mirror of
https://github.com/zadam/trilium.git
synced 2025-11-10 00:19:04 +01:00
feat(search): also support the use of ="exact match search string"
This commit is contained in:
perf3ct
parent
50f0b88eff
commit
4fa4112840
@ -322,4 +322,181 @@ test.describe("Exact Search with Leading = Operator", () => {
|
||||
// Should find both uppercase and lowercase versions
|
||||
expect(ourTestNotes.length).toBe(2);
|
||||
});
|
||||
|
||||
test("Exact phrase matching with multi-word searches", async ({ page }) => {
|
||||
// Create notes with various phrase patterns
|
||||
const note1 = await page.request.post(`${BASE_URL}/api/notes/root/children?target=into&targetBranchId=`, {
|
||||
headers: { "x-csrf-token": csrfToken },
|
||||
data: {
|
||||
title: "exact phrase",
|
||||
content: "This note contains the exact phrase.",
|
||||
type: "text"
|
||||
}
|
||||
});
|
||||
expect(note1.ok()).toBeTruthy();
|
||||
const note1Data = await note1.json();
|
||||
createdNoteIds.push(note1Data.note.noteId);
|
||||
|
||||
const note2 = await page.request.post(`${BASE_URL}/api/notes/root/children?target=into&targetBranchId=`, {
|
||||
headers: { "x-csrf-token": csrfToken },
|
||||
data: {
|
||||
title: "exact phrase match",
|
||||
content: "This note has exact phrase followed by more words.",
|
||||
type: "text"
|
||||
}
|
||||
});
|
||||
expect(note2.ok()).toBeTruthy();
|
||||
const note2Data = await note2.json();
|
||||
createdNoteIds.push(note2Data.note.noteId);
|
||||
|
||||
const note3 = await page.request.post(`${BASE_URL}/api/notes/root/children?target=into&targetBranchId=`, {
|
||||
headers: { "x-csrf-token": csrfToken },
|
||||
data: {
|
||||
title: "phrase exact",
|
||||
content: "This note has the words in reverse order.",
|
||||
type: "text"
|
||||
}
|
||||
});
|
||||
expect(note3.ok()).toBeTruthy();
|
||||
const note3Data = await note3.json();
|
||||
createdNoteIds.push(note3Data.note.noteId);
|
||||
|
||||
const note4 = await page.request.post(`${BASE_URL}/api/notes/root/children?target=into&targetBranchId=`, {
|
||||
headers: { "x-csrf-token": csrfToken },
|
||||
data: {
|
||||
title: "this exact and that phrase",
|
||||
content: "Words are separated but both present.",
|
||||
type: "text"
|
||||
}
|
||||
});
|
||||
expect(note4.ok()).toBeTruthy();
|
||||
const note4Data = await note4.json();
|
||||
createdNoteIds.push(note4Data.note.noteId);
|
||||
|
||||
await page.waitForTimeout(500);
|
||||
|
||||
// Search for exact phrase "exact phrase"
|
||||
const response = await page.request.get(`${BASE_URL}/api/quick-search/='exact phrase'`, {
|
||||
headers: { "x-csrf-token": csrfToken }
|
||||
});
|
||||
|
||||
expect(response.ok()).toBeTruthy();
|
||||
const data = await response.json();
|
||||
|
||||
const ourTestNotes = data.searchResults.filter((result: any) => {
|
||||
const noteId = result.notePath.split("/").pop();
|
||||
return [note1Data.note.noteId, note2Data.note.noteId, note3Data.note.noteId, note4Data.note.noteId].includes(noteId || "");
|
||||
});
|
||||
|
||||
console.log("Exact phrase search '=\"exact phrase\"' found:", ourTestNotes.length, "notes");
|
||||
console.log("Note titles:", ourTestNotes.map((r: any) => r.noteTitle));
|
||||
|
||||
// Should find only notes 1 and 2 (consecutive "exact phrase")
|
||||
// Should NOT find note 3 (reversed order) or note 4 (words separated)
|
||||
expect(ourTestNotes.length).toBe(2);
|
||||
|
||||
const foundTitles = ourTestNotes.map((r: any) => r.noteTitle);
|
||||
expect(foundTitles).toContain("exact phrase");
|
||||
expect(foundTitles).toContain("exact phrase match");
|
||||
expect(foundTitles).not.toContain("phrase exact");
|
||||
expect(foundTitles).not.toContain("this exact and that phrase");
|
||||
});
|
||||
|
||||
test("Exact phrase matching respects word order", async ({ page }) => {
|
||||
// Create notes to test word order sensitivity
|
||||
const noteForward = await page.request.post(`${BASE_URL}/api/notes/root/children?target=into&targetBranchId=`, {
|
||||
headers: { "x-csrf-token": csrfToken },
