@shapeshift-labs/frontier-lang-compiler 0.2.102 → 0.2.104

package/dist/native-region-scanner-main.js

@@ -1,4 +1,4 @@
-import { idFragment, uniqueStrings, upperFirst } from './native-import-utils.js';
+import { idFragment } from './native-import-utils.js';
 import {
   nativeDeclaration,
   nativeImportDeclaration,
@@ -7,30 +7,16 @@ import {
   splitParameters,
   splitTypeParameters
 } from './native-region-scanner-core.js';
-
-function scanPython(input) {
-  const declarations = [];
-  for (const { line, number } of sourceLines(input.sourceText)) {
-    const trimmed = line.trim();
-    let match;
-    if ((match = trimmed.match(/^(?:async\s+)?def\s+([A-Za-z_]\w*)\s*\(([^)]*)\)\s*:/))) {
-      declarations.push(nativeDeclaration(input, number, 'FunctionDef', 'function', match[1], { parameters: splitParameters(match[2]) }, true));
-    } else if ((match = trimmed.match(/^class\s+([A-Za-z_]\w*)/))) {
-      declarations.push(nativeDeclaration(input, number, 'ClassDef', 'class', match[1], {}, true));
-    } else if ((match = trimmed.match(/^(?:from\s+([A-Za-z_][\w.]*)\s+import\s+.+|import\s+([A-Za-z_][\w.]*))/))) {
-      declarations.push(nativeImportDeclaration(input, number, match[1] ?? match[2], 'Import', 'module'));
-    }
-  }
-  return declarations;
-}
+import { braceBlockSpan } from './native-region-scanner-spans.js';
 
 function scanRust(input) {
   const declarations = [];
-  for (const { line, number } of sourceLines(input.sourceText)) {
+  const lines = sourceLines(input.sourceText);
+  for (const [index, { line, number }] of lines.entries()) {
     const trimmed = line.trim();
     let match;
     if ((match = trimmed.match(/^(?:pub(?:\([^)]*\))?\s+)?(?:async\s+)?fn\s+([A-Za-z_]\w*)\s*\(([^)]*)\)/))) {
-      declarations.push(nativeDeclaration(input, number, 'ItemFn', 'function', match[1], { parameters: splitParameters(match[2]) }, trimmed.includes('{')));
+      declarations.push(nativeDeclaration(input, number, 'ItemFn', 'function', match[1], { parameters: splitParameters(match[2]) }, trimmed.includes('{'), { span: trimmed.includes('{') ? braceBlockSpan(input, lines, index) : undefined }));
     } else if ((match = trimmed.match(/^(?:pub(?:\([^)]*\))?\s+)?struct\s+([A-Za-z_]\w*)/))) {
       declarations.push(nativeDeclaration(input, number, 'ItemStruct', 'type', match[1], {}, trimmed.includes('{')));
     } else if ((match = trimmed.match(/^(?:pub(?:\([^)]*\))?\s+)?enum\s+([A-Za-z_]\w*)/))) {
@@ -52,7 +38,8 @@ function scanRust(input) {
 
