Initial Commit

packages/foundation/data/mime/src/charset/charset.test.ts

@@ -0,0 +1,132 @@
+import { describe, expect, it } from "@suaveplan/testing/runner";
+import { sniffBom } from "./sniff-bom.js";
+import { sniffHtmlMeta } from "./sniff-html-meta.js";
+import { sniffXmlDecl } from "./sniff-xml-decl.js";
+
+describe("sniffBom", () => {
+    it("detects UTF-8 BOM (EF BB BF)", { timeout: 5000 }, () => {
+        const bytes = new Uint8Array([0xef, 0xbb, 0xbf, 0x68, 0x65]);
+        const result = sniffBom(bytes);
+        expect(result?.charset).toBe("utf-8");
+        expect(result?.bomLength).toBe(3);
+    });
+
+    it("detects UTF-16 LE BOM (FF FE)", { timeout: 5000 }, () => {
+        const bytes = new Uint8Array([0xff, 0xfe, 0x68, 0x00]);
+        const result = sniffBom(bytes);
+        expect(result?.charset).toBe("utf-16le");
+        expect(result?.bomLength).toBe(2);
+    });
+
+    it("detects UTF-16 BE BOM (FE FF)", { timeout: 5000 }, () => {
+        const bytes = new Uint8Array([0xfe, 0xff, 0x00, 0x68]);
+        const result = sniffBom(bytes);
+        expect(result?.charset).toBe("utf-16be");
+        expect(result?.bomLength).toBe(2);
+    });
+
+    it("detects UTF-32 LE BOM (FF FE 00 00)", { timeout: 5000 }, () => {
+        const bytes = new Uint8Array([0xff, 0xfe, 0x00, 0x00, 0x68]);
+        const result = sniffBom(bytes);
+        expect(result?.charset).toBe("utf-32le");
+        expect(result?.bomLength).toBe(4);
+    });
+
+    it("detects UTF-32 BE BOM (00 00 FE FF)", { timeout: 5000 }, () => {
+        const bytes = new Uint8Array([0x00, 0x00, 0xfe, 0xff, 0x68]);
+        const result = sniffBom(bytes);
+        expect(result?.charset).toBe("utf-32be");
+        expect(result?.bomLength).toBe(4);
+    });
+
+    it("returns null for no BOM", { timeout: 5000 }, () => {
+        const bytes = new Uint8Array([0x68, 0x65, 0x6c, 0x6c, 0x6f]);
+        expect(sniffBom(bytes)).toBeNull();
+    });
+
+    it("returns null for empty buffer", { timeout: 5000 }, () => {
+        expect(sniffBom(new Uint8Array(0))).toBeNull();
+    });
+
+    it("accepts ArrayBuffer", { timeout: 5000 }, () => {
+        const buf = new Uint8Array([0xef, 0xbb, 0xbf, 0x41]).buffer;
+        expect(sniffBom(buf)?.charset).toBe("utf-8");
+    });
+
+    it("UTF-32 LE takes priority over UTF-16 LE when first 4 bytes are FF FE 00 00", {
+        timeout: 5000,
+    }, () => {
+        const bytes = new Uint8Array([0xff, 0xfe, 0x00, 0x00]);
+        expect(sniffBom(bytes)?.charset).toBe("utf-32le");
+    });
+});
+
+describe("sniffHtmlMeta", () => {
+    it("detects charset from <meta charset>", { timeout: 5000 }, () => {
+        const html = '<html><head><meta charset="UTF-8"></head></html>';
+        expect(sniffHtmlMeta(html)).toBe("utf-8");
+    });
+
+    it("detects charset from <meta charset> without quotes", {
+        timeout: 5000,
+    }, () => {
+        const html = "<meta charset=UTF-8>";
+        expect(sniffHtmlMeta(html)).toBe("utf-8");
+    });
+
+    it("detects charset from http-equiv content-type", {
+        timeout: 5000,
+    }, () => {
+        const html =
+            '<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">';
+        expect(sniffHtmlMeta(html)).toBe("iso-8859-1");
+    });
+
+    it("returns null when no charset found", { timeout: 5000 }, () => {
+        expect(
+            sniffHtmlMeta("<html><head><title>Hello</title></head></html>")
+        ).toBeNull();
+    });
+
+    it("only scans first 1024 characters", { timeout: 5000 }, () => {
+        const prefix = "x".repeat(1100);
+        const html = `${prefix}<meta charset="UTF-8">`;
+        expect(sniffHtmlMeta(html)).toBeNull();
+    });
+
+    it("returns lowercase charset", { timeout: 5000 }, () => {
+        expect(sniffHtmlMeta('<meta charset="ISO-8859-1">')).toBe("iso-8859-1");
+    });
+});
+
+describe("sniffXmlDecl", () => {
+    it("detects encoding from XML declaration", { timeout: 5000 }, () => {
+        const xml = '<?xml version="1.0" encoding="UTF-8"?><root/>';
+        expect(sniffXmlDecl(xml)).toBe("utf-8");
+    });
+
+    it("detects encoding with single quotes", { timeout: 5000 }, () => {
+        const xml = "<?xml version='1.0' encoding='ISO-8859-1'?><root/>";
+        expect(sniffXmlDecl(xml)).toBe("iso-8859-1");
+    });
+
+    it("returns null when no encoding attribute", { timeout: 5000 }, () => {
+        const xml = '<?xml version="1.0"?><root/>';
+        expect(sniffXmlDecl(xml)).toBeNull();
+    });
+
+    it("returns null for non-XML content", { timeout: 5000 }, () => {
+        expect(sniffXmlDecl("<html>not xml</html>")).toBeNull();
+    });
+
+    it("only scans first 256 characters", { timeout: 5000 }, () => {
+        const prefix = "x".repeat(300);
+        const xml = `${prefix}<?xml version="1.0" encoding="UTF-8"?>`;
+        expect(sniffXmlDecl(xml)).toBeNull();
+    });
+
+    it("returns lowercase encoding", { timeout: 5000 }, () => {
+        const xml = '<?xml version="1.0" encoding="UTF-16"?>';
+        expect(sniffXmlDecl(xml)).toBe("utf-16");
+    });
+});