feat(search): match quick-open items by pinyin initials
This commit is contained in:
1 parent
299c342d6d
commit
434d44a284
2 files changed
+86
-3
No files matched your search
@@ -115,6 +115,30 @@ describe("useQuickOpen", () => {
|
|||||||
expect(matchQuickOpenText("cardshoplog", label)?.kind).toBe("fuzzy");
|
expect(matchQuickOpenText("cardshoplog", label)?.kind).toBe("fuzzy");
|
||||||
});
|
});
|
||||||
|
|
||||||
|
it("matches Chinese table names by pinyin initials (issue #7912)", () => {
|
||||||
|
// Reported: 全局搜索中文表名输入首字母模糊搜索搜不出来
|
||||||
|
expect(matchQuickOpenText("zzj", "总租金")?.kind).toBe("initials");
|
||||||
|
expect(matchQuickOpenText("yhmx", "用户明细")?.kind).toBe("initials");
|
||||||
|
});
|
||||||
|
|
||||||
|
it("matches Chinese table names by a non-contiguous pinyin-initials subsequence", () => {
|
||||||
|
const match = matchQuickOpenText("zj", "总租金");
|
||||||
|
expect(match?.kind).toBe("fuzzy");
|
||||||
|
expect(match?.indices.map((index) => "总租金"[index]).join("")).toBe("总租金".slice(0, 1) + "总租金".slice(2, 3));
|
||||||
|
});
|
||||||
|
|
||||||
|
it("keeps pinyin-initials highlight indices aligned across unmapped supplementary-plane characters", () => {
|
||||||
|
const match = matchQuickOpenText("bx", "𠮷表x");
|
||||||
|
expect(match?.kind).toBe("fuzzy");
|
||||||
|
expect(match?.indices).toEqual([2, 3]);
|
||||||
|
});
|
||||||
|
|
||||||
|
it("prefers a literal prefix match over pinyin-initials for mixed Han+Latin names", () => {
|
||||||
|
const literal = matchQuickOpenText("abc", "abc表");
|
||||||
|
expect(literal?.kind).toBe("prefix");
|
||||||
|
expect(literal?.score).toBeLessThan(250);
|
||||||
|
});
|
||||||
|
|
||||||
it("puts the exact shop result before prefixed and containing names", () => {
|
it("puts the exact shop result before prefixed and containing names", () => {
|
||||||
vi.mocked(useConnectionStore).mockReturnValue({
|
vi.mocked(useConnectionStore).mockReturnValue({
|
||||||
connections: [
|
connections: [
|
||||||
|
|||||||
@@ -7,6 +7,7 @@ import { useSavedSqlStore } from "@/stores/savedSqlStore";
|
|||||||
import * as api from "@/lib/backend/api";
|
import * as api from "@/lib/backend/api";
|
||||||
import type { SqlFileEntry } from "@/lib/backend/api";
|
import type { SqlFileEntry } from "@/lib/backend/api";
|
||||||
import { getSqlFileFilter, getSqlFileFolderPaths, sqlFileFoldersVersion } from "@/lib/sqlFile/sqlFileFolders";
|
import { getSqlFileFilter, getSqlFileFolderPaths, sqlFileFoldersVersion } from "@/lib/sqlFile/sqlFileFolders";
|
||||||
|
import { containsHan, pinyinFirstLetters } from "@/lib/common/pinyin";
|
||||||
import i18n from "@/i18n";
|
import i18n from "@/i18n";
|
||||||
|
|
||||||
const REMOTE_SEARCH_DEBOUNCE_MS = 180;
|
const REMOTE_SEARCH_DEBOUNCE_MS = 180;
|
||||||
@@ -86,6 +87,23 @@ function rangeIndices(start: number, length: number): number[] {
|
|||||||
return Array.from({ length }, (_, index) => start + index);
|
return Array.from({ length }, (_, index) => start + index);
|
||||||
}
|
}
|
||||||
|
|
||||||
|
const PINYIN_QUERY_RE = /^[a-z0-9]+$/;
|
||||||
|
|
||||||
|
/**
|
||||||
|
* Original-text indices of the characters that feed `pinyinFirstLetters(text)`, in order.
|
||||||
|
* Iterates by Unicode code point (like `pinyinFirstLetters`), not UTF-16 code unit, so
|
||||||
|
* supplementary-plane Han characters (surrogate pairs) stay aligned with the letters they produce.
