indexer-cli 0.2.1

package/dist/languages/python.js

@@ -0,0 +1,292 @@
+"use strict";
+var __importDefault = (this && this.__importDefault) || function (mod) {
+    return (mod && mod.__esModule) ? mod : { "default": mod };
+};
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.plugin = exports.PythonPlugin = void 0;
+const tree_sitter_1 = __importDefault(require("tree-sitter"));
+const tree_sitter_python_1 = __importDefault(require("tree-sitter-python"));
+class PythonPlugin {
+    id = "python";
+    displayName = "Python";
+    fileExtensions = [".py", ".pyi"];
+    frameworks = ["django", "flask", "fastapi"];
+    parser;
+    constructor() {
+        this.parser = new tree_sitter_1.default();
+        const language = tree_sitter_python_1.default;
+        this.parser.setLanguage(language);
+    }
+    parse(file) {
+        const tree = this.parser.parse(file.content);
+        const ast = { source: file.content, tree };
+        const frameworkHint = this.detectFramework(file.content);
+        return {
+            languageId: "python",
+            path: file.path,
+            ast,
+            meta: {
+                languageVersion: "unknown",
+                frameworkHint,
+            },
+        };
+    }
+    extractSymbols(parsed) {
+        const ast = parsed.ast;
+        const symbols = [];
+        this.walk(ast.tree.rootNode, (node) => {
+            if (node.type !== "class_definition" && node.type !== "function_definition") {
+                return;
+            }
+            const nameNode = node.childForFieldName("name");
+            if (!nameNode) {
+                return;
+            }
+            const name = this.nodeText(ast.source, nameNode);
+            const signature = this.firstLine(this.nodeText(ast.source, node));
+            if (node.type === "class_definition") {
+                const classText = this.nodeText(ast.source, node);
+                const baseClasses = this.extractBaseClasses(classText);
+                symbols.push({
+                    id: `${parsed.path}:class:${name}:${nameNode.startPosition.row + 1}`,
+                    kind: "class",
+                    name,
+                    filePath: parsed.path,
+                    range: this.rangeFromNode(nameNode),
+                    exported: !name.startsWith("_"),
+                    signature,
+                    metadata: baseClasses.length > 0 ? { baseClasses } : undefined,
+                });
+                return;
+            }
+            const container = this.findAncestor(node, "class_definition");
+            const containerName = container
+                ? this.nodeText(ast.source, container.childForFieldName("name") ?? container)
+                : undefined;
+            symbols.push({
+                id: `${parsed.path}:function:${name}:${nameNode.startPosition.row + 1}`,
+                kind: container ? "method" : "function",
+                name,
+                filePath: parsed.path,
+                range: this.rangeFromNode(nameNode),
+                exported: !name.startsWith("_"),
+                containerName,
+                signature,
+            });
+        });
+        return symbols;
+    }
+    extractImports(parsed) {
+        const ast = parsed.ast;
+        const imports = [];
+        const lines = ast.source.split(/\r?\n/);
+        this.walk(ast.tree.rootNode, (node) => {
+            if (node.type !== "import_statement" && node.type !== "import_from_statement") {
+                return;
+            }
+            const statement = this.nodeText(ast.source, node).trim();
+            if (!statement) {
+                return;
+            }
+            if (node.type === "import_statement") {
+                const modulePart = statement.replace(/^import\s+/, "");
+                const modules = modulePart
+                    .split(",")
+                    .map((part) => part.trim().split(/\s+as\s+/)[0])
+                    .filter(Boolean);
+                for (const moduleName of modules) {
+                    imports.push({
+                        id: `${parsed.path}:import:${moduleName}:${node.startPosition.row + 1}`,
+                        kind: "import",
+                        spec: moduleName,
+                        filePath: parsed.path,
+                        range: this.rangeForToken(lines, node.startPosition.row, moduleName, node),
+                    });
+                }
+                return;
+            }
