Add support for wikilinks syntax

supporting either

[[url|link text]]

or

[[link text|url]]

examples/s-expr.rs

@@ -86,6 +86,8 @@ fn dump(source: &str) -> io::Result<()> {
         .multiline_block_quotes(true)
         .math_dollars(true)
         .math_code(true)
+        .wikilinks_title_after_pipe(true)
+        .wikilinks_title_before_pipe(true)
         .build()
         .unwrap();
 

fuzz/fuzz_targets/all_options.rs

@@ -23,7 +23,9 @@ fuzz_target!(|s: &str| {
     extension.math_code = true;
     extension.front_matter_delimiter = Some("---".to_string());
     extension.shortcodes = true;
-
+    extension.wikilinks_title_after_pipe = true;
+    extension.wikilinks_title_before_pipe = true;
+
     let mut parse = ParseOptions::default();
     parse.smart = true;
     parse.default_info_string = Some("rust".to_string());

fuzz/fuzz_targets/quadratic.rs

@@ -196,6 +196,8 @@ struct FuzzExtensionOptions {
     math_dollars: bool,
     math_code: bool,
     shortcodes: bool,
+    wikilinks_title_after_pipe: bool,
+    wikilinks_title_before_pipe: bool,
 }
 
 impl FuzzExtensionOptions {
@@ -213,6 +215,8 @@ impl FuzzExtensionOptions {
         extension.math_dollars = self.math_dollars;
         extension.math_code = self.math_code;
         extension.shortcodes = self.shortcodes;
+        extension.wikilinks_title_after_pipe = self.wikilinks_title_after_pipe;
+        extension.wikilinks_title_before_pipe = self.wikilinks_title_before_pipe;
         extension.front_matter_delimiter = None;
         extension.header_ids = None;
         extension

script/cibuild

@@ -39,6 +39,10 @@ if [ x"$SPEC" = "xtrue" ]; then
   python3 spec_tests.py --no-normalize --spec ../../../src/tests/fixtures/math_dollars.md "$PROGRAM_ARG -e math-dollars" \
     || failed=1
   python3 spec_tests.py --no-normalize --spec ../../../src/tests/fixtures/math_code.md "$PROGRAM_ARG -e math-code" \
+    || failed=1
+  python3 spec_tests.py --no-normalize --spec ../../../src/tests/fixtures/wikilinks_title_after_pipe.md "$PROGRAM_ARG -e wikilinks-title-after-pipe" \
+    || failed=1
+  python3 spec_tests.py --no-normalize --spec ../../../src/tests/fixtures/wikilinks_title_before_pipe.md "$PROGRAM_ARG -e wikilinks-title-before-pipe" \
     || failed=1
 
 	python3 spec_tests.py --no-normalize --spec regression.txt "$PROGRAM_ARG" \

src/main.rs

@@ -160,6 +160,8 @@ enum Extension {
     MultilineBlockQuotes,
     MathDollars,
     MathCode,
+    WikilinksTitleAfterPipe,
+    WikilinksTitleBeforePipe,
 }
 
 #[derive(Clone, Copy, Debug, ValueEnum)]
@@ -239,6 +241,8 @@ fn main() -> Result<(), Box<dyn Error>> {
         .multiline_block_quotes(exts.contains(&Extension::MultilineBlockQuotes))
         .math_dollars(exts.contains(&Extension::MathDollars))
         .math_code(exts.contains(&Extension::MathCode))
+        .wikilinks_title_after_pipe(exts.contains(&Extension::WikilinksTitleAfterPipe))
+        .wikilinks_title_before_pipe(exts.contains(&Extension::WikilinksTitleBeforePipe))
         .front_matter_delimiter(cli.front_matter_delimiter);
 
     #[cfg(feature = "shortcodes")]

src/parser/inlines.rs

@@ -183,11 +183,30 @@ impl<'a, 'r, 'o, 'd, 'i, 'c, 'subj> Subject<'a, 'r, 'o, 'd, 'i, 'c, 'subj> {
             '.' => Some(self.handle_period()),
             '[' => {
                 self.pos += 1;
-                let inl =
-                    self.make_inline(NodeValue::Text("[".to_string()), self.pos - 1, self.pos - 1);
-                self.push_bracket(false, inl);
-                self.within_brackets = true;
-                Some(inl)
+
+                let mut wikilink_inl = None;
+
+                if (self.options.extension.wikilinks_title_after_pipe
+                    || self.options.extension.wikilinks_title_before_pipe)
+                    && !self.within_brackets
+                    && self.peek_char() == Some(&(b'['))
+                {
+                    wikilink_inl = self.handle_wikilink();
+                }
+
+                if wikilink_inl.is_none() {
+                    let inl = self.make_inline(
+                        NodeValue::Text("[".to_string()),
+                        self.pos - 1,
+                        self.pos - 1,
+                    );
+                    self.push_bracket(false, inl);
+                    self.within_brackets = true;
+
+                    Some(inl)
+                } else {
+                    wikilink_inl
+                }
             }
             ']' => {
                 self.within_brackets = false;
@@ -1548,6 +1567,117 @@ impl<'a, 'r, 'o, 'd, 'i, 'c, 'subj> Subject<'a, 'r, 'o, 'd, 'i, 'c, 'subj> {
         }
     }
 
