Parse IRI links (#57)

* upgrade rust toolchain to 1.77.2

* don't parse internal markdown links. see #66 in the repo for details

* move parenthesis, bracket and angle parsing into dedicated function

* fix parenthesis in target of labeled link

---------

Co-authored-by: Simon Laux <[email protected]>

Author: farooqkz

.github/workflows/ci.yml

@@ -17,7 +17,7 @@ jobs:
       - uses: actions-rs/toolchain@v1
         with:
           profile: minimal
-          toolchain: 1.64.0
+          toolchain: 1.77.2
           override: true
       - run: rustup component add rustfmt
       - uses: actions-rs/cargo@v1
@@ -31,7 +31,7 @@ jobs:
       - uses: actions/checkout@v2
       - uses: actions-rs/toolchain@v1
         with:
-          toolchain: 1.64.0
+          toolchain: 1.77.2
           components: clippy
           override: true
       - uses: actions-rs/clippy-check@v1
@@ -68,9 +68,9 @@ jobs:
       matrix:
         include:
           - os: ubuntu-latest
-            rust: 1.64.0
+            rust: 1.77.2
           - os: windows-latest
-            rust: 1.64.0
+            rust: 1.77.2
     runs-on: ${{ matrix.os }}
     steps:
       - uses: actions/checkout@master

benches/moar_links.txt

@@ -0,0 +1,20 @@
+Let's add some more links just for testing and benching:
+
+these are some IPv6 links:
+
+gopher://[::1]/
+https://[::1]/سلام
+https://[2345:0425:2CA1:0000:0000:0567:5673:23b5]/hello_world
+https://[2345:425:2CA1:0:0:0567:5673:23b5]/hello_world
+
+an IPvfuture link:
+ftp://mrchickenkiller@[vA.A]/var/log/boot.log
+
+some normal links:
+
+https://www.ietf.org/rfc/rfc3987.txt
+https://iamb.chat/messages/index.html
+https://github.com/deltachat/message-parser/issues/67
+https://far.chickenkiller.com
+gopher://republic.circumlunar.space
+https://far.chickenkiller.com/religion/a-god-who-does-not-care/

benches/my_benchmark.rs

@@ -1,10 +1,13 @@
 use criterion::{black_box, criterion_group, criterion_main, Criterion};
-use deltachat_message_parser::parser::{parse_desktop_set, parse_markdown_text, parse_only_text};
+use deltachat_message_parser::parser::{
+    parse_desktop_set, parse_markdown_text, parse_only_text, LinkDestination,
+};
 
 pub fn criterion_benchmark(c: &mut Criterion) {
     let testdata = include_str!("testdata.md");
     let lorem_ipsum_txt = include_str!("lorem_ipsum.txt");
     let r10s_update_message = include_str!("r10s_update_message.txt");
+    let links = include_str!("moar_links.txt");
 
     c.bench_function("only_text_lorem_ipsum.txt", |b| {
         b.iter(|| parse_only_text(black_box(lorem_ipsum_txt)))
@@ -35,6 +38,10 @@ pub fn criterion_benchmark(c: &mut Criterion) {
     c.bench_function("markdown_r10s_update_message.txt", |b| {
         b.iter(|| parse_markdown_text(black_box(r10s_update_message)))
     });
+
+    c.bench_function("parse_link_moar_links.txt", |b| {
+        b.iter(|| LinkDestination::parse(black_box(links)))
+    });
 }
 
 criterion_group!(benches, criterion_benchmark);

benches/testdata.md

@@ -23,7 +23,7 @@ pub fn parse_text(s: &str, enable_markdown: bool) -> JsValue {
     serde_wasm_bindgen::to_value(&ast).expect("Element converts to JsValue")
 }
 
-/// parses text to json AST (text elements and labled links, to replicate current desktop implementation)
+/// parses text to json AST (text elements and labeled links, to replicate current desktop implementation)
 #[wasm_bindgen]
 pub fn parse_desktop_set(s: &str) -> JsValue {
     serde_wasm_bindgen::to_value(&deltachat_message_parser::parser::parse_desktop_set(s))

message_parser_wasm/src/lib.rs

@@ -1 +1 @@
-1.64.0
+1.77.2

rust-toolchain

@@ -41,6 +41,7 @@ Make email addresses clickable, opens the chat with that contact and creates it
 Make URLs clickable.
 
 - detect all valid hyperlink URLs that have the `://` (protocol://host).
+  - according to [RFC3987](https://www.rfc-editor.org/rfc/rfc3987) and [RFC3988](https://www.rfc-editor.org/rfc/rfc3988)
 
 - other links like `mailto:` (note there is just a single `:`, no `://`) will get separate parsing that includes a whitelisted protocol name, otherwise there will likely be unexpected behavior if user types `hello:world` - will be recognized as link.
 

spec.md

@@ -9,7 +9,7 @@
     clippy::get_last_with_len,
     clippy::get_unwrap,
     clippy::get_unwrap,
-    clippy::integer_arithmetic,
+    clippy::arithmetic_side_effects,
     clippy::match_on_vec_items,
     clippy::match_wild_err_arm,
     clippy::missing_panics_doc,