# HG changeset patch
# User Marcin Kuzminski <marcin@rhodecode.com>
# Date 2020-03-16 14:08:09
# Node ID dc8df3e0e1bf84df3c58a7307168f855d4455479
# Parent  58907ccac5c76a5054b422753067251f26712589

git: use non-unicode author extraction as it's returned as bytes from backend, and we can
get an unicode errors while there's some non-ascii characters.

diff --git a/vcsserver/git.py b/vcsserver/git.py
--- a/vcsserver/git.py
+++ b/vcsserver/git.py
@@ -39,7 +39,7 @@ from dulwich.repo import Repo as Dulwich
 from dulwich.server import update_server_info
 
 from vcsserver import exceptions, settings, subprocessio
-from vcsserver.utils import safe_str, safe_int
+from vcsserver.utils import safe_str, safe_int, safe_unicode
 from vcsserver.base import RepoFactory, obfuscate_qs
 from vcsserver.hgcompat import (
     hg_url as url_parser, httpbasicauthhandler, httpdigestauthhandler)
@@ -840,7 +840,11 @@ class GitRemote(RemoteBase):
                 if author.email:
                     return u"{} <{}>".format(author.name, author.email)
 
-                return u"{}".format(author.raw_name)
+                try:
+                    return u"{}".format(author.name)
+                except Exception:
+                    return u"{}".format(safe_unicode(author.raw_name))
+
         return _author(repo_id, commit_id)
 
     @reraise_safe_exceptions
diff --git a/vcsserver/utils.py b/vcsserver/utils.py
--- a/vcsserver/utils.py
+++ b/vcsserver/utils.py
@@ -37,18 +37,16 @@ def safe_int(val, default=None):
     return val
 
 
-def safe_str(unicode_, to_encoding=['utf8']):
+def safe_str(unicode_, to_encoding=None):
     """
     safe str function. Does few trick to turn unicode_ into string
 
-    In case of UnicodeEncodeError, we try to return it with encoding detected
-    by chardet library if it fails fallback to string with errors replaced
-
     :param unicode_: unicode to encode
+    :param to_encoding: encode to this type UTF8 default
     :rtype: str
     :returns: str object
     """
-
+    to_encoding = to_encoding or ['utf8']
     # if it's not basestr cast to str
     if not isinstance(unicode_, basestring):
         return str(unicode_)
@@ -65,15 +63,38 @@ def safe_str(unicode_, to_encoding=['utf
         except UnicodeEncodeError:
             pass
 
+    return unicode_.encode(to_encoding[0], 'replace')
+
+
+def safe_unicode(str_, from_encoding=None):
+    """
+    safe unicode function. Does few trick to turn str_ into unicode
+
+    :param str_: string to decode
+    :param from_encoding: encode from this type UTF8 default
+    :rtype: unicode
+    :returns: unicode object
+    """
+    from_encoding = from_encoding or ['utf8']
+
+    if isinstance(str_, unicode):
+        return str_
+
+    if not isinstance(from_encoding, (list, tuple)):
+        from_encoding = [from_encoding]
+
     try:
-        import chardet
-        encoding = chardet.detect(unicode_)['encoding']
-        if encoding is None:
-            raise UnicodeEncodeError()
+        return unicode(str_)
+    except UnicodeDecodeError:
+        pass
 
-        return unicode_.encode(encoding)
-    except (ImportError, UnicodeEncodeError):
-        return unicode_.encode(to_encoding[0], 'replace')
+    for enc in from_encoding:
+        try:
+            return unicode(str_, enc)
+        except UnicodeDecodeError:
+            pass
+
+    return unicode(str_, from_encoding[0], 'replace')
 
 
 class AttributeDict(dict):