Merge 340ce2b5f3 into 5f26c94dce

2024-11-07 19:20:29 +00:00 · 2024-11-07 19:20:29 +00:00 · 38f107ccb6
commit 38f107ccb6
parent 5f26c94dce 340ce2b5f3
13 changed files with 270 additions and 85 deletions
--- a/development.md
+++ b/development.md
@ -3,7 +3,7 @@
 ```shell
 cargo fmt -- --check
 cargo test --all-features
-cargo clippy --allow-dirty --allow-staged
+cargo clippy --fix --lib -p orgize --allow-dirty --allow-staged
 ```

 ## Update snapshot testing
@ -18,8 +18,8 @@ cargo insta review

 ```shell
 cargo install cargo-fuzz
-rustup default nightly
-cargo fuzz run fuzz_target_1
+rustup toolchain install nightly
+cargo +nightly fuzz run fuzz_target_1
 ```

 ## Benchmark
--- a/src/ast/generate.js
+++ b/src/ast/generate.js
@ -103,6 +103,16 @@ const nodes = [
  {
    struct: "FnDef",
    kind: ["FN_DEF"],
+    token: [
+      ["label", "FN_LABEL"],
+      ["description", "FN_CONTENT"],
+    ],
+    post_blank: true,
+    affiliated_keywords: true,
+  },
+  {
+    struct: "FnContent",
+    kind: ["FN_CONTENT"],
    post_blank: true,
    affiliated_keywords: true,
  },
@ -188,6 +198,7 @@ const nodes = [
  {
    struct: "FnRef",
    kind: ["FN_REF"],
+    token: [["label", "FN_LABEL"]],
  },
  {
    struct: "Macros",
--- a/src/ast/generated.rs
+++ b/src/ast/generated.rs
@ -831,6 +831,70 @@ impl AstNode for FnDef {
    }
 }
 impl FnDef {
+    /// Beginning position of this element
+    pub fn start(&self) -> TextSize {
+        self.syntax.text_range().start()
+    }
+    /// Ending position of this element
+    pub fn end(&self) -> TextSize {
+        self.syntax.text_range().end()
+    }
+    /// Range of this element
+    pub fn text_range(&self) -> TextRange {
+        self.syntax.text_range()
+    }
+    /// Raw text of this element
+    pub fn raw(&self) -> String {
+        self.syntax.to_string()
+    }
+    pub fn label(&self) -> Option<super::Token> {
+        super::token(&self.syntax, FN_LABEL)
+    }
+    pub fn description(&self) -> Option<super::Token> {
+        super::token(&self.syntax, FN_CONTENT)
+    }
+    pub fn post_blank(&self) -> usize {
+        super::blank_lines(&self.syntax)
+    }
+    pub fn caption(&self) -> Option<AffiliatedKeyword> {
+        affiliated_keyword(&self.syntax, |k| k == "CAPTION")
+    }
+    pub fn header(&self) -> Option<AffiliatedKeyword> {
+        affiliated_keyword(&self.syntax, |k| k == "HEADER")
+    }
+    pub fn name(&self) -> Option<AffiliatedKeyword> {
+        affiliated_keyword(&self.syntax, |k| k == "NAME")
+    }
+    pub fn plot(&self) -> Option<AffiliatedKeyword> {
+        affiliated_keyword(&self.syntax, |k| k == "PLOT")
+    }
+    pub fn results(&self) -> Option<AffiliatedKeyword> {
+        affiliated_keyword(&self.syntax, |k| k == "RESULTS")
+    }
+    pub fn attr(&self, backend: &str) -> Option<AffiliatedKeyword> {
+        affiliated_keyword(&self.syntax, |k| {
+            k.starts_with("ATTR_") && &k[5..] == backend
+        })
+    }
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Hash)]
+pub struct FnContent {
+    pub(crate) syntax: SyntaxNode,
+}
+impl AstNode for FnContent {
+    type Language = OrgLanguage;
+    fn can_cast(kind: SyntaxKind) -> bool {
+        kind == FN_CONTENT
+    }
+    fn cast(node: SyntaxNode) -> Option<FnContent> {
+        Self::can_cast(node.kind()).then(|| FnContent { syntax: node })
+    }
+    fn syntax(&self) -> &SyntaxNode {
+        &self.syntax
+    }
+}
+impl FnContent {
    /// Beginning position of this element
    pub fn start(&self) -> TextSize {
        self.syntax.text_range().start()
@ -1680,6 +1744,9 @@ impl FnRef {
    pub fn raw(&self) -> String {
        self.syntax.to_string()
    }
+    pub fn label(&self) -> Option<super::Token> {
+        super::token(&self.syntax, FN_LABEL)
+    }
 }

