Find valid HTML Tag in String lazarus / 21-General ---------------------------------------------------------------------- Maris · 2023-08-23 13:14 Hi! I wanted to ask for help with probably very simple problem. What I need to do, is to check a string for any valid HTML tag. Examples: Code: Pascal  [Select][+][-]window.onload = function(){var x1 = document.getElementById("main_content_section"); if (x1) { var x = document.getElementsByClassName("geshi");for (var i = 0; i "Some text

more text

" {True}"Some text < p >more text

" {True}"Some text

more text

" {True}"In range < 1 meter and > 1 cm" {False}"Must be >1m" {False} Is there any simple way to do this? Preferably with build-in units to maintain code easier. At least any built-in unit with full list of valid HTML tags would be nice, which I couldn't find. In more details (maybe there is better way in general). What I'm trying to do is to convert Excel with HTML code to .html file to proces it in CAT Tools. Afterwords when everything is done it will be imported back to excel in appropriate cells. Problems showed up when I noticed some columns has plain text only and even more some HTML columns contains plain text in some rows and some of them has symbols > and < which are not encoded, so in general it's a plain text. If symbols > and < are note encoded for non html cells, generated .html file is not valid. Algorithm could be like this: Code: HTML5  [Select][+][-]window.onload = function(){var x1 = document.getElementById("main_content_section"); if (x1) { var x = document.getElementsByClassName("geshi");for (var i = 0; i if "<" or ">" found then  if valid HTML tag found then  begin    add to output as HTML "
CellValue
"  end  else  begin    encode html entities and add to output "
EncodeHTMLEntities(CellValue)
"    (where "e" after cell address means encoded)-----------^  endelse  add to output "
CellValue
" (Text without "<" and ">" is ok to add in between div tags even if it's plain text) Processed file will be imported back based on cell address and Decode Entities will be used for cells marked with "e" (encoded). Generated .html file would look like this: Code: HTML5  [Select][+][-]window.onload = function(){var x1 = document.getElementById("main_content_section"); if (x1) { var x = document.getElementsByClassName("geshi");for (var i = 0; i                   
Some Text or HTML code
Some Text or HTML code
Some Text or HTML code
Some Text or HTML code
Some Text or HTML code
Some Text or HTML code
Some Text or HTML code
Some Text or HTML code
Some Text or HTML code
Some Text or HTML code
  Thank you all in advance! ---------------------------------------------------------------------- Maris · 2023-08-23 14:35 After a while of searching, it looks like there is no ready made solution for finding valid HTML tags in string, so best option would be to search for "" pattern, extract "abc" part and compare it with all possible tags. I found tags are defined in HTMLDefs unit, so this would be useful probably. Any other suggestions would be greatly appreciated. ---------------------------------------------------------------------- cdbc · 2023-08-24 05:43 Hi If you've got this HTMLDef unit, couldn't you then put all these tags into a e.g.: TStringHashTable from "contnrs" unit, as a snappy lookup... And then compare against the lookup-table, it's easy to write a "Find" function for the hashtable... Just my 2 cents worth  Regards Benny ---------------------------------------------------------------------- Zvoni · 2023-08-24 06:47 Untested: Regex-Pattern: (?:<\s*([a-zA-Z]+)\s*.*?>)(?:.*)(<\s*\/*(\1)\s*>) https://regex101.com/r/x1TpVc/1 ---------------------------------------------------------------------- Bart · 2023-08-24 08:23 We do have a fasthtmlparser unit IIRC. Bart ---------------------------------------------------------------------- Zvoni · 2023-08-24 10:08 Quote from: Bart on August 24, 2023, 10:23:29 amWe do have a fasthtmlparser unit IIRC. Bart https://github.com/z505/fast-html-parser if it's tested, then yes: Use tried and trusted ---------------------------------------------------------------------- wp · 2023-08-24 10:19 fasthtmlparser is even included in fpc (folder packages/chm/src). It is my "swiss army knife" when it comes to parsing html, and it is used in TAChart and in fpspreadsheet. ---------------------------------------------------------------------- Maris · 2023-08-24 16:14 Thank you all for answers and suggestions! As it was a bit urgent I made program in simple way. I'll take a look on suggested units and data structures maybe it will be useful to make code more readable or improve performance. Although I'm using OLE objects, so performance cannot be very good anyway. Main problem is that excel files with HTML code can contain anything in-between and there is no guarantee code itself is valid. We receiving files from different customers and we never know what is the source. Could be export from some kind of CMS or anything else. I used TStringList for list of HTML tags, also used old school Pos instead of Regex and could solve task without fully parsing of HTML code, at least for now. The part which is checking if cell contains HTML code or plain text and reading content from excel is as follows: Code: Pascal  [Select][+][-]window.onload = function(){var x1 = document.getElementById("main_content_section"); if (x1) { var x = document.getElementsByClassName("geshi");for (var i = 0; i uses  Classes, SysUtils, Forms, Controls, Graphics, Dialogs, ComCtrls, LazFileUtils,  FileUtil, ExtCtrls, Grids, ActnList, StdCtrls, EditBtn, SynHighlighterHTML,  SynEdit, ComObj, HTMLDefs; type   { TFormHTMLExcel }   TFormHTMLExcel = class(TForm)    ...  