feat: add sampler.py — session value scorer (#17)

scripts/license_checker.py

@@ -1,506 +0,0 @@
-#!/usr/bin/env python3
-"""
-License Checker — Pipeline 5.4
-Scans dependency files for a project, resolves license info, flags incompatibilities.
-
-Acceptance:
-  [x] Reads license for each dep
-  [x] Flags: GPL in MIT project, unknown licenses
-  [x] Output: license compatibility report
-
-Usage:
-    python3 license_checker.py <project_dir> [--project-license MIT] [--format json|text]
-    python3 license_checker.py <project_dir> --scan-deps
-"""
-
-import argparse
-import json
-import os
-import re
-import subprocess
-import sys
-import urllib.request
-import urllib.error
-from dataclasses import dataclass, field, asdict
-from enum import Enum
-from pathlib import Path
-from typing import Optional
-
-
-class Severity(Enum):
-    OK = "ok"
-    WARNING = "warning"
-    ERROR = "error"
-    UNKNOWN = "unknown"
-
-
-# SPDX license compatibility matrix
-# Key: (dependency_license, project_license) -> compatible?
-# Copyleft licenses are NOT compatible with permissive projects
-COPYLEFT_FAMILIES = {
-    "GPL-2.0", "GPL-2.0-only", "GPL-2.0-or-later",
-    "GPL-3.0", "GPL-3.0-only", "GPL-3.0-or-later",
-    "AGPL-3.0", "AGPL-3.0-only", "AGPL-3.0-or-later",
-    "LGPL-2.0", "LGPL-2.1", "LGPL-3.0",
-    "LGPL-2.0-only", "LGPL-2.1-only", "LGPL-3.0-only",
-    "LGPL-2.0-or-later", "LGPL-2.1-or-later", "LGPL-3.0-or-later",
-    "MPL-2.0",  # Weak copyleft — file-level
-    "EUPL-1.1", "EUPL-1.2",
-    "OSL-3.0",
-    "SSPL-1.0",
-    "CC-BY-SA-4.0", "CC-BY-SA-3.0",
-    "CC-BY-NC-4.0", "CC-BY-NC-3.0",
-}
-
-PERMISSIVE_LICENSES = {
-    "MIT", "BSD-2-Clause", "BSD-3-Clause", "Apache-2.0",
-    "ISC", "Unlicense", "CC0-1.0", "0BSD", "BSL-1.0",
-    "Zlib", "PSF-2.0", "Python-2.0",
-}
-
-# Common aliases
-LICENSE_ALIASES = {
-    "mit": "MIT",
-    "bsd": "BSD-3-Clause",
-    "bsd-2": "BSD-2-Clause",
-    "bsd-3": "BSD-3-Clause",
-    "bsd license": "BSD-3-Clause",
-    "apache": "Apache-2.0",
-    "apache 2.0": "Apache-2.0",
-    "apache-2.0": "Apache-2.0",
-    "apache software license": "Apache-2.0",
-    "apache software license 2.0": "Apache-2.0",
-    "gpl": "GPL-3.0",
-    "gpl-2": "GPL-2.0",
-    "gpl-3": "GPL-3.0",
-    "gplv2": "GPL-2.0",
-    "gplv3": "GPL-3.0",
-    "gnu general public license": "GPL-3.0",
-    "gnu general public license v3": "GPL-3.0",
-    "gnu general public license v2": "GPL-2.0",
-    "gnu lesser general public license v2": "LGPL-2.1",
-    "gnu lesser general public license v3": "LGPL-3.0",
-    "lgpl": "LGPL-3.0",
-    "lgpl-2.1": "LGPL-2.1",
-    "lgpl-3": "LGPL-3.0",
-    "agpl": "AGPL-3.0",
-    "agpl-3.0": "AGPL-3.0",
-    "agplv3": "AGPL-3.0",
-    "isc": "ISC",
-    "mpl": "MPL-2.0",
-    "mpl-2.0": "MPL-2.0",
-    "mozilla public license 2.0": "MPL-2.0",
-    "unlicense": "Unlicense",
-    "public domain": "Unlicense",
-    "cc0": "CC0-1.0",
-    "cc0-1.0": "CC0-1.0",
-    "psf": "PSF-2.0",
-    "python software foundation license": "PSF-2.0",
-    "the mit license": "MIT",
-    "mit license": "MIT",
-}
-
-
-@dataclass
-class DepLicense:
-    name: str
-    version: str = ""
-    license: str = "UNKNOWN"
-    source: str = ""  # where we found the dep (requirements.txt, package.json, etc.)
-    severity: Severity = Severity.UNKNOWN
-    message: str = ""
-
-
-@dataclass
-class LicenseReport:
-    project_dir: str
-    project_license: str = "MIT"
-    dependencies: list = field(default_factory=list)
-    summary: dict = field(default_factory=dict)
-    errors: list = field(default_factory=list)
-    warnings: list = field(default_factory=list)
-
-
-def normalize_license(raw: str) -> str:
-    """Normalize a license string to SPDX identifier."""
