rowpipe 2.10.2 → 2.10.4

package/README.md

@@ -56,14 +56,14 @@ brew install litepacks/rowpipe/rowpipe
 
 #### 📦 Debian / Ubuntu (`apt` / `.deb`)
 ```bash
-curl -sLO https://github.com/litepacks/rowpipe/releases/latest/download/rowpipe_2.10.1_amd64.deb
-sudo apt install -y ./rowpipe_2.10.1_amd64.deb
+curl -sLO https://github.com/litepacks/rowpipe/releases/latest/download/rowpipe_2.10.4_amd64.deb
+sudo apt install -y ./rowpipe_2.10.4_amd64.deb
 ```
 
 #### 🪟 Windows & GitHub Releases
 Pre-compiled standalone binaries for **macOS (Apple Silicon)**, **Linux (x64)**, and **Windows (x64)** are available on the [GitHub Releases](https://github.com/litepacks/rowpipe/releases) page:
 * `rowpipe-darwin-arm64.tar.gz` (macOS Apple Silicon M1/M2/M3/M4)
-* `rowpipe-linux-x64.tar.gz` / `rowpipe_2.10.1_amd64.deb` (Linux)
+* `rowpipe-linux-x64.tar.gz` / `rowpipe_2.10.4_amd64.deb` (Linux)
 * `rowpipe-win32-x64.zip` (Windows)
 
 ---
@@ -83,6 +83,126 @@ npm install rowpipe
 
 ---
 
+## Model Context Protocol (MCP) Server
+
+Rowpipe includes a native, production-ready **Model Context Protocol (MCP)** server powered by `mcponce`, enabling AI coding assistants and agents (Claude Desktop, Cursor, Antigravity, OpenDevin, etc.) to inspect, query, transform, convert, and diff massive tabular datasets with bounded $O(1)$ memory.
+
+### 🤖 9 Core Tabular Tools
+1. **`rowpipe_inspect`**: Inspect file size, row/column counts, types, null percentages, approximate distinct rates, and XLSX sheet breakdowns.
+2. **`rowpipe_query`**: Stream-query with filter expressions, column selection, renaming, maps, casting, sorting, and pagination.
+3. **`rowpipe_schema`**: Infer accurate schemas, SQL data types, null rates, and semantic types (`email`, `url`, `uuid`, etc.).
+4. **`rowpipe_stats`**: Calculate summary statistics (min, max, mean, sum, quantiles, approx distinct) for numeric and string columns.
+5. **`rowpipe_sample`**: Reservoir sampling with bounded $O(k)$ memory and deterministic random seeds.
+6. **`rowpipe_convert`**: High-throughput format conversion across CSV, TSV, JSON, JSONL, Parquet, XLSX, and Markdown.
+7. **`rowpipe_diff`**: Key-based row-level dataset diffing with added/removed/changed metrics and schema diffing.
+8. **`rowpipe_profile`**: Deep data profiling, distributions, and quality warnings.
+9. **`rowpipe_table`**: Render clean, aligned Unicode/ASCII preview tables.
+
+### 🌐 Dynamic Resource Templates
+- `rowpipe://metadata/{filePath}`: Returns dataset metadata, size, format, and sheet summaries.
+- `rowpipe://schema/{filePath}`: Returns inferred schema and semantic classifications.
+
+### 💻 MCP CLI Usage
+```bash
+# Start standard MCP stdio server
+rowpipe mcp
+
+# List all registered tools and parameter schemas
+rowpipe mcp tools
+
+# Execute any tool directly from CLI
+rowpipe mcp call rowpipe_inspect --filePath ./sales.csv
+rowpipe mcp call rowpipe_query --filePath ./sales.csv --filter "revenue > 1000" --select "name,revenue"
+
+# Run in background daemon mode
+rowpipe mcp -b
+
+# Install into Claude Desktop or Cursor configuration automatically
+rowpipe mcp install claude
+rowpipe mcp install cursor
+
+# Check server status and view logs
+rowpipe mcp status
+rowpipe mcp logs
+```
+
+### ⚙️ Manual MCP Client Configuration
+
+You can configure `rowpipe` manually in any MCP-compatible client:
+
+#### 1. Claude Desktop
+Add `rowpipe` under `mcpServers` in your Claude Desktop configuration file:
+* **macOS**: `~/Library/Application Support/Claude/claude_desktop_config.json`
+* **Windows**: `%APPDATA%\Claude\claude_desktop_config.json`
+* **Linux**: `~/.config/Claude/claude_desktop_config.json`
+
+**Using installed `rowpipe` binary (Standalone or Global npm):**
+```json
+{
+  "mcpServers": {
+    "rowpipe": {
+      "command": "rowpipe",
+      "args": ["mcp"]
+    }
+  }
+}
+```
+
+**Using `npx` (No installation needed):**
+```json
+{
+  "mcpServers": {
+    "rowpipe": {
+      "command": "npx",
+      "args": ["-y", "rowpipe", "mcp"]
+    }
+  }
+}
+```
+
+#### 2. Cursor
+* **UI**: Go to **Settings** (`Cmd+,` or `Ctrl+,`) → **Features** → **MCP Servers** → **Add New MCP Server**:
+  * **Type**: `command`
+  * **Name**: `rowpipe`
+  * **Command**: `rowpipe mcp` (or `npx -y rowpipe mcp`)
+* **Project Configuration**: Create `.cursor/mcp.json` in your workspace root:
+```json
+{
+  "mcpServers": {
+    "rowpipe": {
+      "command": "rowpipe",
+      "args": ["mcp"]
+    }
+  }
+}
+```
+
+#### 3. Windsurf, Antigravity & VS Code (Cline / Roo Code)
+Add to your client's MCP settings (`mcp_config.json` or `cline_mcp_settings.json`):
+```json
+{
+  "mcpServers": {
+    "rowpipe": {
+      "command": "rowpipe",
+      "args": ["mcp"]
+    }
+  }
+}
+```
+
+### 📦 Programmatic Usage (Node.js / TypeScript)
+```typescript
+import { createRowpipeMcpServer } from "rowpipe";
+
+const server = createRowpipeMcpServer();
+const result = await server.callTool("rowpipe_inspect", {
+  filePath: "./large_dataset.parquet"
+});
+console.log(result.data);
+```
+
+---
+
 ## CLI Usage & Examples
 