 #[derive(Debug, Clone, PartialEq, Eq, Hash)]
--- a/src/export/event.rs
+++ b/src/export/event.rs
@ -19,6 +19,7 @@ pub enum Container {
    DynBlock(DynBlock),

    FnDef(FnDef),
+    FnContent(FnContent),
    Comment(Comment),
    FixedWidth(FixedWidth),
    SpecialBlock(SpecialBlock),
@ -57,6 +58,7 @@ pub enum Event {
    Text(Token),
    Macros(Macros),
    Cookie(Cookie),
+    FnLabel(Token),
    InlineCall(InlineCall),
    InlineSrc(InlineSrc),
    Clock(Clock),
--- a/src/export/html.rs
+++ b/src/export/html.rs
@ -1,3 +1,4 @@
+use rowan::ast::AstNode;
 use rowan::NodeOrToken;
 use std::cmp::min;
 use std::fmt;
@ -6,6 +7,7 @@ use std::fmt::Write as _;
 use super::event::{Container, Event};
 use super::TraversalContext;
 use super::Traverser;
+use crate::ast::token;
 use crate::{SyntaxElement, SyntaxKind, SyntaxNode};

 /// A wrapper for escaping sensitive characters in html.
@ -51,6 +53,9 @@ impl<S: AsRef<str>> fmt::Display for HtmlEscape<S> {
 pub struct HtmlExport {
    output: String,

+    ///TODO: track footnotes and citations within the export struct and
+    /// construct them after the document is fully parsed?
+    //footnotes: HashMap<String, String>,
    in_descriptive_list: Vec<bool>,

    table_row: TableRow,
@ -107,6 +112,55 @@ impl Traverser for HtmlExport {
            }
            Event::Leave(Container::Headline(_)) => {}

+            Event::Enter(Container::FnRef(t)) => {
+                if let Some(label) = t.label() {
+                    let _ = write!(
+                        &mut self.output,
+                        "<a href=\"#footnote_{}\" class=\"footnote-reference\">[{}]",
+                        label.syntax().text(),
+                        label.syntax().text()
+                    );
+                }
+                self.output += "</a>";
+            }
+            Event::Leave(Container::FnRef(_)) => {}
+
+            Event::Enter(Container::FnDef(t)) => {
+                self.output += "<aside ";
+                self.output += r#"class="footnote-definition" "#;
+                self.output += ">";
+
+                if let Some(label) = t.label() {
+                    self.output += "<a ";
+                    let _ = write!(
+                        &mut self.output,
+                        "href=\"#footnote_{}\" ",
+                        label.syntax().text()
+                    );
+                    self.output += "class=\"footnote-reference\" ";
+                    self.output += ">";
+                    let _ = write!(&mut self.output, "[{}]", label.syntax().text());
+                    self.output += "</a>";
+                }
+            }
+            Event::Leave(Container::FnDef(_)) => {
+                self.output += "</aside>";
+            }
+
+            Event::Enter(Container::FnContent(c)) => {
+                self.output += "<span class=\"footnote-content\" ";
+                if let Some(parent) = c.syntax().parent() {
+                    if parent.kind() == SyntaxKind::FN_REF || parent.kind() == SyntaxKind::FN_DEF {
+                        let label = token(&parent, SyntaxKind::FN_LABEL).unwrap();
+                        let _ = write!(&mut self.output, "id=\"footnote_{}\" ", label);
+                    }
+                }
+                self.output += ">";
+            }
+            Event::Leave(Container::FnContent(_)) => {
+                self.output += "</span>";
+            }
+
            Event::Enter(Container::Paragraph(_)) => self.output += "<p>",
            Event::Leave(Container::Paragraph(_)) => self.output += "</p>",