|
||||
data: {
|
||||
title: "Testing Order",
|
||||
content: "This is a test sentence for verification.",
|
||||
type: "text"
|
||||
}
|
||||
});
|
||||
expect(noteForward.ok()).toBeTruthy();
|
||||
const noteForwardData = await noteForward.json();
|
||||
createdNoteIds.push(noteForwardData.note.noteId);
|
||||
|
||||
const noteReverse = await page.request.post(`${BASE_URL}/api/notes/root/children?target=into&targetBranchId=`, {
|
||||
headers: { "x-csrf-token": csrfToken },
|
||||
data: {
|
||||
title: "Order Testing",
|
||||
content: "A sentence test is this for verification.",
|
||||
type: "text"
|
||||
}
|
||||
});
|
||||
expect(noteReverse.ok()).toBeTruthy();
|
||||
const noteReverseData = await noteReverse.json();
|
||||
createdNoteIds.push(noteReverseData.note.noteId);
|
||||
|
||||
await page.waitForTimeout(500);
|
||||
|
||||
// Search for exact phrase "test sentence"
|
||||
const response = await page.request.get(`${BASE_URL}/api/quick-search/='test sentence'`, {
|
||||
headers: { "x-csrf-token": csrfToken }
|
||||
});
|
||||
|
||||
expect(response.ok()).toBeTruthy();
|
||||
const data = await response.json();
|
||||
|
||||
const ourTestNotes = data.searchResults.filter((result: any) => {
|
||||
const noteId = result.notePath.split("/").pop();
|
||||
return noteId === noteForwardData.note.noteId || noteId === noteReverseData.note.noteId;
|
||||
});
|
||||
|
||||
console.log("Exact phrase search '=\"test sentence\"' found:", ourTestNotes.length, "notes");
|
||||
console.log("Note titles:", ourTestNotes.map((r: any) => r.noteTitle));
|
||||
|
||||
// Should find only the forward order note
|
||||
expect(ourTestNotes.length).toBe(1);
|
||||
expect(ourTestNotes[0].noteTitle).toBe("Testing Order");
|
||||
});
|
||||
|
||||
test("Multi-word exact search without quotes", async ({ page }) => {
|
||||
// Test that multi-word search with = but without quotes also does exact phrase matching
|
||||
const notePhrase = await page.request.post(`${BASE_URL}/api/notes/root/children?target=into&targetBranchId=`, {
|
||||
headers: { "x-csrf-token": csrfToken },
|
||||
data: {
|
||||
title: "Quick Test Note",
|
||||
content: "A simple note for multi word testing.",
|
||||
type: "text"
|
||||
}
|
||||
});
|
||||
expect(notePhrase.ok()).toBeTruthy();
|
||||
const notePhraseData = await notePhrase.json();
|
||||
createdNoteIds.push(notePhraseData.note.noteId);
|
||||
|
||||
const noteScattered = await page.request.post(`${BASE_URL}/api/notes/root/children?target=into&targetBranchId=`, {
|
||||
headers: { "x-csrf-token": csrfToken },
|
||||
data: {
|
||||
title: "Word Multi Testing",
|
||||
content: "Words are multi scattered in this testing example.",
|
||||
type: "text"
|
||||
}
|
||||
});
|
||||
expect(noteScattered.ok()).toBeTruthy();
|
||||
const noteScatteredData = await noteScattered.json();
|
||||
createdNoteIds.push(noteScatteredData.note.noteId);
|
||||
|
||||
await page.waitForTimeout(500);
|
||||
|
||||
// Search for "=multi word" without quotes (parser tokenizes as two words)
|
||||
const response = await page.request.get(`${BASE_URL}/api/quick-search/=multi word`, {
|
||||
headers: { "x-csrf-token": csrfToken }
|
||||
});
|
||||
|
||||
expect(response.ok()).toBeTruthy();
|
||||
const data = await response.json();
|
||||
|
||||
const ourTestNotes = data.searchResults.filter((result: any) => {
|
||||
const noteId = result.notePath.split("/").pop();
|
||||
return noteId === notePhraseData.note.noteId || noteId === noteScatteredData.note.noteId;
|
||||
});
|
||||
|
||||
console.log("Multi-word exact search '=multi word' found:", ourTestNotes.length, "notes");
|
||||
console.log("Note titles:", ourTestNotes.map((r: any) => r.noteTitle));
|
||||
|
||||
// Should find only the note with consecutive "multi word" phrase
|
||||
expect(ourTestNotes.length).toBe(1);
|
||||
expect(ourTestNotes[0].noteTitle).toBe("Quick Test Note");
|
||||
});
|
||||
});
|
||||
|
||||
@ -77,15 +77,43 @@ class NoteContentFulltextExp extends Expression {
|
||||
|
||||
