septum 0.1.5 → 0.1.7

package/src/core/parser/extractors/php.ts

@@ -1,66 +1,122 @@
-import type { ExtractedDependency, ExtractedSymbol, ParsedFileAST, Visibility } from "../../../types/index.ts";
-import { findBraceBlockEnd, calculateControlFlowNesting } from "../boundary-tracker.ts";
+import type Parser from "web-tree-sitter";
+import type {
+  ExtractedDependency,
+  ExtractedSymbol,
+  ParsedFileAST,
+  SymbolKind,
+  Visibility,
+} from "../../../types/index.ts";
+import {
+  blankCommentsAndStrings,
+  calculateControlFlowNesting,
+  findBraceBlockEnd,
+} from "../boundary-tracker.ts";
+import { calculateAstControlFlowNesting, parseWithTreeSitter } from "../wasm-loader.ts";
 import type { CodeExtractor } from "./base.ts";
 
+const PHP_CONTROL_FLOW_NODES = new Set([
+  "if_statement",
+  "for_statement",
+  "foreach_statement",
+  "while_statement",
+  "do_statement",
+  "switch_statement",
+  "try_statement",
+  "catch_clause",
+  "match_expression",
+  "conditional_expression",
+]);
+
+const PHP_FUNCTION_BOUNDARY_NODES = new Set([
+  "function_definition",
+  "method_declaration",
+  "anonymous_function_creation_expression",
+  "arrow_function",
+]);
+
 export class PHPExtractor implements CodeExtractor {
   public canHandle(filePath: string): boolean {
     return /\.php$/i.test(filePath);
   }
 
-  public async extract(filePath: string, content: string): Promise<ParsedFileAST> {
+  public async extract(_filePath: string, content: string): Promise<ParsedFileAST> {
+    const astResult = await parseWithTreeSitter("php", content, (rootNode) =>
+      this.extractFromAst(rootNode, content)
+    );
+    if (astResult) {
+      return { ...astResult, status: "parsed" };
+    }
+
+    return { ...this.extractFallback(content), status: "degraded" };
+  }
+
+  private isExternalNamespace(cleaned: string): boolean {
+    return (
+      cleaned.startsWith("Illuminate\\") ||
+      cleaned.startsWith("Symfony\\") ||
+      cleaned.startsWith("Psr\\") ||
+      cleaned.startsWith("Composer\\")
+    );
+  }
+
+  private extractFallback(content: string): ParsedFileAST {
     const symbols: ExtractedSymbol[] = [];
     const dependencies: ExtractedDependency[] = [];
 
     const lines = content.split("\n");
+    const cleanLines = blankCommentsAndStrings(content, { hashComments: true }).split("\n");
 
     // 1. Extract PHP use / import statements
     const useRegex = /^\s*use\s+([^;]+);/;
     for (let i = 0; i < lines.length; i++) {
+      if (!cleanLines[i]?.trim()) continue;
       const line = lines[i];
-      const match = useRegex.exec(line);
+      const match = useRegex.exec(cleanLines[i]);
       if (match) {
         const fullTarget = match[1].trim();
-        // Remove 'function ' or 'const ' prefix if any
-        const cleaned = fullTarget.replace(/^(function|const)\s+/, "");
-        const isExternal =
-          cleaned.startsWith("Illuminate\\") ||
-          cleaned.startsWith("Symfony\\") ||
-          cleaned.startsWith("Psr\\") ||
-          cleaned.startsWith("Composer\\");
+        const cleaned = fullTarget.replace(/^(function|const)\s+/, "").replace(/\s+as\s+[A-Za-z0-9_]+$/i, "").trim();
+        const aliasMatch = /\bas\s+([A-Za-z0-9_]+)$/i.exec(fullTarget);
+        const shortName = cleaned.split("\\").pop() || cleaned;
+        const importedSymbols = Array.from(new Set([shortName, ...(aliasMatch ? [aliasMatch[1]] : [])])).filter(Boolean);
 
         dependencies.push({
           target: cleaned,
           statement: line.trim(),
           line_number: i + 1,
-          is_external: isExternal,
+          is_external: this.isExternalNamespace(cleaned),
+          ...(importedSymbols.length > 0 ? { imported_symbols: importedSymbols } : {}),
         });
       }
     }
 
+
     // 2. Extract PHP Class / Interface / Trait / Enum
     let currentContainer: string | null = null;
     let currentContainerEndLine = 0;
 
-    for (let i = 0; i < lines.length; i++) {
-      const line = lines[i];
-      const trimmed = line.trim();
+    for (let i = 0; i < cleanLines.length; i++) {
+      const trimmed = cleanLines[i].trim();
+      const rawTrimmed = lines[i].trim();
       const lineNum = i + 1;
 
