root/Jyotisha
1
0
Fork 0
Code Issues 1 Pull Requests 1 Actions Packages Projects Releases Wiki Activity

Harden Jyotish runtime evidence gates

Browse Source
This commit is contained in:
732642856
2026-07-05 18:27:33 +08:00
parent 8520921f40
commit 4764482c73
76 changed files with 5628 additions and 253 deletions
Download Patch File Download Diff File Expand all files Collapse all files
scripts/jyotish_api_server.py
+289 -7
View File
@@ -73,8 +73,10 @@ def execute_consultation_workflow(
)
executed_steps = []
known_steps = [
'run_prashna',
'compute_chart',
'run_rectification_gate',
'run_muhurta_panchanga',
'run_historical_event_backtest',
'run_thematic_report',
]
@@ -89,16 +91,45 @@ def execute_consultation_workflow(
'executed_steps': [],
'skipped_steps': known_steps,
}
result['runtime_evidence_log'] = _UNIFIED_CONSULTATION_ORCHESTRATOR.runtime_evidence_log(
surface=surface,
entry_mode=entry_mode,
route_packet=route_packet,
executed_steps=[],
skipped_steps=known_steps,
blind=bool(body.get('blind') or body.get('blind_technical_mode')),
)
if body.get('return_high_rigor_shape'):
result['endpoint'] = 'high_rigor_workflow'
return result
chart = dict(chart_override) if isinstance(chart_override, dict) else {}
prashna = {}
rectification = {}
muhurta_panchanga = {}
computed_chart = bool(chart)
for step in runtime_planner.get('sync_steps', []):
if step == 'run_rectification_gate':
if step == 'run_prashna':
prashna = handler._compute_prashna({
**birth_payload,
'question': body.get('question', 'general'),
'question_text': body.get('question_text', ''),
'horary_number': body.get('horary_number'),
'planets': body.get('planets', {}),
'asc_degree': body.get('asc_degree', 15.5),
})
executed_steps.append('run_prashna')
elif step == 'run_muhurta_panchanga':
muhurta_panchanga = handler._compute_muhurta_panchanga({
**birth_payload,
'reference_date': body.get('reference_date') or body.get('transit_date') or body.get('today') or body.get('current_date'),
'question': question,
'themes': themes,
'activity': body.get('muhurta_activity'),
})
executed_steps.append('run_muhurta_panchanga')
elif step == 'run_rectification_gate':
chart_planets = chart.get('planets') if isinstance(chart, dict) else {}
chart_ascendant = chart.get('ascendant') if isinstance(chart, dict) else {}
rectification = handler._compute_rectification_gate({
@@ -121,6 +152,8 @@ def execute_consultation_workflow(
executed_steps.append('run_historical_event_backtest')
chart_for_theme = dict(chart) if isinstance(chart, dict) else {}
if entry_mode == 'prashna' and isinstance(prashna, dict):
chart_for_theme.setdefault('prashna', prashna)
if isinstance(chart_for_theme.get('modules'), dict):
chart_for_theme.update(chart_for_theme.get('modules', {}).get('chart') or {})
@@ -128,6 +161,7 @@ def execute_consultation_workflow(
prompt_snapshot = (((chart.get('ai_prompt_pack') or {}).get('evidence_snapshot')) or {}) if isinstance(chart, dict) else {}
strict_workflow_contracts = prompt_snapshot.get('strict_workflow_contracts') if isinstance(prompt_snapshot.get('strict_workflow_contracts'), dict) else {}
chart_guided_topics = modules.get('guided_topics') if isinstance(modules.get('guided_topics'), list) else []
audited_remedies = handler._build_audited_remedies_from_guided_topics(chart_guided_topics)
thematic_report = {}
if 'run_thematic_report' in runtime_planner.get('sync_steps', []):
@@ -149,7 +183,30 @@ def execute_consultation_workflow(
executed_steps.append('run_thematic_report')
vedastro_official = handler._high_rigor_vedastro_official_summary(chart)
runtime_truth = vedastro_official.get('runtime_truth') if isinstance(vedastro_official.get('runtime_truth'), dict) else {}
interpretation_source_runtime_coverage = handler._interpretation_source_runtime_coverage(chart)
skipped_steps = [step for step in known_steps if step not in executed_steps]
machine_evidence_packet = _UNIFIED_CONSULTATION_ORCHESTRATOR.machine_evidence_packet(
chart=chart,
route_packet=route_packet,
vedastro_official=vedastro_official,
)
real_case_calibration = _UNIFIED_CONSULTATION_ORCHESTRATOR.real_case_calibration_catalog(
route_packet=route_packet,
