diff --git a/CHANGELOG.md b/CHANGELOG.md index 1229ca73..0acbe2f3 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -16,6 +16,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0 - `req.reply_to` method to the new `crate::sugar::request::RequestReplyExt` trait - `req.disable_link_preview` method to the new `crate::sugar::request::RequestLinkPreviewExt` trait - `stack_size` setter to `DispatcherBuilder` ([PR 1185](https://github.com/teloxide/teloxide/pull/1185)) +- `utils::render` module to render HTML/Markdown-formatted output ([PR 1152](https://github.com/teloxide/teloxide/pull/1152)) ### Changed @@ -563,4 +564,5 @@ This release was yanked because it accidentally [breaks backwards compatibility] ## 0.1.0 - 2020-02-19 ### Added + - This project. diff --git a/Cargo.lock b/Cargo.lock index 866dcc30..e6f465e5 100644 --- a/Cargo.lock +++ b/Cargo.lock @@ -255,9 +255,9 @@ checksum = "428d9aa8fbc0670b7b8d6030a7fadd0f86151cae55e4dbbece15f3780a3dfaf3" [[package]] name = "cc" -version = "1.1.30" +version = "1.1.31" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "b16803a61b81d9eabb7eae2588776c4c1e584b738ede45fdbb4c972cec1e9945" +checksum = "c2e7962b54006dcfcc61cb72735f4d89bb97061dd6a7ed882ec6b8ee53714c6f" dependencies = [ "shlex", ] @@ -583,9 +583,9 @@ checksum = "e8c02a5121d4ea3eb16a80748c74f5549a5665e4c21333c6098f283870fbdea6" [[package]] name = "flume" -version = "0.11.0" +version = "0.11.1" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "55ac459de2512911e4b674ce33cf20befaba382d05b62b008afc1c8b57cbf181" +checksum = "da0e4dd2a88388a1f4ccc7c9ce104604dab68d9f408dc34cd45823d5a9069095" dependencies = [ "futures-core", "futures-sink", @@ -624,9 +624,9 @@ dependencies = [ [[package]] name = "futures" -version = "0.3.30" +version = "0.3.31" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "645c6916888f6cb6350d2550b80fb63e734897a8498abe35cfb732b6487804b0" +checksum = "65bc07b1a8bc7c85c5f2e110c476c7389b4554ba72af57d8445ea63a576b0876" dependencies = [ "futures-channel", "futures-core", @@ -910,9 +910,9 @@ checksum = "9a3a5bfb195931eeb336b2a7b4d761daec841b97f947d34394601737a7bba5e4" [[package]] name = "hyper" -version = "1.4.1" +version = "1.5.0" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "50dfd22e0e76d0f662d429a5f80fcaf3855009297eab6a0a9f8543834744ba05" +checksum = "bbbff0a806a4728c99295b254c8838933b5b082d75e3cb70c8dab21fdfbcfa9a" dependencies = [ "bytes", "futures-channel", @@ -1102,18 +1102,18 @@ checksum = "49f1f14873335454500d59611f1cf4a4b0f786f9ac11f4312a78e4cf2566695b" [[package]] name = "js-sys" -version = "0.3.70" +version = "0.3.72" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "1868808506b929d7b0cfa8f75951347aa71bb21144b7791bae35d9bccfcfe37a" +checksum = "6a88f1bda2bd75b0452a14784937d796722fdebfe50df998aeb3f0b7603019a9" dependencies = [ "wasm-bindgen", ] [[package]] name = "libc" -version = "0.2.159" +version = "0.2.161" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "561d97a539a36e26a9a5fad1ea11a3039a67714694aaa379433e580854bc3dc5" +checksum = "8e9489c2807c139ffd9c1794f4af0ebe86a828db53ecdc7fea2111d0fed085d1" [[package]] name = "libsqlite3-sys" @@ -1276,9 +1276,9 @@ checksum = "1261fe7e33c73b354eab43b1273a57c8f967d0391e80353e51f764ac02cf6775" [[package]] name = "openssl" -version = "0.10.66" +version = "0.10.68" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "9529f4786b70a3e8c61e11179af17ab6188ad8d0ded78c5529441ed39d4bd9c1" +checksum = "6174bc48f102d208783c2c84bf931bb75927a617866870de8a4ea85597f871f5" dependencies = [ "bitflags 2.4.2", "cfg-if", @@ -1308,9 +1308,9 @@ checksum = "ff011a302c396a5197692431fc1948019154afc178baf7d8e37367442a4601cf" [[package]] name = "openssl-sys" -version = "0.9.103" +version = "0.9.104" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "7f9e8deee91df40a943c71b917e5874b951d32a802526c85721ce3b776c929d6" +checksum = "45abf306cbf99debc8195b66b7346498d7b10c210de50418b5ccd7ceba08c741" dependencies = [ "cc", "libc", @@ -1452,9 +1452,9 @@ dependencies = [ [[package]] name = "proc-macro2" -version = "1.0.87" +version = "1.0.88" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "b3e4daa0dcf6feba26f985457cdf104d4b4256fc5a09547140f3631bb076b19a" +checksum = "7c3a7fc5db1e57d5a779a352c8cdb57b29aa4c40cc69c3a68a7fedc815fbf2f9" dependencies = [ "unicode-ident", ] @@ -1772,9 +1772,9 @@ dependencies = [ [[package]] name = "rustls-pki-types" -version = "1.9.0" +version = "1.10.0" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "0e696e35370c65c9c541198af4543ccd580cf17fc25d8e05c5a242b202488c55" +checksum = "16f1201b3c9a7ee8039bcadc17b7e605e2945b27eee7631788c1bd2b0643674b" [[package]] name = "rustls-webpki" @@ -1789,9 +1789,9 @@ dependencies = [ [[package]] name = "rustversion" -version = "1.0.17" +version = "1.0.18" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "955d28af4278de8121b7ebeb796b6a45735dc01436d898801014aced2773a3d6" +checksum = "0e819f2bc632f285be6d7cd36e25940d45b2391dd6d9b939e79de557f7014248" [[package]] name = "ryu" @@ -1875,9 +1875,9 @@ dependencies = [ [[package]] name = "serde_json" -version = "1.0.128" +version = "1.0.132" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "6ff5456707a1de34e7e37f2a6fd3d3f808c318259cbd01ab6377795054b483d8" +checksum = "d726bfaff4b320266d395898905d0eba0345aae23b54aee3a737e260fd46db03" dependencies = [ "itoa", "memchr", @@ -2565,12 +2565,9 @@ checksum = "42ff0bf0c66b8238c6f3b578df37d0b7848e55df8577b3f74f92a69acceeb825" [[package]] name = "unicase" -version = "2.7.0" +version = "2.8.0" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "f7d2d4dafb69621809a81864c9c1b864479e1235c0dd4e199924b9742439ed89" -dependencies = [ - "version_check", -] +checksum = "7e51b68083f157f853b6379db119d1c1be0e6e4dec98101079dec41f6f5cf6df" [[package]] name = "unicode-bidi" @@ -2625,9 +2622,9 @@ dependencies = [ [[package]] name = "uuid" -version = "1.10.0" +version = "1.11.0" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "81dfa00651efa65069b0b6b651f4aaa31ba9e3c3ce0137aaad053604ee7e0314" +checksum = "f8c5f0a0af699448548ad1a2fbf920fb4bee257eae39953ba95cb84891a0446a" dependencies = [ "getrandom", ] @@ -2699,9 +2696,9 @@ dependencies = [ [[package]] name = "wasm-bindgen-futures" -version = "0.4.43" +version = "0.4.45" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "61e9300f63a621e96ed275155c108eb6f843b6a26d053f122ab69724559dc8ed" +checksum = "cc7ec4f8827a71586374db3e87abdb5a2bb3a15afed140221307c3ec06b1f63b" dependencies = [ "cfg-if", "js-sys", @@ -2753,9 +2750,9 @@ dependencies = [ [[package]] name = "web-sys" -version = "0.3.70" +version = "0.3.72" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "26fdeaafd9bd129f65e7c031593c24d62186301e0c72c8978fa1678be7d532c0" +checksum = "f6488b90108c040df0fe62fa815cbdee25124641df01814dd7282749234c6112" dependencies = [ "js-sys", "wasm-bindgen", diff --git a/crates/teloxide-core/src/types/keyboard_button_request_chat.rs b/crates/teloxide-core/src/types/keyboard_button_request_chat.rs index 47f9c4a3..94644aa4 100644 --- a/crates/teloxide-core/src/types/keyboard_button_request_chat.rs +++ b/crates/teloxide-core/src/types/keyboard_button_request_chat.rs @@ -3,6 +3,7 @@ use serde::{Deserialize, Serialize}; use