+    // Handles wikilink syntax
+    //   [[link text|url]]
+    //   [[url|link text]]
+    pub fn handle_wikilink(&mut self) -> Option<&'a AstNode<'a>> {
+        let startpos = self.pos;
+        let (url, title) = self.wikilink_url_title();
+
+        url?;
+
+        let url_clean = strings::clean_url(url.unwrap());
+        let title_clean = match title {
+            Some(title) => entity::unescape_html(title),
+            None => entity::unescape_html(url.unwrap()),
+        };
+
+        let nl = NodeLink {
+            url: String::from_utf8(url_clean).unwrap(),
+            title: String::new(),
+        };
+        let inl = self.make_inline(NodeValue::Link(nl), startpos - 1, self.pos - 1);
+        inl.append(self.make_inline(
+            NodeValue::Text(String::from_utf8(title_clean).unwrap()),
+            startpos - 1,
+            self.pos - 1,
+        ));
+
+        Some(inl)
+    }
+
+    pub fn wikilink_url_title(&mut self) -> (Option<&[u8]>, Option<&[u8]>) {
+        let left_startpos = self.pos;
+
+        if self.peek_char() != Some(&(b'[')) {
+            return (None, None);
+        }
+
+        let found_left = self.wikilink_component();
+
+        if !found_left {
+            self.pos = left_startpos;
+            return (None, None);
+        }
+
+        let left = strings::trim_slice(&self.input[left_startpos + 1..self.pos]);
+
+        if self.peek_char() == Some(&(b']')) && self.peek_char_n(1) == Some(&(b']')) {
+            self.pos += 2;
+            return (Some(left), None);
+        } else if self.peek_char() != Some(&(b'|')) {
+            self.pos = left_startpos;
+            return (None, None);
+        }
+
+        let right_startpos = self.pos;
+        let found_right = self.wikilink_component();
+
+        if !found_right {
+            self.pos = left_startpos;
+            return (None, None);
+        }
+
+        let right = strings::trim_slice(&self.input[right_startpos + 1..self.pos]);
+
+        if self.peek_char() == Some(&(b']')) && self.peek_char_n(1) == Some(&(b']')) {
+            self.pos += 2;
+
+            if self.options.extension.wikilinks_title_after_pipe {
+                (Some(left), Some(right))
+            } else {
+                (Some(right), Some(left))
+            }
+        } else {
+            self.pos = left_startpos;
+            (None, None)
+        }
+    }
+
+    // Locates the edge of a wikilink component (link text or url), and sets the
+    // self.pos to it's end if it's found.
+    pub fn wikilink_component(&mut self) -> bool {
+        let startpos = self.pos;
+
+        if self.peek_char() != Some(&(b'[')) && self.peek_char() != Some(&(b'|')) {
+            return false;
+        }
+
+        self.pos += 1;
+
+        let mut length = 0;
+        let mut c = 0;
+        while unwrap_into_copy(self.peek_char(), &mut c) && c != b'[' && c != b']' && c != b'|' {
+            if c == b'\\' {
+                self.pos += 1;
+                length += 1;
+                if self.peek_char().map_or(false, |&c| ispunct(c)) {
+                    self.pos += 1;
+                    length += 1;
+                }
+            } else {
+                self.pos += 1;
+                length += 1;
+            }
+            if length > MAX_LINK_LABEL_LENGTH {
+                self.pos = startpos;
+                return false;
+            }
+        }
+
+        true
+    }
+
     pub fn spnl(&mut self) {
         self.skip_spaces();
         if self.skip_line_end() {

src/parser/mod.rs

@@ -423,6 +423,36 @@ pub struct ExtensionOptions {
     ///            "<p>Happy Friday! 😄</p>\n");
     /// ```
     pub shortcodes: bool,
+
+    /// Enables wikilinks using title after pipe syntax
+    ///
+    /// ```` md
+    /// [[url|link text]]
+    /// ````
+    ///
+    /// ```
+    /// # use comrak::{markdown_to_html, Options};
+    /// let mut options = Options::default();
+    /// options.extension.wikilinks_title_after_pipe = true;
+    /// assert_eq!(markdown_to_html("[[url|link text]]", &options),
+    ///            "<p><a href="url">link text</a></p>\n");
+    /// ```
+    pub wikilinks_title_after_pipe: bool,
+
+    /// Enables wikilinks using title before pipe syntax
+    ///
+    /// ```` md
+    /// [[link text|url]]
+    /// ````
+    ///
+    /// ```
+    /// # use comrak::{markdown_to_html, Options};
+    /// let mut options = Options::default();
+    /// options.extension.wikilinks_title_after_pipe = true;
+    /// assert_eq!(markdown_to_html("[[link text|url]]", &options),
+    ///            "<p><a href="url">link text</a></p>\n");
+    /// ```
+    pub wikilinks_title_before_pipe: bool,
 }
 