+            const fromMatch = statement.match(/^from\s+([A-Za-z0-9_\.]+)\s+import\s+(.+)$/);
+            if (!fromMatch) {
+                return;
+            }
+            const moduleName = fromMatch[1];
+            const imported = fromMatch[2]
+                .split(",")
+                .map((part) => part.trim().split(/\s+as\s+/)[0])
+                .filter(Boolean);
+            imports.push({
+                id: `${parsed.path}:import:${moduleName}:${node.startPosition.row + 1}`,
+                kind: "import",
+                spec: moduleName,
+                filePath: parsed.path,
+                range: this.rangeForToken(lines, node.startPosition.row, moduleName, node),
+                metadata: imported.length > 0 ? { imported } : undefined,
+            });
+        });
+        return imports;
+    }
+    splitIntoChunks(parsed, opts) {
+        const ast = parsed.ast;
+        const content = ast.source;
+        if (!content.trim()) {
+            return [];
+        }
+        const lines = content.split(/\r?\n/);
+        const importNodes = [];
+        const definitionNodes = [];
+        this.walk(ast.tree.rootNode, (node) => {
+            if (node.type === "import_statement" || node.type === "import_from_statement") {
+                importNodes.push(node);
+            }
+            if (node.type === "function_definition" || node.type === "class_definition") {
+                definitionNodes.push(node);
+            }
+        });
+        const chunks = [];
+        if (importNodes.length > 0) {
+            const importStart = Math.min(...importNodes.map((node) => node.startPosition.row + 1));
+            const importEnd = Math.max(...importNodes.map((node) => node.endPosition.row + 1));
+            const importContent = lines
+                .slice(importStart - 1, importEnd)
+                .join("\n")
+                .trim();
+            if (importContent.length > 0) {
+                chunks.push({
+                    id: `${parsed.path}:chunk:imports`,
+                    filePath: parsed.path,
+                    range: {
+                        startLine: importStart,
+                        startCol: 1,
+                        endLine: importEnd,
+                        endCol: (lines[importEnd - 1]?.length ?? 0) + 1,
+                    },
+                    content: importContent,
+                    languageId: "python",
+                    estimatedTokens: Math.max(1, Math.ceil(importContent.length / 4)),
+                    metadata: {
+                        chunkType: "imports",
+                    },
+                });
+            }
+        }
+        const sortedDefinitions = definitionNodes.sort((a, b) => a.startPosition.row - b.startPosition.row);
+        for (const node of sortedDefinitions) {
+            const nodeContent = this.nodeText(content, node).trim();
+            if (!nodeContent)
+                continue;
+            const range = this.rangeFromNode(node);
+            const firstLine = this.firstLine(nodeContent);
+            const symbolMatch = firstLine.match(/^(?:async\s+def|def|class)\s+([A-Za-z_][A-Za-z0-9_]*)/);
+            chunks.push({
+                id: `${parsed.path}:chunk:${range.startLine}`,
+                filePath: parsed.path,
+                range,
+                content: nodeContent,
+                languageId: "python",
+                estimatedTokens: Math.max(1, Math.ceil(nodeContent.length / 4)),
+                metadata: {
+                    chunkType: node.type === "class_definition" ? "types" : "impl",
+                    primarySymbol: symbolMatch?.[1],
+                },
+            });
+        }
+        if (chunks.length > 0) {
+            return chunks;
+        }
+        const range = {
+            startLine: 1,
+            startCol: 1,
+            endLine: lines.length,
+            endCol: (lines[lines.length - 1]?.length ?? 0) + 1,
+        };
+        return [
+            {
+                id: `${parsed.path}:chunk:1`,
+                filePath: parsed.path,
+                range,
+                content,
+                languageId: "python",
+                estimatedTokens: Math.max(opts.targetTokens, Math.ceil(content.length / 4)),
+                metadata: {
+                    chunkStrategy: "tree-sitter-single-chunk",
+                    chunkType: "impl",
+                },
+            },
+        ];
+    }
+    getEntrypoints(filePaths) {
