feat(misc): embed code links

src/events/message_create.rs

@@ -1,10 +1,16 @@
 use super::*;
 use crate::utils::autorespond::auto_respond;
+use crate::utils::code_embed::utils::handle_code_url;
 use crate::utils::media_channel::handle_media_channel;
 
 pub async fn message_create(ctx: &serenity::Context, new_message: &serenity::Message) {
     let is_media_channel = handle_media_channel(ctx, new_message).await;
-    if !is_media_channel {
-        auto_respond(ctx, new_message).await;
-    }
+
+    if is_media_channel {
+        return;
+    };
+
+    auto_respond(ctx, new_message).await;
+
+    handle_code_url(ctx, new_message).await;
 }

src/utils/code_embed/macros.rs

@@ -0,0 +1,25 @@
+macro_rules! assert_correct_domain {
+    ($url:expr, $expected:expr) => {{
+        if let Some(domain) = $url.domain() {
+            if domain != $expected {
+                return Err(ParserError::WrongParserError(
+                    $expected.to_string(),
+                    domain.to_string(),
+                ));
+            }
+        } else {
+            return Err(ParserError::Error("No domain found".to_string()));
+        }
+    }};
+}
+
+macro_rules! parse_segment {
+    ($segments:expr, $segment:tt) => {
+        $segments.next().ok_or(ParserError::ConversionError(format!(
+            "Failed to parse {}",
+            $segment.to_string()
+        )))
+    };
+}
+
+pub(crate) use {assert_correct_domain, parse_segment};

src/utils/code_embed/mod.rs

@@ -0,0 +1,5 @@
+pub mod macros;
+pub mod url_parser;
+pub mod utils;
+
+use super::*;

