Update instruments import to use integeruniquer.index file (#64)

I knew early on that `integeruniquer.index` could be used to index into `integeruniquer.data`, but I initially thought it was an optimization rather than a necessity. It seems like if there's data past the 1MB threshold in `integeruniquer.data`, then `integeruniquer.index` is actually quite useful.

The file seems to contain `[byte offset, MB offset]` pairs encoded as two 32 bit unsigned little endian integers. Using that to decode the integer arrays encoded in `integeruniquer.data` allows the file in #63 to load.

Fixes #63

import/instruments.ts

@@ -264,6 +264,9 @@ class BinReader {
   hasMore() {
     return this.bytePos < this.view.byteLength
   }
+  bytesLeft() {
+    return this.view.byteLength - this.bytePos
+  }
   readUint8() {
     this.bytePos++
     if (this.bytePos > this.view.byteLength) return 0
@@ -344,10 +347,11 @@ async function getRawSampleList(core: TraceDirectoryTree): Promise<Sample[]> {
 async function getIntegerArrays(samples: Sample[], core: TraceDirectoryTree): Promise<number[][]> {
   const uniquing = getOrThrow(core.subdirectories, 'uniquing')
   const arrayUniquer = getOrThrow(uniquing.subdirectories, 'arrayUniquer')
-  const integeruniquer = getOrThrow(arrayUniquer.files, 'integeruniquer.data')
-  const reader = new BinReader(await readAsArrayBuffer(integeruniquer))
+  const integeruniquerindex = getOrThrow(arrayUniquer.files, 'integeruniquer.index')
+  const integeruniquerdata = getOrThrow(arrayUniquer.files, 'integeruniquer.data')
 
-  let arrays: number[][] = []
+  // integeruniquer.index is a binary file containing an array of [byte offset, MB offset] pairs
+  // that indicate where array data starts in the .data file
 
   // integeruniquer.data is a binary file containing an array of arrays of 64 bit integer.
   // The schema is a 32 byte header followed by a stream of arrays.
@@ -355,14 +359,30 @@ async function getIntegerArrays(samples: Sample[], core: TraceDirectoryTree): Pr
 
   // This table contains the memory addresses of stack frames
 
-  // Header we don't care about
-  reader.seek(32)
+  const indexreader = new BinReader(await readAsArrayBuffer(integeruniquerindex))
+  const datareader = new BinReader(await readAsArrayBuffer(integeruniquerdata))
 
-  while (reader.hasMore()) {
-    let length = reader.readUint32()
+  // Header we don't care about
+  indexreader.seek(32)
+
+  let arrays: number[][] = []
+
+  while (indexreader.hasMore()) {
+    const byteOffset = indexreader.readUint32() + indexreader.readUint32() * (1024 * 1024)
+
+    if (byteOffset === 0) {
+      // The first entry in the index table seems to just indicate the offset of
+      // the header into the data file
+      continue
+    }
+
+    datareader.seek(byteOffset)
+
+    let length = datareader.readUint32()
     let array: number[] = []
+
     while (length--) {
-      array.push(reader.readUint64())
+      array.push(datareader.readUint64())
     }
     arrays.push(array)
   }
@@ -453,7 +473,6 @@ export async function importFromInstrumentsTrace(
   const core = getCoreDirForRun(tree, selectedRun)
   let samples = await getRawSampleList(core)
   const arrays = await getIntegerArrays(samples, core)
-
   const backtraceIDtoStack = new Map<number, FrameInfo[]>()
 
   const profile = new StackListProfileBuilder(lastOf(samples)!.timestamp)

sample/profiles/Instruments/8.3.3/integer-uniquer-data.grammar

@@ -1,13 +1,13 @@
 <?xml version="1.0" encoding="UTF-8"?>
 <ufwb version="1.17">
-    <grammar name="Integer Uniquer Data" start="id:165" author="Jamie Wong">
+    <grammar name="Integer Uniquer Data" start="id:1" author="Jamie Wong">
         <description>Grammar for my file format</description>
-        <structure name="Root" id="165" encoding="ISO_8859-1:1987" endian="little" signed="no">
-            <binary name="Header" id="166" length="32"/>
-            <structure name="Array" id="167" repeatmax="-1">
-                <number name="Size" id="168" fillcolor="CEFFD2" type="integer" length="4"/>
-                <structure name="Data" id="169" repeatmin="Size" repeatmax="Size">
-                    <number name="Value" id="170" type="integer" length="8"/>
+        <structure name="Root" id="1" encoding="ISO_8859-1:1987" endian="little" signed="no">
+            <binary name="integeruniquer.data file" id="2" length="32"/>
+            <structure name="Array" id="3" repeatmax="-1">
+                <number name="Size" id="4" fillcolor="CEFFD2" type="integer" length="4"/>
+                <structure name="Data" id="5" repeatmin="Size" repeatmax="Size">
+                    <number name="Value" id="6" type="integer" length="8"/>
                 </structure>
             </structure>
         </structure>

sample/profiles/Instruments/8.3.3/integer-uniquer-index.grammar

@@ -0,0 +1,13 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<ufwb version="1.17">
+    <grammar name="INDEX grammar" start="id:10" author="Jamie Wong" fileextension="index">
+        <description>Grammar for INDEX files</description>
+        <structure name="integeruniquer.index file" id="10" encoding="ISO_8859-1:1987" endian="big" signed="no">
+            <binary name="Header" id="12" length="32"/>
+            <structure name="Entry" id="16" length="8" repeatmax="-1">
+                <number name="ByteOffset" id="15" type="integer" length="4" endian="little"/>
+                <number name="MegabyteOffset" id="18" type="integer" length="4" endian="little"/>
+            </structure>
+        </structure>
+    </grammar>
+</ufwb>