hillclimb 0.4.0 → 0.4.2

package/dist/cli.js

@@ -13515,7 +13515,7 @@ var EXEC_OPTS = {
   maxBuffer: 50 * 1024 * 1024
 };
 var MAX_ERROR_OUTPUT_CHARS = 2e3;
-var MAX_UNTRACKED_FILE_BYTES = 10 * 1024 * 1024;
+var MAX_SNAPSHOT_FILE_BYTES = 10 * 1024 * 1024;
 var EMPTY_TREE_SHA = "4b825dc642cb6eb9a060e54bf8d69288fbee4904";
 function quoteGitArg(arg) {
   if (/^[A-Za-z0-9_./:=@%+,-]+$/.test(arg)) return arg;
@@ -13619,7 +13619,81 @@ function captureSnapshotSha(repoRoot) {
   }
   return git(repoRoot, ["rev-parse", "HEAD"]);
 }
-function buildUntrackedTree(repoRoot) {
+function recordOmittedSnapshotFile(omittedFiles, file, options = {}) {
+  omittedFiles?.push(file);
+  if (options.log === false) return;
+  const action = file.tracked ? "omitting tracked" : "skipping untracked";
+  appendLog(
+    "warn",
+    `git-traces: ${action} ${file.path} (${file.sizeBytes} bytes > ${MAX_SNAPSHOT_FILE_BYTES} limit)`
+  );
+}
+function parseLsTreeLongZ(output) {
+  const entries = [];
+  for (const record of output.toString("utf-8").split("\0")) {
+    if (!record) continue;
+    const tab = record.indexOf("	");
+    if (tab === -1) continue;
+    const meta = record.slice(0, tab);
+    const filePath = record.slice(tab + 1);
+    const [mode, type, sha, sizeRaw] = meta.trim().split(/\s+/);
+    if (type !== "blob" || !mode || !sha || !sizeRaw) continue;
+    const sizeBytes = Number.parseInt(sizeRaw, 10);
+    if (!Number.isFinite(sizeBytes)) continue;
+    entries.push({ mode, sha, sizeBytes, path: filePath });
+  }
+  return entries;
+}
+function listOversizedTreeFiles(repoRoot, treeSha, options = {}) {
+  const output = gitBuffer(repoRoot, ["ls-tree", "-r", "-l", "-z", treeSha]);
+  const oversized = [];
+  for (const entry of parseLsTreeLongZ(output)) {
+    if (entry.sizeBytes <= MAX_SNAPSHOT_FILE_BYTES) continue;
+    const file = {
+      path: entry.path,
+      sizeBytes: entry.sizeBytes,
+      tracked: true,
+      reason: "file-over-limit",
+      gitObjectSha: entry.sha
+    };
+    oversized.push(file);
+    recordOmittedSnapshotFile(options.omittedFiles, file, {
+      log: options.log
+    });
+  }
+  return oversized;
+}
+function removePathsFromIndex(repoRoot, env, paths) {
+  if (paths.length === 0) return;
+  gitBuffer(repoRoot, ["update-index", "--force-remove", "-z", "--stdin"], {
+    input: `${paths.join("\0")}\0`,
+    env
+  });
+}
+function filterOversizedFilesFromTree(repoRoot, treeSha, options = {}) {
+  const oversizedFiles = listOversizedTreeFiles(repoRoot, treeSha, options);
+  if (oversizedFiles.length === 0) return treeSha;
+  const tmpIndex = path14.join(
+    os6.tmpdir(),
+    `hillclimb-filter-${Date.now()}-${process.pid}`
+  );
+  const env = { ...process.env, GIT_INDEX_FILE: tmpIndex };
+  try {
+    gitWithEnv(repoRoot, ["read-tree", treeSha], env);
+    removePathsFromIndex(
+      repoRoot,
+      env,
+      oversizedFiles.map((file) => file.path)
+    );
+    return gitWithEnv(repoRoot, ["write-tree"], env);
+  } finally {
+    try {
+      fs11.unlinkSync(tmpIndex);
+    } catch {
+    }
+  }
+}
+function buildUntrackedTree(repoRoot, omittedFiles) {
   const list = git(repoRoot, [
     "ls-files",
     "--others",
@@ -13632,11 +13706,13 @@ function buildUntrackedTree(repoRoot) {
     if (!relPath) continue;
     try {
       const stat = fs11.lstatSync(path14.join(repoRoot, relPath));
-      if (stat.size > MAX_UNTRACKED_FILE_BYTES) {
-        appendLog(
-          "warn",
-          `git-traces: skipping untracked ${relPath} (${stat.size} bytes > ${MAX_UNTRACKED_FILE_BYTES} limit)`
-        );
+      if (stat.size > MAX_SNAPSHOT_FILE_BYTES) {
+        recordOmittedSnapshotFile(omittedFiles, {
+          path: relPath,
+          sizeBytes: stat.size,
+          tracked: false,
+          reason: "file-over-limit"
+        });
         continue;
       }
       kept.push(relPath);
@@ -13662,19 +13738,25 @@ function buildUntrackedTree(repoRoot) {
     }
   }
 }
-function buildSnapshotTree(repoRoot, stashSha) {
+function buildSnapshotTree(repoRoot, stashSha, options = {}) {
   const trackedTree = git(repoRoot, ["rev-parse", `${stashSha}^{tree}`]);
-  const untrackedTree = buildUntrackedTree(repoRoot);
