Editor iterators

src/json-crdt-extensions/peritext/editor/Editor.ts

@@ -3,10 +3,17 @@ import {CursorAnchor, SliceBehavior} from '../slice/constants';
 import {PersistedSlice} from '../slice/PersistedSlice';
 import {EditorSlices} from './EditorSlices';
 import {Chars} from '../constants';
+import {ChunkSlice} from '../util/ChunkSlice';
+import {contains, equal} from '../../../json-crdt-patch/clock';
+import {isLetter} from './util';
+import {Anchor} from '../rga/constants';
 import type {ITimestampStruct} from '../../../json-crdt-patch/clock';
 import type {Peritext} from '../Peritext';
 import type {SliceType} from '../slice/types';
 import type {MarkerSlice} from '../slice/MarkerSlice';
+import type {Chunk} from '../../../json-crdt/nodes/rga';
+import type {CharIterator, CharPredicate} from './types';
+import type {Point} from '../rga/Point';
 
 export class Editor<T = string> {
   public readonly saved: EditorSlices<T>;
@@ -96,6 +103,171 @@ export class Editor<T = string> {
     return true;
   }
 
+  /**
+   * Returns a forward iterator through visible text, one character at a time,
+   * starting from a given chunk and offset.
+   *
+   * @param chunk Chunk to start from.
+   * @param offset Offset in the chunk to start from.
+   * @returns The next visible character iterator.
+   */
+  public fwd0(chunk: undefined | Chunk<T>, offset: number): CharIterator<T> {
+    const str = this.txt.str;
+    return () => {
+      if (!chunk) return;
+      const span = chunk.span;
+      const offsetToReturn = offset;
+      const chunkToReturn = chunk;
+      if (offset >= span) return;
+      offset++;
+      if (offset >= span) {
+        offset = 0;
+        chunk = str.next(chunk);
+        while (chunk && chunk.del) chunk = str.next(chunk);
+      }
+      return new ChunkSlice<T>(chunkToReturn, offsetToReturn, 1);
+    };
+  }
+
+  /**
+   * Returns a forward iterator through visible text, one character at a time,
+   * starting from a given ID.
+   *
+   * @param id ID to start from.
+   * @param chunk Chunk to start from.
+   * @returns The next visible character iterator.
+   */
+  public fwd1(id: ITimestampStruct, chunk?: Chunk<T>): CharIterator<T> {
+    const str = this.txt.str;
+    const startFromStrRoot = equal(id, str.id);
+    if (startFromStrRoot) {
+      chunk = str.first();
+      while (chunk && chunk.del) chunk = str.next(chunk);
+      return this.fwd0(chunk, 0);
+    }
+    let offset: number = 0;
+    if (!chunk || !contains(chunk.id, chunk.span, id, 1)) {
+      chunk = str.findById(id);
+      if (!chunk) return () => undefined;
+      offset = id.time - chunk.id.time;
+    } else offset = id.time - chunk.id.time;
+    if (!chunk.del) return this.fwd0(chunk, offset);
+    while (chunk && chunk.del) chunk = str.next(chunk);
+    return this.fwd0(chunk, 0);
+  }
+
+  public bwd0(chunk: undefined | Chunk<T>, offset: number): CharIterator<T> {
+    const txt = this.txt;
+    const str = txt.str;
+    return () => {
+      if (!chunk || offset < 0) return;
+      const offsetToReturn = offset;
+      const chunkToReturn = chunk;
+      offset--;
+      if (offset < 0) {
+        chunk = str.prev(chunk);
+        while (chunk && chunk.del) chunk = str.prev(chunk);
+        if (chunk) offset = chunk.span - 1;
+      }
+      return new ChunkSlice(chunkToReturn, offsetToReturn, 1);
+    };
+  }
+
+  public bwd1(id: ITimestampStruct, chunk?: Chunk<T>): CharIterator<T> {
+    const str = this.txt.str;
+    const startFromStrRoot = equal(id, str.id);
+    if (startFromStrRoot) {
+      chunk = str.last();
+      while (chunk && chunk.del) chunk = str.prev(chunk);
+      return this.bwd0(chunk, chunk ? chunk.span - 1 : 0);