crate::types::{ChatAdministratorRights, RequestId}; /// This object defines the criteria used to request a suitable chat. +/// /// Information about the selected chat will be shared with the bot when the /// corresponding button is pressed. The bot will be granted requested rights in /// the chat if appropriate. [More about requesting chats »] diff --git a/crates/teloxide-core/src/types/keyboard_button_request_users.rs b/crates/teloxide-core/src/types/keyboard_button_request_users.rs index 8c38b831..e7b86f3b 100644 --- a/crates/teloxide-core/src/types/keyboard_button_request_users.rs +++ b/crates/teloxide-core/src/types/keyboard_button_request_users.rs @@ -3,6 +3,7 @@ use serde::{Deserialize, Serialize}; use crate::types::RequestId; /// This object defines the criteria used to request a suitable users. +/// /// Information about the selected users will be shared with the bot when the /// corresponding button is pressed. More about requesting users » /// diff --git a/crates/teloxide/src/stop.rs b/crates/teloxide/src/stop.rs index e7b185ba..9caae098 100644 --- a/crates/teloxide/src/stop.rs +++ b/crates/teloxide/src/stop.rs @@ -53,9 +53,6 @@ impl Future for StopFlag { type Output = (); fn poll(self: Pin<&mut Self>, cx: &mut task::Context<'_>) -> task::Poll { - self.project().0.poll(cx).map(|res| match res { - Err(_aborted) => (), - Ok(unreachable) => match unreachable {}, - }) + self.project().0.poll(cx).map(|_res| ()) } } diff --git a/crates/teloxide/src/utils.rs b/crates/teloxide/src/utils.rs index c7738637..d9aeb979 100644 --- a/crates/teloxide/src/utils.rs +++ b/crates/teloxide/src/utils.rs @@ -3,6 +3,7 @@ pub mod command; pub mod html; pub mod markdown; +pub mod render; pub(crate) mod shutdown_token; pub use teloxide_core::net::client_from_env; diff --git a/crates/teloxide/src/utils/markdown.rs b/crates/teloxide/src/utils/markdown.rs index ec007c29..36e96beb 100644 --- a/crates/teloxide/src/utils/markdown.rs +++ b/crates/teloxide/src/utils/markdown.rs @@ -4,6 +4,10 @@ use teloxide_core::types::{User, UserId}; +pub(super) const ESCAPE_CHARS: [char; 19] = [ + '\\', '_', '*', '[', ']', '(', ')', '~', '`', '>', '#', '+', '-', '=', '|', '{', '}', '.', '!', +]; + /// Applies the bold font style to the string. /// /// Passed string will not be automatically escaped because it can contain @@ -119,11 +123,8 @@ pub fn code_inline(s: &str) -> String { #[must_use = "This function returns a new string, rather than mutating the argument, so calling it \ without using its output does nothing useful"] pub fn escape(s: &str) -> String { - const CHARS: [char; 18] = - ['_', '*', '[', ']', '(', ')', '~', '`', '>', '#', '+', '-', '=', '|', '{', '}', '.', '!']; - s.chars().fold(String::with_capacity(s.len()), |mut s, c| { - if CHARS.contains(&c) { + if ESCAPE_CHARS.contains(&c) { s.push('\\'); } s.push(c); @@ -247,10 +248,11 @@ mod tests { #[test] fn test_escape() { + assert_eq!(escape("\\!"), r"\\\!"); assert_eq!(escape("* foobar *"), r"\* foobar \*"); assert_eq!( escape(r"_ * [ ] ( ) ~ \ ` > # + - = | { } . !"), - r"\_ \* \[ \] \( \) \~ \ \` \> \# \+ \- \= \| \{ \} \. \!", + r"\_ \* \[ \] \( \) \~ \\ \` \> \# \+ \- \= \| \{ \} \. \!", ); } diff --git a/crates/teloxide/src/utils/render.rs b/crates/teloxide/src/utils/render.rs new file mode 100644 index 00000000..a609b233 --- /dev/null +++ b/crates/teloxide/src/utils/render.rs @@ -0,0 +1,250 @@ +//! Utils for rendering HTML and Markdown output. + +use teloxide_core::types::{MessageEntity, MessageEntityKind as MEK}; + +use tag::*; + +pub use helper::RenderMessageTextHelper; + +mod