From e4995a49d3ce34a6916c2cd834a7873083a09c20 Mon Sep 17 00:00:00 2001
From: Bobby Galli <bobbyg603@gmail.com>
Date: Mon, 15 Apr 2024 23:49:45 -0400
Subject: [PATCH] fix: sym file lookup (#107)

* fix: sym file lookup

* fix: shared object matching bug
---
 bin/index.ts     | 11 +++++++++--
 spec/sym.spec.ts |  2 +-
 src/sym.ts       | 14 +++++++++-----
 3 files changed, 19 insertions(+), 8 deletions(-)

diff --git a/bin/index.ts b/bin/index.ts
index f0e5ce6..2f74932 100644
--- a/bin/index.ts
+++ b/bin/index.ts
@@ -113,8 +113,7 @@ import { CommandLineDefinition, argDefinitions, usageDefinitions } from './comma
             const nodeDumpSyms = (await import('node-dump-syms')).dumpSyms;
             symbolFilePaths = symbolFilePaths.map(file => {
                 console.log(`Dumping syms for ${file}...`);
-                const fileNoExt = file.split('.')[0]; // Handle macos .app.dSYM etc.
-                const symFile = join(tmpDir, `${basename(fileNoExt)}.sym`);
+                const symFile = join(tmpDir, `${getSymFileBaseName(file)}.sym`);
                 nodeDumpSyms(file, symFile);
                 return symFile;
             });
@@ -179,6 +178,14 @@ async function getCommandLineOptions(argDefinitions: Array<CommandLineDefinition
         version
     }
 }
+// The rust-minidump-stackwalker symbol lookup implementation removes some extensions from the sym file name for symbol lookups.
+// This is a bit of a mystery and is subject to change when we learn more about how it works.
+// For now, remove any non .so extension in the sym file's base name to satisfy the minidump-stackwalker symbol lookup.
+function getSymFileBaseName(file: string): string {
+    const linuxSoExtensionPattern = /\.so\.?.*$/gm;
+    const fileNoExt = file.split('.')[0];
+    return linuxSoExtensionPattern.test(file) ? basename(file) : basename(fileNoExt);   
+}
 
 function logHelpAndExit() {
     const help = commandLineUsage(usageDefinitions);
diff --git a/spec/sym.spec.ts b/spec/sym.spec.ts
index 6b9c663..1126ced 100644
--- a/spec/sym.spec.ts
+++ b/spec/sym.spec.ts
@@ -18,7 +18,7 @@ describe('getSymFileInfo', () => {
     });
 
     it('should get module name for file with line feeds', async () => {
-        const expected = 'libc++_shared';
+        const expected = 'libc++_shared.so';
 
         const { moduleName } = await getSymFileInfo('./spec/support/android.sym');
 
diff --git a/src/sym.ts b/src/sym.ts
index f097651..ab996a3 100644
--- a/src/sym.ts
+++ b/src/sym.ts
@@ -6,7 +6,7 @@ export async function getSymFileInfo(path: string): Promise<{ dbgId: string, mod
         const matches = Array.from(firstLine?.matchAll(/([0-9a-fA-F]{33,34})\s+(.*)$/gm));
         const dbgId = matches?.at(0)?.at(1) || '';
         const moduleNameWithExt = matches?.at(0)?.at(2) || '';
-        const moduleName = removeNonWindowsExtensions(moduleNameWithExt);
+        const moduleName = removeIgnoredExtensions(moduleNameWithExt);
         return {
             dbgId,
             moduleName
@@ -20,12 +20,16 @@ export async function getSymFileInfo(path: string): Promise<{ dbgId: string, mod
     }
 }
 
-function removeNonWindowsExtensions(moduleName: string): string {
-    const windowsExtensions = ['.dll', '.pdb', '.exe'];
+// The rust-minidump-stackwalker symbol lookup implementation removes some extensions from the module name for symbol lookups.
+// This is a bit of a mystery and is subject to change when we learn more about how it works.
+// For now, remove some module name extensions to satisfy the minidump-stackwalker symbol lookup.
+function removeIgnoredExtensions(moduleName: string): string {
+    // We've seen .pdb, .so, .so.0, and .so.6 in the module lookup
+    const ignoredExtensions = [/\.pdb$/gm, /\.so\.?.*$/gm];
 
-    if (windowsExtensions.some(ext => moduleName.endsWith(ext))) {
+    if (ignoredExtensions.some((regex) => regex.test(moduleName))) {
         return moduleName;
     }
-    
+
     return moduleName.split('.')[0];
 }
\ No newline at end of file