fix(files): parse multi-file and copy git diffs, cap word diffs, readable pie labels

Author: TheodoreSpeaks

apps/sim/app/workspace/[workspaceId]/files/components/file-viewer/mermaid-theme.ts

@@ -57,9 +57,9 @@ export function readMermaidThemeVariables(
     loopTextColor: body,
     sequenceNumberColor: background,
     pie1: body,
-    pie2: token('--text-subtle'),
-    pie3: token('--surface-7'),
-    pie4: line,
+    pie2: line,
+    pie3: token('--text-subtle'),
+    pie4: token('--text-muted'),
     pieStrokeColor: background,
     pieTitleTextColor: text,
     pieSectionTextColor: background,

apps/sim/components/diff/diff-view.tsx

@@ -9,6 +9,8 @@ import type { DiffHunk, DiffLine } from '@/lib/diff/unified'
 const CONTEXT_EDGE = 3
 /** A run of unchanged lines collapses once it is longer than this. */
 const COLLAPSE_AFTER = CONTEXT_EDGE * 2 + 2
+/** Word-level comparison is skipped past this many characters, so huge lines stay cheap. */
+const WORD_DIFF_MAX_CHARS = 1000
 
 const ADD_ROW =
   'bg-[color-mix(in_srgb,var(--badge-success-bg)_35%,transparent)] text-[var(--text-primary)]'
@@ -53,6 +55,7 @@ function wordSegments(lines: DiffLine[]): Map<DiffLine, Segment[]> {
     const adds: DiffLine[] = []
     while (lines[index]?.type === 'add') adds.push(lines[index++])
     for (let pair = 0; pair < Math.min(dels.length, adds.length); pair++) {
+      if (dels[pair].text.length + adds[pair].text.length > WORD_DIFF_MAX_CHARS) continue
       const parts = diffWordsWithSpace(dels[pair].text, adds[pair].text)
       segments.set(
         dels[pair],
@@ -158,9 +161,10 @@ export function DiffView({ hunks }: DiffViewProps) {
           const expand = () => setExpanded((current) => new Set(current).add(hunkIndex))
           return (
             <Fragment key={hunkIndex}>
-              {(hunk.heading || hunkIndex > 0) && (
+              {(hunk.file || hunk.heading || hunkIndex > 0) && (
                 <div className='sticky left-0 col-span-full px-3 pt-2 pb-1 font-sans text-[var(--text-muted)] text-caption'>
-                  {hunk.heading || '⋯'}
+                  {hunk.file && <span className='text-[var(--text-body)]'>{hunk.file} </span>}
+                  {hunk.heading || (hunk.file ? '' : '⋯')}
                 </div>
               )}
               {items.map((item, index) =>

apps/sim/lib/diff/unified.test.ts

@@ -47,6 +47,17 @@ describe('parseUnifiedDiff', () => {
     })
   })
 
+  it('splits a multi-file git diff into hunks per file', () => {
+    const diff = parseUnifiedDiff(
+      'diff --git a/a.ts b/a.ts\n--- a/a.ts\n+++ b/a.ts\n@@ -1,1 +1,1 @@\n-one\n+uno\ndiff --git a/b.ts b/c.ts\nsimilarity index 90%\ncopy from b.ts\ncopy to c.ts\n--- a/b.ts\n+++ b/c.ts\n@@ -3,1 +3,1 @@\n-two\n+dos'
+    )
+    expect(diff.path).toBe('2 files')
+    expect(diff.hunks.map((hunk) => [hunk.file, hunk.lines.map((line) => line.text)])).toEqual([
+      ['a.ts', ['one', 'uno']],
+      ['c.ts', ['two', 'dos']],
+    ])
+  })
+
   it('names each side of a two-document comparison', () => {
     const diff = parseUnifiedDiff(
       '--- sim:knowledge/kb/old\n+++ sim:knowledge/kb/new\n-Refunds are available within 14 days.\n+Annual plans can be refunded within 30 days.'

apps/sim/lib/diff/unified.ts

@@ -14,6 +14,8 @@ export interface DiffLine {
 export interface DiffHunk {
   /** Free text after the `@@ … @@` range, often the enclosing heading or function. */
   heading: string
+  /** The file a hunk belongs to, set only when a git diff spans several files. */
+  file?: string
   lines: DiffLine[]
 }
 
@@ -29,7 +31,7 @@ export interface UnifiedDiff {
 
 /** Git metadata that precedes `---`/`+++` in `git diff` output and carries nothing to render. */
 const GIT_HEADER =
-  /^(?:diff --git |index |new file mode |deleted file mode |similarity |rename |old mode |new mode |Binary files )/
+  /^(?:index |new file mode |deleted file mode |similarity |dissimilarity |rename |copy |old mode |new mode |Binary files )/
 const HUNK_HEADER = /^@@(?: -(\d+)(?:,\d+)? \+(\d+)(?:,\d+)? @@)?\s?(.*?)(?:\s*@@)?$/
 const SIM_SOURCE = /^sim:(file|knowledge)\/([^/\s]+)(?:\/([^/\s]+))?$/
 
@@ -72,28 +74,35 @@ export function parseUnifiedDiff(text: string): UnifiedDiff {
   let oldLine: number | undefined
   let newLine: number | undefined
 
+  let file: string | null = null
   for (const [index, row] of rows.entries()) {
+    if (row.startsWith('diff --git ')) {
+      hunk = null
+      file = null
+      continue
+    }
     if (!hunk && (row.startsWith('--- ') || row.startsWith('+++ '))) {
       const target = row.slice(4).trim()
       const parsed = parseSource(target)
-      if (row.startsWith('--- ')) oldSource = parsed
-      else newSource = parsed
-      if (!parsed && target !== '/dev/null' && (row.startsWith('+++ ') || !path))
-        path = target.replace(/^[ab]\//, '').split('\t')[0]
+      if (row.startsWith('--- ')) oldSource = parsed ?? oldSource
+      else newSource = parsed ?? newSource
+      if (!parsed && target !== '/dev/null' && (row.startsWith('+++ ') || !file))
+        file = target.replace(/^[ab]\//, '').split('\t')[0]
+      path ??= file
       continue
     }
     if (!hunk && GIT_HEADER.test(row)) continue
     const header = HUNK_HEADER.exec(row)
     if (row.startsWith('@@') && header) {
-      hunk = { heading: header[3] ?? '', lines: [] }
+      hunk = { heading: header[3] ?? '', file: file ?? undefined, lines: [] }
       hunks.push(hunk)
       oldLine = header[1] ? Number(header[1]) : undefined
       newLine = header[2] ? Number(header[2]) : undefined
       continue
     }
     if (row.startsWith('\\')) continue
     if (!hunk) {
-      hunk = { heading: '', lines: [] }
+      hunk = { heading: '', file: file ?? undefined, lines: [] }
       hunks.push(hunk)
     }
     const marker = row[0]
@@ -114,5 +123,8 @@ export function parseUnifiedDiff(text: string): UnifiedDiff {
     throw new Error('A diff needs at least one + or - line')
   oldSource ??= newSource
   newSource ??= oldSource
+  const files = new Set(hunks.map((entry) => entry.file))
+  if (files.size <= 1) for (const entry of hunks) entry.file = undefined
+  else path = `${files.size} files`
   return { oldSource, newSource, path: oldSource ? null : path, hunks }
 }