parquetlens 0.1.0 → 0.2.0

package/dist/chunk-NRRDNC7S.js

@@ -0,0 +1,485 @@
+import { createRequire } from 'module';
+import { fileURLToPath } from 'url';
+import { dirname } from 'path';
+const require = createRequire(import.meta.url);
+const __filename = fileURLToPath(import.meta.url);
+const __dirname = dirname(__filename);
+
+// ../../packages/parquet-reader/dist/index.js
+import { createWriteStream, readFileSync, rmSync } from "fs";
+import { promises as fs } from "fs";
+import { randomUUID } from "crypto";
+import { spawnSync } from "child_process";
+import { Buffer } from "buffer";
+import { createRequire as nodeCreateRequire } from "module";
+import { tmpdir } from "os";
+import path from "path";
+import { pipeline } from "stream/promises";
+import {
+  DuckDBAccessMode,
+  DuckDBDataProtocol,
+  FileFlags,
+  NODE_RUNTIME,
+  VoidLogger,
+  createDuckDB,
+  failWith,
+  readString
+} from "@duckdb/duckdb-wasm/blocking";
+function resolveParquetUrl(input) {
+  if (input.startsWith("hf://")) {
+    return resolveHuggingFaceUrl(input);
+  }
+  if (input.startsWith("http://") || input.startsWith("https://")) {
+    return { url: input };
+  }
+  return null;
+}
+function resolveHuggingFaceUrl(input) {
+  const match = input.match(/^hf:\/\/(datasets|models)\/([^@\/]+)\/([^@\/]+)(?:@([^\/]+))?\/(.+)$/);
+  if (!match) {
+    throw new Error(`Invalid hf:// URL: ${input}`);
+  }
+  const [, type, user, repo, branch = "main", filePath] = match;
+  return {
+    url: `https://huggingface.co/${type}/${user}/${repo}/resolve/${branch}/${filePath}`
+  };
+}
+var duckDbPromise = null;
+var httpRuntimePatched = false;
+var httpBuffers = /* @__PURE__ */ new Map();
+async function getDuckDb() {
+  if (!duckDbPromise) {
+    duckDbPromise = (async () => {
+      ensureHttpRuntimeSupport();
+      const bundles = getDuckDbBundles();
+      const db = await createDuckDB(bundles, new VoidLogger(), NODE_RUNTIME);
+      await db.instantiate();
+      db.open({
+        accessMode: DuckDBAccessMode.READ_WRITE,
+        filesystem: {
+          allowFullHTTPReads: true
+        }
+      });
+      return db;
+    })();
+  }
+  return duckDbPromise;
+}
+function getDuckDbBundles() {
+  const localRequire = nodeCreateRequire(import.meta.url);
+  const mvpModule = localRequire.resolve("@duckdb/duckdb-wasm/dist/duckdb-mvp.wasm");
+  const mvpWorker = localRequire.resolve("@duckdb/duckdb-wasm/dist/duckdb-node-mvp.worker.cjs");
+  const ehModule = localRequire.resolve("@duckdb/duckdb-wasm/dist/duckdb-eh.wasm");
+  const ehWorker = localRequire.resolve("@duckdb/duckdb-wasm/dist/duckdb-node-eh.worker.cjs");
+  return {
+    mvp: {
+      mainModule: mvpModule,
+      mainWorker: mvpWorker
+    },
+    eh: {
+      mainModule: ehModule,
+      mainWorker: ehWorker
+    }
+  };
+}
+function ensureHttpRuntimeSupport() {
+  if (httpRuntimePatched) {
+    return;
+  }
+  httpRuntimePatched = true;
+  const nodeOpenFile = NODE_RUNTIME.openFile;
+  const nodeReadFile = NODE_RUNTIME.readFile;
+  const nodeCheckFile = NODE_RUNTIME.checkFile.bind(NODE_RUNTIME);
+  const nodeGlob = NODE_RUNTIME.glob.bind(NODE_RUNTIME);
+  const nodeCloseFile = NODE_RUNTIME.closeFile.bind(NODE_RUNTIME);
+  const nodeGetLastModified = NODE_RUNTIME.getLastFileModificationTime.bind(NODE_RUNTIME);
+  NODE_RUNTIME.openFile = (mod, fileId, flags) => {
+    const file = NODE_RUNTIME.resolveFileInfo(mod, fileId);
+    if (!file || file.dataProtocol !== DuckDBDataProtocol.HTTP) {
+      return nodeOpenFile(mod, fileId, flags);
+    }
+    if (flags & FileFlags.FILE_FLAGS_WRITE || flags & FileFlags.FILE_FLAGS_APPEND) {
+      failWith(mod, `Opening file ${file.fileName} failed: HTTP writes are not supported`);
+      return 0;
+    }
+    if (!(flags & FileFlags.FILE_FLAGS_READ)) {
+      failWith(mod, `Opening file ${file.fileName} failed: unsupported file flags: ${flags}`);
+      return 0;
+    }
+    if (!file.dataUrl) {
+      failWith(mod, `Opening file ${file.fileName} failed: missing data URL`);
+      return 0;
+    }
+    const allowFull = file.allowFullHttpReads ?? true;
+    const forceFull = file.forceFullHttpReads ?? false;
+    if (!forceFull) {
+      try {
+        const probe = requestHttpRange(file.dataUrl, 0, 0);
+        if (probe.status === 206) {
+          const total = parseContentRangeTotal(probe.headers["content-range"]) ?? parseContentLength(probe.headers["content-length"]);
+          if (total !== null) {
+            return buildOpenResult(mod, total, 0);
+          }
+        }
+        if (probe.status === 200 && allowFull) {
+          const dataPtr = writeResponseToHeap(mod, probe.bytes);
+          httpBuffers.set(fileId, { dataPtr, size: probe.bytes.length });
+          return buildOpenResult(mod, probe.bytes.length, dataPtr);
+        }
+      } catch (error) {
+        if (!allowFull) {
+          failWith(mod, `Opening file ${file.fileName} failed: ${String(error)}`);
+          return 0;
+        }
+      }
+    }
+    if (allowFull) {
+      try {
+        const full = requestHttp(file.dataUrl);
+        if (full.status === 200) {
+          const dataPtr = writeResponseToHeap(mod, full.bytes);
+          httpBuffers.set(fileId, { dataPtr, size: full.bytes.length });
+          return buildOpenResult(mod, full.bytes.length, dataPtr);
+        }
+      } catch (error) {
+        failWith(mod, `Opening file ${file.fileName} failed: ${String(error)}`);
+        return 0;
+      }
+    }
+    failWith(mod, `Opening file ${file.fileName} failed: HTTP range requests unavailable`);
+    return 0;
+  };
+  NODE_RUNTIME.readFile = (mod, fileId, buffer, bytes, location) => {
+    if (bytes === 0) {
+      return 0;
+    }
+    const file = NODE_RUNTIME.resolveFileInfo(mod, fileId);
+    if (!file || file.dataProtocol !== DuckDBDataProtocol.HTTP) {
+      return nodeReadFile(mod, fileId, buffer, bytes, location);
+    }
+    const cached = httpBuffers.get(fileId);
+    if (cached) {
+      const sliceStart = Math.max(0, location);
+      const sliceEnd = Math.min(cached.size, location + bytes);
+      const length = Math.max(0, sliceEnd - sliceStart);
+      if (length > 0) {
+        const src = mod.HEAPU8.subarray(cached.dataPtr + sliceStart, cached.dataPtr + sliceEnd);
+        mod.HEAPU8.set(src, buffer);
+      }
