package attach

import (
	"net/http"
	"slices"
	"testing"
)

// TestDiskExtensionTable (plan 05 of phase 19.3, D-15) pins DiskExtension on
// the RESEARCH probe rows and on the consistency of its tables: a client
// extension that belongs to the sniffed type is kept, any other becomes the
// type's canonical one, and a script-capable extension under a generic type
// becomes txt (text/*) or bin.
func TestDiskExtensionTable(t *testing.T) {
	for _, tc := range []struct {
		contentType, clientExt, want string
	}{
		// RESEARCH probe rows.
		{"image/png", "html", "png"},
		{"image/jpeg", "jpeg", "jpeg"},
		{"image/jpeg", "jpg", "jpg"},
		{"image/jpeg", "png", "jpg"},
		{"video/webm", "mkv", "mkv"},
		{"video/webm", "webm", "webm"},
		{"application/zip", "docx", "docx"},
		{"application/zip", "xlsx", "xlsx"},
		{"text/plain; charset=utf-8", "ics", "ics"},
		{"text/plain; charset=utf-8", "html", "txt"},
		{"text/xml; charset=utf-8", "svg", "txt"},
		{"application/octet-stream", "js", "bin"},
		{"application/pdf", "txt", "pdf"},
		// HTML bytes under their own name never stay html.
		{"text/html; charset=utf-8", "html", "txt"},
		{"text/html; charset=utf-8", "txt", "txt"},
		// Normalisation: case, a leading dot, parameters and spaces.
		{"IMAGE/PNG", ".PNG", "png"},
		{" image/jpeg ; q=1", "JPE", "jpe"},
		{"application/octet-stream", ".Docx", "docx"},
		// Extensions outside [a-z0-9]{1,10} fall back like a script-capable one.
		{"application/octet-stream", "", "bin"},
		{"text/plain", "", "txt"},
		{"application/octet-stream", "tar.gz", "bin"},
		{"text/plain", "averyverylongext", "txt"},
		{"application/octet-stream", "ex e", "bin"},
		// Other specific types.
		{"image/gif", "png", "gif"},
		{"image/webp", "webp", "webp"},
		{"application/ogg", "opus", "opus"},
		{"audio/mpeg", "wav", "mp3"},
		{"font/woff2", "woff2", "woff2"},
		{"application/x-gzip", "tgz", "tgz"},
	} {
		if got := DiskExtension(tc.contentType, tc.clientExt); got != tc.want {
			t.Errorf("DiskExtension(%q, %q) = %q, want %q", tc.contentType, tc.clientExt, got, tc.want)
		}
	}

	// Every specific type keeps each of its own extensions, maps a
	// script-capable one to its canonical extension, and lists no
	// script-capable extension itself.
	for ct, exts := range sniffedExtensions {
		if len(exts) == 0 {
			t.Fatalf("%s has no extensions", ct)
		}
		for _, ext := range exts {
			if !extPattern.MatchString(ext) {
				t.Errorf("%s: extension %q does not match the disk name pattern", ct, ext)
			}
			if slices.Contains(scriptCapableExtensions, ext) {
				t.Errorf("%s: script-capable extension %q is kept", ct, ext)
			}
			if got := DiskExtension(ct, ext); got != ext {
				t.Errorf("DiskExtension(%q, %q) = %q", ct, ext, got)
			}
		}
		for _, bad := range scriptCapableExtensions {
			if got := DiskExtension(ct, bad); got != exts[0] {
				t.Errorf("DiskExtension(%q, %q) = %q, want %q", ct, bad, got, exts[0])
			}
		}
	}

	// Under every generic type a script-capable extension is rewritten.
	for _, ct := range []string{"text/plain; charset=utf-8", "text/html; charset=utf-8", "text/xml; charset=utf-8", "text/css"} {
		for _, bad := range scriptCapableExtensions {
			if got := DiskExtension(ct, bad); got != "txt" {
				t.Errorf("DiskExtension(%q, %q) = %q, want txt", ct, bad, got)
			}
		}
	}
	for _, ct := range []string{"application/octet-stream", "application/zip", "application/x-msdownload", ""} {
		for _, bad := range scriptCapableExtensions {
			if got := DiskExtension(ct, bad); got != "bin" {
				t.Errorf("DiskExtension(%q, %q) = %q, want bin", ct, bad, got)
			}
		}
	}

	// The table agrees with what the standard library actually sniffs.
	for body, want := range map[string]string{
		"\x89PNG\r\n\x1a\n":           "png",
		"%PDF-1.7\n":                  "pdf",
		"<html><body>x</body></html>": "txt",
		"<?xml version=\"1.0\"?><svg xmlns=\"http://www.w3.org/2000/svg\"/>": "txt",
	} {
		ct := http.DetectContentType([]byte(body))
		if got := DiskExtension(ct, "html"); got != want {
			t.Errorf("body %q sniffed %q: DiskExtension = %q, want %q", body, ct, got, want)
		}
	}
}