@naturalcycles/nodejs-lib 12.56.1 → 12.60.0

package/src/stream/transform/transformMapSync.ts

@@ -1,8 +1,16 @@
-import { Transform } from 'stream'
-import { AggregatedError, ErrorMode, Mapper, Predicate } from '@naturalcycles/js-lib'
+import {
+  AggregatedError,
+  CommonLogger,
+  END,
+  ErrorMode,
+  Mapper,
+  Predicate,
+  SKIP,
+} from '@naturalcycles/js-lib'
 import { yellow } from '../../colors'
+import { AbortableTransform } from '../pipeline/pipeline'
 import { TransformTyped } from '../stream.model'
-import { notNullishPredicate } from './transformMap'
+import { pipelineClose } from '../stream.util'
 
 export interface TransformMapSyncOptions<IN = any, OUT = IN> {
   /**
@@ -20,9 +28,8 @@ export interface TransformMapSyncOptions<IN = any, OUT = IN> {
    * Predicate to filter outgoing results (after mapper).
    * Allows to not emit all results.
    *
-   * @default to filter out undefined/null values, but pass anything else
-   *
-   * Set to `r => r` (passthrough predicate) to pass ANY value (including undefined/null)
+   * Defaults to "pass everything".
+   * Simpler way to skip individual entries is to return SKIP symbol.
    */
   predicate?: Predicate<OUT>
 
@@ -43,8 +50,12 @@ export interface TransformMapSyncOptions<IN = any, OUT = IN> {
    * @default `stream`
    */
   metric?: string
+
+  logger?: CommonLogger
 }
 
+export class TransformMapSync extends AbortableTransform {}
+
 /**
  * Sync (not async) version of transformMap.
  * Supposedly faster, for cases when async is not needed.
@@ -56,43 +67,49 @@ export function transformMapSync<IN = any, OUT = IN>(
   let index = -1
 
   const {
-    predicate = notNullishPredicate,
+    predicate, // defaults to "no predicate" (pass everything)
     errorMode = ErrorMode.THROW_IMMEDIATELY,
     flattenArrayOutput = false,
     onError,
     metric = 'stream',
     objectMode = true,
+    logger = console,
   } = opt
-  let isRejected = false
+  let isSettled = false
   let errors = 0
   const collectedErrors: Error[] = [] // only used if errorMode == THROW_AGGREGATED
 
-  return new Transform({
+  return new TransformMapSync({
     objectMode,
     ...opt,
-    transform(this: Transform, chunk: IN, _encoding, cb) {
-      // Stop processing if THROW_IMMEDIATELY mode is used
-      if (isRejected && errorMode === ErrorMode.THROW_IMMEDIATELY) {
-        return cb()
-      }
+    transform(this: AbortableTransform, chunk: IN, _, cb) {
+      // Stop processing if isSettled
+      if (isSettled) return cb()
 
-      try {
-        if (!predicate(chunk, ++index)) {
-          cb() // signal that we've finished processing, but emit no output here
-          return
-        }
+      const currentIndex = ++index
 
+      try {
         // map and pass through
-        const v = mapper(chunk, index)
+        const v = mapper(chunk, currentIndex)
+
+        const passedResults = (flattenArrayOutput && Array.isArray(v) ? v : [v]).filter(r => {
+          if (r === END) {
+            isSettled = true // will be checked later
+            return false
+          }
+          return r !== SKIP && (!predicate || predicate(r, currentIndex))
+        })
 
-        if (flattenArrayOutput && Array.isArray(v)) {
-          // Pass each item individually
-          v.forEach(item => this.push(item))
-        } else {
-          cb(null, v)
+        passedResults.forEach(r => this.push(r))
+
+        if (isSettled) {
+          logger.log(`transformMapSync END received at index ${currentIndex}`)
+          pipelineClose('transformMapSync', this, this.sourceReadable, this.streamDone, logger)
         }
+
+        cb() // done processing
       } catch (err) {
-        console.error(err)
+        logger.error(err)
         errors++
 
         logErrorStats()
@@ -104,7 +121,7 @@ export function transformMapSync<IN = any, OUT = IN>(
         }
 
         if (errorMode === ErrorMode.THROW_IMMEDIATELY) {
-          isRejected = true
+          isSettled = true
           // Emit error immediately
           return cb(err as Error)
         }
@@ -134,6 +151,6 @@ export function transformMapSync<IN = any, OUT = IN>(
   function logErrorStats(final = false): void {
     if (!errors) return
 
