Files
Tom Moor 3c18020db5 feat: Weight headings above body text in shared document search (#14087)
* feat: Weight headings above body text in shared document search

* perf: Parse document once when indexing text and headings
2026-10-13 18:44:36 -04:00

313 lines
9.5 KiB
TypeScript

import { SearchIndex } from "./SearchIndex";
describe("SearchIndex", () => {
it("returns no results for an empty query", () => {
const index = new SearchIndex();
index.update([{ id: "1", title: "Engineering Handbook", url: "/doc/1" }]);
expect(index.search("")).toEqual([]);
expect(index.search(" ")).toEqual([]);
});
it("matches titles despite spelling mistakes", () => {
const index = new SearchIndex();
index.update([
{ id: "1", title: "Engineering Handbook", url: "/doc/1" },
{ id: "2", title: "Marketing Plan", url: "/doc/2" },
]);
const results = index.search("enginering handbok");
expect(results[0]?.document.id).toBe("1");
});
it("matches document content and highlights the context", () => {
const index = new SearchIndex();
index.update([
{
id: "1",
title: "Untitled",
url: "/doc/1",
text: "The quarterly revenue numbers exceeded all expectations.",
},
]);
const results = index.search("revenue");
expect(results[0]?.document.id).toBe("1");
expect(results[0]?.context).toContain("<b>revenue</b>");
});
it("highlights only exact matches, not fuzzy character runs", () => {
const index = new SearchIndex();
index.update([
{
id: "1",
title: "Untitled",
url: "/doc/1",
text: "The quarterly revenue numbers exceeded all expectations.",
},
]);
// The typo "numbrs" has no literal occurrence, so only whole contiguous
// matches are marked rather than the scattered characters Fuse matched on.
const results = index.search("revenu numbrs");
expect(results[0]?.document.id).toBe("1");
expect(results[0]?.context).toContain("<b>revenu</b>");
expect(results[0]?.context).not.toContain("numbrs");
expect(results[0]?.context?.match(/<b>/g)).toHaveLength(1);
});
it("highlights the query where it appears inside a longer word", () => {
const index = new SearchIndex();
index.update([
{
id: "1",
title: "Untitled",
url: "/doc/1",
text: "The quarterly revenue numbers exceeded all expectations.",
},
]);
// Partway through typing "revenue", the typed prefix is still marked.
expect(index.search("revenu")[0]?.context).toContain("<b>revenu</b>");
});
it("highlights each term of a multi-word query separately", () => {
const index = new SearchIndex();
index.update([
{
id: "1",
title: "Untitled",
url: "/doc/1",
text: "The quarterly revenue numbers exceeded all expectations.",
},
]);
const context = index.search("revenue numbers")[0]?.context;
expect(context).toContain("<b>revenue numbers</b>");
});
it("marks a match as one contiguous run rather than letter runs", () => {
const index = new SearchIndex();
index.update([
{
id: "1",
title: "Untitled",
url: "/doc/1",
text: "Engineering onboarding covers the deployment pipeline.",
},
]);
const context = index.search("deployment")[0]?.context;
// The whole word is marked once, rather than scattered letter runs.
expect(context).toContain("<b>deployment</b>");
expect(context?.match(/<b>/g)).toHaveLength(1);
});
it("preserves previously indexed content when a later update omits it", () => {
const index = new SearchIndex();
index.update([
{
id: "1",
title: "Report",
url: "/doc/1",
text: "Contains the word pineapple somewhere inside.",
},
]);
// A subsequent update (e.g. a title-only tree pass) must not drop content.
index.update([{ id: "1", title: "Report", url: "/doc/1" }]);
expect(index.search("pineapple")[0]?.document.id).toBe("1");
});
it("clears content when an update sets text to an empty string", () => {
const index = new SearchIndex();
index.update([
{
id: "1",
title: "Report",
url: "/doc/1",
text: "Contains the word pineapple somewhere inside.",
},
]);
// An explicit empty string (e.g. content emptied) must replace stale text.
index.update([{ id: "1", title: "Report", url: "/doc/1", text: "" }]);
expect(index.search("pineapple")).toEqual([]);
});
it("previews content for a title match that has no content match", () => {
const index = new SearchIndex();
index.update([
{
id: "1",
title: "Engineering Handbook",
url: "/doc/1",
text: "Everything a new hire needs to know about our team.",
},
]);
const context = index.search("engineering")[0]?.context;
expect(context).toBe("Everything a new hire needs to know about our team.");