+      if (!trimmed) continue;
+
       if (currentContainer && lineNum > currentContainerEndLine) {
         currentContainer = null;
         currentContainerEndLine = 0;
       }
 
-      // Class / Interface / Trait / Enum header
-      const classMatch = /^(?:abstract\s+|final\s+|readonly\s+)*(class|interface|trait|enum)\s+([A-Za-z0-9_]+)/.exec(trimmed);
+      const classMatch = /^(?:abstract\s+|final\s+|readonly\s+)*\b(class|interface|trait|enum)\s+([A-Za-z0-9_]+)\b/.exec(
+        trimmed
+      );
       if (classMatch) {
         currentContainer = classMatch[2];
-        const endLine = findBraceBlockEnd(lines, i);
+        const endLine = findBraceBlockEnd(cleanLines, i);
         currentContainerEndLine = endLine;
         symbols.push({
           name: classMatch[2],
-          kind: classMatch[1] as any,
-          signature: trimmed.replace(/\{$/, "").trim(),
+          kind: classMatch[1] as SymbolKind,
+          signature: rawTrimmed.replace(/\{$/, "").trim(),
           visibility: "public",
           line_start: lineNum,
           line_end: endLine,
@@ -69,10 +125,8 @@ export class PHPExtractor implements CodeExtractor {
         continue;
       }
 
-      // Methods or Properties in class/interface/trait/enum
       if (currentContainer) {
-        // PHP Constant: const STATUS_ACTIVE = 'active';
-        const constMatch = /^(?:(public|protected|private)\s+)?const\s+([A-Za-z0-9_]+)\s*=/.exec(trimmed);
+        const constMatch = /^(?:(public|protected|private)\s+)?\bconst\s+([A-Za-z0-9_]+)\s*=/.exec(trimmed);
         if (constMatch) {
           const vis = (constMatch[1] as Visibility) || "public";
           const constName = constMatch[2];
@@ -88,15 +142,17 @@ export class PHPExtractor implements CodeExtractor {
           continue;
         }
 
-        // PHP Property: public ?string $foo = null; or private readonly Bar $bar;
-        const propMatch = /^(?:(public|protected|private)\s+)?(?:readonly\s+)?(?:static\s+)?(?:\??[A-Za-z0-9_\\\\]+\s+)?\$([A-Za-z0-9_]+)(?:\s*=[^;]+)?;\s*$/.exec(trimmed);
+        const propMatch =
+          /^(?:(public|protected|private)\s+)?(?:readonly\s+)?(?:static\s+)?(?:\??[A-Za-z0-9_\\]+\s+)?\$([A-Za-z0-9_]+)(?:\s*=[^;]+)?;\s*$/.exec(
+            trimmed
+          );
         if (propMatch && !trimmed.includes("function ")) {
           const vis = (propMatch[1] as Visibility) || "public";
           const propName = `$${propMatch[2]}`;
           symbols.push({
             name: `${currentContainer}::${propName}`,
             kind: "property",
-            signature: `${vis} ${trimmed.replace(/;\s*$/, "")}`,
+            signature: `${vis} ${rawTrimmed.replace(/;\s*$/, "")}`,
             visibility: vis,
             line_start: lineNum,
             line_end: lineNum,
@@ -105,13 +161,16 @@ export class PHPExtractor implements CodeExtractor {
           continue;
         }
 
-        const methodMatch = /^(?:(public|protected|private)\s+)?(?:static\s+)?(?:final\s+)?function\s+([A-Za-z0-9_]+)\s*\(([^)]*)\)(?:\s*:\s*([^;{]+))?/.exec(trimmed);
+        const methodMatch =
+          /^(?:(public|protected|private)\s+)?(?:static\s+)?(?:final\s+)?\bfunction\s+([A-Za-z0-9_]+)\s*\(([^)]*)\)(?:\s*:\s*([^;{]+))?/.exec(
+            trimmed
+          );
         if (methodMatch) {
           const vis = (methodMatch[1] as Visibility) || "public";
           const name = methodMatch[2];
           const params = methodMatch[3].trim();
           const retType = methodMatch[4] ? `: ${methodMatch[4].trim()}` : "";
-          const endLine = findBraceBlockEnd(lines, i);
+          const endLine = findBraceBlockEnd(cleanLines, i);
           const nestingDepth = calculateControlFlowNesting(lines, i, endLine - 1);
           symbols.push({
             name: `${currentContainer}::${name}`,
@@ -127,13 +186,12 @@ export class PHPExtractor implements CodeExtractor {
         }
       }
 
