Imports Google.Cloud.Vision.V1

Public Class GoogleVisionService
    Private ReadOnly _client As ImageAnnotatorClient

    Public Sub New()
        ' Requires GOOGLE_APPLICATION_CREDENTIALS env var
        ' pointing to a service account JSON key file
        _client = ImageAnnotatorClient.Create()
    End Sub

    Public Function DetectText(imagePath As String) As String
        ' WARNING: Image uploaded to Google Cloud
        Dim image = Image.FromFile(imagePath)
        Dim response = _client.DetectText(image)

        If response.Count > 0 Then
            Return response(0).Description
        End If

        Return String.Empty
    End Function
End Class

Imports Google.Cloud.Vision.V1

Public Class GoogleVisionService
    Private ReadOnly _client As ImageAnnotatorClient

    Public Sub New()
        ' Requires GOOGLE_APPLICATION_CREDENTIALS env var
        ' pointing to a service account JSON key file
        _client = ImageAnnotatorClient.Create()
    End Sub

    Public Function DetectText(imagePath As String) As String
        ' WARNING: Image uploaded to Google Cloud
        Dim image = Image.FromFile(imagePath)
        Dim response = _client.DetectText(image)

        If response.Count > 0 Then
            Return response(0).Description
        End If

        Return String.Empty
    End Function
End Class

Public Sub New()
    ' BEFORE: Requires GOOGLE_APPLICATION_CREDENTIALS environment variable
    ' pointing to a service account JSON key file:
    ' export GOOGLE_APPLICATION_CREDENTIALS="/path/to/service-account.json"
    '
    ' The JSON file contains sensitive data:
    ' - private_key: RSA private key
    ' - client_email: service account email
    ' - project_id: GCP project identifier
    '
    ' Security concerns:
    ' - Key file must never be committed to source control
    ' - Key file must be rotated periodically
    ' - Key file must be protected with file system permissions
    ' - Key file compromise grants API access until revoked

    _client = ImageAnnotatorClient.Create()
End Sub

Public Sub New()
    ' BEFORE: Requires GOOGLE_APPLICATION_CREDENTIALS environment variable
    ' pointing to a service account JSON key file:
    ' export GOOGLE_APPLICATION_CREDENTIALS="/path/to/service-account.json"
    '
    ' The JSON file contains sensitive data:
    ' - private_key: RSA private key
    ' - client_email: service account email
    ' - project_id: GCP project identifier
    '
    ' Security concerns:
    ' - Key file must never be committed to source control
    ' - Key file must be rotated periodically
    ' - Key file must be protected with file system permissions
    ' - Key file compromise grants API access until revoked

    _client = ImageAnnotatorClient.Create()
End Sub

Imports System
Imports System.IO
Imports System.Threading.Tasks

Public Class PdfProcessor
    Private _bucketName As String
    Private _visionClient As VisionClient

    Public Async Function ProcessPdfAsync(pdfPath As String) As Task(Of String)
        ' Step 1: Create storage client
        Dim storageClient = StorageClient.Create()
        Dim objectName = $"ocr-input/{Guid.NewGuid()}.pdf"

        ' Step 2: Upload PDF to GCS (document leaves your infrastructure)
        Using stream = File.OpenRead(pdfPath)
            Await storageClient.UploadObjectAsync(_bucketName, objectName, "application/pdf", stream)
        End Using

        ' Step 3: Build async annotation request
        Dim asyncRequest = New AsyncAnnotateFileRequest With {
            .InputConfig = New InputConfig With {
                .GcsSource = New GcsSource With {.Uri = $"gs://{_bucketName}/{objectName}"},
                .MimeType = "application/pdf"
            },
            .Features = {New Feature With {.Type = Feature.Types.Type.DocumentTextDetection}},
            .OutputConfig = New OutputConfig With {
                .GcsDestination = New GcsDestination With {.Uri = $"gs://{_bucketName}/ocr-output/"},
                .BatchSize = 1
            }
        }

        ' Step 4: Submit and wait for async operation
        Dim operation = Await _visionClient.AsyncBatchAnnotateFilesAsync({asyncRequest})
        Dim completedOperation = Await operation.PollUntilCompletedAsync()

        ' Step 5: Download and parse results from GCS output
        Dim outputUri = completedOperation.Result.Responses(0).OutputConfig.GcsDestination.Uri
        Dim text = Await DownloadAndParseResultsAsync(storageClient, outputUri)

        ' Step 6: Clean up input file from GCS
        Await storageClient.DeleteObjectAsync(_bucketName, objectName)

        Return text
    End Function

    Private Async Function DownloadAndParseResultsAsync(storageClient As StorageClient, outputUri As String) As Task(Of String)
        ' Implementation for downloading and parsing results
        Return Await Task.FromResult(String.Empty)
    End Function
End Class

Public Class StorageClient
    Public Shared Function Create() As StorageClient
        Return New StorageClient()
    End Function

