fix: correct outcome shape handling in extractStats and formatOutcome

Rust serializes ReladiffOutcome with serde tag 'mode', producing:
  {mode: 'diff', diff_rows: [...], stats: {rows_table1, rows_table2, exclusive_table1, exclusive_table2, updated, unchanged}}

Previous code checked for {Match: {...}} / {Diff: {...}} shapes that
never matched, causing partitioned diff to report all partitions as
'identical' with 0 rows.

- extractStats(): check outcome.mode === 'diff', read from stats fields
- mergeOutcomes(): aggregate mode-based outcomes correctly
- summarize()/formatOutcome(): display mode-based shape with correct labels

Author: suryaiyer95

packages/opencode/src/altimate/native/connections/data-diff.ts

@@ -229,6 +229,9 @@ function buildPartitionWhereClause(
 
 /**
  * Extract DiffStats from a successful outcome (if present).
+ *
+ * Rust serializes ReladiffOutcome as: {mode: "diff", diff_rows: [...], stats: {...}}
+ * stats fields: rows_table1, rows_table2, exclusive_table1, exclusive_table2, updated, unchanged
  */
 function extractStats(outcome: unknown): {
   rows_source: number
@@ -239,57 +242,54 @@ function extractStats(outcome: unknown): {
   const o = outcome as any
   if (!o) return { rows_source: 0, rows_target: 0, differences: 0, status: "identical" }
 
-  if (o.Match) {
-    return {
-      rows_source: o.Match.row_count ?? 0,
-      rows_target: o.Match.row_count ?? 0,
-      differences: 0,
-      status: "identical",
-    }
-  }
-
-  if (o.Diff) {
-    const d = o.Diff
+  if (o.mode === "diff") {
+    const s = o.stats ?? {}
+    const exclusive1 = Number(s.exclusive_table1 ?? 0)
+    const exclusive2 = Number(s.exclusive_table2 ?? 0)
+    const updated = Number(s.updated ?? 0)
+    const differences = exclusive1 + exclusive2 + updated
     return {
-      rows_source: d.total_source_rows ?? 0,
-      rows_target: d.total_target_rows ?? 0,
-      differences: (d.rows_only_in_source ?? 0) + (d.rows_only_in_target ?? 0) + (d.rows_updated ?? 0),
-      status: "differ",
+      rows_source: Number(s.rows_table1 ?? 0),
+      rows_target: Number(s.rows_table2 ?? 0),
+      differences,
+      status: differences > 0 ? "differ" : "identical",
     }
   }
 
   return { rows_source: 0, rows_target: 0, differences: 0, status: "identical" }
 }
 
 /**
- * Merge two Diff outcomes into one aggregated Diff outcome.
+ * Merge two diff outcomes into one aggregated outcome.
+ *
+ * Both outcomes use the Rust shape: {mode: "diff", diff_rows: [...], stats: {...}}
  */
 function mergeOutcomes(accumulated: unknown, next: unknown): unknown {
+  if (!accumulated) return next
+  if (!next) return accumulated
+
   const a = accumulated as any
   const n = next as any
 