@ -297,6 +351,8 @@ impl Traverser for HtmlExport {
                let _ = write!(&mut self.output, "{}", HtmlEscape(text));
            }

+            Event::FnLabel(_) => {}
+
            Event::LineBreak(_) => self.output += "<br/>",

            Event::Snippet(snippet) => {
--- a/src/export/traverse.rs
+++ b/src/export/traverse.rs
@ -181,6 +181,7 @@ pub trait Traverser {
                    DYN_BLOCK => walk!(DynBlock),
                    FN_DEF => walk!(FnDef),
                    FN_REF => walk!(FnRef),
+                    FN_CONTENT => walk!(FnContent),
                    MACROS => walk!(@Macros),
                    SNIPPET => walk!(@Snippet),
                    TIMESTAMP_ACTIVE | TIMESTAMP_INACTIVE | TIMESTAMP_DIARY => walk!(@Timestamp),
@ -210,12 +211,17 @@ pub trait Traverser {
                    _ => {}
                }
            }
-            SyntaxElement::Token(token) => {
-                if token.kind() == TEXT {
+            SyntaxElement::Token(token) => match token.kind() {
+                TEXT => {
                    self.event(Event::Text(Token(token)), ctx);
                    take_control!();
                }
-            }
+                FN_LABEL => {
+                    self.event(Event::FnLabel(Token(token)), ctx);
+                    take_control!();
+                }
+                _ => {}
+            },
        };
    }
 }
--- a/src/replace.rs
+++ b/src/replace.rs
@ -133,7 +133,7 @@ impl Org {
            ) if level <= new_level
            // non-last headline must ends with a newline
                && (headline.end() == self.document().end()
-                    || replace_with.ends_with(&['\n', '\r'])) =>
+                    || replace_with.ends_with(['\n', '\r'])) =>
            {
                self.replace_headline(headline, range, replace_with)
            }
--- a/src/syntax/element.rs
+++ b/src/syntax/element.rs
@ -286,11 +286,12 @@ fn affiliated_keywords() {
          TEXT@10..25 " a footnote def"
          NEW_LINE@25..26 "\n"
        L_BRACKET@26..27 "["
-        TEXT@27..29 "fn"
+        KEYWORD@27..29 "fn"
        COLON@29..30 ":"
-        TEXT@30..34 "WORD"
+        FN_LABEL@30..34 "WORD"
        R_BRACKET@34..35 "]"
-        TEXT@35..55 " https://orgmode.org"
+        FN_CONTENT@35..55
+          TEXT@35..55 " https://orgmode.org"
    "###
    );

--- a/src/syntax/fn_def.rs
+++ b/src/syntax/fn_def.rs
@ -7,12 +7,13 @@ use nom::{

 use super::{
    combinator::{
-        blank_lines, colon_token, l_bracket_token, r_bracket_token, trim_line_end, GreenElement,
-        NodeBuilder,
+        blank_lines, colon_token, l_bracket_token, node, r_bracket_token, trim_line_end,
+        GreenElement, NodeBuilder,
    },
    input::Input,
    keyword::affiliated_keyword_nodes,
-    SyntaxKind,
+    object::standard_object_nodes,
+    SyntaxKind::*,
 };