 function scanCLike(input) {
   const declarations = [];
-  for (const { line, number } of sourceLines(input.sourceText)) {
+  const lines = sourceLines(input.sourceText);
+  for (const [index, { line, number }] of lines.entries()) {
     const trimmed = line.trim();
     let match;
     if ((match = trimmed.match(/^#\s*include\s+[<"]([^>"]+)[>"]/))) {
@@ -60,30 +47,11 @@ function scanCLike(input) {
     } else if ((match = trimmed.match(/^#\s*define\s+([A-Za-z_]\w*)/))) {
       declarations.push(nativeMacroLoss(input, number, trimmed, 'preprocessor', match[1]));
     } else if ((match = trimmed.match(/^typedef\s+struct(?:\s+([A-Za-z_]\w*))?/))) {
-      declarations.push(nativeDeclaration(input, number, 'TypedefStructDeclaration', 'type', match[1] ?? `anonymous_struct_${number}`, {}, trimmed.includes('{')));
+      declarations.push(nativeDeclaration(input, number, 'TypedefStructDeclaration', 'type', match[1] ?? `anonymous_struct_${number}`, {}, trimmed.includes('{'), { span: trimmed.includes('{') ? braceBlockSpan(input, lines, index) : undefined }));
     } else if ((match = trimmed.match(/^(?:struct|enum)\s+([A-Za-z_]\w*)/))) {
-      declarations.push(nativeDeclaration(input, number, 'TagDeclaration', 'type', match[1], {}, trimmed.includes('{')));
+      declarations.push(nativeDeclaration(input, number, 'TagDeclaration', 'type', match[1], {}, trimmed.includes('{'), { span: trimmed.includes('{') ? braceBlockSpan(input, lines, index) : undefined }));
     } else if ((match = trimmed.match(/^(?:[A-Za-z_][\w\s*:&<>]+)\s+([A-Za-z_]\w*)\s*\(([^;{}]*)\)\s*(?:;|\{)?$/))) {
-      declarations.push(nativeDeclaration(input, number, 'FunctionDeclaration', 'function', match[1], { parameters: splitParameters(match[2]) }, trimmed.endsWith('{')));
-    }
-  }
-  return declarations;
-}
-
-function scanJava(input) {
-  const declarations = [];
-  for (const { line, number } of sourceLines(input.sourceText)) {
-    const trimmed = line.trim();
-    let match;
-    if ((match = trimmed.match(/^package\s+([A-Za-z_][\w.]*);/))) {
-      declarations.push(nativeDeclaration(input, number, 'PackageDeclaration', 'package', match[1], {}, false));
-    } else if ((match = trimmed.match(/^import\s+(?:static\s+)?([A-Za-z_][\w.*]*);/))) {
-      declarations.push(nativeImportDeclaration(input, number, match[1], 'ImportDeclaration', 'package'));
-    } else if ((match = trimmed.match(/^(?:(?:public|protected|private|abstract|final|static|sealed|non-sealed)\s+)*(class|interface|enum|record|@interface)\s+([A-Za-z_$][\w$]*)/))) {
-      const kind = match[1] === '@interface' ? 'AnnotationDeclaration' : `${upperFirst(match[1])}Declaration`;
-      declarations.push(nativeDeclaration(input, number, kind, javaSymbolKind(match[1]), match[2], {}, trimmed.includes('{')));
-    } else if ((match = trimmed.match(/^(?:(?:public|protected|private|abstract|final|static|synchronized|native)\s+)*(?:<[^>]+>\s+)?[A-Za-z_$][\w$<>\[\].?,\s]*\s+([A-Za-z_$][\w$]*)\s*\(([^)]*)\)\s*(?:throws\s+[^{]+)?(?:\{|;)?$/))) {
-      declarations.push(nativeDeclaration(input, number, 'MethodDeclaration', 'method', match[1], { parameters: splitParameters(match[2]) }, trimmed.includes('{')));
+      declarations.push(nativeDeclaration(input, number, 'FunctionDeclaration', 'function', match[1], { parameters: splitParameters(match[2]) }, trimmed.endsWith('{'), { span: trimmed.endsWith('{') ? braceBlockSpan(input, lines, index) : undefined }));
     }
   }
   return declarations;
@@ -92,7 +60,8 @@ function scanJava(input) {
 function scanGo(input) {
   const declarations = [];
   let inImportBlock = false;
-  for (const { line, number } of sourceLines(input.sourceText)) {
+  const lines = sourceLines(input.sourceText);
+  for (const [index, { line, number }] of lines.entries()) {
     const trimmed = line.trim();
     let match;
     if (inImportBlock) {
@@ -120,12 +89,12 @@ function scanGo(input) {
         receiver,
         typeParameters: splitTypeParameters(match[3]),
         parameters: splitParameters(match[4])
-      }, trimmed.includes('{')));
+      }, trimmed.includes('{'), { span: trimmed.includes('{') ? braceBlockSpan(input, lines, index) : undefined }));
     } else if ((match = trimmed.match(/^func\s+([A-Za-z_]\w*)(?:\s*\[([^\]]+)\])?\s*\(([^)]*)\)/))) {
       declarations.push(nativeDeclaration(input, number, 'FuncDecl', 'function', match[1], {
         typeParameters: splitTypeParameters(match[2]),
         parameters: splitParameters(match[3])
-      }, trimmed.includes('{')));
+      }, trimmed.includes('{'), { span: trimmed.includes('{') ? braceBlockSpan(input, lines, index) : undefined }));
     } else if ((match = trimmed.match(/^var\s+([A-Za-z_]\w*)\b/))) {
       declarations.push(nativeDeclaration(input, number, 'VarDecl', 'variable', match[1], {}, false));
     } else if ((match = trimmed.match(/^const\s+([A-Za-z_]\w*)\b/))) {
@@ -135,82 +104,6 @@ function scanGo(input) {
   return declarations;
 }
 
