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)?;
35 fn html_to_site_metadata(html_bytes: &[u8]) -> 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 .and_then(|ogo| Url::parse(&ogo.url).ok());
85 let og_embed_url = page
89 .and_then(|v| Url::parse(&v.url).ok());
92 title: og_title.or(page_title),
93 description: og_description.or(page_description),
94 image: og_image.map(Into::into),
95 embed_video_url: og_embed_url.map(Into::into),
99 #[derive(Deserialize, Debug, Clone)]
100 pub(crate) struct PictrsResponse {
101 files: Vec<PictrsFile>,
105 #[derive(Deserialize, Debug, Clone)]
106 pub(crate) struct PictrsFile {
109 delete_token: String,
112 #[derive(Deserialize, Debug, Clone)]
113 pub(crate) struct PictrsPurgeResponse {
117 #[tracing::instrument(skip_all)]
118 pub(crate) async fn fetch_pictrs(
119 client: &ClientWithMiddleware,
122 ) -> Result<PictrsResponse, LemmyError> {
123 let pictrs_config = settings.pictrs_config()?;
124 is_image_content_type(client, image_url).await?;
126 let fetch_url = format!(
127 "{}image/download?url={}",
129 utf8_percent_encode(image_url.as_str(), NON_ALPHANUMERIC) // TODO this might not be needed
132 let response = client
134 .timeout(REQWEST_TIMEOUT)
138 let response: PictrsResponse = response.json().await.map_err(LemmyError::from)?;
140 if response.msg == "ok" {
143 Err(LemmyError::from_message(&response.msg))
147 /// Purges an image from pictrs
148 /// Note: This should often be coerced from a Result to .ok() in order to fail softly, because:
149 /// - It might fail due to image being not local
150 /// - It might not be an image
151 /// - Pictrs might not be set up
152 pub async fn purge_image_from_pictrs(
153 client: &ClientWithMiddleware,
156 ) -> Result<(), LemmyError> {
157 let pictrs_config = settings.pictrs_config()?;
158 is_image_content_type(client, image_url).await?;
160 let alias = image_url
162 .ok_or_else(|| LemmyError::from_message("Image URL missing path segments"))?
164 .ok_or_else(|| LemmyError::from_message("Image URL missing last path segment"))?;
166 let purge_url = format!("{}/internal/purge?alias={}", pictrs_config.url, alias);
168 let pictrs_api_key = pictrs_config
170 .ok_or_else(|| LemmyError::from_message("pictrs_api_key_not_provided"))?;
171 let response = client
173 .timeout(REQWEST_TIMEOUT)
174 .header("x-api-token", pictrs_api_key)
178 let response: PictrsPurgeResponse = response.json().await.map_err(LemmyError::from)?;
180 if response.msg == "ok" {
183 Err(LemmyError::from_message(&response.msg))
187 /// Both are options, since the URL might be either an html page, or an image
188 /// Returns the SiteMetadata, and a Pictrs URL, if there is a picture associated
189 #[tracing::instrument(skip_all)]
190 pub async fn fetch_site_data(
191 client: &ClientWithMiddleware,
194 ) -> (Option<SiteMetadata>, Option<DbUrl>) {
198 // Ignore errors, since it may be an image, or not have the data.
199 // Warning, this may ignore SSL errors
200 let metadata_option = fetch_site_metadata(client, url).await.ok();
202 let missing_pictrs_file =
203 |r: PictrsResponse| r.files.first().expect("missing pictrs file").file.clone();
205 // Fetch pictrs thumbnail
206 let pictrs_hash = match &metadata_option {
207 Some(metadata_res) => match &metadata_res.image {
208 // Metadata, with image
209 // Try to generate a small thumbnail if there's a full sized one from post-links
210 Some(metadata_image) => fetch_pictrs(client, settings, metadata_image)
212 .map(missing_pictrs_file),
213 // Metadata, but no image
214 None => fetch_pictrs(client, settings, url)
216 .map(missing_pictrs_file),
218 // No metadata, try to fetch the URL as an image
219 None => fetch_pictrs(client, settings, url)
221 .map(missing_pictrs_file),
224 // The full urls are necessary for federation
225 let pictrs_thumbnail = pictrs_hash
228 "{}/pictrs/image/{}",
229 settings.get_protocol_and_hostname(),
237 (metadata_option, pictrs_thumbnail.map(Into::into))
239 None => (None, None),
243 #[tracing::instrument(skip_all)]
244 async fn is_image_content_type(client: &ClientWithMiddleware, url: &Url) -> Result<(), LemmyError> {
245 let response = client.get(url.as_str()).send().await?;
249 .ok_or_else(|| LemmyError::from_message("No Content-Type header"))?
251 .starts_with("image/")
255 Err(LemmyError::from_message("Not an image type."))
259 pub fn build_user_agent(settings: &Settings) -> String {
263 settings.get_protocol_and_hostname()
269 use crate::request::{build_user_agent, fetch_site_metadata, SiteMetadata};
270 use lemmy_utils::settings::SETTINGS;
273 // These helped with testing
275 async fn test_site_metadata() {
276 let settings = &SETTINGS.clone();
277 let client = reqwest::Client::builder()
278 .user_agent(build_user_agent(settings))
282 let sample_url = Url::parse("https://gitlab.com/IzzyOnDroid/repo/-/wikis/FAQ").unwrap();
283 let sample_res = fetch_site_metadata(&client, &sample_url).await.unwrap();
286 title: Some("FAQ · Wiki · IzzyOnDroid / repo · GitLab".to_string()),
288 "The F-Droid compatible repo at https://apt.izzysoft.de/fdroid/".to_string()
291 Url::parse("https://gitlab.com/uploads/-/system/project/avatar/4877469/iod_logo.png")
295 embed_video_url: None,
302 // fn test_pictshare() {
303 // let res = fetch_pictshare("https://upload.wikimedia.org/wikipedia/en/2/27/The_Mandalorian_logo.jpg");
304 // assert!(res.is_ok());
305 // let res_other = fetch_pictshare("https://upload.wikimedia.org/wikipedia/en/2/27/The_Mandalorian_logo.jpgaoeu");
306 // assert!(res_other.is_err());