septum 0.1.3 → 0.1.5

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

@@ -1,226 +1,228 @@
-import type { ExtractedDependency, ExtractedSymbol, ParsedFileAST, Visibility } from "../../../types/index.ts";
-import { findPythonBlockEnd } from "../boundary-tracker.ts";
-import type { CodeExtractor } from "./base.ts";
-
-const PYTHON_STDLIB_AND_EXTERNALS = new Set([
-  "os",
-  "sys",
-  "json",
-  "math",
-  "typing",
-  "re",
-  "datetime",
-  "time",
-  "collections",
-  "itertools",
-  "functools",
-  "pathlib",
-  "io",
-  "logging",
-  "subprocess",
-  "threading",
-  "multiprocessing",
-  "asyncio",
-  "abc",
-  "copy",
-  "enum",
-  "dataclasses",
-  "uuid",
-  "hashlib",
-  "random",
-  "socket",
-  "unittest",
-  // Common 3rd party frameworks
-  "pydantic",
-  "fastapi",
-  "django",
-  "flask",
-  "sqlalchemy",
-  "celery",
-  "numpy",
-  "pandas",
-  "requests",
-  "httpx",
-  "pytest",
-  "starlette",
-  "alembic",
-  "redis",
-]);
-
-export class PythonExtractor implements CodeExtractor {
-  public canHandle(filePath: string): boolean {
-    return /\.py$/i.test(filePath);
-  }
-
-  public async extract(filePath: string, content: string): Promise<ParsedFileAST> {
-    const symbols: ExtractedSymbol[] = [];
-    const dependencies: ExtractedDependency[] = [];
-
-    const lines = content.split("\n");
-
-    // 1. Extract Python Imports
-    for (let i = 0; i < lines.length; i++) {
-      const line = lines[i];
-      const trimmed = line.trim();
-      const lineNum = i + 1;
-
-      // Skip comments
-      if (trimmed.startsWith("#")) continue;
-
-      // Pattern A: from <module> import <symbols>
-      const fromMatch = /^from\s+([.\w]+)\s+import\s+([*\w\s,()]+)/.exec(trimmed);
-      if (fromMatch) {
-        const modulePath = fromMatch[1];
-        const isRelative = modulePath.startsWith(".");
-        const topPackage = isRelative ? "" : modulePath.split(".")[0];
-        const isExternal = !isRelative && PYTHON_STDLIB_AND_EXTERNALS.has(topPackage);
-
-        dependencies.push({
-          target: modulePath,
-          statement: trimmed,
-          line_number: lineNum,
-          is_external: isExternal,
-        });
-        continue;
-      }
-
-      // Pattern B: import <module> [as <alias>]
-      const importMatch = /^import\s+([.\w]+)(?:\s+as\s+\w+)?/.exec(trimmed);
-      if (importMatch) {
-        const modulePath = importMatch[1];
-        const topPackage = modulePath.split(".")[0];
-        const isExternal = PYTHON_STDLIB_AND_EXTERNALS.has(topPackage);
-
-        dependencies.push({
-          target: modulePath,
-          statement: trimmed,
-          line_number: lineNum,
-          is_external: isExternal,
-        });
-      }
-    }
-
-    // 2. Extract Python Classes and Functions
-    let currentClass: string | null = null;
-    let currentClassIndent = -1;
-
-    for (let i = 0; i < lines.length; i++) {
-      const line = lines[i];
-      const trimmed = line.trim();
-      const lineNum = i + 1;
-
-      if (!trimmed || trimmed.startsWith("#")) continue;
-
-      const indentMatch = /^(\s*)/.exec(line);
-      const indent = indentMatch ? indentMatch[1].length : 0;
-
-      // Check if we exited current class scope
-      if (currentClass !== null && indent <= currentClassIndent) {
-        currentClass = null;
-        currentClassIndent = -1;
-      }
-
-      // Class Definition
-      const classMatch = /^class\s+([A-Za-z0-9_]+)(?:\s*\([^)]*\))?\s*:/.exec(trimmed);
-      if (classMatch) {
-        const className = classMatch[1];
-        currentClass = className;
-        currentClassIndent = indent;
-
-        const visibility: Visibility = className.startsWith("__")
-          ? "private"
-          : className.startsWith("_")
-          ? "protected"
-          : "public";
-
-        const endLine = findPythonBlockEnd(lines, i);