-function scanSwift(input) {
-  const declarations = [];
-  const protocols = new Set();
-  for (const { line, number } of sourceLines(input.sourceText)) {
-    const trimmed = line.trim();
-    const declarationLine = trimmed.replace(/^(?:@[A-Za-z_][\w.]+(?:\([^)]*\))?\s+)*/, '');
-    let match;
-    if ((match = declarationLine.match(/^import\s+(?:(?:struct|class|enum|protocol|func|var)\s+)?([A-Za-z_]\w*(?:\.[A-Za-z_]\w*)*)/))) {
-      declarations.push(nativeImportDeclaration(input, number, match[1], 'ImportDecl', 'module'));
-    } else if ((match = declarationLine.match(/^(?:(?:public|private(?:\([^)]*\))?|fileprivate|internal|open|final|indirect)\s+)*(struct|class|enum|protocol|actor)\s+([A-Za-z_]\w*)/))) {
-      if (match[1] === 'protocol') protocols.add(match[2]);
-      declarations.push(nativeDeclaration(input, number, `${upperFirst(match[1])}Decl`, swiftSymbolKind(match[1]), match[2], {}, declarationLine.includes('{')));
-    } else if ((match = declarationLine.match(/^(?:(?:public|private(?:\([^)]*\))?|fileprivate|internal|open)\s+)*extension\s+([A-Za-z_]\w*(?:\.[A-Za-z_]\w*)*)(.*)$/))) {
-      const extensionFields = parseSwiftExtensionTail(match[2]);
-      const isProtocolExtension = protocols.has(match[1]) || /Protocol$/.test(match[1]);
-      declarations.push(nativeDeclaration(input, number, isProtocolExtension ? 'ProtocolExtensionDecl' : 'ExtensionDecl', 'implementation', `${match[1]}.${isProtocolExtension ? 'protocolExtension' : 'extension'}`, {
-        extendedType: match[1],
-        ...extensionFields
-      }, declarationLine.includes('{')));
-    } else if ((match = declarationLine.match(/^(?:(?:public|private(?:\([^)]*\))?|fileprivate|internal|open|static|class|mutating|nonmutating|override|required|convenience|isolated|nonisolated)\s+)*func\s+([A-Za-z_]\w*|`[^`]+`)(?:\s*<([^>]+)>)?\s*\(([^)]*)\)/))) {
-      declarations.push(nativeDeclaration(input, number, 'FunctionDecl', 'function', unquoteSwiftIdentifier(match[1]), {
-        typeParameters: splitTypeParameters(match[2]),
-        parameters: splitParameters(match[3])
-      }, declarationLine.includes('{')));
-    } else if ((match = declarationLine.match(/^(?:(?:public|private(?:\([^)]*\))?|fileprivate|internal|open|static|class|final|lazy|weak|unowned|override|required|nonisolated)\s+)*(let|var)\s+([A-Za-z_]\w*)\b(?::\s*([^={]+))?/))) {
-      declarations.push(nativeDeclaration(input, number, 'PropertyDecl', 'property', match[2], {
-        binding: match[1],
-        valueType: match[3]?.trim()
-      }, declarationLine.includes('{') || declarationLine.includes('=>')));
-    } else if ((match = declarationLine.match(/^(?:(?:public|private(?:\([^)]*\))?|fileprivate|internal|open)\s+)*typealias\s+([A-Za-z_]\w*)\b(?:\s*=\s*(.+))?/))) {
-      declarations.push(nativeDeclaration(input, number, 'TypealiasDecl', 'type', match[1], { target: match[2]?.trim() }, false));
-    }
-  }
-  return declarations;
-}
-
-function scanCSharp(input) {
-  const declarations = [];
-  for (const { line, number } of sourceLines(input.sourceText)) {
-    const trimmed = line.trim();
-    let match;
-    if ((match = trimmed.match(/^using\s+([A-Za-z_]\w*)\s*=\s*(.+?)\s*;/))) {
-      declarations.push(nativeDeclaration(input, number, 'UsingAliasDirective', 'type', match[1], { target: match[2].trim() }, false));
-    } else if ((match = trimmed.match(/^using\s+(?:static\s+)?([A-Za-z_][\w.]*)\s*;/))) {
-      declarations.push(nativeImportDeclaration(input, number, match[1], 'UsingDirective', 'namespace'));
-    } else if ((match = trimmed.match(/^namespace\s+([A-Za-z_][\w.]*)/))) {
-      declarations.push(nativeDeclaration(input, number, 'NamespaceDeclaration', 'namespace', match[1], {}, trimmed.includes('{')));
-    } else if ((match = trimmed.match(/^(?:(?:public|protected|private|internal|static|unsafe|new)\s+)*delegate\s+(.+?)\s+([A-Za-z_]\w*)\s*\(([^)]*)\)\s*;/))) {
-      declarations.push(nativeDeclaration(input, number, 'DelegateDeclaration', 'type', match[2], {
-        returnType: match[1].trim(),
-        parameters: splitParameters(match[3])
-      }, false));
-    } else if ((match = trimmed.match(/^(?:(?:public|protected|private|internal|abstract|sealed|static|partial|readonly|ref|unsafe)\s+)*(class|interface|struct|enum|record(?:\s+(?:class|struct))?)\s+([A-Za-z_]\w*)/))) {
-      declarations.push(nativeDeclaration(input, number, csharpDeclarationKind(match[1]), csharpSymbolKind(match[1]), match[2], { csharpKind: match[1].replace(/\s+/g, ' ') }, trimmed.includes('{')));
-    } else if ((match = trimmed.match(/^(?:(?:public|protected|private|internal|static|virtual|override|async|partial|sealed|abstract|extern|new|unsafe|readonly)\s+)*(?:[A-Za-z_][\w<>\[\].?,\s]*\??|void)\s+([A-Za-z_]\w*)\s*\(([^)]*)\)\s*(?:=>.*|\{|;)?$/))) {
-      const parameters = splitParameters(match[2]);
-      const extensionReceiver = csharpExtensionReceiver(parameters);
-      declarations.push(nativeDeclaration(input, number, extensionReceiver ? 'ExtensionMethodDeclaration' : 'MethodDeclaration', 'method', match[1], {
-        parameters,
-        ...(extensionReceiver ? { extensionReceiver } : {})
-      }, trimmed.includes('{') || trimmed.includes('=>')));
-    } else if ((match = trimmed.match(/^(?:(?:public|protected|private|internal|static|virtual|override|abstract|sealed|new|unsafe)\s+)*event\s+(.+?)\s+([A-Za-z_]\w*)\s*(?:[;{=]|=>)/))) {
-      declarations.push(nativeDeclaration(input, number, 'EventDeclaration', 'event', match[2], {
-        eventType: match[1].trim(),
-        accessors: csharpAccessors(trimmed)
-      }, trimmed.includes('{')));
-    } else if ((match = trimmed.match(/^(?:(?:public|protected|private|internal|static|virtual|override|abstract|sealed|new|required|readonly|unsafe)\s+)*([A-Za-z_][\w<>\[\].?,\s]*\??)\s+([A-Za-z_]\w*)\s*(?:\{|=>)/))) {
-      declarations.push(nativeDeclaration(input, number, 'PropertyDeclaration', 'property', match[2], {
-        propertyType: match[1].trim(),
-        accessors: csharpAccessors(trimmed)
-      }, trimmed.includes('{') || trimmed.includes('=>')));
-    }
-  }
-  return declarations;
-}
-
 function parseGoReceiver(raw) {
   const value = String(raw ?? '').trim();
   const match = value.match(/^(?:(\w+)\s+)?(.+)$/);
@@ -235,67 +128,8 @@ function goReceiverMethodName(receiver, methodName) {
   return receiver?.type ? `${receiver.type}.${methodName}` : methodName;
 }
 