    Public Async Function UploadObjectAsync(bucketName As String, objectName As String, mimeType As String, stream As Stream) As Task
        ' Implementation for uploading object
        Await Task.CompletedTask
    End Function

    Public Async Function DeleteObjectAsync(bucketName As String, objectName As String) As Task
        ' Implementation for deleting object
        Await Task.CompletedTask
    End Function
End Class

Public Class VisionClient
    Public Async Function AsyncBatchAnnotateFilesAsync(requests As AsyncAnnotateFileRequest()) As Task(Of Operation)
        ' Implementation for async batch annotate files
        Return Await Task.FromResult(New Operation())
    End Function
End Class

Public Class Operation
    Public Async Function PollUntilCompletedAsync() As Task(Of OperationResult)
        ' Implementation for polling until completed
        Return Await Task.FromResult(New OperationResult())
    End Function
End Class

Public Class OperationResult
    Public Property Responses As Response()
End Class

Public Class Response
    Public Property OutputConfig As OutputConfig
End Class

Public Class AsyncAnnotateFileRequest
    Public Property InputConfig As InputConfig
    Public Property Features As List(Of Feature)
    Public Property OutputConfig As OutputConfig
End Class

Public Class InputConfig
    Public Property GcsSource As GcsSource
    Public Property MimeType As String
End Class

Public Class GcsSource
    Public Property Uri As String
End Class

Public Class Feature
    Public Property Type As Types.Type

    Public Class Types
        Public Enum Type
            DocumentTextDetection
        End Enum
    End Class
End Class

Public Class OutputConfig
    Public Property GcsDestination As GcsDestination
    Public Property BatchSize As Integer
End Class

Public Class GcsDestination
    Public Property Uri As String
End Class

Imports System
Imports System.IO
Imports System.Threading.Tasks

Public Class PdfProcessor
    Private _bucketName As String
    Private _visionClient As VisionClient

    Public Async Function ProcessPdfAsync(pdfPath As String) As Task(Of String)
        ' Step 1: Create storage client
        Dim storageClient = StorageClient.Create()
        Dim objectName = $"ocr-input/{Guid.NewGuid()}.pdf"

        ' Step 2: Upload PDF to GCS (document leaves your infrastructure)
        Using stream = File.OpenRead(pdfPath)
            Await storageClient.UploadObjectAsync(_bucketName, objectName, "application/pdf", stream)
        End Using

        ' Step 3: Build async annotation request
        Dim asyncRequest = New AsyncAnnotateFileRequest With {
            .InputConfig = New InputConfig With {
                .GcsSource = New GcsSource With {.Uri = $"gs://{_bucketName}/{objectName}"},
                .MimeType = "application/pdf"
            },
            .Features = {New Feature With {.Type = Feature.Types.Type.DocumentTextDetection}},
            .OutputConfig = New OutputConfig With {
                .GcsDestination = New GcsDestination With {.Uri = $"gs://{_bucketName}/ocr-output/"},
                .BatchSize = 1
            }
        }

        ' Step 4: Submit and wait for async operation
        Dim operation = Await _visionClient.AsyncBatchAnnotateFilesAsync({asyncRequest})
        Dim completedOperation = Await operation.PollUntilCompletedAsync()

        ' Step 5: Download and parse results from GCS output
        Dim outputUri = completedOperation.Result.Responses(0).OutputConfig.GcsDestination.Uri
        Dim text = Await DownloadAndParseResultsAsync(storageClient, outputUri)

        ' Step 6: Clean up input file from GCS
        Await storageClient.DeleteObjectAsync(_bucketName, objectName)

        Return text
    End Function

    Private Async Function DownloadAndParseResultsAsync(storageClient As StorageClient, outputUri As String) As Task(Of String)
        ' Implementation for downloading and parsing results
        Return Await Task.FromResult(String.Empty)
    End Function
End Class

Public Class StorageClient
    Public Shared Function Create() As StorageClient
        Return New StorageClient()
    End Function

    Public Async Function UploadObjectAsync(bucketName As String, objectName As String, mimeType As String, stream As Stream) As Task
        ' Implementation for uploading object
        Await Task.CompletedTask
    End Function

    Public Async Function DeleteObjectAsync(bucketName As String, objectName As String) As Task
        ' Implementation for deleting object
        Await Task.CompletedTask
    End Function
End Class

Public Class VisionClient
    Public Async Function AsyncBatchAnnotateFilesAsync(requests As AsyncAnnotateFileRequest()) As Task(Of Operation)
        ' Implementation for async batch annotate files
        Return Await Task.FromResult(New Operation())
    End Function
End Class

Public Class Operation
    Public Async Function PollUntilCompletedAsync() As Task(Of OperationResult)
        ' Implementation for polling until completed
        Return Await Task.FromResult(New OperationResult())
    End Function
End Class

Public Class OperationResult
    Public Property Responses As Response()
End Class