+        return filePaths.filter((filePath) => filePath.endsWith("__main__.py") || filePath.endsWith("manage.py"));
+    }
+    detectFramework(content) {
+        if (content.includes("django")) {
+            return "django";
+        }
+        if (content.includes("fastapi")) {
+            return "fastapi";
+        }
+        if (content.includes("flask")) {
+            return "flask";
+        }
+        return null;
+    }
+    extractBaseClasses(classText) {
+        const firstLine = this.firstLine(classText);
+        const match = firstLine.match(/^class\s+[A-Za-z_]\w*\s*\(([^)]*)\)\s*:/);
+        if (!match) {
+            return [];
+        }
+        return match[1]
+            .split(",")
+            .map((part) => part.trim())
+            .filter(Boolean);
+    }
+    nodeText(source, node) {
+        return source.slice(node.startIndex, node.endIndex);
+    }
+    firstLine(value) {
+        return value.split(/\r?\n/, 1)[0]?.trim() ?? "";
+    }
+    findAncestor(node, type) {
+        let current = node.parent;
+        while (current) {
+            if (current.type === type) {
+                return current;
+            }
+            current = current.parent;
+        }
+        return null;
+    }
+    rangeFromNode(node) {
+        const startCol = node.startPosition.column + 1;
+        const endCol = Math.max(startCol + 1, node.endPosition.column + 1);
+        return {
+            startLine: node.startPosition.row + 1,
+            startCol,
+            endLine: node.endPosition.row + 1,
+            endCol,
+        };
+    }
+    rangeForToken(lines, lineIndex, token, fallbackNode) {
+        const line = lines[lineIndex] ?? "";
+        const tokenIndex = line.indexOf(token);
+        if (tokenIndex < 0) {
+            return this.rangeFromNode(fallbackNode);
+        }
+        return {
+            startLine: lineIndex + 1,
+            startCol: tokenIndex + 1,
+            endLine: lineIndex + 1,
+            endCol: tokenIndex + Math.max(2, token.length + 1),
+        };
+    }
+    walk(node, visitor) {
+        visitor(node);
+        for (const child of node.namedChildren) {
+            this.walk(child, visitor);
+        }
+    }
+}
+exports.PythonPlugin = PythonPlugin;
+exports.plugin = new PythonPlugin();
+exports.default = exports.plugin;
+//# sourceMappingURL=python.js.map

package/dist/languages/python.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"python.js","sourceRoot":"","sources":["../../src/languages/python.ts"],"names":[],"mappings":";;;;;;AAAA,8DAAgC;AAChC,4EAA+C;AAkB/C,MAAa,YAAY;IACd,EAAE,GAAG,QAAQ,CAAA;IACb,WAAW,GAAG,QAAQ,CAAA;IACtB,cAAc,GAAG,CAAC,KAAK,EAAE,MAAM,CAAC,CAAA;IAChC,UAAU,GAAG,CAAC,QAAQ,EAAE,OAAO,EAAE,SAAS,CAAC,CAAA;IAEnC,MAAM,CAAQ;IAE/B;QACE,IAAI,CAAC,MAAM,GAAG,IAAI,qBAAM,EAAE,CAAA;QAC1B,MAAM,QAAQ,GAAG,4BAAiE,CAAA;QAClF,IAAI,CAAC,MAAM,CAAC,WAAW,CAAC,QAAQ,CAAC,CAAA;IACnC,CAAC;IAED,KAAK,CAAC,IAAgB;QACpB,MAAM,IAAI,GAAG,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;QAC5C,MAAM,GAAG,GAAc,EAAC,MAAM,EAAE,IAAI,CAAC,OAAO,EAAE,IAAI,EAAC,CAAA;QACnD,MAAM,aAAa,GAAG,IAAI,CAAC,eAAe,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;QAExD,OAAO;YACL,UAAU,EAAE,QAAQ;YACpB,IAAI,EAAE,IAAI,CAAC,IAAI;YACf,GAAG;YACH,IAAI,EAAE;gBACJ,eAAe,EAAE,SAAS;gBAC1B,aAAa;aACd;SACF,CAAA;IACH,CAAC;IAED,cAAc,CAAC,MAAkB;QAC/B,MAAM,GAAG,GAAG,MAAM,CAAC,GAAgB,CAAA;QACnC,MAAM,OAAO,GAAqB,EAAE,CAAA;QAEpC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,QAAQ,EAAE,CAAC,IAAI,EAAE,EAAE;YACpC,IAAI,IAAI,CAAC,IAAI,KAAK,kBAAkB,IAAI,IAAI,CAAC,IAAI,KAAK,qBAAqB,EAAE,CAAC;gBAC5E,OAAM;YACR,CAAC;YAED,MAAM,QAAQ,GAAG,IAAI,CAAC,iBAAiB,CAAC,MAAM,CAAC,CAAA;YAC/C,IAAI,CAAC,QAAQ,EAAE,CAAC;gBACd,OAAM;YACR,CAAC;YAED,MAAM,IAAI,GAAG,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,MAAM,EAAE,QAAQ,CAAC,CAAA;YAChD,MAAM,SAAS,GAAG,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC,CAAA;YAEjE,IAAI,IAAI,CAAC,IAAI,KAAK,kBAAkB,EAAE,CAAC;gBACrC,MAAM,SAAS,GAAG,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,MAAM,EAAE,IAAI,CAAC,CAAA;gBACjD,MAAM,WAAW,GAAG,IAAI,CAAC,kBAAkB,CAAC,SAAS,CAAC,CAAA;gBAEtD,OAAO,CAAC,IAAI,CAAC;oBACX,EAAE,EAAE,GAAG,MAAM,CAAC,IAAI,UAAU,IAAI,IAAI,QAAQ,CAAC,aAAa,CAAC,GAAG,GAAG,CAAC,EAAE;oBACpE,IAAI,EAAE,OAAO;oBACb,IAAI;oBACJ,QAAQ,EAAE,MAAM,CAAC,IAAI;oBACrB,KAAK,EAAE,IAAI,CAAC,aAAa,CAAC,QAAQ,CAAC;oBACnC,QAAQ,EAAE,CAAC,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC;oBAC/B,SAAS;oBACT,QAAQ,EAAE,WAAW,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,EAAC,WAAW,EAAC,CAAC,CAAC,CAAC,SAAS;iBAC7D,CAAC,CAAA;gBACF,OAAM;YACR,CAAC;YAED,MAAM,SAAS,GAAG,