2023-08-28 10:23:45 +00:00
|
|
|
use crate::{context::LemmyContext, post::SiteMetadata};
|
2022-05-03 17:44:13 +00:00
|
|
|
use encoding::{all::encodings, DecoderTrap};
|
2022-06-02 21:44:47 +00:00
|
|
|
use lemmy_db_schema::newtypes::DbUrl;
|
2022-06-13 19:15:04 +00:00
|
|
|
use lemmy_utils::{
|
2023-07-10 14:50:07 +00:00
|
|
|
error::{LemmyError, LemmyErrorType},
|
2022-06-13 19:15:04 +00:00
|
|
|
settings::structs::Settings,
|
|
|
|
|
version::VERSION,
|
|
|
|
|
REQWEST_TIMEOUT,
|
|
|
|
|
};
|
2022-05-03 17:44:13 +00:00
|
|
|
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);
|
2022-06-02 14:33:41 +00:00
|
|
|
let response = client.get(url.as_str()).send().await?;
|
2022-05-03 17:44:13 +00:00
|
|
|
|
|
|
|
|
// 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();
|
|
|
|
|
|
2023-06-26 13:07:57 +00:00
|
|
|
let tags = html_to_site_metadata(&html_bytes, url)?;
|
2022-05-03 17:44:13 +00:00
|
|
|
|
|
|
|
|
Ok(tags)
|
|
|
|
|
}
|
|
|
|
|
|
2023-06-26 13:07:57 +00:00
|
|
|
fn html_to_site_metadata(html_bytes: &[u8], url: &Url) -> Result<SiteMetadata, LemmyError> {
|
2022-05-03 17:44:13 +00:00
|
|
|
let html = String::from_utf8_lossy(html_bytes);
|
|
|
|
|
|
|
|
|
|
// Make sure the first line is doctype html
|
|
|
|
|
let first_line = html
|
|
|
|
|
.trim_start()
|
|
|
|
|
.lines()
|
|
|
|
|
.next()
|
2023-07-10 14:50:07 +00:00
|
|
|
.ok_or(LemmyErrorType::NoLinesInHtml)?
|
2022-05-03 17:44:13 +00:00
|
|
|
.to_lowercase();
|
|
|
|
|
|
|
|
|
|
if !first_line.starts_with("<!doctype html>") {
|
2023-08-31 13:01:08 +00:00
|
|
|
Err(LemmyErrorType::SiteMetadataPageIsNotDoctypeHtml)?
|
2022-05-03 17:44:13 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
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")
|
2022-11-19 04:33:54 +00:00
|
|
|
.map(std::string::ToString::to_string);
|
2022-05-03 17:44:13 +00:00
|
|
|
let og_title = page
|
|
|
|
|
.opengraph
|
|
|
|
|
.properties
|
|
|
|
|
.get("title")
|
2022-11-19 04:33:54 +00:00
|
|
|
.map(std::string::ToString::to_string);
|
2022-05-03 17:44:13 +00:00
|
|
|
let og_image = page
|
|
|
|
|
.opengraph
|
|
|
|
|
.images
|
2023-02-28 11:34:50 +00:00
|
|
|
.first()
|
2023-06-26 13:07:57 +00:00
|
|
|
// join also works if the target URL is absolute
|
|
|
|
|
.and_then(|ogo| url.join(&ogo.url).ok());
|
2022-06-02 21:44:47 +00:00
|
|
|
let og_embed_url = page
|
|
|
|
|
.opengraph
|
|
|
|
|
.videos
|
|
|
|
|
.first()
|
2023-06-26 13:07:57 +00:00
|
|
|
// join also works if the target URL is absolute
|
|
|
|
|
.and_then(|v| url.join(&v.url).ok());
|
2022-05-03 17:44:13 +00:00
|
|
|
|
|
|
|
|
Ok(SiteMetadata {
|
2022-06-02 21:44:47 +00:00
|
|
|
title: og_title.or(page_title),
|
|
|
|
|
description: og_description.or(page_description),
|
|
|
|
|
image: og_image.map(Into::into),
|
|
|
|
|
embed_video_url: og_embed_url.map(Into::into),
|
2022-05-03 17:44:13 +00:00
|
|
|
})
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#[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,
|
|
|
|
|
}
|
|
|
|
|
|
2022-06-13 19:15:04 +00:00
|
|
|
#[derive(Deserialize, Debug, Clone)]
|
|
|
|
|
pub(crate) struct PictrsPurgeResponse {
|
|
|
|
|
msg: String,
|
|
|
|
|
}
|
|
|
|
|
|
2022-05-03 17:44:13 +00:00
|
|
|
