assistant_tools: Add fetch tool (#26999)

This PR adds a new `fetch` tool to the set of tools the Assistant has
available.

This tool accepts a URL and fetches the content as Markdown.

<img width="1394" alt="Screenshot 2025-03-18 at 11 52 21 AM"
src="https://github.com/user-attachments/assets/e5bcde14-a0dd-4835-9d42-8f45def68f4d"
/>

<img width="1394" alt="Screenshot 2025-03-18 at 11 52 37 AM"
src="https://github.com/user-attachments/assets/3bcce4f5-f61b-40d7-8b30-2c673ce3c06a"
/>

Release Notes:

- N/A
This commit is contained in:
Marshall Bowers 2025-03-18 12:25:51 -04:00 committed by GitHub
parent baaafddbeb
commit ed4e654fdf
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
7 changed files with 170 additions and 6 deletions

2
Cargo.lock generated
View File

@ -715,6 +715,8 @@ dependencies = [
"feature_flags",
"futures 0.3.31",
"gpui",
"html_to_markdown",
"http_client",
"itertools 0.14.0",
"language",
"language_model",

View File

@ -163,7 +163,7 @@ pub fn init(cx: &mut App) -> Arc<HeadlessAppState> {
language::init(cx);
language_model::init(client.clone(), cx);
language_models::init(user_store.clone(), client.clone(), fs.clone(), cx);
assistant_tools::init(cx);
assistant_tools::init(client.http_client().clone(), cx);
context_server::init(cx);
let stdout_is_a_pty = false;
let prompt_builder = PromptBuilder::load(fs.clone(), stdout_is_a_pty, cx);

View File

@ -18,8 +18,10 @@ chrono.workspace = true
collections.workspace = true
feature_flags.workspace = true
futures.workspace = true
itertools.workspace = true
gpui.workspace = true
html_to_markdown.workspace = true
http_client.workspace = true
itertools.workspace = true
language.workspace = true
language_model.workspace = true
project.workspace = true
@ -27,17 +29,17 @@ release_channel.workspace = true
schemars.workspace = true
serde.workspace = true
serde_json.workspace = true
settings.workspace = true
theme.workspace = true
ui.workspace = true
util.workspace = true
workspace.workspace = true
worktree.workspace = true
settings.workspace = true
[dev-dependencies]
rand.workspace = true
collections = { workspace = true, features = ["test-support"] }
gpui = { workspace = true, features = ["test-support"] }
language = { workspace = true, features = ["test-support"] }
project = { workspace = true, features = ["test-support"] }
rand.workspace = true
workspace = { workspace = true, features = ["test-support"] }

View File

@ -2,6 +2,7 @@ mod bash_tool;
mod delete_path_tool;
mod diagnostics_tool;
mod edit_files_tool;
mod fetch_tool;
mod list_directory_tool;
mod now_tool;
mod path_search_tool;
@ -9,13 +10,17 @@ mod read_file_tool;
mod regex_search_tool;
mod thinking_tool;
use std::sync::Arc;
use assistant_tool::ToolRegistry;
use gpui::App;
use http_client::HttpClientWithUrl;
use crate::bash_tool::BashTool;
use crate::delete_path_tool::DeletePathTool;
use crate::diagnostics_tool::DiagnosticsTool;
use crate::edit_files_tool::EditFilesTool;
use crate::fetch_tool::FetchTool;
use crate::list_directory_tool::ListDirectoryTool;
use crate::now_tool::NowTool;
use crate::path_search_tool::PathSearchTool;
@ -23,7 +28,7 @@ use crate::read_file_tool::ReadFileTool;
use crate::regex_search_tool::RegexSearchTool;
use crate::thinking_tool::ThinkingTool;
pub fn init(cx: &mut App) {
pub fn init(http_client: Arc<HttpClientWithUrl>, cx: &mut App) {
assistant_tool::init(cx);
crate::edit_files_tool::log::init(cx);
@ -38,4 +43,5 @@ pub fn init(cx: &mut App) {
registry.register_tool(ReadFileTool);
registry.register_tool(RegexSearchTool);
registry.register_tool(ThinkingTool);
registry.register_tool(FetchTool::new(http_client));
}