-    console.log(`${metric} ${final ? 'final ' : ''}errors: ${yellow(errors)}`)
+    logger.log(`${metric} ${final ? 'final ' : ''}errors: ${yellow(errors)}`)
   }
 }

package/src/stream/transform/transformNoOp.ts

@@ -9,7 +9,7 @@ import { TransformTyped } from '../stream.model'
 export function transformNoOp<T = any>(): TransformTyped<T, T> {
   return new Transform({
     objectMode: true,
-    transform(chunk: T, _encoding, cb) {
+    transform(chunk: T, _, cb) {
       cb(null, chunk)
     },
   })

package/src/stream/transform/transformTap.ts

@@ -1,7 +1,11 @@
 import { Transform } from 'stream'
-import { AsyncMapper } from '@naturalcycles/js-lib'
+import { AsyncMapper, CommonLogger } from '@naturalcycles/js-lib'
 import { TransformOptions, TransformTyped } from '../stream.model'
 
+export interface TransformTapOptions extends TransformOptions {
+  logger?: CommonLogger
+}
+
 /**
  * Similar to RxJS `tap` - allows to run a function for each stream item, without affecting the result.
  * Item is passed through to the output.
@@ -10,20 +14,21 @@ import { TransformOptions, TransformTyped } from '../stream.model'
  */
 export function transformTap<IN>(
   fn: AsyncMapper<IN, any>,
-  opt: TransformOptions = {},
+  opt: TransformTapOptions = {},
 ): TransformTyped<IN, IN> {
-  let index = 0
+  const { logger = console } = opt
+  let index = -1
 
   return new Transform({
     objectMode: true,
     ...opt,
-    async transform(chunk: IN, _encoding, cb) {
+    async transform(chunk: IN, _, cb) {
       // console.log('tap', chunk)
 
       try {
-        await fn(chunk, index++)
+        await fn(chunk, ++index)
       } catch (err) {
-        console.error(err)
+        logger.error(err)
         // suppressed error
       }
 

package/src/stream/transform/transformToArray.ts

@@ -10,7 +10,7 @@ export function transformToArray<IN>(opt: TransformOptions = {}): TransformTyped
   return new Transform({
     objectMode: true,
     ...opt,
-    transform(chunk: IN, _encoding, cb) {
+    transform(chunk: IN, _, cb) {
       res.push(chunk)
       // callback to signal that we processed input, but not emitting any output
       cb()

package/src/stream/transform/transformToString.ts

@@ -14,8 +14,8 @@ export function transformToString(): TransformTyped<Buffer, string> {
   return new Transform({
     objectMode: false,
     readableObjectMode: true,
-    transform(chunk: Buffer, _encoding, cb) {
-      // console.log(`enc: ${_encoding}`, chunk.toString())
+    transform(chunk: Buffer, _, cb) {
+      // console.log(`enc: ${_}`, chunk.toString())
       cb(null, chunk.toString())
     },
   })

package/src/stream/transform/worker/transformMultiThreaded.ts

@@ -115,7 +115,7 @@ export function transformMultiThreaded<IN, OUT>(
         }
       },
     },
-    async function transformMapFn(chunk: IN, _encoding, cb) {
+    async function transformMapFn(chunk: IN, _, cb) {
       // Freezing the index, because it may change due to concurrency
       const currentIndex = ++index
 

package/src/stream/transform/worker/workerClassProxy.js

@@ -10,6 +10,7 @@ if (!workerFile) {
 // console.log(`worker#${workerIndex} created`)
 
 try {
+  // require('esbuild-register') // alternative
   require('ts-node/register/transpile-only')
   require('tsconfig-paths/register')
 } catch {} // require if exists

package/src/stream/writable/writableFork.ts

@@ -7,6 +7,8 @@ import { TransformOptions, WritableTyped } from '../stream.model'
  * Currently does NOT (!) maintain backpressure.
  * Error in the forked pipeline will propagate up to the main pipeline (and log error, to be sure).
  * Will wait until all forked pipelines are completed before completing the stream.
+ *
+ * @experimental
  */
 export function writableFork<T>(
   chains: NodeJS.WritableStream[][],
@@ -29,7 +31,7 @@ export function writableFork<T>(
   return new Writable({
     objectMode: true,
     ...opt,
-    write(chunk: T, _encoding, cb) {
+    write(chunk: T, _, cb) {
       // Push/fork to all sub-streams
       // No backpressure is ensured here, it'll push regardless of the
       readables.forEach(readable => readable.push(chunk))

