MicroFish/backend/app/api/report.py

1024 lines
29 KiB
Python

"""
Report API routes.
Exposes endpoints for generating simulation reports, fetching them and chatting
with the report agent.
"""
import os
import traceback
import threading
from flask import request, jsonify, send_file
from . import report_bp
from ..config import Config
from ..services.report_agent import ReportAgent, ReportManager, ReportStatus
from ..services.simulation_manager import SimulationManager
from ..models.project import ProjectManager
from ..models.task import TaskManager, TaskStatus
from ..utils.logger import get_logger
from ..utils.locale import t, get_locale, set_locale
logger = get_logger('mirofish.api.report')
# ============== Report generation endpoint ==============
@report_bp.route('/generate', methods=['POST'])
def generate_report():
"""
Generate a simulation analysis report (async task).
This is a long-running operation; the endpoint returns a task_id immediately
and progress can be queried via GET /api/report/generate/status.
Request (JSON):
{
"simulation_id": "sim_xxxx", // required, simulation ID
"force_regenerate": false // optional, force regeneration
}
Returns:
{
"success": true,
"data": {
"simulation_id": "sim_xxxx",
"task_id": "task_xxxx",
"status": "generating",
"message": "Report generation task started"
}
}
"""
try:
data = request.get_json() or {}
simulation_id = data.get('simulation_id')
if not simulation_id:
return jsonify({
"success": False,
"error": t('api.requireSimulationId')
}), 400
force_regenerate = data.get('force_regenerate', False)
# Look up the simulation
manager = SimulationManager()
state = manager.get_simulation(simulation_id)
if not state:
return jsonify({
"success": False,
"error": t('api.simulationNotFound', id=simulation_id)
}), 404
# Check whether a report already exists
if not force_regenerate:
existing_report = ReportManager.get_report_by_simulation(simulation_id)
if existing_report and existing_report.status == ReportStatus.COMPLETED:
return jsonify({
"success": True,
"data": {
"simulation_id": simulation_id,
"report_id": existing_report.report_id,
"status": "completed",
"message": t('api.reportAlreadyExists'),
"already_generated": True
}
})
# Look up the project
project = ProjectManager.get_project(state.project_id)
if not project:
return jsonify({
"success": False,
"error": t('api.projectNotFound', id=state.project_id)
}), 404
graph_id = state.graph_id or project.graph_id
if not graph_id:
return jsonify({
"success": False,
"error": t('api.missingGraphIdEnsure')
}), 400
simulation_requirement = project.simulation_requirement
if not simulation_requirement:
return jsonify({
"success": False,
"error": t('api.missingSimRequirement')
}), 400
# Generate report_id up front so we can return it to the client immediately
import uuid
report_id = f"report_{uuid.uuid4().hex[:12]}"
# Create the async task
task_manager = TaskManager()
task_id = task_manager.create_task(
task_type="report_generate",
metadata={
"simulation_id": simulation_id,
"graph_id": graph_id,
"report_id": report_id
}
)
# Capture locale before spawning background thread
current_locale = get_locale()
# Background worker
def run_generate():
set_locale(current_locale)
try:
task_manager.update_task(
task_id,
status=TaskStatus.PROCESSING,
progress=0,
message=t('api.initReportAgent')
)
# Create the Report Agent
agent = ReportAgent(
graph_id=graph_id,
simulation_id=simulation_id,
simulation_requirement=simulation_requirement
)
# Progress callback
def progress_callback(stage, progress, message):
task_manager.update_task(
task_id,
progress=progress,
message=f"[{stage}] {message}"
)
# Generate the report (passing in the pre-generated report_id)
report = agent.generate_report(
progress_callback=progress_callback,
report_id=report_id
)
# Persist the report
ReportManager.save_report(report)
if report.status == ReportStatus.COMPLETED:
task_manager.complete_task(
task_id,
result={
"report_id": report.report_id,
"simulation_id": simulation_id,
"status": "completed"
}
)
else:
task_manager.fail_task(task_id, report.error or t('api.reportGenerateFailed'))
except Exception as e:
logger.error(f"Report generation failed: {str(e)}")
task_manager.fail_task(task_id, str(e))
# Launch background thread
thread = threading.Thread(target=run_generate, daemon=True)
thread.start()
return jsonify({
"success": True,
"data": {
"simulation_id": simulation_id,
"report_id": report_id,
"task_id": task_id,
"status": "generating",
"message": t('api.reportGenerateStarted'),
"already_generated": False
}
})
except Exception as e:
logger.error(f"Failed to start report generation task: {str(e)}")
return jsonify({
"success": False,
"error": str(e),
"traceback": traceback.format_exc()
}), 500
@report_bp.route('/generate/status', methods=['POST'])
def get_generate_status():
"""
Query report-generation progress.