-    if not raw or raw.strip() in ("UNKNOWN", "UNKNOWN:", ""):
-        return "UNKNOWN"
-    cleaned = raw.strip().lower()
-    # Remove version specifiers like "MIT License (MIT)"
-    cleaned = re.sub(r"\(.*?\)", "", cleaned).strip()
-    cleaned = re.sub(r"\s+license$", "", cleaned).strip()
-    cleaned = re.sub(r"^the\s+", "", cleaned).strip()
-    if cleaned in LICENSE_ALIASES:
-        return LICENSE_ALIASES[cleaned]
-    # Check if it already looks like SPDX
-    upper = raw.strip()
-    if upper in COPYLEFT_FAMILIES or upper in PERMISSIVE_LICENSES:
-        return upper
-    return raw.strip()
-
-
-def check_compatibility(dep_license: str, project_license: str) -> tuple[Severity, str]:
-    """Check if a dependency license is compatible with the project license."""
-    if dep_license == "UNKNOWN":
-        return Severity.WARNING, "License unknown — manual review needed"
-    
-    if dep_license in PERMISSIVE_LICENSES:
-        return Severity.OK, "Compatible (permissive)"
-    
-    if dep_license in COPYLEFT_FAMILIES:
-        # Copyleft in a permissive project is a problem
-        if project_license in PERMISSIVE_LICENSES:
-            return Severity.ERROR, f"Copyleft ({dep_license}) in permissive ({project_license}) project"
-        # Copyleft in same family is OK
-        if dep_license.startswith(project_license.split("-")[0]):
-            return Severity.OK, "Compatible (same copyleft family)"
-        return Severity.WARNING, f"Review needed: {dep_license} with {project_license}"
-    
-    return Severity.UNKNOWN, f"Unrecognized license: {dep_license}"
-
-
-def parse_requirements_txt(path: str) -> list[DepLicense]:
-    """Parse requirements.txt format."""
-    deps = []
-    with open(path) as f:
-        for line in f:
-            line = line.strip()
-            if not line or line.startswith("#") or line.startswith("-"):
-                continue
-            # Parse name==version or name>=version etc.
-            match = re.match(r"^([a-zA-Z0-9_.-]+)(?:[>=<!~].*)?$", line)
-            if match:
-                deps.append(DepLicense(name=match.group(1), source="requirements.txt"))
-    return deps
-
-
-def parse_pyproject_toml(path: str) -> list[DepLicense]:
-    """Parse pyproject.toml dependencies."""
-    deps = []
-    try:
-        # Use tomllib (Python 3.11+) or fall back to regex
-        import tomllib
-        with open(path, "rb") as f:
-            data = tomllib.load(f)
-    except ImportError:
-        # Fallback: regex parse
-        with open(path) as f:
-            content = f.read()
-        # Find [project.dependencies] section
-        match = re.search(r"\[project\]\s*dependencies\s*=\s*\[(.*?)\]", content, re.DOTALL)
-        if match:
-            for dep_str in re.findall(r'"([^"]+)"', match.group(1)):
-                name = re.match(r"^([a-zA-Z0-9_.-]+)", dep_str)
-                if name:
-                    deps.append(DepLicense(name=name.group(1), source="pyproject.toml"))
-        return deps
-
-    project_deps = data.get("project", {}).get("dependencies", [])
-    for dep_str in project_deps:
-        name = re.match(r"^([a-zA-Z0-9_.-]+)", dep_str)
-        if name:
-            deps.append(DepLicense(name=name.group(1), source="pyproject.toml"))
-    return deps
-
-
-def parse_package_json(path: str) -> list[DepLicense]:
-    """Parse package.json dependencies."""
-    deps = []
-    with open(path) as f:
-        data = json.load(f)
-    for section in ("dependencies", "devDependencies"):
-        for name, version in data.get(section, {}).items():
-            deps.append(DepLicense(name=name, version=version, source="package.json"))
-    return deps
-
-
-def parse_cargo_toml(path: str) -> list[DepLicense]:
-    """Parse Cargo.toml dependencies (basic)."""
-    deps = []
-    with open(path) as f:
-        for line in f:
-            match = re.match(r'^([a-zA-Z0-9_-]+)\s*=\s*"', line.strip())
-            if match and line.strip()[0] != "[" and line.strip() != "[dependencies]":
-                deps.append(DepLicense(name=match.group(1), source="Cargo.toml"))
-    return deps
-
-
-def parse_go_mod(path: str) -> list[DepLicense]:
-    """Parse go.mod dependencies."""
-    deps = []
-    with open(path) as f:
-        in_require = False
-        for line in f:
-            line = line.strip()
-            if line == "require (":
-                in_require = True
-                continue
-            if line == ")" and in_require:
-                in_require = False
-                continue
-            if in_require:
-                parts = line.split()
-                if len(parts) >= 2:
-                    deps.append(DepLicense(name=parts[0], version=parts[1], source="go.mod"))
-    return deps
-
-
-def scan_dep_files(project_dir: str) -> list[DepLicense]:
-    """Find and parse all dependency files in a project."""