-  const aD = a?.Diff ?? (a?.Match ? { total_source_rows: a.Match.row_count, total_target_rows: a.Match.row_count, rows_only_in_source: 0, rows_only_in_target: 0, rows_updated: 0, rows_identical: a.Match.row_count, sample_diffs: [] } : null)
-  const nD = n?.Diff ?? (n?.Match ? { total_source_rows: n.Match.row_count, total_target_rows: n.Match.row_count, rows_only_in_source: 0, rows_only_in_target: 0, rows_updated: 0, rows_identical: n.Match.row_count, sample_diffs: [] } : null)
-
-  if (!aD && !nD) return { Match: { row_count: 0 } }
-  if (!aD) return next
-  if (!nD) return accumulated
-
-  const merged = {
-    total_source_rows: (aD.total_source_rows ?? 0) + (nD.total_source_rows ?? 0),
-    total_target_rows: (aD.total_target_rows ?? 0) + (nD.total_target_rows ?? 0),
-    rows_only_in_source: (aD.rows_only_in_source ?? 0) + (nD.rows_only_in_source ?? 0),
-    rows_only_in_target: (aD.rows_only_in_target ?? 0) + (nD.rows_only_in_target ?? 0),
-    rows_updated: (aD.rows_updated ?? 0) + (nD.rows_updated ?? 0),
-    rows_identical: (aD.rows_identical ?? 0) + (nD.rows_identical ?? 0),
-    sample_diffs: [...(aD.sample_diffs ?? []), ...(nD.sample_diffs ?? [])].slice(0, 20),
-  }
+  const aS = a.stats ?? {}
+  const nS = n.stats ?? {}
+
+  const rows_table1 = (Number(aS.rows_table1) || 0) + (Number(nS.rows_table1) || 0)
+  const rows_table2 = (Number(aS.rows_table2) || 0) + (Number(nS.rows_table2) || 0)
+  const exclusive_table1 = (Number(aS.exclusive_table1) || 0) + (Number(nS.exclusive_table1) || 0)
+  const exclusive_table2 = (Number(aS.exclusive_table2) || 0) + (Number(nS.exclusive_table2) || 0)
+  const updated = (Number(aS.updated) || 0) + (Number(nS.updated) || 0)
+  const unchanged = (Number(aS.unchanged) || 0) + (Number(nS.unchanged) || 0)
 
-  const totalDiff = merged.rows_only_in_source + merged.rows_only_in_target + merged.rows_updated
-  if (totalDiff === 0) {
-    return { Match: { row_count: merged.total_source_rows, algorithm: "partitioned" } }
+  const totalRows = rows_table1 + rows_table2
+  const totalDiff = exclusive_table1 + exclusive_table2 + updated
+  const diff_percent = totalRows > 0 ? (totalDiff / totalRows) * 100 : 0
+
+  return {
+    mode: "diff",
+    diff_rows: [...(a.diff_rows ?? []), ...(n.diff_rows ?? [])].slice(0, 100),
+    stats: { rows_table1, rows_table2, exclusive_table1, exclusive_table2, updated, unchanged, diff_percent },
   }
-  return { Diff: merged }
 }
 
 /**

packages/opencode/src/altimate/tools/data-diff.ts

@@ -129,16 +129,23 @@ export const DataDiffTool = Tool.define("data_diff", {
 
 function summarize(outcome: any): string {
   if (!outcome) return "complete"
-  if (outcome.Match) return "IDENTICAL ✓"
-  if (outcome.Diff) {
-    const r = outcome.Diff
+
+  // Rust serializes ReladiffOutcome as {mode: "diff"|"profile"|..., stats: {...}, diff_rows: [...]}
+  if (outcome.mode === "diff") {
+    const s = outcome.stats ?? {}
+    const e1 = Number(s.exclusive_table1 ?? 0)
+    const e2 = Number(s.exclusive_table2 ?? 0)
+    const upd = Number(s.updated ?? 0)
+    if (e1 === 0 && e2 === 0 && upd === 0) return "IDENTICAL ✓"
     const parts: string[] = []
-    if (r.rows_only_in_source > 0) parts.push(`${r.rows_only_in_source} only in source`)
-    if (r.rows_only_in_target > 0) parts.push(`${r.rows_only_in_target} only in target`)
-    if (r.rows_updated > 0) parts.push(`${r.rows_updated} updated`)
-    return parts.length ? parts.join(", ") : "differences found"
+    if (e1 > 0) parts.push(`${e1} only in source`)
+    if (e2 > 0) parts.push(`${e2} only in target`)
+    if (upd > 0) parts.push(`${upd} updated`)
+    return parts.join(", ")
   }
-  if (outcome.Profile) return "profile complete"
+  if (outcome.mode === "profile") return "profile complete"
+  if (outcome.mode === "cascade") return "cascade complete"
+
   return "complete"
 }
 
@@ -147,45 +154,54 @@ function formatOutcome(outcome: any, source: string, target: string): string {
 
   const lines: string[] = []
 
