(Tcl) HTML Table to CSV

Demonstrates a method for converting an HTML table to a CSV file.

Note: This example requires Chilkat v9.5.0.77 or greater.

Chilkat Tcl Extension Downloads

Chilkat Tcl Extension Downloads

load ./chilkat.dll

# 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
set http [new_CkHttp]

set bdHtml [new_CkBinData]

set success [CkHttp_QuickGetBd $http "https://example-code.com/data/etf_table.html" $bdHtml]
if {$success != 1} then {
    puts [CkHttp_lastErrorText $http]
    delete_CkHttp $http
    delete_CkBinData $bdHtml
    exit
}

# Convert to XML.
set htx [new_CkHtmlToXml]

CkHtmlToXml_SetHtmlBd $htx $bdHtml

set sbXml [new_CkStringBuilder]

CkHtmlToXml_ToXmlSb $htx $sbXml

set xml [new_CkXml]

CkXml_LoadSb $xml $sbXml 1

# Remove attributes and sub-trees we don't need.
# (In other words, we're getting rid of clutter...)
set numRemoved [CkXml_PruneTag $xml "thead"]
set numRemoved [CkXml_PruneAttribute $xml "style"]
set numRemoved [CkXml_PruneAttribute $xml "class"]

# Scrub the element and attribute content.
CkXml_Scrub $xml "ContentTrimEnds,ContentTrimInside,AttrTrimEnds,AttrTrimInside"

# Let's see what we have...
puts [CkXml_getXml $xml]

# 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:
# 

set i 0
set count_i [CkXml_NumChildrenHavingTag $xml "html|body|div"]
while {$i < $count_i} {
    CkXml_put_I $xml $i
    set table_role [CkXml_chilkatPath $xml "html|body|div[i]|div|table|(role)"]
    set table_data_scrollx [CkXml_chilkatPath $xml "html|body|div[i]|div|table|(data-scrollx)"]
    set table_data_sortdirection [CkXml_chilkatPath $xml "html|body|div[i]|div|table|(data-sortdirection)"]
    set table_data_sorton [CkXml_chilkatPath $xml "html|body|div[i]|div|table|(data-sorton)"]
    set table_id [CkXml_chilkatPath $xml "html|body|div[i]|table|(id)"]
    set table_role [CkXml_chilkatPath $xml "html|body|div[i]|table|(role)"]
    set table_data_scrollx [CkXml_chilkatPath $xml "html|body|div[i]|table|(data-scrollx)"]
    set table_data_sortdirection [CkXml_chilkatPath $xml "html|body|div[i]|table|(data-sortdirection)"]
    set table_data_sorton [CkXml_chilkatPath $xml "html|body|div[i]|table|(data-sorton)"]
    set j 0
    set count_j [CkXml_NumChildrenHavingTag $xml "html|body|div[i]|table|tbody|tr"]
    while {$j < $count_j} {
        CkXml_put_J $xml $j
        set tr_role [CkXml_chilkatPath $xml "html|body|div[i]|table|tbody|tr[j]|(role)"]
        set k 0
        set count_k [CkXml_NumChildrenHavingTag $xml "html|body|div[i]|table|tbody|tr[j]|td"]
        while {$k < $count_k} {
            CkXml_put_K $xml $k
            set text [CkXml_getChildContent $xml "html|body|div[i]|table|tbody|tr[j]|td[k]|text"]
            set k [expr $k + 1]
        }
        set j [expr $j + 1]
    }
    set i [expr $i + 1]
}

# Let's modify the above code to build the CSV.
set csv [new_CkCsv]

CkCsv_SetColumnName $csv 0 "Ticker"
CkCsv_SetColumnName $csv 1 "Name"
CkCsv_SetColumnName $csv 2 "Sector"
CkCsv_SetColumnName $csv 3 "Country"
CkCsv_SetColumnName $csv 4 "Weight"
CkCsv_SetColumnName $csv 5 "Notional Vaue"

set i 0
set count_i [CkXml_NumChildrenHavingTag $xml "html|body|div"]
while {$i < $count_i} {
    CkXml_put_I $xml $i
    set j 0
    set count_j [CkXml_NumChildrenHavingTag $xml "html|body|div[i]|table|tbody|tr"]
    while {$j < $count_j} {
        CkXml_put_J $xml $j
        set k 0
        set count_k [CkXml_NumChildrenHavingTag $xml "html|body|div[i]|table|tbody|tr[j]|td"]
        while {$k < $count_k} {
            CkXml_put_K $xml $k
            CkCsv_SetCell $csv $j $k [CkXml_getChildContent $xml "html|body|div[i]|table|tbody|tr[j]|td[k]|text"]
            set k [expr $k + 1]
        }
        set j [expr $j + 1]
    }
    set i [expr $i + 1]
}

CkCsv_SaveFile $csv "qa_output/brasil_etf.csv"
set csvStr [CkCsv_saveToString $csv]
puts "$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"
# 

delete_CkHttp $http
delete_CkBinData $bdHtml
delete_CkHtmlToXml $htx
delete_CkStringBuilder $sbXml
delete_CkXml $xml
delete_CkCsv $csv