private    HTMLTags: TStringList;    ...    procedure ExcelToHtml(FileName: String; OutputDir: String);    function TagFound(Code: String): Boolean; Inline;   public    constructor Create(AOwner: TComponent); override;    destructor Destroy; override;   end; var  FormHTMLExcel: TFormHTMLExcel; implementation {$R *.lfm} { TFormHTMLExcel } constructor TFormHTMLExcel.Create(AOwner: TComponent);var  i: THTMLElementTag;begin  inherited;   HTMLTags := TStringList.Create;  HTMLTags.Sorted := True;   for i in THTMLElementTag do  begin    HTMLTags.Append(HTMLElementProps[i].Name);  end;  HTMLTags.Append('!DOCTYPE'); {Add Doctype definition to tag list}end; destructor TFormHTMLExcel.Destroy;begin  inherited;   HTMLTags.Free;end; function TFormHTMLExcel.TagFound(Code: String): Boolean; Inline;var  TNStart, TNEnd, NextSpace: Word;begin  TNStart := Pos('<', Code);  if (TNStart > 0) and (TNStart < Length(Code)) then  begin    TNEnd := Pos('>', Code, TNStart + 1);    if TNEnd > 0 then  begin      {Search for next non-space character after <}      repeat      inc(TNStart);      until (Code[TNStart] <> ' ') or (TNStart = Length(Code));      {Condition TNStart = Length(Code) shouldn't happen in theory, but who knows}       {Search for next space or > symbol}    TNEnd := Pos('>', Code, TNStart + 1);      NextSpace := Pos(' ', Code, TNStart + 1);      if (NextSpace > 0) and (NextSpace < TNEnd) then      begin        TNEnd := NextSpace;      end;       if HTMLTags.IndexOf(Copy(Code, TNStart, TNEnd - TNStart)) > -1 then      begin        TagFound := True;      end      else      begin        TagFound := False;      end;    end    else    begin      {If no closing greater then (>) symbol found, it must be plain text or HTML code is not valid       Do not process such code as HTML, consider it as plain text to avoid mistakes in Trados}      TagFound := False;    end;  end  else  begin    TagFound := False;  end;end; procedure TFormHTMLExcel.ExcelToHtml(FileName: String; OutputDir: String);const  AppName = 'Excel.Application';var  App, Workbook: OLEVariant;  InputFile: Variant;  ResultHTML: TStringList;  r, c, s, RowCount, ColumnCount: Word;  CellValue: String;begin  ResultHTML := TStringList.Create;  App := CreateOleObject(AppName);   try    App.Visible := False;    App.DisplayAlerts := False;    InputFile := FileName;     Workbook := App.Workbooks.Open(InputFile);     for s := 1 to Workbook.Sheets.Count do    begin      Workbook.Sheets.Item[s].Activate;       RowCount := App.ActiveSheet.UsedRange.Rows.Count;      ColumnCount := App.ActiveSheet.UsedRange.Columns.Count;       ResultHTML.Clear;      ResultHTML.Append(SynEditHeader.Text);       for c := 1 to ColumnCount do      begin        for r := 1 to RowCount do        begin          CellValue := App.Cells[r, c].Value;          if Length(CellValue) > 0 then          begin            if (not TagFound(CellValue)) and ((Pos('<', CellValue) > 0) or (Pos('>', CellValue) > 0)) then          begin              {Use Encoding only when < and > symbols are found in text, looks like other               HTML special characters does not break structure and can be used as it is}              CellValue := StringReplace(CellValue, '<', '<', [rfReplaceAll]);            CellValue := StringReplace(CellValue, '>', '>', [rfReplaceAll]);            ResultHTML.Append('
' + CellValue + '
');            end            else            begin              ResultHTML.Append('
' + CellValue + '
');            end;          end;        end;      end;       ResultHTML.Append(SynEditFooter.Text);      ResultHTML.SaveToFile(OutputDir + IntToStr(s) + '-' + App.ActiveSheet.Name + '.html');    end;  finally    App.Quit;    App := Unassigned;    ResultHTML.Free;  end;end; In short, program populates String list with tag names in Constructor. This list is used later to find valid tags in string. To read content, program loops through all excel cells and search for any valid HTML tags, if not found encodes < and > symbols only before storing result in .html file. Not fully tested yet, but so far it can distinguish cells with plain text like ">1m2", "<2m2" or similar and encode them before saving to .html file. With more complicated content could be less precise. Another interesting thing is that Lazarus shows bunch of hints like this: Code: Pascal  [Select][+][-]window.onload = function(){var x1 = document.getElementById("main_content_section"); if (x1) { var x = document.getElementsByClassName("geshi");for (var i = 0; i unit1.pas(151,18) Note: Call to subroutine "operator :=(const source:AnsiString):Variant;" marked as inline is not inlined I assume it's implemented in newer versions of Lazarus as I do not remember such hints before. Did a quick search about it and found it is something about Variant variables, but as far as I know there is no other way to work with office OLE objects.