more tools

This commit is contained in:
Dax Raad
2025-05-26 14:09:17 -04:00
parent 113c49457f
commit 80555f13e0
10 changed files with 701 additions and 53 deletions

View File

@@ -3,7 +3,6 @@ import { AppPath } from "./path";
import { Log } from "../util/log";
import { Context } from "../util/context";
import { Config } from "./config";
import { Share } from "../share/share";
export namespace App {
const log = Log.create({ service: "app" });
@@ -35,7 +34,6 @@ export namespace App {
get root() {
return input.directory;
},
service<T extends (app: any) => any>(service: any, init: T) {},
};
return result;

View File

@@ -6,6 +6,7 @@ import { Storage } from "../storage/storage";
import { Log } from "../util/log";
import {
convertToModelMessages,
stepCountIs,
streamText,
type TextUIPart,
type ToolInvocationUIPart,
@@ -169,7 +170,7 @@ export namespace Session {
const model = await LLM.findModel("claude-sonnet-4-20250514");
const result = streamText({
maxSteps: 1000,
stopWhen: stepCountIs(1000),
messages: convertToModelMessages(msgs),
temperature: 0,
tools,

137
js/src/tool/fetch.ts Normal file
View File

@@ -0,0 +1,137 @@
import { z } from "zod";
import { Tool } from "./tool";
import { JSDOM } from "jsdom";
import TurndownService from "turndown";
const MAX_RESPONSE_SIZE = 5 * 1024 * 1024; // 5MB
const DEFAULT_TIMEOUT = 30 * 1000; // 30 seconds
const MAX_TIMEOUT = 120 * 1000; // 2 minutes
const DESCRIPTION = `Fetches content from a URL and returns it in the specified format.
WHEN TO USE THIS TOOL:
- Use when you need to download content from a URL
- Helpful for retrieving documentation, API responses, or web content
- Useful for getting external information to assist with tasks
HOW TO USE:
- Provide the URL to fetch content from
- Specify the desired output format (text, markdown, or html)
- Optionally set a timeout for the request
FEATURES:
- Supports three output formats: text, markdown, and html
- Automatically handles HTTP redirects
- Sets reasonable timeouts to prevent hanging
- Validates input parameters before making requests
LIMITATIONS:
- Maximum response size is 5MB
- Only supports HTTP and HTTPS protocols
- Cannot handle authentication or cookies
- Some websites may block automated requests
TIPS:
- Use text format for plain text content or simple API responses
- Use markdown format for content that should be rendered with formatting
- Use html format when you need the raw HTML structure
- Set appropriate timeouts for potentially slow websites`;
export const Fetch = Tool.define({
name: "fetch",
description: DESCRIPTION,
parameters: z.object({
url: z.string().describe("The URL to fetch content from"),
format: z
.enum(["text", "markdown", "html"])
.describe(
"The format to return the content in (text, markdown, or html)",
),
timeout: z
.number()
.min(0)
.max(MAX_TIMEOUT / 1000)
.describe("Optional timeout in seconds (max 120)")
.optional(),
}),
async execute(params, opts) {
// Validate URL
if (
!params.url.startsWith("http://") &&
!params.url.startsWith("https://")
) {
throw new Error("URL must start with http:// or https://");
}
const timeout = Math.min(
(params.timeout ?? DEFAULT_TIMEOUT / 1000) * 1000,
MAX_TIMEOUT,
);
const controller = new AbortController();
const timeoutId = setTimeout(() => controller.abort(), timeout);
if (opts?.abortSignal) {
opts.abortSignal.addEventListener("abort", () => controller.abort());
}
const response = await fetch(params.url, {
signal: controller.signal,
headers: {
"User-Agent": "opencode/1.0",
},
});
clearTimeout(timeoutId);
if (!response.ok) {
throw new Error(`Request failed with status code: ${response.status}`);
}
// Check content length
const contentLength = response.headers.get("content-length");
if (contentLength && parseInt(contentLength) > MAX_RESPONSE_SIZE) {
throw new Error("Response too large (exceeds 5MB limit)");
}
const arrayBuffer = await response.arrayBuffer();
if (arrayBuffer.byteLength > MAX_RESPONSE_SIZE) {
throw new Error("Response too large (exceeds 5MB limit)");
}
const content = new TextDecoder().decode(arrayBuffer);
const contentType = response.headers.get("content-type") || "";
switch (params.format) {
case "text":
if (contentType.includes("text/html")) {
const text = extractTextFromHTML(content);
return { output: text };
}
return { output: content };
case "markdown":
if (contentType.includes("text/html")) {
const markdown = convertHTMLToMarkdown(content);
return { output: markdown };
}
return { output: "```\n" + content + "\n```" };
case "html":
return { output: content };
default:
return { output: content };
}
},
});
function extractTextFromHTML(html: string): string {
const dom = new JSDOM(html);
const text = dom.window.document.body?.textContent || "";
return text.replace(/\s+/g, " ").trim();
}
function convertHTMLToMarkdown(html: string): string {
const turndownService = new TurndownService();
return turndownService.turndown(html);
}

