X-Git-Url: http://deadsoftware.ru/gitweb?p=d2df-sdl.git;a=blobdiff_plain;f=src%2Fshared%2Futils.pas;h=946e4f5a07c7a31a5422fad1f2f479fb83ffce6a;hp=4c4db7b0df59997698bca6e3d5861cb9ba9a01a0;hb=1b867d9fa61767acfef258fe29b900d3fd1af104;hpb=8e694e6ddeb42110a00233d7ff7345c57ee5d90d

diff --git a/src/shared/utils.pas b/src/shared/utils.pas
index 4c4db7b..946e4f5 100644
--- a/src/shared/utils.pas
+++ b/src/shared/utils.pas
@@ -26,6 +26,17 @@ type
   SSArray = array of ShortString;
 
 
+const wadExtensions: array [0..6] of AnsiString = (
+  '.dfz',
+  '.wad',
+  '.dfwad',
+  '.pk3',
+  '.pak',
+  '.zip',
+  '.dfzip'
+);
+
+
 // ////////////////////////////////////////////////////////////////////////// //
 type
   TUtf8DecoderFast = packed record
@@ -60,6 +71,16 @@ function getFilenameExt (const fn: AnsiString): AnsiString;
 function setFilenameExt (const fn, ext: AnsiString): AnsiString;
 function forceFilenameExt (const fn, ext: AnsiString): AnsiString;
 
+// rewrites slashes to '/'
+function fixSlashes (s: AnsiString): AnsiString;
+
+// replaces all the shitty characters with '_'
+// (everything except alphanumerics, '_', '.')
+function sanitizeFilename (s: AnsiString): AnsiString;
+
+function isAbsolutePath (const s: AnsiString): Boolean;
+function isRootPath (const s: AnsiString): Boolean;
+
 // strips out name from `fn`, leaving trailing slash
 function getFilenamePath (const fn: AnsiString): AnsiString;
 
@@ -87,6 +108,7 @@ function int64ToStrComma (i: Int64): AnsiString;
 
 function upcase1251 (ch: AnsiChar): AnsiChar; inline;
 function locase1251 (ch: AnsiChar): AnsiChar; inline;
+function IsValid1251 (ch: Word): Boolean;
 function IsPrintable1251 (ch: AnsiChar): Boolean;
 
 function toLowerCase1251 (const s: AnsiString): AnsiString;
@@ -98,12 +120,18 @@ function utf8Valid (const s: AnsiString): Boolean;
 
 function utf8to1251 (s: AnsiString): AnsiString;
 
-// `pathname` will be modified if path is valid
-// `lastIsDir` should be `true` if we are searching for directory
-// nobody cares about shitdoze, so i'll use the same code path for it
+// findFileCI takes case-insensitive path, traverses it, and rewrites it to
+// a case-sensetive one (using real on-disk names). return value means 'success'.
+// if some dir or file wasn't found, pathname is undefined (destroyed, but not
+// necessarily cleared).
+// last name assumed to be a file, not directory (unless `lastIsDir` flag is set).
 function findFileCI (var pathname: AnsiString; lastIsDir: Boolean=false): Boolean;
 
-// return fixed AnsiString or empty AnsiString
+// findDiskWad tries to find the wad file using common wad extensions
+// (see `wadExtensions` array).
+// returns real on-disk filename, or empty string.
+// original wad extension is used as a hint for the first try.
+// also, this automatically performs `findFileCI()`.
 function findDiskWad (fname: AnsiString): AnsiString;
 // slashes must be normalized!
 function isWadNamesEqu (wna, wnb: AnsiString): Boolean;
@@ -111,7 +139,7 @@ function isWadNamesEqu (wna, wnb: AnsiString): Boolean;
 // they throws
 function openDiskFileRO (pathname: AnsiString): TStream;
 function createDiskFile (pathname: AnsiString): TStream;
-// creates file if necessary
+// create file if necessary, but don't truncate the existing one
 function openDiskFileRW (pathname: AnsiString): TStream;
 
 // little endian
@@ -219,6 +247,9 @@ function digitInBase (ch: AnsiChar; base: Integer): Integer;
 // double quotes supports c-style escapes
 // function will select quote mode automatically
 function quoteStr (const s: AnsiString): AnsiString;
+// separate single-quote and double-quote escape functions
+function squoteStr (const s: AnsiString): AnsiString;
+function dquoteStr (const s: AnsiString): AnsiString;
 
 
 type
@@ -298,17 +329,6 @@ implementation
 uses
   xstreams;
 
