Changeset 27 for trunk/UAcronym.pas

Timestamp:

Jun 7, 2016, 3:56:46 PM (8 years ago)

Author:

chronos

Message:

• Modified: Now import format supports multiple more generic rules for parsing source content.
• Added: Show total and filtered number of acronyms in acronyms list.

File:

• trunk/UAcronym.pas (modified) (13 diffs)

trunk/UAcronym.pas

@@ -109 +109 @@
   end;
 
-  TImportPattern = record
+  TImportPatternFlag = (ipfNone, ipfNewItem, ipfSkip);
+  TImportVariable = (ivNone, ivAcronym, ivMeaning, ivDescription);
+
+  { TImportPattern }
+
+  TImportPattern = class
     StartString: string;
     EndString: string;
-    NewItem: Boolean;
+    Variable: TImportVariable;
+    Flag: TImportPatternFlag;
+    procedure Assign(Source: TImportPattern);
+    procedure SaveToNode(Node: TDOMNode);
+    procedure LoadFromNode(Node: TDOMNode);
+  end;
+
+  { TImportPatterns }
+
+  TImportPatterns = class(TObjectList)
+    procedure SaveToNode(Node: TDOMNode);
+    procedure LoadFromNode(Node: TDOMNode);
   end;
 
@@ -121 +137 @@
     Name: string;
     Block: TImportPattern;
-    Acronym: TImportPattern;
-    Meaning: TImportPattern;
-    Description: TImportPattern;
+    ItemPatterns: TImportPatterns;
     Formats: TImportFormats;
     procedure Assign(Source: TImportFormat);
     procedure SaveToNode(Node: TDOMNode);
     procedure LoadFromNode(Node: TDOMNode);
+    constructor Create;
+    destructor Destroy; override;
   end;
 
@@ -143 +159 @@
 
   TImportSource = class
+  private
+    procedure DoPassword(Sender: TObject; var RepeatRequest : Boolean);
+  public
     Name: string;
     URL: string;
@@ -148 +167 @@
     LastTime: TDateTime;
     Sources: TImportSources;
+    function DownloadHTTP(URL: string; Stream: TStream): Boolean;
     procedure Process;
     procedure Assign(Source: TImportSource);
@@ -189 +209 @@
 function AcronymComparer(Item1, Item2: Pointer): Integer;
 
+const
+  ImportVariableString: array [TImportVariable] of string = ('None', 'Acronym',
+    'Meaning', 'Description');
+  ImportPatternFlagString: array [TImportPatternFlag] of string = ('None', 'New item',
+    'Skip');
+
 
 implementation
@@ -195 +221 @@
   SWrongFileFormat = 'Wrong file format';
 
+
 function AcronymComparer(Item1, Item2: Pointer): Integer;
 begin
@@ -200 +227 @@
 end;
 
-function DownloadHTTP(URL: string; Stream: TStream): Boolean;
+{ TImportPatterns }
+
+procedure TImportPatterns.SaveToNode(Node: TDOMNode);
+var
+  I: Integer;
+  NewNode2: TDOMNode;
+begin
+  for I := 0 to Count - 1 do
+  with TImportPattern(Items[I]) do begin
+    NewNode2 := Node.OwnerDocument.CreateElement('Pattern');
+    Node.AppendChild(NewNode2);
+    SaveToNode(NewNode2);
+  end;
+end;
+
+procedure TImportPatterns.LoadFromNode(Node: TDOMNode);
+var
+  Node2: TDOMNode;
+  NewItem: TImportPattern;
+begin
+  Count := 0;
+  Node2 := Node.FirstChild;
+  while Assigned(Node2) and (Node2.NodeName = 'Pattern') do begin
+    NewItem := TImportPattern.Create;
+    NewItem.LoadFromNode(Node2);
+    Add(NewItem);
+    Node2 := Node2.NextSibling;
+  end;
+end;
+
+{ TImportPattern }
+
+procedure TImportPattern.Assign(Source: TImportPattern);
+begin
+  StartString := Source.StartString;
+  EndString := Source.EndString;
+  Variable := Source.Variable;
+  Flag := Source.Flag;
+end;
+
+procedure TImportPattern.SaveToNode(Node: TDOMNode);
+begin
+  WriteString(Node, 'StartString', StartString);
+  WriteString(Node, 'EndString', EndString);
+  WriteInteger(Node, 'Variable', Integer(Variable));
+  WriteInteger(Node, 'Flag', Integer(Flag));
+end;
+
+procedure TImportPattern.LoadFromNode(Node: TDOMNode);
+begin
+  StartString := ReadString(Node, 'StartString', '');
+  EndString := ReadString(Node, 'EndString', '');
+  Variable := TImportVariable(ReadInteger(Node, 'Variable', 0));
+  Flag := TImportPatternFlag(ReadInteger(Node, 'Flag', 0));
+end;
+
+procedure TImportSource.DoPassword(Sender: TObject; var RepeatRequest: Boolean);
+begin
+  if TFPHttpClient(Sender).Password = '' then begin
+    TFPHttpClient(Sender).UserName := 'test';
+    TFPHttpClient(Sender).Password := 'test';
+    RepeatRequest := True;
+  end else RepeatRequest := False;
+end;
+
+function TImportSource.DownloadHTTP(URL: string; Stream: TStream): Boolean;
 var
   HTTPClient: TFPHTTPClient;
 begin
   HTTPClient := TFPHttpClient.Create(nil);
+  HTTPClient.OnPassword := DoPassword;
   HTTPClient.Get(URL, Stream);
   HTTPClient.Free;
@@ -232 +325 @@
 