+      return length;
+    }
+    if (!file.dataUrl) {
+      failWith(mod, `Reading file ${file.fileName} failed: missing data URL`);
+      return 0;
+    }
+    try {
+      const response = requestHttpRange(file.dataUrl, location, location + bytes - 1);
+      if (response.status === 206 || response.status === 200 && location === 0) {
+        const length = Math.min(bytes, response.bytes.length);
+        if (length > 0) {
+          mod.HEAPU8.set(response.bytes.subarray(0, length), buffer);
+        }
+        return length;
+      }
+      failWith(mod, `Reading file ${file.fileName} failed with HTTP ${response.status}`);
+      return 0;
+    } catch (error) {
+      failWith(mod, `Reading file ${file.fileName} failed: ${String(error)}`);
+      return 0;
+    }
+  };
+  NODE_RUNTIME.checkFile = (mod, pathPtr, pathLen) => {
+    const path2 = readString(mod, pathPtr, pathLen);
+    if (isHttpUrl(path2)) {
+      const response = requestHttpHead(path2);
+      return response.status === 200 || response.status === 206;
+    }
+    return nodeCheckFile(mod, pathPtr, pathLen);
+  };
+  NODE_RUNTIME.glob = (mod, pathPtr, pathLen) => {
+    const path2 = readString(mod, pathPtr, pathLen);
+    if (isHttpUrl(path2)) {
+      const response = requestHttpHead(path2);
+      if (response.status === 200 || response.status === 206) {
+        mod.ccall("duckdb_web_fs_glob_add_path", null, ["string"], [path2]);
+      }
+      return;
+    }
+    return nodeGlob(mod, pathPtr, pathLen);
+  };
+  NODE_RUNTIME.closeFile = (mod, fileId) => {
+    const cached = httpBuffers.get(fileId);
+    if (cached) {
+      if (cached.dataPtr) {
+        mod._free(cached.dataPtr);
+      }
+      httpBuffers.delete(fileId);
+    }
+    nodeCloseFile(mod, fileId);
+  };
+  NODE_RUNTIME.getLastFileModificationTime = (mod, fileId) => {
+    const file = NODE_RUNTIME.resolveFileInfo(mod, fileId);
+    if (file?.dataProtocol === DuckDBDataProtocol.HTTP) {
+      return Date.now() / 1e3;
+    }
+    return nodeGetLastModified(mod, fileId);
+  };
+}
+function isHttpUrl(value) {
+  return value.startsWith("http://") || value.startsWith("https://");
+}
+function buildOpenResult(mod, size, dataPtr) {
+  const result = mod._malloc(2 * 8);
+  mod.HEAPF64[(result >> 3) + 0] = +size;
+  mod.HEAPF64[(result >> 3) + 1] = dataPtr;
+  return result;
+}
+function writeResponseToHeap(mod, bytes) {
+  const dataPtr = mod._malloc(bytes.byteLength);
+  mod.HEAPU8.set(bytes, dataPtr);
+  return dataPtr;
+}
+function parseContentRangeTotal(contentRange) {
+  if (!contentRange) {
+    return null;
+  }
+  const [, total] = contentRange.split("/");
+  if (!total) {
+    return null;
+  }
+  const parsed = Number.parseInt(total, 10);
+  return Number.isFinite(parsed) ? parsed : null;
+}
+function parseContentLength(contentLength) {
+  if (!contentLength) {
+    return null;
+  }
+  const parsed = Number.parseInt(contentLength, 10);
+  return Number.isFinite(parsed) ? parsed : null;
+}
+function requestHttp(url) {
+  return requestCurl([url]);
+}
+function requestHttpHead(url) {
+  return requestCurl(["-I", url]);
+}
+function requestHttpRange(url, start, end) {
+  return requestCurl(["-r", `${start}-${end}`, url]);
+}
+function requestCurl(args) {
+  const tempPath = path.join(tmpdir(), `parquetlens-http-${randomUUID()}`);
+  try {
+    const result = spawnSync("curl", ["-sS", "-L", "-D", "-", "-o", tempPath, ...args], {
+      encoding: "buffer",
+      maxBuffer: 4 * 1024 * 1024
+    });
+    if (result.error) {
+      if (result.error.code === "ENOENT") {
+        throw new Error("curl not found (required for HTTP range reads)");
+      }
+      throw result.error;
+    }
+    if (result.status !== 0) {
+      const stderr = result.stderr?.toString("utf8").trim();
+      throw new Error(stderr || "curl failed");
+    }
+    const body = readFileSync(tempPath);
+    return parseCurlResponse(Buffer.from(result.stdout ?? []), body);
+  } finally {
+    rmSync(tempPath, { force: true });
+  }
+}
+function parseCurlResponse(headersBuffer, body) {
+  const headerBlob = headersBuffer.toString("latin1");
+  const blocks = headerBlob.split(/\r\n\r\n/).filter(Boolean);
+  const lastBlock = blocks[blocks.length - 1] ?? "";
+  const lines = lastBlock.split(/\r\n/).filter(Boolean);
+  const statusLine = lines.shift() ?? "";
+  const statusToken = statusLine.split(" ")[1] ?? "";
+  const status = Number.parseInt(statusToken, 10);
+  const headers = {};
+  for (const line of lines) {
+    const index = line.indexOf(":");
+    if (index === -1) {
+      continue;
+    }
+    const key = line.slice(0, index).trim().toLowerCase();
+    const value = line.slice(index + 1).trim();
+    headers[key] = value;
+  }
+  return {
+    status: Number.isFinite(status) ? status : 0,
+    bytes: new Uint8Array(body),
+    headers
+  };
+}
+async function openParquetSourceFromPath(filePath) {
+  const db = await getDuckDb();
+  const conn = db.connect();
+  const fileName = buildDuckDbFileName(filePath);
+  db.registerFileURL(fileName, filePath, DuckDBDataProtocol.NODE_FS, true);
+  return createParquetSource(db, conn, fileName);
+}
+async function openParquetSourceFromUrl(input) {
+  const resolved = resolveParquetUrl(input);
+  if (!resolved) {
+    throw new Error("Not a URL");
+  }
+  const db = await getDuckDb();
+  const conn = db.connect();
+  const fileName = buildDuckDbFileName(resolved.url);
+  db.registerFileURL(fileName, resolved.url, DuckDBDataProtocol.HTTP, true);
+  return createParquetSource(db, conn, fileName);
+}
+async function openParquetSource(input) {
+  const resolved = resolveParquetUrl(input);
+  if (resolved) {
+    return openParquetSourceFromUrl(input);
+  }
+  return openParquetSourceFromPath(input);
+}
+async function readParquetTableFromPath(filePath, options) {
+  const source = await openParquetSourceFromPath(filePath);
+  try {
+    return await source.readTable(options);
+  } finally {
+    await source.close();
+  }
+}
+async function readParquetTableFromUrl(input, options) {
+  const source = await openParquetSourceFromUrl(input);
+  try {
+    return await source.readTable(options);
+  } finally {
+    await source.close();
+  }
+}
+async function bufferStdinToTempFile(filenameHint = "stdin.parquet") {
