filter-repo: implement --paths-from-file

This allows the user to put a whole bunch of paths they want to keep (or
want to remove) in a file and then just provide the path to it.  They
can also use globs or regexes (similar to --replace-text) and can also
do renames.  In fact, this allows regex renames, despite the fact that I
never added a --path-rename-regex option.

Signed-off-by: Elijah Newren <newren@gmail.com>

git-filter-repo

@@ -1918,6 +1918,12 @@ class FilteringOptions(object):
         raise SystemExit(_("Error: HelperFilter given invalid option_string: %s")
                          % option_string) # pragma: no cover
 
+  class FileWithPathsFilter(argparse.Action):
+    def __call__(self, parser, namespace, values, option_string=None):
+      if not namespace.path_changes:
+        namespace.path_changes = []
+      namespace.path_changes += FilteringOptions.get_paths_from_file(values)
+
   @staticmethod
   def create_arg_parser():
     # Include usage in the summary, so we can put the description first
@@ -1996,6 +2002,13 @@ class FilteringOptions(object):
                "specified."))
 
     helpers = parser.add_argument_group(title=_("Path shortcuts"))
+    helpers.add_argument('--paths-from-file', metavar='FILENAME',
+        type=os.fsencode,
+        action=FilteringOptions.FileWithPathsFilter, dest='path_changes',
+        help=_("Specify several path filtering and renaming directives, one "
+               "per line.  Lines with '==>' in them specify path renames, "
+               "and lines can begin with 'literal:' (the default), 'glob:', "
+               "or 'regex: ' to specify different matching styles"))
     helpers.add_argument('--subdirectory-filter', metavar='DIRECTORY',
         action=FilteringOptions.HelperFilter, type=os.fsencode,
         help=_("Only look at history that touches the given subdirectory "
@@ -2207,6 +2220,50 @@ class FilteringOptions(object):
           replace_literals.append((line, replacement))
     return {'literals': replace_literals, 'regexes':  replace_regexes}
 
+  @staticmethod
+  def get_paths_from_file(filename):
+    new_path_changes = []
+    with open(filename, 'br') as f:
+      for line in f:
+        line = line.rstrip(b'\r\n')
+
+        # Skip blank lines
+        if not line:
+          continue
+
+        # Determine the replacement
+        match_type, repl = 'literal', None
+        if b'==>' in line:
+          line, repl = line.rsplit(b'==>', 1)
+
+        # See if we need to match via regex
+        match_type = 'match' # a.k.a. 'literal'
+        if line.startswith(b'regex:'):
+          match_type = 'regex'
+          match = re.compile(line[6:])
+        elif line.startswith(b'glob:'):
+          match_type = 'glob'
+          match = line[5:]
+          if repl:
+            raise SystemExit(_("Error: In %s, 'glob:' and '==>' are incompatible (renaming globs makes no sense)" % decode(filename)))
+        else:
+          if line.startswith(b'literal:'):
+            match = line[8:]
+          else:
+            match = line
+          if repl is not None:
+            if match and repl and match.endswith(b'/') != repl.endswith(b'/'):
+              raise SystemExit(_("Error: When rename directories, if OLDNAME "
+                                 "and NEW_NAME are both non-empty and either "
+                                 "ends with a slash then both must."))
+
+        # Record the filter or rename
+        if repl is not None:
+          new_path_changes.append(['rename', match_type, (match, repl)])
+        else:
+          new_path_changes.append(['filter', match_type, match])
+      return new_path_changes
+
   @staticmethod
   def default_options():
     return FilteringOptions.parse_args([], error_on_empty = False)
@@ -2948,9 +3005,11 @@ class RepoFilter(object):
             wanted = True
         elif mod_type == 'rename':
           match, repl = path_exp
-          assert match_type in ('match',)
+          assert match_type in ('match','regex') # glob was translated to regex
           if match_type == 'match' and filename_matches(match, full_pathname):
             full_pathname = full_pathname.replace(match, repl, 1)
+          if match_type == 'regex':
+            full_pathname = match.sub(repl, full_pathname)
       return full_pathname if (wanted == filtering_is_inclusive) else None
 
     # Change the commit message according to callback

t/t9390-filter-repo.sh

@@ -145,6 +145,36 @@ test_expect_success '--path-rename inability to squash' '
 	)
 '
 
+test_expect_success '--paths-from-file' '
+	(
+		git clone file://"$(pwd)"/path_rename paths_from_file &&
+		cd paths_from_file &&
+
+		cat >../path_changes <<-EOF &&
+		literal:values/huge
+		values/huge==>values/gargantuan
+		glob:*rge
+
+		regex:.*med.*
+		regex:^([^/]*)/(.*)ge$==>\2/\1/ge
+		EOF
+
+		git filter-repo --paths-from-file ../path_changes &&
+		git log --format=%n --name-only | sort | uniq >filenames &&
+		# intermediate, medium, two larges, gargantuan, and a blank line
+		test_line_count = 6 filenames &&
+		! grep sequences/tiny filenames &&
+		grep sequences/intermediate filenames &&
+		grep lar/sequences/ge filenames &&
+		grep lar/values/ge filenames &&
+		grep values/gargantuan filenames &&
+		! grep sequences/small filenames &&
+		grep sequences/medium filenames &&
+
+		rm ../path_changes
+	)
+'
+
 test_expect_success 'setup metasyntactic repo' '
 	test_create_repo metasyntactic &&
 	(
@@ -861,7 +891,13 @@ test_expect_success 'other startup error cases and requests for help' '
 		test_i18ngrep ": --use-base-name and --path-rename are incompatible" err &&
 
 		test_must_fail git filter-repo --path-rename foo:bar/ 2>err &&
-		test_i18ngrep "either ends with a slash then both must." err
+		test_i18ngrep "either ends with a slash then both must." err &&
+
+		test_must_fail git filter-repo --paths-from-file <(echo "foo==>bar/") 2>err &&
+		test_i18ngrep "either ends with a slash then both must." err &&
+
+		test_must_fail git filter-repo --paths-from-file <(echo "glob:*.py==>newname") 2>err &&
+		test_i18ngrep "renaming globs makes no sense" err
 	)
 '