fix(cli): fix renku rerun/update with unicode paths

Author: Panaetius

renku/core/commands/cwl_runner.py

@@ -18,8 +18,10 @@
 """Wrap CWL runner."""
 
 import os
+import re
 import shutil
 import sys
+from urllib.parse import unquote
 
 import click
 
@@ -32,11 +34,15 @@ def execute(client, output_file, output_paths=None):
     """Run the generated workflow using cwltool library."""
     output_paths = output_paths or set()
 
+    import cwltool.command_line_tool
     import cwltool.factory
     from cwltool import workflow
     from cwltool.context import LoadingContext, RuntimeContext
     from cwltool.utils import visit_class
 
+    # NOTE: cwltool does not support unicode characters, patch it in (plus emojis 😁)
+    cwltool.command_line_tool.ACCEPTLIST_RE = re.compile("^[\w\-. \u2600-\u26FF]+$")
+
     def construct_tool_object(toolpath_object, *args, **kwargs):
         """Fix missing locations."""
         protocol = "file://"
@@ -71,8 +77,8 @@ def addLocation(d):
 
     def remove_prefix(location, prefix="file://"):
         if location.startswith(prefix):
-            return location[len(prefix) :]
-        return location
+            return unquote(location[len(prefix) :])
+        return unquote(location)
 
     locations = {remove_prefix(output["location"]) for output in outputs.values()}
     # make sure to not move an output if it's containing directory gets moved

renku/core/management/repository.py

@@ -42,6 +42,7 @@
 from renku.core.models.refs import LinkReference
 from renku.core.models.workflow.dependency_graph import DependencyGraph
 from renku.core.utils.migrate import MigrationType
+from renku.core.utils.scm import git_unicode_unescape
 
 from .git import GitCore
 
@@ -319,7 +320,8 @@ def process_commit(self, commit=None, path=None):
 
         if not path:
             # search for activities a file could have been a part of
-            activities = self.activities_for_paths(commit.stats.files.keys(), file_commit=commit, revision="HEAD")
+            paths = [git_unicode_unescape(p) for p in commit.stats.files.keys()]
+            activities = self.activities_for_paths(paths, file_commit=commit, revision="HEAD")
             if len(activities) > 1:
                 raise errors.CommitProcessingError(
                     "Found multiple activities that produced the same entity at commit {}".format(commit)

tests/cli/test_rerun.py

@@ -30,13 +30,25 @@
 from renku.cli import cli
 
 
-def test_simple_rerun(runner, project, run, no_lfs_warning):
+@pytest.mark.parametrize(
+    "source,selected",
+    [
+        ("coffee-orders-☕-by-location test.csv", "😍works.txt"),
+        ("source.txt", "selected.txt"),
+        ("test-愛", "成功"),
+        ("그래프", "성공"),
+        ("يحاول", "نجاح.txt"),
+        ("график.txt", "успех.txt"),
+        ("𒁃.c", "𒁏.txt"),
+    ],
+)
+def test_simple_rerun(runner, project, run, no_lfs_warning, source, selected):
     """Test simple file recreation."""
     greetings = {"hello", "hola", "ahoj"}
 
     cwd = Path(project)
-    source = cwd / "source.txt"
-    selected = cwd / "selected.txt"
+    source = cwd / source
+    selected = cwd / selected
 
     repo = git.Repo(project)