fix(csv-parse): skip event not raised with bom (fix #411)

wdavidw · wdavidw · commit 13263514ef6e · 2024-05-14T00:42:59.000+02:00
diff --git a/demo/issues-esm/lib/411.csv b/demo/issues-esm/lib/411.csv
@@ -1,4 +1,4 @@
-id,first_name,last_name,email,modified_at
+﻿id,first_name,last_name,email,modified_at
 1,Ring,Grinyov,rgrinyov0@weebly.com,2022-02-14
 2,Kylie,Lauderdale,klauderdale1@wsj.com,2022-02-14,
 3,Cammi,Bendix,cbendix2@tuttocitta.it,2022-02-14
diff --git a/demo/issues-esm/lib/411.js b/demo/issues-esm/lib/411.js
@@ -1,38 +1,27 @@
-import path from 'path';
-import { pipeline } from 'stream/promises';
-import { parse as parseCSV } from 'csv-parse';
-import { Writable } from 'stream';
-import { createReadStream } from 'fs';
+import assert from 'node:assert';
+import { createReadStream } from 'node:fs';
+import { Writable } from 'node:stream'
+import { finished } from 'node:stream/promises';
 import desm from "desm";
-const __dirname = desm(import.meta.url);
-
-async function testRecordsSkip() {
-  const errors = [];
-  const records = [];
-
-  const sink = new Writable({
-    objectMode: true,
-    write: (_, __, callback) => {
-      records.push(_);
-      callback();
-    },
-  });
+import { parse } from 'csv-parse';
 
-  const csvSource = createReadStream(path.join(__dirname, '411.csv'));
-  const parser = parseCSV({
-    skip_records_with_error: true,
-    bom: true,
-  });
-  parser.on('skip', function (err) {
-    errors.push(err);
-  });
-
-  await pipeline(csvSource, parser, sink);
-
-  console.log({
-    records,
-    errors,
-  });
-}
+const __dirname = desm(import.meta.url);
+const errors = []
 
-testRecordsSkip().catch(console.error);
+const parser = parse({
+  bom: true,
+  skipRecordsWithError: true,
+});
+// Create a stream and consume its source
+const sink = new Writable ({objectMode: true, write: (_, __, callback) => callback()})
+const outStream = createReadStream(`${__dirname}/411.csv`).pipe(parser).pipe(sink);
+// Catch records with errors
+parser.on('skip', (e) => {
+  errors.push(e);
+});
+// Wait for stream to be consumed
+await finished(outStream);
+// Catch error from skip event
+assert.deepStrictEqual(errors.map(e => e.message), [
+  'Invalid Record Length: expect 5, got 6 on line 3'
+])
diff --git a/packages/csv-parse/lib/index.js b/packages/csv-parse/lib/index.js
@@ -14,10 +14,9 @@ import {CsvError} from './api/CsvError.js';
 class Parser extends Transform {
   constructor(opts = {}){
     super({...{readableObjectMode: true}, ...opts, encoding: null});
-    this.api = transform(opts);
-    this.api.options.on_skip = (err, chunk) => {
+    this.api = transform({on_skip: (err, chunk) => {
       this.emit('skip', err, chunk);
-    };
+    }, ...opts});
     // Backward compatibility
     this.state = this.api.state;
     this.options = this.api.options;
diff --git a/packages/csv-parse/test/option.skip_records_with_error.coffee b/packages/csv-parse/test/option.skip_records_with_error.coffee
@@ -106,8 +106,7 @@ describe 'Option `skip_records_with_error`', ->
     7,8,9,y
     '''
     parser.end()
-    
-
+  
   it 'handle "CSV_RECORD_INCONSISTENT_FIELDS_LENGTH"', (next) ->
     errors = 0
     parser = parse skip_records_with_error: true, (err, records) ->
@@ -130,6 +129,30 @@ describe 'Option `skip_records_with_error`', ->
     '''
     parser.end()
   
+  describe 'with `bom` option', ->
+  
+    it 'handle "CSV_RECORD_INCONSISTENT_FIELDS_LENGTH" with bom (fix #411)', (next) ->
+      errors = 0
+      parser = parse bom: true, skip_records_with_error: true, (err, records) ->
+        records.should.eql [
+          ['a', 'b', 'c', 'd']
+          ['e', 'f', 'g', 'h']
+        ] unless err
+        errors.should.eql 1
+        next err
+      parser.on 'skip', (err) ->
+        assert_error err,
+          message: 'Invalid Record Length: expect 4, got 3 on line 2'
+          code: 'CSV_RECORD_INCONSISTENT_FIELDS_LENGTH'
+          record: ['1', '2', '3']
+        errors++
+      parser.write '''
+      \ufeffa,b,c,d
+      1,2,3
+      e,f,g,h
+      '''
+      parser.end()
+  
   describe 'with `raw` option', ->
     
     it 'print raw record', (next) ->

-Original file line number
+Diff line change
@@ @@ -1,4 +1,4 @@ @@
 -id,first_name,last_name,email,modified_at
 +id,first_name,last_name,email,modified_at
 ,Ring,Grinyov,[email protected],2022-02-14
 ,Kylie,Lauderdale,[email protected],2022-02-14,
 ,Cammi,Bendix,[email protected],2022-02-14