-function parseSwiftExtensionTail(rawTail) {
-  let tail = String(rawTail ?? '').split('{')[0].trim();
-  const fields = {};
-  const whereMatch = tail.match(/\bwhere\b(.+)$/);
-  if (whereMatch) {
-    fields.constraints = whereMatch[1].trim();
-    tail = tail.slice(0, whereMatch.index).trim();
-  }
-  if (tail.startsWith(':')) {
-    fields.conformances = tail.slice(1).split(',').map((part) => part.trim()).filter(Boolean);
-  }
-  return fields;
-}
-
-function unquoteSwiftIdentifier(identifier) {
-  return String(identifier).replace(/^`|`$/g, '');
-}
-
-function javaSymbolKind(kind) {
-  if (kind === 'interface' || kind === '@interface') return 'interface';
-  if (kind === 'enum' || kind === 'record') return 'type';
-  return 'class';
-}
-
-function swiftSymbolKind(kind) {
-  if (kind === 'protocol') return 'protocol';
-  if (kind === 'extension') return 'implementation';
-  if (kind === 'struct' || kind === 'enum' || kind === 'actor') return 'type';
-  return 'class';
-}
-
-function csharpSymbolKind(kind) {
-  const normalized = String(kind).replace(/\s+/g, ' ');
-  if (normalized === 'interface') return 'interface';
-  if (normalized === 'struct' || normalized === 'enum' || normalized.startsWith('record')) return 'type';
-  return 'class';
-}
-
-function csharpDeclarationKind(kind) {
-  const normalized = String(kind).replace(/\s+/g, ' ');
-  if (normalized === 'record struct') return 'RecordStructDeclaration';
-  if (normalized === 'record class') return 'RecordClassDeclaration';
-  if (normalized === 'record') return 'RecordDeclaration';
-  return `${upperFirst(normalized)}Declaration`;
-}
-
-function csharpExtensionReceiver(parameters) {
-  const match = String(parameters?.[0] ?? '').match(/^this\s+(.+?)\s+([A-Za-z_]\w*)$/);
-  return match ? { type: match[1].trim(), name: match[2] } : undefined;
-}
-
-function csharpAccessors(source) {
-  return uniqueStrings([...String(source ?? '').matchAll(/\b(get|set|init|add|remove)\b/g)].map((match) => match[1]));
-}
-
 export {
   scanCLike,
-  scanCSharp,
   scanGo,
-  scanJava,
-  scanPython,
-  scanRust,
-  scanSwift
+  scanRust
 };