const resultNoteSet = new NoteSet();
|
||||
|
||||
// Search through notes with content
|
||||
for (const row of sql.iterateRows<SearchRow>(`
|
||||
SELECT noteId, type, mime, content, isProtected
|
||||
FROM notes JOIN blobs USING (blobId)
|
||||
WHERE type IN ('text', 'code', 'mermaid', 'canvas', 'mindMap')
|
||||
AND isDeleted = 0
|
||||
WHERE type IN ('text', 'code', 'mermaid', 'canvas', 'mindMap')
|
||||
AND isDeleted = 0
|
||||
AND LENGTH(content) < ${MAX_SEARCH_CONTENT_SIZE}`)) {
|
||||
this.findInText(row, inputNoteSet, resultNoteSet);
|
||||
}
|
||||
|
||||
// For exact match with flatText, also search notes WITHOUT content (they may have matching attributes)
|
||||
if (this.flatText && (this.operator === "=" || this.operator === "!=")) {
|
||||
for (const noteId of inputNoteSet.noteIdSet) {
|
||||
// Skip if already found or doesn't exist
|
||||
if (resultNoteSet.hasNoteId(noteId) || !(noteId in becca.notes)) {
|
||||
continue;
|
||||
}
|
||||
|
||||
const note = becca.notes[noteId];
|
||||
const flatText = note.getFlatText();
|
||||
|
||||
// For flatText, only check attribute values (format: #name=value or ~name=value)
|
||||
// Don't match against noteId, type, mime, or title which are also in flatText
|
||||
let matches = false;
|
||||
const phrase = this.tokens.join(" ");
|
||||
const normalizedPhrase = normalizeSearchText(phrase);
|
||||
const normalizedFlatText = normalizeSearchText(flatText);
|
||||
|
||||
// Check if =phrase appears in flatText (indicates attribute value match)
|
||||
matches = normalizedFlatText.includes(`=${normalizedPhrase}`);
|
||||
|
||||
if ((this.operator === "=" && matches) || (this.operator === "!=" && !matches)) {
|
||||
resultNoteSet.add(note);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
return resultNoteSet;
|
||||
}
|
||||
|
||||
@ -103,6 +131,32 @@ class NoteContentFulltextExp extends Expression {
|
||||
return words.some(word => word === normalizedToken);
|
||||
}
|
||||
|
||||
/**
|
||||
* Checks if content contains the exact phrase (consecutive words in order)
|
||||
* This is case-insensitive since content and tokens are already normalized
|
||||
*/
|
||||
private containsExactPhrase(tokens: string[], content: string, checkFlatTextAttributes: boolean = false): boolean {
|
||||
const normalizedTokens = tokens.map(t => normalizeSearchText(t));
|
||||
const normalizedContent = normalizeSearchText(content);
|
||||
|
||||
// Join tokens with single space to form the phrase
|
||||
const phrase = normalizedTokens.join(" ");
|
||||
|
||||
// Check if the phrase appears as a substring (consecutive words)
|
||||
if (normalizedContent.includes(phrase)) {
|
||||
return true;
|
||||
}
|
||||
|
||||
// For flatText, also check if the phrase appears in attribute values
|
||||
// Attributes in flatText appear as "#name=value" or "~name=value"
|
||||
// So we need to check for "=phrase" to match attribute values
|
||||
if (checkFlatTextAttributes && normalizedContent.includes(`=${phrase}`)) {
|
||||
return true;
|
||||
}
|
||||
|
||||
return false;
|
||||
}
|
||||
|
||||
findInText({ noteId, isProtected, content, type, mime }: SearchRow, inputNoteSet: NoteSet, resultNoteSet: NoteSet) {
|
||||
if (!inputNoteSet.hasNoteId(noteId) || !(noteId in becca.notes)) {
|
||||
return;
|
||||
@ -137,9 +191,25 @@ class NoteContentFulltextExp extends Expression {
|
||||
if (this.tokens.length === 1) {
|
||||
const [token] = this.tokens;
|
||||
|
||||
let matches = false;
|
||||
if (this.operator === "=") {
|
||||
matches = this.containsExactWord(token, content);
|
||||
// Also check flatText if enabled (includes attributes)
|
||||
if (!matches && this.flatText) {
|
||||
const flatText = becca.notes[noteId].getFlatText();
|
||||
matches = this.containsExactPhrase([token], flatText, true);
|
||||
}
|
||||
} else if (this.operator === "!=") {
|
||||
matches = !this.containsExactWord(token, content);
|
||||
// For negation, check flatText too
|
||||
if (matches && this.flatText) {
|
||||
const flatText = becca.notes[noteId].getFlatText();