 #[non_exhaustive]

src/tests.rs

@@ -26,6 +26,7 @@ mod superscript;
 mod table;
 mod tagfilter;
 mod tasklist;
+mod wikilinks;
 mod xml;
 
 #[track_caller]
@@ -142,6 +143,8 @@ macro_rules! html_opts {
                 math_code: true,
                 front_matter_delimiter: Some("---".to_string()),
                 shortcodes: true,
+                wikilinks_title_after_pipe: true,
+                wikilinks_title_before_pipe: true,
             },
             parse: $crate::ParseOptions {
                 smart: true,

src/tests/api.rs

@@ -50,6 +50,8 @@ fn exercise_full_api() {
     extension.front_matter_delimiter(None);
     #[cfg(feature = "shortcodes")]
     extension.shortcodes(true);
+    extension.wikilinks_title_after_pipe(true);
+    extension.wikilinks_title_before_pipe(true);
 
     let mut parse = ParseOptionsBuilder::default();
     parse.smart(false);

src/tests/fixtures/wikilinks_title_after_pipe.md

@@ -0,0 +1,47 @@
+---
+title: Wikilinks
+based_on: https://github.com/jgm/commonmark-hs/blob/master/commonmark-extensions/test/wikilinks_title_after_pipe.md
+---
+
+# Wikilinks, title after pipe
+
+Wikilinks can have one of the following forms:
+
+    [[https://example.org]]
+    [[https://example.org|title]]
+    [[name of page]]
+    [[name of page|title]]
+
+With this version of wikilinks, the title comes after the pipe.
+
+```````````````````````````````` example
+[[https://example.org]]
+.
+<p><a href="https://example.org">https://example.org</a></p>
+````````````````````````````````
+
+```````````````````````````````` example
+[[https://example.org|title]]
+.
+<p><a href="https://example.org">title</a></p>
+````````````````````````````````
+
+```````````````````````````````` example
+[[Name of page]]
+.
+<p><a href="Name%20of%20page">Name of page</a></p>
+````````````````````````````````
+
+```````````````````````````````` example
+[[Name of page|Title]]
+.
+<p><a href="Name%20of%20page">Title</a></p>
+````````````````````````````````
+
+HTML entities are recognized both in the name of page and in the link title.
+
+```````````````````````````````` example
+[[Gesch&uuml;tztes Leerzeichen|&#xDC;ber &amp;nbsp;]]
+.
+<p><a href="Gesch%C3%BCtztes%20Leerzeichen">Über &amp;nbsp;</a></p>
+````````````````````````````````

src/tests/fixtures/wikilinks_title_before_pipe.md

@@ -0,0 +1,55 @@
+---
+title: Wikilinks
+based_on: https://github.com/jgm/commonmark-hs/blob/master/commonmark-extensions/test/wikilinks_title_before_pipe.md
+---
+
+# Wikilinks, title before pipe
+
+Wikilinks can have one of the following forms:
+
+    [[https://example.org]]
+    [[title|https://example.org]]
+    [[name of page]]
+    [[title|name of page]]
+
+With this version of wikilinks, the title comes before the pipe.
+
+```````````````````````````````` example
+[[https://example.org]]
+.
+<p><a href="https://example.org">https://example.org</a></p>
+````````````````````````````````
+
+```````````````````````````````` example
+[[title|https://example.org]]
+.
+<p><a href="https://example.org">title</a></p>
+````````````````````````````````
+
+```````````````````````````````` example
+[[Name of page]]
+.
+<p><a href="Name%20of%20page">Name of page</a></p>
+````````````````````````````````
+
+```````````````````````````````` example
+[[Title|Name of page]]
+.
+<p><a href="Name%20of%20page">Title</a></p>
+````````````````````````````````
+
+Regular links should still work!
+
+```````````````````````````````` example
+[Title](Name%20of%20page)
+.
+<p><a href="Name%20of%20page">Title</a></p>
+````````````````````````````````
+
+HTML entities are recognized both in the name of page and in the link title.
+
+```````````````````````````````` example
+[[&#xDC;ber &amp;nbsp;|Gesch&uuml;tztes Leerzeichen]]
+.
+<p><a href="Gesch%C3%BCtztes%20Leerzeichen">Über &amp;nbsp;</a></p>
+````````````````````````````````

src/tests/propfuzz.rs

@@ -23,6 +23,8 @@ fn propfuzz_doesnt_crash(md: String) {
             front_matter_delimiter: None,
             #[cfg(feature = "shortcodes")]
             shortcodes: true,
+            wikilinks_title_after_pipe: true,
+            wikilinks_title_before_pipe: true,
         },
         parse: ParseOptions {
             smart: true,