helper; +mod html; +mod markdown; +mod tag; + +/// Parses text and message entities to produce the final formatted output. +#[derive(Clone, Eq, PartialEq)] +pub struct Renderer<'a> { + text: &'a str, + tags: Vec>, +} + +impl<'a> Renderer<'a> { + /// Creates a new [`Renderer`] instance with given text and message + /// entities. + /// + /// # Arguments + /// + /// - `text`: The input text to be parsed. + /// - `entities`: The message entities (formatting, links, etc.) to be + /// applied to the text. + #[must_use] + pub fn new(text: &'a str, entities: &'a [MessageEntity]) -> Self { + // get the needed size for the new tags that we want to parse from entities + let needed_size: usize = entities + .iter() + .filter(|e| { + matches!( + e.kind, + MEK::Bold + | MEK::Blockquote + | MEK::Italic + | MEK::Underline + | MEK::Strikethrough + | MEK::Spoiler + | MEK::Code + | MEK::Pre { .. } + | MEK::TextLink { .. } + | MEK::TextMention { .. } + | MEK::CustomEmoji { .. } + ) + }) + .count() + * 2; // 2 because we insert two tags for each entity + + let mut tags = Vec::with_capacity(needed_size); + + for (index, entity) in entities.iter().enumerate() { + let kind = match &entity.kind { + MEK::Bold => Kind::Bold, + MEK::Blockquote => Kind::Blockquote, + MEK::Italic => Kind::Italic, + MEK::Underline => Kind::Underline, + MEK::Strikethrough => Kind::Strikethrough, + MEK::Spoiler => Kind::Spoiler, + MEK::Code => Kind::Code, + MEK::Pre { language } => Kind::Pre(language.as_ref().map(String::as_str)), + MEK::TextLink { url } => Kind::TextLink(url.as_str()), + MEK::TextMention { user } => Kind::TextMention(user.id.0), + MEK::CustomEmoji { custom_emoji_id } => Kind::CustomEmoji(custom_emoji_id), + _ => continue, + }; + + // FIXME: maybe instead of clone store all the `kind`s in a seperate + // vector and then just store the index here? + tags.push(Tag::start(kind.clone(), entity.offset, index)); + tags.push(Tag::end(kind, entity.offset + entity.length, index)); + } + + tags.sort_unstable(); + + Self { text, tags } + } + + /// Renders text with a given [`TagWriter`]. + /// + /// This method iterates through the text and the associated position tags + /// and writes the text with the appropriate tags to a buffer, which is then + /// returned as a `String`. + /// + /// If input have no tags we just return the original text as-is. + #[must_use] + fn format(&self, writer: &TagWriter) -> String { + if self.tags.is_empty() { + return self.text.to_owned(); + } + + let mut buffer = + String::with_capacity(self.text.len() + writer.get_extra_size_for_tags(&self.tags)); + let mut tags = self.tags.iter(); + let mut current_tag = tags.next(); + + let mut prev_point = None; + + for (idx, point) in self.text.encode_utf16().enumerate() { + loop { + match current_tag { + Some(tag) if tag.offset == idx => { + (writer.write_tag_fn)(tag, &mut buffer); + current_tag = tags.next(); + } + _ => break, + } + } + + let ch = if let Some(previous) = prev_point.take() { + char::decode_utf16([previous, point]).next().unwrap().unwrap() + } else { + match char::decode_utf16([point]).next().unwrap() { + Ok(c) => c, + Err(unpaired) => { + prev_point = Some(unpaired.unpaired_surrogate()); + continue; + } + } + }; + + (writer.write_char_fn)(ch, &mut buffer); + } + + for tag in current_tag.into_iter().chain(tags) { + (writer.write_tag_fn)(tag, &mut buffer); + } + + buffer + } + + /// Renders and returns the text as an **HTML-formatted** string. + #[must_use] + #[inline] + pub fn as_html(&self) -> String { + self.format(&html::HTML) + } + + /// Renders and returns the text as a **Markdown-formatted** string. + #[must_use] + #[inline] + pub fn as_markdown(&self) -> String { + self.format(&markdown::MARKDOWN) + } +} + +#[cfg(test)] +mod test { + use super::*; + + #[test] + fn test_render_simple() { + let text = "Bold italic Bold italic <underline_"); + assert_eq!(render.as_markdown(), "**Bold** _\ritalic_\r __\rpre, normal and 