});
it("has no context for a document with no content", () => {
const index = new SearchIndex();
index.update([{ id: "1", title: "Engineering Handbook", url: "/doc/1" }]);
expect(index.search("engineering")[0]?.context).toBeUndefined();
});
it("truncates a long preview", () => {
const index = new SearchIndex();
index.update([
{
id: "1",
title: "Engineering Handbook",
url: "/doc/1",
text: "word ".repeat(200),
},
]);
const context = index.search("engineering")[0]?.context ?? "";
expect(context.endsWith("…")).toBe(true);
expect(context.length).toBeLessThan(220);
});
it("orders a title prefix ahead of a shorter mid-title match", () => {
const index = new SearchIndex();
index.update([
// Fuse scores this higher on its own, as the field is shorter and
// position is ignored.
{ id: "mid", title: "Our Design", url: "/doc/mid" },
{
id: "prefix",
title: "Design System Guidelines Handbook",
url: "/doc/prefix",
},
]);
expect(index.search("design")[0]?.document.id).toBe("prefix");
});
it("orders a title prefix ahead of a shorter trailing-word match", () => {
const index = new SearchIndex();
index.update([
{ id: "mid", title: "Team Onboarding", url: "/doc/mid" },
{
id: "prefix",
title: "Onboarding Checklist For New Hires",
url: "/doc/prefix",
},
]);
expect(index.search("onboarding")[0]?.document.id).toBe("prefix");
});
it("orders a title prefix ahead of an equally scored mid-title match", () => {
const index = new SearchIndex();
index.update([
// Both score identically in Fuse, leaving the order arbitrary.
{ id: "mid", title: "A Guide To Design", url: "/doc/mid" },
{ id: "prefix", title: "Design Is A Guide", url: "/doc/prefix" },
]);
expect(index.search("design")[0]?.document.id).toBe("prefix");
});
it("orders a title prefix ahead of a content-only match", () => {
const index = new SearchIndex();
index.update([
{ id: "content", title: "Q3", url: "/doc/content", text: "revenue" },
{
id: "prefix",
title: "Revenue Recognition Policy Handbook",
url: "/doc/prefix",
},
]);
expect(index.search("revenue")[0]?.document.id).toBe("prefix");
});
it("breaks ties within a tier by fuzzy score", () => {
const index = new SearchIndex();
index.update([
{ id: "long", title: "Design System Guidelines", url: "/doc/long" },
{ id: "short", title: "Design", url: "/doc/short" },
]);
// Both are prefix matches, so the closer title wins.
expect(index.search("design")[0]?.document.id).toBe("short");
});
it("weights title matches above content matches", () => {
const index = new SearchIndex();
index.update([
{ id: "1", title: "Onboarding", url: "/doc/1", text: "unrelated body" },
{
id: "2",
title: "Team",
url: "/doc/2",
text: "please read the onboarding guide",
},
]);
const results = index.search("onboarding");
expect(results[0]?.document.id).toBe("1");
});
it("weights heading matches above content matches", () => {
const index = new SearchIndex();
index.update([
{ id: "1", title: "Team", url: "/doc/1", text: "onboarding guide" },
{
id: "2",
title: "People",
url: "/doc/2",
text: "Onboarding\nhow new starters get set up in their first week",
headings: ["Onboarding"],
},
]);
const results = index.search("onboarding");
expect(results[0]?.document.id).toBe("2");
});
it("preserves previously indexed headings when a later update omits them", () => {
const index = new SearchIndex();
index.update([
{ id: "1", title: "Team", url: "/doc/1", headings: ["Onboarding"] },
]);
expect(index.update([{ id: "1", title: "Team", url: "/doc/1" }])).toBe(
false
);
expect(index.search("onboarding")[0]?.document.id).toBe("1");
});
it("reports a change when headings change", () => {
const index = new SearchIndex();
index.update([{ id: "1", title: "A", url: "/doc/1", headings: ["One"] }]);
expect(
index.update([{ id: "1", title: "A", url: "/doc/1", headings: ["One"] }])
).toBe(false);
expect(
index.update([{ id: "1", title: "A", url: "/doc/1", headings: ["Two"] }])
).toBe(true);
});
it("reports whether an update changed the collection", () => {
const index = new SearchIndex();
expect(index.update([{ id: "1", title: "A", url: "/doc/1" }])).toBe(true);
expect(index.update([{ id: "1", title: "A", url: "/doc/1" }])).toBe(false);
});
it("clears all indexed documents", () => {
const index = new SearchIndex();
index.update([{ id: "1", title: "Engineering", url: "/doc/1" }]);
index.clear();
expect(index.search("engineering")).toEqual([]);
});
});