+  const tempDir = await fs.mkdtemp(path.join(tmpdir(), "parquetlens-"));
+  const safeName = filenameHint.replace(/[\\/]/g, "_");
+  const filePath = path.join(tempDir, `${randomUUID()}-${safeName}`);
+  const writeStream = createWriteStream(filePath);
+  await pipeline(process.stdin, writeStream);
+  return {
+    path: filePath,
+    cleanup: async () => {
+      await fs.rm(tempDir, { recursive: true, force: true });
+    }
+  };
+}
+async function readParquetTableFromStdin(filenameHint = "stdin.parquet", options) {
+  const temp = await bufferStdinToTempFile(filenameHint);
+  try {
+    return await readParquetTableFromPath(temp.path, options);
+  } finally {
+    await temp.cleanup();
+  }
+}
+function createParquetSource(db, conn, fileName) {
+  let metadataPromise = null;
+  return {
+    readTable: async (options) => {
+      const query = buildSelectQuery(fileName, options);
+      return conn.query(query);
+    },
+    readMetadata: () => {
+      if (!metadataPromise) {
+        metadataPromise = readParquetMetadata(conn, fileName);
+      }
+      return metadataPromise;
+    },
+    close: async () => {
+      conn.close();
+      db.dropFile(fileName);
+    }
+  };
+}
+function buildDuckDbFileName(input) {
+  const suffix = path.extname(input) || ".parquet";
+  return `parquetlens-${randomUUID()}${suffix}`;
+}
+function buildSelectQuery(fileName, options) {
+  const columns = options?.columns && options.columns.length > 0 ? options.columns : null;
+  const selectList = columns ? columns.map(quoteIdentifier).join(", ") : "*";
+  const limit = options?.limit;
+  const offset = options?.offset;
+  let query = `select ${selectList} from read_parquet(${quoteLiteral(fileName)})`;
+  if (typeof limit === "number") {
+    query += ` limit ${Math.max(0, limit)}`;
+  }
+  if (typeof offset === "number" && offset > 0) {
+    query += ` offset ${Math.max(0, offset)}`;
+  }
+  return query;
+}
+function quoteIdentifier(value) {
+  return `"${value.replace(/"/g, '""')}"`;
+}
+function quoteLiteral(value) {
+  return `'${value.replace(/'/g, "''")}'`;
+}
+async function readParquetMetadata(conn, fileName) {
+  const metadataRows = tableToObjects(
+    conn.query(`select * from parquet_file_metadata(${quoteLiteral(fileName)})`)
+  );
+  const kvRows = tableToObjects(
+    conn.query(`select * from parquet_kv_metadata(${quoteLiteral(fileName)})`)
+  );
+  const createdByRaw = metadataRows[0]?.created_by ?? metadataRows[0]?.createdBy ?? null;
+  const keyValueMetadata = {};
+  for (const row of kvRows) {
+    const key = row.key ?? row.key_name ?? row.name;
+    if (typeof key !== "string" || key.length === 0) {
+      continue;
+    }
+    keyValueMetadata[key] = row.value ?? row.val ?? "";
+  }
+  return {
+    createdBy: normalizeMetadataValue(createdByRaw),
+    keyValueMetadata: normalizeMetadataValues(keyValueMetadata)
+  };
+}
+function tableToObjects(table) {
+  const fields = table.schema.fields.map((field) => field.name);
+  const rows = [];
+  for (const batch of table.batches) {
+    const vectors = fields.map((_, index) => batch.getChildAt(index));
+    for (let rowIndex = 0; rowIndex < batch.numRows; rowIndex += 1) {
+      const row = {};
+      for (let colIndex = 0; colIndex < fields.length; colIndex += 1) {
+        row[fields[colIndex]] = vectors[colIndex]?.get(rowIndex);
+      }
+      rows.push(row);
+    }
+  }
+  return rows;
+}
+function normalizeMetadataValue(value) {
+  if (value === null || value === void 0) {
+    return void 0;
+  }
+  if (value instanceof Uint8Array) {
+    return Buffer.from(value).toString("utf8");
+  }
+  if (typeof value === "string") {
+    return value;
+  }
+  return String(value);
+}
+function normalizeMetadataValues(input) {
+  const normalized = {};
+  for (const [key, value] of Object.entries(input)) {
+    const normalizedValue = normalizeMetadataValue(value);
+    normalized[key] = normalizedValue ?? "";
+  }
+  return normalized;
+}
+
+export {
+  resolveParquetUrl,
+  openParquetSource,
+  readParquetTableFromPath,
+  readParquetTableFromUrl,
+  readParquetTableFromStdin
+};
+//# sourceMappingURL=chunk-NRRDNC7S.js.map

package/dist/chunk-NRRDNC7S.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../packages/parquet-reader/src/index.ts","../../../packages/parquet-reader/src/urls.ts"],"sourcesContent":["import { createWriteStream, readFileSync, rmSync } from \"node:fs\";\nimport { promises as fs } from \"node:fs\";\nimport { randomUUID } from \"node:crypto\";\nimport { spawnSync } from \"node:child_process\";\nimport { Buffer } from \"node:buffer\";\nimport { createRequire as nodeCreateRequire } from \"node:module\";\nimport { tmpdir } from \"node:os\";\nimport path from \"node:path\";\nimport { pipeline } from \"node:stream/promises\";\n\nimport type { Table } from \"apache-arrow\";\nimport {\n  DuckDBAccessMode,\n  DuckDBBundles,\n  DuckDBConnection,\n  DuckDBDataProtocol,\n  DuckDBBindings,\n  FileFlags,\n  NODE_RUNTIME,\n  VoidLogger,\n  createDuckDB,\n  failWith,\n  readString,\n} from \"@duckdb/duckdb-wasm/blocking\";\nimport type { DuckDBModule } from \"@duckdb/duckdb-wasm/blocking\";\n\nimport { resolveParquetUrl } from \"./urls.js\";\n\nexport type TempParquetFile = {\n  path: string;\n  cleanup: () => Promise<void>;\n};\n\nexport type ParquetReadOptions = {\n  batchSize?: number;\n  columns?: string[];\n  limit?: number;\n  offset?: number;\n  rowGroups?: number[];\n};\n\nexport type ParquetFileMetadata = {\n  createdBy?: string;\n  keyValueMetadata: Record<string, string>;\n};\n\nexport type ParquetSource = {\n  readTable: (options?: ParquetReadOptions) => Promise<Table>;\n  readMetadata: () => Promise<ParquetFileMetadata>;\n  close: () => Promise<void>;\n};\n\nlet duckDbPromise: Promise<DuckDBBindings> | null = null;\nlet httpRuntimePatched = false;\n\ntype HttpBuffer = {\n  dataPtr: number;\n  size: number;\n};\n\nconst httpBuffers = new Map<number, HttpBuffer>();\n\ntype WasmModule = DuckDBModule & {\n  HEAPU8: Uint8Array;\n  HEAPF64: Float64Array;\n  _malloc: (size: number) => number;\n  _free: (ptr: number) => void;\n};\n\nasync function getDuckDb(): Promise<DuckDBBindings> {\n  if (!duckDbPromise) {\n    duckDbPromise = (async () => {\n      ensureHttpRuntimeSupport();\n      const bundles = getDuckDbBundles();\n      const db = await