rusty
code", + ); + assert_eq!( + render.as_markdown(), + "Some ```\npre```\n, `normal` and ```rust\nrusty```\n code", + ); + } + + #[test] + fn test_render_nested() { + let text = "Some bold both italics"; + let entities = vec![ + MessageEntity { kind: MEK::Bold, offset: 5, length: 9 }, + MessageEntity { kind: MEK::Italic, offset: 10, length: 12 }, + ]; + + let render = Renderer::new(text, &entities); + + assert_eq!(render.as_html(), "Some bold both italics"); + assert_eq!(render.as_markdown(), "Some **bold _\rboth** italics_\r"); + } + + #[test] + fn test_render_complex() { + let text = "Hi how are you?\nnested entities are cool\nIm in a Blockquote!"; + let entities = vec![ + MessageEntity { kind: MEK::Bold, offset: 0, length: 2 }, + MessageEntity { kind: MEK::Italic, offset: 3, length: 3 }, + MessageEntity { kind: MEK::Underline, offset: 7, length: 3 }, + MessageEntity { kind: MEK::Strikethrough, offset: 11, length: 3 }, + MessageEntity { kind: MEK::Bold, offset: 16, length: 1 }, + MessageEntity { kind: MEK::Bold, offset: 17, length: 5 }, + MessageEntity { kind: MEK::Underline, offset: 17, length: 4 }, + MessageEntity { kind: MEK::Strikethrough, offset: 17, length: 4 }, + MessageEntity { + kind: MEK::TextLink { url: reqwest::Url::parse("https://t.me/").unwrap() }, + offset: 23, + length: 8, + }, + MessageEntity { + kind: MEK::TextLink { url: reqwest::Url::parse("tg://user?id=1234567").unwrap() }, + offset: 32, + length: 3, + }, + MessageEntity { kind: MEK::Code, offset: 36, length: 4 }, + MessageEntity { kind: MEK::Blockquote, offset: 41, length: 19 }, + ]; + + let render = Renderer::new(text, &entities); + + assert_eq!( + render.as_html(), + "Hi how are you?\nnested \ + entities are cool\n\ +
Im in a Blockquote!
" + ); + assert_eq!( + render.as_markdown(), + "**Hi** _\rhow_\r __\rare__\r ~you~?\n**n****__\r~este~__\rd** [entities](https://t.me/) \ + [are](tg://user?id=1234567) `cool`\n>Im in a Blockquote\\!" + ); + } +} diff --git a/crates/teloxide/src/utils/render/helper.rs b/crates/teloxide/src/utils/render/helper.rs new file mode 100644 index 00000000..929c5752 --- /dev/null +++ b/crates/teloxide/src/utils/render/helper.rs @@ -0,0 +1,64 @@ +//! A helpful trait for rendering text/caption and entities back to HTML or +//! Markdown. + +use teloxide_core::types::Message; + +use super::Renderer; + +/// Generates HTML and Markdown representations of text and captions in a +/// Telegram message. +pub trait RenderMessageTextHelper { + /// Returns the HTML representation of the message text, if the message + /// contains text. This method will parse the text and any entities + /// (such as bold, italic, links, etc.) and return the HTML-formatted + /// string. + #[must_use] + fn html_text(&self) -> Option; + + /// Returns the Markdown representation of the message text, if the message + /// contains text. This method will parse the text and any entities + /// (such as bold, italic, links, etc.) and return the + /// Markdown-formatted string. + #[must_use] + fn markdown_text(&self) -> Option; + + /// Returns the HTML representation of the message caption, if the message + /// contains caption. This method will parse the caption and any + /// entities (such as bold, italic, links, etc.) and return the + /// HTML-formatted string. + #[must_use] + fn html_caption(&self) -> Option; + + /// Returns the Markdown representation of the message caption, if the + /// message contains caption. This method will parse the caption and any + /// entities (such as bold, italic, links, etc.) and return the + /// Markdown-formatted string. + #[must_use] + fn markdown_caption(&self) -> Option; +} + +impl RenderMessageTextHelper for Message { + fn html_text(&self) -> Option { + self.text() + .zip(self.entities()) + .map(|(text, entities)| Renderer::new(text, entities).as_html()) + } + + fn markdown_text(&self) -> Option { + self.text() + .zip(self.entities()) + .map(|(text, entities)| Renderer::new(text, entities).as_markdown()) + } + + fn html_caption(&self) -> Option { + self.caption() + .zip(self.caption_entities()) + .map(|(text, entities)| Renderer::new(text, entities).as_html()) + } + + fn markdown_caption(&self) -> Option { + self.caption() + .zip(self.caption_entities()) + .map(|(text, entities)| Renderer::new(text, entities).as_markdown()) + } +} diff --git a/crates/teloxide/src/utils/render/html.rs b/crates/teloxide/src/utils/render/html.rs new file mode 100644 index 00000000..95f76070 --- /dev/null +++ b/crates/teloxide/src/utils/render/html.rs @@ -0,0 +1,70 @@ +use std::fmt::Write; + +use super::{ComplexTag, Kind, Place, SimpleTag, Tag, TagWriter}; + +pub static HTML: TagWriter = TagWriter { + bold: SimpleTag::new("", ""), + blockquote: SimpleTag::new("
", "
"), + italic: SimpleTag::new("", ""), + underline: SimpleTag::new("", ""), + strikethrough: SimpleTag::new("", ""), + spoiler: SimpleTag::new("", ""), + code: SimpleTag::new("", ""), + pre_no_lang: SimpleTag::new("
", "
"), + pre: ComplexTag::new("
", "
"), + text_link: ComplexTag::new("", ""), + text_mention: ComplexTag::new("", ""), + custom_emoji: ComplexTag::new("", ""), + write_tag_fn: write_tag, + write_char_fn: write_char, +}; + +fn write_tag(tag: &Tag, buf: &mut String) { + match tag.kind { + Kind::Bold => buf.push_str(HTML.bold.get_tag(tag.place)), + Kind::Blockquote => buf.push_str(HTML.blockquote.get_tag(tag.place)), + Kind::Italic => buf.push_str(HTML.italic.get_tag(tag.place)), + Kind::Underline => buf.push_str(HTML.underline.get_tag(tag.place)), + Kind::Strikethrough => buf.push_str(HTML.strikethrough.get_tag(tag.place)), + Kind::Spoiler => buf.push_str(HTML.spoiler.get_tag(tag.place)), + Kind::Code => buf.push_str(HTML.code.get_tag(tag.place)), + Kind::Pre(lang) => match tag.place { + Place::Start => match lang { + Some(lang) => write!(buf, "{}{}{}", HTML.pre.start, lang, HTML.pre.middle).unwrap(), + None => buf.push_str(HTML.pre_no_lang.start), + }, + Place::End => buf.push_str(lang.map_or(HTML.pre_no_lang.end, |_| HTML.pre.end)), + }, + Kind::TextLink(url) => match tag.place { + Place::Start => { + write!(buf, "{}{}{}", HTML.text_link.start, url, HTML.text_link.middle).unwrap() + } + Place::End => buf.push_str(HTML.text_link.end), + }, + Kind::TextMention(id) => match tag.place { + Place::Start => { + write!(buf, "{}{}{}", HTML.text_mention.start, id, HTML.text_mention.middle) + .unwrap() + } + Place::End => buf.push_str(HTML.text_mention.end), + }, + Kind::CustomEmoji(custom_emoji_id) => match tag.place { + Place::Start => write!