Compare commits
1 Commits
step35/205
...
step35/155
| Author | SHA1 | Date | |
|---|---|---|---|
|
b4a3501aa3 |
@@ -1,418 +0,0 @@
|
||||
#!/usr/bin/env python3
|
||||
"""
|
||||
knowledge_synthesizer.py — Zero-shot knowledge synthesis for compounding intelligence.
|
||||
|
||||
Given two unrelated knowledge entries, generate a novel hypothesis that connects them.
|
||||
Pipeline: pick unrelated pair → extract entities/relations → find bridging concepts →
|
||||
score plausibility → store if above threshold.
|
||||
|
||||
Usage:
|
||||
python3 scripts/knowledge_synthesizer.py --pair hermes-agent:pitfall:001 global:tool-quirk:001
|
||||
python3 scripts/knowledge_synthesizer.py --auto --threshold 0.75
|
||||
python3 scripts/knowledge_synthesizer.py --dry-run # show candidate pair without synthesizing
|
||||
"""
|
||||
|
||||
import argparse
|
||||
import json
|
||||
import os
|
||||
import sys
|
||||
import time
|
||||
import hashlib
|
||||
from datetime import datetime, timezone
|
||||
from pathlib import Path
|
||||
from typing import Optional, Tuple, List, Dict
|
||||
|
||||
SCRIPT_DIR = Path(__file__).parent.absolute()
|
||||
sys.path.insert(0, str(SCRIPT_DIR))
|
||||
|
||||
REPO_ROOT = SCRIPT_DIR.parent
|
||||
KNOWLEDGE_DIR = REPO_ROOT / "knowledge"
|
||||
TEMPLATE_PATH = SCRIPT_DIR.parent / "templates" / "synthesis-prompt.md"
|
||||
|
||||
# Default API configuration
|
||||
DEFAULT_API_BASE = os.environ.get(
|
||||
"SYNTHESIS_API_BASE",
|
||||
os.environ.get("HARVESTER_API_BASE", "https://api.nousresearch.com/v1")
|
||||
)
|
||||
DEFAULT_API_KEY = os.environ.get("SYNTHESIS_API_KEY", "")
|
||||
DEFAULT_MODEL = os.environ.get(
|
||||
"SYNTHESIS_MODEL",
|
||||
os.environ.get("HARVESTER_MODEL", "xiaomi/mimo-v2-pro")
|
||||
)
|
||||
|
||||
# Places to look for API keys if not in env
|
||||
API_KEY_PATHS = [
|
||||
os.path.expanduser("~/.config/nous/key"),
|
||||
os.path.expanduser("~/.hermes/keymaxxing/active/minimax.key"),
|
||||
os.path.expanduser("~/.config/openrouter/key"),
|
||||
]
|
||||
|
||||
|
||||
def find_api_key() -> str:
|
||||
for path in API_KEY_PATHS:
|
||||
if os.path.exists(path):
|
||||
with open(path) as f:
|
||||
key = f.read().strip()
|
||||
if key:
|
||||
return key
|
||||
return ""
|
||||
|
||||
|
||||
def load_index() -> dict:
|
||||
index_path = KNOWLEDGE_DIR / "index.json"
|
||||
if not index_path.exists():
|
||||
return {"version": 1, "total_facts": 0, "facts": []}
|
||||
with open(index_path) as f:
|
||||
return json.load(f)
|
||||
|
||||
|
||||
def save_index(index: dict) -> None:
|
||||
KNOWLEDGE_DIR.mkdir(parents=True, exist_ok=True)
|
||||
index_path = KNOWLEDGE_DIR / "index.json"
|
||||
with open(index_path, 'w', encoding='utf-8') as f:
|
||||
json.dump(index, f, indent=2, ensure_ascii=False)
|
||||
|
||||
|
||||
def next_sequence(facts: List[dict], domain: str, category: str) -> int:
|
||||
"""Find next sequence number for given domain:category."""
|
||||
prefix = f"{domain}:{category}:"
|
||||
max_seq = 0
|
||||
for fact in facts:
|
||||
fid = fact.get('id', '')
|
||||
if fid.startswith(prefix):
|
||||
try:
|
||||
seq = int(fid.split(':')[-1])
|
||||
max_seq = max(max_seq, seq)
|
||||
except ValueError:
|
||||
continue
|
||||
return max_seq + 1
|
||||
|
||||
|
||||
def generate_id(domain: str, category: str, facts: List[dict]) -> str:
|
||||
"""Generate a new unique ID for synthesized fact."""
|
||||
seq = next_sequence(facts, domain, category)
|
||||
return f"{domain}:{category}:{seq:03d}"
|
||||
|
||||
|
||||
def facts_are_unrelated(f1: dict, f2: dict) -> bool:
|
||||
"""Return True if two facts have no existing 'related' link."""
|
||||
id1, id2 = f1['id'], f2['id']
|
||||
rel1 = set(f1.get('related', []))
|
||||
rel2 = set(f2.get('related', []))
|
||||
return (id2 not in rel1) and (id1 not in rel2)
|
||||
|
||||
|
||||
def find_candidate_pair(facts: List[dict]) -> Optional[Tuple[dict, dict]]:
|
||||
"""Pick two unrelated facts from different domains if possible."""
|
||||
# Prefer cross-domain pairs for more creative synthesis
|
||||
by_domain = {}
|
||||
for f in facts:
|
||||
by_domain.setdefault(f['domain'], []).append(f)
|
||||
|
||||
domains = list(by_domain.keys())
|
||||
if len(domains) < 2:
|
||||
# Not enough domain diversity, pick any unrelated pair
|
||||
for i, f1 in enumerate(facts):
|
||||
for f2 in facts[i+1:]:
|
||||
if facts_are_unrelated(f1, f2):
|
||||
return f1, f2
|
||||
return None
|
||||
|
||||
# Try cross-domain first
|
||||
for d1 in domains:
|
||||
for d2 in domains:
|
||||
if d1 == d2:
|
||||
continue
|
||||
for f1 in by_domain[d1]:
|
||||
for f2 in by_domain[d2]:
|
||||
if facts_are_unrelated(f1, f2):
|
||||
return f1, f2
|
||||
|
||||
# Fallback to any unrelated pair
|
||||
return find_candidate_pair_by_simple(facts)
|
||||
|
||||
|
||||
def find_candidate_pair_by_simple(facts: List[dict]) -> Optional[Tuple[dict, dict]]:
|
||||
for i, f1 in enumerate(facts):
|
||||
for f2 in facts[i+1:]:
|
||||
if facts_are_unrelated(f1, f2):
|
||||
return f1, f2
|
||||
return None
|
||||
|
||||
|
||||
def load_synthesis_prompt() -> str:
|
||||
if TEMPLATE_PATH.exists():
|
||||
return TEMPLATE_PATH.read_text(encoding='utf-8')
|
||||
# Inline fallback
|
||||
return """You are a knowledge synthesis engine. Given two facts, generate a novel hypothesis
|
||||
that connects them in a way no human would typically link.
|
||||
|
||||
TASK:
|
||||
- Fact A: {fact_a}
|
||||
- Fact B: {fact_b}
|
||||
|
||||
OUTPUT a single JSON object:
|
||||
{
|
||||
"hypothesis": "one concise sentence linking the two facts in an actionable way",
|
||||
"plausibility": 0.0-1.0,
|
||||
"bridging_concepts": ["concept1", "concept2"],
|
||||
"suggested_tags": ["tag1", "tag2"]
|
||||
}
|
||||
|
||||
RULES:
|
||||
1. The hypothesis must be a direct logical consequence of combining both facts.
|
||||
2. Do NOT restate either fact — produce a new insight.
|
||||
3. Plausibility should reflect how likely the hypothesis is to be true given the facts.
|
||||
4. If no meaningful connection exists, return {"hypothesis":"","plausibility":0.0}.
|
||||
5. Output ONLY valid JSON, no markdown.
|
||||
"""
|
||||
|
||||
|
||||
def call_synthesis_llm(prompt: str, transcript: str, api_base: str, api_key: str, model: str) -> Optional[dict]:
|
||||
"""Call LLM to synthesize a hypothesis from two facts."""