Request (JSON):
{
"task_id": "task_xxxx", // optional, task_id returned by /generate
"simulation_id": "sim_xxxx" // optional, simulation ID
}
Returns:
{
"success": true,
"data": {
"task_id": "task_xxxx",
"status": "processing|completed|failed",
"progress": 45,
"message": "..."
}
}
"""
try:
data = request.get_json() or {}
task_id = data.get('task_id')
simulation_id = data.get('simulation_id')
# If a simulation_id was provided, first check whether a completed report already exists
if simulation_id:
existing_report = ReportManager.get_report_by_simulation(simulation_id)
if existing_report and existing_report.status == ReportStatus.COMPLETED:
return jsonify({
"success": True,
"data": {
"simulation_id": simulation_id,
"report_id": existing_report.report_id,
"status": "completed",
"progress": 100,
"message": t('api.reportGenerated'),
"already_completed": True
}
})
if not task_id:
return jsonify({
"success": False,
"error": t('api.requireTaskOrSimId')
}), 400
task_manager = TaskManager()
task = task_manager.get_task(task_id)
if not task:
return jsonify({
"success": False,
"error": t('api.taskNotFound', id=task_id)
}), 404
return jsonify({
"success": True,
"data": task.to_dict()
})
except Exception as e:
logger.error(f"Failed to query task status: {str(e)}")
return jsonify({
"success": False,
"error": str(e)
}), 500
# ============== Report retrieval endpoints ==============
@report_bp.route('/<report_id>', methods=['GET'])
def get_report(report_id: str):
"""
Fetch report details.
Returns:
{
"success": true,
"data": {
"report_id": "report_xxxx",
"simulation_id": "sim_xxxx",
"status": "completed",
"outline": {...},
"markdown_content": "...",
"created_at": "...",
"completed_at": "..."
}
}
"""
try:
report = ReportManager.get_report(report_id)
if not report:
return jsonify({
"success": False,
"error": t('api.reportNotFound', id=report_id)
}), 404
return jsonify({
"success": True,
"data": report.to_dict()
})
except Exception as e:
logger.error(f"Failed to fetch report: {str(e)}")
return jsonify({
"success": False,
"error": str(e),
"traceback": traceback.format_exc()
}), 500
@report_bp.route('/by-simulation/<simulation_id>', methods=['GET'])
def get_report_by_simulation(simulation_id: str):
"""
Fetch the report for a given simulation ID.
Returns:
{
"success": true,
"data": {
"report_id": "report_xxxx",
...
}
}
"""
try:
report = ReportManager.get_report_by_simulation(simulation_id)
if not report:
return jsonify({
"success": False,
"error": t('api.noReportForSim', id=simulation_id),
"has_report": False
}), 404
return jsonify({
"success": True,
"data": report.to_dict(),
"has_report": True
})
except Exception as e:
logger.error(f"Failed to fetch report: {str(e)}")
return jsonify({
"success": False,
"error": str(e),
"traceback": traceback.format_exc()
}), 500
@report_bp.route('/list', methods=['GET'])
def list_reports():
"""
List all reports.
Query parameters:
simulation_id: filter by simulation ID (optional).
limit: max number of results to return (default 50).
Returns:
{
"success": true,
"data": [...],
"count": 10
}
"""
try:
simulation_id = request.args.get('simulation_id')
limit = request.args.get('limit', 50, type=int)
reports = ReportManager.list_reports(
simulation_id=simulation_id,
limit=limit
)
return jsonify({
"success": True,
"data": [r.to_dict() for r in reports],
"count": len(reports)
})
except Exception as e:
logger.error(f"Failed to list reports: {str(e)}")
return jsonify({
"success": False,
"error": str(e),
"traceback": traceback.format_exc()
}), 500
@report_bp.route('/<report_id>/download', methods=['GET'])
def download_report(report_id: str):
"""
Download a report (Markdown format).
