dfwad: detect UTF-8

[d2df-editor.git] / src / shared / WADEDITOR_dfzip.pas

diff --git a/src/shared/WADEDITOR_dfzip.pas b/src/shared/WADEDITOR_dfzip.pas
index c73811a3c447bd639755ddd8587c52804610428a..f9e54ad57f88f597a6a56fdf8e0fac7b8b1f4136 100644 (file)
--- a/src/shared/WADEDITOR_dfzip.pas
+++ b/src/shared/WADEDITOR_dfzip.pas
@@ -6,7 +6,7 @@ unit WADEDITOR_dfzip;
 // - File must start with LFH or EOCD signature
 // - EOCD must be located strictly at the end of file
 // - Multi-disk ZIP files are not supported
-// - UTF-8 not supported yet, expected WIN1251 encoding
+// - Expect UTF-8 or CP1251 encoded names
 // - ZIP64 not supported
 // - Encryption not supported
 // - Zero-length file names not supported
@@ -133,6 +133,50 @@ implementation
     ZIP_ENCRYPTION_MASK = (1 << 0) or (1 << 6) or (1 << 13);
     ZIP_UTF8_MASK = (1 << 11);
 
+  function IsASCII(const s: AnsiString): Boolean;
+    var i: Integer;
+  begin
+    for i := 1 to Length(s) do
+    begin
+      if s[i] >= #$80 then
+      begin
+        Result := False;
+        exit;
+      end;
+    end;
+    Result := True;
+  end;
+
+  function IsUTF8(const s: AnsiString): Boolean;
+    var i, j, len: Integer;
+  begin
+    Result := False;
+    i := 1; len := Length(s);
+    while i <= len do
+    begin
+      case Ord(s[i]) of
+        $00..$7F: j := 0;
+        $80..$BF: exit; // invalid encoding
+        $C0..$DF: j := 1;
+        $E0..$EF: j := 2;
+        $F0..$F7: j := 3;
+        otherwise exit; // invalid encoding
+      end;
+      Inc(i);
+      while j > 0 do
+      begin
+        if i > len then exit; // invlid length
+        case Ord(s[i]) of
+          $80..$BF: ; // ok
+          else exit;  // invalid encoding
+        end;
+        Inc(i);
+        Dec(j);
+      end;
+    end;
+    Result := True;
+  end;
+
   procedure ToSectionFile(fname: AnsiString; out section, name: AnsiString); inline;
     var i: SizeInt;
   begin
@@ -728,6 +772,7 @@ implementation
     var mypos, next: UInt64;
     var name: PChar;
     var aname: AnsiString;
+    var cvtbug, utf8: Boolean;
   begin
     mypos := s.Position;
     s.ReadBuffer(sig[0], 4);
@@ -773,6 +818,15 @@ implementation
         e_WriteLog('CDR#' + IntToStr(cdrid) + ' @' + IntToHex(mypos, 8) + ': External Attrib   : $' + IntToHex(eattr, 8), MSG_NOTIFY);
         e_WriteLog('CDR#' + IntToStr(cdrid) + ' @' + IntToHex(mypos, 8) + ': LFH Offset        : $' + IntToHex(offset, 8), MSG_NOTIFY);
       end;
+      cvtbug := False;
+      if (vva = $10) and (vvb = $0A) and (va = $10) and (vb = $00) and (flags = (1 << 10)) and (mtime = 0) and (iattr = 0) and (eattr = 0) then
+      begin
+        // HACK: Editor and wadcvt for long time sets incorrent flag for UTF-8
+        flags := ZIP_UTF8_MASK;
+        cvtbug := True;
+      end;
+      if gWADEditorLogLevel >= DFWAD_LOG_DEBUG then
+        e_WriteLog('CDR#' + IntToStr(cdrid) + ' @' + IntToHex(mypos, 8) + ': WADCVT BUG        : ' + BoolToStr(cvtbug, True), MSG_NOTIFY);
       if (va >= 10) and (va <= ZIP_MAXVERSION) then
       begin
         if (flags and ZIP_ENCRYPTION_MASK) = 0 then
@@ -818,13 +872,18 @@ implementation
                 try
                   s.ReadBuffer(name[0], fnlen);
                   name[fnlen] := #0;
-                  // TODO: when packer version < 63 detect utf8 by hands?
-                  if flags and ZIP_UTF8_MASK = 0 then
-                    aname := win2utf(name)
-                  else
-                    aname := name;
+                  aname := name;
+                  utf8 := True;
+                  if (flags and ZIP_UTF8_MASK = 0) and (IsUTF8(name) = False) then
+                  begin
+                    aname := win2utf(aname);
+                    utf8 := False;
+                  end;
                   if gWADEditorLogLevel >= DFWAD_LOG_DEBUG then
+                  begin
+                    e_WriteLog('CDR#' + IntToStr(cdrid) + ' @' + IntToHex(mypos, 8) + ': UTF-8 Comatible   : ' + BoolToStr(utf8, True), MSG_NOTIFY);
                     e_WriteLog('CDR#' + IntToStr(cdrid) + ' @' + IntToHex(mypos, 8) + ': Name              : "' + aname + '"', MSG_NOTIFY);
+                  end;
                   s.Seek(offset, TSeekOrigin.soBeginning);
                   ReadLFH(s, aname, csize, usize, comp, crc);
                 finally
@@ -1041,20 +1100,6 @@ implementation
     end;
   end;
 
-  function IsASCII(const s: AnsiString): Boolean;
-    var i: Integer;
-  begin
-    for i := 1 to Length(s) do
-    begin
-      if s[i] >= #$80 then
-      begin
-        Result := False;
-        exit;
-      end;
-    end;
-    Result := True;
-  end;
-
   function GetZIPVersion(const afname: AnsiString; flags, comp: UInt16): UInt8;
     var version: UInt8;
   begin