-    all_deps = []
-    parsers = {
-        "requirements.txt": parse_requirements_txt,
-        "requirements-dev.txt": parse_requirements_txt,
-        "requirements_prod.txt": parse_requirements_txt,
-        "pyproject.toml": parse_pyproject_toml,
-        "setup.py": None,  # TODO: parse setup.py
-        "package.json": parse_package_json,
-        "Cargo.toml": parse_cargo_toml,
-        "go.mod": parse_go_mod,
-    }
-    
-    for filename, parser in parsers.items():
-        path = os.path.join(project_dir, filename)
-        if os.path.exists(path) and parser:
-            try:
-                deps = parser(path)
-                all_deps.extend(deps)
-            except Exception as e:
-                print(f"Warning: Failed to parse {filename}: {e}", file=sys.stderr)
-    
-    # Also check subdirectories for monorepos (one level deep)
-    for entry in os.listdir(project_dir):
-        subdir = os.path.join(project_dir, entry)
-        if os.path.isdir(subdir) and not entry.startswith("."):
-            for filename, parser in parsers.items():
-                path = os.path.join(subdir, filename)
-                if os.path.exists(path) and parser:
-                    try:
-                        deps = parser(path)
-                        for d in deps:
-                            d.source = f"{entry}/{filename}"
-                        all_deps.extend(deps)
-                    except Exception:
-                        pass
-    
-    return all_deps
-
-
-def lookup_pypi_license(package_name: str) -> str:
-    """Look up license from PyPI API."""
-    try:
-        url = f"https://pypi.org/pypi/{package_name}/json"
-        req = urllib.request.Request(url, headers={"Accept": "application/json"})
-        resp = urllib.request.urlopen(req, timeout=10)
-        data = json.loads(resp.read())
-        # Try classifiers first
-        for classifier in data.get("info", {}).get("classifiers", []):
-            if classifier.startswith("License ::"):
-                parts = classifier.split(" :: ")
-                if len(parts) >= 3:
-                    return parts[-1]
-        # Fall back to license field
-        lic = data.get("info", {}).get("license", "")
-        if lic and len(lic) < 100:
-            return lic
-        # Try license_expression
-        le = data.get("info", {}).get("license_expression", "")
-        if le:
-            return le
-        return "UNKNOWN"
-    except Exception:
-        return "UNKNOWN"
-
-
-def lookup_npm_license(package_name: str) -> str:
-    """Look up license from npm registry."""
-    try:
-        url = f"https://registry.npmjs.org/{package_name}"
-        req = urllib.request.Request(url, headers={"Accept": "application/json"})
-        resp = urllib.request.urlopen(req, timeout=10)
-        data = json.loads(resp.read())
-        lic = data.get("license", "UNKNOWN")
-        if isinstance(lic, dict):
-            lic = lic.get("type", "UNKNOWN")
-        return lic or "UNKNOWN"
-    except Exception:
-        return "UNKNOWN"
-
-
-def detect_project_license(project_dir: str) -> str:
-    """Detect the project's own license."""
-    for name in ("LICENSE", "LICENSE.md", "LICENSE.txt", "LICENCE", "COPYING"):
-        path = os.path.join(project_dir, name)
-        if os.path.exists(path):
-            with open(path) as f:
-                content = f.read().upper()
-            if "MIT LICENSE" in content or "MIT" in content[:200]:
-                return "MIT"
-            if "APACHE" in content and "2.0" in content:
-                return "Apache-2.0"
-            if "GNU GENERAL PUBLIC LICENSE" in content:
-                if "VERSION 3" in content:
-                    return "GPL-3.0"
-                if "VERSION 2" in content:
-                    return "GPL-2.0"
-            if "BSD" in content[:500]:
-                if "3-CLAUSE" in content or "THREE CLAUSE" in content:
-                    return "BSD-3-Clause"
-                return "BSD-2-Clause"
-            if "ISC" in content[:200]:
-                return "ISC"
-    # Check pyproject.toml
-    pypath = os.path.join(project_dir, "pyproject.toml")
-    if os.path.exists(pypath):
-        with open(pypath) as f:
-            content = f.read()
-        match = re.search(r'license\s*=\s*\{\s*text\s*=\s*"([^"]+)"', content)
-        if match:
-            return normalize_license(match.group(1))
-        match = re.search(r'license\s*=\s*"([^"]+)"', content)
-        if match:
-            return normalize_license(match.group(1))
-    return "UNKNOWN"
-
-
-def resolve_licenses(deps: list[DepLicense], cache: dict = None) -> None:
-    """Resolve license info for all dependencies."""
