test: add parse_csv.test.ts and parse_file.test.ts

tests/parse_csv.test.ts

@@ -0,0 +1,85 @@
+import { describe, expect, it } from "bun:test";
+import type { ColumnName, Separator, Translation } from "../src/interfaces";
+import { getThesaurus } from "../src/utils";
+import "uniformize";
+
+const mockTranslation = ((key: string, options?: any) => {
+	const translations: { [key: string]: string } = {
+		"error.csv.columns": "Invalid column names",
+		"error.csv.separator": `Invalid separator: ${options?.sep}`,
+		"error.csv.header": `Invalid header length: ${options?.len}`,
+		"error.csv.malformed": `Malformed line with length: ${options?.len}`,
+	};
+	return translations[key] || key;
+}) as Translation;
+
+describe("getThesaurus", () => {
+	const columnNames: ColumnName = { term: "Term", synonyms: "Synonyms" };
+
+	it("should parse a valid CSV content", () => {
+		const csvContent = "Term,Synonyms\nword1,synonym1\nword2,synonym2";
+		const separator: Separator = ",";
+		const thesaurus = getThesaurus(csvContent, separator, mockTranslation, columnNames);
+		expect(thesaurus).toEqual({
+			word1: new Set(["synonym1"]),
+			word2: new Set(["synonym2"]),
+		});
+	});
+
+	it("should throw an error for invalid column names", () => {
+		const csvContent = "Invalid,Header\nword1,synonym1";
+		const separator: Separator = ",";
+		expect(() =>
+			getThesaurus(csvContent, separator, mockTranslation, columnNames)
+		).toThrow("Invalid column names");
+	});
+
+	it("should throw an error for invalid separator", () => {
+		const csvContent = "Term;Synonyms\nword1;synonym1";
+		const separator: Separator = ",";
+		expect(() =>
+			getThesaurus(csvContent, separator, mockTranslation, columnNames)
+		).toThrow("Invalid separator: ;");
+	});
+
+	it("should throw an error for invalid header length", () => {
+		const csvContent = "Term,Synonyms,Extra\nword1,synonym1";
+		const separator: Separator = ",";
+		expect(() =>
+			getThesaurus(csvContent, separator, mockTranslation, columnNames)
+		).toThrow("Invalid header length: 3");
+	});
+
+	it("should throw an error for malformed line", () => {
+		const csvContent = "Term,Synonyms\nword1,synonym1\nword2";
+		const separator: Separator = ",";
+		expect(() =>
+			getThesaurus(csvContent, separator, mockTranslation, columnNames)
+		).toThrow("Malformed line with length: 1");
+	});
+
+	it("should not have duplicate terms in the result", () => {
+		const csvContent = "Term,Synonyms\nword1,synonym1\nword1,synonym2";
+		const separator: Separator = ",";
+		const thesaurus = getThesaurus(csvContent, separator, mockTranslation, columnNames);
+		expect(Object.keys(thesaurus).length).toBe(1);
+		expect(thesaurus["word1"]).toEqual(new Set(["synonym1", "synonym2"]));
+	});
+
+	it("should not have duplicate synonyms in the result", () => {
+		const csvContent = "Term,Synonyms\nword1,synonym1\nword1,synonym1";
+		const separator: Separator = ",";
+		const thesaurus = getThesaurus(csvContent, separator, mockTranslation, columnNames);
+		expect(Object.keys(thesaurus).length).toBe(1);
+		expect(thesaurus["word1"]).toEqual(new Set(["synonym1"]));
+	});
+
+	it("should not have duplicate terms and synonyms in the result", () => {
+		const csvContent = "Term,Synonyms\nword1,synonym1\nword2,synonym1";
+		const separator: Separator = ",";
+		const thesaurus = getThesaurus(csvContent, separator, mockTranslation, columnNames);
+		expect(Object.keys(thesaurus).length).toBe(2);
+		expect(thesaurus["word1"]).toEqual(new Set(["synonym1"]));
+		expect(thesaurus["word2"]).toEqual(new Set(["synonym1"]));
+	});
+});

tests/parse_file.test.ts

@@ -0,0 +1,60 @@
+import { describe, expect, it } from "bun:test";
+import type { Thesaurus } from "../src/interfaces";
+import { getTags } from "../src/utils";
+
+describe("getTags", () => {
+	it("should return an empty array if no tags are found", () => {
+		const content = "This is a test content without any tags.";
+		const thesaurus: Thesaurus = {
+			tag1: new Set(["synonym1", "synonym2"]),
+			tag2: new Set(["synonym3", "synonym4"]),
+		};
+
+		const result = getTags(content, thesaurus);
+		expect(result).toEqual([]);
+	});
+
+	it("should return tags if synonyms are found in the content", () => {
+		const content = "This content contains synonym1 and synonym4.";
+		const thesaurus: Thesaurus = {
+			tag1: new Set(["synonym1", "synonym2"]),
+			tag2: new Set(["synonym3", "synonym4"]),
+		};
+
+		const result = getTags(content, thesaurus);
+		expect(result).toEqual(["tag1", "tag2"]);
+	});
+
+	it("should handle case insensitive matches", () => {
+		const content = "This content contains Synonym1 and SYNONYM4.";
+		const thesaurus: Thesaurus = {
+			tag1: new Set(["synonym1", "synonym2"]),
+			tag2: new Set(["synonym3", "synonym4"]),
+		};
+
+		const result = getTags(content, thesaurus);
+		expect(result).toEqual(["tag1", "tag2"]);
+	});
+
+	it("should not return duplicate tags", () => {
+		const content = "This content contains synonym1 and synonym1 again.";
+		const thesaurus: Thesaurus = {
+			tag1: new Set(["synonym1", "synonym2"]),
+			tag2: new Set(["synonym3", "synonym4"]),
+		};
+
+		const result = getTags(content, thesaurus);
+		expect(result).toEqual(["tag1"]);
+	});
+
+	it("should return the two tags related to the synonyms", () => {
+		const content = "This content contains synonym4.";
+		const thesaurus: Thesaurus = {
+			tag1: new Set(["synonym1", "synonym4"]),
+			tag2: new Set(["synonym3", "synonym4"]),
+		};
+
+		const result = getTags(content, thesaurus);
+		expect(result).toEqual(["tag1", "tag2"]);
+	});
+});