+    }
+    let offset: number = 0;
+    if (!chunk || !contains(chunk.id, chunk.span, id, 1)) {
+      chunk = str.findById(id);
+      if (!chunk) return () => undefined;
+      offset = id.time - chunk.id.time;
+    } else offset = id.time - chunk.id.time;
+    if (!chunk.del) return this.bwd0(chunk, offset);
+    while (chunk && chunk.del) chunk = str.prev(chunk);
+    return this.bwd0(chunk, chunk ? chunk.span - 1 : 0);
+  }
+
+  /**
+   * Skips a word in an arbitrary direction. A word is defined by the `predicate`
+   * function, which returns `true` if the character is part of the word.
+   *
+   * @param iterator Character iterator.
+   * @param predicate Predicate function to match characters, returns `true` if
+   *     the character is part of the word.
+   * @param firstLetterFound Whether the first letter has already been found. If
+   *     not, will skip any characters until the first letter, which is matched
+   *     by the `predicate` is found.
+   * @returns Point after the last character skipped.
+   */
+  private skipWord(
+    iterator: CharIterator<T>,
+    predicate: CharPredicate<string>,
+    firstLetterFound: boolean,
+  ): Point<T> | undefined {
+    let next: ChunkSlice<T> | undefined;
+    let prev: ChunkSlice<T> | undefined;
+    while ((next = iterator())) {
+      const char = (next.view() as string)[0];
+      if (firstLetterFound) {
+        if (!predicate(char)) break;
+      } else if (predicate(char)) firstLetterFound = true;
+      prev = next;
+    }
+    if (!prev) return;
+    return this.txt.point(prev.id(), Anchor.After);
+  }
+
+  /**
+   * Skips a word forward. A word is defined by the `predicate` function, which
+   * returns `true` if the character is part of the word.
+   *
+   * @param point Point from which to start skipping.
+   * @param predicate Character class to skip.
+   * @param firstLetterFound Whether the first letter has already been found. If
+   *        not, will skip any characters until the first letter, which is
+   *        matched by the `predicate` is found.
+   * @returns Point after the last character skipped.
+   */
+  public fwdSkipWord(
+    point: Point<T>,
+    predicate: CharPredicate<string> = isLetter,
+    firstLetterFound: boolean = false,
+  ): Point<T> {
+    const firstChar = point.rightChar();
+    if (!firstChar) return point;
+    const fwd = this.fwd1(firstChar.id(), firstChar.chunk);
+    return this.skipWord(fwd, predicate, firstLetterFound) || point;
+  }
+
+  /**
+   * Skips a word backward. A word is defined by the `predicate` function, which
+   * returns `true` if the character is part of the word.
+   *
+   * @param point Point from which to start skipping.
+   * @param predicate Character class to skip.
+   * @param firstLetterFound Whether the first letter has already been found. If
+   *        not, will skip any characters until the first letter, which is
+   *        matched by the `predicate` is found.
+   * @returns Point after the last character skipped.
+   */
+  public bwdSkipWord(
+    point: Point<T>,
+    predicate: CharPredicate<string> = isLetter,
+    firstLetterFound: boolean = false,
+  ): Point<T> {
+    const firstChar = point.leftChar();
+    if (!firstChar) return point;
+    const bwd = this.bwd1(firstChar.id(), firstChar.chunk);
+    const endPoint = this.skipWord(bwd, predicate, firstLetterFound);
+    if (endPoint) endPoint.anchor = Anchor.Before;
+    return endPoint || point;
+  }
+
   /** @deprecated use `.saved.insStack` */
   public insStackSlice(type: SliceType, data?: unknown | ITimestampStruct): PersistedSlice<T> {
     const range = this.cursor.range();