-        symbols.push({
-          name: className,
-          kind: "class",
-          signature: trimmed.replace(/:$/, "").trim(),
-          visibility,
-          line_start: lineNum,
-          line_end: endLine,
-          line_count: Math.max(1, endLine - lineNum + 1),
-        });
-        continue;
-      }
-
-      // Class Attributes or Pydantic fields inside class
-      if (currentClass !== null && indent > currentClassIndent) {
-        const attrMatch = /^([A-Za-z0-9_]+)\s*(?::\s*([^=]+))?(?:\s*=\s*(.+))?$/.exec(trimmed);
-        if (
-          attrMatch &&
-          !trimmed.startsWith("def ") &&
-          !trimmed.startsWith("async def ") &&
-          !["return", "pass", "raise", "assert", "import", "from", "yield"].includes(attrMatch[1])
-        ) {
-          const attrName = attrMatch[1];
-          const visibility: Visibility = attrName.startsWith("__")
-            ? "private"
-            : attrName.startsWith("_")
-            ? "protected"
-            : "public";
-
-          symbols.push({
-            name: `${currentClass}::${attrName}`,
-            kind: "property",
-            signature: trimmed,
-            visibility,
-            line_start: lineNum,
-            line_end: lineNum,
-            line_count: 1,
-          });
-          continue;
-        }
-      }
-
-      // Module-level UPPERCASE constants: MAX_RETRIES = 5
-      if (currentClass === null && indent === 0) {
-        const constMatch = /^([A-Z][A-Z0-9_]*)\s*(?::\s*([^=]+))?\s*=\s*(.+)$/.exec(trimmed);
-        if (constMatch) {
-          const constName = constMatch[1];
-          symbols.push({
-            name: constName,
-            kind: "property",
-            signature: trimmed,
-            visibility: "public",
-            line_start: lineNum,
-            line_end: lineNum,
-            line_count: 1,
-          });
-          continue;
-        }
-      }
-
-      // Function or Method Definition
-      const funcMatch = /^(?:async\s+)?def\s+([A-Za-z0-9_]+)\s*\(([^)]*)\)(?:\s*->\s*[^:]+)?\s*:/.exec(trimmed);
-      if (funcMatch) {
-        const funcName = funcMatch[1];
-        const isMethod = currentClass !== null && indent > currentClassIndent;
-
-        const visibility: Visibility = funcName.startsWith("__")
-          ? "private"
-          : funcName.startsWith("_")
-          ? "protected"
-          : "public";
-
-        const endLine = findPythonBlockEnd(lines, i);
-        symbols.push({
-          name: isMethod ? `${currentClass}::${funcName}` : funcName,
-          kind: isMethod ? "method" : "function",
-          signature: trimmed.replace(/:$/, "").trim(),
-          visibility,
-          line_start: lineNum,
-          line_end: endLine,
-          line_count: Math.max(1, endLine - lineNum + 1),
-        });
-      }
-    }
-
-    return {
-      symbols,
-      dependencies,
-    };
-  }
-}
+import type { ExtractedDependency, ExtractedSymbol, ParsedFileAST, Visibility } from "../../../types/index.ts";
+import { findPythonBlockEnd, calculatePythonNesting } from "../boundary-tracker.ts";
+import type { CodeExtractor } from "./base.ts";
+
+const PYTHON_STDLIB_AND_EXTERNALS = new Set([
+  "os",
+  "sys",
+  "json",
+  "math",
+  "typing",
+  "re",
+  "datetime",
+  "time",
+  "collections",
+  "itertools",
+  "functools",
+  "pathlib",
+  "io",
+  "logging",
+  "subprocess",
+  "threading",
+  "multiprocessing",
+  "asyncio",
+  "abc",
+  "copy",
+  "enum",
+  "dataclasses",
+  "uuid",
+  "hashlib",
+  "random",
+  "socket",
+  "unittest",
+  // Common 3rd party frameworks
+  "pydantic",
+  "fastapi",
+  "django",
+  "flask",
+  "sqlalchemy",
+  "celery",
+  "numpy",
+  "pandas",
+  "requests",
+  "httpx",
+  "pytest",
+  "starlette",
+  "alembic",
+  "redis",
+]);
+
+export class PythonExtractor implements CodeExtractor {