machine_evidence_packet=machine_evidence_packet,
)
runtime_evidence_log = _UNIFIED_CONSULTATION_ORCHESTRATOR.runtime_evidence_log(
surface=surface,
entry_mode=entry_mode,
route_packet=route_packet,
executed_steps=executed_steps,
skipped_steps=skipped_steps,
vedastro_official=vedastro_official,
interpretation_source_runtime_coverage=interpretation_source_runtime_coverage,
machine_evidence_packet=machine_evidence_packet,
real_case_calibration=real_case_calibration,
blind=bool(body.get('blind') or body.get('blind_technical_mode')),
)
result = {
'success': True,
@@ -187,7 +244,15 @@ def execute_consultation_workflow(
'rectification': rectification,
'historical_event_backtest': historical_backtest,
'thematic_report': thematic_report,
'prashna': prashna,
'muhurta_panchanga': muhurta_panchanga,
'audited_remedies': audited_remedies,
'vedastro_official': vedastro_official,
'runtime_truth': runtime_truth,
'interpretation_source_runtime_coverage': interpretation_source_runtime_coverage,
'machine_evidence_packet': machine_evidence_packet,
'real_case_calibration': real_case_calibration,
'runtime_evidence_log': runtime_evidence_log,
'next_questions': handler._high_rigor_next_questions(rectification, historical_backtest),
'boundary': (
'This endpoint composes existing project workflows. It does not claim that every VedAstro callable '
@@ -244,6 +309,15 @@ def _api_chart_cache_ttl_seconds() -> float:
return max(ttl, 0.0)
def _free_tier_queue_enabled_env() -> bool:
raw_values = [
str(os.environ.get("VEDASTRO_FREE_TIER_QUEUE", "")).strip().lower(),
str(os.environ.get("VEDASTRO_FREE_TIER_QUEUE_ENABLED", "")).strip().lower(),
str(os.environ.get("VEDASTRO_ENABLE_FREE_TIER_QUEUE", "")).strip().lower(),
]
return any(value in {"1", "true", "yes", "on"} for value in raw_values)
def _vedastro_runtime_fingerprint() -> dict:
endpoint = os.environ.get('VEDASTRO_API_ENDPOINT', '').strip()
return {
@@ -1623,6 +1697,47 @@ class JyotishAPIHandler(BaseHTTPRequestHandler):
surface=body.get('surface', 'api_web'),
)
def _build_audited_remedies_from_guided_topics(self, guided_topics):
if not isinstance(guided_topics, list):
return {'status': 'blocked', 'reason': 'guided_topics_missing'}
selected_gate = None
selected_topic = None
for topic in guided_topics:
if not isinstance(topic, dict):
continue
gate = topic.get('strict_audit_gate')
if isinstance(gate, dict):
selected_gate = gate
selected_topic = topic
break
if not isinstance(selected_gate, dict):
return {'status': 'blocked', 'reason': 'strict_audit_gate_missing'}
strength_context = selected_gate.get('strength_context') if isinstance(selected_gate.get('strength_context'), dict) else {}
dosha_context = selected_gate.get('dosha_context') if isinstance(selected_gate.get('dosha_context'), list) else []
active_dasha_lord = selected_gate.get('active_dasha_lord') if isinstance(selected_gate.get('active_dasha_lord'), str) else ''
if not strength_context:
return {'status': 'blocked', 'reason': 'strength_context_missing', 'source': 'strict_audit_gate'}
payload = self._compute_remedies({
'shadbala': strength_context,
'doshas': dosha_context,
'dasha_lord': active_dasha_lord,
})
if not isinstance(payload, dict):
payload = {'recommendations': payload}
return {
'status': 'ok',
'source': 'strict_audit_gate',
'topic': (
selected_gate.get('topic')
or selected_topic.get('id')
or selected_topic.get('title')
or 'general'
),
'active_dasha_lord': active_dasha_lord,
'recommendations': payload.get('recommendations') or payload,
'raw': payload,
}
def _compute_high_rigor_workflow_sync(self, body):
body_copy = dict(body or {})
body_copy.pop('async', None)
@@ -1969,6 +2084,8 @@ class JyotishAPIHandler(BaseHTTPRequestHandler):
if not isinstance(prompt_full_snapshot, dict):
prompt_full_snapshot = {}
modules = chart.get('modules') if isinstance(chart, dict) else {}
if not isinstance(modules, dict):
modules = {}
range_scan = modules.get('vedastro_range_scan_result') if isinstance(modules, dict) else {}
if not isinstance(range_scan, dict):
range_scan = {}
@@ -1976,6 +2093,8 @@ class JyotishAPIHandler(BaseHTTPRequestHandler):
official_snapshot = range_scan.get('official_full_snapshot') if isinstance(range_scan, dict) else {}
