Nested clozes and increased cloze meta data (#2141)

* Nested clozes and increased cloze meta data

* Update contributors

* This reverts commit 3423df73f8.

* Update CONTRIBUTORS

* Formating

* Formating

* Formating

* Formating

* Formating

* Formating

* Formating

* Formating

* Code refactor

* Formating

* Formating

* Formating

* Formating and dead code

* Correct test case

* Remove Hint and Close storage of token string

* Update

* Formating

* Formating

* Formating

* Use write! instead of .push_str(&format).

* Formating

CONTRIBUTORS

@@ -106,6 +106,7 @@ Bart Louwers <bart.git@emeel.net>
 Sam Penny <github.com/sam1penny>
 Yutsuten <mateus.etto@gmail.com>
 Zoom <zoomrmc+git@gmail.com>
+TRIAEIOU <github.com/TRIAEIOU>
 Stefan Kangas <stefankangas@gmail.com>
 
 ********************

pylib/tests/test_models.py

@@ -185,41 +185,53 @@ def test_cloze():
     note["Text"] = "hello {{c1::world}}"
     assert col.addNote(note) == 1
     assert (
-        f'hello <span class="cloze" data-cloze="{encode_attribute("world")}">[...]</span>'
+        f'hello <span class="cloze" data-cloze="{encode_attribute("world")}" data-ordinal="1">[...]</span>'
         in note.cards()[0].question()
     )
-    assert 'hello <span class="cloze">world</span>' in note.cards()[0].answer()
+    assert (
+        'hello <span class="cloze" data-ordinal="1">world</span>'
+        in note.cards()[0].answer()
+    )
     # and with a comment
     note = col.new_note(m)
     note["Text"] = "hello {{c1::world::typical}}"
     assert col.addNote(note) == 1
     assert (
-        f'<span class="cloze" data-cloze="{encode_attribute("world")}">[typical]</span>'
+        f'<span class="cloze" data-cloze="{encode_attribute("world")}" data-ordinal="1">[typical]</span>'
         in note.cards()[0].question()
     )
-    assert '<span class="cloze">world</span>' in note.cards()[0].answer()
+    assert (
+        '<span class="cloze" data-ordinal="1">world</span>' in note.cards()[0].answer()
+    )
     # and with 2 clozes
     note = col.new_note(m)
     note["Text"] = "hello {{c1::world}} {{c2::bar}}"
     assert col.addNote(note) == 2
     (c1, c2) = note.cards()
     assert (
-        f'<span class="cloze" data-cloze="{encode_attribute("world")}">[...]</span> bar'
+        f'<span class="cloze" data-cloze="{encode_attribute("world")}" data-ordinal="1">[...]</span> <span class="cloze-inactive" data-ordinal="2">bar</span>'
         in c1.question()
     )
-    assert '<span class="cloze">world</span> bar' in c1.answer()
     assert (
-        f'world <span class="cloze" data-cloze="{encode_attribute("bar")}">[...]</span>'
+        '<span class="cloze" data-ordinal="1">world</span> <span class="cloze-inactive" data-ordinal="2">bar</span>'
+        in c1.answer()
+    )
+    assert (
+        f'<span class="cloze-inactive" data-ordinal="1">world</span> <span class="cloze" data-cloze="{encode_attribute("bar")}" data-ordinal="2">[...]</span>'
         in c2.question()
     )
-    assert 'world <span class="cloze">bar</span>' in c2.answer()
+    assert (
+        '<span class="cloze-inactive" data-ordinal="1">world</span> <span class="cloze" data-ordinal="2">bar</span>'
+        in c2.answer()
+    )
     # if there are multiple answers for a single cloze, they are given in a
     # list
     note = col.new_note(m)
     note["Text"] = "a {{c1::b}} {{c1::c}}"
     assert col.addNote(note) == 1
-    assert '<span class="cloze">b</span> <span class="cloze">c</span>' in (
-        note.cards()[0].answer()
+    assert (
+        '<span class="cloze" data-ordinal="1">b</span> <span class="cloze" data-ordinal="1">c</span>'
+        in (note.cards()[0].answer())
     )
     # if we add another cloze, a card should be generated
     cnt = col.card_count()
@@ -280,7 +292,9 @@ def test_cloze_mathjax():
     assert (
         note.cards()[0]
         .question()
-        .endswith(r'\(a\) <span class="cloze" data-cloze="b">[...]</span> \[ [...] \]')
+        .endswith(
+            r'\(a\) <span class="cloze" data-cloze="b" data-ordinal="1">[...]</span> \[ [...] \]'
+        )
     )
 