+  public canHandle(filePath: string): boolean {
+    return /\.py$/i.test(filePath);
+  }
+
+  public async extract(filePath: string, content: string): Promise<ParsedFileAST> {
+    const symbols: ExtractedSymbol[] = [];
+    const dependencies: ExtractedDependency[] = [];
+
+    const lines = content.split("\n");
+
+    // 1. Extract Python Imports
+    for (let i = 0; i < lines.length; i++) {
+      const line = lines[i];
+      const trimmed = line.trim();
+      const lineNum = i + 1;
+
+      // Skip comments
+      if (trimmed.startsWith("#")) continue;
+
+      // Pattern A: from <module> import <symbols>
+      const fromMatch = /^from\s+([.\w]+)\s+import\s+([*\w\s,()]+)/.exec(trimmed);
+      if (fromMatch) {
+        const modulePath = fromMatch[1];
+        const isRelative = modulePath.startsWith(".");
+        const topPackage = isRelative ? "" : modulePath.split(".")[0];
+        const isExternal = !isRelative && PYTHON_STDLIB_AND_EXTERNALS.has(topPackage);
+
+        dependencies.push({
+          target: modulePath,
+          statement: trimmed,
+          line_number: lineNum,
+          is_external: isExternal,
+        });
+        continue;
+      }
+
+      // Pattern B: import <module> [as <alias>]
+      const importMatch = /^import\s+([.\w]+)(?:\s+as\s+\w+)?/.exec(trimmed);
+      if (importMatch) {
+        const modulePath = importMatch[1];
+        const topPackage = modulePath.split(".")[0];
+        const isExternal = PYTHON_STDLIB_AND_EXTERNALS.has(topPackage);
+
+        dependencies.push({
+          target: modulePath,
+          statement: trimmed,
+          line_number: lineNum,
+          is_external: isExternal,
+        });
+      }
+    }
+
+    // 2. Extract Python Classes and Functions
+    let currentClass: string | null = null;
+    let currentClassIndent = -1;
+
+    for (let i = 0; i < lines.length; i++) {
+      const line = lines[i];
+      const trimmed = line.trim();
+      const lineNum = i + 1;
+
+      if (!trimmed || trimmed.startsWith("#")) continue;
+
+      const indentMatch = /^(\s*)/.exec(line);
+      const indent = indentMatch ? indentMatch[1].length : 0;
+
+      // Check if we exited current class scope
+      if (currentClass !== null && indent <= currentClassIndent) {
+        currentClass = null;
+        currentClassIndent = -1;
+      }
+
+      // Class Definition
+      const classMatch = /^class\s+([A-Za-z0-9_]+)(?:\s*\([^)]*\))?\s*:/.exec(trimmed);
+      if (classMatch) {
+        const className = classMatch[1];
+        currentClass = className;
+        currentClassIndent = indent;
+
+        const visibility: Visibility = className.startsWith("__")
+          ? "private"
+          : className.startsWith("_")
+          ? "protected"
+          : "public";
+
+        const endLine = findPythonBlockEnd(lines, i);
+        symbols.push({
+          name: className,
+          kind: "class",
+          signature: trimmed.replace(/:$/, "").trim(),
+          visibility,
+          line_start: lineNum,
+          line_end: endLine,
+          line_count: Math.max(1, endLine - lineNum + 1),
+        });
+        continue;
+      }
+
+      // Class Attributes or Pydantic fields inside class
+      if (currentClass !== null && indent > currentClassIndent) {
+        const attrMatch = /^([A-Za-z0-9_]+)\s*(?::\s*([^=]+))?(?:\s*=\s*(.+))?$/.exec(trimmed);
+        if (
+          attrMatch &&
+          !trimmed.startsWith("def ") &&
+          !trimmed.startsWith("async def ") &&
+          !["return", "pass", "raise", "assert", "import", "from", "yield"].includes(attrMatch[1])
+        ) {
+          const attrName = attrMatch[1];
+          const visibility: Visibility = attrName.startsWith("__")
+            ? "private"
+            : attrName.startsWith("_")
+            ? "protected"