package/src/stream/writable/writableLimit.ts

@@ -0,0 +1,28 @@
+import { Readable, Writable } from 'stream'
+import { WritableTyped } from '../stream.model'
+
+/**
+ * Allows to stop the Readable stream after the pipeline has processed X number of rows.
+ * It counts OUTPUT rows (not input), because this Writable is always at the end of the Pipeline.
+ * It ensures that everything has been processed before issuing a STOP on the readable.
+ */
+export function writableLimit<T>(readable: Readable, limit: number): WritableTyped<T> {
+  let i = 0
+
+  return new Writable({
+    objectMode: true,
+    write(chunk, _, cb) {
+      if (limit === 0) return cb() // no limit, just passthrough
+
+      i++
+
+      if (i === limit) {
+        console.log(`writableLimit of ${limit} reached`)
+        readable.destroy()
+        cb() // do we need it?
+      } else {
+        cb() // passthrough
+      }
+    },
+  })
+}

package/src/stream/writable/writablePushToArray.ts

@@ -8,7 +8,7 @@ export function writablePushToArray<IN>(arr: IN[], opt: TransformOptions = {}):
   return new Writable({
     objectMode: true,
     ...opt,
-    write(chunk: IN, _encoding, cb) {
+    write(chunk: IN, _, cb) {
       arr.push(chunk)
       // callback to signal that we processed input, but not emitting any output
       cb()

package/src/stream/writable/writableVoid.ts

@@ -1,17 +1,29 @@
 import { Writable } from 'stream'
+import { DeferredPromise } from '@naturalcycles/js-lib'
 import { TransformOptions } from '../stream.model'
 
+export interface WritableVoidOptions extends TransformOptions {
+  /**
+   * If set - it will be Resolved when the Stream is done (after final.cb)
+   */
+  streamDone?: DeferredPromise
+}
+
 /**
  * Use as a "null-terminator" of stream.pipeline.
  * It consumes the stream as quickly as possible without doing anything.
  * Put it in the end of your pipeline in case it ends with Transform that needs a consumer.
  */
-export function writableVoid(opt?: TransformOptions): Writable {
+export function writableVoid(opt: WritableVoidOptions = {}): Writable {
   return new Writable({
     objectMode: true,
     ...opt,
-    write(chunk, _encoding, cb) {
+    write(chunk, _, cb) {
+      cb()
+    },
+    final(cb) {
       cb()
+      opt.streamDone?.resolve()
     },
   })
 }

package/src/util/zip.util.ts

@@ -8,9 +8,10 @@ const gzip = promisify(zlib.gzip.bind(zlib))
 const gunzip = promisify(zlib.gunzip.bind(zlib))
 
 // string > zip
+
 /**
  * zipBuffer uses `deflate`.
- * It's 9 bytes shorter than gzip.
+ * It's 9 bytes shorter than `gzip`.
  */
 export async function zipBuffer(buf: Buffer, options: ZlibOptions = {}): Promise<Buffer> {
   return await deflate(buf, options)
@@ -18,7 +19,7 @@ export async function zipBuffer(buf: Buffer, options: ZlibOptions = {}): Promise
 
 /**
  * gzipBuffer uses `gzip`
- * It's 9 bytes longer than deflate.
+ * It's 9 bytes longer than `deflate`.
  */
 export async function gzipBuffer(buf: Buffer, options: ZlibOptions = {}): Promise<Buffer> {
   return await gzip(buf, options)
@@ -33,11 +34,18 @@ export async function gunzipBuffer(buf: Buffer, options: ZlibOptions = {}): Prom
   return await gunzip(buf, options)
 }
 
-// convenience
+/**
+ * zipString uses `deflate`.
+ * It's 9 bytes shorter than `gzip`.
+ */
 export async function zipString(s: string, options?: ZlibOptions): Promise<Buffer> {
   return await zipBuffer(Buffer.from(s), options)
 }
 
