bug: fix bug in BufferedFormatReader

the undefined state of peekedRow was conflicting with the fact the readRow might return undefined.

Author: livetocode

package-lock.json

@@ -1,6 +1,6 @@
 {
   "name": "tabular-data-differ",
-  "version": "1.1.0",
+  "version": "1.1.1",
   "description": "A very efficient library for diffing two sorted streams of tabular data, such as CSV files.",
   "keywords": [
     "table",

package.json

@@ -1774,6 +1774,44 @@ describe('differ', () => {
                 same: 5        
             });
         });
+        test('should work with real source files (CSV) and support duplicate key handling', async () => {
+            const output = new FakeFormatWriter();
+            const differ = diff({
+                oldSource: './tests/a2.csv',
+                newSource: './tests/b2.csv',
+                keys: ['id'],
+                duplicateKeyHandling: 'keepFirstRow',
+            });
+            await differ.to({
+                destination: {
+                    format: 'custom',
+                    writer: output,
+                }
+            });
+            expect(output.header?.columns).toEqual([ 'id', 'a', 'b', 'c' ]);
+            expect(output.diffs).toEqual([
+                { delta: -1, status: 'deleted', oldRow: [ '01', 'a1', 'b1', 'c1' ] },
+                {
+                  delta: 0,
+                  status: 'modified',
+                  oldRow: [ '04', 'a4', 'b4', 'c4' ],
+                  newRow: [ '04', 'aa4', 'bb4', 'cc4' ]
+                },
+                { delta: -1, status: 'deleted', oldRow: [ '05', 'a5', 'b5', 'c5' ] },
+                { delta: -1, status: 'deleted', oldRow: [ '06', 'a6', 'b6', 'c6' ] },
+                { delta: 1, status: 'added', newRow: [ '10', 'a10', 'b10', 'c10' ] },
+            ]);
+            console.log(output.footer?.stats);
+            expect(output.footer?.stats).toEqual({
+                totalComparisons: 10,
+                totalChanges: 5,
+                changePercent: 50,
+                added: 1,
+                deleted: 3,
+                modified: 1,
+                same: 5
+            });
+        });
         test('should work with http streams (CSV)', async () => {
             const currentDir = process.cwd().replaceAll('\\', '/');
             const stats = await diff({

src/differ.test.ts

@@ -127,10 +127,13 @@ export type FormatWriterFactory = (options: FormatWriterOptions) => FormatWriter
 
 export class BufferedFormatReader implements FormatReader {
     private peekedRow: Row | undefined;
+    private hasPeekedRow = false;
 
     constructor(private reader: FormatReader) {}
 
     open(): Promise<void> {
+        this.hasPeekedRow = false;
+        this.peekedRow = undefined;
         return this.reader.open();
     }
 
@@ -139,17 +142,19 @@ export class BufferedFormatReader implements FormatReader {
     }
 
     async peekRow(): Promise<Row | undefined> {
-        if (this.peekedRow) { 
+        if (this.hasPeekedRow) { 
             return this.peekedRow;
         }
         this.peekedRow = await this.reader.readRow();
+        this.hasPeekedRow = true;
         return this.peekedRow;
     }
 
     async readRow(): Promise<Row | undefined> {
-        if (this.peekedRow) {
+        if (this.hasPeekedRow) {
             const result = this.peekedRow;
             this.peekedRow = undefined;
+            this.hasPeekedRow = false;
             return result;
         }
         return await this.reader.readRow();

src/formats.ts

@@ -0,0 +1,13 @@
+id,a,b,c
+01,a1,b1,c1
+02,a2,b2,c2
+03,a3,b3,c3
+03,a3,b3,c3
+04,a4,b4,c4
+05,a5,b5,c5
+06,a6,b6,c6
+07,a7,b7,c7
+07,a77,b77,c77
+08,a8,b8,c8
+09,a9,b9,c9
+09,a9,b9,c9

tests/a2.csv

@@ -0,0 +1,11 @@
+id,a,b,c
+02,a2,b2,c2
+03,a3,b3,c3
+03,a3,b3,c3
+03,a3,b3,c3
+04,aa4,bb4,cc4
+07,a7,b7,c7
+07,a7b,b7b,c7b
+08,a8,b8,c8
+09,a9,b9,c9
+10,a10,b10,c10