LazUtils: Use TUnicodeCharacterEnumerator for the "for .. in" iterator syntax. Move demo to examples dir and add a GUI demo.

git-svn-id: trunk@56260 -
2025-12-16 03:00:30 +01:00 · 2017-11-03 16:13:46 +00:00 · 2017-11-03 16:13:46 +00:00 · 18570daf7b
commit 18570daf7b
parent 7772e8646b
8 changed files with 300 additions and 32 deletions
--- a/.gitattributes
+++ b/.gitattributes
@ -3126,6 +3126,12 @@ components/lazutils/dynamicarray.pas svneol=native#text/pascal
 components/lazutils/dynhasharray.pp svneol=native#text/pascal
 components/lazutils/dynqueue.pas svneol=native#text/pascal
 components/lazutils/easylazfreetype.pas svneol=native#text/pascal
+components/lazutils/examples/LazUnicode/LazUnicodeConsole.lpi svneol=native#text/plain
+components/lazutils/examples/LazUnicode/LazUnicodeConsole.lpr svneol=native#text/pascal
+components/lazutils/examples/LazUnicode/LazUnicodeGUI.lpi svneol=native#text/plain
+components/lazutils/examples/LazUnicode/LazUnicodeGUI.lpr svneol=native#text/pascal
+components/lazutils/examples/LazUnicode/lazunicodeunit.dfm svneol=native#text/plain
+components/lazutils/examples/LazUnicode/lazunicodeunit.pas svneol=native#text/pascal
 components/lazutils/examples/LookupStringList/ReadMe.txt svneol=native#text/plain
 components/lazutils/examples/LookupStringList/TDedupeDemo.lpi svneol=native#text/plain
 components/lazutils/examples/LookupStringList/TDedupeDemo.lpr svneol=native#text/pascal
@ -3208,8 +3214,6 @@ components/lazutils/maps.pp svneol=native#text/pascal
 components/lazutils/masks.pas svneol=native#text/pascal
 components/lazutils/paswstring.pas svneol=native#text/pascal
 components/lazutils/stringhashlist.pas svneol=native#text/pascal
-components/lazutils/test/LazUnicodeTest.lpi svneol=native#text/plain
-components/lazutils/test/LazUnicodeTest.lpr svneol=native#text/pascal
 components/lazutils/test/TestLazStorageMem.lpi svneol=native#text/plain
 components/lazutils/test/TestLazStorageMem.lpr svneol=native#text/plain
 components/lazutils/test/testlazstoragememcase1.pas svneol=native#text/plain
--- a/components/lazutils/examples/LazUnicode/LazUnicodeConsole.lpi
+++ b/components/lazutils/examples/LazUnicode/LazUnicodeConsole.lpi
@ -9,7 +9,7 @@
      </Flags>
      <SessionStorage Value="InProjectDir"/>
      <MainUnit Value="0"/>
-      <Title Value="LazUnicodeTest"/>
+      <Title Value="LazUnicodeConsole"/>
      <UseAppBundle Value="False"/>
      <ResourceType Value="res"/>
    </General>
@ -19,7 +19,7 @@
        <CompilerOptions>
          <Version Value="11"/>
          <Target>
-            <Filename Value="LazUnicodeTest"/>
+            <Filename Value="LazUnicodeConsole"/>
          </Target>
          <SearchPaths>
            <IncludeFiles Value="$(ProjOutDir)"/>
@ -51,7 +51,7 @@
    </RequiredPackages>
    <Units Count="1">
      <Unit0>
-        <Filename Value="LazUnicodeTest.lpr"/>
+        <Filename Value="LazUnicodeConsole.lpr"/>
        <IsPartOfProject Value="True"/>
      </Unit0>
    </Units>
@ -59,7 +59,7 @@
  <CompilerOptions>
    <Version Value="11"/>
    <Target>
-      <Filename Value="LazUnicodeTest"/>
+      <Filename Value="LazUnicodeConsole"/>
    </Target>
    <SearchPaths>
      <IncludeFiles Value="$(ProjOutDir)"/>
--- a/components/lazutils/examples/LazUnicode/LazUnicodeConsole.lpr
+++ b/components/lazutils/examples/LazUnicode/LazUnicodeConsole.lpr
@ -4,7 +4,7 @@
  Works also in Delphi where String=UnicodeString. Just rename this file as *.dpr.
   For Delphi you must copy units LazUnicode and LazUTF16, both part of LazUtils package.
 }
-program LazUnicodeTest;
+program LazUnicodeConsole;

