pascal/lazarus-ccr
1
0
Fork 0
Code Issues Releases Activity

fpspreadsheet: More improvements for reading numbers in csv files and automatic format detection.

Browse Source 
git-svn-id: https://svn.code.sf.net/p/lazarus-ccr/svn@3672 8e941d3f-bd1b-0410-a28a-d453659cc2b4
This commit is contained in:
wp_xxyyzz
2014-10-20 21:04:20 +00:00
parent 43acbe2908
commit 4473d3b4a5
3 changed files with 155 additions and 75 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
components/fpspreadsheet/examples/spready/spready.lpi
View File

@@ -111,12 +111,10 @@
<ComponentName Value="CSVParamsForm"/> <ComponentName Value="CSVParamsForm"/>
<HasResources Value="True"/> <HasResources Value="True"/>
<ResourceBaseClass Value="Form"/> <ResourceBaseClass Value="Form"/>
<UnitName Value="sCSVParamsForm"/>
</Unit2> </Unit2>
<Unit3> <Unit3>
<Filename Value="sctrls.pas"/> <Filename Value="sctrls.pas"/>
<IsPartOfProject Value="True"/> <IsPartOfProject Value="True"/>
<UnitName Value="sCtrls"/>
</Unit3> </Unit3>
<Unit4> <Unit4>
<Filename Value="sformatsettingsform.pas"/> <Filename Value="sformatsettingsform.pas"/>
@@ -124,7 +122,6 @@
<ComponentName Value="FormatSettingsForm"/> <ComponentName Value="FormatSettingsForm"/>
<HasResources Value="True"/> <HasResources Value="True"/>
<ResourceBaseClass Value="Form"/> <ResourceBaseClass Value="Form"/>
<UnitName Value="sFormatsettingsForm"/>
</Unit4> </Unit4>
</Units> </Units>
</ProjectOptions> </ProjectOptions>

