diff --git a/src/import/callgrind.ts b/src/import/callgrind.ts
index 9821768..6f5a186 100644
--- a/src/import/callgrind.ts
+++ b/src/import/callgrind.ts
@@ -346,7 +346,7 @@ class CallgrindParser {
   private savedFunctionNames: {[id: string]: string} = {}
 
   constructor(contents: TextFileContent, private importedFileName: string) {
-    this.lines = contents.splitLines()
+    this.lines = [...contents.splitLines()]
     this.lineNum = 0
   }
 
diff --git a/src/import/instruments.ts b/src/import/instruments.ts
index 6e5a19c..8dd49bc 100644
--- a/src/import/instruments.ts
+++ b/src/import/instruments.ts
@@ -14,7 +14,7 @@ import {FileSystemDirectoryEntry, FileSystemEntry, FileSystemFileEntry} from './
 import {MaybeCompressedDataReader, TextFileContent} from './utils'
 
 function parseTSV<T>(contents: TextFileContent): T[] {
-  const lines = contents.splitLines().map(l => l.split('\t'))
+  const lines = [...contents.splitLines()].map(l => l.split('\t'))
 
   const headerLine = lines.shift()
   if (!headerLine) return []
diff --git a/src/import/papyrus.ts b/src/import/papyrus.ts
index 2ff69f6..88477bd 100644
--- a/src/import/papyrus.ts
+++ b/src/import/papyrus.ts
@@ -38,9 +38,9 @@ export function importFromPapyrus(papyrusProfile: TextFileContent): Profile {
   const profile = new CallTreeProfileBuilder()
   profile.setValueFormatter(new TimeFormatter('milliseconds'))
 
-  const papyrusProfileLines = papyrusProfile
-    .splitLines()
-    .filter(line => !/^$|^Log closed$|log opened/.exec(line))
+  const papyrusProfileLines = [...papyrusProfile.splitLines()].filter(
+    line => !/^$|^Log closed$|log opened/.exec(line),
+  )
 
   let startValue = -1
   const firstLineParsed = parseLine(papyrusProfileLines[0])
diff --git a/src/import/utils.ts b/src/import/utils.ts
index 67de418..72341f6 100644
--- a/src/import/utils.ts
+++ b/src/import/utils.ts
@@ -8,7 +8,7 @@ export interface ProfileDataSource {
 }
 
 export interface TextFileContent {
-  splitLines(): string[]
+  splitLines(): Iterable<string>
   firstChunk(): string
   parseAsJSON(): any
 }
@@ -19,7 +19,7 @@ export interface TextFileContent {
 //
 // We provide a simple splitLines() which returns simple strings under the
 // assumption that most extremely large text profiles will be broken into many
-// lines.  This isn't true in the general case, but will be true for most common
+// lines. This isn't true in the general case, but will be true for most common
 // large files.
 //
 // See: https://github.com/v8/v8/blob/8b663818fc311217c2cdaaab935f020578bfb7a8/src/objects/string.h#L479-L483
@@ -151,17 +151,26 @@ export class BufferBackedTextFileContent implements TextFileContent {
     }
   }
 
-  splitLines(): string[] {
-    let parts: string[] = this.chunks[0].split('\n')
-    for (let i = 1; i < this.chunks.length; i++) {
-      const chunkParts = this.chunks[i].split('\n')
-      if (chunkParts.length === 0) continue
-      if (parts.length > 0) {
-        parts[parts.length - 1] += chunkParts.shift()
+  splitLines(): Iterable<string> {
+    const iterator = function* (this: BufferBackedTextFileContent) {
+      let lineBuffer: string = ''
+      for (let chunk of this.chunks) {
+        const fragments = chunk.split('\n')
+        for (let i = 0; i < fragments.length; i++) {
+          if (i === 0) lineBuffer += fragments[i]
+          else {
+            yield lineBuffer
+            lineBuffer = fragments[i]
+          }
+        }
       }
-      parts = parts.concat(chunkParts)
+
+      yield lineBuffer
+    }
+
+    return {
+      [Symbol.iterator]: iterator.bind(this),
     }
-    return parts
   }
 
   firstChunk(): string {
diff --git a/src/lib/utils.test.ts b/src/lib/utils.test.ts
index 7d196a1..a8c415d 100644
--- a/src/lib/utils.test.ts
+++ b/src/lib/utils.test.ts
@@ -247,11 +247,18 @@ test('BufferBackedTextFileContent.firstChunk', async () => {
 })
 
 test('BufferBackedTextFileContent.splitLines', async () => {
-  await withMockedFileChunkSizeForTests(2, () => {
-    const str = 'may\nyour\nrope\nbe\nlong'
+  function testWithString(str: string) {
     const buffer = new TextEncoder().encode(str).buffer
     const content = new BufferBackedTextFileContent(buffer)
-    expect(content.splitLines()).toEqual(['may', 'your', 'rope', 'be', 'long'])
+    expect([...content.splitLines()]).toEqual(str.split('\n'))
+  }
+
+  await withMockedFileChunkSizeForTests(2, () => {
+    testWithString('may\nyour\nrope\nbe\nlong')
+    testWithString('one-single-line')
+    testWithString('multiple\n\nnew\n\nlines')
+    testWithString('end-with-new-line\n')
+    testWithString('')
   })
 })