 #[cfg_attr(
@ -20,6 +21,10 @@ use super::{
  tracing::instrument(level = "debug", skip(input), fields(input = input.s))
 )]
 pub fn fn_def_node(input: Input) -> IResult<Input, GreenElement, ()> {
+    crate::lossless_parser!(fn_def_node_base, input)
+}
+
+fn fn_def_node_base(input: Input) -> IResult<Input, GreenElement, ()> {
    let mut parser = map(
        tuple((
            affiliated_keyword_nodes,
@ -42,20 +47,25 @@ pub fn fn_def_node(input: Input) -> IResult<Input, GreenElement, ()> {
            post_blank,
        )| {
            let mut b = NodeBuilder::new();
+
            b.children.extend(affiliated_keywords);
            b.push(l_bracket);
-            b.text(fn_);
+            b.push(fn_.token(KEYWORD));
            b.push(colon);
-            b.text(label);
+            b.push(label.token(FN_LABEL));
            b.push(r_bracket);
-            b.text(content);
+
+            let content_node = node(FN_CONTENT, standard_object_nodes(content));
+            b.push(content_node);
+
            b.ws(ws_);
            b.nl(nl);
            b.children.extend(post_blank);
-            b.finish(SyntaxKind::FN_DEF)
+            b.finish(FN_DEF)
        },
    );
-    crate::lossless_parser!(parser, input)
+    let (i, fn_def) = parser(input)?;
+    Ok((i, fn_def))
 }

 #[test]
@ -66,68 +76,78 @@ fn parse() {
    let to_fn_def = to_ast::<FnDef>(fn_def_node);

    insta::assert_debug_snapshot!(
-         to_fn_def("[fn:1] https://orgmode.org").syntax,
-         @r###"
-    FN_DEF@0..26
+         to_fn_def("[fn:1] *bold* -  https://orgmode.org").syntax,
+         @r#"
+    FN_DEF@0..36
      L_BRACKET@0..1 "["
-      TEXT@1..3 "fn"
+      KEYWORD@1..3 "fn"
      COLON@3..4 ":"
-      TEXT@4..5 "1"
+      FN_LABEL@4..5 "1"
      R_BRACKET@5..6 "]"
-      TEXT@6..26 " https://orgmode.org"
-    "###
+      FN_CONTENT@6..36
+        TEXT@6..7 " "
+        BOLD@7..13
+          STAR@7..8 "*"
+          TEXT@8..12 "bold"
+          STAR@12..13 "*"
+        TEXT@13..36 " -  https://orgmode.org"
+    "#
    );

    insta::assert_debug_snapshot!(
         to_fn_def("[fn:word_1] https://orgmode.org").syntax,
-         @r###"
+         @r#"
    FN_DEF@0..31
      L_BRACKET@0..1 "["
-      TEXT@1..3 "fn"
+      KEYWORD@1..3 "fn"
      COLON@3..4 ":"
-      TEXT@4..10 "word_1"
+      FN_LABEL@4..10 "word_1"
      R_BRACKET@10..11 "]"
-      TEXT@11..31 " https://orgmode.org"
-    "###
+      FN_CONTENT@11..31
+        TEXT@11..31 " https://orgmode.org"
+    "#
    );

    insta::assert_debug_snapshot!(
         to_fn_def("[fn:WORD-1] https://orgmode.org").syntax,
-         @r###"
+         @r#"
    FN_DEF@0..31
      L_BRACKET@0..1 "["
-      TEXT@1..3 "fn"
+      KEYWORD@1..3 "fn"
      COLON@3..4 ":"
-      TEXT@4..10 "WORD-1"
+      FN_LABEL@4..10 "WORD-1"
      R_BRACKET@10..11 "]"
-      TEXT@11..31 " https://orgmode.org"
-    "###
+      FN_CONTENT@11..31
+        TEXT@11..31 " https://orgmode.org"
+    "#
    );

    insta::assert_debug_snapshot!(
         to_fn_def("[fn:WORD]").syntax,
-         @r###"
+         @r#"
    FN_DEF@0..9
      L_BRACKET@0..1 "["
-      TEXT@1..3 "fn"
+      KEYWORD@1..3 "fn"
      COLON@3..4 ":"
-      TEXT@4..8 "WORD"
+      FN_LABEL@4..8 "WORD"
      R_BRACKET@8..9 "]"
-    "###
+      FN_CONTENT@9..9
+    "#
    );

    insta::assert_debug_snapshot!(
         to_fn_def("[fn:1] In particular, the parser requires stars at column 0 to be\n").syntax,
-         @r###"
+         @r#"
    FN_DEF@0..66
      L_BRACKET@0..1 "["
-      TEXT@1..3 "fn"
+      KEYWORD@1..3 "fn"
      COLON@3..4 ":"
-      TEXT@4..5 "1"
+      FN_LABEL@4..5 "1"
      R_BRACKET@5..6 "]"
-      TEXT@6..65 " In particular, the p ..."
+      FN_CONTENT@6..65
+        TEXT@6..65 " In particular, the p ..."
      NEW_LINE@65..66 "\n"
-    "###
+    "#
    );

