mirror of
https://github.com/LemmyNet/lemmy.git
synced 2024-05-10 14:02:39 +00:00
a2a594b763
* Create example for apub lib * some rewriting of apub lib * Add LocalInstance struct for apub lib to avoid using Lemmy Settings * Move ActorType trait to lemmy_apub, because its not needed in library * Use reqwest_retry instead of custom impl, dont specify timeout on every send() * Some improvements to example * Moved inbox handling to library * bug fixes * Move context and serde helpers into library * wip: example changes * Add lemmy_utils feature to build only LemmyError * Rename to activitypub_federation * Remove lemmy_utils dep from activitypub_federation using generic error type * Finish activitypub example * Cleanup and fix tests * Reorganize library files * Remove ApubObject.to_tombstone() * Extract activitypub library into separate git repository
253 lines
7.4 KiB
Rust
253 lines
7.4 KiB
Rust
use crate::post::SiteMetadata;
|
|
use encoding::{all::encodings, DecoderTrap};
|
|
use lemmy_utils::{error::LemmyError, settings::structs::Settings, version::VERSION};
|
|
use percent_encoding::{utf8_percent_encode, NON_ALPHANUMERIC};
|
|
use reqwest_middleware::ClientWithMiddleware;
|
|
use serde::Deserialize;
|
|
use tracing::info;
|
|
use url::Url;
|
|
use webpage::HTML;
|
|
|
|
/// Fetches the post link html tags (like title, description, image, etc)
|
|
#[tracing::instrument(skip_all)]
|
|
pub async fn fetch_site_metadata(
|
|
client: &ClientWithMiddleware,
|
|
url: &Url,
|
|
) -> Result<SiteMetadata, LemmyError> {
|
|
info!("Fetching site metadata for url: {}", url);
|
|
let response = client.get(url.as_str()).send().await?;
|
|
|
|
// Can't use .text() here, because it only checks the content header, not the actual bytes
|
|
// https://github.com/LemmyNet/lemmy/issues/1964
|
|
let html_bytes = response.bytes().await.map_err(LemmyError::from)?.to_vec();
|
|
|
|
let tags = html_to_site_metadata(&html_bytes)?;
|
|
|
|
Ok(tags)
|
|
}
|
|
|
|
fn html_to_site_metadata(html_bytes: &[u8]) -> Result<SiteMetadata, LemmyError> {
|
|
let html = String::from_utf8_lossy(html_bytes);
|
|
|
|
// Make sure the first line is doctype html
|
|
let first_line = html
|
|
.trim_start()
|
|
.lines()
|
|
.into_iter()
|
|
.next()
|
|
.ok_or_else(|| LemmyError::from_message("No lines in html"))?
|
|
.to_lowercase();
|
|
|
|
if !first_line.starts_with("<!doctype html>") {
|
|
return Err(LemmyError::from_message(
|
|
"Site metadata page fetch is not DOCTYPE html",
|
|
));
|
|
}
|
|
|
|
let mut page = HTML::from_string(html.to_string(), None)?;
|
|
|
|
// If the web page specifies that it isn't actually UTF-8, re-decode the received bytes with the
|
|
// proper encoding. If the specified encoding cannot be found, fall back to the original UTF-8
|
|
// version.
|
|
if let Some(charset) = page.meta.get("charset") {
|
|
if charset.to_lowercase() != "utf-8" {
|
|
if let Some(encoding_ref) = encodings().iter().find(|e| e.name() == charset) {
|
|
if let Ok(html_with_encoding) = encoding_ref.decode(html_bytes, DecoderTrap::Replace) {
|
|
page = HTML::from_string(html_with_encoding, None)?;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
let page_title = page.title;
|
|
let page_description = page.description;
|
|
|
|
let og_description = page
|
|
.opengraph
|
|
.properties
|
|
.get("description")
|
|
.map(|t| t.to_string());
|
|
let og_title = page
|
|
.opengraph
|
|
.properties
|
|
.get("title")
|
|
.map(|t| t.to_string());
|
|
let og_image = page
|
|
.opengraph
|
|
.images
|
|
.get(0)
|
|
.and_then(|ogo| Url::parse(&ogo.url).ok());
|
|
|
|
let title = og_title.or(page_title);
|
|
let description = og_description.or(page_description);
|
|
let image = og_image;
|
|
|
|
Ok(SiteMetadata {
|
|
title,
|
|
description,
|
|
image,
|
|
html: None,
|
|
})
|
|
}
|
|
|
|
#[derive(Deserialize, Debug, Clone)]
|
|
pub(crate) struct PictrsResponse {
|
|
files: Vec<PictrsFile>,
|
|
msg: String,
|
|
}
|
|
|
|
#[derive(Deserialize, Debug, Clone)]
|
|
pub(crate) struct PictrsFile {
|
|
file: String,
|
|
#[allow(dead_code)]
|
|
delete_token: String,
|
|
}
|
|
|
|
#[tracing::instrument(skip_all)]
|
|
pub(crate) async fn fetch_pictrs(
|
|
client: &ClientWithMiddleware,
|
|
settings: &Settings,
|
|
image_url: &Url,
|
|
) -> Result<PictrsResponse, LemmyError> {
|
|
if let Some(pictrs_url) = settings.pictrs_url.to_owned() {
|
|
is_image_content_type(client, image_url).await?;
|
|
|
|
let fetch_url = format!(
|
|
"{}/image/download?url={}",
|
|
pictrs_url,
|
|
utf8_percent_encode(image_url.as_str(), NON_ALPHANUMERIC) // TODO this might not be needed
|
|
);
|
|
|
|
let response = client.get(&fetch_url).send().await?;
|
|
|
|
let response: PictrsResponse = response.json().await.map_err(LemmyError::from)?;
|
|
|
|
if response.msg == "ok" {
|
|
Ok(response)
|
|
} else {
|
|
Err(LemmyError::from_message(&response.msg))
|
|
}
|
|
} else {
|
|
Err(LemmyError::from_message("pictrs_url not set up in config"))
|
|
}
|
|
}
|
|
|
|
/// Both are options, since the URL might be either an html page, or an image
|
|
/// Returns the SiteMetadata, and a Pictrs URL, if there is a picture associated
|
|
#[tracing::instrument(skip_all)]
|
|
pub async fn fetch_site_data(
|
|
client: &ClientWithMiddleware,
|
|
settings: &Settings,
|
|
url: Option<&Url>,
|
|
) -> (Option<SiteMetadata>, Option<Url>) {
|
|
match &url {
|
|
Some(url) => {
|
|
// Fetch metadata
|
|
// Ignore errors, since it may be an image, or not have the data.
