From da9aa40c433295f459e5613d306574211dd6c249 Mon Sep 17 00:00:00 2001
From: Durham Goode <durham@fb.com>
Date: Tue, 17 Mar 2020 09:11:11 -0700
Subject: [PATCH] unicode: skip binary filenames coming from Eden

Summary:
Mercurial can't handle these. Let's skip them instead of throwing an
exception.

Reviewed By: quark-zju

Differential Revision: D20016007

fbshipit-source-id: dbf37b9e315d29bcfa1e12f51bc96ed543ff6d84
---
 eden/integration/hg/lib/hg_extension_test_base.py |  3 ++-
 eden/integration/lib/hgrepo.py                    | 11 +++++++++--
 eden/scm/edenscm/mercurial/eden_dirstate_fs.py    |  5 +++++
 3 files changed, 16 insertions(+), 3 deletions(-)

diff --git a/eden/integration/hg/lib/hg_extension_test_base.py b/eden/integration/hg/lib/hg_extension_test_base.py
index b8a850c75a..83a32a4abb 100644
--- a/eden/integration/hg/lib/hg_extension_test_base.py
+++ b/eden/integration/hg/lib/hg_extension_test_base.py
@@ -195,6 +195,7 @@ class EdenHgTestCase(testcase.EdenTestCase, metaclass=abc.ABCMeta):
         msg: Optional[str] = None,
         op: Optional[str] = None,
         check_ignored: bool = True,
+        rev: Optional[str] = None,
     ) -> None:
         """Asserts the output of `hg status` matches the expected state.
 
@@ -204,7 +205,7 @@ class EdenHgTestCase(testcase.EdenTestCase, metaclass=abc.ABCMeta):
 
         'C' is not currently supported.
         """
-        actual_status = self.repo.status(include_ignored=check_ignored)
+        actual_status = self.repo.status(include_ignored=check_ignored, rev=rev)
         self.assertDictEqual(expected, actual_status, msg=msg)
         self.assert_unfinished_operation(op)
 
diff --git a/eden/integration/lib/hgrepo.py b/eden/integration/lib/hgrepo.py
index 7442d6c1fb..7efc4c8863 100644
--- a/eden/integration/lib/hgrepo.py
+++ b/eden/integration/lib/hgrepo.py
@@ -132,7 +132,9 @@ class HgRepository(repobase.Repository):
             cwd=cwd,
             check=check,
         )
-        return typing.cast(str, completed_process.stdout.decode(encoding))
+        return typing.cast(
+            str, completed_process.stdout.decode(encoding, errors="replace")
+        )
 
     def init(self, hgrc: Optional[configparser.ConfigParser] = None) -> None:
         """
@@ -259,7 +261,9 @@ class HgRepository(repobase.Repository):
         json_data = json.loads(output)
         return typing.cast(List[Dict[str, Any]], json_data)
 
-    def status(self, include_ignored: bool = False) -> Dict[str, str]:
+    def status(
+        self, include_ignored: bool = False, rev: Optional[str] = None
+    ) -> Dict[str, str]:
         """Returns the output of `hg status` as a dictionary of {path: status char}.
 
         The status characters are the same as the ones documented by `hg help status`
@@ -267,6 +271,9 @@ class HgRepository(repobase.Repository):
         args = ["status", "--print0"]
         if include_ignored:
             args.append("-mardui")
+        if rev is not None:
+            args.append("--rev")
+            args.append(rev)
 
         output = self.hg(*args)
         status = {}
diff --git a/eden/scm/edenscm/mercurial/eden_dirstate_fs.py b/eden/scm/edenscm/mercurial/eden_dirstate_fs.py
index bdcae67edd..d7a91eefcc 100644
--- a/eden/scm/edenscm/mercurial/eden_dirstate_fs.py
+++ b/eden/scm/edenscm/mercurial/eden_dirstate_fs.py
@@ -9,6 +9,7 @@ from typing import Callable, Iterable, Optional, Tuple
 
 from . import filesystem, perftrace, pycompat, util
 from .EdenThriftClient import ScmFileStatus
+from .i18n import _
 from .pycompat import decodeutf8
 
 
@@ -30,6 +31,10 @@ class eden_filesystem(filesystem.physicalfilesystem):
         IGNORED = ScmFileStatus.IGNORED
 
         for path, code in pycompat.iteritems(edenstatus):
+            if not util.isvalidutf8(path):
+                self.ui.warn(_("skipping invalid utf-8 filename: '%s'\n") % path)
+                continue
+
             path = decodeutf8(path)
             if not match(path):
                 continue