IAAI,CAAC,YAAY,CAAC,IAAI,EAAE,kBAAkB,CAAC,CAAA;YAC7D,MAAM,aAAa,GAAG,SAAS;gBAC7B,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,MAAM,EAAE,SAAS,CAAC,iBAAiB,CAAC,MAAM,CAAC,IAAI,SAAS,CAAC;gBAC7E,CAAC,CAAC,SAAS,CAAA;YAEb,OAAO,CAAC,IAAI,CAAC;gBACX,EAAE,EAAE,GAAG,MAAM,CAAC,IAAI,aAAa,IAAI,IAAI,QAAQ,CAAC,aAAa,CAAC,GAAG,GAAG,CAAC,EAAE;gBACvE,IAAI,EAAE,SAAS,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,UAAU;gBACvC,IAAI;gBACJ,QAAQ,EAAE,MAAM,CAAC,IAAI;gBACrB,KAAK,EAAE,IAAI,CAAC,aAAa,CAAC,QAAQ,CAAC;gBACnC,QAAQ,EAAE,CAAC,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC;gBAC/B,aAAa;gBACb,SAAS;aACV,CAAC,CAAA;QACJ,CAAC,CAAC,CAAA;QAEF,OAAO,OAAO,CAAA;IAChB,CAAC;IAED,cAAc,CAAC,MAAkB;QAC/B,MAAM,GAAG,GAAG,MAAM,CAAC,GAAgB,CAAA;QACnC,MAAM,OAAO,GAAqB,EAAE,CAAA;QACpC,MAAM,KAAK,GAAG,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,OAAO,CAAC,CAAA;QAEvC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,QAAQ,EAAE,CAAC,IAAI,EAAE,EAAE;YACpC,IAAI,IAAI,CAAC,IAAI,KAAK,kBAAkB,IAAI,IAAI,CAAC,IAAI,KAAK,uBAAuB,EAAE,CAAC;gBAC9E,OAAM;YACR,CAAC;YAED,MAAM,SAAS,GAAG,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC,IAAI,EAAE,CAAA;YACxD,IAAI,CAAC,SAAS,EAAE,CAAC;gBACf,OAAM;YACR,CAAC;YAED,IAAI,IAAI,CAAC,IAAI,KAAK,kBAAkB,EAAE,CAAC;gBACrC,MAAM,UAAU,GAAG,SAAS,CAAC,OAAO,CAAC,YAAY,EAAE,EAAE,CAAC,CAAA;gBACtD,MAAM,OAAO,GAAG,UAAU;qBACvB,KAAK,CAAC,GAAG,CAAC;qBACV,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,KAAK,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,CAAC;qBAC/C,MAAM,CAAC,OAAO,CAAC,CAAA;gBAElB,KAAK,MAAM,UAAU,IAAI,OAAO,EAAE,CAAC;oBACjC,OAAO,CAAC,IAAI,CAAC;wBACX,EAAE,EAAE,GAAG,MAAM,CAAC,IAAI,WAAW,UAAU,IAAI,IAAI,CAAC,aAAa,CAAC,GAAG,GAAG,CAAC,EAAE;wBACvE,IAAI,EAAE,QAAQ;wBACd,IAAI,EAAE,UAAU;wBAChB,QAAQ,EAAE,MAAM,CAAC,IAAI;wBACrB,KAAK,EAAE,IAAI,CAAC,aAAa,CAAC,KAAK,EAAE,IAAI,CAAC,aAAa,CAAC,GAAG,EAAE,UAAU,EAAE,IAAI,CAAC;qBAC3E,CAAC,CAAA;gBACJ,CAAC;gBACD,OAAM;YACR,CAAC;YAED,MAAM,SAAS,GAAG,SAAS,CAAC,KAAK,CAAC,4CAA4C,CAAC,CAAA;YAC/E,IAAI,CAAC,SAAS,EAAE,CAAC;gBACf,OAAM;YACR,CAAC;YAED,MAAM,UAAU,GAAG,SAAS,CAAC,CAAC,CAAC,CAAA;YAC/B,MAAM,QAAQ,GAAG,SAAS,CAAC,CAAC,CAAC;iBAC1B,KAAK,CAAC,GAAG,CAAC;iBACV,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,KAAK,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,CAAC;iBAC/C,MAAM,CAAC,OAAO,CAAC,CAAA;YAElB,OAAO,CAAC,IAAI,CAAC;gBACX,EAAE,EAAE,GAAG,MAAM,CAAC,IAAI,WAAW,UAAU,IAAI,IAAI,CAAC,aAAa,CAAC,GAAG,GAAG,CAAC,EAAE;gBACvE,IAAI,EAAE,QAAQ;gBACd,IAAI,EAAE,UAAU;gBAChB,QAAQ,EAAE,MAAM,CAAC,IAAI;gBACrB,KAAK,EAAE,IAAI,CAAC,aAAa,CAAC,KAAK,EAAE,IAAI,CAAC,aAAa,CAAC,GAAG,EAAE,UAAU,EAAE,IAAI,CAAC;gBAC1E,QAAQ,EAAE,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,EAAC,QAAQ,EAAC,CAAC,CAAC,CAAC,SAAS;aACvD,CAAC,CAAA;QACJ,CAAC,CAAC,CAAA;QAEF,OAAO,OAAO,CAAA;IAChB,CAAC;IAED,eAAe,CAAC,MAAkB,EAAE,IAAkB;QACpD,MAAM,GAAG,GAAG,MAAM,CAAC,GAAgB,CAAA;QACnC,MAAM,OAAO,GAAG,GAAG,CAAC,MAAM,CAAA;QAE1B,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,EAAE,CAAC;YACpB,OAAO,EAAE,CAAA;QACX,CAAC;QAED,MAAM,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC,OAAO,CAAC,CAAA;QACpC,MAAM,WAAW,GAAwB,EAAE,CAAA;QAC3C,MAAM,eAAe,GAAwB,EAAE,CAAA;QAE/C,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,QAAQ,EAAE,CAAC,IAAI,EAAE,EAAE;YACpC,IAAI,IAAI,CAAC,IAAI,KAAK,kBAAkB,IAAI,IAAI,CAAC,IAAI,KAAK,uBAAuB,EAAE,CAAC;gBAC9E,WAAW,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YACxB,CAAC;YACD,IAAI,IAAI,CAAC,IAAI,KAAK,qBAAqB,IAAI,IAAI,CAAC,IAAI,KAAK,kBAAkB,EAAE,CAAC;gBAC5E,eAAe,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAC5B,CAAC;QACH,CAAC,CAAC,CAAA;QAEF,MAAM,MAAM,GAAwB,EAAE,CAAA;QAEtC,IAAI,WAAW,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YAC3B,MAAM,WAAW,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,WAAW,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,aAAa,CAAC,GAAG,GAAG,CAAC,CAAC,CAAC,CAAA;YACtF,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,WAAW,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,WAAW,CAAC,GAAG,GAAG,CAAC,CAAC,CAAC,CAAA;YAClF,MAAM,aAAa,GAAG,KAAK;iBACxB,KAAK,CAAC,WAAW,GAAG,CAAC,EAAE,SAAS,CAAC;iBACjC,IAAI,CAAC,IAAI,CAAC;iBACV,IAAI,EAAE,CAAA;YACT,IAAI,aAAa,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBAC7B,MAAM,CAAC,IAAI,CAAC;oBACV,EAAE,EAAE,GAAG,MAAM,CAAC,IAAI,gBAAgB;oBAClC,QAAQ,EAAE,MAAM,CAAC,IAAI;oBACrB,KAAK,EAAE;wBACL,SAAS,EAAE,WAAW;wBACtB,QAAQ,EAAE,CAAC;wBACX,OAAO,EAAE,SAAS;wBAClB,MAAM,EAAE,CAAC,KAAK,CAAC,SAAS,GAAG,CAAC,CAAC,EAAE,MAAM,IAAI,CAAC,CAAC,GAAG,CAAC;qBAChD;oBACD,OAAO,EAAE,aAAa;oBACtB,UAAU,EAAE,QAAQ;oBACpB,eAAe,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,IAAI,CAAC,aAAa,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;oBACjE,QAAQ,EAAE;wBACR,SAAS,EAAE,SAAS;qBACrB;iBACF,CAAC,CAAA;YACJ,CAAC;QACH,CAAC;QAED,MAAM,iBAAiB,GAAG,eAAe,CAAC,IAAI,CAC5C,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,aAAa,CAAC,GAAG,GAAG,CAAC,CAAC,aAAa,CAAC,GAAG,CACpD,CAAA;QACD,KAAK,MAAM,IAAI,IAAI,iBAAiB,EAAE,CAAC;YACrC,MAAM,WAAW,GAAG,IAAI,CAAC,QAAQ,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC,IAAI,EAAE,CAAA;YACvD,IAAI,CAAC,WAAW;gBAAE,SAAQ;YAC1B,MAAM,KAAK,GAAG,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC,CAAA;YACtC,MAAM,SAAS,GAAG,IAAI,CAAC,SAAS,CAAC,WAAW,CAAC,CAAA;YAC7C,MAAM,WAAW,GAAG,SAAS,CAAC,KAAK,CAAC,uDAAuD,CAAC,CAAA;YAC5F,MAAM,CAAC,IAAI,CAAC;gBACV,EAAE,EAAE,GAAG,MAAM,CAAC,IAAI,UAAU,KAAK,CAAC,SAAS,EAAE;gBAC7C,QAAQ,EAAE,MAAM,CAAC,IAAI;gBACrB,KAAK;gBACL,OAAO,EAAE,WAAW;gBACpB,UAAU,EAAE,QAAQ;gBACpB,eAAe,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,IAAI,CAAC,WAAW,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;gBAC/D,QAAQ,EAAE;oBACR,SAAS,EAAE,IAAI,CAAC,IAAI,KAAK,kBAAkB,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM;oBAC9D,aAAa,EAAE,WAAW,EAAE,CAAC,CAAC,CAAC;iBAChC;aACF,CAAC,CAAA;QACJ,CAAC;QAED,IAAI,MAAM,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACtB,OAAO,MAAM,CAAA;QACf,CAAC;QAED,MAAM,KAAK,GAAc;YACvB,SAAS,EAAE,CAAC;YACZ,QAAQ,EAAE,CAAC;YACX,OAAO,EAAE,KAAK,CAAC,MAAM;YACrB,MAAM,EAAE,CAAC,KAAK,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,EAAE,MAAM,IAAI,CAAC,CAAC,GAAG,CAAC;SACnD,CAAA;QAED,OAAO;YACL;gBACE,EAAE,EAAE,GAAG,MAAM,CAAC,IAAI,UAAU;gBAC5B,QAAQ,EAAE,MAAM,CAAC,IAAI;gBACrB,KAAK;gBACL,OAAO;gBACP,UAAU,EAAE,QAAQ;gBACpB,eAAe,EAAE,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,YAAY,EAAE,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;gBAC3E,QAAQ,EAAE;oBACR,aAAa,EAAE,0BAA0B;oBACzC,SAAS,EAAE,MAAM;iBAClB;aACF;SACF,CAAA;IACH,CAAC;IAED,cAAc,CAAC,SAAmB;QAChC,OAAO,SAAS,CAAC,MAAM,CACrB,CAAC,QAAQ,EAAE,EAAE,CAAC,QAAQ,CAAC,QAAQ,CAAC,aAAa,CAAC,IAAI,QAAQ,CAAC,QAAQ,CAAC,WAAW,CAAC,CACjF,CAAA;IACH,CAAC;IAEO,eAAe,CAAC,OAAe;QACrC,IAAI,OAAO,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;YAC/B,OAAO,QAAQ,CAAA;QACjB,CAAC;QACD,IAAI,OAAO,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE,CAAC;YAChC,OAAO,SAAS,CAAA;QAClB,CAAC;QACD,IAAI,OAAO,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;YAC9B,OAAO,OAAO,CAAA;QAChB,CAAC;QACD,OAAO,IAAI,CAAA;IACb,CAAC;IAEO,kBAAkB,CAAC,SAAiB;QAC1C,MAAM,SAAS,GAAG,IAAI,CAAC,SAAS,CAAC,SAAS,CAAC,CAAA;QAC3C,MAAM,KAAK,GAAG,SAAS,CAAC,KAAK,CAAC,yCAAyC,CAAC,CAAA;QACxE,IAAI,CAAC,KAAK,EAAE,CAAC;YACX,OAAO,EAAE,CAAA;QACX,CAAC;QAED,OAAO,KAAK,CAAC,CAAC,CAAC;aACZ,KAAK,CAAC,GAAG,CAAC;aACV,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;aAC1B,MAAM,CAAC,OAAO,CAAC,CAAA;IACpB,CAAC;IAEO,QAAQ,CAAC,MAAc,EAAE,IAAuB;QACtD,OAAO,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,UAAU,EAAE,IAAI,CAAC,QAAQ,CAAC,CAAA;IACrD,CAAC;IAEO,SAAS,CAAC,KAAa;QAC7B,OAAO,KAAK,CAAC,KAAK,CAAC,OAAO,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,IAAI,EAAE,CAAA;IACjD,CAAC;IAEO,YAAY,CAAC,IAAuB,EAAE,IAAY;QACxD,IAAI,OAAO,GAA6B,IAAI,CAAC,MAAM,CAAA;QACnD,OAAO,OAAO,EAAE,CAAC;YACf,IAAI,OAAO,CAAC,IAAI,KAAK,IAAI,EAAE,CAAC;gBAC1B,OAAO,OAAO,CAAA;YAChB,CAAC;YACD,OAAO,GAAG,OAAO,CAAC,MAAM,CAAA;QAC1B,CAAC;QACD,OAAO,IAAI,CAAA;IACb,CAAC;IAEO,aAAa,CAAC,IAAuB;QAC3C,MAAM,QAAQ,GAAG,IAAI,CAAC,aAAa,CAAC,MAAM,GAAG,CAAC,CAAA;QAC9C,MAAM,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC,QAAQ,GAAG,CAAC,EAAE,IAAI,CAAC,WAAW,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;QAClE,OAAO;YACL,SAAS,EAAE,IAAI,CAAC,aAAa,CAAC,GAAG,GAAG,CAAC;YACrC,QAAQ;YACR,OAAO,EAAE,IAAI,CAAC,WAAW,CAAC,GAAG,GAAG,CAAC;YACjC,MAAM;SACP,CAAA;IACH,CAAC;IAEO,aAAa,CACnB,KAAe,EACf,SAAiB,EACjB,KAAa,EACb,YAA+B;QAE/B,MAAM,IAAI,GAAG,KAAK,CAAC,SAAS,CAAC,IAAI,EAAE,CAAA;QACnC,MAAM,UAAU,GAAG,IAAI,CAAC,OAAO,CAAC,KAAK,CAAC,CAAA;QACtC,IAAI,UAAU,GAAG,CAAC,EAAE,CAAC;YACnB,OAAO,IAAI,CAAC,aAAa,CAAC,YAAY,CAAC,CAAA;QACzC,CAAC;QAED,OAAO;YACL,SAAS,EAAE,SAAS,GAAG,CAAC;YACxB,QAAQ,EAAE,UAAU,GAAG,CAAC;YACxB,OAAO,EAAE,SAAS,GAAG,CAAC;YACtB,MAAM,EAAE,UAAU,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC;SACnD,CAAA;IACH,CAAC;IAEO,IAAI,CAAC,IAAuB,EAAE,OAA0C;QAC9E,OAAO,CAAC,IAAI,CAAC,CAAA;QACb,KAAK,MAAM,KAAK,IAAI,IAAI,CAAC,aAAa,EAAE,CAAC;YACvC,IAAI,CAAC,IAAI,CAAC,KAAK,EAAE,OAAO,CAAC,CAAA;QAC3B,CAAC;IACH,CAAC;CACF;AA1UD,oCA0UC;AAEY,QAAA,MAAM,GAAG,IAAI,YAAY,EAAE,CAAA;AACxC,kBAAe,cAAM,CAAA"}

package/dist/languages/ruby.d.ts

@@ -0,0 +1,25 @@
+import type { ChunkOptions, LanguageCodeChunk, LanguageImport, LanguagePlugin, LanguageSymbol, ParsedFile, SourceFile } from "./plugin.js";
