Add WasmFileKeyGenerator for split Wasm symbol files

documentation/symbols/SSQP_Key_Conventions.md

@@ -195,12 +195,26 @@ Example:
 
 ### WASM (WebAssembly) Modules
 
-WebAssembly symbols, which can be used by browser developer tools to provide source-level debugging experiences, are based on the DWARF format. These are indexed by their DWARF Build ID 
-(built with `-Wl,--build-id` arguments) and the name of the module being debugged via the 
-[buildId property](https://chromedevtools.github.io/devtools-protocol/tot/Debugger/#event-scriptParsed), and the symbol file itself is suffixed with `.s` to disambiguate from the WASM file.
+WebAssembly symbols, which can be used by browser developer tools to provide source-level debugging experiences, are based on the DWARF format. These are indexed by the Build ID stored in the `build_id` custom section of the Wasm binary. The Build ID is a byte sequence typically produced by the linker (e.g., with `-Wl,--build-id` arguments). Both the module and its corresponding symbol file contain the same `build_id` section, allowing them to be matched.
+
+The key uses the actual filename of the file being indexed. For split symbol files, toolchains such as Emscripten produce a separate file (e.g., `foo.debug.wasm`) that contains the DWARF debug sections (`.debug_info`, `.debug_line`, etc.) stripped from the original module.
+
+The final key is formatted as follows:
+
+`<file_name>/<build_id_byte_sequence>/<file_name>`
+
+Example (module):
 
 **File name:** `main.wasm`
 
-**Build ID of file:** `e3b0c44298fc1c149afbf4c8996fb92427ae41e4`
+**Build ID bytes:** `0xe3, 0xb0, 0xc4, 0x42, 0x98, 0xfc, 0x1c, 0x14, 0x9a, 0xfb, 0xf4, 0xc8, 0x99, 0x6f, 0xb9, 0x24, 0x27, 0xae, 0x41, 0xe4`
+
+**Lookup key:** `main.wasm/e3b0c44298fc1c149afbf4c8996fb92427ae41e4/main.wasm`
+
+Example (split symbol file):
+
+**File name:** `main.debug.wasm`
+
+**Build ID bytes:** `(same as module)`
 
-**Lookup key:**: `main.wasm.s/e3b0c44298fc1c149afbf4c8996fb92427ae41e4/main.wasm.s`
+**Lookup key:** `main.debug.wasm/e3b0c44298fc1c149afbf4c8996fb92427ae41e4/main.debug.wasm`

src/Microsoft.SymbolStore/KeyGenerators/FileKeyGenerator.cs

@@ -58,6 +58,7 @@ private IEnumerable<KeyGenerator> GetGenerators()
                 yield return new PDBFileKeyGenerator(Tracer, _file);
                 yield return new PortablePDBFileKeyGenerator(Tracer, _file);
                 yield return new PerfMapFileKeyGenerator(Tracer, _file);
+                yield return new WasmFileKeyGenerator(Tracer, _file);
             }
         }
     }