if not isinstance(official_snapshot, dict):
official_snapshot = {}
if not official_snapshot and isinstance(modules.get('vedastro_official_full_snapshot'), dict):
official_snapshot = modules.get('vedastro_official_full_snapshot') or {}
metadata = official_snapshot.get('source_metadata') if isinstance(official_snapshot, dict) else {}
catalog = metadata.get('official_full_capability_catalog') if isinstance(metadata, dict) else {}
if not isinstance(catalog, dict):
@@ -2024,13 +2143,94 @@ class JyotishAPIHandler(BaseHTTPRequestHandler):
if isinstance(selection, dict) and isinstance(selection.get('report_reference'), dict)
}
)
return {
'status': (
prompt_official.get('status')
or official_snapshot.get('status')
or range_scan.get('status')
or 'blocked'
status = (
prompt_official.get('status')
or official_snapshot.get('status')
or range_scan.get('status')
or 'blocked'
)
chart_core_status = 'blocked'
official_primary_evidence = (
primary_contract.get('official_primary_evidence')
or prompt_official.get('official_primary_evidence')
or {}
)
if not isinstance(official_primary_evidence, dict):
official_primary_evidence = {}
chart_core = official_primary_evidence.get('chart_core')
if isinstance(chart_core, dict) and chart_core.get('status'):
chart_core_status = chart_core.get('status')
elif full_snapshot_payload.get('available'):
chart_core_status = 'ok'
event_radar_status = 'blocked'
if (
prompt_official.get('blocked_items')
or prompt_official.get('fallback_used')
or prompt_official.get('conflicts')
):
event_radar_status = 'partial'
elif range_scan.get('status') == 'ok':
event_radar_status = 'ok'
elif range_scan.get('status'):
event_radar_status = 'partial'
runtime_truth = {
'status': status,
'catalog_boundary': 'catalog_recognized_not_full_runtime_execution',
'primary_route': strict_workflow_primary_route or _selected_route,
'routes_available': strict_workflow_routes_available,
'official_execution_layers': {
'chart_core': chart_core_status,
'event_radar': event_radar_status,
'catalog_status': (
prompt_official.get('official_full_capability_catalog_status')
or catalog.get('status')
or range_metadata.get('official_full_capability_catalog_status')
or official_snapshot.get('status')
or 'blocked'
),
},
'fallback_active': bool(
primary_contract.get('fallback_used')
or prompt_official.get('fallback_used')
),
'blocked_items': (
primary_contract.get('blocked_items')
or prompt_official.get('blocked_items')
or []
),
'conflicts': (
primary_contract.get('conflicts')
or prompt_official.get('conflicts')
or []
),
'free_tier_strategy': {
'using_free_tier': not bool(os.environ.get('VEDASTRO_API_KEY', '').strip()),
'queue_enabled': _free_tier_queue_enabled_env(),
'cache_hit': bool(
(((official_snapshot.get('source_metadata') or {}).get('semantic_cache') or {}).get('cache_hit'))
if isinstance(official_snapshot, dict)
else False
),
'guard_status': (
'degraded_or_partial'
if status in {'partial', 'blocked', 'official_snapshot_budget_exhausted'}
or bool(prompt_official.get('blocked_items'))
else 'within_free_tier_strategy'
),
},
}
raw_response = (
official_snapshot.get('raw_response')
or official_snapshot.get('official_raw_response')
or official_snapshot.get('raw_payload')
or official_snapshot.get('raw')
or prompt_full_snapshot.get('raw_response')
or prompt_full_snapshot.get('official_raw_response')
or prompt_full_snapshot.get('raw_payload')
or prompt_full_snapshot.get('raw')
)
return {
'status': status,
'range_scan_status': range_scan.get('status') if isinstance(range_scan, dict) else None,
'event_count': int(range_scan.get('event_count', 0) or 0) if isinstance(range_scan, dict) else 0,
'official_full_capability_catalog_status': (
@@ -2090,12 +2290,58 @@ class JyotishAPIHandler(BaseHTTPRequestHandler):
'verdict': primary_contract.get('verdict'),
'dominant_label': primary_contract.get('dominant_label'),
'main_conflicts': primary_contract.get('main_conflicts') or primary_contract.get('conflicts') or [],
'runtime_truth': runtime_truth,
'raw_response': raw_response,
'boundary': 'VedAstro official snapshot and capability catalog are consumed as primary evidence metadata; execution breadth depends on configured network and sample limits.',