    let config = &ParseConfig::default();
@ -138,7 +158,7 @@ fn parse() {

    insta::assert_debug_snapshot!(
         to_fn_def("#+ATTR_poi: 1\n[fn:WORD-1] https://orgmode.org").syntax,
-         @r###"
+         @r##"
    FN_DEF@0..45
      AFFILIATED_KEYWORD@0..14
        HASH_PLUS@0..2 "#+"
@ -147,11 +167,12 @@ fn parse() {
        TEXT@11..13 " 1"
        NEW_LINE@13..14 "\n"
      L_BRACKET@14..15 "["
-      TEXT@15..17 "fn"
+      KEYWORD@15..17 "fn"
      COLON@17..18 ":"
-      TEXT@18..24 "WORD-1"
+      FN_LABEL@18..24 "WORD-1"
      R_BRACKET@24..25 "]"
-      TEXT@25..45 " https://orgmode.org"
-    "###
+      FN_CONTENT@25..45
+        TEXT@25..45 " https://orgmode.org"
+    "##
    );
 }
--- a/src/syntax/fn_ref.rs
+++ b/src/syntax/fn_ref.rs
@ -31,10 +31,10 @@ fn fn_ref_node_base(input: Input) -> IResult<Input, GreenElement, ()> {
        r_bracket_token,
    ))(input)?;

-    let mut children = vec![l_bracket, fn_.text_token(), colon, label.text_token()];
+    let mut children = vec![l_bracket, fn_.token(KEYWORD), colon, label.token(FN_LABEL)];
    if let Some((colon, definition)) = definition {
        children.push(colon);
-        children.extend(standard_object_nodes(definition));
+        children.push(node(FN_CONTENT, standard_object_nodes(definition)));
    }
    children.push(r_bracket);

@ -64,56 +64,59 @@ fn parse() {

    insta::assert_debug_snapshot!(
        to_fn_ref("[fn:1]").syntax,
-        @r###"
+        @r#"
    FN_REF@0..6
      L_BRACKET@0..1 "["
-      TEXT@1..3 "fn"
+      KEYWORD@1..3 "fn"
      COLON@3..4 ":"
-      TEXT@4..5 "1"
+      FN_LABEL@4..5 "1"
      R_BRACKET@5..6 "]"
-    "###
+    "#
    );

    insta::assert_debug_snapshot!(
        to_fn_ref("[fn:1:2]").syntax,
-        @r###"
+        @r#"
    FN_REF@0..8
      L_BRACKET@0..1 "["
-      TEXT@1..3 "fn"
+      KEYWORD@1..3 "fn"
      COLON@3..4 ":"
-      TEXT@4..5 "1"
+      FN_LABEL@4..5 "1"
      COLON@5..6 ":"
-      TEXT@6..7 "2"
+      FN_CONTENT@6..7
+        TEXT@6..7 "2"
      R_BRACKET@7..8 "]"
-    "###
+    "#
    );

    insta::assert_debug_snapshot!(
        to_fn_ref("[fn::2]").syntax,
-        @r###"
+        @r#"
    FN_REF@0..7
      L_BRACKET@0..1 "["
-      TEXT@1..3 "fn"
+      KEYWORD@1..3 "fn"
      COLON@3..4 ":"
-      TEXT@4..4 ""
+      FN_LABEL@4..4 ""
      COLON@4..5 ":"
-      TEXT@5..6 "2"
+      FN_CONTENT@5..6
+        TEXT@5..6 "2"
      R_BRACKET@6..7 "]"
-    "###
+    "#
    );

