|
Chilkat • HOME • .NET Core C# • Android™ • AutoIt • C • C# • C++ • Chilkat2-Python • CkPython • Classic ASP • DataFlex • Delphi ActiveX • Delphi DLL • Go • Java • Lianja • Mono C# • Node.js • Objective-C • PHP ActiveX • PHP Extension • Perl • PowerBuilder • PowerShell • PureBasic • Ruby • SQL Server • Swift 2 • Swift 3,4,5... • Tcl • Unicode C • Unicode C++ • VB.NET • VBScript • Visual Basic 6.0 • Visual FoxPro • Xojo Plugin
(C) HTML Table to CSVDemonstrates a method for converting an HTML table to a CSV file. Note: This example requires Chilkat v9.5.0.77 or greater.
#include <C_CkHttp.h> #include <C_CkBinData.h> #include <C_CkHtmlToXml.h> #include <C_CkStringBuilder.h> #include <C_CkXml.h> #include <C_CkCsv.h> void ChilkatSample(void) { HCkHttp http; HCkBinData bdHtml; BOOL success; HCkHtmlToXml htx; HCkStringBuilder sbXml; HCkXml xml; int numRemoved; int i; int count_i; const char *table_role; const char *table_data_scrollx; const char *table_data_sortdirection; const char *table_data_sorton; const char *table_id; int j; int count_j; const char *tr_role; int k; int count_k; const char *tagPath; const char *text; HCkCsv csv; const char *csvStr; // 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 http = CkHttp_Create(); bdHtml = CkBinData_Create(); success = CkHttp_QuickGetBd(http,"https://example-code.com/data/etf_table.html",bdHtml); if (success != TRUE) { printf("%s\n",CkHttp_lastErrorText(http)); CkHttp_Dispose(http); CkBinData_Dispose(bdHtml); return; } // Convert to XML. htx = CkHtmlToXml_Create(); CkHtmlToXml_SetHtmlBd(htx,bdHtml); sbXml = CkStringBuilder_Create(); CkHtmlToXml_ToXmlSb(htx,sbXml); xml = CkXml_Create(); CkXml_LoadSb(xml,sbXml,TRUE); // Remove attributes and sub-trees we don't need. // (In other words, we're getting rid of clutter...) numRemoved = CkXml_PruneTag(xml,"thead"); numRemoved = CkXml_PruneAttribute(xml,"style"); numRemoved = CkXml_PruneAttribute(xml,"class"); // Scrub the element and attribute content. CkXml_Scrub(xml,"ContentTrimEnds,ContentTrimInside,AttrTrimEnds,AttrTrimInside"); // Let's see what we have... printf("%s\n",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: // i = 0; count_i = CkXml_NumChildrenHavingTag(xml,"html|body|div"); while (i < count_i) { CkXml_putI(xml,i); table_role = CkXml_chilkatPath(xml,"html|body|div[i]|div|table|(role)"); table_data_scrollx = CkXml_chilkatPath(xml,"html|body|div[i]|div|table|(data-scrollx)"); table_data_sortdirection = CkXml_chilkatPath(xml,"html|body|div[i]|div|table|(data-sortdirection)"); table_data_sorton = CkXml_chilkatPath(xml,"html|body|div[i]|div|table|(data-sorton)"); table_id = CkXml_chilkatPath(xml,"html|body|div[i]|table|(id)"); table_role = CkXml_chilkatPath(xml,"html|body|div[i]|table|(role)"); table_data_scrollx = CkXml_chilkatPath(xml,"html|body|div[i]|table|(data-scrollx)"); table_data_sortdirection = CkXml_chilkatPath(xml,"html|body|div[i]|table|(data-sortdirection)"); table_data_sorton = CkXml_chilkatPath(xml,"html|body|div[i]|table|(data-sorton)"); j = 0; count_j = CkXml_NumChildrenHavingTag(xml,"html|body|div[i]|table|tbody|tr"); while (j < count_j) { CkXml_putJ(xml,j); tr_role = CkXml_chilkatPath(xml,"html|body|div[i]|table|tbody|tr[j]|(role)"); k = 0; count_k = CkXml_NumChildrenHavingTag(xml,"html|body|div[i]|table|tbody|tr[j]|td"); while (k < count_k) { CkXml_putK(xml,k); text = CkXml_getChildContent(xml,"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. csv = CkCsv_Create(); 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"); i = 0; count_i = CkXml_NumChildrenHavingTag(xml,"html|body|div"); while (i < count_i) { CkXml_putI(xml,i); j = 0; count_j = CkXml_NumChildrenHavingTag(xml,"html|body|div[i]|table|tbody|tr"); while (j < count_j) { CkXml_putJ(xml,j); k = 0; count_k = CkXml_NumChildrenHavingTag(xml,"html|body|div[i]|table|tbody|tr[j]|td"); while (k < count_k) { CkXml_putK(xml,k); CkCsv_SetCell(csv,j,k,CkXml_getChildContent(xml,"html|body|div[i]|table|tbody|tr[j]|td[k]|text")); k = k + 1; } j = j + 1; } i = i + 1; } CkCsv_SaveFile(csv,"qa_output/brasil_etf.csv"); csvStr = CkCsv_saveToString(csv); printf("%s\n",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" // CkHttp_Dispose(http); CkBinData_Dispose(bdHtml); CkHtmlToXml_Dispose(htx); CkStringBuilder_Dispose(sbXml); CkXml_Dispose(xml); CkCsv_Dispose(csv); } |
|||||||
© 2000-2024 Chilkat Software, Inc. All Rights Reserved.