|
||||
import urllib.request
|
||||
|
||||
messages = [
|
||||
{"role": "system", "content": prompt},
|
||||
{"role": "user", "content": transcript}
|
||||
]
|
||||
|
||||
payload = json.dumps({
|
||||
"model": model,
|
||||
"messages": messages,
|
||||
"temperature": 0.7, # More creative for synthesis
|
||||
"max_tokens": 512
|
||||
}).encode('utf-8')
|
||||
|
||||
req = urllib.request.Request(
|
||||
f"{api_base}/chat/completions",
|
||||
data=payload,
|
||||
headers={
|
||||
"Authorization": f"Bearer {api_key}",
|
||||
"Content-Type": "application/json"
|
||||
},
|
||||
method="POST"
|
||||
)
|
||||
|
||||
try:
|
||||
with urllib.request.urlopen(req, timeout=60) as resp:
|
||||
result = json.loads(resp.read().decode('utf-8'))
|
||||
content = result["choices"][0]["message"]["content"]
|
||||
return parse_synthesis_response(content)
|
||||
except Exception as e:
|
||||
print(f"ERROR: LLM call failed: {e}", file=sys.stderr)
|
||||
return None
|
||||
|
||||
|
||||
def parse_synthesis_response(content: str) -> Optional[dict]:
|
||||
"""Extract synthesis JSON from LLM response."""
|
||||
try:
|
||||
data = json.loads(content)
|
||||
if isinstance(data, dict) and 'hypothesis' in data:
|
||||
return data
|
||||
except json.JSONDecodeError:
|
||||
pass
|
||||
|
||||
import re
|
||||
json_match = re.search(r'```(?:json)?\s*({.*?})\s*```', content, re.DOTALL)
|
||||
if json_match:
|
||||
try:
|
||||
data = json.loads(json_match.group(1))
|
||||
if isinstance(data, dict) and 'hypothesis' in data:
|
||||
return data
|
||||
except json.JSONDecodeError:
|
||||
pass
|
||||
|
||||
# Try finding any JSON object
|
||||
json_match = re.search(r'(\{.*"hypothesis".*\})', content, re.DOTALL)
|
||||
if json_match:
|
||||
try:
|
||||
return json.loads(json_match.group(1))
|
||||
except json.JSONDecodeError:
|
||||
pass
|
||||
|
||||
return None
|
||||
|
||||
|
||||
def heuristic_synthesis(f1: dict, f2: dict) -> dict:
|
||||
"""Rule-based fallback synthesis when no LLM available."""
|
||||
# Simple bridging: combine tags and domains
|
||||
tags = list(set(f1.get('tags', []) + f2.get('tags', [])))
|
||||
fact1 = f1['fact']
|
||||
fact2 = f2['fact']
|
||||
|
||||
# Very basic heuristic: "By applying X from domain1 to domain2, we can Y"
|
||||
hypothesis = (
|
||||
f"Cross-domain insight: techniques from '{f1['domain']}' "
|
||||
f"might solve problems in '{f2['domain']}'. "
|
||||
f"Specifically: {fact1} could inform {fact2}"
|
||||
)
|
||||
|
||||
return {
|
||||
"hypothesis": hypothesis,
|
||||
"plausibility": 0.4, # Low confidence for heuristic
|
||||
"bridging_concepts": tags[:3],
|
||||
"suggested_tags": tags
|
||||
}
|
||||
|
||||
|
||||
def synthesize_fact(fact1: dict, fact2: dict, api_base: str, api_key: str, model: str,
|
||||
dry_run: bool = False) -> Optional[dict]:
|
||||
"""Generate a synthesized fact from two unrelated facts."""
|
||||
prompt = load_synthesis_prompt()
|
||||
transcript = f"FACT A:\n {fact1['fact']}\n(domain={fact1['domain']}, category={fact1['category']}, tags={fact1.get('tags', [])})\n\nFACT B:\n {fact2['fact']}\n(domain={fact2['domain']}, category={fact2['category']}, tags={fact2.get('tags', [])})"
|
||||
|
||||
if dry_run:
|
||||
print(f"\n[DRY RUN] Would synthesize:")
|
||||
print(f" Fact A: {fact1['fact'][:80]}")
|
||||
print(f" Fact B: {fact2['fact'][:80]}")
|
||||
return None
|
||||
|
||||
result = None
|
||||
if api_key:
|
||||
result = call_synthesis_llm(prompt, transcript, api_base, api_key, model)
|
||||
|
||||
if result is None:
|
||||
print("WARNING: LLM synthesis failed or no API key; using heuristic fallback", file=sys.stderr)
|
||||
result = heuristic_synthesis(fact1, fact2)
|
||||
|
||||
return result
|
||||
|
||||
|
||||
def fingerprint(text: str) -> str:
|
||||
return hashlib.md5(text.lower().strip().encode('utf-8')).hexdigest()
|
||||
|
||||
|
||||
def is_duplicate(hypothesis: str, existing_facts: List[dict]) -> bool:
|
||||
h_fp = fingerprint(hypothesis)
|
||||
for f in existing_facts:
|
||||
if fingerprint(f.get('fact', '')) == h_fp:
|
||||
return True
|
||||
return False
|
||||
|
||||
|
||||
def store_synthesis(synth: dict, source_ids: List[str], index: dict, threshold: float = 0.5) -> bool:
|
||||
"""Store synthesized fact if plausibility exceeds threshold."""
|
||||
plaus = synth.get('plausibility', 0.0)
|
||||
if plaus < threshold:
|
||||
print(f"Skipped: plausibility {plaus:.2f} below threshold {threshold}")
|
||||
return False
|
||||
|
||||
hypothesis = synth['hypothesis'].strip()
|
||||
if not hypothesis or is_duplicate(hypothesis, index['facts']):
|
||||
print(f"Skipped: duplicate or empty hypothesis")
|
||||
return False
|
||||
|
||||
# Build new fact
|
||||
new_fact = {
|
||||
"fact": hypothesis,
|
||||
"category": "pattern", # Synthesized connections become reusable patterns
|
||||
"domain": "global", # Cross-domain synthesis is globally applicable
|
||||
"confidence": round(plaus, 2),
|
||||
"tags": synth.get('suggested_tags', []),
|
||||
"related": source_ids,
|
||||
"first_seen": datetime.now(timezone.utc).strftime("%Y-%m-%d"),
|
||||
"last_confirmed": datetime.now(timezone.utc).strftime("%Y-%m-%d"),
|
||||
"source_count": 1,
|
||||
}
|
||||
|
||||
# Generate ID
|
||||
new_fact['id'] = generate_id("global", "pattern", index['facts'])
|
||||
|
||||
# Update index
|
||||
index['facts'].append(new_fact)
|
||||
index['total_facts'] = len(index['facts'])
|
||||
index['last_updated'] = datetime.now(timezone.utc).isoformat()
|
||||
|
||||
# Write index
|
||||
save_index(index)
|
||||
|
||||
# Append to YAML
|
||||
yaml_path = KNOWLEDGE_DIR / "global" / "patterns.yaml"
|
||||
yaml_path.parent.mkdir(parents=True, exist_ok=True)
|
||||
mode = 'a' if yaml_path.exists() else 'w'
|
||||
with open(yaml_path, mode, encoding='utf-8') as f:
|
||||
if mode == 'w':
|
||||
f.write("---\ndomain: global\ncategory: pattern\nversion: 1\nlast_updated: \"{date}\"\n---\n\n# Synthesized Patterns\n\n".format(date=datetime.now(timezone.utc).strftime("%Y-%m-%d")))
|
||||
f.write(f"\n- id: {new_fact['id']}\n")
|
||||
f.write(f" fact: \"{hypothesis}\"\n")
|
||||
f.write(f" confidence: {plaus}\n")
|
||||
if new_fact['tags']:
|
||||
f.write(f" tags: {json.dumps(new_fact['tags'])}\n")
|
||||
f.write(f" related: {json.dumps(source_ids)}\n")
|
||||
f.write(f" first_seen: \"{new_fact['first_seen']}\"\n")
|
||||
f.write(f" last_confirmed: \"{new_fact['last_confirmed']}\"\n")
|
||||
|
||||
print(f"✓ Stored synthesis as {new_fact['id']}: {hypothesis[:80]}")
|
||||
return True
|
||||
|
||||
|
||||
def main():
|
||||
parser = argparse.ArgumentParser(description="Zero-shot knowledge synthesis")
|
||||
parser.add_argument("--pair", nargs=2, metavar=("ID1", "ID2"),
