@arbidocs/cli 0.3.30 → 0.3.32

package/CHANGELOG.md

@@ -1,5 +1,25 @@
 # Changelog
 
+## v0.3.32
+
+[compare changes](https://github.com/arbicity/ARBI-frontend/compare/v0.3.31...HEAD)
+
+### 🚀 Enhancements
+
+- **cli:** Robust doc reprocess with retry, per-doc fallback, ext filter ([#576](https://github.com/arbicity/ARBI-frontend/pull/576))
+
+## v0.3.31
+
+[compare changes](https://github.com/arbicity/ARBI-frontend/compare/v0.3.30...HEAD)
+
+### 🚀 Enhancements
+
+- **cli:** --interval for reprocess batches, --output for docs export ([#574](https://github.com/arbicity/ARBI-frontend/pull/574))
+
+### 🔥 Performance
+
+- Instant workspace click feedback + bulk reprocess cap ([#573](https://github.com/arbicity/ARBI-frontend/pull/573))
+
 ## v0.3.30
 
 [compare changes](https://github.com/arbicity/ARBI-frontend/compare/v0.3.29...HEAD)

package/dist/index.js

@@ -3637,7 +3637,7 @@ function getLatestVersion(skipCache = false) {
   }
 }
 function getCurrentVersion() {
-  return "0.3.30";
+  return "0.3.32";
 }
 function readChangelog(fromVersion, toVersion) {
   try {
@@ -3690,17 +3690,17 @@ function showChangelog(fromVersion, toVersion) {
 async function checkForUpdates(autoUpdate) {
   try {
     const latest = getLatestVersion();
-    if (!latest || latest === "0.3.30") return;
+    if (!latest || latest === "0.3.32") return;
     if (autoUpdate) {
       warn(`
-Your arbi version is out of date (${"0.3.30"} \u2192 ${latest}). Updating...`);
+Your arbi version is out of date (${"0.3.32"} \u2192 ${latest}). Updating...`);
       child_process.execSync("npm install -g @arbidocs/cli@latest", { stdio: "inherit" });
-      showChangelog("0.3.30", latest);
+      showChangelog("0.3.32", latest);
       console.log(`Updated to ${latest}.`);
     } else {
       warn(
         `
-Your arbi version is out of date (${"0.3.30"} \u2192 ${latest}).
+Your arbi version is out of date (${"0.3.32"} \u2192 ${latest}).
 Run "arbi update" to upgrade, or "arbi update auto" to always stay up to date.`
       );
     }
@@ -3710,9 +3710,9 @@ Run "arbi update" to upgrade, or "arbi update auto" to always stay up to date.`
 function hintUpdateOnError() {
   try {
     const cached = readCache();
-    if (cached && cached.latest !== "0.3.30") {
+    if (cached && cached.latest !== "0.3.32") {
       warn(
-        `Your arbi version is out of date (${"0.3.30"} \u2192 ${cached.latest}). Run "arbi update".`
+        `Your arbi version is out of date (${"0.3.32"} \u2192 ${cached.latest}). Run "arbi update".`
       );
     }
   } catch {
@@ -4669,7 +4669,10 @@ async function fetchDocChoices(arbi, _workspaceId) {
   }));
 }
 function registerDocsCommand(program2) {
-  program2.command("docs").description("List documents in the active workspace").option("-w, --workspace <id>", "Workspace ID (defaults to selected workspace)").option("--json", "Output as JSON").option("--csv", "Output as CSV").option("--ids", "Output only document IDs (one per line)").option("--count", "Output only the count (combine with --status for filtered count)").option("-s, --status <status>", "Filter by status (comma-separated: completed,failed,queued)").option("-f, --folder <pattern>", "Filter by folder (substring match)").option("-e, --ext <extensions>", "Filter by file extension (comma-separated: pdf,docx,eml)").option("-q, --query <text>", "Search file name, title, or summary").option("--sort <field>", "Sort by field (name, status, date, size, created)", "status").option("-n, --limit <n>", "Limit number of results").action(
+  program2.command("docs").description("List documents in the active workspace").option("-w, --workspace <id>", "Workspace ID (defaults to selected workspace)").option("--json", "Output as JSON").option("--csv", "Output as CSV").option("--ids", "Output only document IDs (one per line)").option("--count", "Output only the count (combine with --status for filtered count)").option("-s, --status <status>", "Filter by status (comma-separated: completed,failed,queued)").option("-f, --folder <pattern>", "Filter by folder (substring match)").option("-e, --ext <extensions>", "Filter by file extension (comma-separated: pdf,docx,eml)").option("-q, --query <text>", "Search file name, title, or summary").option("--sort <field>", "Sort by field (name, status, date, size, created)", "status").option("-n, --limit <n>", "Limit number of results").option(
+    "-o, --output <path>",
+    "Write output to a file instead of stdout. Extension determines the format when combined with --ids/--json/--csv (default: csv with doc_id, file_name, folder, status)"
+  ).action(
     (opts) => runAction(async () => {
       const { arbi } = await resolveWorkspace(opts.workspace);
       let data = await sdk.documents.listDocuments(arbi);
@@ -4723,20 +4726,30 @@ function registerDocsCommand(program2) {
         }
         return;
       }
+      const writeOut = (payload, defaultExt) => {
