lib: cache source maps in vm sources

Cache source maps found in sources parsed with `vm.Script`,
`vm.compileFunction`, and `vm.SourceTextModule`. Also, retrieve source
url from V8 parsing results.

Not like filenames returned by `CallSite.getFileName()` in translating
stack traces, when generating source lines prepended to exceptions,
only resource names can be used as an index to find source maps, which
can be source url magic comments instead. Source url magic comments
can be either a file path or a URL. To verify that source urls with
absolute file paths in the source lines are correctly translated,
snapshots should include the full snapshot urls rather than
neutralizing all the path strings in the stack traces.

Author: legendecas

lib/internal/modules/cjs/loader.js

@@ -1471,9 +1471,9 @@ function wrapSafe(filename, content, cjsModuleInstance, format) {
     );
 
     // Cache the source map for the module if present.
-    const { sourceMapURL } = script;
+    const { sourceMapURL, sourceURL } = script;
     if (sourceMapURL) {
-      maybeCacheSourceMap(filename, content, cjsModuleInstance, false, undefined, sourceMapURL);
+      maybeCacheSourceMap(filename, content, cjsModuleInstance, false, sourceURL, sourceMapURL);
     }
 
     return {
@@ -1498,7 +1498,7 @@ function wrapSafe(filename, content, cjsModuleInstance, format) {
 
   // Cache the source map for the module if present.
   if (result.sourceMapURL) {
-    maybeCacheSourceMap(filename, content, cjsModuleInstance, false, undefined, result.sourceMapURL);
+    maybeCacheSourceMap(filename, content, cjsModuleInstance, false, result.sourceURL, result.sourceMapURL);
   }
 
   return result;

lib/internal/modules/esm/translators.js

@@ -118,10 +118,10 @@ translators.set('module', function moduleStrategy(url, source, isMain) {
 function loadCJSModule(module, source, url, filename, isMain) {
   const compileResult = compileFunctionForCJSLoader(source, filename, false /* is_sea_main */, false);
 
-  const { function: compiledWrapper, sourceMapURL } = compileResult;
+  const { function: compiledWrapper, sourceMapURL, sourceURL } = compileResult;
   // Cache the source map for the cjs module if present.
   if (sourceMapURL) {
-    maybeCacheSourceMap(url, source, module, false, undefined, sourceMapURL);
+    maybeCacheSourceMap(url, source, module, false, sourceURL, sourceMapURL);
   }
 
   const cascadedLoader = require('internal/modules/esm/loader').getOrInitializeCascadedLoader();

lib/internal/modules/esm/utils.js

@@ -342,7 +342,7 @@ function compileSourceTextModule(url, source, cascadedLoader) {
   }
   // Cache the source map for the module if present.
   if (wrap.sourceMapURL) {
-    maybeCacheSourceMap(url, source, wrap, false, undefined, wrap.sourceMapURL);
+    maybeCacheSourceMap(url, source, wrap, false, wrap.sourceURL, wrap.sourceMapURL);
   }
   return wrap;
 }

lib/internal/source_map/source_map_cache.js

@@ -148,10 +148,10 @@ function maybeCacheSourceMap(filename, content, moduleInstance, isGeneratedSourc
     return;
   }
 
-  // FIXME: callers should obtain sourceURL from v8 and pass it
-  // rather than leaving it undefined and extract by regex.
-  if (sourceURL === undefined) {
-    sourceURL = extractSourceURLMagicComment(content);
+  if (sourceURL !== undefined) {
+    // SourceURL magic comment content might be a file path or URL.
+    // Normalize the sourceURL to be a file URL if it is a file path.
+    sourceURL = normalizeReferrerURL(sourceURL);
   }
 
   const data = dataFromUrl(filename, sourceMapURL);
@@ -190,7 +190,7 @@ function maybeCacheGeneratedSourceMap(content) {
     return;
   }
   try {
-    maybeCacheSourceMap(sourceURL, content, null, true, sourceURL);
+    maybeCacheSourceMap(sourceURL, content, null, true);
   } catch (err) {
     // This can happen if the filename is not a valid URL.
     // If we fail to cache the source map, we should not fail the whole process.

