Merge pull request #794 from streamich/fragment-converters

`Fragment` exports to other formats
streamich · Dec 21, 2024 · 7e3619d · 7e3619d
2 parents 37aacc1 + 8f5be6c
commit 7e3619d
Show file tree

Hide file tree

Showing 18 changed files with 1,253 additions and 286 deletions.
diff --git a/package.json b/package.json
@@ -136,7 +136,8 @@
     "hyperdyperid": "^1.2.0",
     "sonic-forest": "^1.0.3",
     "thingies": "^2.1.1",
-    "tree-dump": "^1.0.2"
+    "tree-dump": "^1.0.2",
+    "very-small-parser": "^1.8.0"
   },
   "devDependencies": {
     "@biomejs/biome": "^1.9.4",

diff --git a/src/json-crdt-extensions/peritext/__tests__/setup.ts b/src/json-crdt-extensions/peritext/__tests__/setup.ts
@@ -146,6 +146,28 @@ export const setupNumbersWithMultipleChunksAndDeletesKit = (): Kit => {
   });
 };
 
+export const runNumbersKitTestSuite = (runTestSuite: (getKit: () => Kit) => void) => {
+  describe('numbers "0123456789", no edits', () => {
+    runTestSuite(setupNumbersKit);
+  });
+
+  describe('numbers "0123456789", with default schema and tombstones', () => {
+    runTestSuite(setupNumbersWithTombstonesKit);
+  });
+
+  describe('numbers "0123456789", two RGA chunks', () => {
+    runTestSuite(setupNumbersWithTwoChunksKit);
+  });
+
+  describe('numbers "0123456789", with RGA split', () => {
+    runTestSuite(setupNumbersWithRgaSplitKit);
+  });
+
+  describe('numbers "0123456789", with multiple deletes', () => {
+    runTestSuite(setupNumbersWithMultipleChunksAndDeletesKit);
+  });
+};
+
 /**
  * Creates a Peritext instance with text "abcdefghijklmnopqrstuvwxyz", no edits.
  */
@@ -262,19 +284,19 @@ export const runAlphabetKitTestSuite = (runTestSuite: (getKit: () => Kit) => voi
   describe('basic alphabet', () => {
     runTestSuite(setupAlphabetKit);
   });
-  // describe('alphabet with two chunks', () => {
-  //   runTestSuite(setupAlphabetWithTwoChunksKit);
-  // });
-  // describe('alphabet with chunk split', () => {
-  //   runTestSuite(setupAlphabetChunkSplitKit);
-  // });
-  // describe('alphabet with deletes', () => {
-  //   runTestSuite(setupAlphabetWithDeletesKit);
-  // });
-  // describe('alphabet written in reverse', () => {
-  //   runTestSuite(setupAlphabetWrittenInReverse);
-  // });
-  // describe('alphabet written in reverse with deletes', () => {
-  //   runTestSuite(setupAlphabetWrittenInReverseWithDeletes);
-  // });
+  describe('alphabet with two chunks', () => {
+    runTestSuite(setupAlphabetWithTwoChunksKit);
+  });
+  describe('alphabet with chunk split', () => {
+    runTestSuite(setupAlphabetChunkSplitKit);
+  });
+  describe('alphabet with deletes', () => {
+    runTestSuite(setupAlphabetWithDeletesKit);
+  });
+  describe('alphabet written in reverse', () => {
+    runTestSuite(setupAlphabetWrittenInReverse);
+  });
+  describe('alphabet written in reverse with deletes', () => {
+    runTestSuite(setupAlphabetWrittenInReverseWithDeletes);
+  });
 };
diff --git a/src/json-crdt-extensions/peritext/block/Block.ts b/src/json-crdt-extensions/peritext/block/Block.ts
@@ -52,33 +52,6 @@ export class Block<Attr = unknown> extends Range implements IBlock, Printable, S
     return length ? path[length - 1] : '';
   }
 
-  // public htmlTag(): string {
-  //   const tag = this.tag();
-  //   switch (typeof tag) {
-  //     case 'string': return tag.toLowerCase();
-  //     case 'number': return SliceTypeName[tag] || 'div';
-  //     default: return 'div';
-  //   }
-  // }
-
-  // protected jsonMlNode(): JsonMlElement {
-  //   const props: Record<string, string> = {};
-  //   const node: JsonMlElement = ['div', props];
-  //   const tag = this.tag();
-  //   switch (typeof tag) {
-  //     case 'string':
-  //       node[0] = tag;
-  //       break;
-  //     case 'number':
-  //       const tag0 = SliceTypeName[tag];
-  //       if (tag0) node[0] = tag0; else props['data-tag'] = tag + '';
-  //       break;
-  //   }
-  //   const attr = this.attr();
-  //   if (attr !== undefined) props['data-attr'] = JSON.stringify(attr);
-  //   return node;
-  // }
-
   public attr(): Attr | undefined {
     return this.marker?.data() as Attr | undefined;
   }
@@ -112,59 +85,70 @@ export class Block<Attr = unknown> extends Range implements IBlock, Printable, S
     return new UndefEndIter(this.points0(withMarker));
   }
 
