feat(native): native reader for indexer (#14055)

This commit is contained in:
DarkSky
2025-12-07 16:22:11 +08:00
committed by GitHub
parent 69cdeedc4e
commit cf4e37c584
28 changed files with 1376 additions and 569 deletions
@@ -1,6 +1,7 @@
import { AutoReconnectConnection } from '../../connection';
import type {
BlobRecord,
CrawlResult,
DocClock,
DocRecord,
ListedBlobRecord,
@@ -81,6 +82,7 @@ export interface NativeDBApis {
peer: string,
blobId: string
) => Promise<Date | null>;
crawlDocData: (id: string, docId: string) => Promise<CrawlResult>;
}
type NativeDBApisWrapper = NativeDBApis extends infer APIs
@@ -1,5 +1,7 @@
import { share } from '../../connection';
import {
type BlockInfo,
type CrawlResult,
type DocClocks,
type DocRecord,
DocStorageBase,
@@ -79,4 +81,127 @@ export class SqliteDocStorage extends DocStorageBase<SqliteNativeDBOptions> {
updates.map(update => update.timestamp)
);
}
override async crawlDocData(docId: string): Promise<CrawlResult | null> {
const result = await this.db.crawlDocData(docId);
return normalizeNativeCrawlResult(result);
}
}
function normalizeNativeCrawlResult(result: unknown): CrawlResult | null {
if (!isRecord(result)) {
console.warn('[nbstore] crawlDocData returned non-object result');
return null;
}
if (
typeof result.title !== 'string' ||
typeof result.summary !== 'string' ||
!Array.isArray(result.blocks)
) {
console.warn('[nbstore] crawlDocData result missing basic fields');
return null;
}
const { title, summary } = result as { title: string; summary: string };
const rawBlocks = result.blocks as unknown[];
const blocks: BlockInfo[] = [];
for (const block of rawBlocks) {
const normalized = normalizeBlock(block);
if (normalized) {
blocks.push(normalized);
}
}
if (blocks.length === 0) {
console.warn('[nbstore] crawlDocData has no valid blocks');
return null;
}
return {
blocks,
title,
summary,
};
}
function normalizeBlock(block: unknown): BlockInfo | null {
if (!isRecord(block)) {
return null;
}
const blockId = readStringField(block, 'blockId');
const flavour = readStringField(block, 'flavour');
if (!blockId || !flavour) {
return null;
}
return {
blockId,
flavour,
content: readStringArrayField(block, 'content'),
blob: readStringArrayField(block, 'blob'),
refDocId: readStringArrayField(block, 'refDocId'),
refInfo: readStringArrayField(block, 'refInfo'),
parentFlavour: readStringField(block, 'parentFlavour'),
parentBlockId: readStringField(block, 'parentBlockId'),
additional: safeAdditionalField(block),
};
}
function readStringField(
target: Record<string, unknown>,
key: string
): string | undefined {
const value = readField(target, key);
return typeof value === 'string' && value ? value : undefined;
}
function readStringArrayField(
target: Record<string, unknown>,
key: string
): string[] | undefined {
const value = readField(target, key);
if (Array.isArray(value)) {
const filtered = value.filter(
(item): item is string => typeof item === 'string' && item.length > 0
);
return filtered.length ? filtered : undefined;
}
if (typeof value === 'string' && value.length > 0) {
return [value];
}
return undefined;
}
function safeAdditionalField(
target: Record<string, unknown>
): string | undefined {
const value = readField(target, 'additional');
if (typeof value !== 'string' || value.length === 0) {
return undefined;
}
try {
const parsed = JSON.parse(value);
return JSON.stringify(parsed);
} catch {
console.warn(
'[nbstore] ignore invalid additional payload in crawlDocData block'
);
return undefined;
}
}
function readField(target: Record<string, unknown>, key: string) {
return target[key] ?? target[toSnakeCase(key)];
}
function toSnakeCase(key: string) {
return key.replace(/[A-Z]/g, letter => `_${letter.toLowerCase()}`);
}
function isRecord(value: unknown): value is Record<string, unknown> {
return typeof value === 'object' && value !== null;
}