/// <summary>
/// Invoke the BHL "NameSearch" API to determine if the given entity is a scientific name
/// recognized by BHL.
/// </summary>
/// <param name="entity"></param>
/// <returns>True if the entity is a scientific name, False otherwise.</returns>
private static bool IsSciName(EntityRecord entity)
{
bool found = false;
// In the Azure output, Scientific Names are categorized as "Other".
if (entity.Type.ToLower() == "other")
{
// Use the BHL Api to search for the name in BHL
string nameSearchResponse = BHLApi3.NameSearch(entity.Name, BHLApi3.ResponseFormat.Xml, Config.BhlApiKey);
// The response will contain a NameConfirmed element that starts with the search string
// if the name was found. The following is an ugly brute-force evaluation of the XML
// API response.
found = nameSearchResponse.Contains("<NameConfirmed>" + entity.Name);
}
return(found);
}
/// <summary>
/// Download the text to be analyzed.
/// </summary>
private static void DownloadText()
{
// Create the input directory if it does not exist, or clear it if it does
if (Directory.Exists(Config.InputFolder))
{
string[] files = Directory.GetFiles(Config.InputFolder);
foreach (string file in files)
{
File.Delete(file);
}
}
else
{
Directory.CreateDirectory(Config.InputFolder);
}
// Get the item text from the BHL API
string itemMetadataResponse =
BHLApi3.GetItemMetadata(_itemID, true, true, false, BHLApi3.ResponseFormat.Xml, Config.BhlApiKey);
// Extract the text from the API response
XDocument xml = XDocument.Parse(itemMetadataResponse);
foreach (XElement page in xml.Root
.Elements("Result")
.Elements("Item")
.Elements("Pages")
.Elements("Page"))
{
string pageID = page.Element("PageID").Value;
string pageText = page.Element("OcrText").Value;
// Write the text of each page to a file
File.WriteAllText(string.Format("{0}\\{1}.txt", Config.InputFolder, pageID),
pageText, System.Text.Encoding.UTF8);
}
}
