opencode/NSCT---Neutral-Search-Crawler-Tool
Public Access
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(stage11): audio integration — STT for interviews, podcasts, press conferences with timestamped claims

Browse Source
This commit is contained in:
NSCT Agent
2026-08-25 15:31:03 +00:00
parent 3ab875d2bc
commit 10a083aa02
8 changed files with 2832 additions and 65 deletions
Download Patch File Download Diff File Expand all files Collapse all files

182
tests/models/test_vision.py
View File

@@ -84,7 +84,7 @@ class TestVisionCaptureSchema:
"""Tests für VisionCaptureSchema — Pflichtfelder, Defaults, Frozen."""
@pytest.fixture
def valid_kwargs(self):
def base_kwargs(self):
return {
"capture_type": VisionCaptureType.DIAGRAM,
"image_data_url": "data:image/png;base64,abc123",
@@ -93,8 +93,8 @@ class TestVisionCaptureSchema:
"source_url": "https://example.com/chart.png",
}
def test_create_valid_schema(self, valid_kwargs):
schema = VisionCaptureSchema(**valid_kwargs)
def test_create_valid_schema(self, base_kwargs):
schema = VisionCaptureSchema(**base_kwargs)
assert schema.capture_type == VisionCaptureType.DIAGRAM
assert schema.extracted_text == "This is a chart showing revenue growth."
assert schema.source_id == "source-uuid-001"
@@ -104,54 +104,80 @@ class TestVisionCaptureSchema:
assert schema.evidence_level == EvidenceLevel.MEDIUM
assert schema.metadata == {}
def test_defaults(self, valid_kwargs):
schema = VisionCaptureSchema(**valid_kwargs)
def test_defaults(self, base_kwargs):
schema = VisionCaptureSchema(**base_kwargs)
assert schema.entities == []
assert schema.confidence == 0.5
assert schema.confidence_label == VisionConfidence.MEDIUM
assert schema.evidence_level == EvidenceLevel.MEDIUM
assert schema.metadata == {}
def test_frozen(self, valid_kwargs):
schema = VisionCaptureSchema(**valid_kwargs)
def test_frozen(self, base_kwargs):
schema = VisionCaptureSchema(**base_kwargs)
with pytest.raises(Exception):
schema.capture_type = VisionCaptureType.CHART
def test_missing_required_field(self, valid_kwargs):
kwargs = {**valid_kwargs}
del kwargs["extracted_text"]
def test_missing_extracted_text(self, base_kwargs):
with pytest.raises(ValidationError):
VisionCaptureSchema(**kwargs)
VisionCaptureSchema(
capture_type=base_kwargs["capture_type"],
image_data_url=base_kwargs["image_data_url"],
source_id=base_kwargs["source_id"],
source_url=base_kwargs["source_url"],
)
def test_empty_extracted_text(self, valid_kwargs):
kwargs = {**valid_kwargs, "extracted_text": " "}
def test_empty_extracted_text(self, base_kwargs):
with pytest.raises(ValidationError, match="extracted_text darf nicht nur aus Whitespaces bestehen"):
VisionCaptureSchema(**kwargs)
VisionCaptureSchema(
capture_type=base_kwargs["capture_type"],
image_data_url=base_kwargs["image_data_url"],
extracted_text=" ",
source_id=base_kwargs["source_id"],
source_url=base_kwargs["source_url"],
)
def test_empty_source_url(self, valid_kwargs):
kwargs = {**valid_kwargs, "source_url": " "}
def test_empty_source_url(self, base_kwargs):
with pytest.raises(ValidationError, match="source_url darf nicht leer sein"):
VisionCaptureSchema(**kwargs)
VisionCaptureSchema(
capture_type=base_kwargs["capture_type"],
image_data_url=base_kwargs["image_data_url"],
extracted_text=base_kwargs["extracted_text"],
source_id=base_kwargs["source_id"],
source_url=" ",
)
def test_empty_image_data_url(self, valid_kwargs):
kwargs = {**valid_kwargs, "image_data_url": " "}
def test_empty_image_data_url(self, base_kwargs):
with pytest.raises(ValidationError, match="image_data_url darf nicht leer sein"):
VisionCaptureSchema(**kwargs)
VisionCaptureSchema(
capture_type=base_kwargs["capture_type"],
image_data_url=" ",
extracted_text=base_kwargs["extracted_text"],
source_id=base_kwargs["source_id"],
source_url=base_kwargs["source_url"],
)
def test_confidence_bounds(self, valid_kwargs):
schema_low = VisionCaptureSchema(**valid_kwargs, confidence=0.0)