|
||||
help="Synthesize a specific pair by fact ID")
|
||||
parser.add_argument("--auto", action="store_true",
|
||||
help="Automatically pick an unrelated pair")
|
||||
parser.add_argument("--threshold", type=float, default=0.6,
|
||||
help="Plausibility threshold for storage (default: 0.6)")
|
||||
parser.add_argument("--dry-run", action="store_true",
|
||||
help="Show candidate pair without synthesizing or storing")
|
||||
parser.add_argument("--model", default=None,
|
||||
help="LLM model to use (overrides env)")
|
||||
parser.add_argument("--api-base", default=None,
|
||||
help="API base URL (overrides env)")
|
||||
args = parser.parse_args()
|
||||
|
||||
# Resolve API credentials
|
||||
api_base = args.api_base or DEFAULT_API_BASE
|
||||
api_key = find_api_key() or DEFAULT_API_KEY
|
||||
model = args.model or DEFAULT_MODEL
|
||||
|
||||
if not args.dry_run and not args.pair and not args.auto:
|
||||
print("ERROR: Must specify either --pair ID1 ID2 or --auto", file=sys.stderr)
|
||||
parser.print_help()
|
||||
sys.exit(1)
|
||||
|
||||
# Load index
|
||||
index = load_index()
|
||||
facts = index['facts']
|
||||
|
||||
if len(facts) < 2:
|
||||
print("ERROR: Need at least 2 facts in knowledge store to synthesize", file=sys.stderr)
|
||||
sys.exit(1)
|
||||
|
||||
# Select facts
|
||||
f1, f2 = None, None
|
||||
if args.pair:
|
||||
id1, id2 = args.pair
|
||||
f1 = next((f for f in facts if f['id'] == id1), None)
|
||||
f2 = next((f for f in facts if f['id'] == id2), None)
|
||||
if not f1 or not f2:
|
||||
print(f"ERROR: Could not find facts with IDs {id1}, {id2}", file=sys.stderr)
|
||||
sys.exit(1)
|
||||
if not facts_are_unrelated(f1, f2):
|
||||
print(f"WARNING: Facts {id1} and {id2} are already related (may still synthesize)")
|
||||
else:
|
||||
# auto mode
|
||||
pair = find_candidate_pair(facts)
|
||||
if pair is None:
|
||||
print("ERROR: No unrelated fact pairs found — consider lowering threshold or adding more facts", file=sys.stderr)
|
||||
sys.exit(1)
|
||||
f1, f2 = pair
|
||||
print(f"Selected pair:\n {f1['id']}: {f1['fact'][:60]}\n {f2['id']}: {f2['fact'][:60]}")
|
||||
|
||||
# Synthesize
|
||||
synth = synthesize_fact(f1, f2, api_base, api_key, model, dry_run=args.dry_run)
|
||||
if synth is None:
|
||||
sys.exit(0) # dry-run path
|
||||
|
||||
print(f"\nHypothesis: {synth['hypothesis']}")
|
||||
print(f"Plausibility: {synth.get('plausibility', 0.0):.2f}")
|
||||
print(f"Bridging concepts: {synth.get('bridging_concepts', [])}")
|
||||
|
||||
# Store if acceptable
|
||||
store_synthesis(synth, [f1['id'], f2['id']], index, threshold=args.threshold)
|
||||
|
||||
|
||||
if __name__ == '__main__':
|
||||
main()
|
||||
530
scripts/linter_runner.py
Normal file
530
scripts/linter_runner.py
Normal file
@@ -0,0 +1,530 @@
|
||||
#!/usr/bin/env python3
|
||||
"""
|
||||
Linter Runner — detect languages and run linters across a repo.
|
||||
|
||||
Acceptance criteria for #155:
|
||||
[x] Detects language per repo
|
||||
[x] Runs: pylint, eslint, shellcheck, etc.
|
||||
[x] Collects violations (file, line, message, severity)
|
||||
[x] Output: lint report per repo
|
||||
|
||||
Usage:
|
||||
python3 scripts/linter_runner.py --repo .
|
||||
python3 scripts/linter_runner.py --all # Scan all repos in knowledge/repos/
|
||||
python3 scripts/linter_runner.py --repo . --format json # Machine-readable output
|
||||
python3 scripts/linter_runner.py --repo . --fail-on error # Exit non-zero if errors found
|
||||
|
||||
Output format (console):
|
||||
=== Lint Report: repo ===
|
||||
Python: 3 issues (1 error, 2 warnings)
|
||||
Shell: 1 issue (1 error)
|
||||
Total: 4 issues
|
||||
|
||||
Output format (JSON): --format json
|
||||
{"repo": "...", "issues": [...], "summary": {...}}
|
||||
"""
|
||||
|
||||
import argparse
|
||||
import json
|
||||
import os
|
||||
import subprocess
|
||||
import sys
|
||||
import time
|
||||
from dataclasses import dataclass, asdict
|
||||
from pathlib import Path
|
||||
from typing import Optional
|
||||
|
||||
SCRIPT_DIR = Path(__file__).resolve().parent
|
||||
REPO_ROOT = SCRIPT_DIR.parent
|
||||
|
||||
|
||||
@dataclass
|
||||
class Violation:
|
||||
"""A single lint violation."""
|
||||
file: str
|
||||
line: Optional[int]
|
||||
column: Optional[int]
|
||||
message: str
|
||||
severity: str # "error", "warning", "info"
|
||||
linter: str
|
||||
code: Optional[str] = None
|
||||
|
||||
|
||||
@dataclass
|
||||
class LinterResult:
|
||||
"""Result from running a single linter."""
|
||||
linter_name: str
|
||||
language: str
|
||||
violations: list[Violation]
|
||||
timed_out: bool = False
|
||||
error: Optional[str] = None
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# Language detection
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
EXTENSION_TO_LANGUAGE = {
|
||||
".py": "python",
|
||||
".js": "javascript",
|
||||
".ts": "typescript",
|
||||
".jsx": "javascript",
|
||||
".tsx": "typescript",
|
||||
".sh": "shell",
|
||||
".bash": "shell",
|
||||
".zsh": "shell",
|
||||
".yaml": "yaml",
|
||||
".yml": "yaml",
|
||||
".json": "json",
|
||||
".md": "markdown",
|
||||
".rb": "ruby",
|
||||
".go": "go",
|
||||
".rs": "rust",
|
||||
".c": "c",
|
||||
".cpp": "cpp",
|
||||
".h": "c",
|
||||
".java": "java",
|
||||
".php": "php",
|
||||
".swift": "swift",
|
||||
".kt": "kotlin",
|
||||
".scala": "scala",
|
||||
}
|
||||
|
||||
# Which linters to run per language, in order of preference
|
||||
LINTERS_BY_LANGUAGE = {
|
||||
"python": [
|
||||
("pylint", ["pylint", "--output-format=json", "--reports=no"]),
|
||||
("ruff", ["ruff", "check", "--output-format=json"]),
|
||||
("flake8", ["flake8", "--format=json"]),
|
||||
],
|
||||
"javascript": [
|
||||
("eslint", ["eslint", "--format=json", "--max-warnings=0"]),
|
||||
],
|
||||
"typescript": [
|
||||
("eslint", ["eslint", "--format=json", "--max-warnings=0"]),
|
||||
],
|
||||
"shell": [
|
||||
("shellcheck", ["shellcheck", "--format=json1"]),
|
||||
],
|
||||
"yaml": [
|
||||
("yamllint", ["yamllint", "-f", "parsable"]),
|
||||
],
|
||||
"json": [
|
||||
("jsonlinter", ["python3", "-m", "json.tool"]), # Simple syntax check
|
||||
],
|
||||
"markdown": [], # No linter yet
|
||||
"ruby": [
|
||||
("rubocop", ["rubocop", "--format", "json"]),
|
||||
],
|
||||
"go": [
|
||||
("golangci-lint", ["golangci-lint", "run", "--out-format", "json"]),
|
||||
],
|
||||
"rust": [
|
||||
("cargo clippy", ["cargo", "clippy", "--message-format=json"]),
|
||||
],
|
||||
}
|
||||
|
||||
|
||||
def detect_languages(repo_path: Path) -> dict[str, list[Path]]:
|
||||
"""
|
||||
Scan repo and return mapping: language -> list of file paths.