src/Microsoft.SymbolStore/KeyGenerators/WasmFileKeyGenerator.cs

@@ -0,0 +1,248 @@
+// Licensed to the .NET Foundation under one or more agreements.
+// The .NET Foundation licenses this file to you under the MIT license.
+
+using System;
+using System.Collections.Generic;
+using System.Diagnostics;
+using System.IO;
+using System.Text;
+
+namespace Microsoft.SymbolStore.KeyGenerators
+{
+    public class WasmFileKeyGenerator : KeyGenerator
+    {
+        /// <summary>
+        /// Wasm binary magic number: '\0asm'
+        /// </summary>
+        private static readonly byte[] s_wasmMagic = new byte[] { 0x00, 0x61, 0x73, 0x6D };
+
+        /// <summary>
+        /// Wasm binary format version 1
+        /// </summary>
+        private static readonly byte[] s_wasmVersion = new byte[] { 0x01, 0x00, 0x00, 0x00 };
+
+        /// <summary>
+        /// Custom section ID in Wasm binary format
+        /// </summary>
+        private const byte CustomSectionId = 0;
+
+        /// <summary>
+        /// The name of the custom section containing the build ID
+        /// </summary>
+        private const string BuildIdSectionName = "build_id";
+
+        /// <summary>
+        /// Maximum reasonable build ID length (256 bytes). Protects against
+        /// malformed input causing large allocations.
+        /// </summary>
+        private const int MaxBuildIdLength = 256;
+
+        private readonly SymbolStoreFile _file;
+        private byte[] _buildId;
+        private bool _parsed;
+        private bool _isValid;
+
+        public WasmFileKeyGenerator(ITracer tracer, SymbolStoreFile file)
+            : base(tracer)
+        {
+            _file = file ?? throw new ArgumentNullException(nameof(file));
+        }
+
+        public override bool IsValid()
+        {
+            return HasIndexableWasmBuildId();
+        }
+
+        public bool HasIndexableWasmBuildId()
+        {
+            ParseWasmFile();
+            return _isValid;
+        }
+
+        public override IEnumerable<SymbolStoreKey> GetKeys(KeyTypeFlags flags)
+        {
+            if (IsValid())
+            {
+                if ((flags & KeyTypeFlags.IdentityKey) != 0)
+                {
+                    yield return GetKey(_file.FileName, _buildId);
+                }
+            }
+        }
+
+        /// <summary>
+        /// Create a symbol store key for a Wasm file with a build ID.
+        /// </summary>
+        /// <param name="path">file name and path</param>
+        /// <param name="buildId">build ID bytes from the build_id custom section</param>
+        /// <returns>symbol store key</returns>
+        public static SymbolStoreKey GetKey(string path, byte[] buildId)
+        {
+            Debug.Assert(path != null);
+            Debug.Assert(buildId != null && buildId.Length > 0);
+            string file = GetFileName(path).ToLowerInvariant();
+            return BuildKey(path, prefix: null, buildId, file);
+        }
+
+        /// <summary>
+        /// Parses the Wasm file to validate the header and find the buildId custom section.
+        /// </summary>
+        private void ParseWasmFile()
+        {
+            if (_parsed)
+            {
+                return;
+            }
+            _parsed = true;
+            _isValid = false;
+
+            Stream stream = _file.Stream;
+            long prevPosition = stream.Position;
+            try
+            {
+                stream.Position = 0;
+
+                // Validate magic number
+                byte[] magic = new byte[4];
+                if (stream.Read(magic, 0, 4) != 4)
+                {
+                    return;
+                }
+                for (int i = 0; i < 4; i++)
+                {
+                    if (magic[i] != s_wasmMagic[i])
+                    {
+                        return;
+                    }
+                }
+
+                // Validate version
+                byte[] version = new byte[4];
+                if (stream.Read(version, 0, 4) != 4)
+                {
+                    return;
+                }
+                for (int i = 0; i < 4; i++)
+                {
+                    if (version[i] != s_wasmVersion[i])
+                    {
+                        return;
+                    }
+                }
+
+                // Scan sections for the build_id custom section
+                while (stream.Position < stream.Length)
+                {
+                    int sectionId = stream.ReadByte();
+                    if (sectionId == -1)
+                    {
+                        break;
+                    }
+
+                    uint sectionSize = ReadLEB128Unsigned(stream);
+                    long sectionEnd = stream.Position + sectionSize;
+
+                    // Validate that the section doesn't extend beyond the stream
+                    if (sectionEnd > stream.Length)
+                    {
+                        break;
+                    }
+
+                    if (sectionId == CustomSectionId)
+                    {
+                        string name = ReadWasmString(stream, sectionEnd);
+                        if (name == BuildIdSectionName)
+                        {
+                            // The remainder of the section payload is the build ID
+                            int buildIdLength = (int)(sectionEnd - stream.Position);
+                            if (buildIdLength > 0 && buildIdLength <= MaxBuildIdLength)
+                            {
+                                _buildId = new byte[buildIdLength];
+                                if (stream.Read(_buildId, 0, buildIdLength) == buildIdLength)
+                                {
+                                    _isValid = true;
+                                    return;
+                                }
+                            }
+                        }
+                    }
+
+                    stream.Position = sectionEnd;
+                }
+            }
+            catch (Exception ex) when (ex is IOException || ex is OverflowException || ex is ArgumentOutOfRangeException)
+            {
+                Tracer.Verbose("Error parsing Wasm file {0}: {1}", _file.FileName, ex.Message);
+            }
+            finally
+            {
+                stream.Position = prevPosition;
+            }
+        }
+
+        /// <summary>
+        /// Reads an unsigned LEB128-encoded integer from the stream.
+        /// </summary>
+        private static uint ReadLEB128Unsigned(Stream stream)
+        {
+            uint result = 0;
+            int shift = 0;
+
+            while (true)
+            {
+                int b = stream.ReadByte();
+                if (b == -1)
+                {
+                    throw new IOException("Unexpected end of stream reading LEB128 value.");
+                }
+
+                result |= (uint)(b & 0x7F) << shift;
+                if ((b & 0x80) == 0)
+                {
+                    break;
+                }
+
+                shift += 7;
+                if (shift >= 35)
+                {
+                    throw new OverflowException("LEB128 value too large for uint32.");
+                }
+            }
+
+            return result;
+        }
+
+        /// <summary>
+        /// Maximum section name length we'll read. Names longer than this are
+        /// skipped since they cannot match the sections we're looking for.
+        /// </summary>
+        private const int MaxSectionNameLength = 64;
+
+        /// <summary>
+        /// Reads a Wasm string (LEB128 length prefix followed by UTF-8 bytes).
+        /// Returns null if the string is too long or extends past the section boundary.
+        /// </summary>
+        private static string ReadWasmString(Stream stream, long sectionEnd)
+        {
+            uint length = ReadLEB128Unsigned(stream);
+            if (length == 0)
+            {
+                return string.Empty;
+            }
+            if (length > MaxSectionNameLength || stream.Position + length > sectionEnd)
+            {
+                return null;
+            }
+
+            int stringLength = (int)length;
+            byte[] bytes = new byte[stringLength];
+            int bytesRead = stream.Read(bytes, 0, stringLength);
+            if (bytesRead != stringLength)
+            {
+                return null;
+            }
+
+            return Encoding.UTF8.GetString(bytes);
+        }
+    }
+}

