use encoding::{all::encodings, DecoderTrap};
use lemmy_db_schema::newtypes::DbUrl;
use lemmy_utils::{
- error::LemmyError,
+ error::{LemmyError, LemmyErrorType},
settings::structs::Settings,
version::VERSION,
REQWEST_TIMEOUT,
// 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)?;
+ let tags = html_to_site_metadata(&html_bytes, url)?;
Ok(tags)
}
-fn html_to_site_metadata(html_bytes: &[u8]) -> Result<SiteMetadata, LemmyError> {
+fn html_to_site_metadata(html_bytes: &[u8], url: &Url) -> Result<SiteMetadata, LemmyError> {
let html = String::from_utf8_lossy(html_bytes);
// Make sure the first line is doctype html
.trim_start()
.lines()
.next()
- .ok_or_else(|| LemmyError::from_message("No lines in html"))?
+ .ok_or(LemmyErrorType::NoLinesInHtml)?
.to_lowercase();
if !first_line.starts_with("<!doctype html>") {
- return Err(LemmyError::from_message(
- "Site metadata page fetch is not DOCTYPE html",
- ));
+ Err(LemmyErrorType::SiteMetadataPageIsNotDoctypeHtml)?;
}
let mut page = HTML::from_string(html.to_string(), None)?;
.opengraph
.images
.first()
- .and_then(|ogo| Url::parse(&ogo.url).ok());
+ // join also works if the target URL is absolute
+ .and_then(|ogo| url.join(&ogo.url).ok());
let og_embed_url = page
.opengraph
.videos
.first()
- .and_then(|v| Url::parse(&v.url).ok());
+ // join also works if the target URL is absolute
+ .and_then(|v| url.join(&v.url).ok());
Ok(SiteMetadata {
title: og_title.or(page_title),
if response.msg == "ok" {
Ok(response)
} else {
- Err(LemmyError::from_message(&response.msg))
+ Err(LemmyErrorType::PictrsResponseError(response.msg))?
}
}
let alias = image_url
.path_segments()
- .ok_or_else(|| LemmyError::from_message("Image URL missing path segments"))?
+ .ok_or(LemmyErrorType::ImageUrlMissingPathSegments)?
.next_back()
- .ok_or_else(|| LemmyError::from_message("Image URL missing last path segment"))?;
+ .ok_or(LemmyErrorType::ImageUrlMissingLastPathSegment)?;
let purge_url = format!("{}/internal/purge?alias={}", pictrs_config.url, alias);
let pictrs_api_key = pictrs_config
.api_key
- .ok_or_else(|| LemmyError::from_message("pictrs_api_key_not_provided"))?;
+ .ok_or(LemmyErrorType::PictrsApiKeyNotProvided)?;
let response = client
.post(&purge_url)
.timeout(REQWEST_TIMEOUT)
if response.msg == "ok" {
Ok(())
} else {
- Err(LemmyError::from_message(&response.msg))
+ Err(LemmyErrorType::PictrsPurgeResponseError(response.msg))?
}
}
client: &ClientWithMiddleware,
settings: &Settings,
url: Option<&Url>,
+ include_image: bool,
) -> (Option<SiteMetadata>, Option<DbUrl>) {
match &url {
Some(url) => {
// 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();
+ if !include_image {
+ return (metadata_option, None);
+ }
let missing_pictrs_file =
|r: PictrsResponse| r.files.first().expect("missing pictrs file").file.clone();
if response
.headers()
.get("Content-Type")
- .ok_or_else(|| LemmyError::from_message("No Content-Type header"))?
+ .ok_or(LemmyErrorType::NoContentTypeHeader)?
.to_str()?
.starts_with("image/")
{
Ok(())
} else {
- Err(LemmyError::from_message("Not an image type."))
+ Err(LemmyErrorType::NotAnImageType)?
}
}
#[cfg(test)]
mod tests {
- use crate::request::{build_user_agent, fetch_site_metadata, SiteMetadata};
+ #![allow(clippy::unwrap_used)]
+ #![allow(clippy::indexing_slicing)]
+
+ use crate::request::{
+ build_user_agent,
+ fetch_site_metadata,
+ html_to_site_metadata,
+ SiteMetadata,
+ };
use lemmy_utils::settings::SETTINGS;
use url::Url;
// let res_other = fetch_pictshare("https://upload.wikimedia.org/wikipedia/en/2/27/The_Mandalorian_logo.jpgaoeu");
// assert!(res_other.is_err());
// }
+
+ #[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())
+ );
+ }
}