+        if (opts.output) {
+          fs4.writeFileSync(opts.output, payload.endsWith("\n") ? payload : payload + "\n");
+          success(`Wrote ${data.length} rows to ${opts.output} (${defaultExt})`);
+        } else {
+          console.log(payload.endsWith("\n") ? payload.slice(0, -1) : payload);
+        }
+      };
+      const csvEscape = (s) => `"${(s ?? "").replace(/"/g, '""')}"`;
       if (opts.ids) {
-        data.forEach((d) => console.log(d.external_id));
+        writeOut(data.map((d) => d.external_id).join("\n"), "ids");
         return;
       }
       if (opts.json) {
-        console.log(JSON.stringify(data));
+        writeOut(JSON.stringify(data, null, opts.output ? 2 : 0), "json");
         return;
       }
-      if (opts.csv) {
-        console.log("external_id,status,file_name,file_size,folder,n_pages,tokens,doc_date,title");
+      if (opts.csv || opts.output && !opts.ids && !opts.json) {
+        const lines = [
+          "external_id,status,file_name,file_size,folder,n_pages,tokens,doc_date,title"
+        ];
         for (const d of data) {
           const meta = d.doc_metadata;
-          const csvEscape = (s) => `"${(s ?? "").replace(/"/g, '""')}"`;
-          console.log(
+          lines.push(
             [
               d.external_id,
               d.status,
@@ -4750,6 +4763,7 @@ function registerDocsCommand(program2) {
             ].join(",")
           );
         }
+        writeOut(lines.join("\n"), "csv");
         return;
       }
       console.log(chalk2__default.default.dim(`${data.length} documents
@@ -4916,11 +4930,29 @@ function registerDocsCommand(program2) {
       console.log(JSON.stringify(data, null, 2));
     })()
   );
-  doc.command("reprocess [ids...]").description("Reprocess failed/completed documents (sets status back to processing)").option("-s, --status <status>", "Reprocess all docs with this status (e.g. failed)").option("-f, --folder <pattern>", "Filter by folder (substring match)").option("--dry-run", "Show what would be reprocessed without doing it").option("-b, --batch-size <n>", "Batch size for update requests", "50").action(
+  doc.command("reprocess [ids...]").description("Reprocess failed/completed documents (sets status back to processing)").option("-s, --status <status>", "Reprocess all docs with this status (e.g. failed)").option("-f, --folder <pattern>", "Filter by folder (substring match)").option(
+    "-e, --ext <extension>",
+    "Filter by file extension (e.g. pdf, docx). Requires --status."
+  ).option("--dry-run", "Show what would be reprocessed without doing it").option("-b, --batch-size <n>", "Batch size for update requests", "50").option(
+    "-i, --interval <seconds>",
+    "Seconds to wait between batches (float allowed, e.g. 0.5). Default 0 \u2014 send as fast as the server acks.",
+    "0"
+  ).option(
+    "-r, --max-retries <n>",
+    "Retry attempts per failing batch before falling back to per-doc submission.",
+    "3"
+  ).option(
+    "--status-interval <seconds>",
+    "Seconds between progress lines (0 to disable periodic printing).",
+    "5"
+  ).option("-v, --verbose", "Print a log line after each batch in addition to periodic stats.").option("-q, --quiet", "Suppress periodic progress lines \u2014 only print the final summary.").action(
     (ids, opts) => runAction(async () => {
-      const { arbi } = await resolveWorkspace();
+      const { arbi } = await resolveWorkspace(void 0, { skipNotifications: true });
       let docIds;
       if (ids && ids.length > 0) {
+        if (opts.ext || opts.folder) {
+          warn("--ext/--folder are ignored when explicit IDs are given.");
+        }
         docIds = ids;
       } else if (opts.status) {
         const allDocs = await sdk.documents.listDocuments(arbi);
@@ -4931,6 +4963,12 @@ function registerDocsCommand(program2) {
             (d) => (d.folder ?? "").toLowerCase().includes(pattern)
           );
         }
+        if (opts.ext) {
+          const ext = opts.ext.toLowerCase().replace(/^\./, "");
+          filtered = filtered.filter(
+            (d) => (d.file_name ?? "").toLowerCase().endsWith("." + ext)
+          );
+        }
         docIds = filtered.map((d) => d.external_id);
       } else {
         error("Provide document IDs or use --status to select documents to reprocess.");
@@ -4948,18 +4986,115 @@ function registerDocsCommand(program2) {
         return;
       }
       const batchSize = parseInt(opts.batchSize ?? "50", 10);
-      let processed = 0;
-      for (let i = 0; i < docIds.length; i += batchSize) {
-        const batch = docIds.slice(i, i + batchSize);
-        const updates = batch.map((id) => ({ external_id: id, status: "processing" }));
-        await sdk.documents.updateDocuments(
-          arbi,
-          updates
-        );
-        processed += batch.length;