|
||||
Only includes languages we have linters for."""
|
||||
language_files: dict[str, list[Path]] = {lang: [] for lang in LINTERS_BY_LANGUAGE.keys()}
|
||||
|
||||
if not repo_path.exists():
|
||||
return language_files
|
||||
|
||||
exclude_dirs = {".git", ".gitea", "node_modules", "__pycache__", ".venv", "venv", "build", "dist"}
|
||||
|
||||
for root, dirs, files in os.walk(repo_path):
|
||||
# Prune excluded dirs
|
||||
dirs[:] = [d for d in dirs if d not in exclude_dirs]
|
||||
|
||||
for fname in files:
|
||||
file_path = Path(root) / fname
|
||||
suffix = file_path.suffix.lower()
|
||||
lang = EXTENSION_TO_LANGUAGE.get(suffix)
|
||||
if lang and lang in LINTERS_BY_LANGUAGE and LINTERS_BY_LANGUAGE[lang]:
|
||||
language_files[lang].append(file_path)
|
||||
|
||||
# Remove empty languages
|
||||
return {lang: files for lang, files in language_files.items() if files}
|
||||
|
||||
|
||||
def find_linter_executable(name: str) -> Optional[str]:
|
||||
"""Find linter binary in PATH, return full path or None."""
|
||||
for path_dir in os.environ.get("PATH", "").split(os.pathsep):
|
||||
candidate = Path(path_dir) / name
|
||||
if candidate.exists():
|
||||
return str(candidate)
|
||||
# Special handling for multi-word linters like "cargo clippy"
|
||||
if " " in name:
|
||||
primary = name.split()[0]
|
||||
for path_dir in os.environ.get("PATH", "").split(os.pathsep):
|
||||
candidate = Path(path_dir) / primary
|
||||
if candidate.exists():
|
||||
return name # Return full command string
|
||||
return None
|
||||
|
||||
|
||||
def run_linter(
|
||||
linter_name: str,
|
||||
command_template: list[str],
|
||||
files: list[Path],
|
||||
repo_path: Path,
|
||||
) -> LinterResult:
|
||||
"""
|
||||
Execute a linter on a set of files.
|
||||
Returns LinterResult with violations or error.
|
||||
"""
|
||||
# Build command: [linter_bin, args..., files...]
|
||||
# Most linters accept file paths as positional args at the end
|
||||
cmd = [linter_name] if " " not in linter_name else linter_name.split()
|
||||
cmd.extend(command_template[1:]) # Skip the duplicated linter name from template
|
||||
|
||||
# Add file paths, relative to repo root for cleaner output
|
||||
rel_files = [str(f.relative_to(repo_path)) for f in files]
|
||||
cmd.extend(rel_files)
|
||||
|
||||
try:
|
||||
proc = subprocess.run(
|
||||
cmd,
|
||||
cwd=repo_path,
|
||||
capture_output=True,
|
||||
text=True,
|
||||
timeout=60,
|
||||
)
|
||||
except subprocess.TimeoutExpired:
|
||||
return LinterResult(
|
||||
linter_name=linter_name,
|
||||
language="unknown",
|
||||
violations=[],
|
||||
timed_out=True,
|
||||
error="Linter timed out after 60s",
|
||||
)
|
||||
except FileNotFoundError:
|
||||
return LinterResult(
|
||||
linter_name=linter_name,
|
||||
language="unknown",
|
||||
violations=[],
|
||||
error=f"Linter not found: {linter_name}",
|
||||
)
|
||||
|
||||
# Parse output based on linter type
|
||||
violations = parse_linter_output(linter_name, proc.stdout, proc.stderr, repo_path)
|
||||
|
||||
return LinterResult(
|
||||
linter_name=linter_name,
|
||||
language=guess_language_for_linter(linter_name),
|
||||
violations=violations,
|
||||
error=proc.stderr.strip() if proc.returncode != 0 and not violations else None,
|
||||
)
|
||||
|
||||
|
||||
def guess_language_for_linter(linter_name: str) -> str:
|
||||
"""Map linter name back to language category."""
|
||||
mapping = {
|
||||
"pylint": "python",
|
||||
"ruff": "python",
|
||||
"flake8": "python",
|
||||
"eslint": "javascript",
|
||||
"shellcheck": "shell",
|
||||
"yamllint": "yaml",
|
||||
"jsonlinter": "json",
|
||||
"rubocop": "ruby",
|
||||
"golangci-lint": "go",
|
||||
"cargo clippy": "rust",
|
||||
}
|
||||
return mapping.get(linter_name, "unknown")
|
||||
|
||||
|
||||
def parse_linter_output(
|
||||
linter_name: str,
|
||||
stdout: str,
|
||||
stderr: str,
|
||||
repo_path: Path,
|
||||
) -> list[Violation]:
|
||||
"""
|
||||
Parse linter output into Violation objects.
|
||||
Supports JSON output (pylint, ruff, eslint, shellcheck json1, yamllint parsable).
