Fixed len calculation when taring script, unicode issues

arca/backend/docker.py

@@ -447,9 +447,11 @@ def tar_script(self, name, script):
         tar = tarfile.TarFile(fileobj=tarstream, mode='w')
         tarinfo = tarfile.TarInfo(name=name)
 
+        script = script.encode("utf-8")
+
         tarinfo.size = len(script)
         tarinfo.mtime = time.time()
-        tar.addfile(tarinfo, BytesIO(script.encode("utf-8")))
+        tar.addfile(tarinfo, BytesIO(script))
         tar.close()
 
         return tarstream.getvalue()

arca/task.py

@@ -3,6 +3,7 @@
 import re
 
 from pathlib import Path
+from pprint import pformat
 from typing import Optional, Tuple, Any, Dict, Iterable
 
 from .exceptions import TaskMisconfigured
@@ -28,27 +29,27 @@ def __repr__(self):
         return f"Task({self.function_call})"
 
     def build_imports(self):
-        return "\n".join([f"    import {x}" for x in self.imports])
+        return "\r\n".join([f"    import {x}" for x in self.imports])
 
     def build_from_imports(self):
-        return "\n".join([f"    from {x[0]} import {x[1]}" for x in self.from_imports])
+        return "\r\n".join([f"    from {x[0]} import {x[1]}" for x in self.from_imports])
 
     def build_function_call(self):
         if len(self.args) and len(self.kwargs):
             return "{}(*{}, **{})".format(
                 self.function_call,
-                self.args,
-                self.kwargs
+                pformat(self.args),
+                pformat(self.kwargs)
             )
         elif len(self.args):
             return "{}(*{})".format(
                 self.function_call,
-                self.args
+                pformat(self.args)
             )
         elif len(self.kwargs):
             return "{}(**{})".format(
                 self.function_call,
-                self.kwargs
+                pformat(self.kwargs)
             )
         else:
             return f"{self.function_call}()"
@@ -57,19 +58,20 @@ def build_script(self, venv_path: Path=None) -> str:
         result = ""
 
         if venv_path is not None:
-            result += "#!" + str(venv_path.resolve() / "bin" / "python3") + "\n\n"
+            result += "#!" + str(venv_path.resolve() / "bin" / "python3") + "\r\n\r\n"
         else:
-            result += "#!python3\n\n"
-
-        result += "import json\n"
-        result += "import traceback\n"
-        result += "import sys\n"
-        result += "import os\n"
-        result += "sys.path.insert(1, os.getcwd())\n"
-        result += "try:\n"
-
-        result += self.build_imports() + "\n\n"
-        result += self.build_from_imports() + "\n\n"
+            result += "#!python3\r\n"
+
+        result += "# encoding=utf-8\r\n\r\n"
+        result += "import json\r\n"
+        result += "import traceback\r\n"
+        result += "import sys\r\n"
+        result += "import os\r\n"
+        result += "sys.path.insert(1, os.getcwd())\r\n"
+        result += "try:\r\n"
+
+        result += self.build_imports() + "\r\n"
+        result += self.build_from_imports() + "\r\n"
 
         result += f"""
     res = {self.build_function_call()}

tests/common.py

@@ -11,9 +11,21 @@ def return_str_function():
     return "Some string"
 """
 
+TEST_UNICODE = "Nechť již hříšné saxofony ďáblů rozezvučí síň úděsnými tóny waltzu, tanga a\xa0quickstepu.→"
+
 SECOND_RETURN_STR_FUNCTION = """
 def return_str_function():
-    return "Some other string"
+    return "Nechť již hříšné saxofony ďáblů rozezvučí síň úděsnými tóny waltzu, tanga a\xa0quickstepu.→"
+"""
+
+ARG_STR_FUNCTION = """
+def return_str_function(arg):
+    return arg[::-1]
+"""
+
+KWARG_STR_FUNCTION = """
+def return_str_function(*, kwarg):
+    return kwarg[::-1]
 """
 
 RETURN_DJANGO_VERSION_FUNCTION = """

tests/test_backends.py

@@ -9,7 +9,8 @@
 from arca import Arca, VenvBackend, DockerBackend, Task, CurrentEnvironmentBackend
 from arca.exceptions import BuildError, FileOutOfRangeError
 
-from common import BASE_DIR, RETURN_DJANGO_VERSION_FUNCTION, RETURN_STR_FUNCTION, SECOND_RETURN_STR_FUNCTION
+from common import BASE_DIR, RETURN_DJANGO_VERSION_FUNCTION, RETURN_STR_FUNCTION, SECOND_RETURN_STR_FUNCTION, \
+    TEST_UNICODE, ARG_STR_FUNCTION, KWARG_STR_FUNCTION
 
 
 @pytest.mark.parametrize(
@@ -53,13 +54,14 @@ def test_backends(backend, requirements_location, file_location):
     filepath.write_text(RETURN_STR_FUNCTION)
     repo.index.add([str(filepath)])
     repo.index.commit("Initial")
+    repo_url = f"file://{git_dir}"
 
     task = Task(
         "return_str_function",
         from_imports=[("test_file", "return_str_function")]
     )
 
-    result = arca.run(f"file://{git_dir}", "master", task)
+    result = arca.run(repo_url, "master", task)
 
     assert result.output == "Some string"
 