-  public tuples0(): UndefIterator<OverlayTuple<T>> {
+  protected tuples0(): UndefIterator<OverlayTuple<T>> {
     const overlay = this.txt.overlay;
-    const iterator = overlay.tuples0(this.marker);
+    const marker = this.marker;
+    const iterator = overlay.tuples0(marker);
     let closed = false;
     return () => {
       if (closed) return;
-      const pair = iterator();
-      if (!pair) return;
+      let pair = iterator();
+      while (!marker && pair && pair[1] && pair[1].cmpSpatial(this.start) < 0) pair = iterator();
+      if (!pair) return (closed = true), void 0;
       if (!pair[1] || pair[1] instanceof MarkerOverlayPoint) closed = true;
       return pair;
     };
   }
 
-  public tuples(): IterableIterator<OverlayTuple<T>> {
-    return new UndefEndIter(this.tuples0());
-  }
-
+  /**
+   * @todo Consider moving inline-related methods to {@link LeafBlock}.
+   */
   public texts0(): UndefIterator<Inline> {
     const txt = this.txt;
     const iterator = this.tuples0();
-    const blockStart = this.start;
-    const blockEnd = this.end;
+    const start = this.start;
+    const end = this.end;
+    const startIsMarker = txt.overlay.isMarker(start.id);
+    const endIsMarker = txt.overlay.isMarker(end.id);
     let isFirst = true;
     let next = iterator();
+    let closed = false;
     return () => {
+      if (closed) return;
       const pair = next;
       next = iterator();
       if (!pair) return;
-      const [p1, p2] = pair;
-      let start: Point = p1;
-      let end: Point = p2;
+      const [overlayPoint1, overlayPoint2] = pair;
+      let point1: Point = overlayPoint1;
+      let point2: Point = overlayPoint2;
       if (isFirst) {
         isFirst = false;
-        if (blockStart.cmpSpatial(p1) > 0) start = blockStart;
+        if (start.cmpSpatial(overlayPoint1) > 0) point1 = start;
+        if (startIsMarker) {
+          point1 = point1.clone();
+          point1.step(1);
+        }
       }
-      const isLast = !next;
-      if (isLast) if (blockEnd.cmpSpatial(p2) < 0) end = blockEnd;
-      return new Inline(txt, p1, p2, start, end);
+      if (!endIsMarker && end.cmpSpatial(overlayPoint2) < 0) {
+        closed = true;
+        point2 = end;
+      }
+      return new Inline(txt, overlayPoint1, overlayPoint2, point1, point2);
     };
   }
 
+  /**
+   * @todo Consider moving inline-related methods to {@link LeafBlock}.
+   */
   public texts(): IterableIterator<Inline> {
     return new UndefEndIter(this.texts0());
   }
 
   public text(): string {
     let str = '';
-    const iterator = this.texts0();
-    let inline = iterator();
-    while (inline) {
-      str += inline.text();
-      inline = iterator();
-    }
+    const children = this.children;
+    const length = children.length;
+    for (let i = 0; i < length; i++) str += children[i].text();
     return str;
   }
 
@@ -204,6 +188,7 @@ export class Block<Attr = unknown> extends Range implements IBlock, Printable, S
   public toStringName(): string {
     return 'Block';
   }
+
   protected toStringHeader(): string {
     const hash = `#${this.hash.toString(36).slice(-4)}`;
     const tag = this.path.map((step) => formatType(step)).join('.');

diff --git a/src/json-crdt-extensions/peritext/block/Fragment.ts b/src/json-crdt-extensions/peritext/block/Fragment.ts
@@ -71,24 +71,16 @@ export class Fragment extends Range implements Printable, Stateful {
   }
 