def test_confidence_bounds(self, base_kwargs):
schema_low = VisionCaptureSchema(
**base_kwargs, confidence=0.0
)
assert schema_low.confidence == 0.0
schema_high = VisionCaptureSchema(**valid_kwargs, confidence=1.0)
schema_high = VisionCaptureSchema(
**base_kwargs, confidence=1.0
)
assert schema_high.confidence == 1.0
def test_confidence_out_of_bounds_low(self, valid_kwargs):
def test_confidence_out_of_bounds_low(self, base_kwargs):
with pytest.raises(ValidationError):
VisionCaptureSchema(**valid_kwargs, confidence=-0.1)
VisionCaptureSchema(
**base_kwargs, confidence=-0.1
)
def test_confidence_out_of_bounds_high(self, valid_kwargs):
def test_confidence_out_of_bounds_high(self, base_kwargs):
with pytest.raises(ValidationError):
VisionCaptureSchema(**valid_kwargs, confidence=1.1)
VisionCaptureSchema(
**base_kwargs, confidence=1.1
)
def test_all_capture_types(self):
for ct in VisionCaptureType:
@@ -188,9 +214,9 @@ class TestVisionCaptureSchema:
)
assert schema.confidence_label == label
def test_entities_list(self, valid_kwargs):
def test_entities_list(self, base_kwargs):
schema = VisionCaptureSchema(
**valid_kwargs,
**base_kwargs,
entities=[
{"type": "NUMBER", "value": "42", "confidence": 0.9},
{"type": "DATE", "value": "2024-01-15", "confidence": 0.8},
@@ -200,9 +226,9 @@ class TestVisionCaptureSchema:
assert schema.entities[0]["type"] == "NUMBER"
assert schema.entities[1]["value"] == "2024-01-15"
def test_metadata_dict(self, valid_kwargs):
def test_metadata_dict(self, base_kwargs):
schema = VisionCaptureSchema(
**valid_kwargs,
**base_kwargs,
metadata={"model": "qwen2.5-vl-3b", "processing_time": 2.3},
)
assert schema.metadata["model"] == "qwen2.5-vl-3b"
@@ -218,37 +244,40 @@ class TestVisionReportSchema:
"""Tests für VisionReportSchema — Zusammenfassung aller visuellen Evidenzen."""
@pytest.fixture
def valid_kwargs(self):
def base_kwargs(self):
return {
"research_run_id": "run-uuid-001",
"total_captures": 3,
}
def test_create_valid_report(self, valid_kwargs):
report = VisionReportSchema(**valid_kwargs)
def test_create_valid_report(self, base_kwargs):
report = VisionReportSchema(**base_kwargs)
assert report.research_run_id == "run-uuid-001"
assert report.total_captures == 3
assert report.captures == []
assert report.entity_summary == {}
assert report.summary_text == ""
def test_frozen(self, valid_kwargs):
report = VisionReportSchema(**valid_kwargs)
def test_frozen(self, base_kwargs):
report = VisionReportSchema(**base_kwargs)
with pytest.raises(Exception):
report.research_run_id = "new-id"
def test_empty_research_run_id(self, valid_kwargs):
def test_empty_research_run_id(self):
with pytest.raises(ValidationError, match="research_run_id darf nicht leer sein"):
VisionReportSchema(
**valid_kwargs,
research_run_id=" ",
total_captures=3,
)
def test_negative_total_captures(self, valid_kwargs):
def test_negative_total_captures(self):
with pytest.raises(ValidationError):
VisionReportSchema(**valid_kwargs, total_captures=-1)
VisionReportSchema(
research_run_id="run-uuid-001",
total_captures=-1,
)
def test_with_captures(self, valid_kwargs):
def test_with_captures(self):
capture = VisionCaptureSchema(
capture_type=VisionCaptureType.CHART,
image_data_url="data:image/png;base64,xyz",
@@ -257,17 +286,18 @@ class TestVisionReportSchema:
source_url="https://example.com",
)
report = VisionReportSchema(
**valid_kwargs,
research_run_id="run-uuid-001",
total_captures=1,
captures=[capture],
)
assert len(report.captures) == 1
assert report.captures[0].capture_type == VisionCaptureType.CHART
def test_political_summary_rejected(self, valid_kwargs):
def test_political_summary_rejected(self):
with pytest.raises(ValidationError, match="politische Empfehlung"):
VisionReportSchema(
**valid_kwargs,
research_run_id="run-uuid-001",
total_captures=0,
summary_text="Die Regierung sollte handeln.",
)
@@ -281,7 +311,7 @@ class TestVisionRequestSchema:
"""Tests für VisionRequestSchema — API-Request."""
