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

type
  TFaviconLog = reference to procedure(const ALine: string);

// Fetches an icon for AHost (bare hostname, no scheme). Returns a
// "data:image/...;base64,..." string on success, or '' on any failure.
// ALog (optional): called for each fallback step so the host can trace
// exactly which URL hit / missed.
function FetchFaviconDataUri(const AHost: string;
  ALog: TFaviconLog = nil): 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      = 262144;  // 256 KB cap (DDG sometimes serves full-res
                                  // assets; matches handler + JS upload limits)
  HTTP_TIMEOUT_MS     = 5000;
  // DDG returns a generic placeholder for unknown domains. Bigger threshold
  // than 100 to avoid treating its blank globe glyph as a real icon.
  MIN_REAL_ICON_BYTES = 300;
  // Privacy stance: DDG-only fetches. We don't fall back to the site's
  // own /favicon.ico because that would leak DNS to every domain stored
  // in the vault. For sites DDG doesn't index, the user can upload a
  // custom icon via the slideover (soIconField).

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;

// "chat.deepseek.com" → "deepseek.com". Returns '' if S has no dot or
// is already a 2-label hostname (we'd fall back to the same input).
function ExtractSLD(const S: string): string;
var
  DotCount, FirstDot: Integer;
  I: Integer;
begin
  Result := '';
  DotCount := 0;
  FirstDot := 0;
  for I := 1 to Length(S) do
    if S[I] = '.' then
    begin
      Inc(DotCount);
      if FirstDot = 0 then FirstDot := I;
    end;
  if DotCount < 2 then Exit;   // already SLD or no dots
  Result := Copy(S, FirstDot + 1, MaxInt);
end;

function FetchOneIcon(const AUrl: string;
  out ABytes: TBytes): Boolean; forward;

function FetchFaviconDataUri(const AHost: string;
  ALog: TFaviconLog = nil): string;

  procedure Trace(const ALine: string);
  begin
    if Assigned(ALog) then ALog(ALine);
  end;

var
  LHost, LSld, LMime, LBase64, LUrl: string;
  LBytes: TBytes;
  LOk: Boolean;
begin
  Result := '';
  LHost := NormalizeHost(AHost);
  if LHost = '' then
  begin
    Trace('reject: "' + AHost + '" not a valid hostname');
    Exit;
  end;

  // Strategy: prefer the SLD (brand domain) when the host has a subdomain,
  // because DDG often returns a generic placeholder for chat.X.com / app.X.com
  // / etc. (passes our byte threshold but looks wrong) while having the real
  // brand icon under X.com. For bare 2-label hosts we go straight to step 2.

  LSld := ExtractSLD(LHost);
  LOk := False;

  // 1) DDG SLD first when host has a subdomain (e.g. chat.deepseek.com →
  //    try deepseek.com.ico first). Skipped for bare hosts.
  if LSld <> '' then
  begin
    LUrl := Format(ICON_URL_TEMPLATE, [LSld]);
    if FetchOneIcon(LUrl, LBytes) then
    begin
      if Length(LBytes) >= MIN_REAL_ICON_BYTES then
      begin
        LOk := True;
        Trace(Format('OK step1 DDG sld: %s (%d bytes)', [LUrl, Length(LBytes)]));
      end
      else
        Trace(Format('skip step1 DDG sld: %s only %d bytes', [LUrl, Length(LBytes)]));
    end
    else
      Trace('fail step1 DDG sld: ' + LUrl);
  end;

  // 2) DDG full host as fallback (covers brands whose subdomain has its own
  //    distinct icon, OR plain hosts like github.com that have no SLD step).
  if not LOk then
  begin
    var LTry: TBytes;
    LUrl := Format(ICON_URL_TEMPLATE, [LHost]);
    if FetchOneIcon(LUrl, LTry) then
    begin
      if Length(LTry) >= MIN_REAL_ICON_BYTES then
      begin
        LBytes := LTry; LOk := True;
        Trace(Format('OK step2 DDG host: %s (%d bytes)', [LUrl, Length(LTry)]));
      end
      else
        Trace(Format('skip step2 DDG host: %s only %d bytes', [LUrl, Length(LTry)]));
    end
    else
      Trace('fail step2 DDG host: ' + LUrl);
  end;

  if (not LOk) or (Length(LBytes) = 0) then
  begin
    Trace('DDG has no icon for ' + LHost + ' — user can upload a custom one');
    Exit;
  end;

  LMime   := GuessMimeFromBytes(LBytes);
  LBase64 := TNetEncoding.Base64.EncodeBytesToString(LBytes);
  LBase64 := StringReplace(LBase64, #13, '', [rfReplaceAll]);
  LBase64 := StringReplace(LBase64, #10, '', [rfReplaceAll]);
  Result := 'data:' + LMime + ';base64,' + LBase64;
end;

// Low-level HTTP GET. Returns False on any failure (DNS, TLS, non-200,
// oversize). On success ABytes contains the raw image bytes.
// THTTPClient wraps WinHTTP on Windows → native TLS, system cert store,
// zero extra DLLs to ship next to the exe.
function FetchOneIcon(const AUrl: string; out ABytes: TBytes): Boolean;
var
  LHttp: THTTPClient;
  LResp: IHTTPResponse;
  LStream: TMemoryStream;
begin
  Result := False;
  SetLength(ABytes, 0);

  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(AUrl, LStream);
    except
      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(ABytes, LStream.Size);
    LStream.ReadBuffer(ABytes[0], LStream.Size);
    Result := True;
  finally
    LStream.Free;
    LHttp.Free;
  end;
end;

end.