 {$IFDEF FPC}
 {$mode objfpc}{$H+}
@ -68,28 +68,12 @@ begin

  // Use CodePoint enumerator explicitly
  writeln('');
-  writeln('*** Using CodePoint iterator explicitly: ***');
+  writeln('*** Using CodePoint enumerator explicitly: ***');
  cpIter := TCodePointEnumerator.Create(s);
  while cpIter.MoveNext do
    writeln('ch=', cpIter.Current, '  has ', cpIter.CurrentCodeUnitCount, ' codeunits.');
  cpIter.Free;

-  {$IFDEF FPC}
-  // Use for-in loop for CodePoints.
-  writeln('');
-  writeln('*** Using for-in loop for CodePoints : ***');
-  for ch in s do
-    writeln('ch=',ch);
-  {$ENDIF}
-
-  // Use for-in loop for codeunits using a Char variable still works.
-  {    Uncomment to test.
-  writeln('');
-  writeln('*** Using for-in loop for codeunits: ***');
-  for CodeUnit in s do
-    writeln('CodeUnit=',CodeUnit);        // The output makes no sense obviously.
-  }
-
  s_UTF8 := Combining;
  s := s_UTF8;                              // Converts encoding when needed.
  writeln('');
@ -97,11 +81,27 @@ begin

  // Use UnicodeCharacter enumerator explicitly
  writeln('');
-  writeln('*** Using UnicodeCharacter iterator explicitly: ***');
+  writeln('*** Using UnicodeCharacter enumerator explicitly: ***');
  ucIter := TUnicodeCharacterEnumerator.Create(s);
  while ucIter.MoveNext do
    writeln('ch=', ucIter.Current, '  has ', ucIter.CurrentCodePointCount, ' codepoints and ', ucIter.CurrentCodeUnitCount, ' codeunits.');
  ucIter.Free;

+  {$IFDEF FPC}
+  // Use for-in loop for Unicode Characters.
+  writeln('');
+  writeln('*** Using for-in loop for Unicode Characters : ***');
+  for ch in s do
+    writeln('ch=',ch);
+  {$ENDIF}
+
+  // for-in loop for codeunits using a Char variable still works.
+  {    Uncomment to test.
+  writeln('');
+  writeln('*** Using for-in loop for codeunits: ***');
+  for CodeUnit in s do
+    writeln('CodeUnit=',CodeUnit);        // The output makes no sense obviously.
+  }
+
 end.