|
|
// Warning, this may ignore SSL errors
|
|
let metadata_option = fetch_site_metadata(client, url).await.ok();
|
|
|
|
// Fetch pictrs thumbnail
|
|
let pictrs_hash = match &metadata_option {
|
|
Some(metadata_res) => match &metadata_res.image {
|
|
// Metadata, with image
|
|
// Try to generate a small thumbnail if there's a full sized one from post-links
|
|
Some(metadata_image) => fetch_pictrs(client, settings, metadata_image)
|
|
.await
|
|
.map(|r| r.files[0].file.to_owned()),
|
|
// Metadata, but no image
|
|
None => fetch_pictrs(client, settings, url)
|
|
.await
|
|
.map(|r| r.files[0].file.to_owned()),
|
|
},
|
|
// No metadata, try to fetch the URL as an image
|
|
None => fetch_pictrs(client, settings, url)
|
|
.await
|
|
.map(|r| r.files[0].file.to_owned()),
|
|
};
|
|
|
|
// The full urls are necessary for federation
|
|
let pictrs_thumbnail = pictrs_hash
|
|
.map(|p| {
|
|
Url::parse(&format!(
|
|
"{}/pictrs/image/{}",
|
|
settings.get_protocol_and_hostname(),
|
|
p
|
|
))
|
|
.ok()
|
|
})
|
|
.ok()
|
|
.flatten();
|
|
|
|
(metadata_option, pictrs_thumbnail)
|
|
}
|
|
None => (None, None),
|
|
}
|
|
}
|
|
|
|
#[tracing::instrument(skip_all)]
|
|
async fn is_image_content_type(client: &ClientWithMiddleware, url: &Url) -> Result<(), LemmyError> {
|
|
let response = client.get(url.as_str()).send().await?;
|
|
if response
|
|
.headers()
|
|
.get("Content-Type")
|
|
.ok_or_else(|| LemmyError::from_message("No Content-Type header"))?
|
|
.to_str()?
|
|
.starts_with("image/")
|
|
{
|
|
Ok(())
|
|
} else {
|
|
Err(LemmyError::from_message("Not an image type."))
|
|
}
|
|
}
|
|
|
|
pub fn build_user_agent(settings: &Settings) -> String {
|
|
format!(
|
|
"Lemmy/{}; +{}",
|
|
VERSION,
|
|
settings.get_protocol_and_hostname()
|
|
)
|
|
}
|
|
|
|
#[cfg(test)]
|
|
mod tests {
|
|
use crate::request::{build_user_agent, fetch_site_metadata, SiteMetadata};
|
|
use lemmy_utils::settings::structs::Settings;
|
|
use url::Url;
|
|
|
|
// These helped with testing
|
|
#[actix_rt::test]
|
|
async fn test_site_metadata() {
|
|
let settings = Settings::init().unwrap();
|
|
let client = reqwest::Client::builder()
|
|
.user_agent(build_user_agent(&settings))
|
|
.build()
|
|
.unwrap()
|
|
.into();
|
|
let sample_url = Url::parse("https://gitlab.com/IzzyOnDroid/repo/-/wikis/FAQ").unwrap();
|
|
let sample_res = fetch_site_metadata(&client, &sample_url).await.unwrap();
|
|
assert_eq!(
|
|
SiteMetadata {
|
|
title: Some("FAQ · Wiki · IzzyOnDroid / repo · GitLab".to_string()),
|
|
description: Some(
|
|
"The F-Droid compatible repo at https://apt.izzysoft.de/fdroid/".to_string()
|
|
),
|
|
image: Some(
|
|
Url::parse("https://gitlab.com/uploads/-/system/project/avatar/4877469/iod_logo.png")
|
|
.unwrap()
|
|
),
|
|
html: None,
|
|
},
|
|
sample_res
|
|
);
|
|
}
|
|
|
|
// #[test]
|
|
// fn test_pictshare() {
|
|
// let res = fetch_pictshare("https://upload.wikimedia.org/wikipedia/en/2/27/The_Mandalorian_logo.jpg");
|
|
// assert!(res.is_ok());
|
|
// let res_other = fetch_pictshare("https://upload.wikimedia.org/wikipedia/en/2/27/The_Mandalorian_logo.jpgaoeu");
|
|
// assert!(res_other.is_err());
|
|
// }
|
|
}
|