Merge pull request #1 from ianhi/non-ascii-z

Non ascii z

Author: ianhi

jupyterlab_git/git.py

@@ -5,7 +5,6 @@
 import re
 import subprocess
 from urllib.parse import unquote
-from codecs import decode, escape_decode
 
 import pexpect
 import tornado
@@ -169,14 +168,14 @@ async def changed_files(self, base=None, remote=None, single_commit=None):
             }
         """
         if single_commit:
-            cmd = ["git", "diff", "{}^!".format(single_commit), "--name-only"]
+            cmd = ["git", "diff", "{}^!".format(single_commit), "--name-only", "-z"]
         elif base and remote:
             if base == "WORKING":
-                cmd = ["git", "diff", remote, "--name-only"]
+                cmd = ["git", "diff", remote, "--name-only", "-z"]
             elif base == "INDEX":
-                cmd = ["git", "diff", "--staged", remote, "--name-only"]
+                cmd = ["git", "diff", "--staged", remote, "--name-only", "-z"]
             else:
-                cmd = ["git", "diff", base, remote, "--name-only"]
+                cmd = ["git", "diff", base, remote, "--name-only", "-z"]
         else:
             raise tornado.web.HTTPError(
                 400, "Either single_commit or (base and remote) must be provided"
@@ -194,7 +193,7 @@ async def changed_files(self, base=None, remote=None, single_commit=None):
                 response["command"] = " ".join(cmd)
                 response["message"] = error
             else:
-                response["files"] = output.strip().split("\n")
+                response["files"] = output.strip("\x00").split("\x00")
 
         return response
 
@@ -237,7 +236,7 @@ async def status(self, current_path):
         """
         Execute git status command & return the result.
         """
-        cmd = ["git", "status", "--porcelain", "-u"]
+        cmd = ["git", "status", "--porcelain", "-u", "-z"]
         code, my_output, my_error = await execute(
             cmd, cwd=os.path.join(self.root_dir, current_path),
         )
@@ -250,22 +249,24 @@ async def status(self, current_path):
             }
 
         result = []
-        line_array = my_output.splitlines()
-        for line in line_array:
-            to1 = None
-            from_path = line[3:]
-            if line[0] == "R":
-                to0 = line[3:].split(" -> ")
-                to1 = to0[len(to0) - 1]
+        line_iterable = iter(my_output.strip("\x00").split('\x00'))
+        result = []
+        for line in line_iterable:
+            x = line[0]
+            y = line[1]
+            if line[0]=='R':
+                #If file was renamed then we need both this line
+                #and the next line, then we want to move onto the subsequent
+                #line. We can accomplish this by calling next on the iterable
+                to = line[3:]
+                from_path = next(line_iterable)
             else:
-                to1 = line[3:]
-            if to1.startswith('"'):
-                to1 = to1[1:]
-            if to1.endswith('"'):
-                to1 = to1[:-1]
-            to1 = decode(escape_decode(to1)[0],'utf-8')
-            from_path = decode(escape_decode(from_path)[0],'utf-8')
-            result.append({"x": line[0], "y": line[1], "to": to1, "from": from_path})
+                #to and from_path are the same
+                from_path = line[3:]
+                to = line[3:]
+            result.append({"x": x, "y": y, "to": to, "from": from_path})
+
+
         return {"code": code, "files": result}
 
     async def log(self, current_path, history_count=10):
@@ -314,10 +315,10 @@ async def log(self, current_path, history_count=10):
 
     async def detailed_log(self, selected_hash, current_path):
         """
-        Execute git log -1 --stat --numstat --oneline command (used to get
+        Execute git log -1 --stat --numstat --oneline -z command (used to get
         insertions & deletions per file) & return the result.
         """
-        cmd = ["git", "log", "-1", "--stat", "--numstat", "--oneline", selected_hash]
+        cmd = ["git", "log", "-1", "--stat", "--numstat", "--oneline", "-z", selected_hash]
         code, my_output, my_error = await execute(
             cmd, cwd=os.path.join(self.root_dir, current_path),
         )
@@ -329,7 +330,7 @@ async def detailed_log(self, selected_hash, current_path):
         note = [0] * 3
         count = 0
         temp = ""
-        line_array = my_output.splitlines()
+        line_array = re.split("\x00|\n|\r\n|\r", my_output)
         length = len(line_array)
         INSERTION_INDEX = 0
         DELETION_INDEX = 1
@@ -342,7 +343,7 @@ async def detailed_log(self, selected_hash, current_path):
                     note[count] = words[i]
                     count += 1
             for num in range(1, int(length / 2)):
-                line_info = line_array[num].split(maxsplit=2)
+                line_info = line_array[num].split('\t', maxsplit=2)
                 words = line_info[2].split("/")
                 length = len(words)
                 result.append(
@@ -373,14 +374,14 @@ async def diff(self, top_repo_path):
         """
         Execute git diff command & return the result.
         """
-        cmd = ["git", "diff", "--numstat"]
+        cmd = ["git", "diff", "--numstat", "-z"]
         code, my_output, my_error = await execute(cmd, cwd=top_repo_path)
 
