Changeset 185

Timestamp:

Jul 11, 2018, 2:33:00 PM (6 years ago)

Author:

chronos

Message:

• Modified: More robust parsing of acronym meanings.

File:

• trunk/Forms/UFormCheck.pas (modified) (7 diffs)

trunk/Forms/UFormCheck.pas

@@ -53 +53 @@
     procedure FindInSummary;
     procedure FindInContent;
+    function ParseMeaning(Acronym, Text: string; StartIndex: Integer;
+      out Meaning: string; DashSeparator: Boolean = False): Boolean;
     function IsUppercaseAlpha(Text: string): Boolean;
+    function IsAlpha(Text: string): Boolean;
     procedure ReportDifferencies;
+    function WordContainsLetters(Text, Letters: string): Boolean;
   public
     procedure UpdateInterface;
@@ -222 +226 @@
 procedure TFormCheck.FindInContent;
 var
-  I: Integer;
   Text: string;
   StartIndex: Integer;
   EndIndex: Integer;
-  StartIndex2: Integer;
-  StartIndex3: Integer;
-  StartIndex4: Integer;
   Acronym: string;
   Meaning: string;
+  Meaning1: string;
+  Meaning2: string;
+  HasMeaning1: Boolean;
+  HasMeaning2: Boolean;
 begin
   AcronymDbContent.Acronyms.Clear;
@@ -237 +241 @@
   Text := MemoDocument.Lines.Text;
   Text := StringReplace(Text, #9, ' ', [rfReplaceAll]);
+  Text := StringReplace(Text, LineEnding, ' ', [rfReplaceAll]);
   StartIndex := 1;
   repeat
@@ -244 +249 @@
       if EndIndex <> 0 then begin
         Acronym := Trim(Copy(Text, StartIndex + 1, EndIndex - StartIndex - 1));
-        if (Acronym <> '') and IsUppercaseAlpha(Acronym) then begin
-          StartIndex2 := StartIndex;
-          for I := 0 to Length(Acronym) do begin
-            StartIndex3 := PosFromIndexReverse(' ', Text, StartIndex2);
-            StartIndex4 := PosFromIndexReverse('-', Text, StartIndex2);
-            if StartIndex4 > StartIndex3 then StartIndex3 := StartIndex4;
-            StartIndex4 := PosFromIndexReverse('/', Text, StartIndex2);
-            if StartIndex4 > StartIndex3 then StartIndex3 := StartIndex4;
-            StartIndex2 := StartIndex3;
-            if StartIndex2 = 0 then Break
-              else Dec(StartIndex2);
+        if (Length(Acronym) > 1) and IsUppercaseAlpha(Acronym) then begin
+          HasMeaning1 := ParseMeaning(Acronym, Text, StartIndex - 1, Meaning1);
+          if HasMeaning1 then Meaning := Meaning1;
+          HasMeaning2 := ParseMeaning(Acronym, Text, StartIndex - 1, Meaning2, True);
+          if HasMeaning2 then Meaning := Meaning2;
+          if HasMeaning1 and HasMeaning2 then begin
+            if Length(Meaning1) > Length(Meaning2) then Meaning := Meaning1
+              else Meaning := Meaning2;
           end;
-          Meaning := Trim(Copy(Text, StartIndex2 + 1, StartIndex - StartIndex2 - 1));
-          if Assigned(AcronymDbContent.SearchAcronym(Acronym, Meaning)) then
-            MemoReport.Lines.Add(Format(SDuplicateAcronymContent, [Acronym, Meaning]))
-            else AcronymDbContent.AddAcronym(Acronym, Meaning);
+          if HasMeaning1 or HasMeaning2 then begin
+            if Assigned(AcronymDbContent.SearchAcronym(Acronym, Meaning)) then
+              MemoReport.Lines.Add(Format(SDuplicateAcronymContent, [Acronym, Meaning]))
+              else AcronymDbContent.AddAcronym(Acronym, Meaning);
+          end;
         end;
       end;
@@ -265 +268 @@
     end;
   until StartIndex = 0;
+end;
+
+function TFormCheck.ParseMeaning(Acronym, Text: string; StartIndex: Integer;
+  out Meaning: string; DashSeparator: Boolean): Boolean;
+var
+  StartIndex2: Integer;
+  StartIndex3: Integer;
+  StartIndex4: Integer;
+  LetterIndex: Integer;
+  OneWord: string;
+  WordLetterIndex: Integer;
+  WordCount: Integer;
+  WordCountWrong: Integer;
+begin
+  Result := True;
+  Meaning := '';
+  StartIndex2 := StartIndex;
+  LetterIndex := Length(Acronym);
+  WordCount := 0;
+  WordCountWrong := 0;
+  while Length(Acronym) > 0 do begin
+    StartIndex3 := PosFromIndexReverse(' ', Text, StartIndex2);
+    if DashSeparator then begin
+      StartIndex4 := PosFromIndexReverse('-', Text, StartIndex2);
+      if StartIndex4 > StartIndex3 then StartIndex3 := StartIndex4;
+    end;
+
+    if StartIndex3 = 0 then Break;
+    OneWord := Copy(Text, StartIndex3 + 1, StartIndex2 - StartIndex3);
+    if OneWord = '$' then begin
+      // Avoid parsing Bash variables
+      Result := False;
+      Exit;
+    end;
+    if Trim(OneWord) = '' then begin
+      StartIndex2 := StartIndex3 - 1;
+      Continue;
+    end;
+    // Is first letter capital?
+    if (Length(OneWord) > 0) and IsAlpha(OneWord[1]) then begin
+      WordLetterIndex := PosFromIndexReverse(LowerCase(OneWord[1]), LowerCase(Copy(Acronym, 1, LetterIndex)), LetterIndex);
+      if WordLetterIndex > 0 then begin
+        // First letter was found in acronym
+        if WordLetterIndex <= LetterIndex then begin
+          if not WordContainsLetters(LowerCase(OneWord), LowerCase(Copy(Acronym, WordLetterIndex, LetterIndex - WordLetterIndex + 1))) then begin
+            Result := False;
+            Exit;
+          end;
+          LetterIndex := WordLetterIndex - 1;
+        end else begin
+          Dec(LetterIndex);
+        end;
+        WordCountWrong := 0;
+      end else begin
+        Inc(WordCountWrong);
+        if WordCountWrong > 1 then begin
+          Result := False;
+          Exit;
+        end;
+      end;
+    end else begin
+      Inc(WordCountWrong);
+      if WordCountWrong > 1 then begin
+        Result := False;
+        Exit;
+      end;
+    end;
+    StartIndex2 := StartIndex3 - 1;
+    if LetterIndex < 1 then Break;
+    Inc(WordCount);
+    if WordCount > 2 * Length(Acronym) then begin
+      // False acronym in braces with too much words
+      Result := False;
+      Exit;
+    end;
+  end;
+  Meaning := Trim(Copy(Text, StartIndex2 + 1, StartIndex - StartIndex2 - 1));
 end;
 
