Changeset 192 for trunk/Forms/UFormCheck.pas

Timestamp:

Jul 17, 2018, 2:39:01 PM (6 years ago)

Author:

chronos

Message:

• Added: Analysis of acronyms usage (without parentheses) in document content.
• Modified: Better reporting of acronyms with multiple meanings and with different meaning between content and summary acronyms.

File:

• trunk/Forms/UFormCheck.pas (modified) (9 diffs)

trunk/Forms/UFormCheck.pas

@@ -51 +51 @@
     procedure FindInSummary;
     procedure FindInContent;
+    function AllowedSideChar(Before, After: Char): Boolean;
     function ParseMeaning(Acronym, Text: string; StartIndex: Integer;
       out Meaning: string; DashSeparator: Boolean = False): Boolean;
     function IsUppercaseAlpha(Text: string): Boolean;
+    function IsLowercaseAlpha(Text: string): Boolean;
     function IsAlpha(Text: string): Boolean;
     function IsAcronym(Text: string): Boolean;
+    function IsDigit(Text: Char): Boolean;
     procedure ReportDifferencies;
     function WordContainsLetters(Text, Letters: string): Boolean;
@@ -82 +85 @@
   SMissingAcronymContent = 'Warning: Document body acronym %s with meaning "%s" missing from acronym summary.';
   SMissingAcronymSummary = 'Warning: Summary acronym %s with meaning "%s" missing from document body.';
+  SAcronymContentMultipleMeanings = 'Warning: Content acronym %s has multiple meanings: %s.';
+  SAcronymSummaryMultipleMeanings = 'Warning: Summary acronym %s has multiple meanings: %s.';
+  SAcronymWithDifferentMeaning = 'Warning: Acronym %s has different meaning for content "%s" and for summary "%s".';
+  SAcronymWithDifferentMeaningCount = 'Warning: Acronym %s has different meaning count for content (%d) and for summary (%d).';
   SPluralAcronym = 'Note: Acronym %s is defined as plural in document body.';
+  SPluralAcronymUsed = 'Note: Acronym %s is used as plural in document body.';
   SSummaryAcronyms = 'Summary acronyms';
   SContentAcronyms = 'Content acronyms';
+  SAcronymUsedBeforeDefined = 'Acronym %s used before it was defined.';
+
+const
+  MinAcronymLength = 2;
 
 { TFormCheck }
@@ -215 +227 @@
 var
   Text: string;
-  StartIndex: Integer;
-  EndIndex: Integer;
+  Index: Integer;
+  State: (stNone, stAcronymUsage, stAcronymDefinition);
   Acronym: string;
+  AcronymCharBefore: Char;
+  AcronymCharAfter: Char;
+  Lines: TStringList;
+  HasUpperCase: Boolean;
+  HasLowerCase: Boolean;
+  I: Integer;
+  J: Integer;
+  Line: string;
   Meaning: string;
   Meaning1: string;
@@ -224 +244 @@
   HasMeaning2: Boolean;
   Plural: Boolean;
+  StartIndex: Integer;
+  Acro: TAcronym;
 begin
   AcronymDbContent.Acronyms.Clear;
 
+  // Make lowercase lines where all alpha characters are in uppercase
+  // These are usually first level chapter titles or first page text
+  Lines := TStringList.Create;
+  Lines.Assign(MemoDocument.Lines);
+  for I := 0 to Lines.Count - 1 do begin
+    HasLowerCase := False;
+    HasUpperCase := False;
+    Line := Lines[I];
+    for J := 1 to Length(Line) do begin
+      if IsUppercaseAlpha(Line[J]) then HasUpperCase := True;
+      if IsLowercaseAlpha(Line[J]) then HasLowerCase := True;
+    end;
+    if HasUpperCase and not HasLowerCase then
+      Lines[I] := LowerCase(Lines[I]);
+  end;
+
   // Find acronyms usage in text
-  Text := MemoDocument.Lines.Text;
+  Text := Lines.Text;
   Text := StringReplace(Text, #9, ' ', [rfReplaceAll]);
   Text := StringReplace(Text, LineEnding, ' ', [rfReplaceAll]);
-  StartIndex := 1;
+  Index := 1;
+  AcronymCharBefore := ' ';
+  AcronymCharAfter := ' ';
+  State := stNone;
+  StartIndex := 0;
   repeat
-    StartIndex := PosFromIndex('(', Text, StartIndex);
-    if StartIndex <> 0 then begin
-      EndIndex := PosFromIndex(')', Text, StartIndex);
-      if EndIndex <> 0 then begin
-        Acronym := Trim(Copy(Text, StartIndex + 1, EndIndex - StartIndex - 1));
+    if State = stAcronymUsage then begin
+      if not IsUppercaseAlpha(Text[Index]) then begin
+        if Text[Index] = 's' then begin
+          Acronym := Acronym + Text[Index];
+          Inc(Index);
+        end;
+        if (Index + 1) < Length(Text) then AcronymCharAfter := Text[Index + 1]
+          else AcronymCharAfter := ' ';
+        State := stNone;
+
+        // Allow plural acronyms with ending 's' character
+        if (Length(Acronym) >= 1) and (Acronym[Length(Acronym)] = 's') then begin
+          Acronym := Copy(Acronym, 1, Length(Acronym) - 1);
+          Plural := True;
+        end else Plural := False;
+
+        // Acronyms should not contain numbers
+        if (Length(Acronym) >= MinAcronymLength) and
+          AllowedSideChar(AcronymCharBefore, AcronymCharAfter) then begin
+            // If plural acronym then try to remove ending 's' character from the meaning
+            if Plural then MemoReport.Lines.Add(Format(SPluralAcronymUsed, [Acronym]));
+
+            Acro := AcronymDbContent.Acronyms.SearchByName(Acronym);
+            if not Assigned(Acro) then begin
+              MemoReport.Lines.Add(Format(SAcronymUsedBeforeDefined, [Acronym]));
+              AcronymDbContent.AddAcronym(Acronym, '');
+            end;
+        end;
+      end else Acronym := Acronym + Text[Index];
+    end else
+    if State = stAcronymDefinition then begin
+      if Text[Index] = ')' then begin
         // Allow plural acronyms with ending 's' character
         if (Length(Acronym) >= 1) and (Acronym[Length(Acronym)] = 's') then begin
@@ -262 +331 @@
               else AcronymDbContent.AddAcronym(Acronym, Meaning);
           end;
-        end;
+        end else
+          // No acronym inside parenthesis, continue with parsing inside
+          Index := StartIndex + 1;
+        State := stNone;
+      end else begin
+        Acronym := Acronym + Text[Index];
       end;
-      Inc(StartIndex);
-    end;
-  until StartIndex = 0;
+    end else begin
+      if Text[Index] = '(' then begin
+        State := stAcronymDefinition;
+        Acronym := '';
+        StartIndex := Index;
+      end else
+      if IsUppercaseAlpha(Text[Index]) then begin
+        State := stAcronymUsage;
+        Acronym := Text[Index];
+        if (Index - 1) >= 0 then AcronymCharBefore := Text[Index - 1]
+          else AcronymCharBefore := ' ';
+      end;
+    end;
+    Inc(Index);
+  until Index > Length(Text);
+  Lines.Free;
+end;
+
+function TFormCheck.AllowedSideChar(Before, After: Char): Boolean;
+begin
+  Result := ((Before = ' ') or (Before = #10) or (Before = #13) or (Before = ',') or
+    (Before = ';') or (Before = '(') or (Before = ')'))
+    and ((After = ' ') or (After = #10) or (After = #13) or (After = ',') or
+    (After = '.') or (After = ';') or (After = '(') or (After = ')'));
+
 end;
 