   protected build(): void {
-    const {end, root} = this;
+    const {root} = this;
     root.children = [];
     let parent = this.root;
     const txt = this.txt;
     const overlay = txt.overlay;
-    /**
-     * @todo This line always inserts a markerless block at the beginning of
-     * the fragment. But what happens if one actually exists?
-     */
-    this.insertBlock(parent, [CommonSliceType.p], void 0, void 0);
     const iterator = overlay.markerPairs0(this.start, this.end);
-    const checkEnd = !end.isAbsEnd();
     let pair: ReturnType<typeof iterator>;
     while ((pair = iterator())) {
       const [p1, p2] = pair;
-      if (!p1) break;
-      if (checkEnd && p1.cmpSpatial(end) > 0) break;
-      const type = p1.type();
+      const type = p1 ? p1.type() : CommonSliceType.p;
       const path = type instanceof Array ? type : [type];
       const block = this.insertBlock(parent, path, p1, p2);
       if (block.parent) parent = block.parent;

diff --git a/src/json-crdt-extensions/peritext/block/Inline.ts b/src/json-crdt-extensions/peritext/block/Inline.ts
@@ -238,11 +238,6 @@ export class Inline extends Range implements Printable {
     return texts;
   }
 
-  public text(): string {
-    const str = super.text();
-    return this.p1 instanceof MarkerOverlayPoint ? str.slice(1) : str;
-  }
-
   // ------------------------------------------------------------------- export
 
   public toJson(): PeritextMlNode {
@@ -273,12 +268,7 @@ export class Inline extends Range implements Printable {
   }
 
   public toString(tab: string = ''): string {
-    const str = this.text();
-    const truncate = str.length > 32;
-    const text = JSON.stringify(truncate ? str.slice(0, 32) : str) + (truncate ? ' …' : '');
-    const startFormatted = this.p1.toString(tab, true);
-    const range = this.p1.cmp(this.end) === 0 ? startFormatted : `${startFormatted} ↔ ${this.end.toString(tab, true)}`;
-    const header = `Inline ${range} ${text}`;
+    const header = `${super.toString(tab)}`;
     const attr = this.attr();
     const attrKeys = Object.keys(attr);
     const texts = this.texts();

diff --git a/src/json-crdt-extensions/peritext/block/LeafBlock.ts b/src/json-crdt-extensions/peritext/block/LeafBlock.ts
@@ -1,7 +1,7 @@
 import {printTree} from 'tree-dump/lib/printTree';
 import {Block} from './Block';
 import type {Path} from '@jsonjoy.com/json-pointer';
-import type {PeritextMlAttributes, PeritextMlElement, PeritextMlNode} from './types';
+import type {PeritextMlAttributes, PeritextMlElement} from './types';
 
 export interface IBlock<Attr = unknown> {
   readonly path: Path;
@@ -10,6 +10,12 @@ export interface IBlock<Attr = unknown> {
 }
 
 export class LeafBlock<Attr = unknown> extends Block<Attr> {
+  public text(): string {
+    let str = '';
+    for (let iterator = this.texts0(), inline = iterator(); inline; inline = iterator()) str += inline.text();
+    return str;
+  }
+
   // ------------------------------------------------------------------- export
 
   public toJson(): PeritextMlElement {

diff --git a/src/json-crdt-extensions/peritext/block/__tests__/Block.iteration.spec.ts b/src/json-crdt-extensions/peritext/block/__tests__/Block.iteration.spec.ts
@@ -93,32 +93,6 @@ describe('points', () => {
   });
 });
 
-describe('tuples', () => {
-  test('in markup-less document, returns a single pair', () => {
-    const {peritext} = setupHelloWorldKit();
-    peritext.refresh();
-    const blocks = peritext.blocks;
-    const block = blocks.root.children[0]!;
-    const pairs = [...block.tuples()];
-    expect(pairs.length).toBe(1);
-    expect(pairs[0]).toEqual([peritext.overlay.START, peritext.overlay.END]);
-  });
-
-  test('can iterate through all text chunks in two-block documents', () => {
-    const {peritext} = setupTwoBlockDocument();
-    expect(peritext.blocks.root.children.length).toBe(2);
-    const block1 = peritext.blocks.root.children[0]!;
-    const block2 = peritext.blocks.root.children[1]!;
-    const tuples1 = [...block1.tuples()];
-    const tuples2 = [...block2.tuples()];
-    expect(tuples1.length).toBe(3);
-    const text1 = tuples1.map(([p1, p2]) => new Inline(peritext, p1, p2, p1, p2).text()).join('');
-    const text2 = tuples2.map(([p1, p2]) => new Inline(peritext, p1, p2, p1, p2).text()).join('');
-    expect(text1).toBe('hello ');
-    expect(text2).toBe('world');
-  });
-});
-
 describe('texts', () => {
   test('in markup-less document', () => {
     const {peritext} = setupHelloWorldKit();