package/dist/native-region-scanner-php.js

@@ -0,0 +1,80 @@
+import { upperFirst } from './native-import-utils.js';
+import { nativeDeclaration, nativeImportDeclaration, sourceLines, splitParameters } from './native-region-scanner-core.js';
+import { braceBlockSpan } from './native-region-scanner-spans.js';
+
+function scanPhp(input) {
+  const declarations = [];
+  const lines = sourceLines(input.sourceText);
+  const blockStack = [];
+  let braceDepth = 0;
+  for (const [index, { line, number }] of lines.entries()) {
+    const trimmed = line.trim().replace(/^<\?php\s*/, '');
+    const lineStartDepth = depthAfterLeadingClosers(trimmed, braceDepth);
+    while (blockStack.length && blockStack[blockStack.length - 1].bodyDepth > lineStartDepth) blockStack.pop();
+
+    let match;
+    if ((match = trimmed.match(/^namespace\s+([A-Za-z_][\w\\]*)\s*;/))) {
+      declarations.push(nativeDeclaration(input, number, 'NamespaceDefinition', 'namespace', match[1], {}, false));
+    } else if ((match = trimmed.match(/^use\s+([A-Za-z_][\w\\]*)(?:\s+as\s+([A-Za-z_]\w*))?\s*;/))) {
+      declarations.push(nativeImportDeclaration(input, number, match[1], 'UseDeclaration', 'namespace'));
+    } else if ((match = trimmed.match(/^(?:(?:abstract|final|readonly)\s+)*(class|interface|trait|enum)\s+([A-Za-z_]\w*)/))) {
+      const hasBody = trimmed.includes('{');
+      declarations.push(nativeDeclaration(input, number, `${upperFirst(match[1])}Declaration`, phpSymbolKind(match[1]), match[2], {}, hasBody, spanOptions(input, lines, index, hasBody)));
+      if (hasBody) blockStack.push({ kind: match[1], name: match[2], bodyDepth: braceDepth + 1 });
+    } else if ((match = trimmed.match(/^((?:(?:public|protected|private|static|final|abstract)\s+)*)function\s+&?\s*([A-Za-z_]\w*)\s*\(([^)]*)\)/))) {
+      const owner = nearestContainer(blockStack);
+      const target = phpFunctionTarget(owner, match[1], match[2]);
+      const hasBody = trimmed.includes('{');
+      declarations.push(nativeDeclaration(input, number, 'FunctionDeclaration', target.owner ? 'method' : 'function', target.name, {
+        parameters: splitParameters(match[3]),
+        ...(target.owner ? { owner: target.owner, methodName: match[2], receiverKind: target.receiverKind } : {}),
+        ...(match[1].trim() ? { modifiers: splitParameters(match[1].trim().replace(/\s+/g, ',')) } : {})
+      }, hasBody, {
+        ...spanOptions(input, lines, index, hasBody),
+        metadata: {
+          ...(target.owner ? { owner: target.owner, methodName: match[2], receiverKind: target.receiverKind } : {}),
+          ...(match[1].trim() ? { modifiers: splitParameters(match[1].trim().replace(/\s+/g, ',')) } : {})
+        }
+      }));
+      if (hasBody) blockStack.push({ kind: 'function', name: target.name, bodyDepth: braceDepth + 1 });
+    }
+    braceDepth = Math.max(0, braceDepth + braceDelta(line));
+  }
+  return declarations;
+}
+
+function spanOptions(input, lines, index, hasBraceBody) {
+  return hasBraceBody ? { span: braceBlockSpan(input, lines, index) } : {};
+}
+
+function depthAfterLeadingClosers(trimmed, depth) {
+  const closers = String(trimmed).match(/^}+/)?.[0].length ?? 0;
+  return Math.max(0, depth - closers);
+}
+
+function nearestContainer(blockStack) {
+  for (let index = blockStack.length - 1; index >= 0; index -= 1) {
+    if (blockStack[index].kind === 'function') return undefined;
+    if (['class', 'interface', 'trait', 'enum'].includes(blockStack[index].kind)) return blockStack[index];
+  }
+  return undefined;
+}
+
+function phpFunctionTarget(owner, modifiers, methodName) {
+  if (!owner) return { name: methodName };
+  const receiverKind = /\bstatic\b/.test(modifiers) ? 'static' : 'instance';
+  return { name: receiverKind === 'static' ? `${owner.name}.static.${methodName}` : `${owner.name}.${methodName}`, owner: owner.name, receiverKind };
+}
+
+function braceDelta(source) {
+  return [...String(source ?? '')].reduce((delta, char) => delta + (char === '{' ? 1 : char === '}' ? -1 : 0), 0);
+}
+
+function phpSymbolKind(kind) {
+  if (kind === 'interface') return 'interface';
+  if (kind === 'trait') return 'trait';
+  if (kind === 'enum') return 'type';
+  return 'class';
+}
+
+export { scanPhp };

