fix(csv-parse): retain the last emitted record byte offset

packages/csv-parse/lib/api/index.js

@@ -749,7 +749,7 @@ const transform = function (original_options = {}) {
           return;
         }
       }
-      this.info.bytes_records += this.info.bytes;
+      this.info.bytes_records = this.info.bytes;
       push(record);
     },
     // Return a tuple with the error and the casted value

packages/csv-parse/test/api.info.ts

@@ -7,7 +7,7 @@ describe("API info", function () {
       if (!info) return next(Error("Invalid assessment"));
       info.should.eql({
         bytes: 10,
-        bytes_records: 16,
+        bytes_records: 10,
         columns: false,
         comment_lines: 0,
         empty_lines: 0,
@@ -25,7 +25,7 @@ describe("API info", function () {
       if (!info) return next(Error("Invalid assessment"));
       info.should.eql({
         bytes: 11,
-        bytes_records: 17,
+        bytes_records: 11,
         columns: false,
         comment_lines: 0,
         empty_lines: 0,
@@ -61,7 +61,7 @@ describe("API info", function () {
       if (!info) return next(Error("Invalid assessment"));
       info.should.eql({
         bytes: 20,
-        bytes_records: 41,
+        bytes_records: 20,
         columns: false,
         comment_lines: 0,
         empty_lines: 0,

packages/csv-parse/test/info.bytes_records.offset.js

@@ -0,0 +1,53 @@
+import assert from "node:assert/strict";
+import { parse } from "../lib/index.js";
+
+describe("info bytes_records as the last emitted record offset", function () {
+  it("keeps the previous record offset throughout field casting", function (next) {
+    const fieldOffsets = [];
+    parse(
+      "a,b\nc,d\ne,f",
+      {
+        cast(value, info) {
+          fieldOffsets.push(info.bytes_records);
+          return value;
+        },
+      },
+      (err, records, info) => {
+        if (err) return next(err);
+        assert.deepEqual(fieldOffsets, [0, 0, 4, 4, 8, 8]);
+        assert.equal(info.bytes, 11);
+        assert.equal(info.bytes_records, 11);
+        next();
+      },
+    );
+  });
+
+  for (const [name, input, options, expected] of [
+    [
+      "skipped middle record",
+      "a,b\nc,d\ne,f",
+      { on_record: (record) => (record[0] === "c" ? null : record) },
+      11,
+    ],
+    [
+      "skipped final record",
+      "a,b\nc,d\ne,f",
+      { on_record: (record) => (record[0] === "e" ? null : record) },
+      8,
+    ],
+    ["column headers", "x,y\na,b\nc,d", { columns: true }, 11],
+    ["from_line", "a,b\nc,d\ne,f", { from_line: 2 }, 11],
+    ["UTF8 BOM", "\ufeffa,b\nc,d", { bom: true }, 10],
+    ["no emitted records", "a,b\nc,d", { on_record: () => null }, 0],
+    ["one emitted record", "a,b", {}, 3],
+  ]) {
+    it(`reports an absolute offset with ${name}`, function (next) {
+      parse(input, options, (err, records, info) => {
+        if (err) return next(err);
+        assert.equal(info.bytes_records, expected);
+        assert.ok(info.bytes_records <= Buffer.byteLength(input));
+        next();
+      });
+    });
+  }
+});

packages/csv-parse/test/info.bytes_records.ts

@@ -60,7 +60,7 @@ describe("info bytes_records", function () {
     records.should.eql([
       [0, 0],
       [18, 18],
-      [46, 46],
+      [28, 28],
     ]);
   });
 });