-    if cache is None:
-        cache = {}
-    
-    for dep in deps:
-        if dep.name in cache:
-            dep.license = cache[dep.name]
-            continue
-        
-        # Determine ecosystem
-        if dep.source in ("package.json",):
-            raw = lookup_npm_license(dep.name)
-        else:
-            raw = lookup_pypi_license(dep.name)
-        
-        dep.license = normalize_license(raw)
-        cache[dep.name] = dep.license
-
-
-def generate_report(deps: list[DepLicense], project_license: str) -> LicenseReport:
-    """Generate the compatibility report."""
-    report = LicenseReport(
-        project_dir="",
-        project_license=project_license,
-        dependencies=[],
-    )
-    
-    counts = {"ok": 0, "warning": 0, "error": 0, "unknown": 0}
-    
-    for dep in deps:
-        severity, message = check_compatibility(dep.license, project_license)
-        dep.severity = severity
-        dep.message = message
-        counts[severity.value] += 1
-        
-        if severity == Severity.ERROR:
-            report.errors.append(f"{dep.name}: {message}")
-        elif severity == Severity.WARNING:
-            report.warnings.append(f"{dep.name}: {message}")
-        
-        report.dependencies.append(asdict(dep))
-    
-    report.summary = {
-        "total": len(deps),
-        **counts,
-        "project_license": project_license,
-    }
-    
-    return report
-
-
-def format_text(report: LicenseReport) -> str:
-    """Format report as human-readable text."""
-    lines = []
-    lines.append("=" * 60)
-    lines.append("  LICENSE COMPATIBILITY REPORT")
-    lines.append("=" * 60)
-    lines.append(f"  Project License: {report.project_license}")
-    lines.append(f"  Dependencies: {report.summary.get('total', 0)}")
-    lines.append(f"  OK: {report.summary.get('ok', 0)}  "
-                 f"WARN: {report.summary.get('warning', 0)}  "
-                 f"ERR: {report.summary.get('error', 0)}  "
-                 f"UNK: {report.summary.get('unknown', 0)}")
-    lines.append("-" * 60)
-    
-    for dep in report.dependencies:
-        icon = {"ok": "[OK]", "warning": "[!!]", "error": "[XX]", "unknown": "[??]"}
-        sev = dep.get("severity", "unknown")
-        name = dep.get("name", "?")
-        lic = dep.get("license", "?")
-        msg = dep.get("message", "")
-        lines.append(f"  {icon.get(sev, '[ ]')} {name:30s} {lic:20s} {msg}")
-    
-    if report.errors:
-        lines.append("-" * 60)
-        lines.append("  ERRORS:")
-        for e in report.errors:
-            lines.append(f"    - {e}")
-    
-    if report.warnings:
-        lines.append("-" * 60)
-        lines.append("  WARNINGS:")
-        for w in report.warnings:
-            lines.append(f"    - {w}")
-    
-    lines.append("=" * 60)
-    return "\n".join(lines)
-
-
-def main():
-    parser = argparse.ArgumentParser(description="License Checker — Pipeline 5.4")
-    parser.add_argument("project_dir", help="Project directory to scan")
-    parser.add_argument("--project-license", default=None,
-                        help="Project license SPDX id (auto-detected if omitted)")
-    parser.add_argument("--format", choices=["json", "text"], default="text",
-                        help="Output format")
-    parser.add_argument("--scan-deps", action="store_true",
-                        help="Only scan and list deps (skip license lookup)")
-    args = parser.parse_args()
-    
-    project_dir = os.path.abspath(args.project_dir)
-    if not os.path.isdir(project_dir):
-        print(f"Error: {project_dir} is not a directory", file=sys.stderr)
-        sys.exit(1)
-    
-    # Detect project license
-    project_license = args.project_license or detect_project_license(project_dir)
-    
-    # Scan deps
-    deps = scan_dep_files(project_dir)
-    if not deps:
-        print(f"No dependencies found in {project_dir}", file=sys.stderr)
-        sys.exit(0)
-    
-    print(f"Found {len(deps)} dependencies", file=sys.stderr)
-    
-    if args.scan_deps:
-        for d in deps:
-            print(f"  {d.name} ({d.source})")
-        sys.exit(0)
-    
-    # Resolve licenses
-    print("Resolving licenses...", file=sys.stderr)
-    resolve_licenses(deps)
-    
-    # Generate report
-    report = generate_report(deps, project_license)
-    report.project_dir = project_dir
-    
-    if args.format == "json":
-        print(json.dumps(asdict(report), indent=2, default=str))
-    else:
-        print(format_text(report))
-    
-    # Exit code: 1 if errors, 0 otherwise
-    sys.exit(1 if report.errors else 0)
-
-
-if __name__ == "__main__":
-    main()

scripts/sampler.py

@@ -0,0 +1,353 @@
+#!/usr/bin/env python3
+"""
+sampler.py — Score and rank sessions by harvest value.
+
+With 20k+ sessions on disk, we can't harvest all at once. This script
+scores each session by how likely it is to contain valuable knowledge,
+so the harvester processes the best ones first.