package/dist/native-region-scanner-python.js

@@ -0,0 +1,62 @@
+import { nativeDeclaration, nativeImportDeclaration, sourceLines, splitParameters } from './native-region-scanner-core.js';
+import { pythonBlockSpan } from './native-region-scanner-spans.js';
+
+function scanPython(input) {
+  const declarations = [];
+  const lines = sourceLines(input.sourceText);
+  const blockStack = [];
+  let decorators = [];
+  for (const [index, { line, number }] of lines.entries()) {
+    const trimmed = line.trim();
+    if (!trimmed) {
+      decorators = [];
+      continue;
+    }
+    const indent = indentationLength(line);
+    while (blockStack.length && indent <= blockStack[blockStack.length - 1].indent) blockStack.pop();
+
+    let match;
+    if ((match = trimmed.match(/^@([A-Za-z_][\w.]*(?:\([^)]*\))?)/))) {
+      decorators.push(match[1]);
+      continue;
+    }
+    if ((match = trimmed.match(/^(?:async\s+)?def\s+([A-Za-z_]\w*)\s*\(([^)]*)\)\s*:/))) {
+      const owner = blockStack[blockStack.length - 1]?.kind === 'class' ? blockStack[blockStack.length - 1].name : undefined;
+      const name = owner ? `${owner}.${match[1]}` : match[1];
+      declarations.push(nativeDeclaration(input, number, 'FunctionDef', owner ? 'method' : 'function', name, {
+        parameters: splitParameters(match[2]),
+        ...(owner ? { owner, methodName: match[1] } : {}),
+        ...(decorators.length ? { decorators } : {})
+      }, true, {
+        span: pythonBlockSpan(input, lines, index),
+        metadata: {
+          ...(owner ? { owner, methodName: match[1] } : {}),
+          ...(decorators.length ? { decorators } : {})
+        }
+      }));
+      blockStack.push({ kind: 'def', name, indent });
+    } else if ((match = trimmed.match(/^class\s+([A-Za-z_]\w*)/))) {
+      const owner = blockStack[blockStack.length - 1]?.kind === 'class' ? blockStack[blockStack.length - 1].name : undefined;
+      const name = owner ? `${owner}.${match[1]}` : match[1];
+      declarations.push(nativeDeclaration(input, number, 'ClassDef', 'class', name, {
+        ...(owner ? { owner } : {}),
+        ...(decorators.length ? { decorators } : {})
+      }, true, {
+        span: pythonBlockSpan(input, lines, index),
+        metadata: {
+          ...(owner ? { owner } : {}),
+          ...(decorators.length ? { decorators } : {})
+        }
+      }));
+      blockStack.push({ kind: 'class', name, indent });
+    } else if ((match = trimmed.match(/^(?:from\s+([A-Za-z_][\w.]*)\s+import\s+.+|import\s+([A-Za-z_][\w.]*))/))) {
+      declarations.push(nativeImportDeclaration(input, number, match[1] ?? match[2], 'Import', 'module'));
+    }
+    decorators = [];
+  }
+  return declarations;
+}
+
+function indentationLength(line) { return String(line ?? '').match(/^\s*/)?.[0].length ?? 0; }
+
+export { scanPython };

package/dist/native-region-scanner-ruby.js

@@ -0,0 +1,72 @@
+import { nativeDeclaration, nativeImportDeclaration, sourceLines, splitParameters } from './native-region-scanner-core.js';
+import { endKeywordBlockSpan } from './native-region-scanner-spans.js';
+
+function scanRuby(input) {
+  const declarations = [];
+  const lines = sourceLines(input.sourceText);
+  const blockStack = [];
+  for (const [index, { line, number }] of lines.entries()) {
+    const trimmed = line.trim();
+    if (!trimmed) continue;
+    if (/^end\b/.test(trimmed)) {
+      blockStack.pop();
+      continue;
+    }
+
+    let match;
+    if ((match = trimmed.match(/^(?:require|load)\s+['"]([^'"]+)['"]/))) {
+      declarations.push(nativeImportDeclaration(input, number, match[1], 'Require', 'module'));
+    } else if ((match = trimmed.match(/^module\s+([A-Za-z_]\w*(?:::[A-Za-z_]\w*)*)/))) {
+      const name = qualifiedContainerName(blockStack, match[1]);
+      declarations.push(nativeDeclaration(input, number, 'Module', 'module', name, {}, true, endSpanOptions(input, lines, index)));
+      blockStack.push({ kind: 'module', name });
+    } else if ((match = trimmed.match(/^class\s+([A-Za-z_]\w*(?:::[A-Za-z_]\w*)*)/))) {
+      const name = qualifiedContainerName(blockStack, match[1]);
+      declarations.push(nativeDeclaration(input, number, 'Class', 'class', name, {}, true, endSpanOptions(input, lines, index)));
+      blockStack.push({ kind: 'class', name });
+    } else if ((match = trimmed.match(/^def\s+(?:(self|[A-Za-z_]\w*(?:::[A-Za-z_]\w*)*)\.)?([A-Za-z_]\w*[!?=]?)\s*(?:\(([^)]*)\)|([^#=]*))?/))) {
+      const owner = nearestContainer(blockStack);
+      const target = rubyMethodTarget(owner, match[1], match[2]);
+      declarations.push(nativeDeclaration(input, number, 'Def', target.owner ? 'method' : 'function', target.name, {
+        parameters: splitParameters(match[3] ?? match[4]),
+        methodName: match[2],
+        ...(target.owner ? { owner: target.owner, receiverKind: target.receiverKind } : {})
+      }, true, {
+        ...endSpanOptions(input, lines, index),
+        metadata: {
+          methodName: match[2],
+          ...(target.owner ? { owner: target.owner, receiverKind: target.receiverKind } : {})
+        }
+      }));
+      blockStack.push({ kind: 'def', name: target.name });
+    }
+  }
+  return declarations;
+}
+
+function endSpanOptions(input, lines, index) {
+  return { span: endKeywordBlockSpan(input, lines, index) };
+}
+
+function qualifiedContainerName(blockStack, name) {
+  if (name.includes('::')) return name;
+  const owner = nearestContainer(blockStack);
+  return owner?.kind === 'module' ? `${owner.name}::${name}` : name;
+}
+
+function nearestContainer(blockStack) {
+  for (let index = blockStack.length - 1; index >= 0; index -= 1) {
+    if (blockStack[index].kind === 'class' || blockStack[index].kind === 'module') return blockStack[index];
+    if (blockStack[index].kind === 'def') return undefined;
+  }
+  return undefined;
+}
+
+function rubyMethodTarget(owner, receiver, methodName) {
+  if (receiver && receiver !== 'self') return { name: `${receiver}.singleton.${methodName}`, owner: receiver, receiverKind: 'singleton' };
+  if (receiver === 'self' && owner) return { name: `${owner.name}.singleton.${methodName}`, owner: owner.name, receiverKind: 'singleton' };
+  if (owner) return { name: `${owner.name}.instance.${methodName}`, owner: owner.name, receiverKind: 'instance' };
+  return { name: methodName };
+}
+
+export { scanRuby };

