public static async Task Run(
[BlobTrigger("searchabledocuments/{name}.{ext}", Connection = "AzureWebJobsStorage")] Stream myBlob,
string name, string ext, TraceWriter log)
{
// Because suffix filters don't work yet - this should take non-pdfs off the todo list
if (ext.ToLower() != "pdf")
{
return;
}
log.Info($"Text Processing beginning for {name} ({myBlob.Length} Bytes)");
log.Info($"Extracting text from the PDF (including OCR");
var pages = iTextPDFHelper.GetPDFPages(myBlob, log, ocrImages: true);
log.Info($"Calling Text Analytics to determine key phrases");
Dictionary <string, int> keyPhrases = await TextAnalyticsHelper.GetKeyPhrases(pages, log);
log.Info($"Uploading document to Azure Search");
foreach (var page in pages)
{
string pageId = HttpServerUtility.UrlTokenEncode(Encoding.UTF8.GetBytes(name + "." + ext + page.Number));
await AzureSearchHelper.UploadToAzureSearch(pageId, name + "." + ext, page.Number, page.KeyPhrases, page.Text, log);
}
}
