> I am trying to write an application that converts a CSV(or similar)it to
> an XML one.The application looks for a character(the comma - - or anything
> else specified in an Edit box - -), adds a starting and ending tag to the
> line, and writes the line to the new XML file. in the end I should get an
> XML file with the various elements.
{Your task has a number of subtasks.
The first is parsing the input file into lines. You can leave that to a
Tstringlist, if the files you need to handle are not in the
multimegabyte size range. If they are you would be best served by using
the good old Pascal Textfile routines, where a simple ReadLn( filevar, S
) gets you a line.
The second is parsing a line into its elements, based on a separator
character between the elements. This is also not so difficult to do,
especially if you don't need to deal with quoted elements that may
contain the separator. Search the newsgroup archives for "SplitString"
for an example. Tstringlist.Delimitedtext may be of use here, but be
warned that it considers any character <= #32 as a separator *in
addition* to what you define as Delimiter. It can deal with quoted
elements, though.
The second subtask would end with a TStringlist instance containing the
elements to store into the XML file for one line of the input file. This
is the input for the third task: to create a first-level XML element
containing the data. To write valid XML you need not only deal with
proper nesting of XML tags, you also have to properly represent some
characters that have special meaning in XML ('<' and '&' for instance).
I can recommend Berend de Boers xml_generator class
http://www.pobox.com/~berend/delphi for this task, it deals with all the
nastiness behind the scenes and produces syntactically correct XML
without the overhead of a DOM model implementation.
There is something else you need: a list of column names, one name for
each "column" in your XML file. These names will become the node names
for the subnodes of the produced XML. Depending on your input files you
may be able to get these names from the first line (which often is a
header line giving the column names).
Here is sketch (untested!) of the conversion routine: }
type
{: Callback for CSVToXML. If given the callback will be called
after each processed line.
@Param currentline is the 0-based number of the processed line
@Param totallines is the total number of lines. This may be a
raw estimate if the file is not completly loaded in memory.
@Returns true to continue processing, false to stop it. }
TProgressNotification =
function(currentline, totallines: Integer): Boolean of object;
{-- CSVToXML ----------------------------------------------------------}
{: Convert a delimiter-separated file of data to XML
@Param csvfilename is the file to convert
@Param xmlfilename is the xml file to create
@Param aSeparator is the separator for the data
@Param aRootNodeName is the name to use for the root node of the XML
file.
@Param columnnames is an optional list of column names to use as subnode
names. If this parameter is nil the first line of the data file must
contain a header line with the names to use.
@Param onProgress is an optional callback to call afte each processed
line.
@Precondition csvfilename exists
}{ Created 17.3.2003 by P. Below
-----------------------------------------------------------------------}
procedure CSVToXML(const csvfilename, xmlfilename: string;
const aSeparator: Char;
const aRootNodeName: string;
const columnnames: TStrings = nil;
const onProgress: TProgressNotification = nil);
function DoProgress(currentline, totallines: Integer): Boolean;
begin
if Assigned(onProgress) then
Result := onProgress(currentline, totallines)
else
Result := true;
end;
procedure WriteDataline(const line: string; header: TStringlist; xml: TXMLGenerator);
var
elements: TStringlist;
i, max: Integer;
begin
elements := TStringlist.Create;
try
elements.Delimiter := aSeparator;
elements.Delimitedtext := line;
if elements.count > header.count then
max := header.count
else
max := elements.count;
for i := 0 to max - 1 do begin
xml.StartTag(header[i]);
xml.AddData(elements[i]);
xml.StopTag;
end; { For }
finally
elements.Free;
end;
end;
procedure WriteData(data: TStringlist; xml: TXMLGenerator);
var
header: TStringlist;
firstline: Integer;
i: Integer;
begin
header := Tstringlist.Create;
try
firstline := 0;
if assigned(columnnames) then
header.Assign(columnnames)
else begin
header.Delimiter := aSeparator;
header.DelimitedText := data[0];
firstline := 1;
end; { Else }
for i := firstline to data.count - 1 do begin
WriteDataline(data[i], header, xml);
if not DoProgress(i, data.count) then
Break;
end; { For }
finally
header.Free;
end;
end;
procedure SaveStringToFile(const S, filename: string);
var
fs: TFilestream;
begin
fs := TFileStream.Create(filename, fmCreate);
try
if Length(S) > 0 then
fs.WriteBuffer(S[1], Length(S));
finally
fs.free
end;
end; { SaveStringToFile }
var
xml: TXMLGenerator; // from xml_generator unit by Berend de Boers
datafile: Tstringlist;
begin { CSVToXML }
if not FileExists(csvfilename) then
raise Exception.CreateFmt('Input file %s not found', [csvfilename]);
datafile := Tstringlist.Create;
try
datafile.LoadfromFile(csvfilename);
xml := TXMLGenerator.CreateWithEncoding(16 * 1024, encISO_8859_1);
try
xml.StartTag(aRootNodeName);
if datafile.count > 0 then
WriteData(datafile, xml);
xml.StopTag;
SaveStringToFile(xml.AsLatin1, xmlfilename);
finally
xml.Free;
end;
finally
datafile.free;
end;
end; { CSVToXML }