Dart
Dart
HTML Table to CSV
See more HTML-to-XML/Text Examples
Demonstrates a method for converting an HTML table to a CSV file.Note: This example requires Chilkat v9.5.0.77 or greater.
Chilkat Dart Downloads
import 'package:chilkat/chilkat.dart';
void main() {
// This example requires the Chilkat API to have been previously unlocked.
// See Global Unlock Sample for sample code.
// First download the HTML containing the table
final http = CkHttp();
final bdHtml = CkBinData();
try {
http.quickGetBd('https://example-code.com/data/etf_table.html', bdHtml);
} on ChilkatException catch (e) {
print(e.lastErrorText);
return;
}
// Convert to XML.
final htx = CkHtmlToXml();
htx.setHtmlBd(bdHtml);
final sbXml = CkStringBuilder();
htx.toXmlSb(sbXml);
final xml = CkXml();
xml.loadSb(sbXml, true);
// Remove attributes and sub-trees we don't need.
// (In other words, we're getting rid of clutter...)
var numRemoved = xml.pruneTag('thead');
numRemoved = xml.pruneAttribute('style');
numRemoved = xml.pruneAttribute('class');
// Scrub the element and attribute content.
xml.scrub('ContentTrimEnds,ContentTrimInside,AttrTrimEnds,AttrTrimInside');
// Let's see what we have...
print(xml.getXml());
// We have the following XML.
// Copy this XML into the online tool at Generate Parsing Code from XML
// as a starting point for accessing the data..
// <?xml version="1.0" encoding="utf-8"?>
// <root>
// <html>
// <head>
// <meta http-equiv="content-type" content="text/html; charset=UTF-8"/>
// </head>
// <body text="#000000" bgcolor="#FFFFFF">
// <div>
// <div>
// <table role="grid" data-scrollx="true" data-sortdirection="desc" data-sorton="-1"/>
// </div>
// </div>
// <div>
// <table id="topHoldingsTable" role="grid" data-scrollx="true" data-sortdirection="desc" data-sorton="-1">
// <tbody>
// <tr role="row">
// <td>
// <text>ITUB4</text>
// </td>
// <td>
// <text>ITAU UNIBANCO HOLDING PREF SA</text>
// </td>
// <td>
// <text>Financials</text>
// </td>
// <td>
// <text>Brazil</text>
// </td>
// <td>
// <text>10.94</text>
// </td>
// <td>
// <text>998,954,813.73</text>
// </td>
// </tr>
// <tr role="row">
// <td>
// <text>BBDC4</text>
// </td>
// <td>
// <text>BANCO BRADESCO PREF SA</text>
// </td>
// <td>
// <text>Financials</text>
// </td>
// <td>
// <text>Brazil</text>
// </td>
// <td>
// <text>9.01</text>
// </td>
// <td>
// <text>822,164,622.75</text>
// </td>
// </tr>
// ...
// ...
// ...
// </tbody>
// </table>
// </div>
// </body>
// </html>
// </root>
// This is the code generated by the online tool:
//
var i = 0;
var countI = 0;
var tableRole = '';
var tableDataScrollx = '';
var tableDataSortdirection = '';
var tableDataSorton = '';
var tableId = '';
var j = 0;
var countJ = 0;
var trRole = '';
var k = 0;
var countK = 0;
final tagPath = '';
var text = '';
i = 0;
countI = xml.numChildrenHavingTag('html|body|div');
while (i < countI) {
xml.i = i;
tableRole = xml.chilkatPath('html|body|div[i]|div|table|(role)');
tableDataScrollx = xml.chilkatPath('html|body|div[i]|div|table|(data-scrollx)');
tableDataSortdirection = xml.chilkatPath('html|body|div[i]|div|table|(data-sortdirection)');
tableDataSorton = xml.chilkatPath('html|body|div[i]|div|table|(data-sorton)');
tableId = xml.chilkatPath('html|body|div[i]|table|(id)');
tableRole = xml.chilkatPath('html|body|div[i]|table|(role)');
tableDataScrollx = xml.chilkatPath('html|body|div[i]|table|(data-scrollx)');
tableDataSortdirection = xml.chilkatPath('html|body|div[i]|table|(data-sortdirection)');
tableDataSorton = xml.chilkatPath('html|body|div[i]|table|(data-sorton)');
j = 0;
countJ = xml.numChildrenHavingTag('html|body|div[i]|table|tbody|tr');
while (j < countJ) {
xml.j = j;
trRole = xml.chilkatPath('html|body|div[i]|table|tbody|tr[j]|(role)');
k = 0;
countK = xml.numChildrenHavingTag('html|body|div[i]|table|tbody|tr[j]|td');
while (k < countK) {
xml.k = k;
text = xml.getChildContent('html|body|div[i]|table|tbody|tr[j]|td[k]|text');
k++;
}
j++;
}
i++;
}
// Let's modify the above code to build the CSV.
final csv = CkCsv();
csv.setColumnName(0, 'Ticker');
csv.setColumnName(1, 'Name');
csv.setColumnName(2, 'Sector');
csv.setColumnName(3, 'Country');
csv.setColumnName(4, 'Weight');
csv.setColumnName(5, 'Notional Vaue');
i = 0;
countI = xml.numChildrenHavingTag('html|body|div');
while (i < countI) {
xml.i = i;
j = 0;
countJ = xml.numChildrenHavingTag('html|body|div[i]|table|tbody|tr');
while (j < countJ) {
xml.j = j;
k = 0;
countK = xml.numChildrenHavingTag('html|body|div[i]|table|tbody|tr[j]|td');
while (k < countK) {
xml.k = k;
csv.setCell(j, k, xml.getChildContent('html|body|div[i]|table|tbody|tr[j]|td[k]|text'));
k++;
}
j++;
}
i++;
}
csv.saveFile('qa_output/brasil_etf.csv');
final csvStr = csv.saveToString();
print(csvStr);
// Our CSV looks like this:
// Ticker,Name,Sector,Country,Weight,Notional Vaue
// ITUB4,ITAU UNIBANCO HOLDING PREF SA,Financials,Brazil,10.94,"998,954,813.73"
// BBDC4,BANCO BRADESCO PREF SA,Financials,Brazil,9.01,"822,164,622.75"
// VALE3,CIA VALE DO RIO DOCE SH,Materials,Brazil,8.60,"785,290,260.07"
// PETR4,PETROLEO BRASILEIRO PREF SA,Energy,Brazil,5.68,"518,124,434.10"
// PETR3,PETROBRAS,Energy,Brazil,4.86,"443,254,438.53"
// B3SA3,B3 BRASIL BOLSA BALCAO SA,Financials,Brazil,4.57,"417,636,740.16"
// ABEV3,AMBEV SA,Consumer Staples,Brazil,4.57,"417,216,913.63"
// BBAS3,BANCO DO BRASIL SA,Financials,Brazil,3.25,"296,921,232.15"
// ITSA4,ITAUSA INVESTIMENTOS ITAU PREF SA,Financials,Brazil,2.90,"265,153,684.52"
// LREN3,LOJAS RENNER SA,Consumer Discretionary,Brazil,2.25,"205,832,175.98"
//
}