-const wadExtensions: array [0..6] of AnsiString = (
-  '.dfz',
-  '.wad',
-  '.dfwad',
-  '.pk3',
-  '.pak',
-  '.zip',
-  '.dfzip'
-);
-
-
 // ////////////////////////////////////////////////////////////////////////// //
 procedure CopyMemory (Dest: Pointer; Src: Pointer; Len: LongWord); inline;
 begin
@@ -326,6 +346,61 @@ begin
 end;
 
 
+// ////////////////////////////////////////////////////////////////////////// //
+// rewrites slashes to '/'
+function fixSlashes (s: AnsiString): AnsiString;
+{$IFDEF WINDOWS}
+var
+  f: Integer;
+{$ENDIF}
+begin
+  result := s;
+  {$IFDEF WINDOWS}
+  for f := 1 to length(result) do if (result[f] = '\') then result[f] := '/';
+  {$ENDIF}
+end;
+
+// replaces all the shitty characters with '_'
+// (everything except alphanumerics, '_', '.')
+function sanitizeFilename (s: AnsiString): AnsiString;
+var
+  i: Integer;
+const
+  leaveChars: set of Char = [ '0'..'9', 'A'..'Z', 'a'..'z', '_', '.', #192..#255 ];
+  replaceWith: Char = '_';
+begin
+  result := s;
+  for i := 1 to length(result) do
+    if not (result[i] in leaveChars) then
+      result[i] := replaceWith;
+end;
+
+function isAbsolutePath (const s: AnsiString): Boolean;
+begin
+  result := false;
+  if (length(s) = 0) then exit;
+  {$IFDEF WINDOWS}
+  if (s[1] = '/') or (s[1] = '\') then begin result := true; exit; end;
+  if (length(s) > 2) and (s[2] = ':') and ((s[3] = '/') or (s[3] = '\')) then begin result := true; exit; end;
+  {$ELSE}
+  result := (s[1] = '/');
+  {$ENDIF}
+end;
+
+
+function isRootPath (const s: AnsiString): Boolean;
+begin
+  result := false;
+  if (length(s) = 0) then exit;
+  {$IFDEF WINDOWS}
+  if (s = '/') or (s = '\') then begin result := true; exit; end;
+  if (length(s) = 3) and (s[2] = ':') and ((s[3] = '/') or (s[3] = '\')) then begin result := true; exit; end;
+  {$ELSE}
+  result := (s = '/');
+  {$ENDIF}
+end;
+
+
 // ////////////////////////////////////////////////////////////////////////// //
 constructor TSimpleList.TEnumerator.Create (const aitems: TItemArr; acount: Integer);
 begin
@@ -458,7 +533,7 @@ var
 const
   cp1251: array[0..127] of Word = (
     $0402,$0403,$201A,$0453,$201E,$2026,$2020,$2021,$20AC,$2030,$0409,$2039,$040A,$040C,$040B,$040F,
-    $0452,$2018,$2019,$201C,$201D,$2022,$2013,$2014,$003F,$2122,$0459,$203A,$045A,$045C,$045B,$045F,
+    $0452,$2018,$2019,$201C,$201D,$2022,$2013,$2014,$FFFD,$2122,$0459,$203A,$045A,$045C,$045B,$045F,
     $00A0,$040E,$045E,$0408,$00A4,$0490,$00A6,$00A7,$0401,$00A9,$0404,$00AB,$00AC,$00AD,$00AE,$0407,
     $00B0,$00B1,$0406,$0456,$0491,$00B5,$00B6,$00B7,$0451,$2116,$0454,$00BB,$0458,$0405,$0455,$0457,
     $0410,$0411,$0412,$0413,$0414,$0415,$0416,$0417,$0418,$0419,$041A,$041B,$041C,$041D,$041E,$041F,
@@ -648,53 +723,52 @@ end;
 
 
 // ////////////////////////////////////////////////////////////////////////// //
-function quoteStr (const s: AnsiString): AnsiString;
-
-  function squote (const s: AnsiString): AnsiString;
-  var
-    f: Integer;
+function squoteStr (const s: AnsiString): AnsiString;
+var
+  f: Integer;
+begin
+  result := '''';
+  for f := 1 to Length(s) do
   begin
-    result := '''';
-    for f := 1 to Length(s) do
-    begin
-      if (s[f] = '''') then result += '''';
-      result += s[f];
-    end;
-    result += '''';
+    if (s[f] = '''') then result += '''';
+    result += s[f];
   end;
+  result += '''';
+end;
 