+export declare class RubyPlugin implements LanguagePlugin {
+    readonly id = "ruby";
+    readonly displayName = "Ruby";
+    readonly fileExtensions: string[];
+    readonly frameworks: string[];
+    private readonly parser;
+    constructor();
+    parse(file: SourceFile): ParsedFile;
+    extractSymbols(parsed: ParsedFile): LanguageSymbol[];
+    extractImports(parsed: ParsedFile): LanguageImport[];
+    splitIntoChunks(parsed: ParsedFile, opts: ChunkOptions): LanguageCodeChunk[];
+    getEntrypoints(filePaths: string[]): string[];
+    private detectFramework;
+    private extractImportSpec;
+    private getQualifiedContainerName;
+    private resolveMethodVisibility;
+    private nodeText;
+    private firstLine;
+    private rangeFromNode;
+    private rangeForToken;
+    private walk;
+}
+export declare const plugin: RubyPlugin;
+export default plugin;

package/dist/languages/ruby.js

@@ -0,0 +1,328 @@
+"use strict";
+var __importDefault = (this && this.__importDefault) || function (mod) {
+    return (mod && mod.__esModule) ? mod : { "default": mod };
+};
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.plugin = exports.RubyPlugin = void 0;
+const tree_sitter_1 = __importDefault(require("tree-sitter"));
+const tree_sitter_ruby_1 = __importDefault(require("tree-sitter-ruby"));
+const RUBY_REQUIRE_METHODS = new Set(["require", "require_relative"]);
+class RubyPlugin {
+    id = "ruby";
+    displayName = "Ruby";
+    fileExtensions = [".rb"];
+    frameworks = ["rails", "sinatra"];
+    parser;
+    constructor() {
+        this.parser = new tree_sitter_1.default();
+        const language = tree_sitter_ruby_1.default;
+        this.parser.setLanguage(language);
+    }
+    parse(file) {
+        const tree = this.parser.parse(file.content);
+        const ast = { source: file.content, tree };
+        return {
+            languageId: "ruby",
+            path: file.path,
+            ast,
+            meta: {
+                frameworkHint: this.detectFramework(file.content),
+            },
+        };
+    }
+    extractSymbols(parsed) {
+        const ast = parsed.ast;
+        const symbols = [];
+        this.walk(ast.tree.rootNode, (node) => {
+            if (node.type !== "class" &&
+                node.type !== "module" &&
+                node.type !== "method" &&
+                node.type !== "singleton_method") {
+                return;
+            }
+            const nameNode = node.childForFieldName("name");
+            if (!nameNode) {
+                return;
+            }
+            const name = this.nodeText(ast.source, nameNode);
+            const signature = this.firstLine(this.nodeText(ast.source, node));
+            if (node.type === "class" || node.type === "module") {
+                symbols.push({
+                    id: `${parsed.path}:${node.type}:${name}:${nameNode.startPosition.row + 1}`,
+                    kind: node.type === "class" ? "class" : "module",
+                    name,
+                    filePath: parsed.path,
+                    range: this.rangeFromNode(nameNode),
+                    exported: true,
+                    signature,
+                });
+                return;
+            }
+            const containerName = this.getQualifiedContainerName(ast.source, node);
+            const visibility = this.resolveMethodVisibility(ast.source, node);
+            symbols.push({
+                id: `${parsed.path}:${node.type}:${name}:${nameNode.startPosition.row + 1}`,
+                kind: "method",
+                name,
+                filePath: parsed.path,
+                range: this.rangeFromNode(nameNode),
+                exported: visibility === "public" && !name.startsWith("_"),
+                containerName,
+                signature,
+                metadata: node.type === "singleton_method"
+                    ? {
+                        singleton: true,
+                        visibility,
+                    }
+                    : visibility === "public"
+                        ? undefined
+                        : { visibility },
+            });
+        });
+        return symbols;
+    }
+    extractImports(parsed) {
+        const ast = parsed.ast;
+        const imports = [];
+        const lines = ast.source.split(/\r?\n/);
+        this.walk(ast.tree.rootNode, (node) => {
+            if (node.type !== "call") {
+                return;
+            }
+            const methodNode = node.childForFieldName("method");