}
def _interpretation_source_runtime_coverage(self, chart):
modules = chart.get('modules') if isinstance(chart, dict) else {}
if not isinstance(modules, dict):
modules = {}
prompt_pack = chart.get('ai_prompt_pack') if isinstance(chart, dict) else {}
evidence_snapshot = prompt_pack.get('evidence_snapshot') if isinstance(prompt_pack, dict) else {}
interpretation_pack = evidence_snapshot.get('interpretation_source_pack') if isinstance(evidence_snapshot.get('interpretation_source_pack'), dict) else {}
candidates = {
'dasha_timing_layer_used',
'varga_strength_layer_used',
'annual_special_layer_context',
'modifier_obstacle_layer_used',
}
proven_markers = []
guided_topics = modules.get('guided_topics') if isinstance(modules.get('guided_topics'), list) else []
for topic in guided_topics:
if not isinstance(topic, dict):
continue
strict_gate = topic.get('strict_audit_gate')
if not isinstance(strict_gate, dict):
continue
secondary = strict_gate.get('secondary_context')
if not isinstance(secondary, list):
continue
for item in secondary:
if isinstance(item, str) and item in candidates and item not in proven_markers:
proven_markers.append(item)
return {
'source_pack_status': interpretation_pack.get('status') or 'used',
'proven_runtime_markers': proven_markers,
'runtime_visibility_status': 'partial' if proven_markers else 'blocked',
'not_fully_closed': [
'references/open_source_sources/jyotishganit',
'references/open_source_sources/jaimini-tropical',
'references/open_source_sources/VedicAstro',
'references/open_source_sources/rishi-ai-mcp',
'references/open_source_sources/vedic-astro-skills',
'references/open_source_sources/dashaflow',
],
'boundary': 'Inventory/grading exists, but full runtime invocation is only proven for surfaced strict-workflow markers, not every local source asset.',
}
def _high_rigor_next_questions(self, rectification, historical_backtest):
questions = []
summary = rectification.get('summary') if isinstance(rectification, dict) else {}
if not isinstance(summary, dict):
summary = {}
for item in summary.get('recommended_events') or []:
questions.append({
'type': 'yes_no_or_date',
@@ -5424,6 +5670,42 @@ class JyotishAPIHandler(BaseHTTPRequestHandler):
raise BadRequest(str(e)) from e
return {'success': True, 'endpoint': 'panchanga_range', 'report': report}
def _compute_muhurta_panchanga(self, body):
reference_date = body.get('reference_date') or body.get('transit_date') or body.get('today') or datetime.now().strftime('%Y-%m-%d')
if not isinstance(reference_date, str):
raise BadRequest('reference_date must be a string')
date_str = reference_date[:10]
try:
datetime.strptime(date_str, '%Y-%m-%d')
except ValueError as e:
raise BadRequest('reference_date must be YYYY-MM-DD') from e
raw_activity = body.get('activity')
if raw_activity is not None and not isinstance(raw_activity, str):
raise BadRequest('activity must be a string')
activity = (raw_activity or '').strip().lower()
question_text = str(body.get('question') or '')
themes = body.get('themes') if isinstance(body.get('themes'), list) else []
if activity not in {'marriage', 'business', 'travel', 'medical', 'education'}:
if 'marriage' in themes or any(token in question_text for token in ('婚', '恋', 'marry', 'wedding', 'relationship')):
activity = 'marriage'
elif any(token in question_text for token in ('travel', '迁移', '搬家', '出行')):
activity = 'travel'
elif any(token in question_text for token in ('medical', '手术', '治疗', '健康')):
activity = 'medical'
elif any(token in question_text for token in ('education', '学习', '考试', '申请')):
activity = 'education'
else:
activity = 'business'
muhurta = _load_local_module('muhurta')
return muhurta.build_muhurta_sidecar(
date_str=date_str,
activity=activity,
lat=self._get_float(body, 'lat', 0, -90, 90),
lon=self._get_float(body, 'lon', 0, -180, 180),
tz=self._get_float(body, 'tz', 0, -14, 14),
ayanamsa_name=body.get('ayanamsa', 'lahiri'),
)
def _compute_rectification_gate(self, body):
asc_lon = self._asc_lon_from_body(body)
declared_accuracy = body.get('declared_accuracy', body.get('accuracy', 'minute'))