Rust
Rust
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 Rust Downloads
// 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 = chilkat::Http::new();
let bd_html = chilkat::BinData::new();
if http.quick_get_bd("https://example-code.com/data/etf_table.html", &bd_html).is_err() {
println!("{}", http.last_error_text());
return;
}
// Convert to XML.
let htx = chilkat::HtmlToXml::new();
let _ = htx.set_html_bd(&bd_html);
let sb_xml = chilkat::StringBuilder::new();
let _ = htx.to_xml_sb(&sb_xml);
let xml = chilkat::Xml::new();
let _ = xml.load_sb(&sb_xml, true);
// Remove attributes and sub-trees we don't need.
// (In other words, we're getting rid of clutter...)
let mut num_removed = xml.prune_tag("thead");
num_removed = xml.prune_attribute("style");
num_removed = xml.prune_attribute("class");
// Scrub the element and attribute content.
xml.scrub("ContentTrimEnds,ContentTrimInside,AttrTrimEnds,AttrTrimInside");
// Let's see what we have...
println!("{}", xml.get_xml().unwrap_or_default());
// 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:
//
let mut i: i32 = 0;
let mut count_i: i32 = 0;
let mut j: i32 = 0;
let mut count_j: i32 = 0;
let mut k: i32 = 0;
let mut count_k: i32 = 0;
i = 0;
count_i = xml.num_children_having_tag("html|body|div");
while i < count_i {
xml.set_i(i);
let _ = xml.chilkat_path("html|body|div[i]|div|table|(role)").unwrap_or_default();
let _ = xml.chilkat_path("html|body|div[i]|div|table|(data-scrollx)").unwrap_or_default();
let _ = xml.chilkat_path("html|body|div[i]|div|table|(data-sortdirection)").unwrap_or_default();
let _ = xml.chilkat_path("html|body|div[i]|div|table|(data-sorton)").unwrap_or_default();
let _ = xml.chilkat_path("html|body|div[i]|table|(id)").unwrap_or_default();
let _ = xml.chilkat_path("html|body|div[i]|table|(role)").unwrap_or_default();
let _ = xml.chilkat_path("html|body|div[i]|table|(data-scrollx)").unwrap_or_default();
let _ = xml.chilkat_path("html|body|div[i]|table|(data-sortdirection)").unwrap_or_default();
let _ = xml.chilkat_path("html|body|div[i]|table|(data-sorton)").unwrap_or_default();
j = 0;
count_j = xml.num_children_having_tag("html|body|div[i]|table|tbody|tr");
while j < count_j {
xml.set_j(j);
let _ = xml.chilkat_path("html|body|div[i]|table|tbody|tr[j]|(role)").unwrap_or_default();
k = 0;
count_k = xml.num_children_having_tag("html|body|div[i]|table|tbody|tr[j]|td");
while k < count_k {
xml.set_k(k);
let _ = xml.get_child_content("html|body|div[i]|table|tbody|tr[j]|td[k]|text").unwrap_or_default();
k = k + 1;
}
j = j + 1;
}
i = i + 1;
}
// Let's modify the above code to build the CSV.
let csv = chilkat::Csv::new();
let _ = csv.set_column_name(0, "Ticker");
let _ = csv.set_column_name(1, "Name");
let _ = csv.set_column_name(2, "Sector");
let _ = csv.set_column_name(3, "Country");
let _ = csv.set_column_name(4, "Weight");
let _ = csv.set_column_name(5, "Notional Vaue");
i = 0;
count_i = xml.num_children_having_tag("html|body|div");
while i < count_i {
xml.set_i(i);
j = 0;
count_j = xml.num_children_having_tag("html|body|div[i]|table|tbody|tr");
while j < count_j {
xml.set_j(j);
k = 0;
count_k = xml.num_children_having_tag("html|body|div[i]|table|tbody|tr[j]|td");
while k < count_k {
xml.set_k(k);
let _ = csv.set_cell(j, k, &xml.get_child_content("html|body|div[i]|table|tbody|tr[j]|td[k]|text").unwrap_or_default());
k = k + 1;
}
j = j + 1;
}
i = i + 1;
}
let _ = csv.save_file("qa_output/brasil_etf.csv");
let csv_str = csv.save_to_string().unwrap_or_default();
println!("{}", csv_str);
// 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"
//