         if code != 0:
             return {"code": code, "command": " ".join(cmd), "message": my_error}
 
         result = []
-        line_array = my_output.splitlines()
+        line_array = my_output.strip('\x00').split('\x00')
         for line in line_array:
             linesplit = line.split()
             result.append(

jupyterlab_git/tests/test_detailed_log.py

@@ -15,28 +15,35 @@
 async def test_detailed_log():
     with patch("jupyterlab_git.git.execute") as mock_execute:
         # Given
-        process_output = [
+        process_output_first_half = [
             "f29660a (HEAD, origin/feature) Commit message",
-            "10      3       notebook_without_spaces.ipynb",
-            "11      4       Notebook with spaces.ipynb",
-            "12      5       path/notebook_without_spaces.ipynb",
-            "13      6       path/Notebook with spaces.ipynb",
+            "10\t3\tnotebook_without_spaces.ipynb",
+            "11\t4\tNotebook with spaces.ipynb",
+            "12\t5\tpath/notebook_without_spaces.ipynb",
+            "13\t6\tpath/Notebook with spaces.ipynb",
+            "14\t1\tpath/Notebook with λ.ipynb",
+        ]
+        process_output_second_half = [
             " notebook_without_spaces.ipynb      | 13 ++++++++---",
             " Notebook with spaces.ipynb         | 15 +++++++++----",
             " path/notebook_without_spaces.ipynb | 17 ++++++++++-----",
             " path/Notebook with spaces.ipynb    | 19 +++++++++++------",
-            " 4 files changed, 46 insertions(+), 18 deletions(-)",
+            " path/Notebook with \\316\\273.ipynb         | 15 +++++++++++-",
+            " 5 files changed, 50 insertions(+), 19 deletions(-)",
         ]
-        mock_execute.return_value = tornado.gen.maybe_future(
-            (0, "\n".join(process_output), "")
+        process_output_first_half = "\x00".join(process_output_first_half)
+        process_output_second_half = "\n".join(process_output_second_half)
+        process_output = process_output_first_half + "\x00" + process_output_second_half
+        mock_execute._mock_return_value = tornado.gen.maybe_future(
+            (0, process_output, "")
         )
 
         expected_response = {
             "code": 0,
-            "modified_file_note": " 4 files changed, 46 insertions(+), 18 deletions(-)",
-            "modified_files_count": "4",
-            "number_of_insertions": "46",
-            "number_of_deletions": "18",
+            "modified_file_note": " 5 files changed, 50 insertions(+), 19 deletions(-)",
+            "modified_files_count": "5",
+            "number_of_insertions": "50",
+            "number_of_deletions": "19",
             "modified_files": [
                 {
                     "modified_file_path": "notebook_without_spaces.ipynb",
@@ -62,6 +69,12 @@ async def test_detailed_log():
                     "insertion": "13",
                     "deletion": "6",
                 },
+                {
+                    "modified_file_path": "path/Notebook with λ.ipynb",
+                    "modified_file_name": "Notebook with λ.ipynb",
+                    "insertion": "14",
+                    "deletion": "1",
+                },
             ],
         }
 
@@ -83,6 +96,7 @@ async def test_detailed_log():
                 "--stat",
                 "--numstat",
                 "--oneline",
+                "-z",
                 "f29660a2472e24164906af8653babeb48e4bf2ab",
             ],
             cwd=os.path.join("/bin", "test_curr_path"),

jupyterlab_git/tests/test_diff.py

@@ -24,7 +24,7 @@ async def test_changed_files_single_commit():
     with patch("jupyterlab_git.git.execute") as mock_execute:
         # Given
         mock_execute.return_value = tornado.gen.maybe_future(
-            (0, "file1.ipynb\nfile2.py", "")
+            (0, "file1.ipynb\x00file2.py", "")
         )
 