345
js/src/tool/grep.ts Normal file
View File

@@ -0,0 +1,345 @@
import { z } from "zod";
import { Tool } from "./tool";
import { App } from "../app";
import { spawn } from "child_process";
import { promises as fs } from "fs";
import path from "path";
const DESCRIPTION = `Fast content search tool that finds files containing specific text or patterns, returning matching file paths sorted by modification time (newest first).
WHEN TO USE THIS TOOL:
- Use when you need to find files containing specific text or patterns
- Great for searching code bases for function names, variable declarations, or error messages
- Useful for finding all files that use a particular API or pattern
HOW TO USE:
- Provide a regex pattern to search for within file contents
- Set literal_text=true if you want to search for the exact text with special characters (recommended for non-regex users)
- Optionally specify a starting directory (defaults to current working directory)
- Optionally provide an include pattern to filter which files to search
- Results are sorted with most recently modified files first
REGEX PATTERN SYNTAX (when literal_text=false):
- Supports standard regular expression syntax
- 'function' searches for the literal text "function"
- 'log\\..*Error' finds text starting with "log." and ending with "Error"
- 'import\\s+.*\\s+from' finds import statements in JavaScript/TypeScript
COMMON INCLUDE PATTERN EXAMPLES:
- '*.js' - Only search JavaScript files
- '*.{ts,tsx}' - Only search TypeScript files
- '*.go' - Only search Go files
LIMITATIONS:
- Results are limited to 100 files (newest first)
- Performance depends on the number of files being searched
- Very large binary files may be skipped
- Hidden files (starting with '.') are skipped
TIPS:
- For faster, more targeted searches, first use Glob to find relevant files, then use Grep
- When doing iterative exploration that may require multiple rounds of searching, consider using the Agent tool instead
- Always check if results are truncated and refine your search pattern if needed
- Use literal_text=true when searching for exact text containing special characters like dots, parentheses, etc.`;
interface GrepMatch {
path: string;
modTime: number;
lineNum: number;
lineText: string;
}
function escapeRegexPattern(pattern: string): string {
const specialChars = [
"\\",
".",
"+",
"*",
"?",
"(",
")",
"[",
"]",
"{",
"}",
"^",
"$",
"|",
];
let escaped = pattern;
for (const char of specialChars) {
escaped = escaped.replaceAll(char, "\\" + char);
}
return escaped;
}
function globToRegex(glob: string): string {
let regexPattern = glob.replaceAll(".", "\\.");
regexPattern = regexPattern.replaceAll("*", ".*");
regexPattern = regexPattern.replaceAll("?", ".");
// Handle {a,b,c} patterns
regexPattern = regexPattern.replace(/\{([^}]+)\}/g, (match, inner) => {
return "(" + inner.replace(/,/g, "|") + ")";
});
return regexPattern;
}
async function searchWithRipgrep(
pattern: string,
searchPath: string,
include?: string,
): Promise<GrepMatch[]> {
return new Promise((resolve, reject) => {
const args = ["-n", pattern];
if (include) {
args.push("--glob", include);
}
args.push(searchPath);
const rg = spawn("rg", args);
let output = "";
let errorOutput = "";
rg.stdout.on("data", (data) => {
output += data.toString();
});
rg.stderr.on("data", (data) => {
errorOutput += data.toString();
});
rg.on("close", async (code) => {
if (code === 1) {
// No matches found
resolve([]);
return;
}
if (code !== 0) {
reject(new Error(`ripgrep failed: ${errorOutput}`));
return;
}
const lines = output.trim().split("\n");
const matches: GrepMatch[] = [];
for (const line of lines) {
if (!line) continue;
// Parse ripgrep output format: file:line:content
const parts = line.split(":", 3);
if (parts.length < 3) continue;
const filePath = parts[0];
const lineNum = parseInt(parts[1], 10);
const lineText = parts[2];