package/dist/native-region-scanner-scala.js

@@ -0,0 +1,91 @@
+import { upperFirst } from './native-import-utils.js';
+import { nativeDeclaration, nativeImportDeclaration, sourceLines, splitParameters } from './native-region-scanner-core.js';
+import { braceBlockSpan } from './native-region-scanner-spans.js';
+
+function scanScala(input) {
+  const declarations = [];
+  const lines = sourceLines(input.sourceText);
+  const blockStack = [];
+  let braceDepth = 0;
+  for (const [index, { line, number }] of lines.entries()) {
+    const trimmed = line.trim();
+    const lineStartDepth = depthAfterLeadingClosers(trimmed, braceDepth);
+    while (blockStack.length && blockStack[blockStack.length - 1].bodyDepth > lineStartDepth) blockStack.pop();
+
+    let match;
+    if ((match = trimmed.match(/^package\s+([A-Za-z_]\w*(?:\.[A-Za-z_]\w*)*)/))) {
+      declarations.push(nativeDeclaration(input, number, 'PackageClause', 'package', match[1], {}, false));
+    } else if ((match = trimmed.match(/^import\s+(.+?);?$/))) {
+      declarations.push(nativeImportDeclaration(input, number, match[1].trim(), 'Import', 'package'));
+    } else if ((match = trimmed.match(/^(?:(?:private|protected|final|sealed|abstract|case|implicit|lazy|override|inline|transparent|open)\s+)*(class|trait|object|enum)\s+([A-Za-z_]\w*)/))) {
+      const owner = nearestContainer(blockStack);
+      const name = scalaContainerName(owner, match[1], match[2]);
+      const hasBody = trimmed.includes('{') || trimmed.includes(':');
+      declarations.push(nativeDeclaration(input, number, `${upperFirst(match[1])}Def`, scalaSymbolKind(match[1]), name, owner ? { owner: owner.name } : {}, hasBody, spanOptions(input, lines, index, trimmed.includes('{'))));
+      if (trimmed.includes('{')) blockStack.push({ kind: match[1], name, bodyDepth: braceDepth + 1 });
+    } else if ((match = trimmed.match(/^(?:(?:private|protected|final|implicit|override|inline)\s+)*def\s+([A-Za-z_]\w*)\s*(?:\[[^\]]+\])?\s*\(([^)]*)\)/))) {
+      const owner = nearestContainer(blockStack);
+      const target = scalaMethodTarget(owner, match[1]);
+      const hasBody = trimmed.includes('{') || trimmed.includes('=');
+      declarations.push(nativeDeclaration(input, number, 'DefDef', target.owner ? 'method' : 'function', target.name, {
+        parameters: splitParameters(match[2]),
+        methodName: match[1],
+        ...(target.owner ? { owner: target.owner, receiverKind: target.receiverKind } : {})
+      }, hasBody, {
+        ...spanOptions(input, lines, index, trimmed.includes('{')),
+        metadata: {
+          methodName: match[1],
+          ...(target.owner ? { owner: target.owner, receiverKind: target.receiverKind } : {})
+        }
+      }));
+      if (trimmed.includes('{')) blockStack.push({ kind: 'function', name: target.name, bodyDepth: braceDepth + 1 });
+    } else if ((match = trimmed.match(/^(?:(?:private|protected|final|implicit|opaque)\s+)*type\s+([A-Za-z_]\w*)\b/))) {
+      declarations.push(nativeDeclaration(input, number, 'TypeDef', 'type', match[1], {}, false));
+    } else if ((match = trimmed.match(/^(?:(?:private|protected|final|implicit|lazy|override|inline)\s+)*(?:val|var)\s+([A-Za-z_]\w*)\b/))) {
+      declarations.push(nativeDeclaration(input, number, 'ValDef', 'variable', match[1], {}, false));
+    }
+    braceDepth = Math.max(0, braceDepth + braceDelta(line));
+  }
+  return declarations;
+}
+
+function spanOptions(input, lines, index, hasBraceBody) {
+  return hasBraceBody ? { span: braceBlockSpan(input, lines, index) } : {};
+}
+
+function depthAfterLeadingClosers(trimmed, depth) {
+  const closers = String(trimmed).match(/^}+/)?.[0].length ?? 0;
+  return Math.max(0, depth - closers);
+}
+
+function nearestContainer(blockStack) {
+  for (let index = blockStack.length - 1; index >= 0; index -= 1) {
+    if (blockStack[index].kind === 'function') return undefined;
+    if (['class', 'trait', 'object', 'enum'].includes(blockStack[index].kind)) return blockStack[index];
+  }
+  return undefined;
+}
+
+function scalaContainerName(owner, kind, name) {
+  const base = kind === 'object' ? `${name}.object` : name;
+  return owner ? `${owner.name}.${base}` : base;
+}
+
+function scalaMethodTarget(owner, methodName) {
+  if (!owner) return { name: methodName };
+  const receiverKind = owner.kind === 'object' ? 'object' : 'member';
+  return { name: `${owner.name}.${methodName}`, owner: owner.name, receiverKind };
+}
+
+function braceDelta(source) {
+  return [...String(source ?? '')].reduce((delta, char) => delta + (char === '{' ? 1 : char === '}' ? -1 : 0), 0);
+}
+
+function scalaSymbolKind(kind) {
+  if (kind === 'trait') return 'trait';
+  if (kind === 'object') return 'module';
+  if (kind === 'enum') return 'type';
+  return 'class';
+}
+
+export { scanScala };

