Невозможно получить доступ к zip/exe из ADLSv2 в синапс

Вы можете воспользоваться приведенным ниже подходом, чтобы извлечь все образцы изображений из ZIP-файла в фрейм данных Pyspark:

Прочтите ZIP-файл, извлеките его во временный путь к файлу и проверьте, загружены ли изображения по этому пути, используя приведенный ниже код:

  # Download the zip file
blob_client = blob_service_client.get_blob_client(container=container_name, blob = "folder1/sample2.zip")
downloaded_blob = blob_client.download_blob().readall()

# Extract the zip file and upload images to ADLS
with zipfile.ZipFile(BytesIO(downloaded_blob)) as z:
    for file_info in z.infolist():
        if file_info.filename.endswith(('.jpg', '.png')):
            print(f"Extracting and uploading {file_info.filename}")
            with z.open(file_info.filename) as file:
                # Define the blob client for each file
                file_name = os.path.basename(file_info.filename)
                blob_client = blob_service_client.get_blob_client(container=container_name, blob=f"folder1/images/{file_name}")
                blob_client.upload_blob(file, overwrite=True)

# Verify if images are uploaded
blobs_list = blob_service_client.get_container_client(container_name).list_blobs(name_starts_with = "folder1/images/")
for blob in blobs_list:
    print(f"Uploaded file: {blob.name}")

Считайте извлеченные изображения в фрейм данных, используя приведенный ниже код:

df_images = spark \
    .read \
    .format("binaryFile") \
    .option("pathGlobFilter", "*.{jpg,png}") \
    .load("abfss://[email protected]/folder1/images/")
display(df_images)

Вы получите изображения в виде фрейма данных, как показано ниже:

Вот полный код для вашей справки:

import zipfile
from azure.storage.blob import BlobServiceClient
from io import BytesIO
from azure.storage.blob import BlobClient
import os

# Initialize connection to your ADLSv2
account_name = "<accountName>"
account_key = "<accountKey>"
container_name = "<containerName>"
blob_service_client = BlobServiceClient(account_url=f"https://{account_name}.blob.core.windows.net", credential=account_key)

# Download the zip file
blob_client = blob_service_client.get_blob_client(container=container_name, blob = "folder1/sample2.zip")
downloaded_blob = blob_client.download_blob().readall()

# Extract the zip file and upload images to ADLS
with zipfile.ZipFile(BytesIO(downloaded_blob)) as z:
    for file_info in z.infolist():
        if file_info.filename.endswith(('.jpg', '.png')):
            print(f"Extracting and uploading {file_info.filename}")
            with z.open(file_info.filename) as file:
                # Define the blob client for each file
                file_name = os.path.basename(file_info.filename)
                blob_client = blob_service_client.get_blob_client(container=container_name, blob=f"folder1/images/{file_name}")
                blob_client.upload_blob(file, overwrite=True)

# Verify if images are uploaded
blobs_list = blob_service_client.get_container_client(container_name).list_blobs(name_starts_with = "folder1/images/")
for blob in blobs_list:
    print(f"Uploaded file: {blob.name}")

# Read the extracted images
df_images = spark \
    .read \
    .format("binaryFile") \
    .option("pathGlobFilter", "*.{jpg,png}") \
    .load("abfss://[email protected]/folder1/images/")

display(df_images)