src/utils/code_embed/url_parser.rs

@@ -0,0 +1,205 @@
+use std::ffi::OsStr;
+use std::fmt;
+use std::path::Path;
+
+use poise::async_trait;
+use reqwest::Url;
+
+use super::macros::parse_segment;
+use crate::utils::code_embed::macros::assert_correct_domain;
+
+/// A struct that represents a GitHub code URL.
+///
+/// **Note**: The domain of the url has to be github.com.
+pub struct GitHubCodeUrl {
+    pub url: Url,
+}
+
+/// A struct that holds details on code.
+#[derive(Default, std::fmt::Debug)]
+pub struct CodeUrl {
+    pub raw_code_url: String,
+    pub original_code_url: String,
+    pub repo: String,
+    pub user: String,
+    pub branch_or_sha: String,
+    pub relevant_lines: Option<(usize, usize)>,
+    pub language: Option<String>,
+}
+
+/// A struct that holds details on code and a code preview.
+#[derive(Default, std::fmt::Debug)]
+pub struct CodePreview {
+    pub code: CodeUrl,
+    pub preview: Option<String>,
+}
+
+#[async_trait]
+pub trait CodeUrlParser {
+    fn kind(&self) -> &'static str;
+    async fn parse(&self) -> Result<CodePreview, ParserError>;
+    fn parse_code_url(&self) -> Result<CodeUrl, ParserError>;
+}
+
+#[async_trait]
+impl CodeUrlParser for GitHubCodeUrl {
+    fn kind(&self) -> &'static str {
+        "github.com"
+    }
+
+    fn parse_code_url(&self) -> Result<CodeUrl, ParserError> {
+        let mut segments = self
+            .url
+            .path_segments()
+            .ok_or(ParserError::ConversionError(
+                "Failed to convert path segments".to_string(),
+            ))?;
+
+        // parse the segments
+
+        let user = parse_segment!(segments, "user")?;
+        let repo = parse_segment!(segments, "repo")?;
+        let _blob_segment = parse_segment!(segments, "blob"); // GitHub specific segment
+        let branch_or_sha = parse_segment!(segments, "branch or sha")?;
+
+        let mut path = String::new();
+        while let Ok(segment) = parse_segment!(segments, "path") {
+            if segment == "" {
+                continue;
+            }
+            path.push('/');
+            path.push_str(segment);
+        }
+
+        let raw_url = format!(
+            "https://raw.githubusercontent.com/{}/{}/{}{}",
+            user, repo, branch_or_sha, path
+        );
+
+        let mut code_url = CodeUrl {
+            raw_code_url: raw_url,
+            original_code_url: self.url.to_string(),
+            repo: repo.to_string(),
+            user: user.to_string(),
+            branch_or_sha: branch_or_sha.to_string(),
+            ..Default::default()
+        };
+
+        if let Some(fragment) = self.url.fragment() {
+            let mut numbers = fragment
+                .split('-')
+                .map(|s| s.trim_matches('L'))
+                .map(|s| s.parse::<usize>())
+                .collect::<Result<Vec<_>, _>>()
+                .map_err(|_| ParserError::InvalidFragment(fragment.to_string()))?;
+
+            if numbers.len() > 2 {
+                return Err(ParserError::InvalidFragment(fragment.to_string()));
+            }
+
+            let start = numbers.remove(0);
+            let end = numbers.pop().unwrap_or_else(|| start);
+            code_url.relevant_lines = Some((start, end));
+        }
+
+        let mut segments = self.url.path_segments().unwrap();
+        while let Some(segment) = segments.next_back() {
+            if !segment.is_empty() {
+                let extension = Path::new(segment)
+                    .extension()
+                    .and_then(OsStr::to_str)
+                    .map(str::to_string);
+                code_url.language = extension;
+
+                break;
+            }
+        }
+        Ok(code_url)
+    }
+
+    async fn parse(&self) -> Result<CodePreview, ParserError> {
+        assert_correct_domain!(self.url, self.kind());
+
+        let code_url = self.parse_code_url()?;
+
+        // TODO: If the code is huge, downloading could take long. If code_url.relevant_lines is Some, only download up to the relevant lines.
+        let code = reqwest::get(&code_url.raw_code_url)
+            .await
+            .map_err(|_| ParserError::FailedToGetCode("Can't make a request".to_string()))?
+            .text()
+            .await
+            .map_err(|_| ParserError::FailedToGetCode("Can't parse body".to_string()))?;
+
+        let preview = if let Some((start, end)) = code_url.relevant_lines.clone() {
+            let lines = code.lines().collect::<Vec<_>>();
+            let start = start - 1;
+            let end = end - 1;
+
+            if start > end || start >= lines.len() || end >= lines.len() {
+                return Err(ParserError::InvalidFragment(format!("{}-{}", start, end)));
+            }
+
+            let mut code_block = String::new();
+
+            code_block.push_str("```");
+
+            if let Some(language) = code_url.language.clone() {
+                code_block.push_str(&language);
+                code_block.push('\n');
+            }
+
+            code_block.push_str(&lines[start..=end].join("\n"));
+            code_block.push_str("```");
+
+            Some(code_block)
+        } else {
+            None
+        };
+
+        let code_preview = CodePreview {
+            code: code_url,
+            preview,
+        };
+
+        Ok(code_preview)
+    }
+}
+
+#[derive(Debug, Clone)]
+pub enum ParserError {
+    Error(String),
+    WrongParserError(String, String),
+    ConversionError(String),
+    InvalidFragment(String),
+    FailedToGetCode(String),
+}
+
+impl std::error::Error for ParserError {}
+
+impl fmt::Display for ParserError {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        match self {
+            ParserError::Error(e) => {
+                write!(f, "Error: {}", e)
+            },
+            ParserError::WrongParserError(expected, got) => {
+                write!(f, "Expected parser {}, got {}", expected, got)
+            },
+            ParserError::ConversionError(conversion_error) => {
+                write!(f, "Conversion error: {}", conversion_error)
+            },
+            ParserError::InvalidFragment(fragment) => {
+                write!(f, "Invalid fragment: {}", fragment)
+            },
+            ParserError::FailedToGetCode(error) => {
+                write!(f, "Failed to get code: {}", error)
+            },
+        }
+    }
+}
+
+impl From<Box<dyn std::error::Error>> for ParserError {
+    fn from(e: Box<dyn std::error::Error>) -> Self {
+        Self::Error(e.to_string())
+    }
+}

