/// <summary>
/// Loads a finetune collection into memory for useage.
/// </summary>
/// <param name="collectionName">Name of the custom collection to load</param>
/// <returns>An object describing the custom collection</returns>
public async Task <FinetuneCollection> Load(string collectionName)
{
// If the collection is already ready, don't re-load.
FinetuneCollection collectionInfo = await this.Info(collectionName);
if (collectionInfo.LoadStatus == "ready")
{
return(collectionInfo);
}
// Kick off load job
string jobID = await StartModelLoad(collectionName);
while (true)
{
await Task.Delay(pollDelay);
AsyncJobStatus jobStatus = await GetStatus(jobID);
if (jobStatus == AsyncJobStatus.Success)
{
break;
}
else if (jobStatus == AsyncJobStatus.Failure || jobStatus == AsyncJobStatus.Revoked)
{
throw new IndicoAPIException("Load Failed");
}
}
return(await GetResults <FinetuneCollection>(jobID));
}
static async Task Main(string[] args)
{
const string APIKey = "XXXXXX";
const string APIHost = @"https://api-demo.indico.io";
const string FinetuneModel = "1234_5678_9123456789";
if (args.Length == 0)
{
Console.WriteLine("Please supply a path to a pdf file or directory.");
Environment.Exit(1);
}
List <string> validFiles = BuildFileList(args);
Dictionary <string, string> fileTexts = new Dictionary <string, string>();
if (validFiles.Count == 0)
{
return;
}
Indico sdk = new Indico(APIKey, APIHost);
// Extract the contents of each file and write to a txt file.
// Build a list of extraction tasks
IEnumerable <Task <Tuple <string, string> > > extractTasksQuery =
from filePath in validFiles select ExtractDocument(filePath, sdk);
// Kick off all the tasks in the list
List <Task <Tuple <string, string> > > extractTasks = extractTasksQuery.ToList();
while (extractTasks.Count > 0)
{
Task <Tuple <string, string> > firstFinishedTask = await Task.WhenAny(extractTasks);
Tuple <string, string> extractResult = await firstFinishedTask;
extractTasks.Remove(firstFinishedTask);
// Skip any failed documents
if (extractResult.Item2 != null)
{
fileTexts.Add(extractResult.Item1, extractResult.Item2);
}
}
// Load Finetune model
try
{
Console.WriteLine("Loading Model...");
FinetuneCollection collection = await sdk.finetune.Load(FinetuneModel);
Console.WriteLine("Model Loaded: {0}", JsonConvert.SerializeObject(collection, Formatting.Indented));
}
catch (IndicoAPIException iae)
{
Console.WriteLine("Error loading model for predictions. {0}", iae.Message);
return;
}
try
{
// Get Finetune Extraction Predictions
List <List <FinetuneExtraction> > allExtractions =
await sdk.finetune.PredictAnnotation(FinetuneModel, fileTexts.Values.ToArray());
int extractionCount = 0;
string[] extractedFilenames = fileTexts.Keys.ToArray();
StringBuilder sb = new StringBuilder();
for (var i = 0; i < allExtractions.Count; i++)
{
string fileName = Path.GetFileName(extractedFilenames[i]);
foreach (FinetuneExtraction extraction in allExtractions.ElementAt(i))
{
string label = extraction.Label;
sb.Append(fileName + ",");
sb.Append(FormatCSVCell(extraction.Text) + ",");
sb.Append(label + ",");
sb.Append(extraction.Confidence[label]);
sb.AppendLine();
extractionCount++;
}
}
// Write Results to file
string outputPath = Path.Combine(Directory.GetCurrentDirectory(), "output.csv");
File.WriteAllText(outputPath, sb.ToString());
Console.WriteLine("Finetune Extractions complete. {0} extractions found. Report is available here: {1}", extractionCount, outputPath);
}
catch (IndicoAPIException iae)
{
Console.WriteLine("Error getting predictions for documents. {0}", iae.Message);
return;
}
catch (IOException ioe)
{
Console.WriteLine("Error writing predicitons report. {0}", ioe.Message);
}
}