         # When
@@ -39,6 +39,7 @@ async def test_changed_files_single_commit():
                 "diff",
                 "64950a634cd11d1a01ddfedaeffed67b531cb11e^!",
                 "--name-only",
+                "-z",
             ],
             cwd="/bin",
         )
@@ -50,7 +51,7 @@ async def test_changed_files_working_tree():
     with patch("jupyterlab_git.git.execute") as mock_execute:
         # Given
         mock_execute.return_value = tornado.gen.maybe_future(
-            (0, "file1.ipynb\nfile2.py", "")
+            (0, "file1.ipynb\x00file2.py", "")
         )
 
         # When
@@ -60,7 +61,7 @@ async def test_changed_files_working_tree():
 
         # Then
         mock_execute.assert_called_once_with(
-            ["git", "diff", "HEAD", "--name-only"], cwd="/bin"
+            ["git", "diff", "HEAD", "--name-only", "-z"], cwd="/bin"
         )
         assert {"code": 0, "files": ["file1.ipynb", "file2.py"]} == actual_response
 
@@ -70,7 +71,7 @@ async def test_changed_files_index():
     with patch("jupyterlab_git.git.execute") as mock_execute:
         # Given
         mock_execute.return_value = tornado.gen.maybe_future(
-            (0, "file1.ipynb\nfile2.py", "")
+            (0, "file1.ipynb\x00file2.py", "")
         )
 
         # When
@@ -80,7 +81,7 @@ async def test_changed_files_index():
 
         # Then
         mock_execute.assert_called_once_with(
-            ["git", "diff", "--staged", "HEAD", "--name-only"], cwd="/bin"
+            ["git", "diff", "--staged", "HEAD", "--name-only", "-z"], cwd="/bin"
         )
         assert {"code": 0, "files": ["file1.ipynb", "file2.py"]} == actual_response
 
@@ -90,7 +91,7 @@ async def test_changed_files_two_commits():
     with patch("jupyterlab_git.git.execute") as mock_execute:
         # Given
         mock_execute.return_value = tornado.gen.maybe_future(
-            (0, "file1.ipynb\nfile2.py", "")
+            (0, "file1.ipynb\x00file2.py", "")
         )
 
         # When
@@ -100,7 +101,7 @@ async def test_changed_files_two_commits():
 
         # Then
         mock_execute.assert_called_once_with(
-            ["git", "diff", "HEAD", "origin/HEAD", "--name-only"], cwd="/bin"
+            ["git", "diff", "HEAD", "origin/HEAD", "--name-only", "-z"], cwd="/bin"
         )
         assert {"code": 0, "files": ["file1.ipynb", "file2.py"]} == actual_response
 
@@ -118,6 +119,6 @@ async def test_changed_files_git_diff_error():
 
         # Then
         mock_execute.assert_called_once_with(
-            ["git", "diff", "HEAD", "origin/HEAD", "--name-only"], cwd="/bin"
+            ["git", "diff", "HEAD", "origin/HEAD", "--name-only", "-z"], cwd="/bin"
         )
         assert {"code": 128, "message": "error message"} == actual_response

jupyterlab_git/tests/test_status.py

@@ -0,0 +1,46 @@
+# python lib
+import os
+from unittest.mock import Mock, call, patch
+
+import pytest
+import tornado
+
+# local lib
+from jupyterlab_git.git import Git
+
+from .testutils import FakeContentManager
+
+@pytest.mark.asyncio
+async def test_status():
+    with patch("jupyterlab_git.git.execute") as mock_execute:
+        # Given
+        process_output = (
+        "A  notebook with spaces.ipynb",
+        "M  notebook with λ.ipynb",
+        "R  renamed_to_θ.py",
+        "originally_named_π.py",
+        "?? untracked.ipynb",
+        )
+
+        expected_resonse = [
+            {"x": "A", "y": " ", "to": "notebook with spaces.ipynb", "from": "notebook with spaces.ipynb"},
+            {"x": "M", "y": " ", "to": "notebook with λ.ipynb", "from": "notebook with λ.ipynb"},
+            {"x": "R", "y": " ", "to": "renamed_to_θ.py", "from": "originally_named_π.py"},
+            {"x": "?", "y": "?", "to": "untracked.ipynb", "from": "untracked.ipynb"},
+        ]
+        mock_execute.return_value = tornado.gen.maybe_future(
+            (0, "\x00".join(process_output), "")
+
+        )
+
+        # When
+        actual_response = await Git(FakeContentManager("/bin")).status(
+            current_path="test_curr_path"
+        )
+
+        # Then
+        mock_execute.assert_called_once_with(
+            ["git", "status", "--porcelain" , "-u", "-z"], cwd="/bin/test_curr_path"
+        )
+
+        assert {"code": 0, "files": expected_resonse} == actual_response