|
||||||
|
*/
|
||||||
|
function pinyinLetterPositions(text: string): number[] {
|
||||||
|
const positions: number[] = [];
|
||||||
|
let index = 0;
|
||||||
|
for (const char of text) {
|
||||||
|
if (/[\p{Script=Han}a-z0-9]/iu.test(char)) positions.push(index);
|
||||||
|
index += char.length;
|
||||||
|
}
|
||||||
|
return positions;
|
||||||
|
}
|
||||||
|
|
||||||
function matchWordPrefixes(words: IdentifierWord[], query: string): number[] | null {
|
function matchWordPrefixes(words: IdentifierWord[], query: string): number[] | null {
|
||||||
interface PrefixState {
|
interface PrefixState {
|
||||||
queryIndex: number;
|
queryIndex: number;
|
||||||
@@ -150,6 +168,20 @@ export function matchQuickOpenText(query: string, text: string): QuickOpenMatch
|
|||||||
return { kind: "prefix", score: 200 + Math.min(text.length - lowerQuery.length, 99), indices: rangeIndices(0, lowerQuery.length) };
|
return { kind: "prefix", score: 200 + Math.min(text.length - lowerQuery.length, 99), indices: rangeIndices(0, lowerQuery.length) };
|
||||||
}
|
}
|
||||||
|
|
||||||
|
// DataGrip-style pinyin-initials matching for Chinese identifiers, e.g. "总租金" via "zzj".
|
||||||
|
// Only tried after literal matches fail, so a mixed Han+Latin name that literally prefix-matches
|
||||||
|
// (e.g. "abc表" via "abc") keeps its better literal-prefix score instead of being intercepted here.
|
||||||
|
const isPinyinQuery = PINYIN_QUERY_RE.test(lowerQuery) && containsHan(text);
|
||||||
|
if (isPinyinQuery) {
|
||||||
|
const pinyinLetters = pinyinFirstLetters(text);
|
||||||
|
if (pinyinLetters === lowerQuery) {
|
||||||
|
return { kind: "initials", score: 150 + Math.min(text.length, 99), indices: pinyinLetterPositions(text) };
|
||||||
|
}
|
||||||
|
if (pinyinLetters.startsWith(lowerQuery)) {
|
||||||
|
return { kind: "prefix", score: 250 + Math.min(text.length - lowerQuery.length, 99), indices: pinyinLetterPositions(text).slice(0, lowerQuery.length) };
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
const wordPrefixIndices = matchWordPrefixes(words, lowerQuery);
|
const wordPrefixIndices = matchWordPrefixes(words, lowerQuery);
|
||||||
if (wordPrefixIndices) {
|
if (wordPrefixIndices) {
|
||||||
return { kind: "word-prefix", score: 300 + Math.min(text.length - lowerQuery.length, 99), indices: wordPrefixIndices };
|
return { kind: "word-prefix", score: 300 + Math.min(text.length - lowerQuery.length, 99), indices: wordPrefixIndices };
|
||||||
@@ -168,10 +200,37 @@ export function matchQuickOpenText(query: string, text: string): QuickOpenMatch
|
|||||||
indices.push(index);
|
indices.push(index);
|
||||||
queryIndex++;
|
queryIndex++;
|
||||||
}
|
}
|
||||||
if (queryIndex !== lowerQuery.length) return null;
|
if (queryIndex === lowerQuery.length) {
|
||||||
|
const span = indices[indices.length - 1] - indices[0] + 1;
|
||||||
|
return { kind: "fuzzy", score: 500 + Math.min(span - lowerQuery.length, 99), indices };
|
||||||
|
}
|
||||||
|
|
||||||
const span = indices[indices.length - 1] - indices[0] + 1;
|
// Non-contiguous pinyin-initials fallback, e.g. "zj" matching "总租金" (pinyin initials "zzj"),
|
||||||
return { kind: "fuzzy", score: 500 + Math.min(span - lowerQuery.length, 99), indices };
|
// matching the ordered-subsequence behavior every other pinyin call site in the app already has.
|
||||||
|
if (isPinyinQuery) {
|
||||||
|
const pinyinLetters = pinyinFirstLetters(text);
|
||||||
|
const positions = pinyinLetterPositions(text);
|
||||||
|
const letterIndices: number[] = [];
|
||||||
|
let letterQueryIndex = 0;
|
||||||
|
let letterCount = 0;
|
||||||
|
// Iterate the initials by code point, keeping letterCount aligned with
|
||||||
|
// positions: unmapped supplementary-plane Han characters occupy two code
|
||||||
|
// units in pinyinLetters but still correspond to exactly one position.
|
||||||
|
for (const letter of pinyinLetters) {
|
||||||
|
if (letterQueryIndex >= lowerQuery.length) break;
|
||||||
|
if (letter === lowerQuery[letterQueryIndex]) {
|
||||||
|
letterIndices.push(positions[letterCount] ?? letterCount);
|
||||||
|
letterQueryIndex += 1;
|
||||||
|
}
|
||||||
|
letterCount += 1;
|
||||||
|
}
|
||||||
|
if (letterQueryIndex === lowerQuery.length) {
|
||||||
|
const pinyinSpan = letterIndices[letterIndices.length - 1] - letterIndices[0] + 1;
|
||||||
|
return { kind: "fuzzy", score: 500 + Math.min(pinyinSpan - lowerQuery.length, 99), indices: letterIndices };
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
return null;
|
||||||
}
|
}
|
||||||
|
|
||||||
interface MatchedItem extends QuickOpenItem {
|
interface MatchedItem extends QuickOpenItem {
|
||||||
|
|||||||
Reference in new issue
Block a user