-      // Standalone Global Function in PHP
-      const funcMatch = /^(?:final\s+)?function\s+([A-Za-z0-9_]+)\s*\(([^)]*)\)(?:\s*:\s*([^;{]+))?/.exec(trimmed);
+      const funcMatch = /^(?:final\s+)?\bfunction\s+([A-Za-z0-9_]+)\s*\(([^)]*)\)(?:\s*:\s*([^;{]+))?/.exec(trimmed);
       if (funcMatch && !currentContainer) {
         const name = funcMatch[1];
         const params = funcMatch[2].trim();
         const retType = funcMatch[3] ? `: ${funcMatch[3].trim()}` : "";
-        const endLine = findBraceBlockEnd(lines, i);
+        const endLine = findBraceBlockEnd(cleanLines, i);
         const nestingDepth = calculateControlFlowNesting(lines, i, endLine - 1);
         symbols.push({
           name,
@@ -150,4 +208,208 @@ export class PHPExtractor implements CodeExtractor {
 
     return { symbols, dependencies };
   }
+
+  private extractFromAst(rootNode: Parser.SyntaxNode, content: string): ParsedFileAST {
+    const symbols: ExtractedSymbol[] = [];
+    const dependencies: ExtractedDependency[] = [];
+    const lines = content.split("\n");
+
+    const walk = (node: Parser.SyntaxNode): void => {
+      if (node.type === "namespace_use_declaration") {
+        const rawStmt = (lines[node.startPosition.row] ?? node.text).trim();
+        const bodyMatch = /^\s*use\s+([^;]+);/i.exec(node.text.replace(/\r?\n/g, " "));
+        if (bodyMatch) {
+          const fullTarget = bodyMatch[1].trim();
+          const cleaned = fullTarget.replace(/^(function|const)\s+/i, "").replace(/\s+as\s+[A-Za-z0-9_]+$/i, "").trim();
+          const aliasMatch = /\bas\s+([A-Za-z0-9_]+)$/i.exec(fullTarget);
+          const shortName = cleaned.split("\\").pop() || cleaned;
+          const importedSymbols = Array.from(
+            new Set([shortName, ...(aliasMatch ? [aliasMatch[1]] : [])])
+          ).filter(Boolean);
+
+          dependencies.push({
+            target: cleaned,
+            statement: rawStmt,
+            line_number: node.startPosition.row + 1,
+            is_external: this.isExternalNamespace(cleaned),
+            ...(importedSymbols.length > 0 ? { imported_symbols: importedSymbols } : {}),
+          });
+        }
+        return;
+      }
+
+      if (
+        node.type === "class_declaration" ||
+        node.type === "interface_declaration" ||
+        node.type === "trait_declaration" ||
+        node.type === "enum_declaration"
+      ) {
+        const nameNode = node.childForFieldName("name") ?? node.namedChildren.find((c) => c.type === "name");
+        if (!nameNode) return;
+
+        const containerName = nameNode.text;
+        const kindMap: Record<string, SymbolKind> = {
+          class_declaration: "class",
+          interface_declaration: "interface",
+          trait_declaration: "trait",
+          enum_declaration: "enum",
+        };
+        const lineStart = node.startPosition.row + 1;
+        const lineEnd = Math.max(lineStart, node.endPosition.row + 1);
+        const headerLine = (lines[node.startPosition.row] ?? "").trim().replace(/\{$/, "").trim();
+
+        symbols.push({
+          name: containerName,
+          kind: kindMap[node.type] ?? "class",
+          signature: headerLine || `${kindMap[node.type]} ${containerName}`,
+          visibility: "public",
+          line_start: lineStart,
+          line_end: lineEnd,
+          line_count: Math.max(1, lineEnd - lineStart + 1),
+        });
+
+        const bodyNode =
+          node.childForFieldName("body") ?? node.namedChildren.find((c) => c.type === "declaration_list" || c.type === "enum_declaration_list");
+        if (bodyNode) {
+          this.extractPhpContainerMembers(containerName, bodyNode, lines, symbols);
+        }
+        return;
+      }
+
+      if (node.type === "function_definition") {
+        const nameNode = node.childForFieldName("name") ?? node.namedChildren.find((c) => c.type === "name");
+        if (!nameNode) return;
+
+        const fnName = nameNode.text;
+        const paramsNode = node.childForFieldName("parameters") ?? node.namedChildren.find((c) => c.type === "formal_parameters");
+        const retNode = node.childForFieldName("return_type");
+        const paramsText = paramsNode ? paramsNode.text : "()";
+        const retText = retNode ? `: ${retNode.text.replace(/^:\s*/, "")}` : "";
+        const lineStart = node.startPosition.row + 1;