-        console.log(`  [${processed}/${docIds.length}] Triggered reprocessing...`);
+      const maxRetries = parseInt(opts.maxRetries ?? "3", 10);
+      const intervalSec = parseFloat(opts.interval ?? "0");
+      const intervalMs = Number.isFinite(intervalSec) && intervalSec > 0 ? intervalSec * 1e3 : 0;
+      const statusIntervalSec = parseFloat(opts.statusInterval ?? "5");
+      const statusIntervalMs = Number.isFinite(statusIntervalSec) && statusIntervalSec > 0 ? statusIntervalSec * 1e3 : 0;
+      const total = docIds.length;
+      const startMs = Date.now();
+      let submitted = 0;
+      let succeeded = 0;
+      let failed = 0;
+      const failedIds = [];
+      const formatDuration = (ms) => {
+        const s = Math.floor(ms / 1e3);
+        if (s < 60) return `${s}s`;
+        const m = Math.floor(s / 60);
+        if (m < 60) return `${m}m${s % 60}s`;
+        const h = Math.floor(m / 60);
+        return `${h}h${m % 60}m`;
+      };
+      const printStatus = (prefix = "") => {
+        const elapsed = Date.now() - startMs;
+        const rate = submitted > 0 ? submitted / elapsed * 1e3 : 0;
+        const remaining = total - submitted;
+        const etaMs = rate > 0 ? remaining / rate * 1e3 : 0;
+        const bar = `submitted=${submitted}/${total} ok=${succeeded} fail=${failed} rate=${rate.toFixed(1)}/s elapsed=${formatDuration(elapsed)}` + (rate > 0 && remaining > 0 ? ` eta=${formatDuration(etaMs)}` : "");
+        console.log(`${prefix}${bar}`);
+      };
+      let statusTimer;
+      if (!opts.quiet && statusIntervalMs > 0) {
+        statusTimer = setInterval(() => printStatus("  "), statusIntervalMs);
+      }
+      const sleep = (ms) => new Promise((resolve3) => setTimeout(resolve3, ms));
+      const backoff = async (attempt) => {
+        const delayMs = Math.min(1e3 * Math.pow(2, attempt), 3e4);
+        await sleep(delayMs);
+      };
+      const submitBatch = async (batchIds) => {
+        const updates = batchIds.map((id) => ({
+          external_id: id,
+          status: "processing"
+        }));
+        try {
+          await sdk.documents.updateDocuments(
+            arbi,
+            updates
+          );
+          return true;
+        } catch {
+          return false;
+        }
+      };
+      try {
+        for (let i = 0; i < docIds.length; i += batchSize) {
+          const batch = docIds.slice(i, i + batchSize);
+          let batchOk = false;
+          for (let attempt = 0; attempt <= maxRetries; attempt++) {
+            batchOk = await submitBatch(batch);
+            if (batchOk) break;
+            if (attempt < maxRetries) {
+              await backoff(attempt);
+            }
+          }
+          if (batchOk) {
+            succeeded += batch.length;
+            submitted += batch.length;
+            if (opts.verbose) {
+              console.log(`  [${submitted}/${total}] Triggered reprocessing...`);
+            }
+          } else {
+            console.log(
+              chalk2__default.default.dim(
+                `  batch of ${batch.length} failed after ${maxRetries} retries, falling back to per-doc`
+              )
+            );
+            for (const id of batch) {
+              let docOk = await submitBatch([id]);
+              if (!docOk) {
+                await sleep(500);
+                docOk = await submitBatch([id]);
+              }
+              if (docOk) {
+                succeeded += 1;
+              } else {
+                failed += 1;
+                failedIds.push(id);
+              }
+              submitted += 1;
+            }
+          }
+          if (intervalMs > 0 && i + batchSize < docIds.length) {
+            await sleep(intervalMs);
+          }
+        }
+      } finally {
+        if (statusTimer) clearInterval(statusTimer);
+      }
+      console.log("");
+      printStatus();
+      if (failed > 0) {
+        warn(`${failed} document(s) failed to reprocess.`);
+        const preview = failedIds.slice(0, 20);
+        preview.forEach((id) => console.log(`  ${chalk2__default.default.red("\u2717")} ${id}`));
+        if (failedIds.length > preview.length) {
+          console.log(chalk2__default.default.dim(`  ... and ${failedIds.length - preview.length} more`));
+        }
+      }
+      if (succeeded > 0) {
+        success(`Triggered reprocessing for ${succeeded} document(s).`);
       }
-      success(`Triggered reprocessing for ${docIds.length} document(s).`);
     })()
   );
 }
@@ -7745,7 +7880,7 @@ console.info = (...args) => {
   _origInfo(...args);
 };
 var program = new commander.Command();
-program.name("arbi").description("ARBI CLI \u2014 interact with ARBI from the terminal").version("0.3.30");
+program.name("arbi").description("ARBI CLI \u2014 interact with ARBI from the terminal").version("0.3.32");
 registerConfigCommand(program);
 registerLoginCommand(program);
 registerRegisterCommand(program);