rf: Factor ColumnsMap creation out of TSV loading

Author: effigies

src/files/tsv.test.ts

@@ -64,7 +64,7 @@ Deno.test('TSV loading', async (t) => {
     try {
       await loadTSV(file)
     } catch (e: any) {
-      assertObjectMatch(e, { code: 'TSV_EQUAL_ROWS', location: '/mismatched_row.tsv', line: 3 })
+      assertObjectMatch(e, { code: 'TSV_EQUAL_ROWS', line: 3 })
     }
   })
 

src/files/tsv.ts

@@ -8,6 +8,47 @@ import type { BIDSFile } from '../types/filetree.ts'
 import { filememoizeAsync } from '../utils/memoize.ts'
 import { createUTF8Stream } from './streams.ts'
 
+async function loadColumns(
+  reader: ReadableStreamDefaultReader<string>,
+  headers: string[],
+  maxRows: number,
+): Promise<ColumnsMap> {
+  // Initialize columns in array for construction efficiency
+  const initialCapacity = maxRows >= 0 ? maxRows : 1000
+  const columns: string[][] = headers.map(() => new Array<string>(initialCapacity))
+
+  maxRows = maxRows >= 0 ? maxRows : Infinity
+  let rowIndex = 0 // Keep in scope after loop
+  for (; rowIndex < maxRows; rowIndex++) {
+    const { done, value } = await reader.read()
+    if (done) break
+
+    // Expect a newline at the end of the file, but otherwise error on empty lines
+    if (!value) {
+      const nextRow = await reader.read()
+      if (nextRow.done) break
+      throw { code: 'TSV_EMPTY_LINE', line: rowIndex + 2 }
+    }
+
+    const values = value.split('\t')
+    if (values.length !== headers.length) {
+      throw { code: 'TSV_EQUAL_ROWS', line: rowIndex + 2 }
+    }
+    columns.forEach((column, columnIndex) => {
+      // Double array size if we exceed the current capacity
+      if (rowIndex >= column.length) {
+        column.length = column.length * 2
+      }
+      column[rowIndex] = values[columnIndex]
+    })
+  }
+
+  // Construct map, truncating columns to number of rows read
+  return new ColumnsMap(
+    headers.map((header, index) => [header, columns[index].slice(0, rowIndex)]),
+  )
+}
+
 async function _loadTSV(file: BIDSFile, maxRows: number = -1): Promise<ColumnsMap> {
   const reader = file.stream
     .pipeThrough(createUTF8Stream())
@@ -26,40 +67,7 @@ async function _loadTSV(file: BIDSFile, maxRows: number = -1): Promise<ColumnsMa
       }
     }
 
-    // Initialize columns in array for construction efficiency
-    const initialCapacity = maxRows >= 0 ? maxRows : 1000
-    const columns: string[][] = headers.map(() => new Array<string>(initialCapacity))
-
-    maxRows = maxRows >= 0 ? maxRows : Infinity
-    let rowIndex = 0 // Keep in scope after loop
-    for (; rowIndex < maxRows; rowIndex++) {
-      const { done, value } = await reader.read()
-      if (done) break
-
-      // Expect a newline at the end of the file, but otherwise error on empty lines
-      if (!value) {
-        const nextRow = await reader.read()
-        if (nextRow.done) break
-        throw { code: 'TSV_EMPTY_LINE', location: file.path, line: rowIndex + 2 }
-      }
-
-      const values = value.split('\t')
-      if (values.length !== headers.length) {
-        throw { code: 'TSV_EQUAL_ROWS', location: file.path, line: rowIndex + 2 }
-      }
-      columns.forEach((column, columnIndex) => {
-        // Double array size if we exceed the current capacity
-        if (rowIndex >= column.length) {
-          column.length = column.length * 2
-        }
-        column[rowIndex] = values[columnIndex]
-      })
-    }
-
-    // Construct map, truncating columns to number of rows read
-    return new ColumnsMap(
-      headers.map((header, index) => [header, columns[index].slice(0, rowIndex)]),
-    )
+    return await loadColumns(reader, headers, maxRows)
   } finally {
     await reader.cancel()
   }