|
||||
matches = !this.containsExactPhrase([token], flatText, true);
|
||||
}
|
||||
}
|
||||
|
||||
if (
|
||||
(this.operator === "=" && this.containsExactWord(token, content)) ||
|
||||
(this.operator === "!=" && !this.containsExactWord(token, content)) ||
|
||||
matches ||
|
||||
(this.operator === "*=" && content.endsWith(token)) ||
|
||||
(this.operator === "=*" && content.startsWith(token)) ||
|
||||
(this.operator === "*=*" && content.includes(token)) ||
|
||||
@ -152,10 +222,26 @@ class NoteContentFulltextExp extends Expression {
|
||||
} else {
|
||||
// Multi-token matching with fuzzy support and phrase proximity
|
||||
if (this.operator === "~=" || this.operator === "~*") {
|
||||
// Fuzzy phrase matching
|
||||
if (this.matchesWithFuzzy(content, noteId)) {
|
||||
resultNoteSet.add(becca.notes[noteId]);
|
||||
}
|
||||
} else if (this.operator === "=" || this.operator === "!=") {
|
||||
// Exact phrase matching for = and !=
|
||||
let matches = this.containsExactPhrase(this.tokens, content, false);
|
||||
|
||||
// Also check flatText if enabled (includes attributes)
|
||||
if (!matches && this.flatText) {
|
||||
const flatText = becca.notes[noteId].getFlatText();
|
||||
matches = this.containsExactPhrase(this.tokens, flatText, true);
|
||||
}
|
||||
|
||||
if ((this.operator === "=" && matches) ||
|
||||
(this.operator === "!=" && !matches)) {
|
||||
resultNoteSet.add(becca.notes[noteId]);
|
||||
}
|
||||
} else {
|
||||
// Other operators: check all tokens present (any order)
|
||||
const nonMatchingToken = this.tokens.find(
|
||||
(token) =>
|
||||
!this.tokenMatchesContent(token, content, noteId)
|
||||
|
||||
@ -14,24 +14,35 @@ type Comparator<T> = (comparedValue: T) => (val: string) => boolean;
|
||||
|
||||
const stringComparators: Record<string, Comparator<string>> = {
|
||||
"=": (comparedValue) => (val) => {
|
||||
// For the = operator, check if the value contains the exact word (word-boundary matching)
|
||||
// For the = operator, check if the value contains the exact word or phrase
|
||||
// This is case-insensitive since both values are already lowercased
|
||||
if (!val) return false;
|
||||
|
||||
const normalizedVal = normalizeSearchText(val);
|
||||
const normalizedCompared = normalizeSearchText(comparedValue);
|
||||
|
||||
// Split into words and check for exact match
|
||||
// If comparedValue has multiple words, check for exact phrase
|
||||
if (normalizedCompared.includes(" ")) {
|
||||
return normalizedVal.includes(normalizedCompared);
|
||||
}
|
||||
|
||||
// For single word, split into words and check for exact match
|
||||
const words = normalizedVal.split(/\s+/);
|
||||
return words.some(word => word === normalizedCompared);
|
||||
},
|
||||
"!=": (comparedValue) => (val) => {
|
||||
// Negation of exact word match
|
||||
// Negation of exact word/phrase match
|
||||
if (!val) return true;
|
||||
|
||||
const normalizedVal = normalizeSearchText(val);
|
||||
const normalizedCompared = normalizeSearchText(comparedValue);
|
||||
|
||||
// If comparedValue has multiple words, check for exact phrase
|
||||
if (normalizedCompared.includes(" ")) {
|
||||
return !normalizedVal.includes(normalizedCompared);
|
||||
}
|
||||
|
||||
// For single word, split into words and check for exact match
|
||||
const words = normalizedVal.split(/\s+/);
|
||||
return !words.some(word => word === normalizedCompared);
|
||||
},
|
||||
|
||||
@ -38,11 +38,14 @@ function getFulltext(_tokens: TokenData[], searchContext: SearchContext, leading
|
||||
|
||||
if (!searchContext.fastSearch) {
|
||||
// For exact match with "=", we need different behavior
|
||||
if (leadingOperator === "=" && tokens.length === 1) {
|
||||
// Exact match on title OR exact match on content
|
||||
if (leadingOperator === "=" && tokens.length >= 1) {
|
||||
// Exact match on title OR exact match on content OR exact match in flat text (includes attributes)
|
||||
// For multi-word, join tokens with space to form exact phrase
|
||||
const titleSearchValue = tokens.join(" ");
