Skip to content

Upgrade: Handled Test cases Like Large Doc chunking One doc test case… #3

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
techieworld2 wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Upgrade: Handled Test cases Like Large Doc chunking One doc test case… #3

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
40 changes: 0 additions & 40 deletions specgap/.env.example
View file
Open in desktop

This file was deleted.

91 changes: 52 additions & 39 deletions specgap/app/main.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -26,7 +26,8 @@
from app.services.patch_pack import build_patch_pack_files
from app.services.tech_engine import analyze_tech_gaps
from app.services.biz_engine import analyze_proposal_leverage
from app.services.cross_check import run_cross_check
from app.services.cross_check import run_cross_check, run_smart_comparison
Copy link

Copilot AI Feb 8, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

run_cross_check is imported but never used in main.py (the code paths now call run_smart_comparison). Removing the unused import helps keep dependencies clear and avoids lint failures if you run static checks.

Suggested change
from app.services.cross_check import run_cross_check, run_smart_comparison
from app.services.cross_check import run_smart_comparison

Copilot uses AI. Check for mistakes.
from app.services.chunker import condense_large_document


# ============== LOGGING SETUP ==============
Expand Down Expand Up @@ -133,6 +134,14 @@ async def run_council_session(

logger.info(f"Council session started for: {file_names}")

# Condense large documents for council efficiency (Test Case 1: 200-page PDFs)
if len(combined_text) > settings.MAX_CONTEXT_CHARS:
logger.info(
f"Large document detected ({len(combined_text):,} chars), "
f"condensing for council (limit: {settings.MAX_CONTEXT_CHARS:,})..."
)
combined_text = await condense_large_document(combined_text)

initial_state = {
"combined_context": combined_text,
"domain": domain,
Expand Down Expand Up @@ -238,13 +247,15 @@ async def run_deep_analysis(
"""
combined_text = ""
file_names = []

file_texts = {}

for f in files:
await f.seek(0)
text, _ = await extract_text_from_file(f)
combined_text += f"\n=== SOURCE DOCUMENT: {f.filename} ===\n{text}"
file_names.append(f.filename)

file_texts[f.filename] = text

logger.info(f"Deep analysis started for: {file_names}")

try:
Expand All @@ -255,14 +266,20 @@ async def run_deep_analysis(
# Run Biz Engine
logger.info("[Deep Audit] Running Legal Leverage Analysis...")
legal_report = await analyze_proposal_leverage(combined_text)

# Run Cross-Check

tech_valid = not tech_report.get("error") or bool(tech_report.get("critical_gaps"))
legal_valid = not legal_report.get("error") or bool(legal_report.get("trap_clauses"))

if not tech_valid:
logger.warning("Tech engine returned error, cross-check will run without tech context")
if not legal_valid:
logger.warning("Legal engine returned error, cross-check will run without legal context")

logger.info("[Deep Audit] Running Cross-Check Synthesis...")
synthesis = await run_cross_check(
tech_text=combined_text,
proposal_text=combined_text,
tech_report=tech_report,
legal_report=legal_report
synthesis = await run_smart_comparison(
file_texts=file_texts,
tech_report=tech_report if tech_valid else None,
legal_report=legal_report if legal_valid else None
)

logger.info("Deep analysis completed successfully")
Expand Down Expand Up @@ -298,7 +315,6 @@ async def run_deep_analysis_legacy(
return await run_deep_analysis(files, domain)


# ============== FULL SPECTRUM ENDPOINT ==============

@app.post("/api/v1/audit/full-spectrum", tags=["Audit"])
async def run_full_spectrum_analysis(
Expand All @@ -316,17 +332,27 @@ async def run_full_spectrum_analysis(
"""
combined_text = ""
file_names = []
file_texts = {}

for f in files:
await f.seek(0)
text, _ = await extract_text_from_file(f)
combined_text += f"\n=== SOURCE DOCUMENT: {f.filename} ===\n{text}"
file_names.append(f.filename)

file_texts[f.filename] = text

logger.info(f"Full spectrum analysis started for: {file_names}")

council_text = combined_text
if len(combined_text) > settings.MAX_CONTEXT_CHARS:
logger.info(
f"Large document detected ({len(combined_text):,} chars), "
f"condensing for council..."
)
council_text = await condense_large_document(combined_text)

council_state = {
"combined_context": combined_text,
"combined_context": council_text,
"domain": domain,
"round_1_drafts": {},
"round_2_drafts": {},
Expand All @@ -344,11 +370,14 @@ async def run_full_spectrum_analysis(
logger.info("[Full Spectrum] Running Deep Analysis...")
tech_report = await analyze_tech_gaps(combined_text)
legal_report = await analyze_proposal_leverage(combined_text)
synthesis = await run_cross_check(
tech_text=combined_text,
proposal_text=combined_text,
tech_report=tech_report,
legal_report=legal_report

tech_valid = not tech_report.get("error") or bool(tech_report.get("critical_gaps"))
legal_valid = not legal_report.get("error") or bool(legal_report.get("trap_clauses"))

synthesis = await run_smart_comparison(
file_texts=file_texts,
tech_report=tech_report if tech_valid else None,
legal_report=legal_report if legal_valid else None
)

logger.info("Full spectrum analysis completed successfully")
Expand Down Expand Up @@ -393,12 +422,7 @@ async def run_full_spectrum_legacy(
async def classify_uploaded_document(
file: UploadFile = File(..., description="Document to classify")
):
"""
Classify a document to determine recommended analysis agents.