createDuckDB(bundles, new VoidLogger(), NODE_RUNTIME);\n      await db.instantiate();\n      db.open({\n        accessMode: DuckDBAccessMode.READ_WRITE,\n        filesystem: {\n          allowFullHTTPReads: true,\n        },\n      });\n      return db;\n    })();\n  }\n\n  return duckDbPromise;\n}\n\nfunction getDuckDbBundles(): DuckDBBundles {\n  const localRequire = nodeCreateRequire(import.meta.url);\n  const mvpModule = localRequire.resolve(\"@duckdb/duckdb-wasm/dist/duckdb-mvp.wasm\");\n  const mvpWorker = localRequire.resolve(\"@duckdb/duckdb-wasm/dist/duckdb-node-mvp.worker.cjs\");\n  const ehModule = localRequire.resolve(\"@duckdb/duckdb-wasm/dist/duckdb-eh.wasm\");\n  const ehWorker = localRequire.resolve(\"@duckdb/duckdb-wasm/dist/duckdb-node-eh.worker.cjs\");\n\n  return {\n    mvp: {\n      mainModule: mvpModule,\n      mainWorker: mvpWorker,\n    },\n    eh: {\n      mainModule: ehModule,\n      mainWorker: ehWorker,\n    },\n  };\n}\n\nfunction ensureHttpRuntimeSupport(): void {\n  if (httpRuntimePatched) {\n    return;\n  }\n  httpRuntimePatched = true;\n\n  const nodeOpenFile = NODE_RUNTIME.openFile as unknown as (\n    mod: WasmModule,\n    fileId: number,\n    flags: FileFlags,\n  ) => number;\n  const nodeReadFile = NODE_RUNTIME.readFile as unknown as (\n    mod: WasmModule,\n    fileId: number,\n    buffer: number,\n    bytes: number,\n    location: number,\n  ) => number;\n  const nodeCheckFile = NODE_RUNTIME.checkFile.bind(NODE_RUNTIME);\n  const nodeGlob = NODE_RUNTIME.glob.bind(NODE_RUNTIME);\n  const nodeCloseFile = NODE_RUNTIME.closeFile.bind(NODE_RUNTIME);\n  const nodeGetLastModified = NODE_RUNTIME.getLastFileModificationTime.bind(NODE_RUNTIME);\n\n  NODE_RUNTIME.openFile = (mod: WasmModule, fileId: number, flags: FileFlags): number => {\n    const file = NODE_RUNTIME.resolveFileInfo(mod, fileId);\n    if (!file || file.dataProtocol !== DuckDBDataProtocol.HTTP) {\n      return nodeOpenFile(mod, fileId, flags);\n    }\n\n    if (flags & FileFlags.FILE_FLAGS_WRITE || flags & FileFlags.FILE_FLAGS_APPEND) {\n      failWith(mod, `Opening file ${file.fileName} failed: HTTP writes are not supported`);\n      return 0;\n    }\n\n    if (!(flags & FileFlags.FILE_FLAGS_READ)) {\n      failWith(mod, `Opening file ${file.fileName} failed: unsupported file flags: ${flags}`);\n      return 0;\n    }\n\n    if (!file.dataUrl) {\n      failWith(mod, `Opening file ${file.fileName} failed: missing data URL`);\n      return 0;\n    }\n\n    const allowFull = file.allowFullHttpReads ?? true;\n    const forceFull = file.forceFullHttpReads ?? false;\n\n    if (!forceFull) {\n      try {\n        const probe = requestHttpRange(file.dataUrl, 0, 0);\n        if (probe.status === 206) {\n          const total =\n            parseContentRangeTotal(probe.headers[\"content-range\"]) ??\n            parseContentLength(probe.headers[\"content-length\"]);\n          if (total !== null) {\n            return buildOpenResult(mod, total, 0);\n          }\n        }\n\n        if (probe.status === 200 && allowFull) {\n          const dataPtr = writeResponseToHeap(mod, probe.bytes);\n          httpBuffers.set(fileId, { dataPtr, size: probe.bytes.length });\n          return buildOpenResult(mod, probe.bytes.length, dataPtr);\n        }\n      } catch (error) {\n        if (!allowFull) {\n          failWith(mod, `Opening file ${file.fileName} failed: ${String(error)}`);\n          return 0;\n        }\n      }\n    }\n\n    if (allowFull) {\n      try {\n        const full = requestHttp(file.dataUrl);\n        if (full.status === 200) {\n          const dataPtr = writeResponseToHeap(mod, full.bytes);\n          httpBuffers.set(fileId, { dataPtr, size: full.bytes.length });\n          return buildOpenResult(mod, full.bytes.length, dataPtr);\n        }\n      } catch (error) {\n        failWith(mod, `Opening file ${file.fileName} failed: ${String(error)}`);\n        return 0;\n      }\n    }\n\n    failWith(mod, `Opening file ${file.fileName} failed: HTTP range requests unavailable`);\n    return 0;\n  };\n\n  NODE_RUNTIME.readFile = (\n    mod: WasmModule,\n    fileId: number,\n    buffer: number,\n    bytes: number,\n    location: number,\n  ): number => {\n    if (bytes === 0) {\n      return 0;\n    }\n\n    const file = NODE_RUNTIME.resolveFileInfo(mod, fileId);\n    if (!file || file.dataProtocol !== DuckDBDataProtocol.HTTP) {\n      return nodeReadFile(mod, fileId, buffer, bytes, location);\n    }\n\n    const cached = httpBuffers.get(fileId);\n    if (cached) {\n      const sliceStart = Math.max(0, location);\n      const sliceEnd = Math.min(cached.size, location + bytes);\n      const length = Math.max(0, sliceEnd - sliceStart);\n      if (length > 0) {\n        const src = mod.HEAPU8.subarray(cached.dataPtr + sliceStart, cached.dataPtr + sliceEnd);\n        mod.HEAPU8.set(src, buffer);\n      }\n      return length;\n    }\n\n    if (!file.dataUrl) {\n      failWith(mod, `Reading file ${file.fileName} failed: missing data URL`);\n      return 0;\n    }\n\n    try {\n      const response = requestHttpRange(file.dataUrl, location, location + bytes - 1);\n      if (response.status === 206 || (response.status === 200 && location === 0)) {\n        const length = Math.min(bytes, response.bytes.length);\n        if (length > 0) {\n          mod.HEAPU8.set(response.bytes.subarray(0, length), buffer);\n        }\n        return length;\n      }\n\n      failWith(mod, `Reading file ${file.fileName} failed with HTTP ${response.status}`);\n      return 0;\n    } catch (error) {\n      failWith(mod, `Reading file ${file.fileName} failed: ${String(error)}`);\n      return 0;\n    }\n  };\n\n  NODE_RUNTIME.checkFile = (mod: DuckDBModule, pathPtr: number, pathLen: number): boolean => {\n    const path = readString(mod, pathPtr, pathLen);\n    if (isHttpUrl(path)) {\n      const response = requestHttpHead(path);\n      return response.status === 200 || response.status === 206;\n    }\n    return nodeCheckFile(mod, pathPtr, pathLen);\n  };\n\n  NODE_RUNTIME.glob = (mod: DuckDBModule, pathPtr: number, pathLen: number): void => {\n    const path = readString(mod, pathPtr, pathLen);\n    if (isHttpUrl(path)) {\n      const response = requestHttpHead(path);\n      if (response.status === 200 || response.status === 206) {\n        