refactor: change parser to parse raw text content

2025-12-18 22:44:26 +01:00 · 2023-04-06 19:19:09 +09:00
parent 8e1714f476
commit eda4382524
3 changed files with 153 additions and 161 deletions
--- a/src/components/textNote/TextNoteContentDisplay.tsx
+++ b/src/components/textNote/TextNoteContentDisplay.tsx
@@ -1,5 +1,5 @@
 import { For } from 'solid-js';
-import parseTextNote, { type ParsedTextNoteNode } from '@/core/parseTextNote';
+import parseTextNote, { resolveTagReference, type ParsedTextNoteNode } from '@/core/parseTextNote';
 import type { Event as NostrEvent } from 'nostr-tools';
 import PlainTextDisplay from '@/components/textNote/PlainTextDisplay';
 import MentionedUserDisplay from '@/components/textNote/MentionedUserDisplay';
@@ -21,19 +21,23 @@ const TextNoteContentDisplay = (props: TextNoteContentDisplayProps) => {
  const { config } = useConfig();
  const event = () => eventWrapper(props.event);
  return (
-    <For each={parseTextNote(props.event)}>
+    <For each={parseTextNote(props.event.content)}>
      {(item: ParsedTextNoteNode) => {
        if (item.type === 'PlainText') {
          return <PlainTextDisplay plainText={item} />;
        }
-        if (item.type === 'MentionedUser') {
-          return <MentionedUserDisplay pubkey={item.pubkey} />;
+        if (item.type === 'TagReference') {
+          const resolved = resolveTagReference(item, props.event);
+          if (resolved == null) return null;
+          if (resolved.type === 'MentionedUser') {
+            return <MentionedUserDisplay pubkey={resolved.pubkey} />;
          }
-        if (item.type === 'MentionedEvent') {
+          if (resolved.type === 'MentionedEvent') {
            if (props.embedding) {
-            return <MentionedEventDisplay mentionedEvent={item} />;
+              return <MentionedEventDisplay mentionedEvent={resolved} />;
+            }
+            return <EventLink eventId={resolved.eventId} />;
          }
-          return <EventLink eventId={item.eventId} />;
        }
        if (item.type === 'Bech32Entity') {
          if (item.data.type === 'note' && props.embedding) {
--- a/src/core/parseTextNote.test.ts
+++ b/src/core/parseTextNote.test.ts
@@ -1,7 +1,8 @@
 import assert from 'assert';
 import { describe, it } from 'vitest';
+import { type Event as NostrEvent } from 'nostr-tools';

-import parseTextNote, { type ParsedTextNoteNode } from './parseTextNote';
+import parseTextNote, { resolveTagReference, type ParsedTextNoteNode, TagReference } from './parseTextNote';

 describe('parseTextNote', () => {
  /*
@@ -21,15 +22,7 @@ describe('parseTextNote', () => {
  */

  it('should parse text note with the url with hash', () => {
-    const parsed = parseTextNote({
-      id: '',
-      sig: '',
-      kind: 1,
-      content: 'this is url\nhttps://github.com/syusui-s/rabbit/#readme #rabbit',
-      tags: [],
-      created_at: 1678377182,
-      pubkey: '9366708117c4a7edf9178acdce538c95059b9eb3394808cdd90564094172d972',
-    });
+    const parsed = parseTextNote('this is url\nhttps://github.com/syusui-s/rabbit/#readme #rabbit');

    const expected: ParsedTextNoteNode[] = [
      { type: 'PlainText', content: 'this is url\n' },
@@ -42,15 +35,7 @@ describe('parseTextNote', () => {
  });

  it('should parse text note with the url with hash and hashtag', () => {
-    const parsed = parseTextNote({
-      id: '',
-      sig: '',
-      kind: 1,
-      content: 'this is url\nhttps://github.com/syusui-s/rabbit/#readme #rabbit',
-      tags: [],
-      created_at: 1678377182,
-      pubkey: '9366708117c4a7edf9178acdce538c95059b9eb3394808cdd90564094172d972',
-    });
+    const parsed = parseTextNote('this is url\nhttps://github.com/syusui-s/rabbit/#readme #rabbit');