+
+Scoring strategy:
+  - Recency: last 7d=3pts, last 30d=2pts, older=1pt
+  - Length: >50 messages=3pts, >20=2pts, <20=1pt
+  - Repo uniqueness: first session for a repo=5pts, otherwise=1pt
+  - Outcome: failure=3pts (most to learn), success=2pts, unknown=1pt
+  - Tool calls: >10 tool invocations=2pts (complex sessions)
+
+Usage:
+    python3 sampler.py --count 100                          # Top 100 sessions
+    python3 sampler.py --repo the-nexus --count 20          # Top 20 for a repo
+    python3 sampler.py --since 2026-04-01                   # All sessions since date
+    python3 sampler.py --count 50 --min-score 8             # Only high-value sessions
+    python3 sampler.py --count 100 --output sample.json     # Save to file
+"""
+
+import argparse
+import json
+import os
+import sys
+import time
+from datetime import datetime, timedelta, timezone
+from pathlib import Path
+from typing import Optional
+
+
+# --- Fast session scanning (no full parse) ---
+
+def scan_session_fast(path: str) -> dict:
+    """Extract scoring metadata from a session without parsing the full JSONL.
+    
+    Reads only: first line, last ~20 lines, and line count. This processes
+    20k sessions in seconds instead of minutes.
+    """
+    meta = {
+        'path': path,
+        'message_count': 0,
+        'has_tool_calls': False,
+        'tool_call_count': 0,
+        'first_timestamp': '',
+        'last_timestamp': '',
+        'is_failure': False,
+        'repos_mentioned': set(),
+        'first_role': '',
+        'last_content_preview': '',
+    }
+    
+    try:
+        file_size = os.path.getsize(path)
+        if file_size == 0:
+            return meta
+        
+        with open(path, 'r', encoding='utf-8', errors='replace') as f:
+            # Read first line for timestamp + role
+            first_line = f.readline().strip()
+            if first_line:
+                try:
+                    first_msg = json.loads(first_line)
+                    meta['first_timestamp'] = first_msg.get('timestamp', '')
+                    meta['first_role'] = first_msg.get('role', '')
+                except json.JSONDecodeError:
+                    pass
+            
+            # Fast line count + collect tail lines
+            # For the tail, seek to near end of file
+            tail_lines = []
+            line_count = 1  # already read first
+            
+            if file_size > 8192:
+                # Seek to last 8KB for tail sampling
+                f.seek(max(0, file_size - 8192))
+                f.readline()  # skip partial line
+                for line in f:
+                    line = line.strip()
+                    if line:
+                        tail_lines.append(line)
+                    line_count += 1
+                # We lost the exact count for big files — estimate from file size
+                # Average JSONL line is ~500 bytes
+                if line_count < 100:
+                    line_count = max(line_count, file_size // 500)
+            else:
+                # Small file — read all
+                for line in f:
+                    line = line.strip()
+                    if line:
+                        tail_lines.append(line)
+                    line_count += 1
+            
+            meta['message_count'] = line_count
+            
+            # Parse tail lines for outcome, tool calls, repos
+            for line in tail_lines[-30:]:  # last 30 non-empty lines
+                try:
+                    msg = json.loads(line)
+                    
+                    # Track last timestamp
+                    ts = msg.get('timestamp', '')
+                    if ts:
+                        meta['last_timestamp'] = ts
+                    
+                    # Count tool calls
+                    if msg.get('tool_calls'):
+                        meta['has_tool_calls'] = True
+                        meta['tool_call_count'] += len(msg['tool_calls'])
+                    
+                    # Detect failure signals in content
+                    content = ''
+                    if isinstance(msg.get('content'), str):
+                        content = msg['content'].lower()
+                    elif isinstance(msg.get('content'), list):
+                        for part in msg['content']:
+                            if isinstance(part, dict) and part.get('type') == 'text':
+                                content += part.get('text', '').lower()
+                    
+                    if content:
+                        meta['last_content_preview'] = content[:200]
+                        failure_signals = ['error', 'failed', 'cannot', 'unable', 
+                                          'exception', 'traceback', 'rejected', 'denied']
+                        if any(sig in content for sig in failure_signals):
+                            meta['is_failure'] = True
+                    
+                    # Extract repo references from tool call arguments
+                    if msg.get('tool_calls'):
+                        for tc in msg['tool_calls']:
+                            args = tc.get('function', {}).get('arguments', '')
+                            if isinstance(args, str):
+                                # Look for repo patterns
+                                for pattern in ['Timmy_Foundation/', 'Rockachopa/', 'compounding-intelligence', 'the-nexus', 'timmy-home', 'hermes-agent', 'the-beacon', 'the-door']:
+                                    if pattern in args:
+                                        repo = pattern.rstrip('/')
+                                        meta['repos_mentioned'].add(repo)
+                
+                except json.JSONDecodeError:
+                    continue
+    
+    except (IOError, OSError):
+        pass
+    
+    meta['repos_mentioned'] = list(meta['repos_mentioned'])
+    return meta
+
+
+# --- Filename timestamp parsing ---
+
+def parse_session_timestamp(filename: str) -> Optional[datetime]:
+    """Parse timestamp from session filename.