|
||||
"""
|
||||
violations: list[Violation] = []
|
||||
|
||||
if linter_name in ("pylint", "ruff", "eslint"):
|
||||
# JSON array output
|
||||
try:
|
||||
data = json.loads(stdout)
|
||||
except json.JSONDecodeError:
|
||||
return []
|
||||
|
||||
if linter_name == "pylint":
|
||||
for msg in data:
|
||||
violations.append(Violation(
|
||||
file=msg.get("path", "").lstrip("./"),
|
||||
line=msg.get("line"),
|
||||
column=msg.get("column"),
|
||||
message=msg.get("message", ""),
|
||||
severity="error" if msg.get("type") == "error" else "warning",
|
||||
linter=linter_name,
|
||||
code=msg.get("symbol"),
|
||||
))
|
||||
elif linter_name == "ruff":
|
||||
for entry in data:
|
||||
violations.append(Violation(
|
||||
file=entry.get("filename", "").lstrip("./"),
|
||||
line=entry.get("location", {}).get("row"),
|
||||
column=entry.get("location", {}).get("column"),
|
||||
message=entry.get("message", ""),
|
||||
severity="error", # ruff treats all as errors
|
||||
linter=linter_name,
|
||||
code=entry.get("code"),
|
||||
))
|
||||
elif linter_name == "eslint":
|
||||
for entry in data:
|
||||
violations.append(Violation(
|
||||
file=entry.get("fileName", "").lstrip("./"),
|
||||
line=entry.get("range", {}).get("start", {}).get("line"),
|
||||
column=entry.get("range", {}).get("start", {}).get("column"),
|
||||
message=entry.get("message", ""),
|
||||
severity=entry.get("severity", 1) and "error" or "warning",
|
||||
linter=linter_name,
|
||||
code=entry.get("ruleId"),
|
||||
))
|
||||
|
||||
elif linter_name == "shellcheck":
|
||||
# shellcheck --format=json1
|
||||
try:
|
||||
data = json.loads(stdout)
|
||||
for issue in data.get("issues", []):
|
||||
violations.append(Violation(
|
||||
file=issue.get("file", "").lstrip("./"),
|
||||
line=issue.get("line"),
|
||||
column=issue.get("column"),
|
||||
message=issue.get("message", ""),
|
||||
severity="error" if issue.get("level") == "error" else "warning",
|
||||
linter=linter_name,
|
||||
code=str(issue.get("code")),
|
||||
))
|
||||
except json.JSONDecodeError:
|
||||
pass
|
||||
|
||||
elif linter_name == "yamllint":
|
||||
# parsable: file:line:col: level message [rule]
|
||||
# Example: test.yaml:3:1: [error] wrong document start (document-start)
|
||||
for line in stdout.splitlines():
|
||||
parts = line.split(":")
|
||||
if len(parts) >= 4:
|
||||
file_rel = parts[0].lstrip("./")
|
||||
line_num = int(parts[1]) if parts[1].isdigit() else None
|
||||
col_num = int(parts[2]) if parts[2].isdigit() else None
|
||||
rest = ":".join(parts[3:]).strip()
|
||||
# Parse: "[error] message (rule)"
|
||||
import re
|
||||
m = re.match(r'\[(\w+)\]\s+(.+?)(?:\s+\(([^)]+)\))?$', rest)
|
||||
if m:
|
||||
severity = m.group(1).lower()
|
||||
message = m.group(2)
|
||||
code = m.group(3)
|
||||
violations.append(Violation(
|
||||
file=file_rel,
|
||||
line=line_num,
|
||||
column=col_num,
|
||||
message=message,
|
||||
severity=severity,
|
||||
linter=linter_name,
|
||||
code=code,
|
||||
))
|
||||
|
||||
elif linter_name == "jsonlinter":
|
||||
# json.tool syntax check — no violations, just exit code
|
||||
if proc.returncode != 0:
|
||||
violations.append(Violation(
|
||||
file="(multiple)",
|
||||
line=None,
|
||||
column=None,
|
||||
message="JSON syntax error (run json.tool on each file individually)",
|
||||
severity="error",
|
||||
linter="json.tool",
|
||||
))
|
||||
|
||||
return violations
|
||||
|
||||
|
||||
def run_linters_for_language(
|
||||
language: str,
|
||||
files: list[Path],
|
||||
repo_path: Path,
|
||||
) -> LinterResult:
|
||||
"""
|
||||
Run the first available linter for this language.
|
||||
Returns the first successful run, or aggregates all errors if none available.
|
||||
"""
|
||||
linter_options = LINTERS_BY_LANGUAGE.get(language, [])
|
||||
if not linter_options:
|
||||
return LinterResult(linter_name="none", language=language, violations=[],
|
||||
error=f"No linter configured for {language}")
|
||||
|
||||
for linter_name, cmd_template in linter_options:
|
||||
# Check if linter exists
|
||||
if not find_linter_executable(linter_name):
|
||||
continue # Try next linter for this language
|
||||
|
||||
result = run_linter(linter_name, cmd_template, files, repo_path)
|
||||
if not result.error and not result.timed_out:
|
||||
return result
|
||||
# If this linter failed to start (not found), try next
|
||||
if result.error and "not found" in result.error.lower():
|
||||
continue
|
||||
|
||||
# All linters failed
|
||||
errors = []
|
||||
for linter_name, _ in linter_options:
|
||||
if find_linter_executable(linter_name):
|
||||
errors.append(f"{linter_name}: not runnable")
|
||||
else:
|
||||
errors.append(f"{linter_name}: not installed")
|
||||
return LinterResult(
|
||||
linter_name="/".join(l[0] for l in linter_options),
|
||||
language=language,
|
||||
violations=[],
|
||||
error="; ".join(errors),
|
||||
)
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# Main
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
def parse_args():
|
||||
p = argparse.ArgumentParser(description="Linter Runner for compounding-intelligence")
|
||||
p.add_argument("--repo", type=str, help="Path to repository (absolute or relative)")
|
||||
p.add_argument("--all", action="store_true", help="Scan all repos in knowledge/repos/")
|
||||
p.add_argument("--format", choices=["console", "json"], default="console",
|
||||
help="Output format (default: console)")
|
||||
p.add_argument("--fail-on", choices=["error", "warning", "any"], default="error",
|
||||
help="Exit non-zero if any violations at this level are found")
|
||||
p.add_argument("--output", type=str, help="Write report to file (default: stdout)")
|
||||
return p.parse_args()
|
||||
|
||||
|
||||
def main():
|
||||
args = parse_args()
|
||||
|
||||
if not args.repo and not args.all:
|
||||
print("ERROR: Must specify --repo <path> or --all", file=sys.stderr)
|
||||
sys.exit(2)
|
||||
|
||||
repos_to_scan = []
|
||||
if args.repo:
|
||||
repos_to_scan.append(Path(args.repo).resolve())
|
||||
if args.all:
|
||||
repos_dir = REPO_ROOT / "knowledge" / "repos"
|
||||
if repos_dir.exists():
|
||||
for yaml_file in repos_dir.glob("*.yaml"):
|
||||
# Extract repo name from filename
|
||||
repos_to_scan.append(REPO_ROOT / yaml_file.stem)
|
||||
else:
|
||||
print(f"WARNING: knowledge/repos/ not found, --all has nothing to scan", file=sys.stderr)
|
||||
|
||||
all_results: dict[str, dict] = {}
|
||||
exit_code = 0
|
||||
|
||||
for repo_path in repos_to_scan:
|
||||
if not repo_path.exists():
|
||||
print(f"WARNING: Repo not found: {repo_path}", file=sys.stderr)
|
||||
continue
|
||||
|
||||
repo_name = repo_path.name
|
||||
print(f"\n=== Scanning: {repo_name} ===") if args.format == "console" else None
|
||||
|
||||
lang_files = detect_languages(repo_path)
|
||||
results_by_lang: dict[str, LinterResult] = {}
|
||||
|
||||
for language, files in sorted(lang_files.items()):
|
||||
# Limit files for sanity (first 200 for now)
|
||||
if len(files) > 200:
|
||||
print(f" {language}: {len(files)} files (limiting to first 200)", file=sys.stderr)
|
||||
files = files[:200]
|
||||
|
||||
result = run_linters_for_language(language, files, repo_path)
|
||||
results_by_lang[language] = result
|
||||
|
||||
if args.format == "console":
|
||||
_print_language_result(language, result, repo_name)
|
||||
else:
|
||||
pass # JSON aggregation below
|
||||
|
||||
# Build summary
|
||||
total_issues = sum(len(r.violations) for r in results_by_lang.values())
|
||||
total_errors = sum(1 for v in (v for r in results_by_lang.values() for v in r.violations)
|
||||
if v.severity == "error")
|
||||
total_warnings = sum(1 for v in (v for r in results_by_lang.values() for v in r.violations)
|
||||
if v.severity == "warning")
|
||||
|
||||
if args.format == "console":
|
||||
print(f" Summary: {total_issues} issues ({total_errors} errors, {total_warnings} warnings)")
|
||||
else:
|
||||
all_results[repo_name] = {
|
||||
"languages": {lang: _result_to_dict(res) for lang, res in results_by_lang.items()},
|
||||
"summary": {
|
||||
"total_issues": total_issues,
|
||||
"errors": total_errors,
|
||||
"warnings": total_warnings,
|
||||
},
|
||||
}
|
||||
|
||||
# Determine exit code based on --fail-on
|
||||
if args.fail_on == "error" and total_errors > 0:
|
||||
exit_code = 1
|
||||
elif args.fail_on == "warning" and total_issues > 0:
|
||||
exit_code = 1
|
||||
elif args.fail_on == "any" and total_issues > 0:
|
||||
exit_code = 1
|
||||
|
||||
if args.format == "json":
|
||||
output = json.dumps({"repos": all_results, "meta": {"scanned": len(repos_to_scan)}}, indent=2)
|
||||
if args.output:
|
||||
Path(args.output).write_text(output)
|
||||
else:
|
||||
print(output)
|
||||
|
||||
sys.exit(exit_code)
|
||||
|
||||
|
||||
def _print_language_result(language: str, result: LinterResult, repo_name: str):
|
||||
"""Pretty-print a single language's lint results."""