@@ -71,11 +73,11 @@ def test_backends(backend, requirements_location, file_location):
     repo.index.add([str(filepath)])
     repo.index.commit("Updated function")
 
-    result = arca.run(f"file://{git_dir}", "master", task)
-    assert result.output == "Some other string"
+    result = arca.run(repo_url, "master", task)
+    assert result.output == TEST_UNICODE
 
     # in the other branch there's still the original
-    result = arca.run(f"file://{git_dir}", "new_branch", task)
+    result = arca.run(repo_url, "new_branch", task)
     assert result.output == "Some string"
 
     repo.branches.master.checkout()
@@ -94,7 +96,7 @@ def test_backends(backend, requirements_location, file_location):
     repo.index.add([str(requirements_path)])
     repo.index.commit("Added requirements, changed to version")
 
-    result = arca.run(f"file://{git_dir}", "master", task)
+    result = arca.run(repo_url, "master", task)
 
     assert result.output == "1.11.4"
 
@@ -111,15 +113,15 @@ def test_backends(backend, requirements_location, file_location):
     repo.index.add([str(requirements_path)])
     repo.index.commit("Updated requirements")
 
-    result = arca.run(f"file://{git_dir}", "master", task)
+    result = arca.run(repo_url, "master", task)
     assert result.output == "1.11.5"
 
     django_task = Task(
         "django.get_version",
         imports=["django"]
     )
 
-    result = arca.run(f"file://{git_dir}", "master", django_task)
+    result = arca.run(repo_url, "master", django_task)
     assert result.output == "1.11.5"
 
     django_task_error = Task(
@@ -132,6 +134,26 @@ def test_backends(backend, requirements_location, file_location):
     if isinstance(backend, CurrentEnvironmentBackend):
         backend._uninstall("django")
 
+    filepath.write_text(ARG_STR_FUNCTION)
+    repo.index.add([str(filepath)])
+    repo.index.commit("Argument function")
+
+    assert arca.run(repo_url, "master", Task(
+        "return_str_function",
+        from_imports=[("test_file", "return_str_function")],
+        args=[TEST_UNICODE]
+    )).output == TEST_UNICODE[::-1]
+
+    filepath.write_text(KWARG_STR_FUNCTION)
+    repo.index.add([str(filepath)])
+    repo.index.commit("Keyword argument function")
+
+    assert arca.run(repo_url, "master", Task(
+        "return_str_function",
+        from_imports=[("test_file", "return_str_function")],
+        kwargs={"kwarg": TEST_UNICODE}
+    )).output == TEST_UNICODE[::-1]
+
 
 @pytest.mark.parametrize(
     "backend,file_location", list(itertools.product(

tests/test_single_pull.py

@@ -6,7 +6,7 @@
 from git import Repo
 
 from arca import Arca, DockerBackend, Task, VenvBackend, CurrentEnvironmentBackend
-from common import RETURN_STR_FUNCTION, SECOND_RETURN_STR_FUNCTION, BASE_DIR
+from common import RETURN_STR_FUNCTION, SECOND_RETURN_STR_FUNCTION, BASE_DIR, TEST_UNICODE
 
 
 @pytest.mark.parametrize(
@@ -59,7 +59,7 @@ def test_single_pull(mocker, backend):
     arca.pull_again(repo_url, "master")
 
     result = arca.run(repo_url, "master", task)
-    assert result.output == "Some other string"
+    assert result.output == TEST_UNICODE
     assert arca._pull.call_count == 2
 
 
@@ -111,9 +111,9 @@ def test_pull_efficiency(mocker, backend):
     repo.index.commit("Updated function")
 
     result = arca.run(repo_url, "master", task)
-    assert result.output == "Some other string"
+    assert result.output == TEST_UNICODE
     assert arca._pull.call_count == 3
 
     result = arca.run(repo_url, "master", task)
-    assert result.output == "Some other string"
+    assert result.output == TEST_UNICODE
     assert arca._pull.call_count == 4

tests/test_task.py

@@ -9,7 +9,7 @@
 
 
 def _clean_lines(x: str) -> str:
-    return "\n".join([line for line in x.split("\n") if line.strip()])
+    return "\n".join([line.replace("\r", "") for line in x.split("\n") if line.replace("\r", "").strip()])
 
 
 @pytest.mark.parametrize(["imports", "res"], (
@@ -23,6 +23,7 @@ def _clean_lines(x: str) -> str:
 def test_imports(imports, res):
     task = Task("func", imports=imports)
     assert _clean_lines(task.build_script(ENV_PATH)) == _clean_lines("""#!/usr/bin/python3
+# encoding=utf-8
 import json
 import traceback
 import sys
@@ -48,6 +49,7 @@ def test_imports(imports, res):
 def test_from_imports(from_imports, res):
     task = Task("func", from_imports=from_imports)
     assert _clean_lines(task.build_script(ENV_PATH)) == _clean_lines("""#!/usr/bin/python3
+# encoding=utf-8
 import json
 import traceback
 import sys
@@ -73,6 +75,7 @@ def test_from_imports(from_imports, res):
 def test_function_call(args, kwargs, res):
     task = Task("func", args=args, kwargs=kwargs)
     assert _clean_lines(task.build_script(ENV_PATH)) == _clean_lines("""#!/usr/bin/python3
+# encoding=utf-8
 import json
 import traceback
 import sys