try {
const stats = await fs.stat(filePath);
matches.push({
path: filePath,
modTime: stats.mtime.getTime(),
lineNum,
lineText,
});
} catch {
// Skip files we can't access
continue;
}
}
resolve(matches);
});
rg.on("error", (err) => {
reject(err);
});
});
}
async function searchFilesWithRegex(
pattern: string,
rootPath: string,
include?: string,
): Promise<GrepMatch[]> {
const matches: GrepMatch[] = [];
const regex = new RegExp(pattern);
let includePattern: RegExp | undefined;
if (include) {
const regexPattern = globToRegex(include);
includePattern = new RegExp(regexPattern);
}
async function walkDir(dir: string) {
if (matches.length >= 200) return;
try {
const entries = await fs.readdir(dir, { withFileTypes: true });
for (const entry of entries) {
if (matches.length >= 200) break;
const fullPath = path.join(dir, entry.name);
if (entry.isDirectory()) {
// Skip hidden directories
if (entry.name.startsWith(".")) continue;
await walkDir(fullPath);
} else if (entry.isFile()) {
// Skip hidden files
if (entry.name.startsWith(".")) continue;
if (includePattern && !includePattern.test(fullPath)) {
continue;
}
try {
const content = await fs.readFile(fullPath, "utf-8");
const lines = content.split("\n");
for (let i = 0; i < lines.length; i++) {
if (regex.test(lines[i])) {
const stats = await fs.stat(fullPath);
matches.push({
path: fullPath,
modTime: stats.mtime.getTime(),
lineNum: i + 1,
lineText: lines[i],
});
break; // Only first match per file
}
}
} catch {
// Skip files we can't read
continue;
}
}
}
} catch {
// Skip directories we can't read
return;
}
}
await walkDir(rootPath);
return matches;
}
async function searchFiles(
pattern: string,
rootPath: string,
include?: string,
limit: number = 100,
): Promise<{ matches: GrepMatch[]; truncated: boolean }> {
let matches: GrepMatch[];
try {
matches = await searchWithRipgrep(pattern, rootPath, include);
} catch {
matches = await searchFilesWithRegex(pattern, rootPath, include);
}
// Sort by modification time (newest first)
matches.sort((a, b) => b.modTime - a.modTime);
const truncated = matches.length > limit;
if (truncated) {
matches = matches.slice(0, limit);
}
return { matches, truncated };
}
export const grep = Tool.define({
name: "grep",
description: DESCRIPTION,
parameters: z.object({
pattern: z
.string()
.describe("The regex pattern to search for in file contents"),
path: z
.string()
.describe(
"The directory to search in. Defaults to the current working directory.",
)
.optional(),
include: z
.string()
.describe(
'File pattern to include in the search (e.g. "*.js", "*.{ts,tsx}")',
)
.optional(),
literal_text: z
.boolean()
.describe(
"If true, the pattern will be treated as literal text with special regex characters escaped. Default is false.",
)
.optional(),
}),
async execute(params) {
if (!params.pattern) {
throw new Error("pattern is required");
}
const app = await App.use();
const searchPath = params.path || app.root;
// If literal_text is true, escape the pattern
const searchPattern = params.literal_text
? escapeRegexPattern(params.pattern)
: params.pattern;
const { matches, truncated } = await searchFiles(
searchPattern,
searchPath,
params.include,
100,
);
let output: string;
if (matches.length === 0) {
output = "No files found";
} else {
const lines = [`Found ${matches.length} matches`];
let currentFile = "";
for (const match of matches) {
if (currentFile !== match.path) {
if (currentFile !== "") {
lines.push("");
}
currentFile = match.path;
lines.push(`${match.path}:`);
}
if (match.lineNum > 0) {
lines.push(` Line ${match.lineNum}: ${match.lineText}`);
} else {
lines.push(` ${match.path}`);
}
}
if (truncated) {
lines.push("");
lines.push(
"(Results are truncated. Consider using a more specific path or pattern.)",
);
}
output = lines.join("\n");
}
return {
metadata: {
matches: matches.length,
truncated,
},
output,
};
},
});