 
@@ -310,26 +324,26 @@ def test_chained_mods():
     col.models.update(m)
 
     note = col.newNote()
-    q1 = '<span style="color:red">phrase</span>'
-    a1 = "<b>sentence</b>"
-    q2 = '<span style="color:red">en chaine</span>'
-    a2 = "<i>chained</i>"
+    a1 = '<span style="color:red">phrase</span>'
+    h1 = "<b>sentence</b>"
+    a2 = '<span style="color:red">en chaine</span>'
+    h2 = "<i>chained</i>"
     note[
         "Text"
     ] = "This {{{{c1::{}::{}}}}} demonstrates {{{{c1::{}::{}}}}} clozes.".format(
-        q1,
         a1,
-        q2,
+        h1,
         a2,
+        h2,
     )
     assert col.addNote(note) == 1
     assert (
-        f'This <span class="cloze" data-cloze="{encode_attribute("phrase")}">[sentence]</span>'
-        f' demonstrates <span class="cloze" data-cloze="{encode_attribute("en chaine")}">[chained]</span> clozes.'
+        'This <span class="cloze" data-cloze="phrase" data-ordinal="1">[sentence]</span>'
+        f' demonstrates <span class="cloze" data-cloze="{encode_attribute("en chaine")}" data-ordinal="1">[chained]</span> clozes.'
         in note.cards()[0].question()
     )
     assert (
-        f'This <span class="cloze">phrase</span> demonstrates <span class="cloze">en chaine</span> clozes.'
+        'This <span class="cloze" data-ordinal="1">phrase</span> demonstrates <span class="cloze" data-ordinal="1">en chaine</span> clozes.'
         in note.cards()[0].answer()
     )
 

rslib/src/cloze.rs

@@ -1,26 +1,20 @@
 // Copyright: Ankitects Pty Ltd and contributors
 // License: GNU AGPL, version 3 or later; http://www.gnu.org/licenses/agpl.html
 
-use std::{borrow::Cow, collections::HashSet};
-
+use htmlescape::encode_attribute;
 use lazy_static::lazy_static;
+use nom::{
+    branch::alt,
+    bytes::complete::{tag, take_while},
+    combinator::map,
+    IResult,
+};
 use regex::{Captures, Regex};
+use std::{borrow::Cow, collections::HashSet, fmt::Write};
 
 use crate::{latex::contains_latex, template::RenderContext, text::strip_html_preserving_entities};
 
 lazy_static! {
-    static ref CLOZE: Regex = Regex::new(
-        r#"(?xsi)
-            \{\{
-            c(\d+)::    # 1 = cloze number
-            (.*?)       # 2 = clozed text
-            (?:
-              ::(.*?)   # 3 = optional hint
-            )?
-            \}\}
-        "#
-    )
-    .unwrap();
     static ref MATHJAX: Regex = Regex::new(
         r#"(?xsi)
             (\\[(\[])       # 1 = mathjax opening tag
@@ -31,99 +25,276 @@ lazy_static! {
     .unwrap();
 }
 
-mod cloze_caps {
-    // cloze ordinal
-    pub const ORD: usize = 1;
-    // the occluded text
-    pub const TEXT: usize = 2;
-    // optional hint
-    pub const HINT: usize = 3;
-}
-
 mod mathjax_caps {
     pub const OPENING_TAG: usize = 1;
     pub const INNER_TEXT: usize = 2;
     pub const CLOSING_TAG: usize = 3;
 }
 