 ### 1. Inspect Dataset & Multi-Sheet Excel / Parquet Analysis

package/dist/cli/commands/mcp.d.ts

@@ -0,0 +1,17 @@
+export interface McpCliCommandOptions {
+    background?: boolean;
+    dev?: boolean;
+}
+/**
+ * Handles `rowpipe mcp [args...]`
+ *
+ * Forwards arguments to the mcponce MCP application runner, enabling:
+ * - stdio bridge mode (`rowpipe mcp`)
+ * - daemon background mode (`rowpipe mcp --background`)
+ * - developer mode with stderr logging (`rowpipe mcp --dev`)
+ * - tool listings and cli invocations (`rowpipe mcp tools`, `rowpipe mcp call ...`)
+ * - client installation (`rowpipe mcp install claude`, etc.)
+ * - server status, logs, stop, restart (`rowpipe mcp status`, `rowpipe mcp logs`, etc.)
+ */
+export declare function mcpCommand(args?: string[]): Promise<void>;
+//# sourceMappingURL=mcp.d.ts.map

package/dist/cli/commands/mcp.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"mcp.d.ts","sourceRoot":"","sources":["../../../src/cli/commands/mcp.ts"],"names":[],"mappings":"AAGA,MAAM,WAAW,oBAAoB;IACnC,UAAU,CAAC,EAAE,OAAO,CAAC;IACrB,GAAG,CAAC,EAAE,OAAO,CAAC;CACf;AAED;;;;;;;;;;GAUG;AACH,wBAAsB,UAAU,CAAC,IAAI,GAAE,MAAM,EAAO,GAAG,OAAO,CAAC,IAAI,CAAC,CA2BnE"}