+    
+    Common formats:
+      session_20260413_123456_hash.jsonl
+      20260413_123456_hash.jsonl
+    """
+    stem = Path(filename).stem
+    parts = stem.split('_')
+    
+    # Try session_YYYYMMDD_HHMMSS format
+    for i, part in enumerate(parts):
+        if len(part) == 8 and part.isdigit():
+            date_part = part
+            time_part = parts[i + 1] if i + 1 < len(parts) and len(parts[i + 1]) == 6 else '000000'
+            try:
+                return datetime.strptime(f"{date_part}_{time_part}", '%Y%m%d_%H%M%S').replace(tzinfo=timezone.utc)
+            except ValueError:
+                continue
+    
+    # Fallback: use file modification time
+    return None
+
+
+# --- Scoring ---
+
+def score_session(meta: dict, now: datetime, seen_repos: set) -> tuple[int, dict]:
+    """Score a session for harvest value. Returns (score, breakdown)."""
+    score = 0
+    breakdown = {}
+    
+    # 1. Recency
+    ts = parse_session_timestamp(os.path.basename(meta['path']))
+    if ts is None:
+        # Fallback to mtime
+        try:
+            ts = datetime.fromtimestamp(os.path.getmtime(meta['path']), tz=timezone.utc)
+        except OSError:
+            ts = now - timedelta(days=365)
+    
+    age_days = (now - ts).days
+    if age_days <= 7:
+        recency = 3
+    elif age_days <= 30:
+        recency = 2
+    else:
+        recency = 1
+    score += recency
+    breakdown['recency'] = recency
+    
+    # 2. Length
+    count = meta['message_count']
+    if count > 50:
+        length = 3
+    elif count > 20:
+        length = 2
+    else:
+        length = 1
+    score += length
+    breakdown['length'] = length
+    
+    # 3. Repo uniqueness (first session mentioning a repo gets bonus)
+    repo_score = 0
+    for repo in meta.get('repos_mentioned', []):
+        if repo not in seen_repos:
+            seen_repos.add(repo)
+            repo_score = max(repo_score, 5)
+        else:
+            repo_score = max(repo_score, 1)
+    score += repo_score
+    breakdown['repo_unique'] = repo_score
+    
+    # 4. Outcome
+    if meta.get('is_failure'):
+        outcome = 3
+    elif meta.get('last_content_preview', '').strip():
+        outcome = 2  # has some content = likely completed
+    else:
+        outcome = 1
+    score += outcome
+    breakdown['outcome'] = outcome
+    
+    # 5. Tool calls
+    if meta.get('tool_call_count', 0) > 10:
+        tool = 2
+    else:
+        tool = 0
+    score += tool
+    breakdown['tool_calls'] = tool
+    
+    return score, breakdown
+
+
+# --- Main ---
+
+def main():
+    parser = argparse.ArgumentParser(description="Score and rank sessions for harvesting")
+    parser.add_argument('--sessions-dir', default=os.path.expanduser('~/.hermes/sessions'),
+                        help='Directory containing session files')
+    parser.add_argument('--count', type=int, default=100, help='Number of top sessions to return')
+    parser.add_argument('--repo', default='', help='Filter to sessions mentioning this repo')
+    parser.add_argument('--since', default='', help='Only score sessions after this date (YYYY-MM-DD)')
+    parser.add_argument('--min-score', type=int, default=0, help='Minimum score threshold')
+    parser.add_argument('--output', default='', help='Output file (JSON). Default: stdout')
+    parser.add_argument('--format', choices=['json', 'paths', 'table'], default='table',
+                        help='Output format: json (full), paths (one per line), table (human)')
+    parser.add_argument('--top-percent', type=float, default=0, help='Return top N%% instead of --count')
+    
+    args = parser.parse_args()
+    
+    sessions_dir = Path(args.sessions_dir)
+    if not sessions_dir.is_dir():
+        print(f"ERROR: Sessions directory not found: {sessions_dir}", file=sys.stderr)
+        sys.exit(1)
+    
+    # Find all JSONL files
+    print(f"Scanning {sessions_dir}...", file=sys.stderr)
+    t0 = time.time()
+    
+    session_files = list(sessions_dir.glob('*.jsonl'))
+    total = len(session_files)
+    print(f"Found {total} session files", file=sys.stderr)
+    
+    # Parse since date
+    since_dt = None
+    if args.since:
+        since_dt = datetime.strptime(args.since, '%Y-%m-%d').replace(tzinfo=timezone.utc)
+    
+    # Score all sessions
+    now = datetime.now(timezone.utc)
+    seen_repos = set()  # Track repos for uniqueness scoring
+    scored = []
+    
+    for i, sf in enumerate(session_files):
+        # Date filter (fast path: check filename first)
