Password-Manager/delphi-backend/Source/PM.Favicon.pas

unit PM.Favicon;

{
  Favicon proxy — fetches a website's icon and returns a base64 data URI
  ready to drop into an <img src="...">.

  Source: DuckDuckGo's icons service (icons.duckduckgo.com/ip3/<host>.ico)
    - No tracking pixels / analytics on the icon endpoints
    - Returns a 32×32 PNG (or ICO) with the proper MIME type
    - Centralised: only DDG sees the list of domains the user looks up,
      vs hitting each site's /favicon.ico directly (which would leak the
      full vault contents to every site listed)
    - Falls back to a generic globe glyph for unknown sites
  HTTPS only; 5s timeout; cap response at 64 KB; no redirects beyond 3.

  Threat model: this is the ONLY outbound network call from Delphi (HIBP is
  done client-side). The user explicitly opts in via Settings. Failure
  modes (DNS, TLS, 4xx, oversize) all return '' — caller falls back to
  the first-letter avatar.
}

interface

// Fetches an icon for AHost (bare hostname, no scheme). Returns a
// "data:image/...;base64,..." string on success, or '' on any failure.
function FetchFaviconDataUri(const AHost: string): string;

implementation

uses
  System.SysUtils, System.Classes, System.NetEncoding,
  System.Net.HttpClient, System.Net.URLClient;

const
  ICON_URL_TEMPLATE  = 'https://icons.duckduckgo.com/ip3/%s.ico';
  MAX_ICON_BYTES     = 65536;   // 64 KB cap (matches handler's SetEntryIcon limit)
  HTTP_TIMEOUT_MS    = 5000;

function NormalizeHost(const ARaw: string): string;
var
  S: string;
  SlashPos, ColonPos, I: Integer;
  Ch: Char;
begin
  // Accept anything user-typed: "https://www.github.com/login", "github.com",
  // "GitHub.com:8443". Return lowercase bare hostname, or '' if the input
  // doesn't look like a real domain — defense in depth alongside the JS
  // faviconHost() validation (so a future bridge caller can't leak a
  // brand label like "Gitea" upstream).
  Result := '';
  S := Trim(ARaw).ToLower;
  if S.StartsWith('https://') then S := Copy(S, 9, MaxInt)
  else if S.StartsWith('http://')  then S := Copy(S, 8, MaxInt);
  if S.StartsWith('www.') then S := Copy(S, 5, MaxInt);
  SlashPos := Pos('/', S);
  if SlashPos > 0 then S := Copy(S, 1, SlashPos - 1);
  ColonPos := Pos(':', S);
  if ColonPos > 0 then S := Copy(S, 1, ColonPos - 1);

  if (S = '') or (Length(S) > 253) then Exit;
  // Must contain a dot, no leading/trailing dot, no consecutive dots,
  // only [a-z0-9.-] characters.
  if Pos('.', S) < 2 then Exit;
  if S.StartsWith('.') or S.EndsWith('.') or S.Contains('..') then Exit;
  for I := 1 to Length(S) do
  begin
    Ch := S[I];
    if not (((Ch >= 'a') and (Ch <= 'z')) or
            ((Ch >= '0') and (Ch <= '9')) or
            (Ch = '.') or (Ch = '-')) then
      Exit;
  end;
  Result := S;
end;

function GuessMimeFromBytes(const ABytes: TBytes): string;
begin
  // Lightweight magic-byte sniffing. Saves a Content-Type round-trip parse.
  Result := 'image/x-icon';   // safe default for an .ico fetch
  if Length(ABytes) < 8 then Exit;
  // PNG  : 89 50 4E 47 0D 0A 1A 0A
  if (ABytes[0] = $89) and (ABytes[1] = $50) and (ABytes[2] = $4E) and (ABytes[3] = $47) then
    Exit('image/png');
  // GIF  : "GIF8"
  if (ABytes[0] = Ord('G')) and (ABytes[1] = Ord('I')) and
     (ABytes[2] = Ord('F')) and (ABytes[3] = Ord('8')) then
    Exit('image/gif');
  // JPEG : FF D8 FF
  if (ABytes[0] = $FF) and (ABytes[1] = $D8) and (ABytes[2] = $FF) then
    Exit('image/jpeg');
  // SVG  : "<svg" or "<?xml" (text-prefixed)
  if (ABytes[0] = Ord('<')) then Exit('image/svg+xml');
  // ICO  : 00 00 01 00
  if (ABytes[0] = $00) and (ABytes[1] = $00) and
     (ABytes[2] = $01) and (ABytes[3] = $00) then
    Exit('image/x-icon');
end;

function FetchFaviconDataUri(const AHost: string): string;
var
  LHost, LUrl, LMime, LBase64: string;
  LHttp: THTTPClient;
  LResp: IHTTPResponse;
  LStream: TMemoryStream;
  LBytes: TBytes;
begin
  Result := '';
  LHost := NormalizeHost(AHost);
  if LHost = '' then Exit;

  LUrl := Format(ICON_URL_TEMPLATE, [LHost]);

  // THTTPClient wraps WinHTTP on Windows → native TLS, system cert store,
  // zero extra DLLs to ship next to the exe (unlike Indy + OpenSSL which
  // fails silently when libcrypto/libssl are missing).
  LHttp := THTTPClient.Create;
  LStream := TMemoryStream.Create;
  try
    LHttp.ConnectionTimeout := HTTP_TIMEOUT_MS;
    LHttp.ResponseTimeout   := HTTP_TIMEOUT_MS;
    LHttp.HandleRedirects   := True;
    LHttp.MaxRedirects      := 3;
    LHttp.UserAgent         := 'PMServer/1.0 (favicon-fetch)';
    LHttp.CustHeaders.Add('Accept',
      'image/png,image/x-icon,image/*;q=0.8,*/*;q=0.1');

    try
      LResp := LHttp.Get(LUrl, LStream);
    except
      // Any DNS / connect / TLS failure → silent ''.
      Exit;
    end;

    if (LResp = nil) or (LResp.StatusCode <> 200) then Exit;
    if LStream.Size <= 0 then Exit;
    if LStream.Size > MAX_ICON_BYTES then Exit;

    LStream.Position := 0;
    SetLength(LBytes, LStream.Size);
    LStream.ReadBuffer(LBytes[0], LStream.Size);

    LMime   := GuessMimeFromBytes(LBytes);
    LBase64 := TNetEncoding.Base64.EncodeBytesToString(LBytes);
    // Strip CR/LF that the encoder inserts every 76 chars — invalid inside
    // an <img src="..."> attribute and bloats the cached blob.
    LBase64 := StringReplace(LBase64, #13, '', [rfReplaceAll]);
    LBase64 := StringReplace(LBase64, #10, '', [rfReplaceAll]);

    Result := 'data:' + LMime + ';base64,' + LBase64;
  finally
    LStream.Free;
    LHttp.Free;
  end;
end;

end.