Returns the report as a Markdown file.
"""
try:
report = ReportManager.get_report(report_id)
if not report:
return jsonify({
"success": False,
"error": t('api.reportNotFound', id=report_id)
}), 404
md_path = ReportManager._get_report_markdown_path(report_id)
if not os.path.exists(md_path):
# If the markdown file is missing, create a temporary one on the fly
import tempfile
with tempfile.NamedTemporaryFile(mode='w', suffix='.md', delete=False) as f:
f.write(report.markdown_content)
temp_path = f.name
return send_file(
temp_path,
as_attachment=True,
download_name=f"{report_id}.md"
)
return send_file(
md_path,
as_attachment=True,
download_name=f"{report_id}.md"
)
except Exception as e:
logger.error(f"Failed to download report: {str(e)}")
return jsonify({
"success": False,
"error": str(e),
"traceback": traceback.format_exc()
}), 500
@report_bp.route('/<report_id>', methods=['DELETE'])
def delete_report(report_id: str):
"""Delete a report."""
try:
success = ReportManager.delete_report(report_id)
if not success:
return jsonify({
"success": False,
"error": t('api.reportNotFound', id=report_id)
}), 404
return jsonify({
"success": True,
"message": t('api.reportDeleted', id=report_id)
})
except Exception as e:
logger.error(f"Failed to delete report: {str(e)}")
return jsonify({
"success": False,
"error": str(e),
"traceback": traceback.format_exc()
}), 500
# ============== Report Agent chat endpoint ==============
@report_bp.route('/chat', methods=['POST'])
def chat_with_report_agent():
"""
Chat with the Report Agent.
The Report Agent can autonomously call retrieval tools to answer questions
while it converses with the user.
Request (JSON):
{
"simulation_id": "sim_xxxx", // required, simulation ID
"message": "Please explain the public-opinion trend", // required, user message
"chat_history": [ // optional, conversation history
{"role": "user", "content": "..."},
{"role": "assistant", "content": "..."}
]
}
Returns:
{
"success": true,
"data": {
"response": "Agent reply ...",
"tool_calls": [list of tools invoked],
"sources": [information sources]
}
}
"""
try:
data = request.get_json() or {}
simulation_id = data.get('simulation_id')
message = data.get('message')
chat_history = data.get('chat_history', [])
if not simulation_id:
return jsonify({
"success": False,
"error": t('api.requireSimulationId')
}), 400
if not message:
return jsonify({
"success": False,
"error": t('api.requireMessage')
}), 400
# Look up the simulation and the project
manager = SimulationManager()
state = manager.get_simulation(simulation_id)
if not state:
return jsonify({
"success": False,
"error": t('api.simulationNotFound', id=simulation_id)
}), 404
project = ProjectManager.get_project(state.project_id)
if not project:
return jsonify({
"success": False,
"error": t('api.projectNotFound', id=state.project_id)
}), 404
graph_id = state.graph_id or project.graph_id
if not graph_id:
return jsonify({
"success": False,
"error": t('api.missingGraphId')
}), 400
simulation_requirement = project.simulation_requirement or ""
# Create the agent and run the chat
agent = ReportAgent(
graph_id=graph_id,
simulation_id=simulation_id,
simulation_requirement=simulation_requirement
)
result = agent.chat(message=message, chat_history=chat_history)
return jsonify({
"success": True,
"data": result
})
except Exception as e:
logger.error(f"Chat failed: {str(e)}")
return jsonify({
"success": False,
"error": str(e),
"traceback": traceback.format_exc()
}), 500
# ============== Report progress and per-section endpoints ==============
@report_bp.route('/<report_id>/progress', methods=['GET'])
def get_report_progress(report_id: str):
"""
Get the (live) report-generation progress.
Returns:
{
"success": true,
"data": {
"status": "generating",
"progress": 45,
"message": "Generating section: Key Findings",
"current_section": "Key Findings",
"completed_sections": ["Executive Summary", "Simulation Background"],
"updated_at": "2025-12-09T..."
}
}
"""
try:
progress = ReportManager.get_progress(report_id)
if not progress:
return jsonify({
"success": False,
"error": t('api.reportProgressNotAvail', id=report_id)
}), 404
return jsonify({
"success": True,
"data": progress
})
except Exception as e:
logger.error(f"Failed to fetch report progress: {str(e)}")
return jsonify({
"success": False,
"error": str(e),
"traceback": traceback.format_exc()
}), 500
@report_bp.route('/<report_id>/sections', methods=['GET'])
def get_report_sections(report_id: str):
"""
Get the list of already-generated sections (per-section output).