src/tests/Microsoft.SymbolStore.UnitTests/KeyGeneratorTests.cs

@@ -33,6 +33,7 @@ public void FileKeyGenerator()
             PEFileKeyGeneratorInternal(fileGenerator: true);
             PortablePDBFileKeyGeneratorInternal(fileGenerator: true);
             PerfMapFileKeyGeneratorInternal(fileGenerator: true);
+            WasmFileKeyGeneratorInternal(fileGenerator: true);
         }
 
 
@@ -531,5 +532,77 @@ public void SourceFileKeyGenerator()
                 Assert.True(clrKeys.Count() == 0);
             }
         }
+        [Fact]
+        public void WasmFileKeyGenerator()
+        {
+            WasmFileKeyGeneratorInternal(fileGenerator: false);
+        }
+
+        private void WasmFileKeyGeneratorInternal(bool fileGenerator)
+        {
+            // Test 1: Plain Wasm module with build_id (not a symbol file)
+            const string WasmModulePath = "TestBinaries/test_module.wasm";
+            using (Stream stream = File.OpenRead(WasmModulePath))
+            {
+                var file = new SymbolStoreFile(stream, WasmModulePath);
+                KeyGenerator generator = fileGenerator ? (KeyGenerator)new FileKeyGenerator(_tracer, file) : new WasmFileKeyGenerator(_tracer, file);
+
+                Assert.True(generator.IsValid());
+
+                IEnumerable<SymbolStoreKey> identityKey = generator.GetKeys(KeyTypeFlags.IdentityKey);
+                Assert.True(identityKey.Count() == 1);
+                Assert.True(identityKey.First().Index == "test_module.wasm/deadbeef0123456789abcdeffedcba98/test_module.wasm");
+
+                IEnumerable<SymbolStoreKey> symbolKey = generator.GetKeys(KeyTypeFlags.SymbolKey);
+                Assert.True(!symbolKey.Any());
+
+                IEnumerable<SymbolStoreKey> clrKeys = generator.GetKeys(KeyTypeFlags.ClrKeys);
+                Assert.True(!clrKeys.Any());
+            }
+
+            // Test 2: Wasm symbol file with build_id and .debug_info section
+            const string WasmSymbolPath = "TestBinaries/test_module_symbols.wasm";
+            using (Stream stream = File.OpenRead(WasmSymbolPath))
+            {
+                var file = new SymbolStoreFile(stream, WasmSymbolPath);
+                KeyGenerator generator = fileGenerator ? (KeyGenerator)new FileKeyGenerator(_tracer, file) : new WasmFileKeyGenerator(_tracer, file);
+
+                Assert.True(generator.IsValid());
+
+                IEnumerable<SymbolStoreKey> identityKey = generator.GetKeys(KeyTypeFlags.IdentityKey);
+                Assert.True(identityKey.Count() == 1);
+                Assert.True(identityKey.First().Index == "test_module_symbols.wasm/deadbeef0123456789abcdeffedcba98/test_module_symbols.wasm");
+
+                IEnumerable<SymbolStoreKey> symbolKey = generator.GetKeys(KeyTypeFlags.SymbolKey);
+                Assert.True(!symbolKey.Any());
+            }
+
+            // Test 3: Wasm file without build_id should be invalid
+            const string WasmNoBuildIdPath = "TestBinaries/test_module_no_buildid.wasm";
+            using (Stream stream = File.OpenRead(WasmNoBuildIdPath))
+            {
+                var file = new SymbolStoreFile(stream, WasmNoBuildIdPath);
+                var generator = new WasmFileKeyGenerator(_tracer, file);
+
+                Assert.False(generator.IsValid());
+
+                IEnumerable<SymbolStoreKey> identityKey = generator.GetKeys(KeyTypeFlags.IdentityKey);
+                Assert.True(!identityKey.Any());
+            }
+
+            // Test 4: Wasm file with a custom section name longer than 64 chars before build_id
+            const string WasmLongNamePath = "TestBinaries/test_module_long_section_name.wasm";
+            using (Stream stream = File.OpenRead(WasmLongNamePath))
+            {
+                var file = new SymbolStoreFile(stream, WasmLongNamePath);
+                var generator = new WasmFileKeyGenerator(_tracer, file);
+
+                Assert.True(generator.IsValid());
+
+                IEnumerable<SymbolStoreKey> identityKey = generator.GetKeys(KeyTypeFlags.IdentityKey);
+                Assert.True(identityKey.Count() == 1);
+                Assert.True(identityKey.First().Index == "test_module_long_section_name.wasm/deadbeef0123456789abcdeffedcba98/test_module_long_section_name.wasm");
+            }
+        }
     }
 }