    const expected: ParsedTextNoteNode[] = [
      { type: 'PlainText', content: 'this is url\n' },
@@ -63,15 +48,7 @@ describe('parseTextNote', () => {
  });

  it('should parse text note which includes punycode URL', () => {
-    const parsed = parseTextNote({
-      id: '',
-      sig: '',
-      kind: 1,
-      content: 'This is Japanese domain: https://xn--p8j9a0d9c9a.xn--q9jyb4c/',
-      tags: [],
-      created_at: 1678377182,
-      pubkey: '9366708117c4a7edf9178acdce538c95059b9eb3394808cdd90564094172d972',
-    });
+    const parsed = parseTextNote('This is Japanese domain: https://xn--p8j9a0d9c9a.xn--q9jyb4c/');

    const expected: ParsedTextNoteNode[] = [
      { type: 'PlainText', content: 'This is Japanese domain: ' },
@@ -82,16 +59,7 @@ describe('parseTextNote', () => {
  });

  it('should parse text note which includes image URLs', () => {
-    const parsed = parseTextNote({
-      id: '',
-      sig: '',
-      kind: 1,
-      content:
-        'https://i.gyazo.com/8f177b9953fdb9513ad00d0743d9c608.png\nhttps://i.gyazo.com/346ad7260f6a999720c2d13317ff795f.jpg',
-      tags: [],
-      created_at: 1678377182,
-      pubkey: '9366708117c4a7edf9178acdce538c95059b9eb3394808cdd90564094172d972',
-    });
+    const parsed = parseTextNote('https://i.gyazo.com/8f177b9953fdb9513ad00d0743d9c608.png\nhttps://i.gyazo.com/346ad7260f6a999720c2d13317ff795f.jpg');

    const expected: ParsedTextNoteNode[] = [
      { type: 'URL', content: 'https://i.gyazo.com/8f177b9953fdb9513ad00d0743d9c608.png' },
@@ -103,15 +71,7 @@ describe('parseTextNote', () => {
  });

  it('should parse text note which includes URL with + symbol', () => {
-    const parsed = parseTextNote({
-      id: '',
-      sig: '',
-      kind: 1,
-      content: 'this is my page\nhttps://example.com/abc+def?q=ghi+jkl#lmn+opq',
-      tags: [],
-      created_at: 1678377182,
-      pubkey: '9366708117c4a7edf9178acdce538c95059b9eb3394808cdd90564094172d972',
-    });
+    const parsed = parseTextNote('this is my page\nhttps://example.com/abc+def?q=ghi+jkl#lmn+opq');

    const expected: ParsedTextNoteNode[] = [
      { type: 'PlainText', content: 'this is my page\n' },
@@ -123,15 +83,7 @@ describe('parseTextNote', () => {

  //
  it('should parse text note which includes URL with + symbol', () => {
-    const parsed = parseTextNote({
-      id: '',
-      sig: '',
-      kind: 1,
-      content: 'I wrote this page\nhttps://example.com/test(test)?q=(q)#(h)',
-      tags: [],
-      created_at: 1678377182,
-      pubkey: '9366708117c4a7edf9178acdce538c95059b9eb3394808cdd90564094172d972',
-    });
+    const parsed = parseTextNote('I wrote this page\nhttps://example.com/test(test)?q=(q)#(h)');

    const expected: ParsedTextNoteNode[] = [
      { type: 'PlainText', content: 'I wrote this page\n' },
@@ -145,15 +97,7 @@ describe('parseTextNote', () => {
  });