Useful for understanding what type of document you're uploading
before running a full analysis.
"""

await file.seek(0)
text, metadata = await extract_text_from_file(file)
classification = await classify_document(text, file.filename)
Expand All @@ -415,11 +439,7 @@ async def classify_uploaded_document(
async def extract_document_text(
file: UploadFile = File(..., description="Document to extract text from")
):
"""
Extract text from a document without analysis.

Useful for previewing what the AI will see.
"""

await file.seek(0)
content = await file.read()
file_hash = compute_file_hash(content)
Expand All @@ -437,7 +457,6 @@ async def extract_document_text(
}


# ============== AUDIT HISTORY ==============

@app.get("/api/v1/audits", tags=["History"])
async def list_audits(
Expand All @@ -446,9 +465,7 @@ async def list_audits(
audit_type: str = Query(None, description="Filter by audit type"),
risk_level: str = Query(None, description="Filter by risk level")
):
"""
List saved audit records with optional filtering.
"""

Copy link

Copilot AI Feb 8, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The documentation string for this endpoint was removed, which means the generated OpenAPI docs will no longer include an operation description here. If the project relies on Swagger/ReDoc for usability, consider restoring the docstring (or setting summary/description on the route decorator).

Suggested change
"""
List stored audits with optional filtering and pagination.
This endpoint returns a paginated list of audit records, which can be
filtered by audit type and risk level for easier browsing of history.
"""

Copilot uses AI. Check for mistakes.
from app.core.database import get_db_session

with get_db_session() as db:
Expand Down Expand Up @@ -481,9 +498,7 @@ async def list_audits(

@app.get("/api/v1/audits/statistics", tags=["History"])
async def get_audit_statistics():
"""
Get aggregate statistics for dashboard.
"""

from app.core.database import get_db_session

with get_db_session() as db:
Expand All @@ -497,9 +512,7 @@ async def get_audit_statistics():

@app.get("/api/v1/audits/{audit_id}", tags=["History"])
async def get_audit_detail(audit_id: str):
"""
Get detailed audit record by ID.
"""

from app.core.database import get_db_session

with get_db_session() as db:
Expand Down
15 changes: 14 additions & 1 deletion specgap/app/services/__init__.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,8 +6,11 @@
from .workflow import council_app, CouncilState
from .tech_engine import analyze_tech_gaps
from .biz_engine import analyze_proposal_leverage
from .cross_check import run_cross_check
from .cross_check import run_cross_check, run_smart_comparison, run_single_doc_audit
from .patch_pack import build_patch_pack_files
from .safe_parse import safe_parse_llm_response, extract_json
from .sanitizer import sanitize_document_text, wrap_as_document_context
from .chunker import chunk_document, condense_large_document
from .parser import (
extract_text_from_file,
extract_text_from_pdf,
Expand All @@ -28,8 +31,18 @@
"analyze_tech_gaps",
"analyze_proposal_leverage",
"run_cross_check",
"run_smart_comparison",
"run_single_doc_audit",
"build_patch_pack_files",

# Utilities (Test Case fixes)
"safe_parse_llm_response",
"extract_json",
"sanitize_document_text",
"wrap_as_document_context",
"chunk_document",
"condense_large_document",

# Parser
"extract_text_from_file",
"extract_text_from_pdf",
Expand Down
19 changes: 12 additions & 7 deletions specgap/app/services/biz_engine.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,6 +10,8 @@
from app.core.config import model_text, settings
from app.core.logging import get_logger
from app.core.exceptions import AIModelError, AIResponseParseError
from app.services.safe_parse import safe_parse_llm_response
from app.services.sanitizer import wrap_as_document_context

logger = get_logger("biz_engine")

Expand Down Expand Up @@ -98,7 +100,7 @@ async def analyze_proposal_leverage(
logger.warning(f"Truncating input from {len(proposal_text):,} to {max_chars:,} chars")
proposal_text = proposal_text[:max_chars] + "\n\n[...content truncated...]"

full_prompt = f"{LEGAL_SYSTEM_PROMPT}\n\n--- BUSINESS PROPOSAL TEXT ---\n{proposal_text}"
full_prompt = f"{LEGAL_SYSTEM_PROMPT}\n\n{wrap_as_document_context(proposal_text, label='BUSINESS PROPOSAL')}"

last_error = None
for attempt in range(max_retries):
Expand All @@ -115,8 +117,15 @@ async def analyze_proposal_leverage(
details="Empty response"
)

cleaned = _clean_json_response(response.text)
result = json.loads(cleaned)
result = safe_parse_llm_response(
response.text,
expected_keys=["leverage_score", "trap_clauses"]
)

Comment on lines +120 to +124
Copy link

Copilot AI Feb 8, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

After switching to safe_parse_llm_response, _clean_json_response is no longer referenced in this file. Consider deleting it to reduce dead code and avoid confusion about which parser to use.

Copilot uses AI. Check for mistakes.
if result.get("parse_error"):
last_error = AIResponseParseError(agent="biz_engine", raw_response=response.text)
logger.warning(f"JSON parse error on attempt {attempt + 1}: {result.get('error_message')}")
continue # Retry with next attempt

# Validate and set defaults
if "leverage_score" not in result:
Expand All @@ -136,10 +145,6 @@ async def analyze_proposal_leverage(

return result

except json.JSONDecodeError as e:
last_error = AIResponseParseError(agent="biz_engine", raw_response=response.text if response else None)
logger.warning(f"JSON parse error on attempt {attempt + 1}: {e}")

except Exception as e:
last_error = e
logger.warning(f"Legal analysis attempt {attempt + 1} failed: {e}")
Expand Down
Loading
Loading