-pub fn reveal_cloze_text(text: &str, cloze_ord: u16, question: bool) -> Cow<str> {
-    let mut cloze_ord_was_in_text = false;
+#[derive(Debug)]
+enum Token<'a> {
+    OpenCloze(u16),
+    Text(&'a str),
+    CloseCloze,
+}
 
-    let output = CLOZE.replace_all(text, |caps: &Captures| {
-        let captured_ord = caps
-            .get(cloze_caps::ORD)
-            .unwrap()
-            .as_str()
-            .parse()
-            .unwrap_or(0);
-
-        let text = caps.get(cloze_caps::TEXT).unwrap().as_str().to_owned();
-        if captured_ord != cloze_ord {
-            // other cloze deletions are unchanged
-            return text;
-        } else {
-            cloze_ord_was_in_text = true;
+/// Tokenize string
+fn tokenize(mut text: &str) -> impl Iterator<Item = Token> {
+    fn open_cloze(text: &str) -> IResult<&str, Token> {
+        // opening brackets and 'c'
+        let (text, _opening_brackets_and_c) = tag("{{c")(text)?;
+        // following number
+        let (text, digits) = take_while(|c: char| c.is_ascii_digit())(text)?;
+        let digits: u16 = match digits.parse() {
+            Ok(digits) => digits,
+            Err(_) => {
+                // not a valid number; fail to recognize
+                return Err(nom::Err::Error(nom::error::make_error(
+                    text,
+                    nom::error::ErrorKind::Digit,
+                )));
+            }
+        };
+        // ::
+        let (text, _colons) = tag("::")(text)?;
+        Ok((text, Token::OpenCloze(digits)))
     }
 