  it('should parse text note which includes wss URL', () => {
-    const parsed = parseTextNote({
-      id: '',
-      sig: '',
-      kind: 1,
-      content: 'this is my using relays: wss://relay.damus.io, wss://relay.snort.social',
-      tags: [],
-      created_at: 1678377182,
-      pubkey: '9366708117c4a7edf9178acdce538c95059b9eb3394808cdd90564094172d972',
-    });
+    const parsed = parseTextNote('this is my using relays: wss://relay.damus.io, wss://relay.snort.social');

    const expected: ParsedTextNoteNode[] = [
      { type: 'PlainText', content: 'this is my using relays: ' },
@@ -166,50 +110,20 @@ describe('parseTextNote', () => {
  });

  it('should parse text note with pubkey mentions', () => {
-    const parsed = parseTextNote({
-      id: '',
-      sig: '',
-      kind: 1,
-      content: 'this is pubkey\n#[0] #[1]',
-      tags: [
-        ['p', '9366708117c4a7edf9178acdce538c95059b9eb3394808cdd90564094172d972'],
-        ['p', '80d3a41d8a00679c0105faac2cdf7643c9ba26835cff096bf7f9c7a0eee8c8fc'],
-      ],
-      created_at: 1678377182,
-      pubkey: '9366708117c4a7edf9178acdce538c95059b9eb3394808cdd90564094172d972',
-    });
+    const parsed = parseTextNote('this is pubkey\n#[0] #[1]');

    const expected: ParsedTextNoteNode[] = [
      { type: 'PlainText', content: 'this is pubkey\n' },
-      {
-        type: 'MentionedUser',
-        tagIndex: 0,
-        content: '#[0]',
-        pubkey: '9366708117c4a7edf9178acdce538c95059b9eb3394808cdd90564094172d972',
-      },
+      { type: 'TagReference', tagIndex: 0, content: '#[0]'},
      { type: 'PlainText', content: ' ' },
-      {
-        type: 'MentionedUser',
-        tagIndex: 1,
-        content: '#[1]',
-        pubkey: '80d3a41d8a00679c0105faac2cdf7643c9ba26835cff096bf7f9c7a0eee8c8fc',
-      },
+      { type: 'TagReference', tagIndex: 1, content: '#[1]'},
    ];

    assert.deepStrictEqual(parsed, expected);
  });

  it('should parse text note which includes npub string', () => {
-    const parsed = parseTextNote({
-      id: '',
-      sig: '',
-      kind: 1,
-      content:
-        'this is pubkey\nnpub1srf6g8v2qpnecqg9l2kzehmkg0ym5f5rtnlsj6lhl8r6pmhger7q5mtt3q\nhello',
-      tags: [],
-      created_at: 1678377182,
-      pubkey: '9366708117c4a7edf9178acdce538c95059b9eb3394808cdd90564094172d972',
-    });
+    const parsed = parseTextNote('this is pubkey\nnpub1srf6g8v2qpnecqg9l2kzehmkg0ym5f5rtnlsj6lhl8r6pmhger7q5mtt3q\nhello');