@@ -361 +457 @@
 end;
 
+function TFormCheck.IsLowercaseAlpha(Text: string): Boolean;
+var
+  I: Integer;
+begin
+  I := 1;
+  Result := True;
+  while (I <= Length(Text)) do begin
+    if not (Text[I] in ['a'..'z']) then begin
+      Result := False;
+      Break;
+    end;
+    Inc(I);
+  end;
+end;
+
 function TFormCheck.IsAlpha(Text: string): Boolean;
 var
@@ -383 +494 @@
 end;
 
+function TFormCheck.IsDigit(Text: Char): Boolean;
+begin
+  Result := Text in ['0'..'9'];
+end;
+
 procedure TFormCheck.ReportDifferencies;
 var
@@ -388 +504 @@
   J: Integer;
   Acronym: TAcronym;
+  Acronym2: TAcronym;
   Meaning: TAcronymMeaning;
-begin
+  Meaning2: TAcronymMeaning;
+begin
+  // In content but not in summary
   for I := 0 to AcronymDbContent.Acronyms.Count - 1 do begin
     Acronym := TAcronym(AcronymDbContent.Acronyms[I]);
+    if Acronym.Meanings.Count > 1 then
+      MemoReport.Lines.Add(Format(SAcronymContentMultipleMeanings, [Acronym.Name, Acronym.Meanings.GetNames]));
+    Acronym2 := AcronymDbSummary.Acronyms.SearchByName(Acronym.Name);
+    if not Assigned(Acronym2) then
     for J := 0 to Acronym.Meanings.Count - 1 do begin
       Meaning := TAcronymMeaning(Acronym.Meanings[J]);
@@ -399 +522 @@
   end;
 
+  // In summary but not in content
   for I := 0 to AcronymDbSummary.Acronyms.Count - 1 do begin
     Acronym := TAcronym(AcronymDbSummary.Acronyms[I]);
+    if Acronym.Meanings.Count > 1 then
+      MemoReport.Lines.Add(Format(SAcronymSummaryMultipleMeanings, [Acronym.Name, Acronym.Meanings.GetNames]));
+    Acronym2 := AcronymDbContent.Acronyms.SearchByName(Acronym.Name);
+    if not Assigned(Acronym2) then
     for J := 0 to Acronym.Meanings.Count - 1 do begin
       Meaning := TAcronymMeaning(Acronym.Meanings[J]);
       if not Assigned(AcronymDbContent.SearchAcronym(Acronym.Name, Meaning.Name, [sfCaseInsensitive])) then
         MemoReport.Lines.Add(Format(SMissingAcronymSummary, [Acronym.Name, Meaning.Name]));
+    end;
+  end;
+
+  // With different meaning
+  for I := 0 to AcronymDbSummary.Acronyms.Count - 1 do begin
+    Acronym := TAcronym(AcronymDbSummary.Acronyms[I]);
+    Acronym2 := AcronymDbContent.Acronyms.SearchByName(Acronym.Name);
+    if Assigned(Acronym2) then begin
+      if (Acronym.Meanings.Count = 1) and (Acronym2.Meanings.Count = 1) then begin
+        Meaning := TAcronymMeaning(Acronym.Meanings[0]);
+        Meaning2 := TAcronymMeaning(Acronym2.Meanings[0]);
+        if Meaning.Name <> Meaning2.Name then
+          MemoReport.Lines.Add(Format(SAcronymWithDifferentMeaning, [Acronym.Name, Meaning.Name, Meaning2.Name]));
+      end else
+        MemoReport.Lines.Add(Format(SAcronymWithDifferentMeaningCount, [Acronym.Name, Acronym.Meanings.Count, Acronym2.Meanings.Count]));
     end;
   end;