The frontend can poll this endpoint to stream section content without
waiting for the whole report to finish.
Returns:
{
"success": true,
"data": {
"report_id": "report_xxxx",
"sections": [
{
"filename": "section_01.md",
"section_index": 1,
"content": "## Executive Summary\n\n..."
},
...
],
"total_sections": 3,
"is_complete": false
}
}
"""
try:
sections = ReportManager.get_generated_sections(report_id)
# Read the report status
report = ReportManager.get_report(report_id)
is_complete = report is not None and report.status == ReportStatus.COMPLETED
return jsonify({
"success": True,
"data": {
"report_id": report_id,
"sections": sections,
"total_sections": len(sections),
"is_complete": is_complete
}
})
except Exception as e:
logger.error(f"Failed to list sections: {str(e)}")
return jsonify({
"success": False,
"error": str(e),
"traceback": traceback.format_exc()
}), 500
@report_bp.route('/<report_id>/section/<int:section_index>', methods=['GET'])
def get_single_section(report_id: str, section_index: int):
"""
Get the content of a single section.
Returns:
{
"success": true,
"data": {
"filename": "section_01.md",
"content": "## Executive Summary\n\n..."
}
}
"""
try:
section_path = ReportManager._get_section_path(report_id, section_index)
if not os.path.exists(section_path):
return jsonify({
"success": False,
"error": t('api.sectionNotFound', index=f"{section_index:02d}")
}), 404
with open(section_path, 'r', encoding='utf-8') as f:
content = f.read()
return jsonify({
"success": True,
"data": {
"filename": f"section_{section_index:02d}.md",
"section_index": section_index,
"content": content
}
})
except Exception as e:
logger.error(f"Failed to fetch section content: {str(e)}")
return jsonify({
"success": False,
"error": str(e),
"traceback": traceback.format_exc()
}), 500
# ============== Report status check endpoint ==============
@report_bp.route('/check/<simulation_id>', methods=['GET'])
def check_report_status(simulation_id: str):
"""
Check whether a simulation has a report, and its status.
Used by the frontend to decide whether the Interview feature is unlocked.
Returns:
{
"success": true,
"data": {
"simulation_id": "sim_xxxx",
"has_report": true,
"report_status": "completed",
"report_id": "report_xxxx",
"interview_unlocked": true
}
}
"""
try:
report = ReportManager.get_report_by_simulation(simulation_id)
has_report = report is not None
report_status = report.status.value if report else None
report_id = report.report_id if report else None
# The Interview feature unlocks only when the report is complete
interview_unlocked = has_report and report.status == ReportStatus.COMPLETED
return jsonify({
"success": True,
"data": {
"simulation_id": simulation_id,
"has_report": has_report,
"report_status": report_status,
"report_id": report_id,
"interview_unlocked": interview_unlocked
}
})
except Exception as e:
logger.error(f"Failed to check report status: {str(e)}")
return jsonify({
"success": False,
"error": str(e),
"traceback": traceback.format_exc()
}), 500
# ============== Agent log endpoint ==============
@report_bp.route('/<report_id>/agent-log', methods=['GET'])
def get_agent_log(report_id: str):
"""
Fetch the Report Agent's detailed execution log.
Streams every step of the report-generation process, including:
- report start, planning start/finish
- each section's start, tool calls, LLM responses, finish
- report completion or failure
Query parameters:
from_line: line number to start reading from (optional, default 0, used for incremental reads)
Returns:
{
"success": true,
"data": {
"logs": [
{
"timestamp": "2025-12-13T...",
"elapsed_seconds": 12.5,
"report_id": "report_xxxx",
"action": "tool_call",
"stage": "generating",
"section_title": "Executive Summary",
"section_index": 1,
"details": {
"tool_name": "insight_forge",
"parameters": {...},
...
}
},
...
