1use std::rc::Rc;
2use std::sync::Arc;
3use std::{borrow::Cow, cell::RefCell};
4
5use crate::schema::json_schema_for;
6use anyhow::{Context as _, Result, anyhow, bail};
7use assistant_tool::{ActionLog, Tool, ToolResult};
8use futures::AsyncReadExt as _;
9use gpui::{AnyWindowHandle, App, AppContext as _, Entity, Task};
10use html_to_markdown::{TagHandler, convert_html_to_markdown, markdown};
11use http_client::{AsyncBody, HttpClientWithUrl};
12use language_model::{LanguageModel, LanguageModelRequest, LanguageModelToolSchemaFormat};
13use project::Project;
14use schemars::JsonSchema;
15use serde::{Deserialize, Serialize};
16use ui::IconName;
17use util::markdown::MarkdownEscaped;
18
19#[derive(Debug, PartialEq, Eq, PartialOrd, Ord, Hash, Clone, Copy)]
20enum ContentType {
21 Html,
22 Plaintext,
23 Json,
24}
25
26#[derive(Debug, Serialize, Deserialize, JsonSchema)]
27pub struct FetchToolInput {
28 /// The URL to fetch.
29 url: String,
30}
31
32pub struct FetchTool {
33 http_client: Arc<HttpClientWithUrl>,
34}
35
36impl FetchTool {
37 pub fn new(http_client: Arc<HttpClientWithUrl>) -> Self {
38 Self { http_client }
39 }
40
41 async fn build_message(http_client: Arc<HttpClientWithUrl>, url: &str) -> Result<String> {
42 let url = if !url.starts_with("https://") && !url.starts_with("http://") {
43 Cow::Owned(format!("https://{url}"))
44 } else {
45 Cow::Borrowed(url)
46 };
47
48 let mut response = http_client.get(&url, AsyncBody::default(), true).await?;
49
50 let mut body = Vec::new();
51 response
52 .body_mut()
53 .read_to_end(&mut body)
54 .await
55 .context("error reading response body")?;
56
57 if response.status().is_client_error() {
58 let text = String::from_utf8_lossy(body.as_slice());
59 bail!(
60 "status error {}, response: {text:?}",
61 response.status().as_u16()
62 );
63 }
64
65 let Some(content_type) = response.headers().get("content-type") else {
66 bail!("missing Content-Type header");
67 };
68 let content_type = content_type
69 .to_str()
70 .context("invalid Content-Type header")?;
71 let content_type = match content_type {
72 "text/html" => ContentType::Html,
73 "text/plain" => ContentType::Plaintext,
74 "application/json" => ContentType::Json,
75 _ => ContentType::Html,
76 };
77
78 match content_type {
79 ContentType::Html => {
80 let mut handlers: Vec<TagHandler> = vec![
81 Rc::new(RefCell::new(markdown::WebpageChromeRemover)),
82 Rc::new(RefCell::new(markdown::ParagraphHandler)),
83 Rc::new(RefCell::new(markdown::HeadingHandler)),
84 Rc::new(RefCell::new(markdown::ListHandler)),
85 Rc::new(RefCell::new(markdown::TableHandler::new())),
86 Rc::new(RefCell::new(markdown::StyledTextHandler)),
87 ];
88 if url.contains("wikipedia.org") {
89 use html_to_markdown::structure::wikipedia;
90
91 handlers.push(Rc::new(RefCell::new(wikipedia::WikipediaChromeRemover)));
92 handlers.push(Rc::new(RefCell::new(wikipedia::WikipediaInfoboxHandler)));
93 handlers.push(Rc::new(
94 RefCell::new(wikipedia::WikipediaCodeHandler::new()),
95 ));
96 } else {
97 handlers.push(Rc::new(RefCell::new(markdown::CodeHandler)));
98 }
99
100 convert_html_to_markdown(&body[..], &mut handlers)
101 }
102 ContentType::Plaintext => Ok(std::str::from_utf8(&body)?.to_owned()),
103 ContentType::Json => {
104 let json: serde_json::Value = serde_json::from_slice(&body)?;
105
106 Ok(format!(
107 "```json\n{}\n```",
108 serde_json::to_string_pretty(&json)?
109 ))
110 }
111 }
112 }
113}
114
115impl Tool for FetchTool {
116 fn name(&self) -> String {
117 "fetch".to_string()
118 }
119
120 fn needs_confirmation(&self, _: &serde_json::Value, _: &App) -> bool {
121 true
122 }
123
124 fn description(&self) -> String {
125 include_str!("./fetch_tool/description.md").to_string()
126 }
127
128 fn icon(&self) -> IconName {
129 IconName::Globe
130 }
131
132 fn input_schema(&self, format: LanguageModelToolSchemaFormat) -> Result<serde_json::Value> {
133 json_schema_for::<FetchToolInput>(format)
134 }
135
136 fn ui_text(&self, input: &serde_json::Value) -> String {
137 match serde_json::from_value::<FetchToolInput>(input.clone()) {
138 Ok(input) => format!("Fetch {}", MarkdownEscaped(&input.url)),
139 Err(_) => "Fetch URL".to_string(),
140 }
141 }
142
143 fn run(
144 self: Arc<Self>,
145 input: serde_json::Value,
146 _request: Arc<LanguageModelRequest>,
147 _project: Entity<Project>,
148 _action_log: Entity<ActionLog>,
149 _model: Arc<dyn LanguageModel>,
150 _window: Option<AnyWindowHandle>,
151 cx: &mut App,
152 ) -> ToolResult {
153 let input = match serde_json::from_value::<FetchToolInput>(input) {
154 Ok(input) => input,
155 Err(err) => return Task::ready(Err(anyhow!(err))).into(),
156 };
157
158 let text = cx.background_spawn({
159 let http_client = self.http_client.clone();
160 async move { Self::build_message(http_client, &input.url).await }
161 });
162
163 cx.foreground_executor()
164 .spawn(async move {
165 let text = text.await?;
166 if text.trim().is_empty() {
167 bail!("no textual content found");
168 }
169
170 Ok(text.into())
171 })
172 .into()
173 }
174}