mod.ccall(\"duckdb_web_fs_glob_add_path\", null, [\"string\"], [path]);\n      }\n      return;\n    }\n\n    return nodeGlob(mod, pathPtr, pathLen);\n  };\n\n  NODE_RUNTIME.closeFile = (mod: DuckDBModule, fileId: number): void => {\n    const cached = httpBuffers.get(fileId);\n    if (cached) {\n      if (cached.dataPtr) {\n        (mod as WasmModule)._free(cached.dataPtr);\n      }\n      httpBuffers.delete(fileId);\n    }\n    nodeCloseFile(mod, fileId);\n  };\n\n  NODE_RUNTIME.getLastFileModificationTime = (mod: DuckDBModule, fileId: number): number => {\n    const file = NODE_RUNTIME.resolveFileInfo(mod, fileId);\n    if (file?.dataProtocol === DuckDBDataProtocol.HTTP) {\n      return Date.now() / 1000;\n    }\n    return nodeGetLastModified(mod, fileId);\n  };\n}\n\nfunction isHttpUrl(value: string): boolean {\n  return value.startsWith(\"http://\") || value.startsWith(\"https://\");\n}\n\nfunction buildOpenResult(mod: WasmModule, size: number, dataPtr: number): number {\n  const result = mod._malloc(2 * 8);\n  mod.HEAPF64[(result >> 3) + 0] = +size;\n  mod.HEAPF64[(result >> 3) + 1] = dataPtr;\n  return result;\n}\n\nfunction writeResponseToHeap(mod: WasmModule, bytes: Uint8Array): number {\n  const dataPtr = mod._malloc(bytes.byteLength);\n  mod.HEAPU8.set(bytes, dataPtr);\n  return dataPtr;\n}\n\nfunction parseContentRangeTotal(contentRange: string | null): number | null {\n  if (!contentRange) {\n    return null;\n  }\n  const [, total] = contentRange.split(\"/\");\n  if (!total) {\n    return null;\n  }\n  const parsed = Number.parseInt(total, 10);\n  return Number.isFinite(parsed) ? parsed : null;\n}\n\nfunction parseContentLength(contentLength: string | null): number | null {\n  if (!contentLength) {\n    return null;\n  }\n  const parsed = Number.parseInt(contentLength, 10);\n  return Number.isFinite(parsed) ? parsed : null;\n}\n\ntype HttpResponse = {\n  status: number;\n  bytes: Uint8Array;\n  headers: Record<string, string>;\n};\n\nfunction requestHttp(url: string): HttpResponse {\n  return requestCurl([url]);\n}\n\nfunction requestHttpHead(url: string): HttpResponse {\n  return requestCurl([\"-I\", url]);\n}\n\nfunction requestHttpRange(url: string, start: number, end: number): HttpResponse {\n  return requestCurl([\"-r\", `${start}-${end}`, url]);\n}\n\nfunction requestCurl(args: string[]): HttpResponse {\n  const tempPath = path.join(tmpdir(), `parquetlens-http-${randomUUID()}`);\n  try {\n    const result = spawnSync(\"curl\", [\"-sS\", \"-L\", \"-D\", \"-\", \"-o\", tempPath, ...args], {\n      encoding: \"buffer\",\n      maxBuffer: 4 * 1024 * 1024,\n    });\n\n    if (result.error) {\n      if ((result.error as NodeJS.ErrnoException).code === \"ENOENT\") {\n        throw new Error(\"curl not found (required for HTTP range reads)\");\n      }\n      throw result.error;\n    }\n    if (result.status !== 0) {\n      const stderr = result.stderr?.toString(\"utf8\").trim();\n      throw new Error(stderr || \"curl failed\");\n    }\n\n    const body = readFileSync(tempPath);\n    return parseCurlResponse(Buffer.from(result.stdout ?? []), body);\n  } finally {\n    rmSync(tempPath, { force: true });\n  }\n}\n\nfunction parseCurlResponse(headersBuffer: Buffer, body: Buffer): HttpResponse {\n  const headerBlob = headersBuffer.toString(\"latin1\");\n  const blocks = headerBlob.split(/\\r\\n\\r\\n/).filter(Boolean);\n  const lastBlock = blocks[blocks.length - 1] ?? \"\";\n  const lines = lastBlock.split(/\\r\\n/).filter(Boolean);\n  const statusLine = lines.shift() ?? \"\";\n  const statusToken = statusLine.split(\" \")[1] ?? \"\";\n  const status = Number.parseInt(statusToken, 10);\n  const headers: Record<string, string> = {};\n\n  for (const line of lines) {\n    const index = line.indexOf(\":\");\n    if (index === -1) {\n      continue;\n    }\n    const key = line.slice(0, index).trim().toLowerCase();\n    const value = line.slice(index + 1).trim();\n    headers[key] = value;\n  }\n\n  return {\n    status: Number.isFinite(status) ? status : 0,\n    bytes: new Uint8Array(body),\n    headers,\n  };\n}\n\nexport async function openParquetSourceFromPath(filePath: string): Promise<ParquetSource> {\n  const db = await getDuckDb();\n  const conn = db.connect();\n  const fileName = buildDuckDbFileName(filePath);\n\n  db.registerFileURL(fileName, filePath, DuckDBDataProtocol.NODE_FS, true);\n\n  return createParquetSource(db, conn, fileName);\n}\n\nexport async function openParquetSourceFromUrl(input: string): Promise<ParquetSource> {\n  const resolved = resolveParquetUrl(input);\n  if (!resolved) {\n    throw new Error(\"Not a URL\");\n  }\n\n  const db = await getDuckDb();\n  const conn = db.connect();\n  const fileName = buildDuckDbFileName(resolved.url);\n\n  db.registerFileURL(fileName, resolved.url, DuckDBDataProtocol.HTTP, true);\n\n  return createParquetSource(db, conn, fileName);\n}\n\nexport async function openParquetSourceFromBuffer(buffer: Uint8Array): Promise<ParquetSource> {\n  const db = await getDuckDb();\n  const conn = db.connect();\n  const fileName = buildDuckDbFileName(\"buffer\");\n\n  db.registerFileBuffer(fileName, buffer);\n\n  return createParquetSource(db, conn, fileName);\n}\n\nexport async function openParquetSource(input: string): Promise<ParquetSource> {\n  const resolved = resolveParquetUrl(input);\n  if (resolved) {\n    return openParquetSourceFromUrl(input);\n  }\n\n  return openParquetSourceFromPath(input);\n}\n\nexport async function readParquetTableFromBuffer(\n  buffer: Uint8Array,\n  options?: ParquetReadOptions,\n): Promise<Table> {\n  const source = await openParquetSourceFromBuffer(buffer);\n\n  try {\n    return await source.readTable(options);\n  } finally {\n    await source.close();\n  }\n}\n\nexport async function readParquetTableFromPath(\n  filePath: string,\n  options?: ParquetReadOptions,\n): Promise<Table> {\n  const source = await openParquetSourceFromPath(filePath);\n\n  try {\n    return await source.readTable(options);\n  } finally {\n    await source.close();\n  }\n}\n\nexport async function readParquetTableFromUrl(\n  input: string,\n  options?: ParquetReadOptions,\n): Promise<Table> {\n  const source = await openParquetSourceFromUrl(input);\n\n  try {\n    return await source.readTable(options);\n  } finally {\n    