public static List <AssetData> RetrieveAssetData(string blockId, List <Block> blocks)
{
//Create a new TableToolbox with the blocks returned by AWS Textract
TableToolbox tableToolbox = new TableToolbox(blocks);
//Filter the Toolbox to only remember the blocks that are descendants of the Table
tableToolbox.FilterToChildren(blockId);
//Build the lookup dictionary for the relevant blocks
tableToolbox.BuildBlockLookup();
//Build an [x,y] 2-Dimensional array of the Table's Cells
tableToolbox.ConstructTable(blockId);
return(FindAssetsInTable(tableToolbox));
}
private static List <AssetData> FindAssetsInTable(TableToolbox tableToolbox)
{
//Navigate the cells of the table to find column indexes for the information that must be located by matching a header string
int[] assetIndex = tableToolbox.FindIndex(new List <string>()
{
"asset", "holding", "assets", "holdings"
});
int[] quantityIndex = tableToolbox.FindIndex(new List <string>()
{
"quantity", "amount", "#", "quan"
});
//If there is not an index for assets (assetColumnIndex has -1 as x coordinate), return empty list;
if (assetIndex[0] == -1)
{
return(new List <AssetData>());
}
//Else, iterate through each row in the table and add assets to the list
List <AssetData> assets = new List <AssetData>();
for (int row = 0; row < tableToolbox.Table.GetLength(1); row++)
{
//Evaluate if the row contains a valid asset
try
{
//Get all text from the cell at index (x,y) in the assumed quantity column
string quantityString = tableToolbox.GetAllTextFromCell(quantityIndex[0], row);
//Strip quantityString's decimal place (Round down)
quantityString = quantityString.Substring(0, quantityString.IndexOf('.'));
//Attempt to convert the string to an int
var quantity = int.Parse(quantityString);
assets.Add(new AssetData(
tableToolbox.GetAllTextFromCell(assetIndex[0], row),
quantity));
}
catch
{
// ignored exceptions, catch will assume failures to execute above code as indication of a non-valid asset
}
}
return(assets);
}
