Skip to content

[Sample] get raw json for analyze_return_raw_json samples and tests #11892

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
chienyuanchang wants to merge 2 commits into
base: sdkauto/azure-ai-contentunderstanding-5580124
Choose a base branch
from
Open

[Sample] get raw json for analyze_return_raw_json samples and tests #11892

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
bcfcd51
[Sample] get raw json for analyze_return_raw_json samples and tests
Dec 8, 2025
6cdca09
[Sample] print out markdown for get_raw_json sample
Dec 8, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
101 changes: 83 additions & 18 deletions ...ure-ai-contentunderstanding/samples/async_samples/sample_analyze_return_raw_json_async.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,14 +9,18 @@

DESCRIPTION:
This sample demonstrates how to access the raw JSON response from analysis operations
using protocol methods. This is useful for advanced scenarios where you need direct access
to the JSON structure.
using the 'cls' callback parameter (async version). This is useful for advanced scenarios
where you need direct access to the JSON structure.

The Content Understanding SDK provides two approaches for accessing analysis results:

1. Object model approach (recommended): Returns strongly-typed AnalyzeResult objects
2. Protocol method approach: Returns raw BinaryData containing the JSON response
that are easier to navigate and use. This is shown in sample_analyze_binary_async.py.

2. Protocol method approach: Returns raw HTTP response containing the JSON. This sample
demonstrates this approach for advanced scenarios.

For production use, prefer the object model approach as it provides:
IMPORTANT: For production use, prefer the object model approach as it provides:
- Type safety
- IntelliSense support
- Easier navigation of results
Expand Down Expand Up @@ -66,21 +70,31 @@ async def main() -> None:

print(f"Analyzing {file_path} with prebuilt-documentSearch...")

# Use the standard method which returns an AnalyzeResult
# Then serialize to JSON for raw access
# Use the 'cls' callback parameter to get the raw HTTP response
# The 'cls' parameter allows us to intercept the response and return custom data
# We return a tuple: (deserialized_object, raw_http_response)
# Note: For production use, prefer the object model approach (without cls parameter)
# which returns AnalyzeResult objects that are easier to work with
poller = await client.begin_analyze_binary(
analyzer_id="prebuilt-documentSearch",
binary_input=file_bytes,
content_type="application/pdf",
cls=lambda pipeline_response, deserialized_obj, response_headers: (
deserialized_obj,
pipeline_response.http_response,
),
)
result = await poller.result()

# Convert to dictionary and then to JSON
result_dict = result.as_dict()
# Wait for completion and get both the deserialized object and raw HTTP response
_, raw_http_response = await poller.result()
# [END analyze_return_raw_json]

# [START parse_raw_json]
# Parse the raw JSON response
response_json = raw_http_response.json()

# Pretty-print the JSON
pretty_json = json.dumps(result_dict, indent=2, ensure_ascii=False, default=str)
pretty_json = json.dumps(response_json, indent=2, ensure_ascii=False)

# Create output directory if it doesn't exist
output_dir = Path(__file__).parent.parent / "sample_output"
Expand All @@ -94,17 +108,68 @@ async def main() -> None:
with open(output_path, "w", encoding="utf-8") as f:
f.write(pretty_json)

print(f"\nRaw JSON response saved to: {output_path}")
print(f"Raw JSON response saved to: {output_path}")
print(f"File size: {len(pretty_json):,} characters")

# Show a preview of the JSON structure
print("\nJSON Structure Preview:")
print("=" * 50)
preview = pretty_json[:2000] + "..." if len(pretty_json) > 2000 else pretty_json
print(preview)
print("=" * 50)
# [END parse_raw_json]

# [START extract_from_raw_json]
# Extract key information from raw JSON
Copy link

@yungshinlintw yungshinlintw Dec 8, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can we follow samples/sample_analyze_binary.py to print out markdown and equivalent document structure to show two ways to access the same functionalities?

Copy link
Author

@chienyuanchang chienyuanchang Dec 8, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes, updated

# This demonstrates accessing the same data that would be available via the object model
if "result" in response_json:
result_data = response_json["result"]

if "analyzerId" in result_data:
print(f"\nAnalyzer ID: {result_data['analyzerId']}")

if "contents" in result_data and isinstance(result_data["contents"], list):
print(f"Contents count: {len(result_data['contents'])}")

if len(result_data["contents"]) > 0:
first_content = result_data["contents"][0]

if "kind" in first_content:
print(f"Content kind: {first_content['kind']}")
if "mimeType" in first_content:
print(f"MIME type: {first_content['mimeType']}")

# Extract markdown content from raw JSON
# Object model equivalent: content.markdown
print("\nMarkdown Content (from raw JSON):")
print("=" * 50)
if "markdown" in first_content and first_content["markdown"]:
print(first_content["markdown"])
else:
print("No markdown content available.")
print("=" * 50)

# Extract document properties from raw JSON
# Object model equivalent: document_content.start_page_number, etc.
if first_content.get("kind") == "document":
print("\nDocument Information (from raw JSON):")
if "startPageNumber" in first_content:
print(f" Start page: {first_content['startPageNumber']}")
if "endPageNumber" in first_content:
print(f" End page: {first_content['endPageNumber']}")

start_page = first_content.get("startPageNumber")
end_page = first_content.get("endPageNumber")
if start_page and end_page:
total_pages = end_page - start_page + 1
print(f" Total pages: {total_pages}")

