Allow and warn on RECORD lines with more than three elements.

Author: cjerdonek

news/6165.bugfix

@@ -0,0 +1 @@
+Allow ``RECORD`` lines with more than three elements, and display a warning.

src/pip/_internal/wheel.py

@@ -262,17 +262,24 @@ def sorted_outrows(outrows):
 def get_csv_rows_for_installed(
     old_csv_rows,  # type: Iterable[List[str]]
     installed,  # type: Dict[str, str]
-    changed,  # set
+    changed,  # type: set
     generated,  # type: List[str]
-    lib_dir,
+    lib_dir,  # type: str
 ):
     # type: (...) -> List[InstalledCSVRow]
     installed_rows = []  # type: List[InstalledCSVRow]
-    for fpath, digest, length in old_csv_rows:
+    for row in old_csv_rows:
+        if len(row) > 3:
+            logger.warning(
+                'RECORD line has more than three elements: {}'.format(row)
+            )
+        fpath = row[0]
         fpath = installed.pop(fpath, fpath)
         if fpath in changed:
             digest, length = rehash(fpath)
-        installed_rows.append((fpath, digest, str(length)))
+            row[1] = digest
+            row[2] = length
+        installed_rows.append(tuple(row))
     for f in generated:
         digest, length = rehash(f)
         installed_rows.append((normpath(f, lib_dir), digest, str(length)))

tests/unit/test_wheel.py

@@ -96,7 +96,7 @@ def call_get_csv_rows_for_installed(tmpdir, text):
     return outrows
 
 
-def test_get_csv_rows_for_installed(tmpdir):
+def test_get_csv_rows_for_installed(tmpdir, caplog):
     text = textwrap.dedent("""\
     a,b,c
     d,e,f
@@ -108,15 +108,32 @@ def test_get_csv_rows_for_installed(tmpdir):
         ('d', 'e', 'f'),
     ]
     assert outrows == expected
+    # Check there were no warnings.
+    assert len(caplog.records) == 0
 
 
-def test_get_csv_rows_for_installed__long_lines(tmpdir):
+def test_get_csv_rows_for_installed__long_lines(tmpdir, caplog):
     text = textwrap.dedent("""\
     a,b,c,d
-    e,f,g,h
+    e,f,g
+    h,i,j,k
     """)
     outrows = call_get_csv_rows_for_installed(tmpdir, text)
 
+    expected = [
+        ('a', 'b', 'c', 'd'),
+        ('e', 'f', 'g'),
+        ('h', 'i', 'j', 'k'),
+    ]
+    assert outrows == expected
+
+    messages = [rec.message for rec in caplog.records]
+    expected = [
+        "RECORD line has more than three elements: ['a', 'b', 'c', 'd']",
+        "RECORD line has more than three elements: ['h', 'i', 'j', 'k']"
+    ]
+    assert messages == expected
+
 
 def test_wheel_version(tmpdir, data):
     future_wheel = 'futurewheel-1.9-py2.py3-none-any.whl'