#[tracing::instrument(skip_all)]
|
|
|
|
|
pub(crate) async fn fetch_pictrs(
|
|
|
|
|
client: &ClientWithMiddleware,
|
|
|
|
|
settings: &Settings,
|
|
|
|
|
image_url: &Url,
|
|
|
|
|
) -> Result<PictrsResponse, LemmyError> {
|
2022-06-13 19:15:04 +00:00
|
|
|
let pictrs_config = settings.pictrs_config()?;
|
|
|
|
|
is_image_content_type(client, image_url).await?;
|
2022-05-03 17:44:13 +00:00
|
|
|
|
2022-06-13 19:15:04 +00:00
|
|
|
let fetch_url = format!(
|
2022-08-16 15:20:41 +00:00
|
|
|
"{}image/download?url={}",
|
2022-06-13 19:15:04 +00:00
|
|
|
pictrs_config.url,
|
|
|
|
|
utf8_percent_encode(image_url.as_str(), NON_ALPHANUMERIC) // TODO this might not be needed
|
|
|
|
|
);
|
2022-05-03 17:44:13 +00:00
|
|
|
|
2022-06-13 19:15:04 +00:00
|
|
|
let response = client
|
|
|
|
|
.get(&fetch_url)
|
|
|
|
|
.timeout(REQWEST_TIMEOUT)
|
|
|
|
|
.send()
|
|
|
|
|
.await?;
|
2022-05-03 17:44:13 +00:00
|
|
|
|
2022-06-13 19:15:04 +00:00
|
|
|
let response: PictrsResponse = response.json().await.map_err(LemmyError::from)?;
|
2022-05-03 17:44:13 +00:00
|
|
|
|
2022-06-13 19:15:04 +00:00
|
|
|
if response.msg == "ok" {
|
|
|
|
|
Ok(response)
|
|
|
|
|
} else {
|
2023-07-10 14:50:07 +00:00
|
|
|
Err(LemmyErrorType::PictrsResponseError(response.msg))?
|
2022-06-13 19:15:04 +00:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// Purges an image from pictrs
|
|
|
|
|
/// Note: This should often be coerced from a Result to .ok() in order to fail softly, because:
|
|
|
|
|
/// - It might fail due to image being not local
|
|
|
|
|
/// - It might not be an image
|
|
|
|
|
/// - Pictrs might not be set up
|
|
|
|
|
pub async fn purge_image_from_pictrs(
|
|
|
|
|
image_url: &Url,
|
2023-08-28 10:23:45 +00:00
|
|
|
context: &LemmyContext,
|
2022-06-13 19:15:04 +00:00
|
|
|
) -> Result<(), LemmyError> {
|
2023-08-28 10:23:45 +00:00
|
|
|
let pictrs_config = context.settings().pictrs_config()?;
|
|
|
|
|
is_image_content_type(context.client(), image_url).await?;
|
2022-06-13 19:15:04 +00:00
|
|
|
|
|
|
|
|
let alias = image_url
|
|
|
|
|
.path_segments()
|
2023-07-10 14:50:07 +00:00
|
|
|
.ok_or(LemmyErrorType::ImageUrlMissingPathSegments)?
|
2022-06-13 19:15:04 +00:00
|
|
|
.next_back()
|
2023-07-10 14:50:07 +00:00
|
|
|
.ok_or(LemmyErrorType::ImageUrlMissingLastPathSegment)?;
|
2022-06-13 19:15:04 +00:00
|
|
|
|
|
|
|
|
let purge_url = format!("{}/internal/purge?alias={}", pictrs_config.url, alias);
|
|
|
|
|
|
2022-07-11 20:38:37 +00:00
|
|
|
let pictrs_api_key = pictrs_config
|
|
|
|
|
.api_key
|
2023-07-10 14:50:07 +00:00
|
|
|
.ok_or(LemmyErrorType::PictrsApiKeyNotProvided)?;
|
2023-08-28 10:23:45 +00:00
|
|
|
let response = context
|
|
|
|
|
.client()
|
2022-06-13 19:15:04 +00:00
|
|
|
.post(&purge_url)
|
|
|
|
|
.timeout(REQWEST_TIMEOUT)
|
2022-07-11 20:38:37 +00:00
|
|
|
.header("x-api-token", pictrs_api_key)
|
2022-06-13 19:15:04 +00:00
|
|
|
.send()
|
|
|
|
|
.await?;
|
|
|
|
|
|
|
|
|
|
let response: PictrsPurgeResponse = response.json().await.map_err(LemmyError::from)?;
|
|
|
|
|
|
|
|
|
|
if response.msg == "ok" {
|
|
|
|
|
Ok(())
|
2022-05-03 17:44:13 +00:00
|
|
|
} else {
|
2023-07-10 14:50:07 +00:00
|
|
|
Err(LemmyErrorType::PictrsPurgeResponseError(response.msg))?
