feat: add utf8bom and utf16 detection detection

src/index.spec.ts

@@ -184,4 +184,22 @@ describe("Tests the public API", () => {
     expect(result).toHaveLength(0);
     expect(result).toEqual([]);
   });
+
+  it("detects utf8", () => {
+    const file = getBytes("a.utf8");
+    const result = filetypemime(file);
+    expect(result).toContain("text/plain; charset=UTF-8");
+  });
+
+  it("detects utf16le", () => {
+    const file = getBytes("a.utf16le");
+    const result = filetypemime(file);
+    expect(result).toContain("text/plain; charset=UTF-16LE");
+  });
+
+  it("detects utf16be", () => {
+    const file = getBytes("a.utf16be");
+    const result = filetypemime(file);
+    expect(result).toContain("text/plain; charset=UTF-16BE");
+  });
 });

src/index.ts

@@ -52,9 +52,20 @@ const walkTree = (
 
 export default filetypeinfo;
 
-export const filetypename = (bytes: number[] | Uint8Array | Uint8ClampedArray): string[] =>
-  filetypeinfo(bytes).map((e) => e.typename);
-export const filetypemime = (bytes: number[] | Uint8Array | Uint8ClampedArray): string[] =>
-  filetypeinfo(bytes).map((e) => (e.mime ? e.mime : ""));
-export const filetypeextension = (bytes: number[] | Uint8Array | Uint8ClampedArray): string[] =>
-  filetypeinfo(bytes).map((e) => (e.extension ? e.extension : ""));
+export const filetypename = (
+  bytes: number[] | Uint8Array | Uint8ClampedArray
+): string[] => filetypeinfo(bytes).map((e) => e.typename);
+
+export const filetypemime = (
+  bytes: number[] | Uint8Array | Uint8ClampedArray
+): string[] =>
+  filetypeinfo(bytes)
+    .map((e) => (e.mime ? e.mime : null))
+    .filter((x) => x !== null) as string[];
+
+export const filetypeextension = (
+  bytes: number[] | Uint8Array | Uint8ClampedArray
+): string[] =>
+  filetypeinfo(bytes)
+    .map((e) => (e.extension ? e.extension : null))
+    .filter((x) => x !== null) as string[];

src/model/pattern-tree.ts

@@ -1170,6 +1170,25 @@ add(
   0x4
 );
 
-add("hl2demo", ["48", "4C", "32", "44", "45", "4D", "4F"]);
-
+add("hl2demo", ["0x48", "0x4C", "0x32", "0x44", "0x45", "0x4D", "0x4F"]);
+add("txt", ["0xEF", "0xBB", "0xBF"], {
+  mime: "text/plain; charset=UTF-8",
+  extension: "txt",
+});
+add("txt", ["0xFF", "0xFE"], {
+  mime: "text/plain; charset=UTF-16LE",
+  extension: "txt",
+});
+add("txt", ["0xFE", "0xFF"], {
+  mime: "text/plain; charset=UTF-16BE",
+  extension: "txt",
+});
+add("txt", ["0xFF", "0xFE", "0x00", "0x00"], {
+  mime: "text/plain; charset=UTF-32LE",
+  extension: "txt",
+});
+add("txt", ["00", "00", "FE", "FF"], {
+  mime: "text/plain; charset=UTF-32BE",
+  extension: "txt",
+});
 export default (): Tree => tree as Tree;

src/testfiles/a.utf8

@@ -0,0 +1 @@
+thusisutf8