-  if (outcome.Match) {
-    lines.push(`✓ Tables are IDENTICAL`)
-    const m = outcome.Match
-    if (m.row_count != null) lines.push(`  Rows checked: ${m.row_count.toLocaleString()}`)
-    if (m.algorithm) lines.push(`  Algorithm: ${m.algorithm}`)
-    return lines.join("\n")
-  }
+  // Rust serializes ReladiffOutcome as {mode: "diff", diff_rows: [...], stats: {...}}
+  // stats: rows_table1, rows_table2, exclusive_table1, exclusive_table2, updated, unchanged
+  if (outcome.mode === "diff") {
+    const s = outcome.stats ?? {}
+    const rows1 = Number(s.rows_table1 ?? 0)
+    const rows2 = Number(s.rows_table2 ?? 0)
+    const e1 = Number(s.exclusive_table1 ?? 0)
+    const e2 = Number(s.exclusive_table2 ?? 0)
+    const updated = Number(s.updated ?? 0)
+    const unchanged = Number(s.unchanged ?? 0)
+
+    if (e1 === 0 && e2 === 0 && updated === 0) {
+      lines.push(`✓ Tables are IDENTICAL`)
+      if (rows1 > 0) lines.push(`  Rows checked: ${rows1.toLocaleString()}`)
+      return lines.join("\n")
+    }
 
-  if (outcome.Diff) {
-    const r = outcome.Diff
     lines.push(`✗ Tables DIFFER`)
     lines.push(``)
     lines.push(`  Source:  ${source}`)
     lines.push(`  Target:  ${target}`)
     lines.push(``)
 
-    if (r.total_source_rows != null) lines.push(`  Source rows:        ${r.total_source_rows.toLocaleString()}`)
-    if (r.total_target_rows != null) lines.push(`  Target rows:        ${r.total_target_rows.toLocaleString()}`)
-    if (r.rows_only_in_source > 0) lines.push(`  Only in source:     ${r.rows_only_in_source.toLocaleString()}`)
-    if (r.rows_only_in_target > 0) lines.push(`  Only in target:     ${r.rows_only_in_target.toLocaleString()}`)
-    if (r.rows_updated > 0) lines.push(`  Updated rows:       ${r.rows_updated.toLocaleString()}`)
-    if (r.rows_identical > 0) lines.push(`  Identical rows:     ${r.rows_identical.toLocaleString()}`)
+    if (rows1 > 0) lines.push(`  Source rows:        ${rows1.toLocaleString()}`)
+    if (rows2 > 0) lines.push(`  Target rows:        ${rows2.toLocaleString()}`)
+    if (e1 > 0) lines.push(`  Only in source:     ${e1.toLocaleString()}`)
+    if (e2 > 0) lines.push(`  Only in target:     ${e2.toLocaleString()}`)
+    if (updated > 0) lines.push(`  Updated rows:       ${updated.toLocaleString()}`)
+    if (unchanged > 0) lines.push(`  Identical rows:     ${unchanged.toLocaleString()}`)
 
-    if (r.sample_diffs?.length) {
+    const diffRows = outcome.diff_rows ?? []
+    if (diffRows.length > 0) {
       lines.push(``)
-      lines.push(`  Sample differences (first ${r.sample_diffs.length}):`)
-      for (const d of r.sample_diffs.slice(0, 5)) {
-        lines.push(`    key=${JSON.stringify(d.key)} col=${d.column}: ${d.source_value} → ${d.target_value}`)
+      lines.push(`  Sample differences (first ${Math.min(diffRows.length, 5)}):`)
+      for (const d of diffRows.slice(0, 5)) {
+        const label = d.sign === "-" ? "source only" : "target only"
+        lines.push(`    [${label}] ${d.values?.join(" | ")}`)
       }
     }
 
     return lines.join("\n")
   }
 
-  if (outcome.Profile) {
-    const p = outcome.Profile
+  if (outcome.mode === "profile") {
+    const cols = outcome.column_stats ?? outcome.columns ?? []
     lines.push(`Column Profile Comparison`)
     lines.push(``)
-    for (const col of p.columns ?? []) {
+    for (const col of cols) {
       const verdict = col.verdict === "match" ? "✓" : col.verdict === "within_tolerance" ? "~" : "✗"
       lines.push(`  ${verdict} ${col.column}: ${col.verdict}`)
       if (col.source_stats && col.target_stats) {