+/**
+ * gzipString uses `gzip`.
+ * It's 9 bytes longer than `deflate`.
+ */
 export async function gzipString(s: string, options?: ZlibOptions): Promise<Buffer> {
   return await gzipBuffer(Buffer.from(s), options)
 }

package/dist/stream/transform/legacy/transformMap.d.ts

@@ -1,17 +0,0 @@
-import { AsyncMapper } from '@naturalcycles/js-lib';
-import { TransformTyped } from '../../stream.model';
-import { TransformMapOptions } from '../transformMap';
-export declare function notNullishPredicate(item: any): boolean;
-/**
- * Like pMap, but for streams.
- * Inspired by `through2`.
- * Main feature is concurrency control (implemented via `through2-concurrent`) and convenient options.
- * Using this allows native stream .pipe() to work and use backpressure.
- *
- * Only works in objectMode (due to through2Concurrent).
- *
- * Concurrency defaults to 16.
- *
- * If an Array is returned by `mapper` - it will be flattened and multiple results will be emitted from it. Tested by Array.isArray().
- */
-export declare function transformMapLegacy<IN = any, OUT = IN>(mapper: AsyncMapper<IN, OUT>, opt?: TransformMapOptions<IN, OUT>): TransformTyped<IN, OUT>;

package/dist/stream/transform/legacy/transformMap.js

@@ -1,94 +0,0 @@
-"use strict";
-Object.defineProperty(exports, "__esModule", { value: true });
-exports.transformMapLegacy = exports.notNullishPredicate = void 0;
-const js_lib_1 = require("@naturalcycles/js-lib");
-const through2Concurrent = require("through2-concurrent");
-const colors_1 = require("../../../colors");
-function notNullishPredicate(item) {
-    return item !== undefined && item !== null;
-}
-exports.notNullishPredicate = notNullishPredicate;
-/**
- * Like pMap, but for streams.
- * Inspired by `through2`.
- * Main feature is concurrency control (implemented via `through2-concurrent`) and convenient options.
- * Using this allows native stream .pipe() to work and use backpressure.
- *
- * Only works in objectMode (due to through2Concurrent).
- *
- * Concurrency defaults to 16.
- *
- * If an Array is returned by `mapper` - it will be flattened and multiple results will be emitted from it. Tested by Array.isArray().
- */
-function transformMapLegacy(mapper, opt = {}) {
-    const { concurrency = 16, predicate = notNullishPredicate, errorMode = js_lib_1.ErrorMode.THROW_IMMEDIATELY, flattenArrayOutput, onError, beforeFinal, metric = 'stream', logger = console, } = opt;
-    let index = -1;
-    let isRejected = false;
-    let errors = 0;
-    const collectedErrors = []; // only used if errorMode == THROW_AGGREGATED
-    return through2Concurrent.obj({
-        maxConcurrency: concurrency,
-        // autoDestroy: true,
-        async final(cb) {
-            // console.log('transformMap final')
-            logErrorStats(logger, true);
-            await beforeFinal?.(); // call beforeFinal if defined
-            if (collectedErrors.length) {
-                // emit Aggregated error
-                cb(new js_lib_1.AggregatedError(collectedErrors));
-            }
-            else {
-                // emit no error
-                cb();
-            }
-        },
-    }, async function transformMapFn(chunk, _encoding, cb) {
-        index++;
-        // console.log({chunk, _encoding})
-        // Stop processing if THROW_IMMEDIATELY mode is used
-        if (isRejected && errorMode === js_lib_1.ErrorMode.THROW_IMMEDIATELY)
-            return cb();
-        try {
-            const currentIndex = index; // because we need to pass it to 2 functions - mapper and predicate. Refers to INPUT index (since it may return multiple outputs)
-            const res = await mapper(chunk, currentIndex);
-            const passedResults = await (0, js_lib_1.pFilter)(flattenArrayOutput && Array.isArray(res) ? res : [res], async (r) => await predicate(r, currentIndex));
-            if (passedResults.length === 0) {
-                cb(); // 0 results
-            }
-            else {
-                passedResults.forEach(r => {
-                    this.push(r);
-                    // cb(null, r)
-                });
-                cb(); // done processing
-            }
-        }
-        catch (err) {
-            logger.error(err);
-            errors++;
-            logErrorStats(logger);
-            if (onError) {
-                try {
-                    onError(err, chunk);
-                }
-                catch { }
-            }
-            if (errorMode === js_lib_1.ErrorMode.THROW_IMMEDIATELY) {
-                isRejected = true;
-                // Emit error immediately
-                return cb(err);
-            }
-            if (errorMode === js_lib_1.ErrorMode.THROW_AGGREGATED) {
-                collectedErrors.push(err);
-            }
-            // Tell input stream that we're done processing, but emit nothing to output - not error nor result
-            cb();
-        }
-    });
-    function logErrorStats(logger, final = false) {
-        if (!errors)
-            return;
-        logger.log(`${metric} ${final ? 'final ' : ''}errors: ${(0, colors_1.yellow)(errors)}`);
-    }
-}
-exports.transformMapLegacy = transformMapLegacy;