package/dist/cli/commands/mcp.js

@@ -0,0 +1,40 @@
+import { fileURLToPath } from "node:url";
+import { createRowpipeMcpServer } from "../../mcp/server.js";
+/**
+ * Handles `rowpipe mcp [args...]`
+ *
+ * Forwards arguments to the mcponce MCP application runner, enabling:
+ * - stdio bridge mode (`rowpipe mcp`)
+ * - daemon background mode (`rowpipe mcp --background`)
+ * - developer mode with stderr logging (`rowpipe mcp --dev`)
+ * - tool listings and cli invocations (`rowpipe mcp tools`, `rowpipe mcp call ...`)
+ * - client installation (`rowpipe mcp install claude`, etc.)
+ * - server status, logs, stop, restart (`rowpipe mcp status`, `rowpipe mcp logs`, etc.)
+ */
+export async function mcpCommand(args = []) {
+    const mcpEntrypoint = fileURLToPath(new URL("../../mcp/index.js", import.meta.url));
+    const app = createRowpipeMcpServer({
+        name: "rowpipe",
+        background: args.includes("--background") || args.includes("-b"),
+    });
+    // Adjust process.argv for mcponce CLI parsing
+    // mcponce inspects process.argv.slice(2)
+    const nodeExec = process.argv[0] ?? process.execPath;
+    const mcpIdx = process.argv.findIndex((a) => a === "mcp");
+    if (mcpIdx !== -1) {
+        process.argv = [
+            nodeExec,
+            mcpEntrypoint,
+            ...process.argv.slice(mcpIdx + 1),
+        ];
+    }
+    else {
+        process.argv = [
+            nodeExec,
+            mcpEntrypoint,
+            ...args,
+        ];
+    }
+    await app.run();
+}
+//# sourceMappingURL=mcp.js.map

package/dist/cli/commands/mcp.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"mcp.js","sourceRoot":"","sources":["../../../src/cli/commands/mcp.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,aAAa,EAAE,MAAM,UAAU,CAAC;AACzC,OAAO,EAAE,sBAAsB,EAAE,MAAM,qBAAqB,CAAC;AAO7D;;;;;;;;;;GAUG;AACH,MAAM,CAAC,KAAK,UAAU,UAAU,CAAC,OAAiB,EAAE;IAClD,MAAM,aAAa,GAAG,aAAa,CAAC,IAAI,GAAG,CAAC,oBAAoB,EAAE,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC;IAEpF,MAAM,GAAG,GAAG,sBAAsB,CAAC;QACjC,IAAI,EAAE,SAAS;QACf,UAAU,EAAE,IAAI,CAAC,QAAQ,CAAC,cAAc,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC;KACjE,CAAC,CAAC;IAEH,8CAA8C;IAC9C,yCAAyC;IACzC,MAAM,QAAQ,GAAG,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,OAAO,CAAC,QAAQ,CAAC;IACrD,MAAM,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,KAAK,KAAK,CAAC,CAAC;IAC1D,IAAI,MAAM,KAAK,CAAC,CAAC,EAAE,CAAC;QAClB,OAAO,CAAC,IAAI,GAAG;YACb,QAAQ;YACR,aAAa;YACb,GAAG,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC;SAClC,CAAC;IACJ,CAAC;SAAM,CAAC;QACN,OAAO,CAAC,IAAI,GAAG;YACb,QAAQ;YACR,aAAa;YACb,GAAG,IAAI;SACR,CAAC;IACJ,CAAC;IAED,MAAM,GAAG,CAAC,GAAG,EAAE,CAAC;AAClB,CAAC"}

package/dist/cli/index.d.ts

@@ -1,3 +1,4 @@
 #!/usr/bin/env node
-export {};
+export declare function getCliVersion(): string;
+export declare function isStdinInteractive(): boolean;
 //# sourceMappingURL=index.d.ts.map

package/dist/cli/index.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../src/cli/index.ts"],"names":[],"mappings":""}
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../src/cli/index.ts"],"names":[],"mappings":";AA+FA,wBAAgB,aAAa,IAAI,MAAM,CAWtC;AAED,wBAAgB,kBAAkB,IAAI,OAAO,CAa5C"}