76
components/fpspreadsheet/fpscsv.pas
View File

@@ -14,8 +14,8 @@ type
FWorksheetName: String; FWorksheetName: String;
function IsBool(AText: String; out AValue: Boolean): Boolean; function IsBool(AText: String; out AValue: Boolean): Boolean;
function IsDateTime(AText: String; out ADateTime: TDateTime): Boolean; function IsDateTime(AText: String; out ADateTime: TDateTime): Boolean;
function IsNumber(AText: String; out ANumber: Double; function IsNumber(AText: String; out ANumber: Double; out ANumFormat: TsNumberFormat;
out ACurrencySymbol, AWarning: String): Boolean; out ADecimals: Integer; out ACurrencySymbol, AWarning: String): Boolean;
function IsQuotedText(var AText: String): Boolean; function IsQuotedText(var AText: String): Boolean;
procedure ReadCellValue(ARow, ACol: Cardinal; AText: String); procedure ReadCellValue(ARow, ACol: Cardinal; AText: String);
protected protected
@@ -207,10 +207,13 @@ begin
end; end;
function TsCSVReader.IsNumber(AText: String; out ANumber: Double; function TsCSVReader.IsNumber(AText: String; out ANumber: Double;
out ANumFormat: TsNumberFormat; out ADecimals: Integer;
out ACurrencySymbol, AWarning: String): Boolean; out ACurrencySymbol, AWarning: String): Boolean;
var var
p: Integer; p: Integer;
decsep, thsep: Char;
begin begin
Result := false;
AWarning := ''; AWarning := '';
// To detect whether the text is a currency value we look for the currency // To detect whether the text is a currency value we look for the currency
@@ -223,11 +226,8 @@ begin
if p > 0 then begin if p > 0 then begin
Delete(AText, p, Length(ACurrencySymbol)); Delete(AText, p, Length(ACurrencySymbol));
AText := Trim(AText); AText := Trim(AText);
if AText = '' then begin if AText = '' then
Result := false;
ACurrencySymbol := '';
exit; exit;
end;
// Negative financial values are often enclosed by parenthesis // Negative financial values are often enclosed by parenthesis
if ((AText[1] = '(') and (AText[Length(AText)] = ')')) then if ((AText[1] = '(') and (AText[Length(AText)] = ')')) then
AText := '-' + Trim(Copy(AText, 2, Length(AText)-2)); AText := '-' + Trim(Copy(AText, 2, Length(AText)-2));
@@ -235,11 +235,57 @@ begin
ACurrencySymbol := ''; ACurrencySymbol := '';
if CSVParams.AutoDetectNumberFormat then if CSVParams.AutoDetectNumberFormat then
Result := TryStrToFloatAuto(AText, ANumber, AWarning) Result := TryStrToFloatAuto(AText, ANumber, decsep, thsep, AWarning)
else else begin
Result := TryStrToFloat(AText, ANumber, CSVParams.FormatSettings); Result := TryStrToFloat(AText, ANumber, CSVParams.FormatSettings);
if Result then
begin
if pos(CSVParams.FormatSettings.DecimalSeparator, AText) = 0
then decsep := #0
else decsep := CSVParams.FormatSettings.DecimalSeparator;
if pos(CSVParams.FormatSettings.ThousandSeparator, AText) = 0
then thsep := #0
else thsep := CSVParams.FormatSettings.ThousandSeparator;
end;
end;
if not Result then ACurrencySymbol := ''; // Try to determine the number format
if Result then
begin
if thsep <> #0 then
ANumFormat := nfFixedTh
else
ANumFormat := nfGeneral;
// count number of decimal places and try to catch special formats
ADecimals := 0;
if decsep <> #0 then
begin
// Go to the decimal separator and search towards the end of the string
p := pos(decsep, AText) + 1;
while (p <= Length(AText)) do begin
// exponential format
if AText[p] in ['+', '-', 'E', 'e'] then
begin
ANumFormat := nfExp;
break;
end else
// percent format
if AText[p] = '%' then
begin
ANumFormat := nfPercentage;
break;
end else
begin
inc(p);
inc(ADecimals);
end;
end;
if (ADecimals > 0) and (ADecimals < 9) and (ANumFormat = nfGeneral) then
// "no formatting" assumed if there are "many" decimals
ANumFormat := nfFixed;
end;
end else
ACurrencySymbol := '';
end; end;
function TsCSVReader.IsQuotedText(var AText: String): Boolean; function TsCSVReader.IsQuotedText(var AText: String): Boolean;
@@ -268,6 +314,7 @@ var
currSym: string; currSym: string;
warning: String; warning: String;
nf: TsNumberFormat; nf: TsNumberFormat;
decs: Integer;
begin begin
// Empty strings are blank cells -- nothing to do // Empty strings are blank cells -- nothing to do
if AText = '' then if AText = '' then
@@ -282,7 +329,10 @@ begin
// Remove quotes // Remove quotes
if (AText[1] = CSVParams.QuoteChar) and (AText[Length(AText)] = CSVParams.QuoteChar) then if (AText[1] = CSVParams.QuoteChar) and (AText[Length(AText)] = CSVParams.QuoteChar) then
Delete(AText, 2, Length(AText)-2); begin
Delete(AText, Length(AText), 1);
Delete(AText, 1, 1);
end;
{ {
// Quoted text is a TEXT cell // Quoted text is a TEXT cell
@@ -294,12 +344,12 @@ begin
} }
// Check for a NUMBER or CURRENCY cell // Check for a NUMBER or CURRENCY cell
if IsNumber(AText, dblValue, currSym, warning) then if IsNumber(AText, dblValue, nf, decs, currSym, warning) then
begin begin
if currSym <> '' then if currSym <> '' then
FWorksheet.WriteCurrency(ARow, ACol, dblValue, nfCurrency, 2, currSym) FWorksheet.WriteCurrency(ARow, ACol, dblValue, nfCurrency, decs, currSym)
else else
FWorksheet.WriteNumber(ARow, ACol, dblValue); FWorksheet.WriteNumber(ARow, ACol, dblValue, nf, decs);
if warning <> '' then if warning <> '' then
FWorkbook.AddErrorMsg('Cell %s: %s', [GetCellString(ARow, ACol), warning]); FWorkbook.AddErrorMsg('Cell %s: %s', [GetCellString(ARow, ACol), warning]);
exit; exit;