],
"total_lines": 25,
"from_line": 0,
"has_more": false
}
}
"""
try:
from_line = request.args.get('from_line', 0, type=int)
log_data = ReportManager.get_agent_log(report_id, from_line=from_line)
return jsonify({
"success": True,
"data": log_data
})
except Exception as e:
logger.error(f"Failed to fetch agent log: {str(e)}")
return jsonify({
"success": False,
"error": str(e),
"traceback": traceback.format_exc()
}), 500
@report_bp.route('/<report_id>/agent-log/stream', methods=['GET'])
def stream_agent_log(report_id: str):
"""
Get the full agent log (one-shot dump).
Returns:
{
"success": true,
"data": {
"logs": [...],
"count": 25
}
}
"""
try:
logs = ReportManager.get_agent_log_stream(report_id)
return jsonify({
"success": True,
"data": {
"logs": logs,
"count": len(logs)
}
})
except Exception as e:
logger.error(f"Failed to fetch agent log: {str(e)}")
return jsonify({
"success": False,
"error": str(e),
"traceback": traceback.format_exc()
}), 500
# ============== Console log endpoint ==============
@report_bp.route('/<report_id>/console-log', methods=['GET'])
def get_console_log(report_id: str):
"""
Fetch the Report Agent's console-style log output.
Streams the console output (INFO, WARNING, etc.) emitted during report
generation. Unlike the structured JSON returned by the agent-log endpoint,
this is plain text in console format.
Query parameters:
from_line: line number to start reading from (optional, default 0, used for incremental reads)
Returns:
{
"success": true,
"data": {
"logs": [
"[19:46:14] INFO: Search complete: found 15 related facts",
"[19:46:14] INFO: Graph search: graph_id=xxx, query=...",
...
],
"total_lines": 100,
"from_line": 0,
"has_more": false
}
}
"""
try:
from_line = request.args.get('from_line', 0, type=int)
log_data = ReportManager.get_console_log(report_id, from_line=from_line)
return jsonify({
"success": True,
"data": log_data
})
except Exception as e:
logger.error(f"Failed to fetch console log: {str(e)}")
return jsonify({
"success": False,
"error": str(e),
"traceback": traceback.format_exc()
}), 500
@report_bp.route('/<report_id>/console-log/stream', methods=['GET'])
def stream_console_log(report_id: str):
"""
Get the full console log (one-shot dump).
Returns:
{
"success": true,
"data": {
"logs": [...],
"count": 100
}
}
"""
try:
logs = ReportManager.get_console_log_stream(report_id)
return jsonify({
"success": True,
"data": {
"logs": logs,
"count": len(logs)
}
})
except Exception as e:
logger.error(f"Failed to fetch console log: {str(e)}")
return jsonify({
"success": False,
"error": str(e),
"traceback": traceback.format_exc()
}), 500
# ============== Tool-call endpoints (for debugging) ==============
@report_bp.route('/tools/search', methods=['POST'])
def search_graph_tool():
"""
Graph search tool endpoint (for debugging).
Request (JSON):
{
"graph_id": "mirofish_xxxx",
"query": "search query",
"limit": 10
}
"""
try:
data = request.get_json() or {}
graph_id = data.get('graph_id')
query = data.get('query')
limit = data.get('limit', 10)
if not graph_id or not query:
return jsonify({
"success": False,
"error": t('api.requireGraphIdAndQuery')
}), 400
from ..services.zep_tools import ZepToolsService
tools = ZepToolsService()
result = tools.search_graph(
graph_id=graph_id,
query=query,
limit=limit
)
return jsonify({
"success": True,
"data": result.to_dict()
})
except Exception as e:
logger.error(f"Graph search failed: {str(e)}")
return jsonify({
"success": False,
"error": str(e),
"traceback": traceback.format_exc()
}), 500
@report_bp.route('/tools/statistics', methods=['POST'])
def get_graph_statistics_tool():
"""
Graph statistics tool endpoint (for debugging).
Request (JSON):
{
"graph_id": "mirofish_xxxx"
}
"""
try:
data = request.get_json() or {}
graph_id = data.get('graph_id')
if not graph_id:
return jsonify({
"success": False,
"error": t('api.requireGraphId')
}), 400
from ..services.zep_tools import ZepToolsService
tools = ZepToolsService()
result = tools.get_graph_statistics(graph_id)
return jsonify({
"success": True,
"data": result
})
except Exception as e:
logger.error(f"Failed to fetch graph statistics: {str(e)}")
return jsonify({
"success": False,
"error": str(e),
"traceback": traceback.format_exc()
}), 500