+            : "public";
+
+          symbols.push({
+            name: `${currentClass}::${attrName}`,
+            kind: "property",
+            signature: trimmed,
+            visibility,
+            line_start: lineNum,
+            line_end: lineNum,
+            line_count: 1,
+          });
+          continue;
+        }
+      }
+
+      // Module-level UPPERCASE constants: MAX_RETRIES = 5
+      if (currentClass === null && indent === 0) {
+        const constMatch = /^([A-Z][A-Z0-9_]*)\s*(?::\s*([^=]+))?\s*=\s*(.+)$/.exec(trimmed);
+        if (constMatch) {
+          const constName = constMatch[1];
+          symbols.push({
+            name: constName,
+            kind: "property",
+            signature: trimmed,
+            visibility: "public",
+            line_start: lineNum,
+            line_end: lineNum,
+            line_count: 1,
+          });
+          continue;
+        }
+      }
+
+      // Function or Method Definition
+      const funcMatch = /^(?:async\s+)?def\s+([A-Za-z0-9_]+)\s*\(([^)]*)\)(?:\s*->\s*[^:]+)?\s*:/.exec(trimmed);
+      if (funcMatch) {
+        const funcName = funcMatch[1];
+        const isMethod = currentClass !== null && indent > currentClassIndent;
+
+        const visibility: Visibility = funcName.startsWith("__")
+          ? "private"
+          : funcName.startsWith("_")
+          ? "protected"
+          : "public";
+
+        const endLine = findPythonBlockEnd(lines, i);
+        const nestingDepth = calculatePythonNesting(lines, i, endLine - 1);
+        symbols.push({
+          name: isMethod ? `${currentClass}::${funcName}` : funcName,
+          kind: isMethod ? "method" : "function",
+          signature: trimmed.replace(/:$/, "").trim(),
+          visibility,
+          line_start: lineNum,
+          line_end: endLine,
+          line_count: Math.max(1, endLine - lineNum + 1),
+          nesting_depth: nestingDepth,
+        });
+      }
+    }
+
+    return {
+      symbols,
+      dependencies,
+    };
+  }
+}

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

@@ -1,70 +1,70 @@
-import type {
-  SemanticSliceExtractor,
-  VerticalSliceCandidate,
-} from "./semantic-extractor.interface.ts";
-import { LaravelSemanticExtractor } from "./laravel-semantic.ts";
-import { McpCliSemanticExtractor } from "./mcp-cli-semantic.ts";
-import { NestJsSemanticExtractor } from "./nestjs-semantic.ts";
-
-export class SemanticSliceExtractorRegistry {
-  private static extractors: SemanticSliceExtractor[] = [
-    new LaravelSemanticExtractor(),
-    new McpCliSemanticExtractor(),
-    new NestJsSemanticExtractor(),
-  ];
-
-  /**
-   * Registers a new semantic slice extractor strategy.
-   * Prepends to the list to give higher priority to custom extractors.
-   */
-  public static register(extractor: SemanticSliceExtractor): void {
-    this.extractors.unshift(extractor);
-  }
-
-  /**
-   * Returns all extractors capable of handling the specified project topology.
-   */
-  public static getExtractors(projectRoot: string, framework?: string): SemanticSliceExtractor[] {
-    return this.extractors.filter((extractor) => extractor.canHandle(projectRoot, framework));
-  }
-
-  /**
-   * Executes all matched extractors and combines their vertical slices.
-   */
-  public static async extractAllSlices(
-    projectRoot: string,
-    framework?: string,
-    domainId?: number
-  ): Promise<VerticalSliceCandidate[]> {
-    const matched = this.getExtractors(projectRoot, framework);
-    const combinedSlices: VerticalSliceCandidate[] = [];
-
-    for (const extractor of matched) {
-      const slices = await extractor.extractSlices(projectRoot, domainId);
-      combinedSlices.push(...slices);
-    }
-
-    return combinedSlices;
-  }
-
-  /**
-   * Synchronous execution of all matched semantic extractors.