package/dist/cli/index.js

@@ -10,8 +10,9 @@ process.emitWarning = function (warning, ...args) {
     }
     return Reflect.apply(_origEmitWarning, process, [warning, ...args]);
 };
-import { existsSync } from "node:fs";
+import { existsSync, readFileSync, fstatSync } from "node:fs";
 import { Command } from "commander";
+import { VERSION } from "../version.js";
 import { RowpipeError } from "../core/errors.js";
 import { castCommand } from "./commands/cast.js";
 import { cleanCommand } from "./commands/clean.js";
@@ -77,6 +78,7 @@ import { testCommand } from "./commands/test.js";
 import { serveCommand } from "./commands/serve.js";
 import { reportCommand } from "./commands/report.js";
 import { fetchCommand } from "./commands/fetch.js";
+import { mcpCommand } from "./commands/mcp.js";
 // Handle broken pipe gracefully when piping to head/less
 process.stdout.on("error", (err) => {
     if (err.code === "EPIPE") {
@@ -84,10 +86,89 @@ process.stdout.on("error", (err) => {
     }
 });
 const program = new Command();
+export function getCliVersion() {
+    try {
+        const pkgUrl = new URL("../../package.json", import.meta.url);
+        if (existsSync(pkgUrl)) {
+            const pkg = JSON.parse(readFileSync(pkgUrl, "utf-8"));
+            if (pkg.version)
+                return pkg.version;
+        }
+    }
+    catch {
+        // fallback
+    }
+    return VERSION;
+}
+export function isStdinInteractive() {
+    if (process.stdin.isTTY) {
+        return true;
+    }
+    try {
+        const stat = fstatSync(0);
+        if (stat.isCharacterDevice() && !stat.isFIFO() && !stat.isFile() && !stat.isSocket()) {
+            return true;
+        }
+    }
+    catch {
+        // ignore
+    }
+    return false;
+}
+const currentVersion = getCliVersion();
 program
     .name("rowpipe")
-    .description("Stream-first tabular data toolkit for CSV, TSV, JSON, JSONL, XLSX, Parquet, and Markdown")
-    .version("2.10.0");
+    .description("Stream-first tabular data toolkit for CSV, TSV, JSON, JSONL, XLSX, Apache Parquet, and Markdown")
+    .version(currentVersion, "-v, --version", "Output the current version")
+    .option("-V", "Output the current version", () => {
+    console.log(currentVersion);
+    process.exit(0);
+});
+program
+    .command("version")
+    .description("Output version information")
+    .action(() => {
+    console.log(currentVersion);
+    process.exit(0);
+});
+program.hook("preAction", (thisCommand, actionCommand) => {
+    const isRoot = actionCommand === program || actionCommand.name() === "rowpipe";
+    const firstArg = actionCommand.args[0];
+    // If explicit empty string argument was passed (e.g. rowpipe "" or rowpipe inspect "")
+    if (firstArg !== undefined && typeof firstArg === "string" && firstArg.trim() === "") {
+        actionCommand.outputHelp();
+        process.exit(0);
+    }
+    // If interactive terminal (no piped stdin data), prevent hanging waiting for stdin
+    if (isStdinInteractive()) {
+        if (isRoot) {
+            if (!firstArg || firstArg === "-") {
+                actionCommand.outputHelp();
+                process.exit(0);
+            }
+        }
+        else {
+            const nonStreamCommands = new Set([
+                "db",
+                "diff",
+                "diff-files",
+                "join",
+                "fuzzy-join",
+                "concat",
+                "generate",
+                "completion",
+                "version",
+                "fetch",
+            ]);
+            if (!nonStreamCommands.has(actionCommand.name())) {
+                if (!firstArg || firstArg === "-") {
+                    actionCommand.outputHelp();
+                    process.exit(0);
+                }
+            }
+        }
+    }
+});
 // Global & Pipeline options on root command
 program
     .argument("[input]", "Input dataset file path or '-' for stdin")
@@ -149,10 +230,16 @@ program
     .option("--on-error <strategy>", "Error handling strategy: abort (default), skip, or log", "abort")
     .option("--bad-rows-log <file>", "Write malformed or rejected rows to dead-letter log file")
     .action(async (input = "-", cmdOptions) => {
-    // If running root command with no input file and stdin is an interactive TTY, display help
-    if (input === "-" && process.stdin.isTTY) {
-        program.outputHelp();
-        return;
+    const rawInput = typeof input === "string" ? input.trim() : "";
+    if (!rawInput || rawInput === "-") {
+        if (isStdinInteractive()) {
+            program.outputHelp();
+            return;
+        }
+        input = "-";
+    }
+    else {
+        input = rawInput;
     }
     await unifiedPipelineCommand(input, cmdOptions);
 });
@@ -1569,6 +1656,14 @@ program
         opts.maxPages = parseInt(cmdOptions.maxPages, 10);
     await fetchCommand(url, opts);
 });
+// 65. mcp
+program
+    .command("mcp [args...]")
+    .description("Model Context Protocol (MCP) server for Claude Desktop, Cursor, Antigravity, and AI agents")
+    .allowUnknownOption(true)
+    .action(async (args) => {
+    await mcpCommand(args || []);
+});
 async function main() {
     if (process.argv.length <= 2 && process.stdin.isTTY) {
         program.outputHelp();