Public Class Response
    Public Property OutputConfig As OutputConfig
End Class

Public Class AsyncAnnotateFileRequest
    Public Property InputConfig As InputConfig
    Public Property Features As List(Of Feature)
    Public Property OutputConfig As OutputConfig
End Class

Public Class InputConfig
    Public Property GcsSource As GcsSource
    Public Property MimeType As String
End Class

Public Class GcsSource
    Public Property Uri As String
End Class

Public Class Feature
    Public Property Type As Types.Type

    Public Class Types
        Public Enum Type
            DocumentTextDetection
        End Enum
    End Class
End Class

Public Class OutputConfig
    Public Property GcsDestination As GcsDestination
    Public Property BatchSize As Integer
End Class

Public Class GcsDestination
    Public Property Uri As String
End Class

Imports System.Drawing

Public Function ExtractDocumentStructure(imagePath As String) As DocumentStructure
    Dim image = Image.FromFile(imagePath)
    Dim annotation = _client.DetectDocumentText(image)

    Dim structure As New DocumentStructure With {
        .FullText = annotation.Text,
        .Pages = New List(Of PageInfo)()
    }

    ' Navigate: Pages -> Blocks -> Paragraphs -> Words -> Symbols
    For Each page In annotation.Pages
        Dim pageInfo As New PageInfo With {
            .Confidence = page.Confidence,
            .Paragraphs = New List(Of ParagraphInfo)()
        }

        For Each block In page.Blocks
            For Each paragraph In block.Paragraphs
                ' Must concatenate symbols to get paragraph text
                Dim text = String.Join("", paragraph.Words _
                    .SelectMany(Function(w) w.Symbols) _
                    .Select(Function(s) s.Text))

                pageInfo.Paragraphs.Add(New ParagraphInfo With {
                    .Text = text,
                    .Confidence = paragraph.Confidence
                })
            Next
        Next
        structure.Pages.Add(pageInfo)
    Next

    Return structure
End Function

Imports System.Drawing

Public Function ExtractDocumentStructure(imagePath As String) As DocumentStructure
    Dim image = Image.FromFile(imagePath)
    Dim annotation = _client.DetectDocumentText(image)

    Dim structure As New DocumentStructure With {
        .FullText = annotation.Text,
        .Pages = New List(Of PageInfo)()
    }

    ' Navigate: Pages -> Blocks -> Paragraphs -> Words -> Symbols
    For Each page In annotation.Pages
        Dim pageInfo As New PageInfo With {
            .Confidence = page.Confidence,
            .Paragraphs = New List(Of ParagraphInfo)()
        }

        For Each block In page.Blocks
            For Each paragraph In block.Paragraphs
                ' Must concatenate symbols to get paragraph text
                Dim text = String.Join("", paragraph.Words _
                    .SelectMany(Function(w) w.Symbols) _
                    .Select(Function(s) s.Text))

                pageInfo.Paragraphs.Add(New ParagraphInfo With {
                    .Text = text,
                    .Confidence = paragraph.Confidence
                })
            Next
        Next
        structure.Pages.Add(pageInfo)
    Next

    Return structure
End Function

Public Function ExtractDocumentStructure(imagePath As String) As DocumentStructure
    Dim result = _ocr.Read(imagePath)

    ' Direct access - no symbol concatenation, no nested loops
    Return New DocumentStructure With {
        .FullText = result.Text,
        .Confidence = result.Confidence,
        .Paragraphs = result.Paragraphs.Select(Function(p) New ParagraphInfo With {
            .Text = p.Text,
            .Confidence = p.Confidence
        }).ToList(),
        .Lines = result.Lines.Select(Function(l) New LineInfo With {
            .Text = l.Text,
            .X = l.X,
            .Y = l.Y
        }).ToList()
    }
End Function

Public Function ExtractDocumentStructure(imagePath As String) As DocumentStructure
    Dim result = _ocr.Read(imagePath)

    ' Direct access - no symbol concatenation, no nested loops
    Return New DocumentStructure With {
        .FullText = result.Text,
        .Confidence = result.Confidence,
        .Paragraphs = result.Paragraphs.Select(Function(p) New ParagraphInfo With {
            .Text = p.Text,
            .Confidence = p.Confidence
        }).ToList(),
        .Lines = result.Lines.Select(Function(l) New LineInfo With {
            .Text = l.Text,
            .X = l.X,
            .Y = l.Y
        }).ToList()
    }
End Function

Imports System.Threading.Tasks

Public Async Function ProcessPdfAsync(pdfPath As String) As Task(Of String)
    Return Await Task.Run(Function()
                              Using input As New OcrInput()
                                  input.LoadPdf(pdfPath)
                                  Return _ocr.Read(input).Text
                              End Using
                          End Function)
End Function

Imports System.Threading.Tasks

Public Async Function ProcessPdfAsync(pdfPath As String) As Task(Of String)
    Return Await Task.Run(Function()
                              Using input As New OcrInput()
                                  input.LoadPdf(pdfPath)
                                  Return _ocr.Read(input).Text
                              End Using
                          End Function)
End Function