-   */
-  public static extractAllSlicesSync(
-    projectRoot: string,
-    framework?: string,
-    domainId?: number
-  ): VerticalSliceCandidate[] {
-    const matched = this.getExtractors(projectRoot, framework);
-    const combinedSlices: VerticalSliceCandidate[] = [];
-
-    for (const extractor of matched) {
-      const res = extractor.extractSlices(projectRoot, domainId);
-      if (Array.isArray(res)) {
-        combinedSlices.push(...res);
-      }
-    }
-
-    return combinedSlices;
-  }
-}
+import type {
+  SemanticSliceExtractor,
+  VerticalSliceCandidate,
+} from "./semantic-extractor.interface.ts";
+import { LaravelSemanticExtractor } from "./laravel-semantic.ts";
+import { McpCliSemanticExtractor } from "./mcp-cli-semantic.ts";
+import { NestJsSemanticExtractor } from "./nestjs-semantic.ts";
+
+export class SemanticSliceExtractorRegistry {
+  private static extractors: SemanticSliceExtractor[] = [
+    new LaravelSemanticExtractor(),
+    new McpCliSemanticExtractor(),
+    new NestJsSemanticExtractor(),
+  ];
+
+  /**
+   * Registers a new semantic slice extractor strategy.
+   * Prepends to the list to give higher priority to custom extractors.
+   */
+  public static register(extractor: SemanticSliceExtractor): void {
+    this.extractors.unshift(extractor);
+  }
+
+  /**
+   * Returns all extractors capable of handling the specified project topology.
+   */
+  public static getExtractors(projectRoot: string, framework?: string): SemanticSliceExtractor[] {
+    return this.extractors.filter((extractor) => extractor.canHandle(projectRoot, framework));
+  }
+
+  /**
+   * Executes all matched extractors and combines their vertical slices.
+   */
+  public static async extractAllSlices(
+    projectRoot: string,
+    framework?: string,
+    domainId?: number
+  ): Promise<VerticalSliceCandidate[]> {
+    const matched = this.getExtractors(projectRoot, framework);
+    const combinedSlices: VerticalSliceCandidate[] = [];
+
+    for (const extractor of matched) {
+      const slices = await extractor.extractSlices(projectRoot, domainId);
+      combinedSlices.push(...slices);
+    }
+
+    return combinedSlices;
+  }
+
+  /**
+   * Synchronous execution of all matched semantic extractors.
+   */
+  public static extractAllSlicesSync(
+    projectRoot: string,
+    framework?: string,
+    domainId?: number
+  ): VerticalSliceCandidate[] {
+    const matched = this.getExtractors(projectRoot, framework);
+    const combinedSlices: VerticalSliceCandidate[] = [];
+
+    for (const extractor of matched) {
+      const res = extractor.extractSlices(projectRoot, domainId);
+      if (Array.isArray(res)) {
+        combinedSlices.push(...res);
+      }
+    }
+
+    return combinedSlices;
+  }
+}

package/src/core/parser/extractors/semantic-extractor.interface.ts

@@ -1,21 +1,21 @@
-import type { VerticalSliceRecord } from "../../../types/index.ts";
-
-export type VerticalSliceCandidate = Omit<VerticalSliceRecord, "id" | "created_at">;
-
-export interface SemanticSliceExtractor {
-  readonly id: string;
-  readonly name: string;
-
-  /**
-   * Determines if this extractor can handle the project topology.
-   */
-  canHandle(projectRoot: string, framework?: string): boolean;
-
-  /**
-   * Extracts vertical slices from the codebase.
-   */
-  extractSlices(
-    projectRoot: string,
-    domainId?: number
-  ): Promise<VerticalSliceCandidate[]> | VerticalSliceCandidate[];
-}
+import type { VerticalSliceRecord } from "../../../types/index.ts";
+
+export type VerticalSliceCandidate = Omit<VerticalSliceRecord, "id" | "created_at">;
+
+export interface SemanticSliceExtractor {
+  readonly id: string;
+  readonly name: string;
+
+  /**
+   * Determines if this extractor can handle the project topology.
+   */
+  canHandle(projectRoot: string, framework?: string): boolean;
+
+  /**
+   * Extracts vertical slices from the codebase.
+   */
+  extractSlices(
+    projectRoot: string,
+    domainId?: number
+  ): Promise<VerticalSliceCandidate[]> | VerticalSliceCandidate[];
+}