From 0f6f5b9c53c6325a39e7a87057551c3fb1da97e9 Mon Sep 17 00:00:00 2001
From: mattias <nc-gaertnma@netcologne.de>
Date: Sun, 24 Apr 2011 08:36:36 +0000
Subject: [PATCH] IDE: TSimpleHTMLControl.HTMLToCaption: skip header

git-svn-id: trunk@30445 -
---
 ide/helpmanager.pas   | 59 +++++++++++++++++++++++++++++++++----------
 lcl/lconvencoding.pas |  3 ++-
 2 files changed, 47 insertions(+), 15 deletions(-)

diff --git a/ide/helpmanager.pas b/ide/helpmanager.pas
index 3cc278682d..dbf2ab530a 100644
--- a/ide/helpmanager.pas
+++ b/ide/helpmanager.pas
@@ -35,7 +35,7 @@ interface
 uses
   // FCL+LCL
   Classes, SysUtils, AVL_Tree, LCLProc, LCLIntf, LCLType, Forms, Controls, Buttons,
-  StdCtrls, Dialogs, ExtCtrls, FileProcs, Graphics, ButtonPanel,
+  StdCtrls, Dialogs, ExtCtrls, FileProcs, Graphics, ButtonPanel, LConvEncoding,
   // CodeTools
   BasicCodeTools, CodeToolManager, CodeAtom, CodeCache, CustomCodeTool, CodeTree,
   PascalParserTool, FindDeclarationTool,
@@ -265,19 +265,29 @@ function TSimpleHTMLControl.HTMLToCaption(const s: string; MaxLines: integer
 var
   p: Integer;
   EndPos: Integer;
-  CurTag: String;
   NewTag: String;
   Line: Integer;
   sp: LongInt;
+  InHeader: Boolean;
+  CurTagName: String;
 begin
   Result:=s;
   //debugln(['TSimpleHTMLControl.HTMLToCaption HTML="',Result,'"']);
   Line:=1;
   p:=1;
+  // remove UTF8 BOM
+  if copy(Result,1,3)=UTF8BOM then
+    Result:=copy(s,4,length(Result));
+  InHeader:=false; // it could be a snippet
   while p<=length(Result) do begin
     if Result[p]='<' then begin
       // removes html tags
       EndPos:=p+1;
+      if (EndPos<=length(Result)) and (Result[EndPos]='/') then inc(EndPos);
+      while (EndPos<=length(Result))
+      and (not (Result[EndPos] in [' ','>','"','/',#9,#10,#13])) do
+        inc(EndPos);
+      CurTagName:=UpperCase(copy(Result,p+1,EndPos-p-1));
       while (EndPos<=length(Result)) do begin
         if Result[EndPos]='"' then begin
           // skip " tag
@@ -291,26 +301,47 @@ begin
         end;
         inc(EndPos);
       end;
-      CurTag:=copy(Result,p,EndPos-p);
 
-      if ((SysUtils.CompareText(CurTag,'<P>')=0)
-        or (SysUtils.CompareText(CurTag,'</P>')=0))
-      then begin
+      if CurTagName='HTML' then
+      begin
+        // it's a whole page
+        InHeader:=true;
+      end;
+      if CurTagName='BODY' then
+      begin
+        // start of body => ignore header
+        InHeader:=false;
+        Result:=copy(Result,EndPos,length(Result));
+        p:=1;
+        EndPos:=1;
+        Line:=1;
+      end;
+      if CurTagName='/BODY' then
+      begin
+        // end of body
+        Result:=copy(Result,1,p-1);
+        break;
+      end;
+
+      if (CurTagName='P') or (CurTagName='/P') then begin
         // add a line break if there is not already one
         sp:=p;
         while (sp>1) and (Result[sp-1] in [' ',#9]) do dec(sp);
         if (sp>1) and (not (Result[sp-1] in [#10,#13])) then
-          CurTag:='<BR>';
+          CurTagName:='BR';
+      end;
+      if (CurTagName='DIV') or (CurTagName='/DIV')
+      then begin
+        // add a line break if not in first line
+        if Line>1 then
+          CurTagName:='BR';
       end;
 
-      if (p>1)
-      and ((SysUtils.CompareText(CurTag,'<BR>')=0)
-        or (SysUtils.CompareText(CurTag,'<DIV>')=0)
-        or (SysUtils.CompareText(copy(CurTag,1,5),'<DIV ')=0)
-        or (SysUtils.CompareText(CurTag,'</DIV>')=0))
-      then begin
+      if CurTagName='BR' then
+      begin
         NewTag:=LineEnding;
-        inc(Line);
+        if not InHeader then
+          inc(Line);
         if Line>MaxLines then begin
           Result:=copy(Result,1,p)+LineEnding+'...';
           break;
diff --git a/lcl/lconvencoding.pas b/lcl/lconvencoding.pas
index e77cad4b87..91bc4581c5 100644
--- a/lcl/lconvencoding.pas
+++ b/lcl/lconvencoding.pas
@@ -31,6 +31,7 @@ const
   EncodingUTF8BOM = 'utf8bom'; // UTF-8 with byte order mark
   EncodingUCS2LE = 'ucs2le'; // UCS 2 byte little endian
   EncodingUCS2BE = 'ucs2be'; // UCS 2 byte big endian
+  UTF8BOM = #$EF#$BB#$BF;
 
 function GuessEncoding(const s: string): string;
 
@@ -5842,7 +5843,7 @@ begin
   end;
   
   // try UTF-8 BOM (Byte Order Mark)
-  if CompareI(@s[1],#$EF#$BB#$BF,3) then begin
+  if CompareI(@s[1],UTF8BOM,3) then begin
     Result:=EncodingUTF8BOM;
     exit;
   end;