@pytest.fixture
def valid_kwargs(self):
def base_kwargs(self):
return {
"research_run_id": "run-uuid-001",
"source_id": "source-uuid-001",
@@ -289,48 +319,74 @@ class TestVisionRequestSchema:
"image_data": "data:image/png;base64,iVBORw0KGgoAAA==",
}
def test_create_valid_request(self, valid_kwargs):
request = VisionRequestSchema(**valid_kwargs)
def test_create_valid_request(self, base_kwargs):
request = VisionRequestSchema(**base_kwargs)
assert request.research_run_id == "run-uuid-001"
assert request.source_id == "source-uuid-001"
assert request.source_url == "https://example.com/image.png"
assert request.capture_type == VisionCaptureType.RAW_IMAGE
assert request.prompt is None
def test_frozen(self, valid_kwargs):
request = VisionRequestSchema(**valid_kwargs)
def test_frozen(self, base_kwargs):
request = VisionRequestSchema(**base_kwargs)
with pytest.raises(Exception):
request.research_run_id = "new-id"
def test_defaults(self, valid_kwargs):
request = VisionRequestSchema(**valid_kwargs)
def test_defaults(self, base_kwargs):
request = VisionRequestSchema(**base_kwargs)
assert request.capture_type == VisionCaptureType.RAW_IMAGE
assert request.prompt is None
def test_with_prompt(self, valid_kwargs):
def test_with_prompt(self, base_kwargs):
request = VisionRequestSchema(
**valid_kwargs,
**base_kwargs,
prompt="Extrahiere alle Zahlen und Daten aus dem Diagramm.",
)
assert request.prompt == "Extrahiere alle Zahlen und Daten aus dem Diagramm."
def test_empty_image_data(self, valid_kwargs):
def test_empty_image_data(self):
with pytest.raises(ValidationError, match="image_data darf nicht leer sein"):
VisionRequestSchema(**valid_kwargs, image_data=" ")
VisionRequestSchema(
research_run_id="run-uuid-001",
source_id="source-uuid-001",
source_url="https://example.com/image.png",
image_data=" ",
)
def test_empty_source_url(self, valid_kwargs):
def test_empty_source_url(self):
with pytest.raises(ValidationError, match="source_url darf nicht leer sein"):
VisionRequestSchema(**valid_kwargs, source_url=" ")
VisionRequestSchema(
research_run_id="run-uuid-001",
source_id="source-uuid-001",
source_url=" ",
image_data="data:image/png;base64,abc",
)
def test_empty_research_run_id(self, valid_kwargs):
def test_empty_research_run_id(self):
with pytest.raises(ValidationError):
VisionRequestSchema(**valid_kwargs, research_run_id=" ")
VisionRequestSchema(
research_run_id=" ",
source_id="source-uuid-001",
source_url="https://example.com/image.png",
image_data="data:image/png;base64,abc",
)
def test_empty_source_id(self, valid_kwargs):
def test_empty_source_id(self):
with pytest.raises(ValidationError):
VisionRequestSchema(**valid_kwargs, source_id=" ")
VisionRequestSchema(
research_run_id="run-uuid-001",
source_id=" ",
source_url="https://example.com/image.png",
image_data="data:image/png;base64,abc",
)
def test_capture_type_override(self, valid_kwargs):
def test_capture_type_override(self):
for ct in VisionCaptureType:
request = VisionRequestSchema(**valid_kwargs, capture_type=ct)
request = VisionRequestSchema(
research_run_id="run-uuid-001",
source_id="source-uuid-001",
source_url="https://example.com/image.png",
image_data="data:image/png;base64,abc",
capture_type=ct,
)
assert request.capture_type == ct