-        let text_attr;
-        let replacement;
+    fn close_cloze(text: &str) -> IResult<&str, Token> {
+        map(tag("}}"), |_| Token::CloseCloze)(text)
+    }
+
+    /// Match a run of text until an open/close marker is encountered.
+    fn normal_text(text: &str) -> IResult<&str, Token> {
+        if text.is_empty() {
+            return Err(nom::Err::Error(nom::error::make_error(
+                text,
+                nom::error::ErrorKind::Eof,
+            )));
+        }
+        let mut index = 0;
+        let mut other_token = alt((open_cloze, close_cloze));
+        while other_token(&text[index..]).is_err() && index < text.len() {
+            index += 1;
+        }
+        Ok((&text[index..], Token::Text(&text[0..index])))
+    }
+
+    std::iter::from_fn(move || {
+        if text.is_empty() {
+            None
+        } else {
+            let (remaining_text, token) =
+                alt((open_cloze, close_cloze, normal_text))(text).unwrap();
+            text = remaining_text;
+            Some(token)
+        }
+    })
+}
+
+#[derive(Debug)]
+enum TextOrCloze<'a> {
+    Text(&'a str),
+    Cloze(ExtractedCloze<'a>),
+}
+
+#[derive(Debug)]
+struct ExtractedCloze<'a> {
+    ordinal: u16,
+    nodes: Vec<TextOrCloze<'a>>,
+    hint: Option<&'a str>,
+}
+
+impl ExtractedCloze<'_> {
+    /// Return the cloze's hint, or "..." if none was provided.
+    fn hint(&self) -> &str {
+        self.hint.unwrap_or("...")
+    }
+
+    fn clozed_text(&self) -> Cow<str> {
+        // happy efficient path?
+        if self.nodes.len() == 1 {
+            if let TextOrCloze::Text(text) = self.nodes.last().unwrap() {
+                return (*text).into();
+            }
+        }
+
+        let mut buf = String::new();
+        for node in &self.nodes {
+            match node {
+                TextOrCloze::Text(text) => buf.push_str(text),
+                TextOrCloze::Cloze(cloze) => buf.push_str(&cloze.clozed_text()),
+            }
+        }
+
+        buf.into()
+    }
+}
+
+fn parse_text_with_clozes(text: &str) -> Vec<TextOrCloze<'_>> {
+    let mut open_clozes: Vec<ExtractedCloze> = vec![];
+    let mut output = vec![];
+    for token in tokenize(text) {
+        match token {
+            Token::OpenCloze(ordinal) => open_clozes.push(ExtractedCloze {
+                ordinal,
+                nodes: Vec::with_capacity(1), // common case
+                hint: None,
+            }),
+            Token::Text(mut text) => {
+                if let Some(cloze) = open_clozes.last_mut() {
+                    // extract hint if found
+                    if let Some((head, tail)) = text.split_once("::") {
+                        text = head;
+                        cloze.hint = Some(tail);
+                    }
+                    cloze.nodes.push(TextOrCloze::Text(text));
+                } else {
+                    output.push(TextOrCloze::Text(text));
+                }
+            }
+            Token::CloseCloze => {
+                // take the currently active cloze
+                if let Some(cloze) = open_clozes.pop() {
+                    let target = if let Some(outer_cloze) = open_clozes.last_mut() {
+                        // and place it into the cloze layer above
+                        &mut outer_cloze.nodes
+                    } else {
+                        // or the top level if no other clozes active
+                        &mut output
+                    };
+                    target.push(TextOrCloze::Cloze(cloze));
+                } else {
+                    // closing marker outside of any clozes
+                    output.push(TextOrCloze::Text("}}"))
+                }
+            }
+        }
+    }
+    output
+}
+
+fn reveal_cloze_text_in_nodes(
+    node: &TextOrCloze,
+    cloze_ord: u16,
+    question: bool,
+    output: &mut Vec<String>,
+) {
+    if let TextOrCloze::Cloze(cloze) = node {
+        if cloze.ordinal == cloze_ord {
             if question {
-            text_attr = format!(r#" data-cloze="{}""#, htmlescape::encode_attribute(&text));
-            // hint provided?
-            if let Some(hint) = caps.get(cloze_caps::HINT) {
-                replacement = format!("[{}]", hint.as_str());
+                output.push(cloze.hint().into())
             } else {
-                replacement = "[...]".to_string();
+                output.push(cloze.clozed_text().into())
             }
+        }
+        for node in &cloze.nodes {
+            reveal_cloze_text_in_nodes(node, cloze_ord, question, output);
+        }
+    }
+}
+
+fn reveal_cloze(
+    cloze: &ExtractedCloze,
+    cloze_ord: u16,
+    question: bool,
+    active_cloze_found_in_text: &mut bool,
+    buf: &mut String,
+) {
+    let active = cloze.ordinal == cloze_ord;
+    *active_cloze_found_in_text |= active;
+    match (question, active) {
+        (true, true) => {
+            // question side with active cloze; all inner content is elided
+            let mut content_buf = String::new();
+            for node in &cloze.nodes {
+                match node {
+                    TextOrCloze::Text(text) => content_buf.push_str(text),
+                    TextOrCloze::Cloze(cloze) => reveal_cloze(
+                        cloze,
+                        cloze_ord,
+                        question,
+                        active_cloze_found_in_text,
+                        &mut content_buf,
+                    ),
+                }
+            }
+            write!(
+                buf,
+                r#"<span class="cloze" data-cloze="{}" data-ordinal="{}">[{}]</span>"#,
+                encode_attribute(&content_buf),
+                cloze.ordinal,
+                cloze.hint()
+            )
+            .unwrap();
+        }
+        (false, true) => {
+            write!(
+                buf,
+                r#"<span class="cloze" data-ordinal="{}">"#,
+                cloze.ordinal
+            )
+            .unwrap();
+            for node in &cloze.nodes {
+                match node {
+                    TextOrCloze::Text(text) => buf.push_str(text),
+                    TextOrCloze::Cloze(cloze) => {
+                        reveal_cloze(cloze, cloze_ord, question, active_cloze_found_in_text, buf)
+                    }
+                }
+            }
+            buf.push_str("</span>");
+        }
+        (_, false) => {
+            // question or answer side inactive cloze; text shown, children may be active
+            write!(
+                buf,
+                r#"<span class="cloze-inactive" data-ordinal="{}">"#,
+                cloze.ordinal
+            )
+            .unwrap();
+            for node in &cloze.nodes {
+                match node {
+                    TextOrCloze::Text(text) => buf.push_str(text),
+                    TextOrCloze::Cloze(cloze) => {
+                        reveal_cloze(cloze, cloze_ord, question, active_cloze_found_in_text, buf)
+                    }
+                }
+            }
+            buf.push_str("</span>")
+        }
+    }
+}
+
+pub fn reveal_cloze_text(text: &str, cloze_ord: u16, question: bool) -> Cow<str> {
+    let mut buf = String::new();
+    let mut active_cloze_found_in_text = false;
+    for node in &parse_text_with_clozes(text) {
+        match node {
+            // top-level text is indiscriminately added
+            TextOrCloze::Text(text) => buf.push_str(text),
+            TextOrCloze::Cloze(cloze) => reveal_cloze(
+                cloze,
+                cloze_ord,
+                question,
+                &mut active_cloze_found_in_text,
+                &mut buf,
+            ),
+        }
+    }
+    if active_cloze_found_in_text {
+        buf.into()
     } else {
-            text_attr = "".to_string();
-            replacement = text;
-        }
-
-        format!(r#"<span class="cloze"{}>{}</span>"#, text_attr, replacement)
-    });
-
-    if !cloze_ord_was_in_text {
-        return "".into();
-    }
-
-    // if no cloze deletions are found, Anki returns an empty string
-    match output {
-        Cow::Borrowed(_) => "".into(),
-        other => other,
+        Cow::from("")
     }
 }
 
