Support for Anki Cloze Syntax

src/AnkiParser.ts

@@ -0,0 +1,296 @@
+/*
+Source / Inspiration: https://github.com/ankitects/anki/blob/80d807e08a6d3148f973829c48fe633a760546c5/rslib/src/cloze.rs
+
+The "BNF" for this.
+
+<number>       ::= [0-9]+
+<separator>    ::= ::
+<c_open>       ::= {{c<number>
+<c_close>      ::= }}
+<text>         ::= .*
+<text_or_card> ::= <text> | <card>
+<card_inner>   ::= <text> | <card> | <text_or_card>?<separator><text_or_card>?
+<card>         ::= <text>?<c_open><separator><card_inner><separator><c_close><text>?
+*/
+
+import { CardFrontBack, QuestionType_ClozeUtil } from "./QuestionType";
+
+// The types of tokens.
+enum TokenKind {
+    Open = "C_OPEN",
+    Close = "C_CLOSE",
+    Separator = "SEPARATOR",
+    Text = "TEXT",
+}
+
+interface OpenToken {
+    kind: TokenKind.Open;
+    ordinal: number;
+    i: number;
+}
+interface CloseToken {
+    kind: TokenKind.Close;
+    i: number;
+}
+interface SeparatorToken {
+    kind: TokenKind.Separator;
+    i: number;
+}
+interface TextToken {
+    kind: TokenKind.Text;
+    text: string;
+    i: number;
+}
+
+type Token = OpenToken | CloseToken | SeparatorToken | TextToken;
+
+// Returns the "Anki tokens" of a given text, given an anchor.
+// This does not return the TEXT token.
+function anki_tokens(
+    input: string,
+    i: number,
+): OpenToken | CloseToken | SeparatorToken | undefined {
+    // The following statements are sorted high -> low `i`.
+    // Avoids any similar statements being mixed with each other.
+
+    // C_OPEN
+    if (input.slice(i, i + 3) === "{{c") {
+        i += 3;
+
+        // Extract the number
+        let num = "";
+        while (input[i] !== ":") {
+            num += input[i];
+            i++;
+        }
+
+        // Attempt to parse the number
+        const number = parseInt(num);
+        if (isNaN(number)) return;
+
+        // Return the object
+        return {
+            kind: TokenKind.Open,
+            ordinal: number,
+            i,
+        } satisfies OpenToken;
+    }
+    // SEPARATOR
+    else if (input.slice(i, i + 2) === "::") {
+        return {
+            kind: TokenKind.Separator,
+            i: i + 2,
+        } satisfies SeparatorToken;
+    }
+    // C_CLOSE
+    else if (input.slice(i, i + 2) === "}}") {
+        return {
+            kind: TokenKind.Close,
+            i: i + 2,
+        } satisfies CloseToken;
+    }
+}
+
+// Fully gets the "Anki tokens" of a given text.
+// This does return the TEXT token.
+// NOTE: This function does not error and does not check for invalid syntax
+function text_anki_tokens(input: string): Token[] {
+    // Vars
+    const tokens: Token[] = [];
+    let i = 0;
+    let buf_start;
+
+    // Loop through the input
+    while (i < input.length) {
+        // Grab the token at position `i`
+        const token = anki_tokens(input, i);
+        if (!token) {
+            // No token found, wait until the next token...
+            if (buf_start === undefined) buf_start = i;
+            i++;
+            continue;
+        }
+
+        // Assume any "in between text" is TEXT
+        if (buf_start !== undefined) {
+            tokens.push({
+                kind: TokenKind.Text,
+                text: input.slice(buf_start, i),
+                i,
+            });
+
+            // Reset for next
+            buf_start = undefined;
+        }
+
+        // Add the token
+        tokens.push(token);
+        i = token.i;
+    }
+
+    // We still have some text left over
+    if (buf_start !== undefined) {
+        tokens.push({
+            kind: TokenKind.Text,
+            text: input.slice(buf_start),
+            i: buf_start,
+        });
+    }
+
+    // Return the tokens
+    return tokens;
+}
+
+interface ExtractedCloze {
+    kind: "cloze";
+    ordinal: number;
+    nodes: (TextToken | ExtractedCloze)[];
+    hint?: string;
+}
+
+/// Parses the tokens into a tree of clozes and text.
+function parse_anki_tokens(tokens: Token[]): (ExtractedCloze | string)[] {
+    const open_clozes: ExtractedCloze[] = [];
+    const output: (ExtractedCloze | string)[] = [];
+
+    // Loop through the tokens
+    for (let i = 0; i < tokens.length; i++) {
+        // Make sure the token exists
+        const token = tokens[i];
+        if (!token) break;
+
+        switch (token.kind) {
+            // Open a cloze
+            case TokenKind.Open: {
+                open_clozes.push({
+                    kind: "cloze",
+                    ordinal: token.ordinal,
+                    nodes: [],
+                });
+                break;
+            }
+            // Add text to the cloze
+            case TokenKind.Text: {
+                const last_open_cloze = open_clozes[open_clozes.length - 1];
+
+                if (!last_open_cloze) {
+                    output.push(token.text);
+                    break;
+                }
+
+                // Look ahead to see whether is a hint or not
+                if (
+                    tokens[i - 2].kind != TokenKind.Open &&
+                    tokens[i - 1].kind == TokenKind.Separator &&
+                    tokens[i + 1].kind == TokenKind.Close
+                ) {
+                    last_open_cloze.hint = token.text;
+                    break;
+                }
+
+                last_open_cloze.nodes.push(token);
+                break;
+            }
+            // Close a cloze
+            case TokenKind.Close: {
+                const cloze = open_clozes.pop();
+                if (!cloze) {
+                    output.push("}}");
+                    break;
+                }
+                const last_cloze = open_clozes[open_clozes.length - 1];
+
+                if (last_cloze) {
+                    last_cloze.nodes.push(cloze);
+                } else {
+                    output.push(cloze);
+                }
+
+                break;
+            }
+            // Ignore
+            case TokenKind.Separator: {
+                break;
+            }
+        }
+    }
+
+    return output;
+}
+
+// Reveal / hides a singular cloze ordinal
+function reveal_cloze(
+    cloze: ExtractedCloze,
+    cloze_ord: number,
+    show_question: boolean,
+    data: { buf: string },
+) {
+    // We want to hide the cloze
+    const ord_match = cloze.ordinal === cloze_ord;
+    if (ord_match && show_question) {
+        data.buf += QuestionType_ClozeUtil.renderClozeFront(cloze.hint);
+        return;
+    }
+
+    // Show all of the text inside the nodes
+    for (const node of cloze.nodes) {
+        if (node.kind == TokenKind.Text) {
+            data.buf += ord_match ? QuestionType_ClozeUtil.renderClozeBack(node.text) : node.text;
+        } else if (node.kind === "cloze") {
+            reveal_cloze(node, cloze_ord, show_question, data);
+        }
+    }
+}
+
+// Reveal / hides all cloze ordinals
+function reveal_cloze_text(
+    tokens: (ExtractedCloze | string)[],
+    cloze_ord: number,
+    question: boolean,
+): string {
+    const data = { buf: "" };
+
+    for (const node of tokens) {
+        if (typeof node === "string") {
+            data.buf += node;
+        } else {
+            reveal_cloze(node, cloze_ord, question, data);
+        }
+    }
+
+    return data.buf;
+}
+
+// Recursively loops to find all of the ordinals
+function get_ordinals(tokens: (ExtractedCloze | string)[], ordinals: Set<number>) {
+    for (const token of tokens) {
+        if (typeof token === "string") continue;
+        ordinals.add(token.ordinal);
+        get_ordinals(token.nodes.filter((x) => x.kind == "cloze") as ExtractedCloze[], ordinals);
+    }
+}
+
+// This can probably be done better.
+function tokens_to_cards(tokens: (ExtractedCloze | string)[]): CardFrontBack[] {
+    // Grab each ordinal
+    const ordinals = new Set<number>();
+    get_ordinals(tokens, ordinals);
+
+    // Create the cards
+    const cards: CardFrontBack[] = [];
+    for (const ordinal of ordinals) {
+        cards.push(
+            new CardFrontBack(
+                reveal_cloze_text(tokens, ordinal, true),
+                reveal_cloze_text(tokens, ordinal, false),
+            ),
+        );
+    }
+
+    return cards;
+}
+
+// Converts text to cards (assumes the regex has already been used)
+export function text_to_cards(text: string): CardFrontBack[] {
+    return tokens_to_cards(parse_anki_tokens(text_anki_tokens(text)));
+}