View File

@@ -1,6 +1,7 @@
export * from "./tool";
export * from "./bash";
export * from "./edit";
export * from "./fetch";
export * from "./glob";
export * from "./grep";
export * from "./view";
export * from "./ls";
export * from "./ls";

View File

@@ -65,9 +65,8 @@ export const ls = Tool.define({
searchPath = path.join(app.root, searchPath);
}
try {
await fs.promises.stat(searchPath);
} catch (err) {
const stat = await fs.promises.stat(searchPath).catch(() => null);
if (!stat) {
return {
metadata: {},
output: `Path does not exist: ${searchPath}`,
@@ -88,7 +87,7 @@ export const ls = Tool.define({
return {
metadata: {
numberOfFiles: files.length,
count: files.length,
truncated,
},
output,
@@ -110,40 +109,38 @@ async function listDirectory(
return;
}
try {
const entries = await fs.promises.readdir(dir, { withFileTypes: true });
const entries = await fs.promises
.readdir(dir, { withFileTypes: true })
.catch(() => []);
for (const entry of entries) {
const fullPath = path.join(dir, entry.name);
for (const entry of entries) {
const fullPath = path.join(dir, entry.name);
if (shouldSkip(fullPath, ignorePatterns)) {
continue;
if (shouldSkip(fullPath, ignorePatterns)) {
continue;
}
if (entry.isDirectory()) {
if (fullPath !== initialPath) {
results.push(fullPath + path.sep);
}
if (entry.isDirectory()) {
if (fullPath !== initialPath) {
results.push(fullPath + path.sep);
}
if (results.length < limit) {
await walk(fullPath);
} else {
truncated = true;
return;
}
} else if (entry.isFile()) {
if (fullPath !== initialPath) {
results.push(fullPath);
}
if (results.length < limit) {
await walk(fullPath);
} else {
truncated = true;
return;
}
} else if (entry.isFile()) {
if (fullPath !== initialPath) {
results.push(fullPath);
}
if (results.length >= limit) {
truncated = true;
return;
}
if (results.length >= limit) {
truncated = true;
return;
}
}
} catch (err) {
// Skip directories we don't have permission to access
}
}
@@ -200,13 +197,9 @@ function shouldSkip(filePath: string, ignorePatterns: string[]): boolean {
}
for (const pattern of ignorePatterns) {
try {
const glob = new Bun.Glob(pattern);
if (glob.match(base)) {
return true;
}
} catch (err) {
// Skip invalid patterns
const glob = new Bun.Glob(pattern);
if (glob.match(base)) {
return true;
}
}
@@ -272,7 +265,7 @@ function printTree(tree: TreeNode[], rootPath: string): string {
let result = `- ${rootPath}${path.sep}\n`;
for (const node of tree) {
printNode(node, 1, result);
result = printNode(node, 1, result);
}
return result;
@@ -296,4 +289,3 @@ function printNode(node: TreeNode, level: number, result: string): string {
return result;
}