await source.close();\n  }\n}\n\nexport async function readParquetMetadataFromBuffer(\n  buffer: Uint8Array,\n): Promise<ParquetFileMetadata> {\n  const source = await openParquetSourceFromBuffer(buffer);\n\n  try {\n    return await source.readMetadata();\n  } finally {\n    await source.close();\n  }\n}\n\nexport async function bufferStdinToTempFile(\n  filenameHint = \"stdin.parquet\",\n): Promise<TempParquetFile> {\n  const tempDir = await fs.mkdtemp(path.join(tmpdir(), \"parquetlens-\"));\n  const safeName = filenameHint.replace(/[\\\\/]/g, \"_\");\n  const filePath = path.join(tempDir, `${randomUUID()}-${safeName}`);\n  const writeStream = createWriteStream(filePath);\n\n  await pipeline(process.stdin, writeStream);\n\n  return {\n    path: filePath,\n    cleanup: async () => {\n      await fs.rm(tempDir, { recursive: true, force: true });\n    },\n  };\n}\n\nexport async function readParquetTableFromStdin(\n  filenameHint = \"stdin.parquet\",\n  options?: ParquetReadOptions,\n): Promise<Table> {\n  const temp = await bufferStdinToTempFile(filenameHint);\n\n  try {\n    return await readParquetTableFromPath(temp.path, options);\n  } finally {\n    await temp.cleanup();\n  }\n}\n\nfunction createParquetSource(\n  db: DuckDBBindings,\n  conn: DuckDBConnection,\n  fileName: string,\n): ParquetSource {\n  let metadataPromise: Promise<ParquetFileMetadata> | null = null;\n\n  return {\n    readTable: async (options?: ParquetReadOptions) => {\n      const query = buildSelectQuery(fileName, options);\n      return conn.query(query);\n    },\n    readMetadata: () => {\n      if (!metadataPromise) {\n        metadataPromise = readParquetMetadata(conn, fileName);\n      }\n      return metadataPromise;\n    },\n    close: async () => {\n      conn.close();\n      db.dropFile(fileName);\n    },\n  };\n}\n\nfunction buildDuckDbFileName(input: string): string {\n  const suffix = path.extname(input) || \".parquet\";\n  return `parquetlens-${randomUUID()}${suffix}`;\n}\n\nfunction buildSelectQuery(fileName: string, options?: ParquetReadOptions): string {\n  const columns = options?.columns && options.columns.length > 0 ? options.columns : null;\n  const selectList = columns ? columns.map(quoteIdentifier).join(\", \") : \"*\";\n  const limit = options?.limit;\n  const offset = options?.offset;\n\n  let query = `select ${selectList} from read_parquet(${quoteLiteral(fileName)})`;\n\n  if (typeof limit === \"number\") {\n    query += ` limit ${Math.max(0, limit)}`;\n  }\n\n  if (typeof offset === \"number\" && offset > 0) {\n    query += ` offset ${Math.max(0, offset)}`;\n  }\n\n  return query;\n}\n\nfunction quoteIdentifier(value: string): string {\n  return `\"${value.replace(/\"/g, '\"\"')}\"`;\n}\n\nfunction quoteLiteral(value: string): string {\n  return `'${value.replace(/'/g, \"''\")}'`;\n}\n\nasync function readParquetMetadata(\n  conn: DuckDBConnection,\n  fileName: string,\n): Promise<ParquetFileMetadata> {\n  const metadataRows = tableToObjects(\n    conn.query(`select * from parquet_file_metadata(${quoteLiteral(fileName)})`),\n  );\n  const kvRows = tableToObjects(\n    conn.query(`select * from parquet_kv_metadata(${quoteLiteral(fileName)})`),\n  );\n\n  const createdByRaw = metadataRows[0]?.created_by ?? metadataRows[0]?.createdBy ?? null;\n  const keyValueMetadata: Record<string, unknown> = {};\n\n  for (const row of kvRows) {\n    const key = row.key ?? row.key_name ?? row.name;\n    if (typeof key !== \"string\" || key.length === 0) {\n      continue;\n    }\n    keyValueMetadata[key] = row.value ?? row.val ?? \"\";\n  }\n\n  return {\n    createdBy: normalizeMetadataValue(createdByRaw),\n    keyValueMetadata: normalizeMetadataValues(keyValueMetadata),\n  };\n}\n\nfunction tableToObjects(table: Table): Record<string, unknown>[] {\n  const fields = table.schema.fields.map((field) => field.name);\n  const rows: Record<string, unknown>[] = [];\n\n  for (const batch of table.batches) {\n    const vectors = fields.map((_, index) => batch.getChildAt(index));\n\n    for (let rowIndex = 0; rowIndex < batch.numRows; rowIndex += 1) {\n      const row: Record<string, unknown> = {};\n\n      for (let colIndex = 0; colIndex < fields.length; colIndex += 1) {\n        row[fields[colIndex]] = vectors[colIndex]?.get(rowIndex);\n      }\n\n      rows.push(row);\n    }\n  }\n\n  return rows;\n}\n\nfunction normalizeMetadataValue(value: unknown): string | undefined {\n  if (value === null || value === undefined) {\n    return undefined;\n  }\n\n  if (value instanceof Uint8Array) {\n    return Buffer.from(value).toString(\"utf8\");\n  }\n\n  if (typeof value === \"string\") {\n    return value;\n  }\n\n  return String(value);\n}\n\nfunction normalizeMetadataValues(input: Record<string, unknown>): Record<string, string> {\n  const normalized: Record<string, string> = {};\n\n  for (const [key, value] of Object.entries(input)) {\n    const normalizedValue = normalizeMetadataValue(value);\n    normalized[key] = normalizedValue ?? \"\";\n  }\n\n  return normalized;\n}\n\nexport { resolveParquetUrl } from \"./urls.js\";\nexport type { ResolvedParquetUrl } from \"./urls.js\";\n","export type ResolvedParquetUrl = {\n  url: string;\n};\n\nexport function resolveParquetUrl(input: string): ResolvedParquetUrl | null {\n  if (input.startsWith(\"hf://\")) {\n    return resolveHuggingFaceUrl(input);\n  }\n\n  if (input.startsWith(\"http://\") || input.startsWith(\"https://\")) {\n    return { url: input };\n  }\n\n  return null;\n}\n\nfunction resolveHuggingFaceUrl(input: string): ResolvedParquetUrl {\n  const match = input.match(/^hf:\\/\\/(datasets|models)\\/([^@\\/]+)\\/([^@\\/]+)(?:@([^\\/]+))?\\/(.