 procedure TImportFormat.Assign(Source: TImportFormat);
+var
+  I: Integer;
 begin
   Name := Source.Name;
   Block.StartString := Source.Block.StartString;
   Block.EndString := Source.Block.EndString;
-  Acronym.StartString := Source.Acronym.StartString;
-  Acronym.EndString := Source.Acronym.EndString;
-  Meaning.StartString := Source.Meaning.StartString;
-  Meaning.EndString := Source.Meaning.EndString;
-  Description.StartString := Source.Description.StartString;
-  Description.EndString := Source.Description.EndString;
+  while ItemPatterns.Count < Source.ItemPatterns.Count do
+    ItemPatterns.Add(TImportPattern.Create);
+  if ItemPatterns.Count > Source.ItemPatterns.Count then
+    ItemPatterns.Count := Source.ItemPatterns.Count;
+  for I := 0 to ItemPatterns.Count - 1 do begin
+    TImportPattern(ItemPatterns[I]).Assign(TImportPattern(Source.ItemPatterns[I]));
+  end;
 end;
 
 procedure TImportFormat.SaveToNode(Node: TDOMNode);
+var
+  NewNode: TDOMNode;
 begin
   WriteInteger(Node, 'Id', Id);
@@ -250 +348 @@
   WriteString(Node, 'BlockStartString', Block.StartString);
   WriteString(Node, 'BlockEndString', Block.EndString);
-  WriteString(Node, 'AcronymStartString', Acronym.StartString);
-  WriteString(Node, 'AcronymEndString', Acronym.EndString);
-  WriteString(Node, 'MeaningStartString', Meaning.StartString);
-  WriteString(Node, 'MeaningEndString', Meaning.EndString);
-  WriteString(Node, 'DescriptionStartString', Description.StartString);
-  WriteString(Node, 'DescriptionEndString', Description.EndString);
+
+  NewNode := Node.OwnerDocument.CreateElement('Patterns');
+  Node.AppendChild(NewNode);
+  ItemPatterns.SaveToNode(NewNode);
 end;
 
 procedure TImportFormat.LoadFromNode(Node: TDOMNode);
+var
+  NewNode: TDOMNode;
 begin
   Id := ReadInteger(Node, 'Id', 0);
@@ -264 +362 @@
   Block.StartString := ReadString(Node, 'BlockStartString', '');
   Block.EndString := ReadString(Node, 'BlockEndString', '');
-  Acronym.StartString := ReadString(Node, 'AcronymStartString', '');
-  Acronym.EndString := ReadString(Node, 'AcronymEndString', '');
-  Meaning.StartString := ReadString(Node, 'MeaningStartString', '');
-  Meaning.EndString := ReadString(Node, 'MeaningEndString', '');
-  Description.StartString := ReadString(Node, 'DescriptionStartString', '');
-  Description.EndString := ReadString(Node, 'DescriptionEndString', '');
+
+  NewNode := Node.FindNode('Patterns');
+  if Assigned(NewNode) then
+    ItemPatterns.LoadFromNode(NewNode);
+end;
+
+constructor TImportFormat.Create;
+begin
+  Block := TImportPattern.Create;
+  ItemPatterns := TImportPatterns.Create;
+end;
+
+destructor TImportFormat.Destroy;
+begin
+  Block.Free;
+  ItemPatterns.Free;
+  inherited Destroy;
 end;
 