+            if (!methodNode) {
+                return;
+            }
+            const kind = this.nodeText(ast.source, methodNode);
+            if (!RUBY_REQUIRE_METHODS.has(kind)) {
+                return;
+            }
+            const statement = this.nodeText(ast.source, node).trim();
+            const spec = this.extractImportSpec(statement);
+            if (!spec) {
+                return;
+            }
+            imports.push({
+                id: `${parsed.path}:${kind}:${spec}:${node.startPosition.row + 1}`,
+                kind: "require",
+                spec,
+                filePath: parsed.path,
+                range: this.rangeForToken(lines, node.startPosition.row, spec, node),
+                metadata: {
+                    syntax: kind,
+                },
+            });
+        });
+        return imports;
+    }
+    splitIntoChunks(parsed, opts) {
+        const ast = parsed.ast;
+        const content = ast.source;
+        if (!content.trim()) {
+            return [];
+        }
+        const lines = content.split(/\r?\n/);
+        const importNodes = [];
+        const definitionNodes = [];
+        this.walk(ast.tree.rootNode, (node) => {
+            if (node.type === "call") {
+                const methodNode = node.childForFieldName("method");
+                const methodName = methodNode ? this.nodeText(content, methodNode) : "";
+                if (RUBY_REQUIRE_METHODS.has(methodName) &&
+                    this.extractImportSpec(this.nodeText(content, node))) {
+                    importNodes.push(node);
+                }
+                return;
+            }
+            if (node.type === "class" || node.type === "module") {
+                const nameNode = node.childForFieldName("name");
+                const primarySymbol = nameNode
+                    ? this.nodeText(content, nameNode)
+                    : undefined;
+                definitionNodes.push({
+                    node,
+                    chunkType: "types",
+                    primarySymbol,
+                });
+                return;
+            }
+            if (node.type === "method" || node.type === "singleton_method") {
+                const nameNode = node.childForFieldName("name");
+                const primarySymbol = nameNode
+                    ? this.nodeText(content, nameNode)
+                    : undefined;
+                definitionNodes.push({
+                    node,
+                    chunkType: "impl",
+                    primarySymbol,
+                });
+            }
+        });
+        const chunks = [];
+        if (importNodes.length > 0) {
+            const importStart = Math.min(...importNodes.map((node) => node.startPosition.row + 1));
+            const importEnd = Math.max(...importNodes.map((node) => node.endPosition.row + 1));
+            const importContent = lines
+                .slice(importStart - 1, importEnd)
+                .join("\n")
+                .trim();
+            if (importContent.length > 0) {
+                chunks.push({
+                    id: `${parsed.path}:chunk:imports`,
+                    filePath: parsed.path,
+                    range: {
+                        startLine: importStart,
+                        startCol: 1,
+                        endLine: importEnd,
+                        endCol: (lines[importEnd - 1]?.length ?? 0) + 1,
+                    },
+                    content: importContent,
+                    languageId: "ruby",
+                    estimatedTokens: Math.max(1, Math.ceil(importContent.length / 4)),
+                    metadata: {
+                        chunkType: "imports",
+                    },
+                });
+            }
+        }
+        const sortedDefinitions = definitionNodes.sort((a, b) => a.node.startPosition.row - b.node.startPosition.row);
+        for (const definition of sortedDefinitions) {
+            const nodeContent = this.nodeText(content, definition.node).trim();
+            if (!nodeContent) {
+                continue;
+            }
+            const range = this.rangeFromNode(definition.node);
+            chunks.push({
+                id: `${parsed.path}:chunk:${range.startLine}`,
+                filePath: parsed.path,
+                range,
+                content: nodeContent,
+                languageId: "ruby",
+                estimatedTokens: Math.max(1, Math.ceil(nodeContent.length / 4)),
+                metadata: {