+        if since_dt:
+            ts = parse_session_timestamp(sf.name)
+            if ts and ts < since_dt:
+                continue
+        
+        meta = scan_session_fast(str(sf))
+        
+        # Repo filter
+        if args.repo:
+            repos = meta.get('repos_mentioned', [])
+            if args.repo.lower() not in [r.lower() for r in repos]:
+                # Also check filename
+                if args.repo.lower() not in sf.name.lower():
+                    continue
+        
+        score, breakdown = score_session(meta, now, seen_repos)
+        
+        if score >= args.min_score:
+            scored.append({
+                'path': str(sf),
+                'filename': sf.name,
+                'score': score,
+                'breakdown': breakdown,
+                'message_count': meta['message_count'],
+                'repos': meta['repos_mentioned'],
+                'is_failure': meta['is_failure'],
+            })
+        
+        if (i + 1) % 5000 == 0:
+            elapsed = time.time() - t0
+            print(f"  Scanned {i + 1}/{total} ({elapsed:.1f}s)", file=sys.stderr)
+    
+    elapsed = time.time() - t0
+    print(f"Scored {len(scored)} sessions in {elapsed:.1f}s", file=sys.stderr)
+    
+    # Sort by score descending
+    scored.sort(key=lambda x: x['score'], reverse=True)
+    
+    # Apply count or percent
+    if args.top_percent > 0:
+        count = max(1, int(len(scored) * args.top_percent / 100))
+    else:
+        count = args.count
+    scored = scored[:count]
+    
+    # Output
+    if args.output:
+        with open(args.output, 'w', encoding='utf-8') as f:
+            json.dump(scored, f, indent=2)
+        print(f"Wrote {len(scored)} sessions to {args.output}", file=sys.stderr)
+    elif args.format == 'json':
+        json.dump(scored, sys.stdout, indent=2)
+    elif args.format == 'paths':
+        for s in scored:
+            print(s['path'])
+    else:  # table
+        print(f"{'SCORE':>5}  {'MSGS':>5}  {'REPOS':<25}  {'FILE'}")
+        print(f"{'-'*5}  {'-'*5}  {'-'*25}  {'-'*40}")
+        for s in scored:
+            repos = ', '.join(s['repos'][:2]) if s['repos'] else '-'
+            fail = ' FAIL' if s['is_failure'] else ''
+            print(f"{s['score']:>5}  {s['message_count']:>5}  {repos:<25}  {s['filename'][:40]}{fail}")
+
+
+if __name__ == '__main__':
+    main()

tests/test_license_checker.py

@@ -1,186 +0,0 @@
-#!/usr/bin/env python3
-"""Tests for license_checker.py — Pipeline 5.4"""
-
-import json
-import os
-import sys
-import tempfile
-import unittest
-
-# Add scripts dir to path
-sys.path.insert(0, os.path.dirname(__file__))
-
-from license_checker import (
-    normalize_license,
-    check_compatibility,
-    parse_requirements_txt,
-    parse_package_json,
-    parse_pyproject_toml,
-    parse_go_mod,
-    detect_project_license,
-    scan_dep_files,
-    generate_report,
-    format_text,
-    Severity,
-    DepLicense,
-)
-
-
-class TestNormalizeLicense(unittest.TestCase):
-    def test_mit_aliases(self):
-        for alias in ["mit", "MIT License", "The MIT License", "MIT license"]:
-            self.assertEqual(normalize_license(alias), "MIT")
-
-    def test_apache_aliases(self):
-        for alias in ["Apache 2.0", "Apache-2.0", "apache software license"]:
-            self.assertEqual(normalize_license(alias), "Apache-2.0")
-
-    def test_gpl_aliases(self):
-        self.assertEqual(normalize_license("GPL-3.0"), "GPL-3.0")
-        self.assertEqual(normalize_license("gplv3"), "GPL-3.0")
-
-    def test_unknown(self):
-        self.assertEqual(normalize_license(""), "UNKNOWN")
-        self.assertEqual(normalize_license("UNKNOWN"), "UNKNOWN")
-
-    def test_already_spdx(self):
-        self.assertEqual(normalize_license("BSD-3-Clause"), "BSD-3-Clause")
-
-
-class TestCheckCompatibility(unittest.TestCase):
-    def test_permissive_ok(self):
-        sev, msg = check_compatibility("MIT", "MIT")
-        self.assertEqual(sev, Severity.OK)
-
-    def test_gpl_in_mit_error(self):
-        sev, msg = check_compatibility("GPL-3.0", "MIT")
-        self.assertEqual(sev, Severity.ERROR)
-
-    def test_unknown_warning(self):
-        sev, msg = check_compatibility("UNKNOWN", "MIT")
-        self.assertEqual(sev, Severity.WARNING)
-
-    def test_apache_in_mit_ok(self):
-        sev, msg = check_compatibility("Apache-2.0", "MIT")
-        self.assertEqual(sev, Severity.OK)
-
-    def test_lgpl_in_mit_error(self):