@@ -390 +499 @@
   Stream: TMemoryStream;
   S: string;
+  SS: string;
   NewAcronym: TAcronymEntry;
   P: Integer;
-  NewName: string;
+  Q: Integer;
+  I: Integer;
+  T: string;
+  LastLength: Integer;
 begin
   Stream := TMemoryStream.Create;
@@ -401 +514 @@
       SetLength(S, Stream.Size);
       Stream.Read(S[1], Length(S));
+
+      // Find main block
       if Format.Block.StartString <> '' then begin
         P := Pos(Format.Block.StartString, S);
@@ -412 +527 @@
       end;
 
+      // Remove unneeded items
       repeat
-        P := Pos(Format.Acronym.StartString, S);
-        if P > 0 then begin
-          // Acronym
-          Delete(S, 1, P + Length(Format.Acronym.StartString) - 1);
-          P := Pos(Format.Acronym.EndString, S);
-          NewName := Trim(StripHTML(Copy(S, 1, P - 1)));
-          if NewName <> '' then NewAcronym.Name := NewName;
-          Delete(S, 1, P + Length(Format.Acronym.EndString) - 1);
-
-          // Meaning
-          if Length(Format.Meaning.StartString) > 0 then begin
-            P := Pos(Format.Meaning.StartString, S);
-            Delete(S, 1, P + Length(Format.Meaning.StartString) - 1);
+        LastLength := Length(S);
+        for I := 0 to Format.ItemPatterns.Count - 1 do
+        with TImportPattern(Format.ItemPatterns[I]) do
+        if Flag = ipfSkip then begin
+          P := Pos(StartString, S);
+          if P > 0 then begin
+            SS := Copy(S, P + Length(StartString), Length(S));
+            Q := Pos(EndString, SS);
+            if Q > 0 then begin
+              Delete(S, P, Q + Length(EndString) + Length(StartString) - 1);
+            end;
           end;
-          P := Pos(Format.Meaning.EndString, S);
-          NewAcronym.Meaning := Trim(StripHTML(Copy(S, 1, P - 1)));
-          Delete(S, 1, P + Length(Format.Meaning.EndString) - 1);
-          Sources.AcronymDb.AddAcronym(NewAcronym.Name, NewAcronym.Meaning);
-        end else Break;
-      until False;
+        end;
+      until Length(S) = LastLength;
+
+      // Find items
+      repeat
+        LastLength := Length(S);
+        for I := 0 to Format.ItemPatterns.Count - 1 do
+        with TImportPattern(Format.ItemPatterns[I]) do
+        if Flag <> ipfSkip then begin
+          if Length(StartString) > 0 then begin
+            P := Pos(StartString, S);
+            if P > 0 then Delete(S, 1, P + Length(StartString) - 1);
+          end;
+
+          if ((Length(StartString) > 0) and (P > 0)) or (Length(StartString) = 0) then begin
+            P := Pos(EndString, S);
+            T := StripHTML(Copy(S, 1, P - 1));
+            T := StringReplace(T, '&quot;', '"', [rfReplaceAll]);
+            T := StringReplace(T, '&trade;', 'TM', [rfReplaceAll]);
+            T := StringReplace(T, '&amp;', '&', [rfReplaceAll]);
+            T := Trim(T);
+            case Variable of
+              ivAcronym: NewAcronym.Name := T;
+              ivMeaning: NewAcronym.Meaning := T;
+            end;
+            Delete(S, 1, P + Length(EndString) - 1);
+
+            if (Flag = ipfNewItem) and (Trim(NewAcronym.Name) <> '') and
+              (Trim(NewAcronym.Meaning) <> '') then
+              Sources.AcronymDb.AddAcronym(NewAcronym.Name, NewAcronym.Meaning);
+          end;
+        end;
+      until Length(S) = LastLength;
     end;
   finally