|
||||
status = "✓"
|
||||
if result.error:
|
||||
status = "✗"
|
||||
print(f" {language}: {result.error}")
|
||||
elif result.timed_out:
|
||||
status = "⌛"
|
||||
print(f" {language}: timed out")
|
||||
else:
|
||||
n_violations = len(result.violations)
|
||||
if n_violations == 0:
|
||||
print(f" {language}: clean")
|
||||
else:
|
||||
errors = sum(1 for v in result.violations if v.severity == "error")
|
||||
warnings = n_violations - errors
|
||||
print(f" {language}: {n_violations} issues ({errors} errors, {warnings} warnings)")
|
||||
# Show first 3 violations as preview
|
||||
for v in result.violations[:3]:
|
||||
loc = f"{v.file}:{v.line or '?'}"
|
||||
print(f" {loc} [{v.severity.upper()}] {v.message[:70]}")
|
||||
if len(result.violations) > 3:
|
||||
print(f" ... and {len(result.violations) - 3} more")
|
||||
|
||||
|
||||
def _result_to_dict(result: LinterResult) -> dict:
|
||||
return {
|
||||
"linter": result.linter_name,
|
||||
"language": result.language,
|
||||
"violations": [asdict(v) for v in result.violations],
|
||||
"timed_out": result.timed_out,
|
||||
"error": result.error,
|
||||
}
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
main()
|
||||
@@ -1,235 +0,0 @@
|
||||
#!/usr/bin/env python3
|
||||
"""
|
||||
Tests for knowledge_synthesizer.py — zero-shot knowledge synthesis pipeline.
|
||||
|
||||
Run with: python3 scripts/test_knowledge_synthesizer.py
|
||||
Or via pytest: pytest scripts/test_knowledge_synthesizer.py
|
||||
"""
|
||||
|
||||
import json
|
||||
import os
|
||||
import sys
|
||||
import os
|
||||
import tempfile
|
||||
from pathlib import Path
|
||||
|
||||
# Add scripts dir to path for importing sibling module
|
||||
SCRIPT_DIR = Path(__file__).resolve().parent
|
||||
sys.path.insert(0, str(SCRIPT_DIR))
|
||||
|
||||
import importlib.util
|
||||
spec = importlib.util.spec_from_file_location(
|
||||
"ks", os.path.join(str(SCRIPT_DIR), "knowledge_synthesizer.py")
|
||||
)
|
||||
ks = importlib.util.module_from_spec(spec)
|
||||
spec.loader.exec_module(ks)
|
||||
|
||||
|
||||
# ── Test data helpers ─────────────────────────────────────────────
|
||||
|
||||
SAMPLE_FACTS = [
|
||||
{
|
||||
"id": "global:pitfall:001",
|
||||
"fact": "Branch protection requires 1 approval on main for Gitea merges",
|
||||
"category": "pitfall",
|
||||
"domain": "global",
|
||||
"confidence": 0.95,
|
||||
"tags": ["git", "merge"],
|
||||
"related": []
|
||||
},
|
||||
{
|
||||
"id": "global:tool-quirk:001",
|
||||
"fact": "Gitea token stored at ~/.config/gitea/token not GITEA_TOKEN",
|
||||
"category": "tool-quirk",
|
||||
"domain": "global",
|
||||
"confidence": 0.95,
|
||||
"tags": ["gitea", "auth"],
|
||||
"related": ["global:pitfall:001"]
|
||||
},
|
||||
{
|
||||
"id": "hermes-agent:pitfall:001",
|
||||
"fact": "deploy-crons.py leaves jobs in mixed model format",
|
||||
"category": "pitfall",
|
||||
"domain": "hermes-agent",
|
||||
"confidence": 0.95,
|
||||
"tags": ["cron"],
|
||||
"related": []
|
||||
},
|
||||
]
|
||||
|
||||
|
||||
def make_index(facts, tmp_dir: Path) -> Path:
|
||||
index = {
|
||||
"version": 1,
|
||||
"last_updated": "2026-04-13T20:00:00Z",
|
||||
"total_facts": len(facts),
|
||||
"facts": facts,
|
||||
}
|
||||
path = tmp_dir / "index.json"
|
||||
with open(path, "w") as f:
|
||||
json.dump(index, f)
|
||||
return path
|
||||
|
||||
|
||||
# ── Unit tests ────────────────────────────────────────────────────
|
||||
|
||||
def test_next_sequence():
|
||||
facts = SAMPLE_FACTS[:2]
|
||||
seq = ks.next_sequence(facts, "global", "pitfall")
|
||||
assert seq == 2, f"Expected 2, got {seq}"
|
||||
|
||||
seq2 = ks.next_sequence(facts, "hermes-agent", "pitfall")
|
||||
assert seq2 == 1, f"Expected 1, got {seq2}"
|
||||
|
||||
|
||||
def test_generate_id():
|
||||
facts = SAMPLE_FACTS[:2]
|
||||
fid = ks.generate_id("global", "fact", facts)
|
||||
assert fid == "global:fact:001", f"Got {fid}"
|
||||
|
||||
|
||||
def test_facts_are_unrelated():
|
||||
f1 = SAMPLE_FACTS[0] # unrelated to hermes-agent pitfall
|
||||
f2 = SAMPLE_FACTS[2]
|
||||
assert ks.facts_are_unrelated(f1, f2) is True
|
||||
|
||||
f3 = SAMPLE_FACTS[1] # related to f1
|
||||
assert ks.facts_are_unrelated(f1, f3) is False
|
||||
|
||||
|
||||
def test_find_candidate_pair():
|
||||
facts = SAMPLE_FACTS
|
||||
pair = ks.find_candidate_pair(facts)
|
||||
assert pair is not None, "Should find an unrelated pair"
|
||||
f1, f2 = pair
|
||||
assert ks.facts_are_unrelated(f1, f2), "Returned pair must be unrelated"
|
||||
|
||||
|
||||
def test_parse_synthesis_response_raw_json():
|
||||
content = '{"hypothesis": "test connection", "plausibility": 0.8, "bridging_concepts": ["x"], "suggested_tags": ["a"]}'
|
||||
result = ks.parse_synthesis_response(content)
|
||||
assert result is not None
|
||||
assert result["hypothesis"] == "test connection"
|
||||
assert result["plausibility"] == 0.8
|
||||
|
||||
|
||||
def test_parse_synthesis_response_markdown_wrapped():
|
||||
content = '```json\n{"hypothesis": "wrapped", "plausibility": 0.5}\n```'
|
||||
result = ks.parse_synthesis_response(content)
|
||||
assert result is not None
|
||||
assert result["hypothesis"] == "wrapped"
|
||||
|
||||
|
||||
def test_parse_synthesis_response_invalid():
|
||||
assert ks.parse_synthesis_response("not json") is None
|
||||
assert ks.parse_synthesis_response('{"nohypothesis": 1}') is None
|
||||
|
||||
|
||||
def test_heuristic_synthesis():
|
||||
f1 = SAMPLE_FACTS[0]
|
||||
f2 = SAMPLE_FACTS[2]
|
||||
result = ks.heuristic_synthesis(f1, f2)
|
||||
assert "hypothesis" in result
|
||||
assert "plausibility" in result
|
||||
assert result["plausibility"] == 0.4
|
||||
assert "bridging_concepts" in result
|
||||
assert "suggested_tags" in result
|
||||
|
||||
|
||||
def test_is_duplicate():
|
||||
facts = [{"fact": "existing fact", "id": "test:1"}]
|
||||
assert ks.is_duplicate("existing fact", facts) is True
|
||||
assert ks.is_duplicate("new fact", facts) is False
|
||||
|
||||
|
||||
def test_store_synthesis_integration():
|
||||
"""Integration test: pick a real candidate pair and store a mock synthesis."""