-        sev, msg = check_compatibility("LGPL-3.0", "MIT")
-        self.assertEqual(sev, Severity.ERROR)
-
-
-class TestParseRequirements(unittest.TestCase):
-    def test_basic(self):
-        with tempfile.NamedTemporaryFile(mode="w", suffix=".txt", delete=False) as f:
-            f.write("requests>=2.28.0\nflask==2.3.0\n# comment\npytest\n")
-            f.flush()
-            deps = parse_requirements_txt(f.name)
-        os.unlink(f.name)
-        names = [d.name for d in deps]
-        self.assertIn("requests", names)
-        self.assertIn("flask", names)
-        self.assertIn("pytest", names)
-        self.assertEqual(len(deps), 3)
-
-    def test_skip_flags(self):
-        with tempfile.NamedTemporaryFile(mode="w", suffix=".txt", delete=False) as f:
-            f.write("-r other.txt\n--index-url https://pypi.org\nreal-dep\n")
-            f.flush()
-            deps = parse_requirements_txt(f.name)
-        os.unlink(f.name)
-        self.assertEqual(len(deps), 1)
-        self.assertEqual(deps[0].name, "real-dep")
-
-
-class TestParsePackageJson(unittest.TestCase):
-    def test_basic(self):
-        data = {
-            "dependencies": {"express": "^4.18.0", "lodash": "^4.17.21"},
-            "devDependencies": {"jest": "^29.0.0"},
-        }
-        with tempfile.NamedTemporaryFile(mode="w", suffix=".json", delete=False) as f:
-            json.dump(data, f)
-            f.flush()
-            deps = parse_package_json(f.name)
-        os.unlink(f.name)
-        names = [d.name for d in deps]
-        self.assertIn("express", names)
-        self.assertIn("jest", names)
-        self.assertEqual(len(deps), 3)
-
-
-class TestParseGoMod(unittest.TestCase):
-    def test_basic(self):
-        content = """module example.com/mymod
-
-go 1.21
-
-require (
-    github.com/gin-gonic/gin v1.9.1
-    github.com/stretchr/testify v1.8.4
-)
-"""
-        with tempfile.NamedTemporaryFile(mode="w", suffix=".mod", delete=False) as f:
-            f.write(content)
-            f.flush()
-            deps = parse_go_mod(f.name)
-        os.unlink(f.name)
-        self.assertEqual(len(deps), 2)
-        self.assertEqual(deps[0].name, "github.com/gin-gonic/gin")
-
-
-class TestDetectProjectLicense(unittest.TestCase):
-    def test_mit_file(self):
-        with tempfile.TemporaryDirectory() as d:
-            with open(os.path.join(d, "LICENSE"), "w") as f:
-                f.write("MIT License\n\nCopyright (c) 2024...\n")
-            self.assertEqual(detect_project_license(d), "MIT")
-
-    def test_apache_file(self):
-        with tempfile.TemporaryDirectory() as d:
-            with open(os.path.join(d, "LICENSE"), "w") as f:
-                f.write("Apache License Version 2.0...")
-            self.assertEqual(detect_project_license(d), "Apache-2.0")
-
-    def test_no_license(self):
-        with tempfile.TemporaryDirectory() as d:
-            self.assertEqual(detect_project_license(d), "UNKNOWN")
-
-
-class TestScanDeps(unittest.TestCase):
-    def test_multi_ecosystem(self):
-        with tempfile.TemporaryDirectory() as d:
-            with open(os.path.join(d, "requirements.txt"), "w") as f:
-                f.write("flask\nrequests\n")
-            with open(os.path.join(d, "package.json"), "w") as f:
-                json.dump({"dependencies": {"express": "^4.0.0"}}, f)
-            deps = scan_dep_files(d)
-            names = [d.name for d in deps]
-            self.assertIn("flask", names)
-            self.assertIn("express", names)
-
-
-class TestGenerateReport(unittest.TestCase):
-    def test_basic(self):
-        deps = [
-            DepLicense(name="flask", license="BSD-3-Clause", source="requirements.txt"),
-            DepLicense(name="gpl-pkg", license="GPL-3.0", source="requirements.txt"),
-            DepLicense(name="unknown-pkg", license="UNKNOWN", source="requirements.txt"),
-        ]
-        report = generate_report(deps, "MIT")
-        self.assertEqual(report.summary["ok"], 1)
-        self.assertEqual(report.summary["error"], 1)
-        self.assertEqual(report.summary["warning"], 1)
-        self.assertEqual(len(report.errors), 1)
-        self.assertIn("gpl-pkg", report.errors[0])
-
-    def test_format_text(self):
-        deps = [DepLicense(name="flask", license="BSD-3-Clause", source="requirements.txt")]
-        report = generate_report(deps, "MIT")
-        text = format_text(report)
-        self.assertIn("LICENSE COMPATIBILITY REPORT", text)
-        self.assertIn("flask", text)
-
-
-if __name__ == "__main__":
-    unittest.main()