151
components/fpspreadsheet/fpsutils.pas
View File

@@ -130,7 +130,7 @@ function FormatDateTime(const FormatStr: string; DateTime: TDateTime;
const FormatSettings: TFormatSettings; Options : TFormatDateTimeOptions = []): string; const FormatSettings: TFormatSettings; Options : TFormatDateTimeOptions = []): string;
function TryStrToFloatAuto(AText: String; out ANumber: Double; function TryStrToFloatAuto(AText: String; out ANumber: Double;
out AWarning: String): Boolean; out ADecimalSeparator, AThousandSeparator: Char; out AWarning: String): Boolean;
function TwipsToPts(AValue: Integer): Single; function TwipsToPts(AValue: Integer): Single;
function PtsToTwips(AValue: Single): Integer; function PtsToTwips(AValue: Single): Integer;
@@ -1481,17 +1481,21 @@ end;
Is needed for reading CSV files. Is needed for reading CSV files.
-------------------------------------------------------------------------------} -------------------------------------------------------------------------------}
function TryStrToFloatAuto(AText: String; out ANumber: Double; function TryStrToFloatAuto(AText: String; out ANumber: Double;
out AWarning: String): Boolean; out ADecimalSeparator, AThousandSeparator: Char; out AWarning: String): Boolean;
var var
i: Integer; i: Integer;
testSep: Char; testSep: Char;
testSepPos: Integer; testSepPos: Integer;
lastDigitPos: Integer;
isPercent: Boolean; isPercent: Boolean;
isExp: Boolean;
fs: TFormatSettings; fs: TFormatSettings;
done: Boolean; done: Boolean;
begin begin
Result := false; Result := false;
AWarning := ''; AWarning := '';
ADecimalSeparator := #0;
AThousandSeparator := #0;
if AText = '' then if AText = '' then
exit; exit;
@@ -1506,65 +1510,89 @@ begin
// but no decimal separator misinterprets the thousand separator as a // but no decimal separator misinterprets the thousand separator as a
// decimal separator. // decimal separator.
done := false; // Indicates that both decimal and thousand separators are found done := false; // Indicates that both decimal and thousand separators are found
testSep := #0; // Separator candidate to be tested testSep := #0; // Separator candidate to be tested
testSepPos := 0; // Position of this separator chandidate in the string testSepPos := 0; // Position of this separator candidate in the string
lastDigitPos := 0; // Position of the last numerical digit
isExp := false; // Flag for exponential format
isPercent := false; // Flag for percentage format
i := Length(AText); // Start at end... i := Length(AText); // Start at end...
while i >= 1 do // ...and search towards start while i >= 1 do // ...and search towards start
begin begin
if AText[i] in ['.', ','] then case AText[i] of
begin '0'..'9':
if testSep = #0 then begin if (lastDigitPos = 0) and (AText[i] in ['0'..'9']) then
testSep := AText[i]; lastDigitPos := i;
testSepPos := i;
end; 'e', 'E':
// This is the right-most separator candidate in the text isExp := true;
// It can be a decimal or a thousand separator.
dec(i); '%':
while i >= 1 do isPercent := true;
begin
if not (AText[i] in ['0'..'9']) then begin '+', '-':
Result := false; ;
exit;
end; '.', ',':
// If we find the testSep character again it must be a thousand separator.
if (AText[i] = testSep) then
begin begin
// ... but only if there are 3 numerical digits in between if testSep = #0 then begin
if (testSepPos - i = 4) then testSep := AText[i];
begin testSepPos := i;
fs.ThousandSeparator := testSep; end;