|
||||
with tempfile.TemporaryDirectory() as tmp:
|
||||
tmp_path = Path(tmp)
|
||||
# Create fake knowledge dir with index
|
||||
kdir = tmp_path / "knowledge"
|
||||
kdir.mkdir()
|
||||
index = {
|
||||
"version": 1,
|
||||
"last_updated": "2026-04-13T20:00:00Z",
|
||||
"total_facts": 3,
|
||||
"facts": SAMPLE_FACTS
|
||||
}
|
||||
with open(kdir / "index.json", "w") as f:
|
||||
json.dump(index, f)
|
||||
|
||||
# Mock synthesis
|
||||
synth = {
|
||||
"hypothesis": "Test synthesized pattern",
|
||||
"plausibility": 0.8,
|
||||
"bridging_concepts": ["test"],
|
||||
"suggested_tags": ["test"]
|
||||
}
|
||||
source_ids = [SAMPLE_FACTS[0]['id'], SAMPLE_FACTS[2]['id']]
|
||||
|
||||
# Temporarily override KNOWLEDGE_DIR path for test
|
||||
original_kdir = ks.KNOWLEDGE_DIR
|
||||
ks.KNOWLEDGE_DIR = kdir
|
||||
try:
|
||||
stored = ks.store_synthesis(synth, source_ids, index, threshold=0.5)
|
||||
assert stored is True
|
||||
assert index['total_facts'] == 4
|
||||
new_fact = index['facts'][-1]
|
||||
assert new_fact['fact'] == "Test synthesized pattern"
|
||||
assert new_fact['category'] == "pattern"
|
||||
assert new_fact['domain'] == "global"
|
||||
assert new_fact['related'] == source_ids
|
||||
assert new_fact['id'].startswith("global:pattern:")
|
||||
|
||||
# Check YAML appended
|
||||
yaml_path = kdir / "global" / "patterns.yaml"
|
||||
assert yaml_path.exists()
|
||||
content = yaml_path.read_text()
|
||||
assert "Test synthesized pattern" in content
|
||||
finally:
|
||||
ks.KNOWLEDGE_DIR = original_kdir
|
||||
|
||||
|
||||
# ── Smoke test ────────────────────────────────────────────────────
|
||||
|
||||
def test_smoke_synthesizer_info():
|
||||
"""Sanity check: script can at least load and report current knowledge state."""
|
||||
index = ks.load_index()
|
||||
total = index.get('total_facts', 0)
|
||||
facts = index.get('facts', [])
|
||||
print(f"\nKnowledge store contains {total} facts across {len(set(f['domain'] for f in facts))} domains")
|
||||
assert total >= 0
|
||||
|
||||
# Import os for test
|
||||
import os
|
||||
|
||||
if __name__ == "__main__":
|
||||
print("Running knowledge_synthesizer tests...\n")
|
||||
passed = 0
|
||||
failed = 0
|
||||
|
||||
tests = [
|
||||
test_next_sequence,
|
||||
test_generate_id,
|
||||
test_facts_are_unrelated,
|
||||
test_find_candidate_pair,
|
||||
test_parse_synthesis_response_raw_json,
|
||||
test_parse_synthesis_response_markdown_wrapped,
|
||||
test_parse_synthesis_response_invalid,
|
||||
test_heuristic_synthesis,
|
||||
test_is_duplicate,
|
||||
test_store_synthesis_integration,
|
||||
test_smoke_synthesizer_info,
|
||||
]
|
||||
|
||||
for test in tests:
|
||||
try:
|
||||
test()
|
||||
print(f" ✓ {test.__name__}")
|
||||
passed += 1
|
||||
except Exception as e:
|
||||
import traceback; traceback.print_exc(); print(f" ✗ {test.__name__}: {e}")
|
||||
failed += 1
|
||||
|
||||
print(f"\n{passed} passed, {failed} failed")
|
||||
sys.exit(0 if failed == 0 else 1)
|
||||
@@ -1,47 +0,0 @@
|
||||
# Knowledge Synthesis Prompt
|
||||
|
||||
## System Prompt
|
||||
|
||||
You are a knowledge synthesis engine. Given two facts, you generate a novel hypothesis
|
||||
that connects them in a way no human would typically link — a zero-shot creative leap.
|
||||
|
||||
## Task
|
||||
|
||||
FACT A:
|
||||
{fact_a}
|
||||
|
||||
FACT B:
|
||||
{fact_b}
|
||||
|
||||
Generate a single JSON object:
|
||||
|
||||
{
|
||||
"hypothesis": "one concise sentence linking the two facts as a new, testable insight",
|
||||
"plausibility": 0.0-1.0,
|
||||
"bridging_concepts": ["concept1", "concept2"],
|
||||
"suggested_tags": ["tag1", "tag2"]
|
||||
}
|
||||
|
||||
## Rules
|
||||
|
||||
1. The hypothesis must be a logical consequence of combining both facts.
|
||||
2. DO NOT restate either fact — produce genuinely new insight.
|
||||
3. Plausibility should reflect confidence given only these two facts.
|
||||
4. If no meaningful connection exists, return {"hypothesis":"","plausibility":0.0}.
|
||||
5. Output ONLY valid JSON — no markdown, no explanation.
|
||||
|
||||
## Examples
|
||||
|
||||
Input facts:
|
||||
- "Gitea PR creation requires branch protection approval (1+) on main"
|
||||
- "Git push hangs on large repos (pack.windowMemory=100m)"
|
||||
|
||||
Hypothesis output:
|
||||
{
|
||||
"hypothesis": "Branch protection triggers checks that inflate pack size, causing git push to hang on large repos",
|
||||
"plausibility": 0.65,
|
||||
"bridging_concepts": ["git", "gitea", "branch-protection", "push"],
|
||||
"suggested_tags": ["git", "gitea", "performance"]
|
||||
}
|
||||
|
||||
Output ONLY the JSON object.
|
||||
222
tests/test_linter_runner.py
Normal file
222
tests/test_linter_runner.py
Normal file
@@ -0,0 +1,222 @@
|
||||
#!/usr/bin/env python3
|
||||
"""Tests for linter_runner module (Issue #155).
|
||||
|
||||
Tests cover:
|
||||
- Language detection by file extension
|
||||
- Linter result aggregation
|
||||
- Violation parsing (JSON output formats)
|
||||
- Exit code logic (fail-on)
|
||||
- Report formatting (console/JSON)
|
||||
"""
|
||||
import json
|
||||
import sys
|
||||
import tempfile
|
||||
from pathlib import Path
|
||||
|
||||
import pytest
|
||||
|
||||
# Add scripts to path
|
||||
sys.path.insert(0, str(Path(__file__).parent.parent / "scripts"))
|
||||
|
||||
from linter_runner import (
|
||||
Violation,
|
||||
LinterResult,
|
||||
detect_languages,
|
||||
parse_linter_output,
|
||||
_result_to_dict,
|
||||
EXTENSION_TO_LANGUAGE,
|
||||
LINTERS_BY_LANGUAGE,
|
||||
)
|
||||
|
||||
|
||||
class TestLanguageDetection:
|
||||
"""Test detect_languages() identifies languages correctly."""