+        const lineEnd = Math.max(lineStart, node.endPosition.row + 1);
+        const nestingDepth = calculateAstControlFlowNesting(
+          node,
+          PHP_CONTROL_FLOW_NODES,
+          PHP_FUNCTION_BOUNDARY_NODES
+        );
+
+        symbols.push({
+          name: fnName,
+          kind: "function",
+          signature: `function ${fnName}${paramsText}${retText}`,
+          visibility: "public",
+          line_start: lineStart,
+          line_end: lineEnd,
+          line_count: Math.max(1, lineEnd - lineStart + 1),
+          nesting_depth: nestingDepth,
+        });
+        return;
+      }
+
+      for (const child of node.namedChildren) {
+        walk(child);
+      }
+    };
+
+    walk(rootNode);
+    return { symbols, dependencies };
+  }
+
+  private extractPhpContainerMembers(
+    containerName: string,
+    bodyNode: Parser.SyntaxNode,
+    lines: string[],
+    symbols: ExtractedSymbol[]
+  ): void {
+    for (const member of bodyNode.namedChildren) {
+      const mStart = member.startPosition.row + 1;
+      const mEnd = Math.max(mStart, member.endPosition.row + 1);
+      const mCount = Math.max(1, mEnd - mStart + 1);
+      const vis = this.extractPhpVisibility(member);
+
+      if (member.type === "method_declaration") {
+        const nameNode = member.childForFieldName("name") ?? member.namedChildren.find((c) => c.type === "name");
+        if (!nameNode) continue;
+
+        const methodName = nameNode.text;
+        const paramsNode =
+          member.childForFieldName("parameters") ??
+          member.namedChildren.find((c) => c.type === "formal_parameters");
+        const retNode = member.childForFieldName("return_type");
+        const paramsText = paramsNode ? paramsNode.text : "()";
+        const retText = retNode ? `: ${retNode.text.replace(/^:\s*/, "")}` : "";
+        const nestingDepth = calculateAstControlFlowNesting(
+          member,
+          PHP_CONTROL_FLOW_NODES,
+          PHP_FUNCTION_BOUNDARY_NODES
+        );
+
+        symbols.push({
+          name: `${containerName}::${methodName}`,
+          kind: "method",
+          signature: `${vis} function ${methodName}${paramsText}${retText}`,
+          visibility: vis,
+          line_start: mStart,
+          line_end: mEnd,
+          line_count: mCount,
+          nesting_depth: nestingDepth,
+        });
+        continue;
+      }
+
+      if (member.type === "property_declaration") {
+        const propElem = member.namedChildren.find((c) => c.type === "property_element");
+        const varNode =
+          propElem?.namedChildren.find((c) => c.type === "variable_name") ??
+          member.namedChildren.find((c) => c.type === "variable_name");
+        if (!varNode) continue;
+
+        const propName = varNode.text.startsWith("$") ? varNode.text : `$${varNode.text}`;
+        const rawLine = (lines[member.startPosition.row] ?? member.text).trim().replace(/;\s*$/, "");
+
+        symbols.push({
+          name: `${containerName}::${propName}`,
+          kind: "property",
+          signature: rawLine.startsWith(vis) ? rawLine : `${vis} ${rawLine}`,
+          visibility: vis,
+          line_start: mStart,
+          line_end: mEnd,
+          line_count: mCount,
+        });
+        continue;
+      }
+
+      if (member.type === "const_declaration") {
+        const constElem = member.namedChildren.find((c) => c.type === "const_element");
+        const nameNode =
+          constElem?.namedChildren.find((c) => c.type === "name") ??
+          member.namedChildren.find((c) => c.type === "name");
+        if (!nameNode) continue;
+
+        const constName = nameNode.text;
+        symbols.push({
+          name: `${containerName}::${constName}`,
+          kind: "property",
+          signature: `${vis} const ${constName}`,
+          visibility: vis,
+          line_start: mStart,
+          line_end: mEnd,
+          line_count: mCount,
+        });
+      }
+    }
+  }
+
+  private extractPhpVisibility(node: Parser.SyntaxNode): Visibility {
+    for (const child of node.children) {
+      if (child.type === "visibility_modifier") {
+        const t = child.text.trim().toLowerCase();
+        if (t === "private" || t === "protected" || t === "public") {
+          return t;
+        }
+      }
+    }
+    return "public";
+  }
 }
+