@@ -282 +362 @@
 end;
 
+function TFormCheck.IsAlpha(Text: string): Boolean;
+var
+  I: Integer;
+begin
+  I := 1;
+  Result := True;
+  while (I <= Length(Text)) do begin
+    if not (Text[I] in ['A'..'Z']) and not (Text[I] in ['a'..'z']) then begin
+      Result := False;
+      Break;
+    end;
+    Inc(I);
+  end;
+end;
+
 procedure TFormCheck.ReportDifferencies;
 var
@@ -304 +399 @@
       if not Assigned(AcronymDbContent.SearchAcronym(Acronym.Name, Meaning.Name, [sfCaseInsensitive])) then
         MemoReport.Lines.Add(Format(SMissingAcronymSummary, [Acronym.Name, Meaning.Name]));
+    end;
+  end;
+end;
+
+function TFormCheck.WordContainsLetters(Text, Letters: string): Boolean;
+var
+  I: Integer;
+  LetterIndex: Integer;
+begin
+  Result := True;
+  for I := 1 to Length(Letters) do begin
+    LetterIndex := Pos(Letters[I], Text);
+    if LetterIndex > 0 then begin
+      Text := Copy(Text, LetterIndex + 1, Length(Text));
+    end else begin
+      Result := False;
+      Break;
     end;
   end;