-  if (!untrackedTree || untrackedTree === EMPTY_TREE_SHA) {
-    return trackedTree;
-  }
+  const filteredTrackedTree = filterOversizedFilesFromTree(
+    repoRoot,
+    trackedTree,
+    {
+      omittedFiles: options.omittedFiles
+    }
+  );
+  const untrackedTree = buildUntrackedTree(repoRoot, options.omittedFiles);
+  if (!untrackedTree || untrackedTree === EMPTY_TREE_SHA)
+    return filteredTrackedTree;
   const tmpIndex = path14.join(
     os6.tmpdir(),
     `hillclimb-index-${Date.now()}-${process.pid}`
   );
   const env = { ...process.env, GIT_INDEX_FILE: tmpIndex };
   try {
-    gitWithEnv(repoRoot, ["read-tree", trackedTree], env);
+    gitWithEnv(repoRoot, ["read-tree", filteredTrackedTree], env);
     const untrackedList = gitBuffer(
       repoRoot,
       ["ls-tree", "-r", untrackedTree],
@@ -13727,11 +13809,22 @@ function createBundleFromTree(repoRoot, treeSha, sessionId, label) {
 }
 function createTreeDiffPatchGz(repoRoot, fromTreeSha, toTreeSha) {
   if (fromTreeSha === toTreeSha) return null;
+  const filteredFromTreeSha = filterOversizedFilesFromTree(
+    repoRoot,
+    fromTreeSha,
+    {
+      log: false
+    }
+  );
+  const filteredToTreeSha = filterOversizedFilesFromTree(repoRoot, toTreeSha, {
+    log: false
+  });
+  if (filteredFromTreeSha === filteredToTreeSha) return null;
   const diff = gitBuffer(repoRoot, [
     "diff",
     "--binary",
-    fromTreeSha,
-    toTreeSha
+    filteredFromTreeSha,
+    filteredToTreeSha
   ]);
   if (diff.length === 0) return null;
   return Buffer.from(gzipSync(diff));
@@ -13778,7 +13871,7 @@ function parseDirtyFilesFromStatus(status) {
     return pathPart;
   }).filter(Boolean);
 }
-function buildBaselineMetadata(repoRoot, sessionId, tool, baselineSha, cliVersion, epoch, prevHeadSha, transitionKind, startedAt) {
+function buildBaselineMetadata(repoRoot, sessionId, tool, baselineSha, cliVersion, epoch, prevHeadSha, transitionKind, startedAt, omittedFiles = []) {
   const headSha = safeGit(repoRoot, ["rev-parse", "HEAD"]) ?? "unknown";
   const branch = safeGit(repoRoot, ["rev-parse", "--abbrev-ref", "HEAD"]) ?? null;
   const remoteUrl = safeGit(repoRoot, ["config", "--get", "remote.origin.url"]) ?? null;
@@ -13802,7 +13895,8 @@ function buildBaselineMetadata(repoRoot, sessionId, tool, baselineSha, cliVersio
       branch,
       remoteUrl,
       isDirty: dirtyFiles.length > 0,
-      dirtyFiles
+      dirtyFiles,
+      ...omittedFiles.length > 0 ? { omittedFiles } : {}
     },
     author: { name: authorName, email: authorEmail },
     hostname: os6.hostname(),
@@ -14246,6 +14340,10 @@ function freezeEpochBaseline(params) {
       sessionId,
       epoch
     );
+    const omittedFiles = [];
+    const baselineTreeSha = buildSnapshotTree(repoRoot, baselineSha, {
+      omittedFiles
+    });
     const baselineMetadata = buildBaselineMetadata(
       repoRoot,
       sessionId,
@@ -14255,9 +14353,9 @@ function freezeEpochBaseline(params) {
       epoch,
       prevHeadSha,
       transitionKind,
-      startedAt
+      startedAt,
+      omittedFiles
     );
-    const baselineTreeSha = buildSnapshotTree(repoRoot, baselineSha);
     pinFrozenBaselineTree(repoRoot, sessionId, epoch, baselineTreeSha);
     return { baselineSha, baselineTreeSha, baselineMetadata, headSha };
   } catch (err) {
@@ -14303,6 +14401,10 @@ function buildEpochBaselineArtifacts(params) {
   } = params;
   const prefix = epochPrefix(epoch);
   try {
+    const omittedFiles = [];
+    const baselineTreeSha = buildSnapshotTree(repoRoot, baselineSha, {
+      omittedFiles
+    });
     const metadata = buildBaselineMetadata(
       repoRoot,
       sessionId,
@@ -14312,9 +14414,9 @@ function buildEpochBaselineArtifacts(params) {
       epoch,
       prevHeadSha,
       transitionKind,
-      startedAt
+      startedAt,
+      omittedFiles
     );
-    const baselineTreeSha = buildSnapshotTree(repoRoot, baselineSha);
     return buildFrozenEpochBaselineArtifacts({
       repoRoot,
       sessionId,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "hillclimb",
-  "version": "0.4.0",
+  "version": "0.4.2",
   "description": "Extract and export AI coding tool logs grouped by repo",
   "license": "MIT",
   "type": "module",