package/src/stream/transform/legacy/transformMap.ts

@@ -1,133 +0,0 @@
-import { Transform } from 'stream'
-import {
-  AggregatedError,
-  AsyncMapper,
-  CommonLogger,
-  ErrorMode,
-  pFilter,
-} from '@naturalcycles/js-lib'
-import through2Concurrent = require('through2-concurrent')
-import { yellow } from '../../../colors'
-import { TransformTyped } from '../../stream.model'
-import { TransformMapOptions } from '../transformMap'
-
-export function notNullishPredicate(item: any): boolean {
-  return item !== undefined && item !== null
-}
-
-/**
- * Like pMap, but for streams.
- * Inspired by `through2`.
- * Main feature is concurrency control (implemented via `through2-concurrent`) and convenient options.
- * Using this allows native stream .pipe() to work and use backpressure.
- *
- * Only works in objectMode (due to through2Concurrent).
- *
- * Concurrency defaults to 16.
- *
- * If an Array is returned by `mapper` - it will be flattened and multiple results will be emitted from it. Tested by Array.isArray().
- */
-export function transformMapLegacy<IN = any, OUT = IN>(
-  mapper: AsyncMapper<IN, OUT>,
-  opt: TransformMapOptions<IN, OUT> = {},
-): TransformTyped<IN, OUT> {
-  const {
-    concurrency = 16,
-    predicate = notNullishPredicate,
-    errorMode = ErrorMode.THROW_IMMEDIATELY,
-    flattenArrayOutput,
-    onError,
-    beforeFinal,
-    metric = 'stream',
-    logger = console,
-  } = opt
-
-  let index = -1
-  let isRejected = false
-  let errors = 0
-  const collectedErrors: Error[] = [] // only used if errorMode == THROW_AGGREGATED
-
-  return through2Concurrent.obj(
-    {
-      maxConcurrency: concurrency,
-      // autoDestroy: true,
-      async final(cb) {
-        // console.log('transformMap final')
-
-        logErrorStats(logger, true)
-
-        await beforeFinal?.() // call beforeFinal if defined
-
-        if (collectedErrors.length) {
-          // emit Aggregated error
-          cb(new AggregatedError(collectedErrors))
-        } else {
-          // emit no error
-          cb()
-        }
-      },
-    },
-    async function transformMapFn(
-      this: Transform,
-      chunk: IN,
-      _encoding: any,
-      cb: (...args: any[]) => any,
-    ) {
-      index++
-      // console.log({chunk, _encoding})
-
-      // Stop processing if THROW_IMMEDIATELY mode is used
-      if (isRejected && errorMode === ErrorMode.THROW_IMMEDIATELY) return cb()
-
-      try {
-        const currentIndex = index // because we need to pass it to 2 functions - mapper and predicate. Refers to INPUT index (since it may return multiple outputs)
-        const res = await mapper(chunk, currentIndex)
-        const passedResults = await pFilter(
-          flattenArrayOutput && Array.isArray(res) ? res : [res],
-          async r => await predicate(r, currentIndex),
-        )
-
-        if (passedResults.length === 0) {
-          cb() // 0 results
-        } else {
-          passedResults.forEach(r => {
-            this.push(r)
-            // cb(null, r)
-          })
-          cb() // done processing
-        }
-      } catch (err) {
-        logger.error(err)
-
-        errors++
-
-        logErrorStats(logger)
-
-        if (onError) {
-          try {
-            onError(err, chunk)
-          } catch {}
-        }
-
-        if (errorMode === ErrorMode.THROW_IMMEDIATELY) {
-          isRejected = true
-          // Emit error immediately
-          return cb(err)
-        }
-
-        if (errorMode === ErrorMode.THROW_AGGREGATED) {
-          collectedErrors.push(err as Error)
-        }
-
-        // Tell input stream that we're done processing, but emit nothing to output - not error nor result
-        cb()
-      }
-    },
-  )
-
-  function logErrorStats(logger: CommonLogger, final = false): void {
-    if (!errors) return
-
-    logger.log(`${metric} ${final ? 'final ' : ''}errors: ${yellow(errors)}`)
-  }
-}