( + buf, + "{}{}{}", + HTML.custom_emoji.start, custom_emoji_id, HTML.custom_emoji.middle + ) + .unwrap(), + Place::End => buf.push_str(HTML.custom_emoji.end), + }, + } +} + +fn write_char(ch: char, buf: &mut String) { + match ch { + '&' => buf.push_str("&"), + '<' => buf.push_str("<"), + '>' => buf.push_str(">"), + c => buf.push(c), + } +} diff --git a/crates/teloxide/src/utils/render/markdown.rs b/crates/teloxide/src/utils/render/markdown.rs new file mode 100644 index 00000000..822274fb --- /dev/null +++ b/crates/teloxide/src/utils/render/markdown.rs @@ -0,0 +1,73 @@ +use std::fmt::Write; + +use crate::utils::markdown::ESCAPE_CHARS; + +use super::{ComplexTag, Kind, Place, SimpleTag, Tag, TagWriter}; + +pub static MARKDOWN: TagWriter = TagWriter { + bold: SimpleTag::new("**", "**"), + blockquote: SimpleTag::new(">", ""), + italic: SimpleTag::new("_\r", "_\r"), + underline: SimpleTag::new("__\r", "__\r"), + strikethrough: SimpleTag::new("~", "~"), + spoiler: SimpleTag::new("||", "||"), + code: SimpleTag::new("`", "`"), + pre_no_lang: SimpleTag::new("```\n", "```\n"), + pre: ComplexTag::new("```", "\n", "```\n"), + text_link: ComplexTag::new("[", "](", ")"), + text_mention: ComplexTag::new("[", "](tg://user?id=", ")"), + custom_emoji: ComplexTag::new("[", "](tg://emoji?id=", ")"), + write_tag_fn: write_tag, + write_char_fn: write_char, +}; + +fn write_tag(tag: &Tag, buf: &mut String) { + match tag.kind { + Kind::Bold => buf.push_str(MARKDOWN.bold.get_tag(tag.place)), + Kind::Blockquote => buf.push_str(MARKDOWN.blockquote.get_tag(tag.place)), + Kind::Italic => buf.push_str(MARKDOWN.italic.get_tag(tag.place)), + Kind::Underline => buf.push_str(MARKDOWN.underline.get_tag(tag.place)), + Kind::Strikethrough => buf.push_str(MARKDOWN.strikethrough.get_tag(tag.place)), + Kind::Spoiler => buf.push_str(MARKDOWN.spoiler.get_tag(tag.place)), + Kind::Code => buf.push_str(MARKDOWN.code.get_tag(tag.place)), + Kind::Pre(lang) => match tag.place { + Place::Start => match lang { + Some(lang) => { + write!(buf, "{}{}{}", MARKDOWN.pre.start, lang, MARKDOWN.pre.middle).unwrap() + } + None => buf.push_str(MARKDOWN.pre_no_lang.start), + }, + Place::End => buf.push_str(lang.map_or(MARKDOWN.pre_no_lang.end, |_| MARKDOWN.pre.end)), + }, + Kind::TextLink(url) => match tag.place { + Place::Start => buf.push_str(MARKDOWN.text_link.start), + Place::End => { + write!(buf, "{}{}{}", MARKDOWN.text_link.middle, url, MARKDOWN.text_link.end) + .unwrap() + } + }, + Kind::TextMention(id) => match tag.place { + Place::Start => buf.push_str(MARKDOWN.text_mention.start), + Place::End => { + write!(buf, "{}{}{}", MARKDOWN.text_mention.middle, id, MARKDOWN.text_mention.end) + .unwrap() + } + }, + Kind::CustomEmoji(custom_emoji_id) => match tag.place { + Place::Start => buf.push_str(MARKDOWN.custom_emoji.start), + Place::End => write!( + buf, + "{}{}{}", + MARKDOWN.custom_emoji.middle, custom_emoji_id, MARKDOWN.custom_emoji.end + ) + .unwrap(), + }, + } +} + +fn write_char(ch: char, buf: &mut String) { + if ESCAPE_CHARS.contains(&ch) { + buf.push('\\'); + } + buf.push(ch); +} diff --git a/crates/teloxide/src/utils/render/tag.rs b/crates/teloxide/src/utils/render/tag.rs new file mode 100644 index 00000000..2d250e2a --- /dev/null +++ b/crates/teloxide/src/utils/render/tag.rs @@ -0,0 +1,155 @@ +use std::cmp::Ordering; + +#[derive(Clone)] +pub struct Tag<'a> { + pub place: Place, + pub kind: Kind<'a>, + pub offset: usize, + pub index: usize, +} + +impl<'a> Tag<'a> { + #[inline(always)] + pub const fn start(kind: Kind<'a>, offset: usize, index: usize) -> Self { + Self { place: Place::Start, kind, offset, index } + } + + #[inline(always)] + pub const fn end(kind: Kind<'a>, offset: usize, index: usize) -> Self { + Self { place: Place::End, kind, offset, index } + } +} + +impl<'a> Eq for Tag<'a> {} + +impl<'a> PartialEq for