    const expected: ParsedTextNoteNode[] = [
      { type: 'PlainText', content: 'this is pubkey\n' },
@@ -227,3 +141,63 @@ describe('parseTextNote', () => {
    assert.deepStrictEqual(parsed, expected);
  });
 });
+
+describe('resolveTagReference', () => {
+  it('should resolve a tag reference refers a user', () => {
+    const tagReference: TagReference = {
+      type: 'TagReference',
+      tagIndex: 1,
+      content: '#[1]',
+    };
+    const dummyEvent: NostrEvent = {
+      id: '',
+      sig: '',
+      kind: 1,
+      content: '#[1]',
+      tags: [
+        ['p', '9366708117c4a7edf9178acdce538c95059b9eb3394808cdd90564094172d972'],
+        ['p', '80d3a41d8a00679c0105faac2cdf7643c9ba26835cff096bf7f9c7a0eee8c8fc'],
+      ],
+      created_at: 1678377182,
+      pubkey: '9366708117c4a7edf9178acdce538c95059b9eb3394808cdd90564094172d972',
+    };
+    const result = resolveTagReference(tagReference, dummyEvent);
+    const expected = {
+      type: 'MentionedUser',
+      tagIndex: 1,
+      content: '#[1]',
+      pubkey: '80d3a41d8a00679c0105faac2cdf7643c9ba26835cff096bf7f9c7a0eee8c8fc',
+    };
+
+    assert.deepStrictEqual(result, expected);
+  });
+
+  it('should resolve a tag reference refers an other text note', () => {
+    const tagReference: TagReference = {
+      type: 'TagReference',
+      tagIndex: 1,
+      content: '#[1]',
+    };
+    const dummyEvent: NostrEvent = {
+      id: '',
+      sig: '',
+      kind: 1,
+      content: '',
+      tags: [
+        ['p', '80d3a41d8a00679c0105faac2cdf7643c9ba26835cff096bf7f9c7a0eee8c8fc'],
+        ['e', 'b9cefcb857fa487d5794156e85b30a7f98cb21721040631210262091d86ff6f212', '', 'reply'],
+      ],
+      created_at: 1678377182,
+      pubkey: '9366708117c4a7edf9178acdce538c95059b9eb3394808cdd90564094172d972',
+    };
+    const result = resolveTagReference(tagReference, dummyEvent);
+    const expected = {
+      type: 'MentionedEvent',
+      tagIndex: 1,
+      marker: 'reply',
+      content: '#[1]',
+      eventId: 'b9cefcb857fa487d5794156e85b30a7f98cb21721040631210262091d86ff6f212',
+    };
+    assert.deepStrictEqual(result, expected);
+  });
+});
--- a/src/core/parseTextNote.ts
+++ b/src/core/parseTextNote.ts
@@ -11,19 +11,10 @@ export type PlainText = {
  content: string;
 };

-export type MentionedEvent = {
-  type: 'MentionedEvent';
-  content: string;
+export type TagReference = {
+  type: 'TagReference';
  tagIndex: number;
-  eventId: string;
-  marker: 'reply' | 'root' | 'mention' | undefined;
-};
-
-export type MentionedUser = {
-  type: 'MentionedUser';
  content: string;
-  tagIndex: number;
-  pubkey: string;
 };

 export type Bech32Entity = {
@@ -46,16 +37,25 @@ export type UrlText = {
  content: string;
 };

-export type ParsedTextNoteNode =
-  | PlainText
-  | MentionedEvent
-  | MentionedUser
-  | Bech32Entity
-  | HashTag
-  | UrlText;
+export type ParsedTextNoteNode = PlainText | TagReference | Bech32Entity | HashTag | UrlText;

 export type ParsedTextNote = ParsedTextNoteNode[];

+export type MentionedEvent = {
+  type: 'MentionedEvent';
+  content: string;
+  tagIndex: number;
+  eventId: string;
+  marker: 'reply' | 'root' | 'mention' | undefined;
+};
+
+export type MentionedUser = {
+  type: 'MentionedUser';
+  content: string;
+  tagIndex: number;
+  pubkey: string;
+};
+
 const tagRefRegex = /(?:#\[(?<idx>\d+)\])/g;
 const hashTagRegex = /#(?<hashtag>[^[-^`:-@!-/{-~\d\s][^[-^`:-@!-/{-~\s]+)/g;
 // raw NIP-19 codes, NIP-21 links (NIP-27)
@@ -64,19 +64,19 @@ const mentionRegex = /(?:nostr:)?(?<mention>(npub|note|nprofile|nevent)1[ac-hj-n
 const urlRegex =
  /(?<url>(?:https?|wss?):\/\/[-a-zA-Z0-9.:]+(?:\/[-[\]~!$&'()*+.,:;@&=%\w]+|\/)*(?:\?[-[\]~!$&'()*+.,/:;%@&=\w?]+)?(?:#[-[\]~!$&'()*+.,/:;%@\w&=?#]+)?)/g;

