib_dir), digest, length))
+                outrows.append((normpath(f, lib_dir), digest, length))
             for f in installed:
-                writer.writerow((installed[f], '', ''))
+                outrows.append((installed[f], '', ''))
+            for row in sorted(outrows):
+                writer.writerow(row)
     shutil.move(temp_record, record)
 
 

tests/functional/test_install_wheel.py

@@ -217,6 +217,23 @@ def test_install_from_wheel_no_deps(script, data):
     assert pkg_folder not in result.files_created
 
 
+def test_wheel_record_lines_in_deterministic_order(script, data):
+    to_install = data.packages.join("simplewheel-1.0-py2.py3-none-any.whl")
+    result = script.pip('install', to_install)
+
+    dist_info_folder = script.site_packages / 'simplewheel-1.0.dist-info'
+    record_path = dist_info_folder / 'RECORD'
+
+    assert dist_info_folder in result.files_created, str(result)
+    assert record_path in result.files_created, str(result)
+
+    record_path = result.files_created[record_path].full
+    record_lines = [
+        p for p in Path(record_path).read_text().split('\n') if p
+    ]
+    assert record_lines == sorted(record_lines)
+
+
 @pytest.mark.network
 def test_install_user_wheel(script, virtualenv, data, common_wheels):
     """