package/dist/native-region-scanner-spans.js

@@ -0,0 +1,74 @@
+function pythonBlockSpan(input, lines, index) {
+  const baseIndent = indentationLength(lines[index]?.line);
+  let end = index;
+  for (let cursor = index + 1; cursor < lines.length; cursor += 1) {
+    const line = lines[cursor].line;
+    if (line.trim() && indentationLength(line) <= baseIndent) break;
+    end = cursor;
+  }
+  return lineSpan(input, lines[index], lines[end] ?? lines[index]);
+}
+
+function braceBlockSpan(input, lines, index) {
+  let depth = 0;
+  let seenOpen = false;
+  let end = index;
+  for (let cursor = index; cursor < lines.length; cursor += 1) {
+    for (const char of lines[cursor].line) {
+      if (char === '{') {
+        seenOpen = true;
+        depth += 1;
+      } else if (char === '}') depth -= 1;
+    }
+    end = cursor;
+    if (seenOpen && depth <= 0) break;
+  }
+  return lineSpan(input, lines[index], lines[end] ?? lines[index]);
+}
+
+function endKeywordBlockSpan(input, lines, index) {
+  let depth = 0;
+  let end = index;
+  for (let cursor = index; cursor < lines.length; cursor += 1) {
+    const trimmed = lines[cursor].line.trim();
+    if (endBlockStart(trimmed)) depth += 1;
+    if (/^end\b/.test(trimmed)) depth -= 1;
+    end = cursor;
+    if (depth <= 0 && cursor > index) break;
+  }
+  return lineSpan(input, lines[index], lines[end] ?? lines[index]);
+}
+
+function terminatedBlockSpan(input, lines, index, terminator) {
+  let end = index;
+  for (let cursor = index; cursor < lines.length; cursor += 1) {
+    end = cursor;
+    if (terminator.test(lines[cursor].line.trim())) break;
+  }
+  return lineSpan(input, lines[index], lines[end] ?? lines[index]);
+}
+
+function sqlStatementSpan(input, lines, index) {
+  let end = index;
+  let inDollarQuote = false;
+  for (let cursor = index; cursor < lines.length; cursor += 1) {
+    const line = lines[cursor].line;
+    const markerCount = (line.match(/\$\$/g) ?? []).length;
+    if (markerCount % 2 === 1) inDollarQuote = !inDollarQuote;
+    end = cursor;
+    if (!inDollarQuote && /;\s*$/.test(line.trim())) break;
+  }
+  return lineSpan(input, lines[index], lines[end] ?? lines[index]);
+}
+
+function lineSpan(input, startLine, endLine) {
+  return { sourceId: input.sourceHash, path: input.sourcePath, startLine: startLine.number, endLine: endLine.number, startColumn: 1, endColumn: endLine.line.length + 1 };
+}
+
+function indentationLength(line) { return String(line ?? '').match(/^\s*/)?.[0].length ?? 0; }
+
+function endBlockStart(line) {
+  return /^(?:class|module|def|defp?|defmodule|function|if|unless|case|while|for|begin|try|receive)\b/.test(line);
+}
+
+export { braceBlockSpan, endKeywordBlockSpan, pythonBlockSpan, sqlStatementSpan, terminatedBlockSpan };