-  function dquote (const s: AnsiString): AnsiString;
-  var
-    f: Integer;
-    ch: AnsiChar;
+function dquoteStr (const s: AnsiString): AnsiString;
+var
+  f: Integer;
+  ch: AnsiChar;
+begin
+  result := '"';
+  for f := 1 to Length(s) do
   begin
-    result := '"';
-    for f := 1 to Length(s) do
+    ch := s[f];
+         if (ch = #0) then result += '\z'
+    else if (ch = #9) then result += '\t'
+    else if (ch = #10) then result += '\n'
+    else if (ch = #13) then result += '\r'
+    else if (ch = #27) then result += '\e'
+    else if (ch < ' ') or (ch = #127) then
     begin
-      ch := s[f];
-           if (ch = #0) then result += '\z'
-      else if (ch = #9) then result += '\t'
-      else if (ch = #10) then result += '\n'
-      else if (ch = #13) then result += '\r'
-      else if (ch = #27) then result += '\e'
-      else if (ch < ' ') or (ch = #127) then
-      begin
-        result += '\x';
-        result += LowerCase(IntToHex(Integer(ch), 2));
-      end
-      else if (ch = '"') or (ch = '\') then
-      begin
-        result += '\';
-        result += ch;
-      end
-      else
-      begin
-        result += ch;
-      end;
+      result += '\x';
+      result += LowerCase(IntToHex(Integer(ch), 2));
+    end
+    else if (ch = '"') or (ch = '\') then
+    begin
+      result += '\';
+      result += ch;
+    end
+    else
+    begin
+      result += ch;
     end;
-    result += '"';
   end;
+  result += '"';
+end;
 
+function quoteStr (const s: AnsiString): AnsiString;
 var
   needSingle: Boolean = false;
   f: Integer;
@@ -702,9 +776,9 @@ begin
   for f := 1 to Length(s) do
   begin
     if (s[f] = '''') then begin needSingle := true; continue; end;
-    if (s[f] < ' ') or (s[f] = #127) then begin result := dquote(s); exit; end;
+    if (s[f] < ' ') or (s[f] = #127) then begin result := dquoteStr(s); exit; end;
   end;
-  if needSingle then result := squote(s) else result := ''''+s+'''';
+  if needSingle then result := squoteStr(s) else result := ''''+s+'''';
 end;
 
 
@@ -878,7 +952,7 @@ end;
 function isWadPath (const fn: AnsiString): Boolean;
 var
   pos: Integer;
-  s: AnsiString;
+  s, wext: AnsiString;
 begin
   result := false;
   pos := 1;
@@ -890,10 +964,13 @@ begin
       if (pos-4 > 1) and (fn[pos-4] = '.') and ((fn[pos+1] = '\') or (fn[pos+1] = '/')) then
       begin
         s := Copy(fn, pos-4, 4);
-        if StrEquCI1251(s, '.wad') or StrEquCI1251(s, '.pk3') or StrEquCI1251(s, '.zip') or StrEquCI1251(s, '.dfz') then
+        for wext in wadExtensions do
         begin
-          result := true;
-          exit;
+          if strEquCI1251(s, wext) then
+          begin
+            result := true;
+            exit;
+          end;
         end;
       end;
     end;
@@ -962,9 +1039,14 @@ begin
   result := ch;
 end;
 
+function IsValid1251 (ch: Word): Boolean;
+begin
+  result := ((ch = Ord('?')) or (wc2shitmap[ch] <> '?')) and (wc2shitmap[ch] <> #$98)
+end;
+
 function IsPrintable1251 (ch: AnsiChar): Boolean;
 begin
-  result := (ch >= #32) and (ch <> #127)
+  result := (ch >= #32) and (ch <> #127) and (ch <> #$98)
 end;
 
 
@@ -1117,9 +1199,9 @@ end;
 
 
 // ////////////////////////////////////////////////////////////////////////// //
-// `pathname` will be modified if path is valid
-// `lastIsDir` should be `true` if we are searching for directory
-// nobody cares about shitdoze, so i'll use the same code path for it
+// findFileCI eats case-insensitive path, traverses it and rewrites it to a
+// case-sensetive. result value means success.
+// if file/dir not founded than pathname is in undefined state!
 function findFileCI (var pathname: AnsiString; lastIsDir: Boolean=false): Boolean;
 var
   sr: TSearchRec;