View File

@ -0,0 +1,153 @@
use std::cell::RefCell;
use std::rc::Rc;
use std::sync::Arc;
use anyhow::{anyhow, bail, Context as _, Result};
use assistant_tool::{ActionLog, Tool};
use futures::AsyncReadExt as _;
use gpui::{App, AppContext as _, Entity, Task};
use html_to_markdown::{convert_html_to_markdown, markdown, TagHandler};
use http_client::{AsyncBody, HttpClientWithUrl};
use language_model::LanguageModelRequestMessage;
use project::Project;
use schemars::JsonSchema;
use serde::{Deserialize, Serialize};
#[derive(Debug, PartialEq, Eq, PartialOrd, Ord, Hash, Clone, Copy)]
enum ContentType {
Html,
Plaintext,
Json,
}
#[derive(Debug, Serialize, Deserialize, JsonSchema)]
pub struct FetchToolInput {
/// The URL to fetch.
url: String,
}
pub struct FetchTool {
http_client: Arc<HttpClientWithUrl>,
}
impl FetchTool {
pub fn new(http_client: Arc<HttpClientWithUrl>) -> Self {
Self { http_client }
}
async fn build_message(http_client: Arc<HttpClientWithUrl>, url: &str) -> Result<String> {
let mut url = url.to_owned();
if !url.starts_with("https://") && !url.starts_with("http://") {
url = format!("https://{url}");
}
let mut response = http_client.get(&url, AsyncBody::default(), true).await?;
let mut body = Vec::new();
response
.body_mut()
.read_to_end(&mut body)
.await
.context("error reading response body")?;
if response.status().is_client_error() {
let text = String::from_utf8_lossy(body.as_slice());
bail!(
"status error {}, response: {text:?}",
response.status().as_u16()
);
}
let Some(content_type) = response.headers().get("content-type") else {
bail!("missing Content-Type header");
};
let content_type = content_type
.to_str()
.context("invalid Content-Type header")?;
let content_type = match content_type {
"text/html" => ContentType::Html,
"text/plain" => ContentType::Plaintext,
"application/json" => ContentType::Json,
_ => ContentType::Html,
};
match content_type {
ContentType::Html => {
let mut handlers: Vec<TagHandler> = vec![
Rc::new(RefCell::new(markdown::WebpageChromeRemover)),
Rc::new(RefCell::new(markdown::ParagraphHandler)),
Rc::new(RefCell::new(markdown::HeadingHandler)),
Rc::new(RefCell::new(markdown::ListHandler)),
Rc::new(RefCell::new(markdown::TableHandler::new())),
Rc::new(RefCell::new(markdown::StyledTextHandler)),
];
if url.contains("wikipedia.org") {
use html_to_markdown::structure::wikipedia;
handlers.push(Rc::new(RefCell::new(wikipedia::WikipediaChromeRemover)));
handlers.push(Rc::new(RefCell::new(wikipedia::WikipediaInfoboxHandler)));
handlers.push(Rc::new(
RefCell::new(wikipedia::WikipediaCodeHandler::new()),
));
} else {
handlers.push(Rc::new(RefCell::new(markdown::CodeHandler)));
}
convert_html_to_markdown(&body[..], &mut handlers)
}
ContentType::Plaintext => Ok(std::str::from_utf8(&body)?.to_owned()),
ContentType::Json => {
let json: serde_json::Value = serde_json::from_slice(&body)?;
Ok(format!(
"```json\n{}\n```",
serde_json::to_string_pretty(&json)?
))
}
}
}
}
impl Tool for FetchTool {
fn name(&self) -> String {
"fetch".to_string()
}
fn description(&self) -> String {
include_str!("./fetch_tool/description.md").to_string()
}
fn input_schema(&self) -> serde_json::Value {
let schema = schemars::schema_for!(FetchToolInput);
serde_json::to_value(&schema).unwrap()
}
fn run(
self: Arc<Self>,
input: serde_json::Value,
_messages: &[LanguageModelRequestMessage],
_project: Entity<Project>,
_action_log: Entity<ActionLog>,
cx: &mut App,
) -> Task<Result<String>> {
let input = match serde_json::from_value::<FetchToolInput>(input) {
Ok(input) => input,
Err(err) => return Task::ready(Err(anyhow!(err))),
};
let text = cx.background_spawn({
let http_client = self.http_client.clone();
let url = input.url.clone();
async move { Self::build_message(http_client, &url).await }
});
cx.foreground_executor().spawn(async move {
let text = text.await?;
if text.trim().is_empty() {
bail!("no textual content found");
}
Ok(text)
})
}
}

View File

@ -0,0 +1 @@
Fetches a URL and returns the content as Markdown.

View File

@ -472,7 +472,7 @@ fn main() {
prompt_builder.clone(),
cx,
);
assistant_tools::init(cx);
assistant_tools::init(app_state.client.http_client(), cx);
repl::init(app_state.fs.clone(), cx);
extension_host::init(
extension_host_proxy,