|
2022-05-03 17:44:13 +00:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// 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>,
|
2023-06-30 10:42:42 +00:00
|
|
|
include_image: bool,
|
2022-06-02 21:44:47 +00:00
|
|
|
) -> (Option<SiteMetadata>, Option<DbUrl>) {
|
2022-05-03 17:44:13 +00:00
|
|
|
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();
|
2023-06-30 10:42:42 +00:00
|
|
|
if !include_image {
|
|
|
|
|
return (metadata_option, None);
|
|
|
|
|
}
|
2022-05-03 17:44:13 +00:00
|
|
|
|
2023-02-28 11:34:50 +00:00
|
|
|
let missing_pictrs_file =
|
|
|
|
|
|r: PictrsResponse| r.files.first().expect("missing pictrs file").file.clone();
|
|
|
|
|
|
2022-05-03 17:44:13 +00:00
|
|
|
// 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
|
2023-02-28 11:34:50 +00:00
|
|
|
.map(missing_pictrs_file),
|
2022-05-03 17:44:13 +00:00
|
|
|
// Metadata, but no image
|
|
|
|
|
None => fetch_pictrs(client, settings, url)
|
|
|
|
|
.await
|
2023-02-28 11:34:50 +00:00
|
|
|
.map(missing_pictrs_file),
|
2022-05-03 17:44:13 +00:00
|
|
|
},
|
|
|
|
|
// No metadata, try to fetch the URL as an image
|
|
|
|
|
None => fetch_pictrs(client, settings, url)
|
|
|
|
|
.await
|
2023-02-28 11:34:50 +00:00
|
|
|
.map(missing_pictrs_file),
|
2022-05-03 17:44:13 +00:00
|
|
|
};
|
|
|
|
|
|
|
|
|
|
// 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();
|
|
|
|
|
|
2022-06-02 21:44:47 +00:00
|
|
|
(metadata_option, pictrs_thumbnail.map(Into::into))
|
2022-05-03 17:44:13 +00:00
|
|
|
}
|
|
|
|
|
None => (None, None),
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#[tracing::instrument(skip_all)]
|
|
|
|
|
async fn is_image_content_type(client: &ClientWithMiddleware, url: &Url) -> Result<(), LemmyError> {
|
2022-06-02 14:33:41 +00:00
|
|
|
let response = client.get(url.as_str()).send().await?;
|
2022-05-03 17:44:13 +00:00
|
|
|
if response
|
|
|
|
|
.headers()
|
|
|
|
|
.get("Content-Type")
|
2023-07-10 14:50:07 +00:00
|
|
|
.ok_or(LemmyErrorType::NoContentTypeHeader)?
|
2022-05-03 17:44:13 +00:00
|
|
|
.to_str()?
|
|
|
|
|
.starts_with("image/")
|
|
|
|
|
{
|
|
|
|
|
Ok(())
|
|
|
|
|
} else {
|
2023-07-10 14:50:07 +00:00
|
|
|
Err(LemmyErrorType::NotAnImageType)?