 pub fn reveal_cloze_text_only(text: &str, cloze_ord: u16, question: bool) -> Cow<str> {
-    CLOZE
-        .captures_iter(text)
-        .filter(|caps| {
-            let captured_ord = caps
-                .get(cloze_caps::ORD)
-                .unwrap()
-                .as_str()
-                .parse()
-                .unwrap_or(0);
-
-            captured_ord == cloze_ord
-        })
-        .map(|caps| {
-            let cloze = if question {
-                // hint provided?
-                if let Some(hint) = caps.get(cloze_caps::HINT) {
-                    hint.as_str()
-                } else {
-                    "..."
+    let mut output = Vec::new();
+    for node in &parse_text_with_clozes(text) {
+        reveal_cloze_text_in_nodes(node, cloze_ord, question, &mut output);
     }
-            } else {
-                caps.get(cloze_caps::TEXT).unwrap().as_str()
-            };
-
-            cloze
-        })
-        .collect::<Vec<_>>()
-        .join(", ")
-        .into()
+    output.join(", ").into()
 }
 
 /// If text contains any LaTeX tags, render the front and back
@@ -144,7 +315,9 @@ pub fn expand_clozes_to_reveal_latex(text: &str) -> String {
 }
 
 pub(crate) fn contains_cloze(text: &str) -> bool {
-    CLOZE.is_match(text)
+    parse_text_with_clozes(text)
+        .iter()
+        .any(|node| matches!(node, TextOrCloze::Cloze(_)))
 }
 
 pub fn cloze_numbers_in_string(html: &str) -> HashSet<u16> {
@@ -153,13 +326,18 @@ pub fn cloze_numbers_in_string(html: &str) -> HashSet<u16> {
     set
 }
 
+fn add_cloze_numbers_in_text_with_clozes(nodes: &[TextOrCloze], set: &mut HashSet<u16>) {
+    for node in nodes {
+        if let TextOrCloze::Cloze(cloze) = node {
+            set.insert(cloze.ordinal);
+            add_cloze_numbers_in_text_with_clozes(&cloze.nodes, set);
+        }
+    }
+}
+
 #[allow(clippy::implicit_hasher)]
 pub fn add_cloze_numbers_in_string(field: &str, set: &mut HashSet<u16>) {
-    for cap in CLOZE.captures_iter(field) {
-        if let Ok(n) = cap[1].parse() {
-            set.insert(n);
-        }
-    }
+    add_cloze_numbers_in_text_with_clozes(&parse_text_with_clozes(field), set)
 }
 
 fn strip_html_inside_mathjax(text: &str) -> Cow<str> {
@@ -232,6 +410,76 @@ mod test {
         );
     }
 
