fix(core): git hashing handles "unusual" pathname characters

See: https://git-scm.com/docs/git-status#_pathname_format_notes_and_z

packages/workspace/src/core/hasher/git-hasher.spec.ts

@@ -1,5 +1,5 @@
 import { dirSync } from 'tmp';
-import { rmdirSync } from 'fs-extra';
+import { removeSync } from 'fs-extra';
 import { execSync } from 'child_process';
 import { getFileHashes } from './git-hasher';
 
@@ -14,7 +14,7 @@ describe('git-hasher', () => {
   });
 
   afterEach(() => {
-    rmdirSync(dir, { recursive: true });
+    removeSync(dir);
   });
 
   it('should work', () => {
@@ -112,6 +112,28 @@ describe('git-hasher', () => {
     expect([...getFileHashes(dir).keys()]).toEqual([`${dir}/moda.txt`]);
   });
 
+  it('should handle special characters in filenames', () => {
+    run(`echo AAA > "a-ū".txt`);
+    run(`echo BBB > "b-ū".txt`);
+    run(`git add .`);
+    run(`git commit -am init`);
+    expect([...getFileHashes(dir).keys()]).toEqual([
+      `${dir}/a-ū.txt`,
+      `${dir}/b-ū.txt`,
+    ]);
+
+    run(`mv a-ū.txt moda-ū.txt`);
+    run(`git add .`);
+    run(`echo modified >> moda-ū.txt`);
+    expect([...getFileHashes(dir).keys()]).toEqual([
+      `${dir}/b-ū.txt`,
+      `${dir}/moda-ū.txt`,
+    ]);
+
+    run(`rm "moda-ū.txt"`);
+    expect([...getFileHashes(dir).keys()]).toEqual([`${dir}/b-ū.txt`]);
+  });
+
   function run(command: string) {
     return execSync(command, { cwd: dir, stdio: ['pipe', 'pipe', 'pipe'] });
   }

packages/workspace/src/core/hasher/git-hasher.ts

@@ -6,7 +6,7 @@ function parseGitLsTree(output: string): Map<string, string> {
   const changes: Map<string, string> = new Map<string, string>();
   if (output) {
     const gitRegex: RegExp = /([0-9]{6})\s(blob|commit)\s([a-f0-9]{40})\s*(.*)/;
-    output.split('\n').forEach((line) => {
+    output.split('\0').forEach((line) => {
       if (line) {
         const matches: RegExpMatchArray | null = line.match(gitRegex);
         if (matches && matches[3] && matches[4]) {
@@ -27,31 +27,18 @@ function parseGitStatus(output: string): Map<string, string> {
   if (!output) {
     return changes;
   }
-  output
-    .trim()
-    .split('\n')
-    .forEach((line) => {
-      const [changeType, ...filenames] = line
-        .trim()
-        .match(/(?:[^\s"]+|"[^"]*")+/g)
-        .map((r) => (r.startsWith('"') ? r.substring(1, r.length - 1) : r))
-        .filter((r) => !!r);
 
-      if (changeType && filenames && filenames.length > 0) {
-        // the before filename we mark as deleted, so we remove it from the map
-        // changeType can be A/D/R/RM etc
-        // if it R and RM, we need to split the output into before and after
-        // the before part gets marked as deleted
-        if (changeType[0] === 'R') {
-          changes.set(filenames[0], 'D');
-          changes.set(filenames[filenames.length - 1], changeType);
-        } else if (changeType === '??') {
-          changes.set(filenames.join(' '), changeType);
-        } else {
-          changes.set(filenames[filenames.length - 1], changeType);
+  var chunks = output.split('\0');
+  for (let i = 0; i < chunks.length; i++) {
+    const chunk = chunks[i];
+    if (chunk.length) {
+      const change = chunk[0];
+      if (change === 'R') {
+        changes.set(chunks[++i], 'D');
+      }
+      changes.set(chunk.substring(2).trim(), change);
     }
   }
-    });
   return changes;
 }
 
@@ -92,7 +79,9 @@ function getGitHashForFiles(
 }
 
 function gitLsTree(path: string): Map<string, string> {
-  return parseGitLsTree(spawnProcess('git', ['ls-tree', 'HEAD', '-r'], path));
+  return parseGitLsTree(
+    spawnProcess('git', ['ls-tree', 'HEAD', '-r', '-z'], path)
+  );
 }
 
 function gitStatus(
@@ -101,7 +90,7 @@ function gitStatus(
   const deletedFiles: string[] = [];
   const filesToHash: string[] = [];
   parseGitStatus(
-    spawnProcess('git', ['status', '-s', '-u', '.'], path)
+    spawnProcess('git', ['status', '-s', '-u', '-z', '.'], path)
   ).forEach((changeType: string, filename: string) => {
     if (changeType !== 'D') {
       filesToHash.push(filename);