src/Question.ts

@@ -13,6 +13,7 @@ export enum CardType {
     MultiLineBasic,
     MultiLineReversed,
     Cloze,
+    AnkiCloze,
 }
 
 //

src/QuestionType.ts

@@ -1,3 +1,4 @@
+import { text_to_cards } from "./AnkiParser";
 import { CardType } from "./Question";
 import { SRSettings } from "./settings";
 
@@ -136,9 +137,15 @@ class QuestionType_Cloze implements IQuestionTypeHandler {
     }
 }
 
+class QuestionType_AnkiCloze implements IQuestionTypeHandler {
+    expand(questionText: string, _settings: SRSettings): CardFrontBack[] {
+        return text_to_cards(questionText);
+    }
+}
+
 export class QuestionType_ClozeUtil {
-    static renderClozeFront(): string {
-        return "<span style='color:#2196f3'>[...]</span>";
+    static renderClozeFront(hint?: string): string {
+        return `<span style='color:#2196f3'>[${hint || "..."}]</span>`;
     }
 
     static renderClozeBack(str: string): string {
@@ -165,6 +172,9 @@ export class QuestionTypeFactory {
             case CardType.Cloze:
                 handler = new QuestionType_Cloze();
                 break;
+            case CardType.AnkiCloze:
+                handler = new QuestionType_AnkiCloze();
+                break;
         }
         return handler;
     }

src/parser.ts

@@ -1,5 +1,8 @@
 import { CardType } from "./Question";
 
+const ANKI_RE =
+    /((?:.+\n)*(?:.*\{\{[^{}]*::[^{}]*\}\}.*)(?:\n(?:^.{1,3}$|^.{4}(?<!<!--).*))*)(\n<!--SR:.+-->)?/gm;
+
 /**
  * Returns flashcards found in `text`
  *
@@ -25,6 +28,14 @@ export function parse(
     let cardType: CardType | null = null;
     let lineNo = 0;
 
+    let match;
+    while ((match = ANKI_RE.exec(text)) !== null) {
+        const upTo = text.substring(0, match.index);
+        cards.push([CardType.AnkiCloze, match[0], upTo.split("\n").length]);
+
+        text = upTo + text.substring(match.index + match[0].length);
+    }
+
     const lines: string[] = text.replaceAll("\r\n", "\n").split("\n");
     for (let i = 0; i < lines.length; i++) {
         const currentLine = lines[i];

tests/unit/parser.test.ts

@@ -161,6 +161,10 @@ test("Test parsing of cloze cards", () => {
     ]);
 });
 
+test("Testing parsing of anki cloze cards", () => {
+    // TODO!
+});
+
 test("Test parsing of a mix of card types", () => {
     expect(
         parse(