--- a/components/lazutils/examples/LazUnicode/LazUnicodeGUI.lpi
+++ b/components/lazutils/examples/LazUnicode/LazUnicodeGUI.lpi
@ -0,0 +1,82 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<CONFIG>
+  <ProjectOptions>
+    <Version Value="10"/>
+    <General>
+      <SessionStorage Value="InProjectDir"/>
+      <MainUnit Value="0"/>
+      <Title Value="LazUnicodeGUI"/>
+      <UseAppBundle Value="False"/>
+      <ResourceType Value="res"/>
+      <Icon Value="0"/>
+    </General>
+    <BuildModes Count="1">
+      <Item1 Name="Default" Default="True"/>
+    </BuildModes>
+    <PublishOptions>
+      <Version Value="2"/>
+    </PublishOptions>
+    <RunParams>
+      <local>
+        <FormatVersion Value="1"/>
+      </local>
+    </RunParams>
+    <RequiredPackages Count="1">
+      <Item1>
+        <PackageName Value="LCL"/>
+      </Item1>
+    </RequiredPackages>
+    <Units Count="2">
+      <Unit0>
+        <Filename Value="LazUnicodeGUI.lpr"/>
+        <IsPartOfProject Value="True"/>
+      </Unit0>
+      <Unit1>
+        <Filename Value="lazunicodeunit.pas"/>
+        <IsPartOfProject Value="True"/>
+        <ComponentName Value="Form1"/>
+        <HasResources Value="True"/>
+        <ResourceBaseClass Value="Form"/>
+        <UnitName Value="LazUnicodeUnit"/>
+      </Unit1>
+    </Units>
+  </ProjectOptions>
+  <CompilerOptions>
+    <Version Value="11"/>
+    <Target>
+      <Filename Value="LazUnicodeGUI"/>
+    </Target>
+    <SearchPaths>
+      <IncludeFiles Value="$(ProjOutDir)"/>
+      <UnitOutputDirectory Value="lib/$(TargetCPU)-$(TargetOS)"/>
+    </SearchPaths>
+    <Parsing>
+      <SyntaxOptions>
+        <SyntaxMode Value="Delphi"/>
+      </SyntaxOptions>
+    </Parsing>
+    <Linking>
+      <Options>
+        <Win32>
+          <GraphicApplication Value="True"/>
+        </Win32>
+      </Options>
+    </Linking>
+    <Other>
+      <CustomOptions Value="-dBorland -dVer150 -dDelphi7 -dCompiler6_Up -dPUREPASCAL"/>
+    </Other>
+  </CompilerOptions>
+  <Debugging>
+    <Exceptions Count="3">
+      <Item1>
+        <Name Value="EAbort"/>
+      </Item1>
+      <Item2>
+        <Name Value="ECodetoolError"/>
+      </Item2>
+      <Item3>
+        <Name Value="EFOpenError"/>
+      </Item3>
+    </Exceptions>
+  </Debugging>
+</CONFIG>
--- a/components/lazutils/examples/LazUnicode/LazUnicodeGUI.lpr
+++ b/components/lazutils/examples/LazUnicode/LazUnicodeGUI.lpr
@ -0,0 +1,21 @@
+program LazUnicodeGUI;
+
+{$mode objfpc}{$H+}
+
+uses
+  {$IFDEF UNIX}{$IFDEF UseCThreads}
+  cthreads,
+  {$ENDIF}{$ENDIF}
+  Interfaces, // this includes the LCL widgetset
+  Forms, LazUnicodeUnit
+  { you can add units after this };
+
+{$R *.res}
+
+begin
+  RequireDerivedFormResource:=True;
+  Application.Initialize;
+  Application.CreateForm(TForm1, Form1);
+  Application.Run;
+end.
+
--- a/components/lazutils/examples/LazUnicode/lazunicodeunit.dfm
+++ b/components/lazutils/examples/LazUnicode/lazunicodeunit.dfm
@ -0,0 +1,30 @@
+object Form1: TForm1
+  Left = 153
+  Height = 528
+  Top = 77
+  Width = 608
+  Caption = 'LazUnicode Demo'
+  ClientHeight = 528
+  ClientWidth = 608
+  LCLVersion = '1.9.0.0'
+  object Button1: TButton
+    Left = 0
+    Height = 29
+    Top = 0
+    Width = 608
+    Align = alTop
+    AutoSize = True
+    Caption = 'Test LazUnicode functions and enumerators'
+    OnClick = Button1Click
+    TabOrder = 0
+  end
+  object Memo1: TMemo
+    Left = 0
+    Height = 499
+    Top = 29
+    Width = 608
+    Align = alClient
+    ScrollBars = ssAutoBoth
+    TabOrder = 1
+  end
+end
--- a/components/lazutils/examples/LazUnicode/lazunicodeunit.pas
+++ b/components/lazutils/examples/LazUnicode/lazunicodeunit.pas
@ -0,0 +1,130 @@
+unit LazUnicodeUnit;
+
+{$IFDEF FPC}
+ {$mode objfpc}{$H+}
+{$ENDIF}
+
+// For testing the UTF16 version.
+{$IF DEFINED(FPC) and DEFINED(UseUTF16)}
+{$ModeSwitch UnicodeStrings}   // Sets also FPC_UNICODESTRINGS.
+{$ENDIF}
+
+{$IF DEFINED(FPC_UNICODESTRINGS) or not DEFINED(FPC)}
+ {$DEFINE ReallyUseUTF16}       // FPC with UTF-16 or Delphi
+{$ENDIF}
+
+interface
+
+uses
+  Classes, SysUtils, FileUtil, Forms, Controls, Graphics, Dialogs, StdCtrls,
+  LazUnicode;
+
+type
+
+  { TForm1 }
+
+  TForm1 = class(TForm)
+    Button1: TButton;
+    Memo1: TMemo;
+    procedure Button1Click(Sender: TObject);
+  private
+
+  public
+
+  end;
+
+var
+  Form1: TForm1;
+
+const