src/utils/code_embed/utils.rs

@@ -0,0 +1,128 @@
+use chrono::Utc;
+use poise::serenity_prelude::{ButtonStyle, ReactionType};
+use reqwest::Url;
+use tracing::{debug, error, trace};
+
+use super::*;
+use crate::utils::bot::get_data_lock;
+use crate::utils::code_embed::url_parser::{CodePreview, CodeUrlParser, GitHubCodeUrl};
+
+pub async fn handle_code_url(ctx: &serenity::Context, new_message: &serenity::Message) {
+    let data_lock = get_data_lock(ctx).await;
+    let configuration = &data_lock.read().await.configuration;
+
+    let mut urls: Vec<Url> = Vec::new();
+
+    fn get_all_http_urls(string: &str, out: &mut Vec<Url>) {
+        fn get_http_url(slice: &str) -> Option<(&str, usize)> {
+            if let Some(start) = slice.find("http") {
+                debug!("HTTP url start: {}", start);
+
+                let new_slice = &slice[start..];
+
+                if let Some(end) = new_slice
+                    .find(" ")
+                    .or(new_slice.find("\n"))
+                    .and_then(|slice_end| Some(start + slice_end))
+                {
+                    debug!("HTTP url end: {}", end);
+
+                    let url = &slice[start..end];
+                    return Some((url, end));
+                }
+            }
+
+            None
+        }
+
+        if let Some((url, next_start_index)) = get_http_url(string) {
+            if let Ok(url) = Url::parse(url) {
+                out.push(url);
+            } else {
+                error!("Failed to parse url: {}", url);
+            }
+
+            get_all_http_urls(&string[next_start_index..], out);
+        }
+    }
+    get_all_http_urls(&new_message.content, &mut urls);
+
+    let mut code_previews: Vec<CodePreview> = Vec::new();
+
+    for url in urls {
+        // TODO: Add support for other domains by using the provider pattern
+        let code_url = GitHubCodeUrl {
+            url: url.clone(),
+        };
+
+        match code_url.parse().await {
+            Err(e) => error!("Failed to parse url: {} ({:?})", url, e),
+            Ok(code_preview) => code_previews.push(code_preview),
+        }
+    }
+
+    if code_previews.is_empty() {
+        return; // Nothing to do
+    }
+
+    if let Err(err) = new_message
+        .channel_id
+        .send_message(&ctx.http, |m| {
+            let mut message = m;
+
+            for code_preview in code_previews {
+                message = message.add_embed(|e| {
+                    let mut e = e
+                        .title("Code preview")
+                        .url(code_preview.code.original_code_url)
+                        .color(configuration.general.embed_color)
+                        .field(
+                            "Raw link",
+                            format!("[Click here]({})", code_preview.code.raw_code_url),
+                            true,
+                        )
+                        .field("Branch/Sha", code_preview.code.branch_or_sha, true);
+
+                    if let Some(preview) = code_preview.preview {
+                        e = e.field("Preview", preview, false)
+                    }
+
+                    let guild = &new_message.guild(&ctx.cache).unwrap();
+                    if let Some(url) = &guild.icon_url() {
+                        e = e.footer(|f| {
+                            f.icon_url(url).text(format!(
+                                "{} • {}",
+                                guild.name,
+                                Utc::today().format("%Y/%m/%d")
+                            ))
+                        })
+                    }
+
+                    e.field(
+                        format!("Original message by {}", new_message.author.tag()),
+                        new_message.content.clone(),
+                        false,
+                    )
+                });
+            }
+
+            message.content(
+                new_message
+                    .mentions
+                    .iter()
+                    .map(|m| format!("<@{}>", m.id))
+                    .collect::<Vec<_>>()
+                    .join(" "),
+            )
+        })
+        .await
+    {
+        error!(
+            "Failed to reply to the message from {}. Error: {:?}",
+            new_message.author.tag(),
+            err
+        );
+    }
+
+    new_message.delete(&ctx.http).await;
+}

src/utils/mod.rs

@@ -2,6 +2,7 @@ use poise::serenity_prelude::{self as serenity, Member, RoleId};
 
 pub mod autorespond;
 pub mod bot;
+pub mod code_embed;
 pub mod decancer;
 pub mod embed;
 pub mod macros;