Tag<'a> { + fn eq(&self, other: &Self) -> bool { + // We don't check kind here + self.place == other.place && self.offset == other.offset && self.index == other.index + } +} + +impl<'a> Ord for Tag<'a> { + fn cmp(&self, other: &Self) -> Ordering { + self.offset.cmp(&other.offset).then_with(|| self.place.cmp(&other.place)).then_with(|| { + match other.place { + Place::Start => self.index.cmp(&other.index), + Place::End => other.index.cmp(&self.index), + } + }) + } +} + +impl<'a> PartialOrd for Tag<'a> { + fn partial_cmp(&self, other: &Self) -> Option { + Some(self.cmp(other)) + } +} + +#[derive(Clone, Copy, PartialEq, Eq, PartialOrd, Ord)] +pub enum Place { + // HACK: `End` needs to be first because of the `Ord` Implementation. + // the reason is when comparing tags we want the `End` to be first if the offset + // is the same. + End, + Start, +} + +#[derive(Clone, PartialEq, Eq)] +pub enum Kind<'a> { + Bold, + Blockquote, + Italic, + Underline, + Strikethrough, + Spoiler, + Code, + Pre(Option<&'a str>), + TextLink(&'a str), + TextMention(u64), + CustomEmoji(&'a str), +} + +pub struct SimpleTag { + pub start: &'static str, + pub end: &'static str, +} + +impl SimpleTag { + #[inline] + pub const fn new(start: &'static str, end: &'static str) -> Self { + Self { start, end } + } + + pub const fn get_tag(&self, place: Place) -> &'static str { + match place { + Place::Start => self.start, + Place::End => self.end, + } + } +} + +pub struct ComplexTag { + pub start: &'static str, + pub middle: &'static str, + pub end: &'static str, +} + +impl ComplexTag { + #[inline] + pub const fn new(start: &'static str, middle: &'static str, end: &'static str) -> Self { + Self { start, middle, end } + } +} + +pub struct TagWriter { + pub bold: SimpleTag, + pub blockquote: SimpleTag, + pub italic: SimpleTag, + pub underline: SimpleTag, + pub strikethrough: SimpleTag, + pub spoiler: SimpleTag, + pub code: SimpleTag, + pub pre_no_lang: SimpleTag, + pub pre: ComplexTag, + pub text_link: ComplexTag, + pub text_mention: ComplexTag, + pub custom_emoji: ComplexTag, + pub write_tag_fn: fn(&Tag, buf: &mut String), + pub write_char_fn: fn(char, buf: &mut String), +} + +impl TagWriter { + pub fn get_extra_size_for_tags(&self, tags: &[Tag]) -> usize { + tags.iter() + .map(|tag| match tag.kind { + Kind::Bold => self.bold.get_tag(tag.place).len(), + Kind::Blockquote => self.blockquote.get_tag(tag.place).len(), + Kind::Italic => self.italic.get_tag(tag.place).len(), + Kind::Underline => self.underline.get_tag(tag.place).len(), + Kind::Strikethrough => self.strikethrough.get_tag(tag.place).len(), + Kind::Spoiler => self.spoiler.get_tag(tag.place).len(), + Kind::Code => self.code.get_tag(tag.place).len(), + Kind::Pre(lang) => match tag.place { + Place::Start => lang + .map_or(self.pre_no_lang.start.len(), |l| self.pre.start.len() + l.len()), + Place::End => lang.map_or(self.pre_no_lang.end.len(), |_| { + self.pre.middle.len() + self.pre.end.len() + }), + }, + Kind::TextLink(url) => match tag.place { + Place::Start => self.text_link.start.len() + url.len(), + Place::End => self.text_link.middle.len() + self.text_link.end.len(), + }, + Kind::TextMention(id) => match tag.place { + Place::Start => self.text_mention.start.len() + id.ilog10() as usize + 1, + Place::End => self.text_mention.middle.len() + self.text_mention.end.len(), + }, + Kind::CustomEmoji(custom_emoji_id) => match tag.place { + Place::Start => self.custom_emoji.start.len() + custom_emoji_id.len(), + Place::End => self.custom_emoji.middle.len() + self.custom_emoji.end.len(), + }, + }) + .sum() + } +}