// The decimal separator is the "other" character. // This is the right-most separator candidate in the text
if testSep = '.' then // It can be a decimal or a thousand separator.
fs.DecimalSeparator := ',' // Therefore, we continue searching from here.
else dec(i);
fs.DecimalSeparator := '.'; while i >= 1 do
done := true; begin
i := 0; if not (AText[i] in ['0'..'9', '+', '-']) then
end else exit;
begin
Result := false; // If we find the testSep character again it must be a thousand separator.
exit; if (AText[i] = testSep) then
begin
// ... but only if there are 3 numerical digits in between
if (testSepPos - i = 4) then
begin
fs.ThousandSeparator := testSep;
// The decimal separator is the "other" character.
if testSep = '.' then
fs.DecimalSeparator := ','
else
fs.DecimalSeparator := '.';
ADecimalSeparator := fs.DecimalSeparator;
AThousandSeparator := fs.ThousandSeparator;
done := true;
i := 0;
end else
begin
Result := false;
exit;
end;
end
else
// If we find the "other" separator character, then testSep was a
// decimal separator and the current character is a thousand separator.
// But there must be 3 digits in between.
if AText[i] in ['.', ','] then
begin
if testSepPos - i <> 4 then // no 3 digits in between --> no number, maybe a date.
exit;
fs.DecimalSeparator := testSep;
fs.ThousandSeparator := AText[i];
ADecimalSeparator := fs.DecimalSeparator;
AThousandSeparator := fs.ThousandSeparator;
done := true;
i := 0;
end;
dec(i);
end; end;
end
else
// If we find the "other" separator character, then testSep was a
// decimal separator and the current character is a thousand separator.
if AText[i] in ['.',','] then
begin
fs.DecimalSeparator := testSep;
fs.ThousandSeparator := AText[i];
done := true;
i := 0;
end; end;
dec(i);
end; else
end else exit; // Non-numeric character found, no need to continue
if not (AText[i] in ['0'..'9', '+', '-', 'e', 'E', '%']) then
begin
Result := false;
AWarning := '';
exit;
end; end;
dec(i); dec(i);
end; end;
@@ -1576,9 +1604,10 @@ begin
// type is found and it is at the third position from the string's end it // type is found and it is at the third position from the string's end it
// might by a thousand separator or a decimal separator. We assume the // might by a thousand separator or a decimal separator. We assume the
// latter case, but create a warning. // latter case, but create a warning.
if Length(AText) - testSepPos = 3 then if (lastDigitPos - testSepPos = 3) and not isPercent then
AWarning := Format(rsAmbiguousDecThouSeparator, [AText]); AWarning := Format(rsAmbiguousDecThouSeparator, [AText]);
fs.DecimalSeparator := testSep; fs.DecimalSeparator := testSep;
ADecimalSeparator := fs.DecimalSeparator;
// Make sure that the thousand separator is different from the decimal sep. // Make sure that the thousand separator is different from the decimal sep.
if testSep = '.' then fs.ThousandSeparator := ',' else fs.ThousandSeparator := '.'; if testSep = '.' then fs.ThousandSeparator := ',' else fs.ThousandSeparator := '.';
end; end;
@@ -1595,9 +1624,13 @@ begin
// Try string-to-number conversion // Try string-to-number conversion
Result := TryStrToFloat(AText, ANumber, fs); Result := TryStrToFloat(AText, ANumber, fs);
// If successful take care of the percentage sign // If successful ...
if Result and isPercent then if Result then
ANumber := ANumber * 0.01; begin
// ... take care of the percentage sign
if isPercent then
ANumber := ANumber * 0.01;
end;
end; end;