+)$/);\n\n  if (!match) {\n    throw new Error(`Invalid hf:// URL: ${input}`);\n  }\n\n  const [, type, user, repo, branch = \"main\", filePath] = match;\n\n  return {\n    url: `https://huggingface.co/${type}/${user}/${repo}/resolve/${branch}/${filePath}`,\n  };\n}\n"],"mappings":";;;;;;;;AAAA,SAAS,mBAAmB,cAAc,cAAc;AACxD,SAAS,YAAY,UAAU;AAC/B,SAAS,kBAAkB;AAC3B,SAAS,iBAAiB;AAC1B,SAAS,cAAc;AACvB,SAAS,iBAAiB,yBAAyB;AACnD,SAAS,cAAc;AACvB,OAAO,UAAU;AACjB,SAAS,gBAAgB;AAGzB;EACE;EAGA;EAEA;EACA;EACA;EACA;EACA;EACA;OACK;ACnBA,SAAS,kBAAkB,OAA0C;AAC1E,MAAI,MAAM,WAAW,OAAO,GAAG;AAC7B,WAAO,sBAAsB,KAAK;EACpC;AAEA,MAAI,MAAM,WAAW,SAAS,KAAK,MAAM,WAAW,UAAU,GAAG;AAC/D,WAAO,EAAE,KAAK,MAAM;EACtB;AAEA,SAAO;AACT;AAEA,SAAS,sBAAsB,OAAmC;AAChE,QAAM,QAAQ,MAAM,MAAM,sEAAsE;AAEhG,MAAI,CAAC,OAAO;AACV,UAAM,IAAI,MAAM,sBAAsB,KAAK,EAAE;EAC/C;AAEA,QAAM,CAAC,EAAE,MAAM,MAAM,MAAM,SAAS,QAAQ,QAAQ,IAAI;AAExD,SAAO;IACL,KAAK,0BAA0B,IAAI,IAAI,IAAI,IAAI,IAAI,YAAY,MAAM,IAAI,QAAQ;EACnF;AACF;ADwBA,IAAI,gBAAgD;AACpD,IAAI,qBAAqB;AAOzB,IAAM,cAAc,oBAAI,IAAwB;AAShD,eAAe,YAAqC;AAClD,MAAI,CAAC,eAAe;AAClB,qBAAiB,YAAY;AAC3B,+BAAyB;AACzB,YAAM,UAAU,iBAAiB;AACjC,YAAM,KAAK,MAAM,aAAa,SAAS,IAAI,WAAW,GAAG,YAAY;AACrE,YAAM,GAAG,YAAY;AACrB,SAAG,KAAK;QACN,YAAY,iBAAiB;QAC7B,YAAY;UACV,oBAAoB;QACtB;MACF,CAAC;AACD,aAAO;IACT,GAAG;EACL;AAEA,SAAO;AACT;AAEA,SAAS,mBAAkC;AACzC,QAAM,eAAe,kBAAkB,YAAY,GAAG;AACtD,QAAM,YAAY,aAAa,QAAQ,0CAA0C;AACjF,QAAM,YAAY,aAAa,QAAQ,qDAAqD;AAC5F,QAAM,WAAW,aAAa,QAAQ,yCAAyC;AAC/E,QAAM,WAAW,aAAa,QAAQ,oDAAoD;AAE1F,SAAO;IACL,KAAK;MACH,YAAY;MACZ,YAAY;IACd;IACA,IAAI;MACF,YAAY;MACZ,YAAY;IACd;EACF;AACF;AAEA,SAAS,2BAAiC;AACxC,MAAI,oBAAoB;AACtB;EACF;AACA,uBAAqB;AAErB,QAAM,eAAe,aAAa;AAKlC,QAAM,eAAe,aAAa;AAOlC,QAAM,gBAAgB,aAAa,UAAU,KAAK,YAAY;AAC9D,QAAM,WAAW,aAAa,KAAK,KAAK,YAAY;AACpD,QAAM,gBAAgB,aAAa,UAAU,KAAK,YAAY;AAC9D,QAAM,sBAAsB,aAAa,4BAA4B,KAAK,YAAY;AAEtF,eAAa,WAAW,CAAC,KAAiB,QAAgB,UAA6B;AACrF,UAAM,OAAO,aAAa,gBAAgB,KAAK,MAAM;AACrD,QAAI,CAAC,QAAQ,KAAK,iBAAiB,mBAAmB,MAAM;AAC1D,aAAO,aAAa,KAAK,QAAQ,KAAK;IACxC;AAEA,QAAI,QAAQ,UAAU,oBAAoB,QAAQ,UAAU,mBAAmB;AAC7E,eAAS,KAAK,gBAAgB,KAAK,QAAQ,wCAAwC;AACnF,aAAO;IACT;AAEA,QAAI,EAAE,QAAQ,UAAU,kBAAkB;AACxC,eAAS,KAAK,gBAAgB,KAAK,QAAQ,oCAAoC,KAAK,EAAE;AACtF,aAAO;IACT;AAEA,QAAI,CAAC,KAAK,SAAS;AACjB,eAAS,KAAK,gBAAgB,KAAK,QAAQ,2BAA2B;AACtE,aAAO;IACT;AAEA,UAAM,YAAY,KAAK,sBAAsB;AAC7C,UAAM,YAAY,KAAK,sBAAsB;AAE7C,QAAI,CAAC,WAAW;AACd,UAAI;AACF,cAAM,QAAQ,iBAAiB,KAAK,SAAS,GAAG,CAAC;AACjD,YAAI,MAAM,WAAW,KAAK;AACxB,gBAAM,QACJ,uBAAuB,MAAM,QAAQ,eAAe,CAAC,KACrD,mBAAmB,MAAM,QAAQ,gBAAgB,CAAC;AACpD,cAAI,UAAU,MAAM;AAClB,mBAAO,gBAAgB,KAAK,OAAO,CAAC;UACtC;QACF;AAEA,YAAI,MAAM,WAAW,OAAO,WAAW;AACrC,gBAAM,UAAU,oBAAoB,KAAK,MAAM,KAAK;AACpD,sBAAY,IAAI,QAAQ,EAAE,SAAS,MAAM,MAAM,MAAM,OAAO,CAAC;AAC7D,iBAAO,gBAAgB,KAAK,MAAM,MAAM,QAAQ,OAAO;QACzD;MACF,SAAS,OAAO;AACd,YAAI,CAAC,WAAW;AACd,mBAAS,KAAK,gBAAgB,KAAK,QAAQ,YAAY,OAAO,KAAK,CAAC,EAAE;AACtE,iBAAO;QACT;MACF;IACF;AAEA,QAAI,WAAW;AACb,UAAI;AACF,cAAM,OAAO,YAAY,KAAK,OAAO;AACrC,YAAI,KAAK,WAAW,KAAK;AACvB,gBAAM,UAAU,oBAAoB,KAAK,KAAK,KAAK;AACnD,sBAAY,IAAI,QAAQ,EAAE,SAAS,MAAM,KAAK,MAAM,OAAO,CAAC;AAC5D,iBAAO,gBAAgB,KAAK,KAAK,MAAM,QAAQ,OAAO;QACxD;MACF,SAAS,OAAO;AACd,iBAAS,KAAK,gBAAgB,KAAK,QAAQ,YAAY,OAAO,KAAK,CAAC,EAAE;AACtE,eAAO;MACT;IACF;AAEA,aAAS,KAAK,gBAAgB,KAAK,QAAQ,0CAA0C;AACrF,WAAO;EACT;AAEA,eAAa,WAAW,CACtB,KACA,QACA,QACA,OACA,aACW;AACX,QAAI,UAAU,GAAG;AACf,aAAO;IACT;AAEA,UAAM,OAAO,aAAa,gBAAgB,KAAK,MAAM;AACrD,QAAI,CAAC,QAAQ,KAAK,iBAAiB,mBAAmB,MAAM;AAC1D,aAAO,aAAa,KAAK,QAAQ,QAAQ,OAAO,QAAQ;IAC1D;AAEA,UAAM,SAAS,YAAY,IAAI,MAAM;AACrC,QAAI,QAAQ;AACV,YAAM,aAAa,KAAK,IAAI,GAAG,QAAQ;AACvC,YAAM,WAAW,KAAK,IAAI,OAAO,MAAM,WAAW,KAAK;AACvD,YAAM,SAAS,KAAK,IAAI,GAAG,WAAW,UAAU;AAChD,UAAI,SAAS,GAAG;AACd,cAAM,MAAM,IAAI,OAAO,SAAS,OAAO,UAAU,YAAY,OAAO,UAAU,QAAQ;AACtF,YAAI,OAAO,IAAI,KAAK,MAAM;MAC5B;AACA,aAAO;IACT;AAEA,QAAI,CAAC,KAAK,SAAS;AACjB,eAAS,KAAK,gBAAgB,KAAK,QAAQ,2BAA2B;AACtE,aAAO;IACT;AAEA,QAAI;AACF,YAAM,WAAW,iBAAiB,KAAK,SAAS,UAAU,WAAW,QAAQ,CAAC;AAC9E,UAAI,SAAS,WAAW,OAAQ,SAAS,WAAW,OAAO,aAAa,GAAI;AAC1E,cAAM,SAAS,KAAK,IAAI,OAAO,SAAS,MAAM,MAAM;AACpD,YAAI,SAAS,GAAG;AACd,cAAI,OAAO,IAAI,SAAS,MAAM,SAAS,GAAG,MAAM,GAAG,MAAM;QAC3D;AACA,eAAO;MACT;AAEA,eAAS,KAAK,gBAAgB,KAAK,QAAQ,qBAAqB,SAAS,MAAM,EAAE;AACjF,aAAO;IACT,SAAS,OAAO;AACd,eAAS,KAAK,gBAAgB,KAAK,QAAQ,YAAY,OAAO,KAAK,CAAC,EAAE;AACtE,aAAO;IACT;EACF;AAEA,eAAa,YAAY,CAAC,KAAmB,SAAiB,YAA6B;AACzF,UAAMA,QAAO,WAAW,KAAK,SAAS,OAAO;AAC7C,QAAI,UAAUA,KAAI,GAAG;AACnB,YAAM,WAAW,gBAAgBA,KAAI;AACrC,aAAO,SAAS,WAAW,OAAO,SAAS,WAAW;IACxD;AACA,WAAO,cAAc,KAAK,SAAS,OAAO;EAC5C;AAEA,eAAa,OAAO,CAAC,KAAmB,SAAiB,YAA0B;AACjF,UAAMA,QAAO,WAAW,KAAK,SAAS,OAAO;AAC7C,QAAI,UAAUA,KAAI,GAAG;AACnB,YAAM,WAAW,gBAAgBA,KAAI;AACrC,UAAI,SAAS,WAAW,OAAO,SAAS,WAAW,KAAK;AACtD,YAAI,MAAM,+BAA+B,MAAM,CAAC,QAAQ,GAAG,CAACA,KAAI,CAAC;MACnE;AACA;IACF;AAEA,WAAO,SAAS,KAAK,SAAS,OAAO;EACvC;AAEA,eAAa,YAAY,CAAC,KAAmB,WAAyB;AACpE,UAAM,SAAS,YAAY,IAAI,MAAM;AACrC,QAAI,QAAQ;AACV,UAAI,OAAO,SAAS;AACjB,YAAmB,MAAM,OAAO,OAAO;MAC1C;AACA,kBAAY,OAAO,MAAM;IAC3B;AACA,kBAAc,KAAK,MAAM;EAC3B;AAEA,eAAa,8BAA8B,CAAC,KAAmB,WAA2B;AACxF,UAAM,OAAO,aAAa,gBAAgB,KAAK,MAAM;AACrD,QAAI,MAAM,iBAAiB,mBAAmB,MAAM;AAClD,aAAO,KAAK,IAAI,IAAI;IACtB;AACA,WAAO,oBAAoB,KAAK,MAAM;EACxC;AACF;AAEA,SAAS,UAAU,OAAwB;AACzC,SAAO,MAAM,WAAW,SAAS,KAAK,MAAM,WAAW,UAAU;AACnE;AAEA,SAAS,gBAAgB,KAAiB,MAAc,SAAyB;AAC/E,QAAM,SAAS,IAAI,QAAQ,IAAI,CAAC;AAChC,MAAI,SAAS,UAAU,KAAK,CAAC,IAAI,CAAC;AAClC,MAAI,SAAS,UAAU,KAAK,CAAC,IAAI;AACjC,SAAO;AACT;AAEA,SAAS,oBAAoB,KAAiB,OAA2B;AACvE,QAAM,UAAU,IAAI,QAAQ,MAAM,UAAU;AAC5C,MAAI,OAAO,IAAI,OAAO,OAAO;AAC7B,SAAO;AACT;AAEA,SAAS,uBAAuB,cAA4C;AAC1E,MAAI,CAAC,cAAc;AACjB,WAAO;EACT;AACA,QAAM,CAAC,EAAE,KAAK,IAAI,aAAa,MAAM,GAAG;AACxC,MAAI,CAAC,OAAO;AACV,WAAO;EACT;AACA,QAAM,SAAS,OAAO,SAAS,OAAO,EAAE;AACxC,SAAO,OAAO,SAAS,MAAM,IAAI,SAAS;AAC5C;AAEA,SAAS,mBAAmB,eAA6C;AACvE,MAAI,CAAC,eAAe;AAClB,WAAO;EACT;AACA,QAAM,SAAS,OAAO,SAAS,eAAe,EAAE;AAChD,SAAO,OAAO,SAAS,MAAM,IAAI,SAAS;AAC5C;AAQA,SAAS,YAAY,KAA2B;AAC9C,SAAO,YAAY,CAAC,GAAG,CAAC;AAC1B;AAEA,SAAS,gBAAgB,KAA2B;AAClD,SAAO,YAAY,CAAC,MAAM,GAAG,CAAC;AAChC;AAEA,SAAS,iBAAiB,KAAa,OAAe,KAA2B;AAC/E,SAAO,YAAY,CAAC,MAAM,GAAG,KAAK,IAAI,GAAG,IAAI,GAAG,CAAC;AACnD;AAEA,SAAS,YAAY,MAA8B;AACjD,QAAM,WAAW,KAAK,KAAK,OAAO,GAAG,oBAAoB,WAAW,CAAC,EAAE;AACvE,MAAI;AACF,UAAM,SAAS,UAAU,QAAQ,CAAC,OAAO,MAAM,MAAM,KAAK,MAAM,UAAU,GAAG,IAAI,GAAG;MAClF,UAAU;MACV,WAAW,IAAI,OAAO;IACxB,CAAC;AAED,QAAI,OAAO,OAAO;AAChB,UAAK,OAAO,MAAgC,SAAS,UAAU;AAC7D,cAAM,IAAI,MAAM,gDAAgD;MAClE;AACA,YAAM,OAAO;IACf;AACA,QAAI,OAAO,WAAW,GAAG;AACvB,YAAM,SAAS,OAAO,QAAQ,SAAS,MAAM,EAAE,KAAK;AACpD,YAAM,IAAI,MAAM,UAAU,aAAa;IACzC;AAEA,UAAM,OAAO,aAAa,QAAQ;AAClC,WAAO,kBAAkB,OAAO,KAAK,OAAO,UAAU,CAAC,CAAC,GAAG,IAAI;EACjE,UAAA;AACE,WAAO,UAAU,EAAE,OAAO,KAAK,CAAC;EAClC;AACF;AAEA,SAAS,kBAAkB,eAAuB,MAA4B;AAC5E,QAAM,aAAa,cAAc,SAAS,QAAQ;AAClD,QAAM,SAAS,WAAW,MAAM,UAAU,EAAE,OAAO,OAAO;AAC1D,QAAM,YAAY,OAAO,OAAO,SAAS,CAAC,KAAK;AAC/C,QAAM,QAAQ,UAAU,MAAM,MAAM,EAAE,OAAO,OAAO;AACpD,QAAM,aAAa,MAAM,MAAM,KAAK;AACpC,QAAM,cAAc,WAAW,MAAM,GAAG,EAAE,CAAC,KAAK;AAChD,QAAM,SAAS,OAAO,SAAS,aAAa,EAAE;AAC9C,QAAM,UAAkC,CAAC;AAEzC,aAAW,QAAQ,OAAO;AACxB,UAAM,QAAQ,KAAK,QAAQ,GAAG;AAC9B,QAAI,UAAU,IAAI;AAChB;IACF;AACA,UAAM,MAAM,KAAK,MAAM,GAAG,KAAK,EAAE,KAAK,EAAE,YAAY;AACpD,UAAM,QAAQ,KAAK,MAAM,QAAQ,CAAC,EAAE,KAAK;AACzC,YAAQ,GAAG,IAAI;EACjB;AAEA,SAAO;IACL,QAAQ,OAAO,SAAS,MAAM,IAAI,SAAS;IAC3C,OAAO,IAAI,WAAW,IAAI;IAC1B;EACF;AACF;AAEA,eAAsB,0BAA0B,UAA0C;AACxF,QAAM,KAAK,MAAM,UAAU;AAC3B,QAAM,OAAO,GAAG,QAAQ;AACxB,QAAM,WAAW,oBAAoB,QAAQ;AAE7C,KAAG,gBAAgB,UAAU,UAAU,mBAAmB,SAAS,IAAI;AAEvE,SAAO,oBAAoB,IAAI,MAAM,QAAQ;AAC/C;AAEA,eAAsB,yBAAyB,OAAuC;AACpF,QAAM,WAAW,kBAAkB,KAAK;AACxC,MAAI,CAAC,UAAU;AACb,UAAM,IAAI,MAAM,WAAW;EAC7B;AAEA,QAAM,KAAK,MAAM,UAAU;AAC3B,QAAM,OAAO,GAAG,QAAQ;AACxB,QAAM,WAAW,oBAAoB,SAAS,GAAG;AAEjD,KAAG,gBAAgB,UAAU,SAAS,KAAK,mBAAmB,MAAM,IAAI;AAExE,SAAO,oBAAoB,IAAI,MAAM,QAAQ;AAC/C;AAYA,eAAsB,kBAAkB,OAAuC;AAC7E,QAAM,WAAW,kBAAkB,KAAK;AACxC,MAAI,UAAU;AACZ,WAAO,yBAAyB,KAAK;EACvC;AAEA,SAAO,0BAA0B,KAAK;AACxC;AAeA,eAAsB,yBACpB,UACA,SACgB;AAChB,QAAM,SAAS,MAAM,0BAA0B,QAAQ;AAEvD,MAAI;AACF,WAAO,MAAM,OAAO,UAAU,OAAO;EACvC,UAAA;AACE,UAAM,OAAO,MAAM;EACrB;AACF;AAEA,eAAsB,wBACpB,OACA,SACgB;AAChB,QAAM,SAAS,MAAM,yBAAyB,KAAK;AAEnD,MAAI;AACF,WAAO,MAAM,OAAO,UAAU,OAAO;EACvC,UAAA;AACE,UAAM,OAAO,MAAM;EACrB;AACF;AAcA,eAAsB,sBACpB,eAAe,iBACW;AAC1B,QAAM,UAAU,MAAM,GAAG,QAAQ,KAAK,KAAK,OAAO,GAAG,cAAc,CAAC;AACpE,QAAM,WAAW,aAAa,QAAQ,UAAU,GAAG;AACnD,QAAM,WAAW,KAAK,KAAK,SAAS,GAAG,WAAW,CAAC,IAAI,QAAQ,EAAE;AACjE,QAAM,cAAc,kBAAkB,QAAQ;AAE9C,QAAM,SAAS,QAAQ,OAAO,WAAW;AAEzC,SAAO;IACL,MAAM;IACN,SAAS,YAAY;AACnB,YAAM,GAAG,GAAG,SAAS,EAAE,WAAW,MAAM,OAAO,KAAK,CAAC;IACvD;EACF;AACF;AAEA,eAAsB,0BACpB,eAAe,iBACf,SACgB;AAChB,QAAM,OAAO,MAAM,sBAAsB,YAAY;AAErD,MAAI;AACF,WAAO,MAAM,yBAAyB,KAAK,MAAM,OAAO;EAC1D,UAAA;AACE,UAAM,KAAK,QAAQ;EACrB;AACF;AAEA,SAAS,oBACP,IACA,MACA,UACe;AACf,MAAI,kBAAuD;AAE3D,SAAO;IACL,WAAW,OAAO,YAAiC;AACjD,YAAM,QAAQ,iBAAiB,UAAU,OAAO;AAChD,aAAO,KAAK,MAAM,KAAK;IACzB;IACA,cAAc,MAAM;AAClB,UAAI,CAAC,iBAAiB;AACpB,0BAAkB,oBAAoB,MAAM,QAAQ;MACtD;AACA,aAAO;IACT;IACA,OAAO,YAAY;AACjB,WAAK,MAAM;AACX,SAAG,SAAS,QAAQ;IACtB;EACF;AACF;AAEA,SAAS,oBAAoB,OAAuB;AAClD,QAAM,SAAS,KAAK,QAAQ,KAAK,KAAK;AACtC,SAAO,eAAe,WAAW,CAAC,GAAG,MAAM;AAC7C;AAEA,SAAS,iBAAiB,UAAkB,SAAsC;AAChF,QAAM,UAAU,SAAS,WAAW,QAAQ,QAAQ,SAAS,IAAI,QAAQ,UAAU;AACnF,QAAM,aAAa,UAAU,QAAQ,IAAI,eAAe,EAAE,KAAK,IAAI,IAAI;AACvE,QAAM,QAAQ,SAAS;AACvB,QAAM,SAAS,SAAS;AAExB,MAAI,QAAQ,UAAU,UAAU,sBAAsB,aAAa,QAAQ,CAAC;AAE5E,MAAI,OAAO,UAAU,UAAU;AAC7B,aAAS,UAAU,KAAK,IAAI,GAAG,KAAK,CAAC;EACvC;AAEA,MAAI,OAAO,WAAW,YAAY,SAAS,GAAG;AAC5C,aAAS,WAAW,KAAK,IAAI,GAAG,MAAM,CAAC;EACzC;AAEA,SAAO;AACT;AAEA,SAAS,gBAAgB,OAAuB;AAC9C,SAAO,IAAI,MAAM,QAAQ,MAAM,IAAI,CAAC;AACtC;AAEA,SAAS,aAAa,OAAuB;AAC3C,SAAO,IAAI,MAAM,QAAQ,MAAM,IAAI,CAAC;AACtC;AAEA,eAAe,oBACb,MACA,UAC8B;AAC9B,QAAM,eAAe;IACnB,KAAK,MAAM,uCAAuC,aAAa,QAAQ,CAAC,GAAG;EAC7E;AACA,QAAM,SAAS;IACb,KAAK,MAAM,qCAAqC,aAAa,QAAQ,CAAC,GAAG;EAC3E;AAEA,QAAM,eAAe,aAAa,CAAC,GAAG,cAAc,aAAa,CAAC,GAAG,aAAa;AAClF,QAAM,mBAA4C,CAAC;AAEnD,aAAW,OAAO,QAAQ;AACxB,UAAM,MAAM,IAAI,OAAO,IAAI,YAAY,IAAI;AAC3C,QAAI,OAAO,QAAQ,YAAY,IAAI,WAAW,GAAG;AAC/C;IACF;AACA,qBAAiB,GAAG,IAAI,IAAI,SAAS,IAAI,OAAO;EAClD;AAEA,SAAO;IACL,WAAW,uBAAuB,YAAY;IAC9C,kBAAkB,wBAAwB,gBAAgB;EAC5D;AACF;AAEA,SAAS,eAAe,OAAyC;AAC/D,QAAM,SAAS,MAAM,OAAO,OAAO,IAAI,CAAC,UAAU,MAAM,IAAI;AAC5D,QAAM,OAAkC,CAAC;AAEzC,aAAW,SAAS,MAAM,SAAS;AACjC,UAAM,UAAU,OAAO,IAAI,CAAC,GAAG,UAAU,MAAM,WAAW,KAAK,CAAC;AAEhE,aAAS,WAAW,GAAG,WAAW,MAAM,SAAS,YAAY,GAAG;AAC9D,YAAM,MAA+B,CAAC;AAEtC,eAAS,WAAW,GAAG,WAAW,OAAO,QAAQ,YAAY,GAAG;AAC9D,YAAI,OAAO,QAAQ,CAAC,IAAI,QAAQ,QAAQ,GAAG,IAAI,QAAQ;MACzD;AAEA,WAAK,KAAK,GAAG;IACf;EACF;AAEA,SAAO;AACT;AAEA,SAAS,uBAAuB,OAAoC;AAClE,MAAI,UAAU,QAAQ,UAAU,QAAW;AACzC,WAAO;EACT;AAEA,MAAI,iBAAiB,YAAY;AAC/B,WAAO,OAAO,KAAK,KAAK,EAAE,SAAS,MAAM;EAC3C;AAEA,MAAI,OAAO,UAAU,UAAU;AAC7B,WAAO;EACT;AAEA,SAAO,OAAO,KAAK;AACrB;AAEA,SAAS,wBAAwB,OAAwD;AACvF,QAAM,aAAqC,CAAC;AAE5C,aAAW,CAAC,KAAK,KAAK,KAAK,OAAO,QAAQ,KAAK,GAAG;AAChD,UAAM,kBAAkB,uBAAuB,KAAK;AACpD,eAAW,GAAG,IAAI,mBAAmB;EACvC;AAEA,SAAO;AACT;","names":["path"]}

package/dist/main.js

@@ -7,12 +7,16 @@ const __filename = fileURLToPath(import.meta.url);
 const __dirname = dirname(__filename);
 import {
   readParquetTableFromPath,
-  readParquetTableFromStdin
-} from "./chunk-573AA4JN.js";
+  readParquetTableFromStdin,
+  readParquetTableFromUrl,
+  resolveParquetUrl
+} from "./chunk-NRRDNC7S.js";
 
 // src/main.ts
 import { spawnSync } from "child_process";
 import path from "path";
+import { fileURLToPath } from "url";
+var __filename = typeof globalThis.__filename !== "undefined" ? globalThis.__filename : fileURLToPath(import.meta.url);
 var DEFAULT_LIMIT = 20;
 function parseArgs(argv) {
   const options = {
@@ -112,7 +116,7 @@ function readOptionValue(arg, name, next) {
   return null;
 }
 function printUsage() {
-  const helpText = `parquetlens <file|-> [options]
+  const helpText = `parquetlens <file|url|-> [options]
 
 options:
   --limit, --limit=<n>       number of rows to show (default: ${DEFAULT_LIMIT})
@@ -127,7 +131,8 @@ options:
 examples:
   parquetlens data.parquet --limit 25
   parquetlens data.parquet --columns=city,state
-  parquetlens data.parquet --tui
+  parquetlens hf://datasets/cfahlgren1/hub-stats/daily_papers.parquet
+  parquetlens https://huggingface.co/datasets/cfahlgren1/hub-stats/resolve/main/daily_papers.parquet
   parquetlens data.parquet --plain
   parquetlens - < input.parquet
 `;
@@ -204,11 +209,14 @@ async function loadTable(input, readOptions) {
   const stdinFallback = process.stdin.isTTY ? void 0 : "-";
   const source = input ?? stdinFallback;
   if (!source) {
-    throw new Error("missing input file (pass a path or pipe stdin)");
+    throw new Error("missing input file (pass a path, URL, or pipe stdin)");
   }
   if (source === "-") {
     return readParquetTableFromStdin("stdin.parquet", readOptions);
   }
+  if (resolveParquetUrl(source)) {
+    return readParquetTableFromUrl(source, readOptions);
+  }
   return readParquetTableFromPath(source, readOptions);
 }
 async function main() {
@@ -227,7 +235,9 @@ async function main() {
   const wantsTui = resolveTuiMode(options.tuiMode, options);
   if (wantsTui) {
     if (!input || input === "-") {
-      process.stderr.write("parquetlens: tui mode requires a file path (stdin not supported)\n");
+      process.stderr.write(
+        "parquetlens: tui mode requires a file path or URL (stdin not supported)\n"
+      );
       process.exitCode = 1;
       return;
     }