1 use crate::post::SiteMetadata;
2 use encoding::{all::encodings, DecoderTrap};
3 use lemmy_db_schema::newtypes::DbUrl;
6 settings::structs::Settings,
10 use percent_encoding::{utf8_percent_encode, NON_ALPHANUMERIC};
11 use reqwest_middleware::ClientWithMiddleware;
12 use serde::Deserialize;
17 /// Fetches the post link html tags (like title, description, image, etc)
18 #[tracing::instrument(skip_all)]
19 pub async fn fetch_site_metadata(
20 client: &ClientWithMiddleware,
22 ) -> Result<SiteMetadata, LemmyError> {
23 info!("Fetching site metadata for url: {}", url);
24 let response = client.get(url.as_str()).send().await?;
26 // Can't use .text() here, because it only checks the content header, not the actual bytes
27 // https://github.com/LemmyNet/lemmy/issues/1964
28 let html_bytes = response.bytes().await.map_err(LemmyError::from)?.to_vec();
30 let tags = html_to_site_metadata(&html_bytes, url)?;
35 fn html_to_site_metadata(html_bytes: &[u8], url: &Url) -> Result<SiteMetadata, LemmyError> {
36 let html = String::from_utf8_lossy(html_bytes);
38 // Make sure the first line is doctype html
43 .ok_or_else(|| LemmyError::from_message("No lines in html"))?
46 if !first_line.starts_with("<!doctype html>") {
47 return Err(LemmyError::from_message(
48 "Site metadata page fetch is not DOCTYPE html",
52 let mut page = HTML::from_string(html.to_string(), None)?;
54 // If the web page specifies that it isn't actually UTF-8, re-decode the received bytes with the
55 // proper encoding. If the specified encoding cannot be found, fall back to the original UTF-8
57 if let Some(charset) = page.meta.get("charset") {
58 if charset.to_lowercase() != "utf-8" {
59 if let Some(encoding_ref) = encodings().iter().find(|e| e.name() == charset) {
60 if let Ok(html_with_encoding) = encoding_ref.decode(html_bytes, DecoderTrap::Replace) {
61 page = HTML::from_string(html_with_encoding, None)?;
67 let page_title = page.title;
68 let page_description = page.description;
70 let og_description = page
74 .map(std::string::ToString::to_string);
79 .map(std::string::ToString::to_string);
84 // join also works if the target URL is absolute
85 .and_then(|ogo| url.join(&ogo.url).ok());
86 let og_embed_url = page
90 // join also works if the target URL is absolute
91 .and_then(|v| url.join(&v.url).ok());
94 title: og_title.or(page_title),
95 description: og_description.or(page_description),
96 image: og_image.map(Into::into),
97 embed_video_url: og_embed_url.map(Into::into),
101 #[derive(Deserialize, Debug, Clone)]
102 pub(crate) struct PictrsResponse {
103 files: Vec<PictrsFile>,
107 #[derive(Deserialize, Debug, Clone)]
108 pub(crate) struct PictrsFile {
111 delete_token: String,
114 #[derive(Deserialize, Debug, Clone)]
115 pub(crate) struct PictrsPurgeResponse {
119 #[tracing::instrument(skip_all)]
120 pub(crate) async fn fetch_pictrs(
121 client: &ClientWithMiddleware,
124 ) -> Result<PictrsResponse, LemmyError> {
125 let pictrs_config = settings.pictrs_config()?;
126 is_image_content_type(client, image_url).await?;
128 let fetch_url = format!(
129 "{}image/download?url={}",
131 utf8_percent_encode(image_url.as_str(), NON_ALPHANUMERIC) // TODO this might not be needed
134 let response = client
136 .timeout(REQWEST_TIMEOUT)
140 let response: PictrsResponse = response.json().await.map_err(LemmyError::from)?;
142 if response.msg == "ok" {
145 Err(LemmyError::from_message(&response.msg))
149 /// Purges an image from pictrs
150 /// Note: This should often be coerced from a Result to .ok() in order to fail softly, because:
151 /// - It might fail due to image being not local
152 /// - It might not be an image
153 /// - Pictrs might not be set up
154 pub async fn purge_image_from_pictrs(
155 client: &ClientWithMiddleware,
158 ) -> Result<(), LemmyError> {
159 let pictrs_config = settings.pictrs_config()?;
160 is_image_content_type(client, image_url).await?;
162 let alias = image_url
164 .ok_or_else(|| LemmyError::from_message("Image URL missing path segments"))?
