/// <summary>
/// This method retrives all the links on a page (or in a frame,
/// in the case of Webtender.com).
/// </summary>
/// <param name="sourceCode">The raw HTML source code from getSourceCode().</param>
/// <returns>string</returns>
public static string[] getLinksToScrape(string sourceCode)
{
string snipOfSourceCode = "";
string linkRoot = "http://webtender.com";
string links = "";
string[] snipValue = {};
int numLinks = TextTool.CountStringOccurrences(sourceCode, "<LI>");
snipOfSourceCode = textFollowing(sourceCode, "<UL>");
snipValue = snipOfSourceCode.Split('T');
// Begins separating the link text.
snipOfSourceCode = textFollowing(sourceCode, "<UL>");
snipValue = Regex.Split(snipOfSourceCode, "</A>");
// Builds each link into the snipValue array.
for (int count = 0; count < numLinks; count++)
{
string[] temp = {};
temp = Regex.Split(snipValue[count], "\"");
snipValue[count] = linkRoot + temp[1];
}
// DEBUG: The following loop is for testing purposes.
for (int count = 0; count < numLinks; count++)
{
links += snipValue[count];
}
return(snipValue);
}
/// <summary>
/// Carves out the recipe information: title, measurement of each ingredient,
/// name of each ingredient, mixing instructions, and the "creator's comments",
/// referred to in this method as "comments".
/// </summary>
/// <param name="sourceCode">This parameter comes from the getSourceCode() method
/// and provides the raw HTML page source.</param>
/// <returns>string[]</returns>
public static string getRecipeDetails(string sourceCode)
{
string testString;
string snipOfSourceCode;
string smallSnip;
string[] snipValue;
string[] recipeDetails = {};
string recipeSQLstatement = "INSERT INTO TestTable (recipeName, meas, ingrName, mixInstructions) VALUES (\"";
// Gets recipe title
try
{
// Gets recipe title.
snipOfSourceCode = textFollowing(sourceCode, "<H1>");
snipValue = snipOfSourceCode.Split('<');
testString = recipeSQLstatement += snipValue[0] + "\", ";
// Gets measurement of first ingredient
smallSnip = textFollowing(sourceCode, "<LI>");
snipValue = Regex.Split(smallSnip, " <A");
recipeSQLstatement += "\"" + snipValue[0] + "\", ";
// Gets name of first ingredient
snipOfSourceCode = textFollowing(snipValue[1], ">");
snipValue = Regex.Split(snipOfSourceCode, "</A");
recipeSQLstatement += "\"" + snipValue[0] + "\", ";
// Gets mixing instructions
snipOfSourceCode = textFollowing(smallSnip, "<P>");
snipValue = Regex.Split(snipOfSourceCode, "</P>");
recipeSQLstatement += "\"" + snipValue[0] + "\"";
recipeSQLstatement += ");";
}
catch
{
MessageBox.Show("It appears that you are scraping a links page \nwithout checking the box.",
"Error",
MessageBoxButtons.OK,
MessageBoxIcon.Exclamation,
MessageBoxDefaultButton.Button1);
}
// Gets measurements and ingredients
int numIngredients = TextTool.CountStringOccurrences(sourceCode, "<LI>");
//numIngredients -= 5;
numIngredients = 1; // DEBUG: Testing on a non-relational table
for (int counter = 0; counter < numIngredients; counter++)
{
}
return(recipeSQLstatement);
}