+    #[test]
+    fn nested_cloze_plain_text() {
+        assert_eq!(
+            strip_html(reveal_cloze_text("foo {{c1::bar {{c2::baz}}}}", 1, true).as_ref()),
+            "foo [...]"
+        );
+        assert_eq!(
+            strip_html(reveal_cloze_text("foo {{c1::bar {{c2::baz}}}}", 1, false).as_ref()),
+            "foo bar baz"
+        );
+        assert_eq!(
+            strip_html(reveal_cloze_text("foo {{c1::bar {{c2::baz}}::qux}}", 2, true).as_ref()),
+            "foo bar [...]"
+        );
+        assert_eq!(
+            strip_html(reveal_cloze_text("foo {{c1::bar {{c2::baz}}::qux}}", 2, false).as_ref()),
+            "foo bar baz"
+        );
+        assert_eq!(
+            strip_html(reveal_cloze_text("foo {{c1::bar {{c2::baz}}::qux}}", 1, true).as_ref()),
+            "foo [qux]"
+        );
+        assert_eq!(
+            strip_html(reveal_cloze_text("foo {{c1::bar {{c2::baz}}::qux}}", 1, false).as_ref()),
+            "foo bar baz"
+        );
+    }
+
+    #[test]
+    fn nested_cloze_html() {
+        assert_eq!(
+            cloze_numbers_in_string("{{c2::te{{c1::s}}}}t{{"),
+            vec![1, 2].into_iter().collect::<HashSet<u16>>()
+        );
+        assert_eq!(
+            reveal_cloze_text("foo {{c1::bar {{c2::baz}}}}", 1, true),
+            format!(
+                r#"foo <span class="cloze" data-cloze="{}" data-ordinal="1">[...]</span>"#,
+                htmlescape::encode_attribute(
+                    r#"bar <span class="cloze-inactive" data-ordinal="2">baz</span>"#
+                )
+            )
+        );
+        assert_eq!(
+            reveal_cloze_text("foo {{c1::bar {{c2::baz}}}}", 1, false),
+            r#"foo <span class="cloze" data-ordinal="1">bar <span class="cloze-inactive" data-ordinal="2">baz</span></span>"#
+        );
+        assert_eq!(
+            reveal_cloze_text("foo {{c1::bar {{c2::baz}}::qux}}", 2, true),
+            r#"foo <span class="cloze-inactive" data-ordinal="1">bar <span class="cloze" data-cloze="baz" data-ordinal="2">[...]</span></span>"#
+        );
+        assert_eq!(
+            reveal_cloze_text("foo {{c1::bar {{c2::baz}}::qux}}", 2, false),
+            r#"foo <span class="cloze-inactive" data-ordinal="1">bar <span class="cloze" data-ordinal="2">baz</span></span>"#
+        );
+        assert_eq!(
+            reveal_cloze_text("foo {{c1::bar {{c2::baz}}::qux}}", 1, true),
+            format!(
+                r#"foo <span class="cloze" data-cloze="{}" data-ordinal="1">[qux]</span>"#,
+                htmlescape::encode_attribute(
+                    r#"bar <span class="cloze-inactive" data-ordinal="2">baz</span>"#
+                )
+            )
+        );
+        assert_eq!(
+            reveal_cloze_text("foo {{c1::bar {{c2::baz}}::qux}}", 1, false),
+            r#"foo <span class="cloze" data-ordinal="1">bar <span class="cloze-inactive" data-ordinal="2">baz</span></span>"#
+        );
+    }
+
     #[test]
     fn mathjax_html() {
         // escaped angle brackets should be preserved

rslib/src/template_filters.rs

@@ -256,7 +256,7 @@ field</a>
         assert_eq!(strip_html(&cloze_filter(text, &ctx)).as_ref(), "[...] two");
         assert_eq!(
             cloze_filter(text, &ctx),
-            r#"<span class="cloze" data-cloze="one">[...]</span> two"#
+            r#"<span class="cloze" data-cloze="one" data-ordinal="1">[...]</span> <span class="cloze-inactive" data-ordinal="2">two</span>"#
         );
 
         ctx.card_ord = 1;