[ContentUnderstanding] Add ContentRange samples for document, video, and audio.

[changjian-wang]

This pull request introduces a new ContentRange value type for specifying content ranges in analysis requests, and updates both the SDK and sample code to support and demonstrate its usage. The changes improve flexibility for users to restrict analysis to specific document pages or time ranges in audio/video content, and update documentation and sample code to showcase these features.

ContentRange value type introduction and integration

• Added the new ContentRange class in models/_content_range.py, providing methods to construct ranges for document pages and audio/video time intervals, combine multiple ranges, and convert to string. This enables precise specification of content segments to analyze.
• Integrated ContentRange into the SDK: updated begin_analyze_binary and its async variant to accept either a string or a ContentRange object for the content_range parameter, converting it to string as needed. Documentation for these methods was updated to reflect the new parameter type and usage. [1] [2] [3] [4] [5] [6]
• Exported ContentRange in models/_patch.py for public access. [1] [2]

Sample code updates

• Updated sample_analyze_binary_async.py to demonstrate analyzing specific pages and combined page ranges using the new ContentRange class, including example output for these scenarios. [1] [2]
• Updated sample_analyze_url_async.py to show how to restrict analysis to a single page using ContentRange, and changed the document URL to a more complex sample. Also imported timedelta for potential time range usage. [1] [2] [3] [4]

Asset metadata update

• Updated the asset tag in assets.json to reflect the new version, ensuring asset tracking aligns with these changes.

[Copilot]

Pull request overview

This PR introduces a new ContentRange value type for the Azure AI Content Understanding SDK, enabling users to specify content ranges (document pages or audio/video time intervals) when analyzing content. The class provides factory methods for constructing ranges and is integrated into the begin_analyze_binary API.

Changes:

• Added ContentRange class with factory methods (page, pages, pages_from, time_range, time_range_from, combine) and exported it in the models namespace.
• Updated begin_analyze_binary (sync and async) to accept ContentRange objects in addition to raw strings for the content_range parameter.
• Added comprehensive sample code and tests demonstrating ContentRange usage for document, video, and audio analysis scenarios.

Reviewed changes

Copilot reviewed 12 out of 12 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
models/_content_range.py	New ContentRange class with factory methods, equality, hashing, and string conversion
models/_patch.py	Exports ContentRange in __all__
_patch.py	Updates sync begin_analyze_binary to accept ContentRange and convert to string
aio/_patch.py	Updates async begin_analyze_binary to accept ContentRange and convert to string
tests/test_content_range.py	Unit tests for ContentRange construction, validation, equality, and integration with AnalysisInput
tests/samples/test_sample_analyze_url.py	Integration tests for ContentRange with document, video, and audio URL analysis
tests/samples/test_sample_analyze_binary.py	Integration tests for ContentRange with binary document analysis
samples/sample_analyze_url.py	Sync sample showing ContentRange usage for URL-based analysis
samples/async_samples/sample_analyze_url_async.py	Async sample showing ContentRange usage for URL-based analysis
samples/sample_analyze_binary.py	Sync sample showing ContentRange usage for binary analysis
samples/async_samples/sample_analyze_binary_async.py	Async sample showing ContentRange usage for async binary analysis
assets.json	Updated asset tag for new test recordings

[github-actions]

API Change Check

APIView identified API level changes in this PR and created the following API reviews

azure-ai-contentunderstanding

[changjian-wang]

@copilot
Update — 2026-03-13
For Python, the samples should explicitly demonstrate passing the content range as a raw string directly (not via any helper), covering the same scenarios as the .NET/Java typed ContentRange samples:
"1-3" for a document page range
"9-" for all pages from page 9 onward
"0-5000" for an audio/video time window (milliseconds)
"1-3,5,9-" for combined ranges
This ensures Python/JS samples serve as a direct equivalent reference for the same functionality, just expressed as strings instead of typed factory methods.

[Copilot]

@changjian-wang I've opened a new pull request, #45788, to work on those changes. Once the pull request is ready, I'll request review from you.

[changjian-wang]

@copilot check test_sample_analyze_binary.py and test_sample_analyze_url.py, we need more precise assertions.

[Copilot]

@changjian-wang I've opened a new pull request, #45791, to work on those changes. Once the pull request is ready, I'll request review from you.

[changjian-wang]

@copilot analyze audio with contentrange need implement and follow this:

1. "0-5000" — first 5 seconds
2. "10000-" — from 10 seconds onward
3. "1200-3651" — sub-second precision
4. "0-3000,30000-" — combined time ranges

[Copilot]

@changjian-wang I've opened a new pull request, #45792, to work on those changes. Once the pull request is ready, I'll request review from you.

[changjian-wang]

@copilot The assersion in file test_sample_analyze_binary.py is not precise enough as it did not verify the specific pages extracted. We should look at the specific page number from the extracted results. We need use assert equal to verify page count.

[Copilot]

@changjian-wang I've opened a new pull request, #45793, to work on those changes. Once the pull request is ready, I'll request review from you.