+                    chunkType: definition.chunkType,
+                    primarySymbol: definition.primarySymbol,
+                },
+            });
+        }
+        if (chunks.length > 0) {
+            return chunks;
+        }
+        return [
+            {
+                id: `${parsed.path}:chunk:1`,
+                filePath: parsed.path,
+                range: {
+                    startLine: 1,
+                    startCol: 1,
+                    endLine: lines.length,
+                    endCol: (lines[lines.length - 1]?.length ?? 0) + 1,
+                },
+                content,
+                languageId: "ruby",
+                estimatedTokens: Math.max(opts.targetTokens, Math.ceil(content.length / 4)),
+                metadata: {
+                    chunkStrategy: "tree-sitter-single-chunk",
+                    chunkType: "impl",
+                },
+            },
+        ];
+    }
+    getEntrypoints(filePaths) {
+        return filePaths.filter((filePath) => /\/bin\/.+\.rb$/.test(filePath) ||
+            /\/(?:main|app|cli|boot)\.rb$/.test(filePath));
+    }
+    detectFramework(content) {
+        if (content.includes("Rails") || content.includes("ActiveRecord")) {
+            return "rails";
+        }
+        if (content.includes("Sinatra")) {
+            return "sinatra";
+        }
+        return null;
+    }
+    extractImportSpec(statement) {
+        const match = statement.match(/^(?:require|require_relative)\s+(?:\(?\s*)?["']?([^"'\)]+)["']?/);
+        return match?.[1]?.trim() || null;
+    }
+    getQualifiedContainerName(source, node) {
+        const containers = [];
+        let current = node.parent;
+        while (current) {
+            if (current.type === "class" || current.type === "module") {
+                const nameNode = current.childForFieldName("name");
+                if (nameNode) {
+                    containers.push(this.nodeText(source, nameNode));
+                }
+            }
+            current = current.parent;
+        }
+        if (containers.length === 0) {
+            return undefined;
+        }
+        return containers.reverse().join("::");
+    }
+    resolveMethodVisibility(source, node) {
+        let sibling = node.previousNamedSibling;
+        while (sibling) {
+            if (sibling.type === "identifier" &&
+                (this.nodeText(source, sibling) === "private" ||
+                    this.nodeText(source, sibling) === "protected" ||
+                    this.nodeText(source, sibling) === "public")) {
+                return this.nodeText(source, sibling);
+            }
+            if (sibling.type === "call") {
+                const methodNode = sibling.childForFieldName("method");
+                if (methodNode) {
+                    const methodName = this.nodeText(source, methodNode);
+                    if (methodName === "private" ||
+                        methodName === "protected" ||
+                        methodName === "public") {
+                        return methodName;
+                    }
+                }
+            }
+            sibling = sibling.previousNamedSibling;
+        }
+        return "public";
+    }
+    nodeText(source, node) {
+        return source.slice(node.startIndex, node.endIndex);
+    }
+    firstLine(value) {
+        return value.split(/\r?\n/, 1)[0]?.trim() ?? "";
+    }
+    rangeFromNode(node) {
+        const startCol = node.startPosition.column + 1;
+        const endCol = Math.max(startCol + 1, node.endPosition.column + 1);
+        return {
+            startLine: node.startPosition.row + 1,
+            startCol,
+            endLine: node.endPosition.row + 1,
+            endCol,
+        };
+    }
+    rangeForToken(lines, lineIndex, token, fallbackNode) {
+        const line = lines[lineIndex] ?? "";
+        const tokenIndex = line.indexOf(token);
+        if (tokenIndex < 0) {
+            return this.rangeFromNode(fallbackNode);
+        }
+        return {
+            startLine: lineIndex + 1,
+            startCol: tokenIndex + 1,
+            endLine: lineIndex + 1,
+            endCol: tokenIndex + Math.max(2, token.length + 1),
+        };
+    }
+    walk(node, visitor) {
+        visitor(node);
+        for (const child of node.namedChildren) {
+            this.walk(child, visitor);
+        }
+    }
+}
+exports.RubyPlugin = RubyPlugin;
+exports.plugin = new RubyPlugin();
+exports.default = exports.plugin;
+//# sourceMappingURL=ruby.js.map