-const parseTextNote = (event: NostrEvent): ParsedTextNote => {
+const parseTextNote = (textNoteContent: string) => {
  const matches = [
-    ...event.content.matchAll(tagRefRegex),
-    ...event.content.matchAll(hashTagRegex),
-    ...event.content.matchAll(mentionRegex),
-    ...event.content.matchAll(urlRegex),
+    ...textNoteContent.matchAll(tagRefRegex),
+    ...textNoteContent.matchAll(hashTagRegex),
+    ...textNoteContent.matchAll(mentionRegex),
+    ...textNoteContent.matchAll(urlRegex),
  ].sort((a, b) => (a.index as number) - (b.index as number));
  let pos = 0;
  const result: ParsedTextNote = [];

  const pushPlainText = (index: number | undefined) => {
    if (index != null && pos !== index) {
-      const content = event.content.slice(pos, index);
+      const content = textNoteContent.slice(pos, index);
      const plainText: PlainText = { type: 'PlainText', content };
      result.push(plainText);
    }
@@ -94,34 +94,13 @@ const parseTextNote = (event: NostrEvent): ParsedTextNote => {
      result.push(url);
    } else if (match.groups?.idx) {
      const tagIndex = parseInt(match.groups.idx, 10);
-      const tag = event.tags[tagIndex];
-      if (tag == null) return;
-
      pushPlainText(index);

-      const tagName = tag[0];
-      if (tagName === 'p') {
-        const mentionedUser: MentionedUser = {
-          type: 'MentionedUser',
+      result.push({
+        type: 'TagReference',
        tagIndex,
        content: match[0],
-          pubkey: tag[1],
-        };
-        result.push(mentionedUser);
-      } else if (tagName === 'e') {
-        const mention = eventWrapper(event)
-          .taggedEvents()
-          .find((ev) => ev.index === tagIndex);
-
-        const mentionedEvent: MentionedEvent = {
-          type: 'MentionedEvent',
-          tagIndex,
-          content: match[0],
-          eventId: tag[1],
-          marker: mention?.marker,
-        };
-        result.push(mentionedEvent);
-      }
+      });
    } else if (match.groups?.mention) {
      pushPlainText(index);
      try {
@@ -133,7 +112,7 @@ const parseTextNote = (event: NostrEvent): ParsedTextNote => {
        };
        result.push(bech32Entity);
      } catch (e) {
-        console.error(`failed to parse Bech32 entity (NIP-19) but ignore this: ${match[0]}`);
+        console.warn(`failed to parse Bech32 entity (NIP-19): ${match[0]}`);
        pushPlainText(index + match[0].length);
        return;
      }
@@ -150,8 +129,8 @@ const parseTextNote = (event: NostrEvent): ParsedTextNote => {
    pos = index + match[0].length;
  });

-  if (pos !== event.content.length) {
-    const content = event.content.slice(pos);
+  if (pos !== textNoteContent.length) {
+    const content = textNoteContent.slice(pos);
    const plainText: PlainText = { type: 'PlainText', content };
    result.push(plainText);
  }
@@ -159,4 +138,39 @@ const parseTextNote = (event: NostrEvent): ParsedTextNote => {
  return result;
 };

+export const resolveTagReference = (
+  { tagIndex, content }: TagReference,
+  event: NostrEvent,
+): MentionedUser | MentionedEvent | null => {
+  const tag = event.tags[tagIndex];
+  if (tag == null) return null;
+
+  const tagName = tag[0];
+
+  if (tagName === 'p') {
+    return {
+      type: 'MentionedUser',
+      tagIndex,
+      content,
+      pubkey: tag[1],
+    } satisfies MentionedUser;
+  }
+
+  if (tagName === 'e') {
+    const mention = eventWrapper(event)
+      .taggedEvents()
+      .find((ev) => ev.index === tagIndex);
+
+    return {
+      type: 'MentionedEvent',
+      tagIndex,
+      content,
+      eventId: tag[1],
+      marker: mention?.marker,
+    } satisfies MentionedEvent;
+  }
+
+  return null;
+};
+
 export default parseTextNote;