src/json-crdt-extensions/peritext/editor/__tests__/Editor-cursor.spec.ts

@@ -0,0 +1,109 @@
+import {Model} from '../../../../json-crdt/model';
+import {Peritext} from '../../Peritext';
+import {Point} from '../../rga/Point';
+import {Editor} from '../Editor';
+
+const setup = (insert = (editor: Editor) => editor.insert('Hello world!'), sid?: number) => {
+  const model = Model.withLogicalClock(sid);
+  model.api.root({
+    text: '',
+    slices: [],
+  });
+  const peritext = new Peritext(model, model.api.str(['text']).node, model.api.arr(['slices']).node);
+  const editor = peritext.editor;
+  insert(editor);
+  return {model, peritext, editor};
+};
+
+describe('.fwdSkipWord()', () => {
+  test('can go to the end of a word', () => {
+    const {editor} = setup((editor) => editor.insert('Hello world!'));
+    editor.cursor.setAt(0);
+    const point = editor.fwdSkipWord(editor.cursor.end);
+    editor.cursor.end.set(point!);
+    expect(editor.cursor.text()).toBe('Hello');
+  });
+
+  test('can skip whitespace between words', () => {
+    const {editor} = setup((editor) => editor.insert('Hello world!'));
+    editor.cursor.setAt(5);
+    const point = editor.fwdSkipWord(editor.cursor.end);
+    editor.cursor.end.set(point!);
+    expect(editor.cursor.text()).toBe(' world');
+  });
+
+  test('skipping stops before exclamation mark', () => {
+    const {editor} = setup((editor) => editor.insert('Hello world!'));
+    editor.cursor.setAt(6);
+    const point = editor.fwdSkipWord(editor.cursor.end);
+    editor.cursor.end.set(point!);
+    expect(editor.cursor.text()).toBe('world');
+  });
+
+  test('can skip to the end of string', () => {
+    const {editor} = setup((editor) => editor.insert('Hello world!'));
+    editor.cursor.setAt(11);
+    const point = editor.fwdSkipWord(editor.cursor.end);
+    expect(point instanceof Point).toBe(true);
+    editor.cursor.end.set(point!);
+    expect(editor.cursor.text()).toBe('!');
+  });
+
+  test('can skip various character classes', () => {
+    const {editor} = setup((editor) =>
+      editor.insert("const {editor} = setup(editor => editor.insert('Hello world!'));"),
+    );
+    editor.cursor.setAt(0);
+    const move = (): string => {
+      const point = editor.fwdSkipWord(editor.cursor.end);
+      if (point) editor.cursor.end.set(point);
+      return editor.cursor.text();
+    };
+    expect(move()).toBe('const');
+    expect(move()).toBe('const {editor');
+    expect(move()).toBe('const {editor} = setup');
+    expect(move()).toBe('const {editor} = setup(editor');
+    expect(move()).toBe('const {editor} = setup(editor => editor');
+    expect(move()).toBe('const {editor} = setup(editor => editor.insert');
+    expect(move()).toBe("const {editor} = setup(editor => editor.insert('Hello");
+    expect(move()).toBe("const {editor} = setup(editor => editor.insert('Hello world");
+    expect(move()).toBe("const {editor} = setup(editor => editor.insert('Hello world!'));");
+  });
+});
+
+describe('.bwdSkipWord()', () => {
+  test('can skip over simple text.', () => {
+    const {editor} = setup((editor) => editor.insert('Hello world!\nfoo bar baz'));
+    editor.cursor.setAt(editor.txt.str.length());
+    const move = (): string => {
+      const point = editor.bwdSkipWord(editor.cursor.start);
+      if (point) editor.cursor.start.set(point);
+      return editor.cursor.text();
+    };
+    expect(move()).toBe('baz');
+    expect(move()).toBe('bar baz');
+    expect(move()).toBe('foo bar baz');
+    expect(move()).toBe('world!\nfoo bar baz');
+    expect(move()).toBe('Hello world!\nfoo bar baz');
+  });
+
+  test('can skip various character classes', () => {
+    const {editor} = setup((editor) =>
+      editor.insert("const {editor} = setup(editor => editor.insert('Hello world!'));"),
+    );
+    editor.cursor.setAt(editor.txt.str.length());
+    const move = (): string => {
+      const point = editor.bwdSkipWord(editor.cursor.start);
+      if (point) editor.cursor.start.set(point);
+      return editor.cursor.text();
+    };
+    expect(move()).toBe("world!'));");
+    expect(move()).toBe("Hello world!'));");
+    expect(move()).toBe("insert('Hello world!'));");
+    expect(move()).toBe("editor.insert('Hello world!'));");
+    expect(move()).toBe("editor => editor.insert('Hello world!'));");
+    expect(move()).toBe("setup(editor => editor.insert('Hello world!'));");
+    expect(move()).toBe("editor} = setup(editor => editor.insert('Hello world!'));");
+    expect(move()).toBe("const {editor} = setup(editor => editor.insert('Hello world!'));");
+  });
+});