166 .ok_or_else(|| LemmyError::from_message("Image URL missing last path segment"))?;
168 let purge_url = format!("{}/internal/purge?alias={}", pictrs_config.url, alias);
170 let pictrs_api_key = pictrs_config
172 .ok_or_else(|| LemmyError::from_message("pictrs_api_key_not_provided"))?;
173 let response = client
175 .timeout(REQWEST_TIMEOUT)
176 .header("x-api-token", pictrs_api_key)
180 let response: PictrsPurgeResponse = response.json().await.map_err(LemmyError::from)?;
182 if response.msg == "ok" {
185 Err(LemmyError::from_message(&response.msg))
189 /// Both are options, since the URL might be either an html page, or an image
190 /// Returns the SiteMetadata, and a Pictrs URL, if there is a picture associated
191 #[tracing::instrument(skip_all)]
192 pub async fn fetch_site_data(
193 client: &ClientWithMiddleware,
196 ) -> (Option<SiteMetadata>, Option<DbUrl>) {
200 // Ignore errors, since it may be an image, or not have the data.
201 // Warning, this may ignore SSL errors
202 let metadata_option = fetch_site_metadata(client, url).await.ok();
204 let missing_pictrs_file =
205 |r: PictrsResponse| r.files.first().expect("missing pictrs file").file.clone();
207 // Fetch pictrs thumbnail
208 let pictrs_hash = match &metadata_option {
209 Some(metadata_res) => match &metadata_res.image {
210 // Metadata, with image
211 // Try to generate a small thumbnail if there's a full sized one from post-links
212 Some(metadata_image) => fetch_pictrs(client, settings, metadata_image)
214 .map(missing_pictrs_file),
215 // Metadata, but no image
216 None => fetch_pictrs(client, settings, url)
218 .map(missing_pictrs_file),
220 // No metadata, try to fetch the URL as an image
221 None => fetch_pictrs(client, settings, url)
223 .map(missing_pictrs_file),
226 // The full urls are necessary for federation
227 let pictrs_thumbnail = pictrs_hash
230 "{}/pictrs/image/{}",
231 settings.get_protocol_and_hostname(),
239 (metadata_option, pictrs_thumbnail.map(Into::into))
241 None => (None, None),
245 #[tracing::instrument(skip_all)]
246 async fn is_image_content_type(client: &ClientWithMiddleware, url: &Url) -> Result<(), LemmyError> {
247 let response = client.get(url.as_str()).send().await?;
251 .ok_or_else(|| LemmyError::from_message("No Content-Type header"))?
253 .starts_with("image/")
257 Err(LemmyError::from_message("Not an image type."))
261 pub fn build_user_agent(settings: &Settings) -> String {
265 settings.get_protocol_and_hostname()
271 use crate::request::{
274 html_to_site_metadata,
277 use lemmy_utils::settings::SETTINGS;
280 // These helped with testing
282 async fn test_site_metadata() {
283 let settings = &SETTINGS.clone();
284 let client = reqwest::Client::builder()
285 .user_agent(build_user_agent(settings))
289 let sample_url = Url::parse("https://gitlab.com/IzzyOnDroid/repo/-/wikis/FAQ").unwrap();
290 let sample_res = fetch_site_metadata(&client, &sample_url).await.unwrap();
293 title: Some("FAQ · Wiki · IzzyOnDroid / repo · GitLab".to_string()),
295 "The F-Droid compatible repo at https://apt.izzysoft.de/fdroid/".to_string()
298 Url::parse("https://gitlab.com/uploads/-/system/project/avatar/4877469/iod_logo.png")
302 embed_video_url: None,
309 // fn test_pictshare() {
310 // let res = fetch_pictshare("https://upload.wikimedia.org/wikipedia/en/2/27/The_Mandalorian_logo.jpg");
311 // assert!(res.is_ok());
312 // let res_other = fetch_pictshare("https://upload.wikimedia.org/wikipedia/en/2/27/The_Mandalorian_logo.jpgaoeu");
313 // assert!(res_other.is_err());
317 fn test_resolve_image_url() {
318 // url that lists the opengraph fields
319 let url = Url::parse("https://example.com/one/two.html").unwrap();
322 let html_bytes = b"<!DOCTYPE html><html><head><meta property='og:image' content='/image.jpg'></head><body></body></html>";
323 let metadata = html_to_site_metadata(html_bytes, &url).expect("Unable to parse metadata");
326 Some(Url::parse("https://example.com/image.jpg").unwrap().into())
330 let html_bytes = b"<!DOCTYPE html><html><head><meta property='og:image' content='image.jpg'></head><body></body></html>";
331 let metadata = html_to_site_metadata(html_bytes, &url).expect("Unable to parse metadata");
335 Url::parse("https://example.com/one/image.jpg")
342 let html_bytes = b"<!DOCTYPE html><html><head><meta property='og:image' content='https://cdn.host.com/image.jpg'></head><body></body></html>";
343 let metadata = html_to_site_metadata(html_bytes, &url).expect("Unable to parse metadata");
346 Some(Url::parse("https://cdn.host.com/image.jpg").unwrap().into())
349 // protocol relative url
350 let html_bytes = b"<!DOCTYPE html><html><head><meta property='og:image' content='//example.com/image.jpg'></head><body></body></html>";
351 let metadata = html_to_site_metadata(html_bytes, &url).expect("Unable to parse metadata");
354 Some(Url::parse("https://example.com/image.jpg").unwrap().into())