# Extract pages information
# Object model equivalent: document_content.pages
if "pages" in first_content and first_content["pages"]:
pages = first_content["pages"]
unit = first_content.get("unit", "units")
print(f"\nPages ({len(pages)}):")
for page in pages:
page_num = page.get("pageNumber")
width = page.get("width")
height = page.get("height")
print(f" Page {page_num}: {width} x {height} {unit}")
# [END extract_from_raw_json]

if not isinstance(credential, AzureKeyCredential):
await credential.close()

Expand Down
101 changes: 83 additions & 18 deletions ...tentunderstanding/azure-ai-contentunderstanding/samples/sample_analyze_return_raw_json.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,14 +9,18 @@

DESCRIPTION:
This sample demonstrates how to access the raw JSON response from analysis operations
using protocol methods. This is useful for advanced scenarios where you need direct access
to the JSON structure.
using the 'cls' callback parameter. This is useful for advanced scenarios where you need
direct access to the JSON structure.

The Content Understanding SDK provides two approaches for accessing analysis results:

1. Object model approach (recommended): Returns strongly-typed AnalyzeResult objects
2. Protocol method approach: Returns raw BinaryData containing the JSON response
that are easier to navigate and use. This is shown in sample_analyze_binary.py.

2. Protocol method approach: Returns raw HTTP response containing the JSON. This sample
demonstrates this approach for advanced scenarios.

For production use, prefer the object model approach as it provides:
IMPORTANT: For production use, prefer the object model approach as it provides:
- Type safety
- IntelliSense support
- Easier navigation of results
Expand Down Expand Up @@ -66,21 +70,31 @@ def main() -> None:

print(f"Analyzing {file_path} with prebuilt-documentSearch...")

# Use the standard method which returns an AnalyzeResult
# Then serialize to JSON for raw access
# Use the 'cls' callback parameter to get the raw HTTP response
# The 'cls' parameter allows us to intercept the response and return custom data
# We return a tuple: (deserialized_object, raw_http_response)
# Note: For production use, prefer the object model approach (without cls parameter)
# which returns AnalyzeResult objects that are easier to work with
poller = client.begin_analyze_binary(
analyzer_id="prebuilt-documentSearch",
binary_input=file_bytes,
content_type="application/pdf",
cls=lambda pipeline_response, deserialized_obj, response_headers: (
deserialized_obj,
pipeline_response.http_response,
),
)
result = poller.result()

# Convert to dictionary and then to JSON
result_dict = result.as_dict()
# Wait for completion and get both the deserialized object and raw HTTP response
_, raw_http_response = poller.result()
# [END analyze_return_raw_json]

# [START parse_raw_json]
# Parse the raw JSON response
response_json = raw_http_response.json()

# Pretty-print the JSON
pretty_json = json.dumps(result_dict, indent=2, ensure_ascii=False, default=str)
pretty_json = json.dumps(response_json, indent=2, ensure_ascii=False)

# Create output directory if it doesn't exist
output_dir = Path(__file__).parent / "sample_output"
Expand All @@ -94,17 +108,68 @@ def main() -> None:
with open(output_path, "w", encoding="utf-8") as f:
f.write(pretty_json)

print(f"\nRaw JSON response saved to: {output_path}")
print(f"Raw JSON response saved to: {output_path}")
print(f"File size: {len(pretty_json):,} characters")

# Show a preview of the JSON structure
print("\nJSON Structure Preview:")
print("=" * 50)
preview = pretty_json[:2000] + "..." if len(pretty_json) > 2000 else pretty_json
print(preview)
print("=" * 50)
# [END parse_raw_json]

# [START extract_from_raw_json]
# Extract key information from raw JSON
# This demonstrates accessing the same data that would be available via the object model
if "result" in response_json:
result_data = response_json["result"]

if "analyzerId" in result_data:
print(f"\nAnalyzer ID: {result_data['analyzerId']}")

if "contents" in result_data and isinstance(result_data["contents"], list):
print(f"Contents count: {len(result_data['contents'])}")

if len(result_data["contents"]) > 0:
first_content = result_data["contents"][0]

if "kind" in first_content:
print(f"Content kind: {first_content['kind']}")
if "mimeType" in first_content:
print(f"MIME type: {first_content['mimeType']}")

# Extract markdown content from raw JSON
# Object model equivalent: content.markdown
print("\nMarkdown Content (from raw JSON):")
print("=" * 50)
if "markdown" in first_content and first_content["markdown"]:
print(first_content["markdown"])
else:
print("No markdown content available.")
print("=" * 50)

# Extract document properties from raw JSON
# Object model equivalent: document_content.start_page_number, etc.
if first_content.get("kind") == "document":
print("\nDocument Information (from raw JSON):")
if "startPageNumber" in first_content:
print(f" Start page: {first_content['startPageNumber']}")
if "endPageNumber" in first_content:
print(f" End page: {first_content['endPageNumber']}")

start_page = first_content.get("startPageNumber")
end_page = first_content.get("endPageNumber")
if start_page and end_page:
total_pages = end_page - start_page + 1
print(f" Total pages: {total_pages}")

# Extract pages information
# Object model equivalent: document_content.pages
if "pages" in first_content and first_content["pages"]:
pages = first_content["pages"]
unit = first_content.get("unit", "units")
print(f"\nPages ({len(pages)}):")
for page in pages:
page_num = page.get("pageNumber")
width = page.get("width")
height = page.get("height")
print(f" Page {page_num}: {width} x {height} {unit}")
# [END extract_from_raw_json]


if __name__ == "__main__":
main()
Loading