lib/internal/vm.js

@@ -25,6 +25,7 @@ const {
 const {
   getOptionValue,
 } = require('internal/options');
+const { maybeCacheSourceMap } = require('internal/source_map/source_map_cache');
 const {
   privateSymbols: {
     contextify_context_private_symbol,
@@ -151,6 +152,7 @@ function internalCompileFunction(
   }
 
   registerImportModuleDynamically(result.function, importModuleDynamically);
+  maybeCacheSourceMap(filename, code, result.function, false, result.sourceURL, result.sourceMapURL);
 
   return result;
 }

lib/internal/vm/module.js

@@ -81,6 +81,7 @@ const kPerContextModuleId = Symbol('kPerContextModuleId');
 const kLink = Symbol('kLink');
 
 const { isContext } = require('internal/vm');
+const { maybeCacheSourceMap } = require('internal/source_map/source_map_cache');
 
 function isModule(object) {
   if (typeof object !== 'object' || object === null || !ObjectPrototypeHasOwnProperty(object, kWrap)) {
@@ -150,6 +151,7 @@ class Module {
       registry.callbackReferrer = this;
       const { registerModule } = require('internal/modules/esm/utils');
       registerModule(this[kWrap], registry);
+      maybeCacheSourceMap(identifier, sourceText, this, false, this[kWrap].sourceURL, this[kWrap].sourceMapURL);
     } else {
       assert(syntheticEvaluationSteps);
       this[kWrap] = new ModuleWrap(identifier, context,

lib/vm.js

@@ -63,6 +63,7 @@ const {
   isContext: _isContext,
   registerImportModuleDynamically,
 } = require('internal/vm');
+const { maybeCacheSourceMap } = require('internal/source_map/source_map_cache');
 const {
   vm_dynamic_import_main_context_default,
   vm_context_no_contextify,
@@ -127,6 +128,7 @@ class Script extends ContextifyScript {
     }
 
     registerImportModuleDynamically(this, importModuleDynamically);
+    maybeCacheSourceMap(filename, code, this, false, this.sourceURL, this.sourceMapURL);
   }
 
   runInThisContext(options) {

src/env_properties.h

@@ -332,6 +332,7 @@
   V(sni_context_string, "sni_context")                                         \
   V(source_string, "source")                                                   \
   V(source_map_url_string, "sourceMapURL")                                     \
+  V(source_url_string, "sourceURL")                                            \
   V(specifier_string, "specifier")                                             \
   V(stack_string, "stack")                                                     \
   V(standard_name_string, "standardName")                                      \

src/module_wrap.cc

@@ -291,6 +291,13 @@ void ModuleWrap::New(const FunctionCallbackInfo<Value>& args) {
         return;
       }
 
+      if (that->Set(context,
+                    realm->env()->source_url_string(),
+                    module->GetUnboundModuleScript()->GetSourceURL())
+              .IsNothing()) {
+        return;
+      }
+
       if (that->Set(context,
                     realm->env()->source_map_url_string(),
                     module->GetUnboundModuleScript()->GetSourceMappingURL())

src/node_contextify.cc

@@ -1098,6 +1098,13 @@ void ContextifyScript::New(const FunctionCallbackInfo<Value>& args) {
     return;
   }
 
+  if (args.This()
+          ->Set(env->context(),
+                env->source_url_string(),
+                v8_script->GetSourceURL())
+          .IsNothing())
+    return;
+
   if (args.This()
           ->Set(env->context(),
                 env->source_map_url_string(),
@@ -1556,6 +1563,15 @@ Local<Object> ContextifyContext::CompileFunctionAndCacheResult(
   Local<Object> result = Object::New(isolate);
   if (result->Set(parsing_context, env->function_string(), fn).IsNothing())
     return Object::New(env->isolate());
+
+  // ScriptOrigin::ResourceName() returns SourceURL magic comment content if
+  // present.
+  if (result
+          ->Set(parsing_context,
+                env->source_url_string(),
+                fn->GetScriptOrigin().ResourceName())
+          .IsNothing())
+    return Object::New(env->isolate());
   if (result
           ->Set(parsing_context,
                 env->source_map_url_string(),
@@ -1793,12 +1809,16 @@ static void CompileFunctionForCJSLoader(
   std::vector<Local<Name>> names = {
       env->cached_data_rejected_string(),
       env->source_map_url_string(),
+      env->source_url_string(),
       env->function_string(),
       FIXED_ONE_BYTE_STRING(isolate, "canParseAsESM"),
   };
   std::vector<Local<Value>> values = {
       Boolean::New(isolate, cache_rejected),
       fn.IsEmpty() ? undefined : fn->GetScriptOrigin().SourceMapUrl(),
+      // ScriptOrigin::ResourceName() returns SourceURL magic comment content if
+      // present.
+      fn.IsEmpty() ? undefined : fn->GetScriptOrigin().ResourceName(),
       fn.IsEmpty() ? undefined : fn.As<Value>(),
       Boolean::New(isolate, can_parse_as_esm),
   };