|
||||
return new OrExp([
|
||||
new PropertyComparisonExp(searchContext, "title", "=", tokens[0]),
|
||||
new NoteContentFulltextExp("=", { tokens, flatText: false })
|
||||
new PropertyComparisonExp(searchContext, "title", "=", titleSearchValue),
|
||||
new NoteContentFulltextExp("=", { tokens, flatText: false }),
|
||||
new NoteContentFulltextExp("=", { tokens, flatText: true })
|
||||
]);
|
||||
}
|
||||
return new OrExp([new NoteFlatTextExp(tokens), new NoteContentFulltextExp(operator, { tokens, flatText: true })]);
|
||||
|
||||
@ -304,10 +304,13 @@ describe("Search", () => {
|
||||
|
||||
const searchContext = new SearchContext();
|
||||
|
||||
// Test 1: With = and quotes, treat as multi-word exact match (both words must match)
|
||||
// Test 1: With = and quotes, treat as exact phrase match (consecutive words in order)
|
||||
let searchResults = searchService.findResultsWithQuery("='exact phrase'", searchContext);
|
||||
// With current implementation, this searches for notes containing both "exact" and "phrase" words
|
||||
expect(searchResults.length).toEqual(4); // All notes with both words
|
||||
// Should match only notes containing the exact phrase "exact phrase"
|
||||
expect(searchResults.length).toEqual(3); // Only notes with consecutive "exact phrase"
|
||||
expect(findNoteByTitle(searchResults, "exact phrase")).toBeTruthy();
|
||||
expect(findNoteByTitle(searchResults, "exact phrase match")).toBeTruthy();
|
||||
expect(findNoteByTitle(searchResults, "this exact phrase here")).toBeTruthy();
|
||||
|
||||
// Test 2: Without =, quoted phrase should find substring/contains matches
|
||||
searchResults = searchService.findResultsWithQuery("'exact phrase'", searchContext);
|
||||
@ -316,9 +319,10 @@ describe("Search", () => {
|
||||
expect(findNoteByTitle(searchResults, "exact phrase match")).toBeTruthy();
|
||||
expect(findNoteByTitle(searchResults, "this exact phrase here")).toBeTruthy();
|
||||
|
||||
// Test 3: Verify word order doesn't matter with exact word matching
|
||||
// Test 3: Verify word order matters with exact phrase matching
|
||||
searchResults = searchService.findResultsWithQuery("='phrase exact'", searchContext);
|
||||
expect(searchResults.length).toEqual(4); // All notes with both words
|
||||
expect(searchResults.length).toEqual(1); // Only "phrase exact" matches
|
||||
expect(findNoteByTitle(searchResults, "phrase exact")).toBeTruthy();
|
||||
});
|
||||
|
||||
it("leading = operator case sensitivity", () => {
|
||||
@ -368,15 +372,15 @@ describe("Search", () => {
|
||||
expect(findNoteByTitle(searchResults, "test.note")).toBeTruthy();
|
||||
|
||||
// For phrases with spaces, use quotes to keep them together
|
||||
// With word-boundary matching, this finds all notes with both words
|
||||
// With exact phrase matching, this finds notes with the consecutive phrase
|
||||
searchResults = searchService.findResultsWithQuery("='test note'", searchContext);
|
||||
expect(searchResults.length).toEqual(1); // Only "test note" has both words as separate tokens
|
||||
expect(searchResults.length).toEqual(1); // Only "test note" has the exact phrase
|
||||
expect(findNoteByTitle(searchResults, "test note")).toBeTruthy();
|
||||
|
||||
// Without quotes, "test note" is tokenized as two separate words
|
||||
// and will match all notes containing both "test" AND "note" words
|
||||
// Without quotes, "test note" is tokenized as two separate tokens
|
||||
// and will be treated as an exact phrase search with = operator
|
||||
searchResults = searchService.findResultsWithQuery("=test note", searchContext);
|
||||
expect(searchResults.length).toEqual(1); // Only "test note" has both as separate words
|
||||
expect(searchResults.length).toEqual(1); // Only "test note" has the exact phrase
|
||||
|
||||
// Without =, should find all matches containing "test" substring
|
||||
searchResults = searchService.findResultsWithQuery("test", searchContext);
|
||||
|
||||
Loading…
x
Reference in New Issue
Block a user