|
2022-05-03 17:44:13 +00:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
pub fn build_user_agent(settings: &Settings) -> String {
|
|
|
|
|
format!(
|
|
|
|
|
"Lemmy/{}; +{}",
|
|
|
|
|
VERSION,
|
|
|
|
|
settings.get_protocol_and_hostname()
|
|
|
|
|
)
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#[cfg(test)]
|
|
|
|
|
mod tests {
|
2023-07-17 15:04:14 +00:00
|
|
|
#![allow(clippy::unwrap_used)]
|
|
|
|
|
#![allow(clippy::indexing_slicing)]
|
|
|
|
|
|
2023-06-26 13:07:57 +00:00
|
|
|
use crate::request::{
|
|
|
|
|
build_user_agent,
|
|
|
|
|
fetch_site_metadata,
|
|
|
|
|
html_to_site_metadata,
|
|
|
|
|
SiteMetadata,
|
|
|
|
|
};
|
2022-06-22 20:24:54 +00:00
|
|
|
use lemmy_utils::settings::SETTINGS;
|
2022-05-03 17:44:13 +00:00
|
|
|
use url::Url;
|
|
|
|
|
|
|
|
|
|
// These helped with testing
|
2023-06-26 08:24:11 +00:00
|
|
|
#[tokio::test]
|
2022-05-03 17:44:13 +00:00
|
|
|
async fn test_site_metadata() {
|
2022-11-19 04:33:54 +00:00
|
|
|
let settings = &SETTINGS.clone();
|
2022-05-03 17:44:13 +00:00
|
|
|
let client = reqwest::Client::builder()
|
2022-06-22 20:24:54 +00:00
|
|
|
.user_agent(build_user_agent(settings))
|
2022-05-03 17:44:13 +00:00
|
|
|
.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()
|
2022-06-02 21:44:47 +00:00
|
|
|
.into()
|
2022-05-03 17:44:13 +00:00
|
|
|
),
|
2022-06-02 21:44:47 +00:00
|
|
|
embed_video_url: None,
|
2022-05-03 17:44:13 +00:00
|
|
|
},
|
|
|
|
|
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());
|
|
|
|
|
// }
|
2023-06-26 13:07:57 +00:00
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
|
fn test_resolve_image_url() {
|
|
|
|
|
// url that lists the opengraph fields
|
|
|
|
|
let url = Url::parse("https://example.com/one/two.html").unwrap();
|
|
|
|
|
|
|
|
|
|
// root relative url
|
|
|
|
|
let html_bytes = b"<!DOCTYPE html><html><head><meta property='og:image' content='/image.jpg'></head><body></body></html>";
|
|
|
|
|
let metadata = html_to_site_metadata(html_bytes, &url).expect("Unable to parse metadata");
|
|
|
|
|
assert_eq!(
|
|
|
|
|
metadata.image,
|
|
|
|
|
Some(Url::parse("https://example.com/image.jpg").unwrap().into())
|
|
|
|
|
);
|
|
|
|
|
|
|
|
|
|
// base relative url
|
|
|
|
|
let html_bytes = b"<!DOCTYPE html><html><head><meta property='og:image' content='image.jpg'></head><body></body></html>";
|
|
|
|
|
let metadata = html_to_site_metadata(html_bytes, &url).expect("Unable to parse metadata");
|
|
|
|
|
assert_eq!(
|
|
|
|
|
metadata.image,
|
|
|
|
|
Some(
|
|
|
|
|
Url::parse("https://example.com/one/image.jpg")
|
|
|
|
|
.unwrap()
|
|
|
|
|
.into()
|
|
|
|
|
)
|
|
|
|
|
);
|
|
|
|
|
|
|
|
|
|
// absolute url
|
|
|
|
|
let html_bytes = b"<!DOCTYPE html><html><head><meta property='og:image' content='https://cdn.host.com/image.jpg'></head><body></body></html>";
|
|
|
|
|
let metadata = html_to_site_metadata(html_bytes, &url).expect("Unable to parse metadata");
|
|
|
|
|
assert_eq!(
|
|
|
|
|
metadata.image,
|
|
|
|
|
Some(Url::parse("https://cdn.host.com/image.jpg").unwrap().into())
|
|
|
|
|
);
|
|
|
|
|
|
|
|
|
|
// protocol relative url
|
|
|
|
|
let html_bytes = b"<!DOCTYPE html><html><head><meta property='og:image' content='//example.com/image.jpg'></head><body></body></html>";
|
|
|
|
|
let metadata = html_to_site_metadata(html_bytes, &url).expect("Unable to parse metadata");
|
|
|
|
|
assert_eq!(
|
|
|
|
|
metadata.image,
|
|
|
|
|
Some(Url::parse("https://example.com/image.jpg").unwrap().into())
|
|
|
|
|
);
|
|
|
|
|
}
|
2022-05-03 17:44:13 +00:00
|
|
|
}
|