+  Eyes = '👀';
+  Thai = 'ฃ';   // No idea what it means.
+  WineGlass = '🍷';
+  Heart = '💓';  // or '♡';
+  // Accents in combining codepoints. Last one has 2 consecutive combining marks.
+  Combining = 'ÓÓỐỐỚỚÒÒỒỒỎỎỔỔỞỞỌỌBあC'#$CC#$81#$CC#$B2;
+  //ArEnStr1 = 'مAرBحCبDاE';
+
+implementation
+
+{$R *.dfm}
+
+{ TForm1 }
+
+procedure TForm1.Button1Click(Sender: TObject);
+var
+  s_UTF8: {$IFDEF FPC}AnsiString{$ELSE}String{$ENDIF};
+  s, ch: String;
+  //CodeUnit: Char;
+  i: NativeInt;
+  cpIter: TCodePointEnumerator;
+  ucIter: TUnicodeCharacterEnumerator;
+begin
+  Memo1.Lines.Clear;
+  s_UTF8 := 'Pöö ' + Eyes + Thai + '. Have ' + WineGlass + ' for FPC' + Heart + 'Lazarus';
+  s := s_UTF8;                              // Converts encoding when needed.
+  {$IFDEF ReallyUseUTF16}
+  Memo1.Lines.Add('Using UnicodeString + UTF-16');
+  {$ELSE}
+  Memo1.Lines.Add('Using AnsiString + UTF-8');
+  {$ENDIF}
+  Memo1.Lines.Add(Format('Sizeof(Char) = %d.', [Sizeof(Char)]));
+  Memo1.Lines.Add('Testing with string: ' + s);
+
+  Memo1.Lines.Add('');
+  ch := CodePointCopy(s, 14, 1);       // Should return the wine glass.
+  Memo1.Lines.Add('Testing CodePointCopy. SubString = "' + ch + '"');
+
+  i := CodePointLength(s);             // Should return 30.
+  Memo1.Lines.Add(Format('Testing CodePointLength. Result = %d.', [i]));
+
+  // Constant must be assigned to AnsiString when using the UTF-8 system.
+  s_UTF8 := WineGlass;
+  i := CodePointPos(s_UTF8, s);        // Should return 14.
+  Memo1.Lines.Add(Format('Testing CodePointPos. Result = %d.', [i]));
+  s_UTF8 := '☐';
+  i := CodePointPos(s_UTF8, s);        // Should return 0.
+  Memo1.Lines.Add(Format('Testing CodePointPos for non-existent char. Result = %d.', [i]));
+
+  // Use CodePoint enumerator explicitly
+  Memo1.Lines.Add('');
+  Memo1.Lines.Add('*** Using CodePoint enumerator explicitly: ***');
+  cpIter := TCodePointEnumerator.Create(s);
+  while cpIter.MoveNext do
+    Memo1.Lines.Add(Format('ch=%s has %d codeunits.',
+                           [cpIter.Current, cpIter.CurrentCodeUnitCount]));
+  cpIter.Free;
+
+  s_UTF8 := Combining;
+  s := s_UTF8;                              // Converts encoding when needed.
+  Memo1.Lines.Add('');
+  Memo1.Lines.Add('Testing with string: ' + s);
+
+  // Use UnicodeCharacter enumerator explicitly
+  Memo1.Lines.Add('');
+  Memo1.Lines.Add('*** Using UnicodeCharacter enumerator explicitly: ***');
+  ucIter := TUnicodeCharacterEnumerator.Create(s);
+  while ucIter.MoveNext do
+    Memo1.Lines.Add(Format('ch=%s has %d codepoints and %d codeunits.',
+           [ucIter.Current, ucIter.CurrentCodePointCount, ucIter.CurrentCodeUnitCount]));
+  ucIter.Free;
+
+  {$IFDEF FPC}
+  // Use for-in loop for Unicode Characters.
+  Memo1.Lines.Add('');
+  Memo1.Lines.Add('*** Using for-in loop for Unicode Characters : ***');
+  for ch in s do
+    Memo1.Lines.Add('ch=' + ch);
+  {$ENDIF}
+
+  // for-in loop for codeunits using a Char variable still works.
+  {    Uncomment to test.
+  Memo1.Lines.Add('');
+  Memo1.Lines.Add('*** Using for-in loop for codeunits: ***');
+  for CodeUnit in s do
+    Memo1.Lines.Add('CodeUnit=',CodeUnit);        // The output makes no sense obviously.
+  }
+end;
+
+end.
+
--- a/components/lazutils/lazunicode.pas
+++ b/components/lazutils/lazunicode.pas
@ -93,12 +93,13 @@ type
  end;

  {$IFDEF FPC}
-  // Enumerator for CodePoints is used for for-in loop now.
-  operator Enumerator(A: String): TCodePointEnumerator;
+  // Enumerator for CodePoints could be used for the for-in loop.
+  //operator Enumerator(A: String): TCodePointEnumerator;

-  // This enumerator combines diacritical marks. Not enabled by default because
-  //  there are many more rules for combining codepoints.
-  //operator Enumerator(A: String): TUnicodeCharacterEnumerator;
+  // This enumerator combines diacritical marks.
+  // It is used by default although there are more rules for combining codepoints.
+  // Diacritical marks cover rules for most western languages.
+  operator Enumerator(A: String): TUnicodeCharacterEnumerator;
  {$ENDIF}

 implementation
@ -311,9 +312,9 @@ end;
 // Enumerator
 //---
 {$IFDEF FPC}
-operator Enumerator(A: String): TCodePointEnumerator;
+operator Enumerator(A: String): TUnicodeCharacterEnumerator;
 begin
-  Result := TCodePointEnumerator.Create(A);
+  Result := TUnicodeCharacterEnumerator.Create(A);
 end;
 {$ENDIF}