REVAI-4573: Update SDKs to support new model

[Greyvend]

Description of Work

Migrates the Python SDK's speech-to-text submit_job_url to use source_config instead of media_url, enabling support for the v3 transcriber (machine_v3). Also fixes the verbatim option to be sent when explicitly set to False (previously skipped due to a falsy check).

What Changed

• src/rev_ai/apiclient.py — _create_job_options_payload:
  • media_url param now builds source_config: {url: media_url} in the payload instead of sending media_url directly
  • Added conflict validation: raises ValueError if both media_url and source_config are provided
  • Fixed if verbatim: → if verbatim is not None: so verbatim=False is correctly sent to the API
• Tests (TDD approach — tests committed first, verified failing, then implementation committed):
  • Updated existing assertions from media_url to source_config in job, summarization, and translation tests
  • Added new tests: media_url/source_config conflict, verbatim=True, verbatim=False, verbatim=None

Screenshots

Integration test script (test_source_config_migration.py)

"""
REVAI-4573: Integration test script for source_config migration.

Demonstrates that submit_job_url now sends source_config instead of media_url,
enabling v3 transcriber support. Also verifies verbatim=False is sent correctly
and that media_url + source_config conflict raises ValueError.
"""

import sys
sys.path.insert(0, 'src')

from rev_ai import apiclient
from rev_ai.models.customer_url_data import CustomerUrlData

LOCAL_API = "http://localhost:5000"
TOKEN = "<your_token>"
MEDIA_URL = "https://www.rev.ai/FTC_Sample_1.mp3"


def test_submit_job_url_sends_source_config():
    """Test 1: submit_job_url(media_url=...) sends source_config in payload."""
    print("=" * 60)
    print("TEST 1: submit_job_url sends source_config instead of media_url")
    print("=" * 60)
    client = apiclient.RevAiAPIClient(TOKEN, url=LOCAL_API)
    try:
        job = client.submit_job_url(
            media_url=MEDIA_URL,
            transcriber='machine_v3',
            language='en'
        )
        print(f"  Job ID:     {job.id}")
        print(f"  Status:     {job.status}")
        print(f"  Created:    {job.created_on}")
        print("  RESULT: PASS")
    except Exception as e:
        print(f"  Response/Error: {e}")
        print("  RESULT: FAIL (see error above)")
    print()


def test_submit_job_url_with_source_config_object():
    """Test 2: submit_job_url(source_config=CustomerUrlData(...)) still works."""
    print("=" * 60)
    print("TEST 2: submit_job_url with source_config object directly")
    print("=" * 60)
    client = apiclient.RevAiAPIClient(TOKEN, url=LOCAL_API)
    try:
        job = client.submit_job_url(
            source_config=CustomerUrlData(
                url=MEDIA_URL,
                auth_headers={"Authorization": "Bearer media_token"}
            ),
            transcriber='machine_v3',
            language='en'
        )
        print(f"  Job ID:     {job.id}")
        print(f"  Status:     {job.status}")
        print(f"  Created:    {job.created_on}")
        print("  RESULT: PASS")
    except Exception as e:
        print(f"  Response/Error: {e}")
        print("  RESULT: FAIL (see error above)")
    print()


def test_media_url_and_source_config_conflict():
    """Test 3: Passing both media_url and source_config raises ValueError."""
    print("=" * 60)
    print("TEST 3: media_url + source_config conflict raises ValueError")
    print("=" * 60)
    client = apiclient.RevAiAPIClient(TOKEN, url=LOCAL_API)
    try:
        client.submit_job_url(
            media_url=MEDIA_URL,
            source_config=CustomerUrlData(url=MEDIA_URL)
        )
        print("  RESULT: FAIL (no error raised)")
    except ValueError as e:
        print(f"  ValueError: {e}")
        print("  RESULT: PASS")
    except Exception as e:
        print(f"  Unexpected error: {e}")
        print("  RESULT: FAIL")
    print()


def test_verbatim_false_is_sent():
    """Test 4: verbatim=False is included in the payload."""
    print("=" * 60)
    print("TEST 4: verbatim=False is sent in payload")
    print("=" * 60)
    client = apiclient.RevAiAPIClient(TOKEN, url=LOCAL_API)
    try:
        job = client.submit_job_url(
            media_url=MEDIA_URL,
            transcriber='machine_v3',
            verbatim=False,
            language='en'
        )
        print(f"  Job ID:     {job.id}")
        print(f"  Status:     {job.status}")
        print(f"  Verbatim:   {job.verbatim}")
        print("  RESULT: PASS")
    except Exception as e:
        print(f"  Response/Error: {e}")
        print("  RESULT: FAIL (see error above)")
    print()


if __name__ == '__main__':
    print(f"Rev AI Python SDK — source_config migration test")
    print(f"API endpoint: {LOCAL_API}")
    print()
    test_submit_job_url_sends_source_config()
    test_submit_job_url_with_source_config_object()
    test_media_url_and_source_config_conflict()
    test_verbatim_false_is_sent()
    print("All tests completed.")

Test run output (against local API on localhost:5000)

Rev AI Python SDK — source_config migration test
API endpoint: http://localhost:5000

============================================================
TEST 1: submit_job_url sends source_config instead of media_url
============================================================
  Job ID:     NhZpnf4hLCQV
  Status:     in_progress
  Created:    2026-02-19T13:12:09.322Z
  RESULT: PASS

============================================================
TEST 2: submit_job_url with source_config object directly
============================================================
  Job ID:     gUDieEdBoHeo
  Status:     in_progress
  Created:    2026-02-19T13:12:09.338Z
  RESULT: PASS

============================================================
TEST 3: media_url + source_config conflict raises ValueError
============================================================
  ValueError: media_url is not compatible with source_config. Use source_config for all URL-based submissions.
  RESULT: PASS

============================================================
TEST 4: verbatim=False is sent in payload
============================================================
  Job ID:     QGgphget9T0I
  Status:     in_progress
  Verbatim:   False
  RESULT: PASS

All tests completed.

Additional Comments

Changes are scoped to the speech-to-text API client only (RevAiAPIClient). The GenericApiClient used by language identification, sentiment analysis, and topic extraction clients is untouched.

[Greyvend]

@dmtrrk the CI builds are failing. There is some dependency problem with the setup. Should we address that as well?

[dmtrrk]

@dmtrrk the CI builds are failing. There is some dependency problem with the setup. Should we address that as well?

yes, looks like this should be a good time to update the workflow. let's just call pytest direwctly, e.g.

- name: Test
  run: python -m pytest -q

[dmtrrk]

it looks like we need to make this fix to all boolean options