Swift
Swift
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 Swift Downloads
func chilkatTest() {
var success: Bool = false
// 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
let http = CkoHttp()!
let bdHtml = CkoBinData()!
success = http.quickGetBd(url: "https://example-code.com/data/etf_table.html", binData: bdHtml)
if success != true {
print("\(http.lastErrorText!)")
return
}
// Convert to XML.
let htx = CkoHtmlToXml()!
htx.setHtmlBd(bd: bdHtml)
let sbXml = CkoStringBuilder()!
htx.toXmlSb(sb: sbXml)
let xml = CkoXml()!
xml.loadSb(sb: sbXml, autoTrim: true)
// Remove attributes and sub-trees we don't need.
// (In other words, we're getting rid of clutter...)
var numRemoved: Int = xml.pruneTag(tag: "thead").intValue
numRemoved = xml.pruneAttribute(attrName: "style").intValue
numRemoved = xml.pruneAttribute(attrName: "class").intValue
// Scrub the element and attribute content.
xml.scrub(directives: "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: Int
var count_i: Int
var table_role: String?
var table_data_scrollx: String?
var table_data_sortdirection: String?
var table_data_sorton: String?
var table_id: String?
var j: Int
var count_j: Int
var tr_role: String?
var k: Int
var count_k: Int
var tagPath: String?
var text: String?
i = 0
count_i = xml.numChildrenHavingTag(tag: "html|body|div").intValue
while i < count_i {
xml.i = i
table_role = xml.chilkatPath(cmd: "html|body|div[i]|div|table|(role)")
table_data_scrollx = xml.chilkatPath(cmd: "html|body|div[i]|div|table|(data-scrollx)")
table_data_sortdirection = xml.chilkatPath(cmd: "html|body|div[i]|div|table|(data-sortdirection)")
table_data_sorton = xml.chilkatPath(cmd: "html|body|div[i]|div|table|(data-sorton)")
table_id = xml.chilkatPath(cmd: "html|body|div[i]|table|(id)")
table_role = xml.chilkatPath(cmd: "html|body|div[i]|table|(role)")
table_data_scrollx = xml.chilkatPath(cmd: "html|body|div[i]|table|(data-scrollx)")
table_data_sortdirection = xml.chilkatPath(cmd: "html|body|div[i]|table|(data-sortdirection)")
table_data_sorton = xml.chilkatPath(cmd: "html|body|div[i]|table|(data-sorton)")
j = 0
count_j = xml.numChildrenHavingTag(tag: "html|body|div[i]|table|tbody|tr").intValue
while j < count_j {
xml.j = j
tr_role = xml.chilkatPath(cmd: "html|body|div[i]|table|tbody|tr[j]|(role)")
k = 0
count_k = xml.numChildrenHavingTag(tag: "html|body|div[i]|table|tbody|tr[j]|td").intValue
while k < count_k {
xml.k = k
text = xml.getChildContent(tagPath: "html|body|div[i]|table|tbody|tr[j]|td[k]|text")
k = k + 1
}
j = j + 1
}
i = i + 1
}
// Let's modify the above code to build the CSV.
let csv = CkoCsv()!
csv.setColumnName(index: 0, columnName: "Ticker")
csv.setColumnName(index: 1, columnName: "Name")
csv.setColumnName(index: 2, columnName: "Sector")
csv.setColumnName(index: 3, columnName: "Country")
csv.setColumnName(index: 4, columnName: "Weight")
csv.setColumnName(index: 5, columnName: "Notional Vaue")
i = 0
count_i = xml.numChildrenHavingTag(tag: "html|body|div").intValue
while i < count_i {
xml.i = i
j = 0
count_j = xml.numChildrenHavingTag(tag: "html|body|div[i]|table|tbody|tr").intValue
while j < count_j {
xml.j = j
k = 0
count_k = xml.numChildrenHavingTag(tag: "html|body|div[i]|table|tbody|tr[j]|td").intValue
while k < count_k {
xml.k = k
csv.setCell(row: j, col: k, content: xml.getChildContent(tagPath: "html|body|div[i]|table|tbody|tr[j]|td[k]|text"))
k = k + 1
}
j = j + 1
}
i = i + 1
}
csv.saveFile(path: "qa_output/brasil_etf.csv")
var csvStr: String? = 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"
//
}