|
||||
|
||||
def test_detects_python_files(self, tmp_path: Path):
|
||||
(tmp_path / "main.py").write_text("print('hello')")
|
||||
(tmp_path / "lib" / "utils.py").mkdir(parents=True)
|
||||
(tmp_path / "lib" / "utils.py").write_text("def foo(): pass")
|
||||
|
||||
result = detect_languages(tmp_path)
|
||||
assert "python" in result
|
||||
assert len(result["python"]) == 2
|
||||
|
||||
def test_detects_javascript_files(self, tmp_path: Path):
|
||||
(tmp_path / "app.js").write_text("console.log('hi')")
|
||||
(tmp_path / "component.jsx").write_text("<div/>")
|
||||
|
||||
result = detect_languages(tmp_path)
|
||||
assert "javascript" in result
|
||||
assert len(result["javascript"]) == 2
|
||||
|
||||
def test_detects_shell_files(self, tmp_path: Path):
|
||||
(tmp_path / "setup.sh").write_text("#!/bin/bash\necho hi")
|
||||
(tmp_path / "build.sh").write_text("make")
|
||||
|
||||
result = detect_languages(tmp_path)
|
||||
assert "shell" in result
|
||||
assert len(result["shell"]) == 2
|
||||
|
||||
def test_detects_yaml_files(self, tmp_path: Path):
|
||||
(tmp_path / "config.yml").write_text("key: value")
|
||||
(tmp_path / "env.yaml").write_text("env: test")
|
||||
|
||||
result = detect_languages(tmp_path)
|
||||
assert "yaml" in result
|
||||
assert len(result["yaml"]) == 2
|
||||
|
||||
def test_ignores_git_directory(self, tmp_path: Path):
|
||||
git_dir = tmp_path / ".git"
|
||||
git_dir.mkdir()
|
||||
(git_dir / "config").write_text("placeholder")
|
||||
(tmp_path / "script.py").write_text("print(1)")
|
||||
|
||||
result = detect_languages(tmp_path)
|
||||
assert "python" in result
|
||||
assert not any(".git" in str(f) for f in result.get("python", []))
|
||||
|
||||
def test_returns_empty_for_nonexistent_path(self):
|
||||
result = detect_languages(Path("/nonexistent/path/xyz"))
|
||||
assert result == {}
|
||||
|
||||
def test_mixed_languages(self, tmp_path: Path):
|
||||
(tmp_path / "app.py").write_text("")
|
||||
(tmp_path / "main.js").write_text("")
|
||||
(tmp_path / "deploy.sh").write_text("")
|
||||
|
||||
result = detect_languages(tmp_path)
|
||||
langs = set(result.keys())
|
||||
assert {"python", "javascript", "shell"} <= langs
|
||||
|
||||
def test_limits_files_to_known_languages(self, tmp_path: Path):
|
||||
(tmp_path / "readme.txt").write_text("text")
|
||||
(tmp_path / "data.csv").write_text("a,b,c")
|
||||
|
||||
result = detect_languages(tmp_path)
|
||||
assert len(result) == 0
|
||||
|
||||
|
||||
class TestViolationParsing:
|
||||
"""Test parse_linter_output parses various linter formats."""
|
||||
|
||||
def test_parses_pylint_json(self):
|
||||
stdout = json.dumps([
|
||||
{"type": "error", "module": "test.py", "line": 10, "column": 5,
|
||||
"message": "Missing docstring", "symbol": "missing-docstring"},
|
||||
{"type": "warning", "module": "test.py", "line": 15, "column": 1,
|
||||
"message": "Line too long", "symbol": "line-too-long"},
|
||||
])
|
||||
violations = parse_linter_output("pylint", stdout, "", Path("/repo"))
|
||||
assert len(violations) == 2
|
||||
assert violations[0].severity == "error"
|
||||
assert violations[0].message == "Missing docstring"
|
||||
assert violations[1].severity == "warning"
|
||||
assert violations[1].code == "line-too-long"
|
||||
|
||||
def test_parses_ruff_json(self):
|
||||
stdout = json.dumps([
|
||||
{"filename": "src/main.py", "location": {"row": 5, "column": 1},
|
||||
"code": "E501", "message": "Line too long"},
|
||||
])
|
||||
violations = parse_linter_output("ruff", stdout, "", Path("/repo"))
|
||||
assert len(violations) == 1
|
||||
assert violations[0].file == "src/main.py"
|
||||
assert violations[0].line == 5
|
||||
assert violations[0].code == "E501"
|
||||
|
||||
def test_parses_eslint_json(self):
|
||||
stdout = json.dumps([
|
||||
{"fileName": "app.js", "range": {"start": {"line": 2, "column": 0}},
|
||||
"message": "Unexpected console statement", "severity": 2, "ruleId": "no-console"},
|
||||
])
|
||||
violations = parse_linter_output("eslint", stdout, "", Path("/repo"))
|
||||
assert len(violations) == 1
|
||||
assert violations[0].severity == "error"
|
||||
assert violations[0].code == "no-console"
|
||||
|
||||
def test_parses_shellcheck_json1(self):
|
||||
stdout = json.dumps({
|
||||
"issues": [
|
||||
{"file": "script.sh", "line": 3, "column": 1,
|
||||
"message": "Quote this to prevent word splitting", "level": "warning", "code": "SC2086"},
|
||||
]
|
||||
})
|
||||
violations = parse_linter_output("shellcheck", stdout, "", Path("/repo"))
|
||||
assert len(violations) == 1
|
||||
assert violations[0].severity == "warning"
|
||||
assert violations[0].code == "SC2086"
|
||||
|
||||
def test_parses_yamllint_parsable(self):
|
||||
stdout = "config.yaml:3:1: [error] wrong document start (document-start)\n"
|
||||
violations = parse_linter_output("yamllint", stdout, "", Path("/repo"))
|
||||
assert len(violations) == 1
|
||||
assert violations[0].file == "config.yaml"
|
||||
assert violations[0].line == 3
|
||||
assert violations[0].severity == "error"
|
||||
assert violations[0].code == "document-start"
|
||||
|
||||
def test_returns_empty_on_invalid_json(self):
|
||||
stdout = "Not valid JSON"
|
||||
violations = parse_linter_output("pylint", stdout, "", Path("/repo"))
|
||||
assert violations == []
|
||||
|
||||
def test_strips_leading_slash_from_paths(self):
|
||||
stdout = json.dumps([{"type": "error", "module": "/repo/src/test.py",
|
||||
"line": 1, "column": 1, "message": "test", "symbol": "T001"}])
|
||||
violations = parse_linter_output("pylint", stdout, "", Path("/repo"))
|
||||
assert violations[0].file == "src/test.py"
|
||||
|
||||
|
||||
class TestLinterResult:
|
||||
"""Test LinterResult and JSON serialization."""
|
||||
|
||||
def test_result_to_dict_roundtrip(self):
|
||||
v = Violation(file="test.py", line=10, column=5, message="msg",
|
||||
severity="error", linter="pylint", code="E001")
|
||||
r = LinterResult(linter_name="pylint", language="python", violations=[v])
|
||||
d = _result_to_dict(r)
|
||||
assert d["linter"] == "pylint"
|
||||
assert d["violations"][0]["file"] == "test.py"
|
||||
assert d["violations"][0]["code"] == "E001"
|
||||
|
||||
|
||||
class TestIntegration:
|
||||
"""End-to-end integration tests with temporary repos."""
|
||||
|
||||
def test_linter_runner_accepts_repo_path(self, tmp_path: Path):
|
||||
(tmp_path / "main.py").write_text("print('hello')")
|
||||
(tmp_path / "bad.py").write_text("import unused_module\nx=1")
|
||||
|
||||
from linter_runner import detect_languages, run_linters_for_language
|
||||
|
||||
langs = detect_languages(tmp_path)
|
||||
assert "python" in langs
|
||||
|
||||
result = run_linters_for_language("python", langs["python"][:1], tmp_path)
|
||||
assert result.language == "python"
|
||||
assert result.violations or result.error # either linter output or not-installed
|
||||
|
||||
def test_json_output_structure(self, tmp_path: Path):
|
||||
(tmp_path / "script.py").write_text("print(1)")
|
||||
|
||||
from linter_runner import detect_languages, run_linters_for_language, _result_to_dict
|
||||
|
||||
langs = detect_languages(tmp_path)
|
||||
if "python" not in langs:
|
||||
pytest.skip("No Python files detected")
|
||||
|
||||
result = run_linters_for_language("python", langs["python"], tmp_path)
|
||||
report = {
|
||||
"repo": tmp_path.name,
|
||||
"languages": {"python": _result_to_dict(result)},
|
||||
"summary": {
|
||||
"total_issues": len(result.violations),
|
||||
"errors": sum(1 for v in result.violations if v.severity == "error"),
|
||||
},
|
||||
}
|
||||
json.dumps(report) # should not raise
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
print("Run: pytest tests/test_linter_runner.py -v")
|
||||
|
||||
Reference in New Issue
Block a user