    insta::assert_debug_snapshot!(
        to_fn_ref("[fn::[]]").syntax,
-        @r###"
+        @r#"
    FN_REF@0..8
      L_BRACKET@0..1 "["
-      TEXT@1..3 "fn"
+      KEYWORD@1..3 "fn"
      COLON@3..4 ":"
-      TEXT@4..4 ""
+      FN_LABEL@4..4 ""
      COLON@4..5 ":"
-      TEXT@5..7 "[]"
+      FN_CONTENT@5..7
+        TEXT@5..7 "[]"
      R_BRACKET@7..8 "]"
-    "###
+    "#
    );

    let config = &ParseConfig::default();
--- a/src/syntax/mod.rs
+++ b/src/syntax/mod.rs
@ -199,6 +199,8 @@ pub enum SyntaxKind {
    COOKIE,
    RADIO_TARGET,
    FN_REF,
+    FN_LABEL,
+    FN_CONTENT,
    LATEX_FRAGMENT,
    MACROS,
    SNIPPET,
--- a/src/syntax/object.rs
+++ b/src/syntax/object.rs
@ -305,7 +305,7 @@ fn parse() {

    insta::assert_debug_snapshot!(
        t("~org-inlinetask-min-level~[fn:oiml:The default value of \n~org-inlinetask-min-level~ is =15=.]"),
-        @r###"
+        @r#"
    PARAGRAPH@0..93
      CODE@0..26
        TILDE@0..1 "~"
@ -313,23 +313,24 @@ fn parse() {
        TILDE@25..26 "~"
      FN_REF@26..93
        L_BRACKET@26..27 "["
-        TEXT@27..29 "fn"
+        KEYWORD@27..29 "fn"
        COLON@29..30 ":"
-        TEXT@30..34 "oiml"
+        FN_LABEL@30..34 "oiml"
        COLON@34..35 ":"
-        TEXT@35..57 "The default value of \n"
-        CODE@57..83
-          TILDE@57..58 "~"
-          TEXT@58..82 "org-inlinetask-min-level"
-          TILDE@82..83 "~"
-        TEXT@83..87 " is "
-        VERBATIM@87..91
-          EQUAL@87..88 "="
-          TEXT@88..90 "15"
-          EQUAL@90..91 "="
-        TEXT@91..92 "."
+        FN_CONTENT@35..92
+          TEXT@35..57 "The default value of \n"
+          CODE@57..83
+            TILDE@57..58 "~"
+            TEXT@58..82 "org-inlinetask-min-level"
+            TILDE@82..83 "~"
+          TEXT@83..87 " is "
+          VERBATIM@87..91
+            EQUAL@87..88 "="
+            TEXT@88..90 "15"
+            EQUAL@90..91 "="
+          TEXT@91..92 "."
        R_BRACKET@92..93 "]"
-    "###
+    "#
    );

    insta::assert_debug_snapshot!(
--- a/tests/html.rs
+++ b/tests/html.rs
@ -174,3 +174,18 @@ fn line_break() {
        @r###""<main><section><p>aa<br/>bb</p></section></main>""###
    );
 }
+
+#[test]
+fn footnote() {
+    insta::assert_debug_snapshot!(
+        Org::parse("[fn:1] In particular, the parser requires stars at column 0 to be\n").to_html(),
+        @r##""<main><section><aside class=\"footnote-definition\" ><a href=\"#footnote_1\" class=\"footnote-reference\" >[1]</a><span class=\"footnote-content\" id=\"footnote_1\" > In particular, the parser requires stars at column 0 to be</span></aside></section></main>""##
+    );
+    // "~org-inlinetask-min-level~[fn:oiml:The default value of \n~org-inlinetask-min-level~ is =15=.]"
+    insta::assert_debug_snapshot!(
+        Org::parse(
+            "~org-inlinetask-min-level~[fn:oiml:The default value of \n~org-inlinetask-min-level~ is =15=.]"
+        ).to_html(),
+        @r##""<main><section><p><code>org-inlinetask-min-level</code><a href=\"#footnote_oiml\" class=\"footnote-reference\">[oiml]</a><span class=\"footnote-content\" id=\"footnote_oiml\" >The default value of \n<code>org-inlinetask-min-level</code> is <code>15</code>.</span></p></section></main>""##
+    );
+}