feat: fix session_pair_harvester to use role/content format (#91)

- Harvester used old message fields (from/value) but Hermes sessions use role/content
- Import session_reader to normalize conversations properly
- Update extract function to operate on normalized role/content messages
- Change predecessor lookup from "human"/"gpt" to "user"/"assistant"
- Add comprehensive smoke tests (8 tests, all pass)
- Verify extraction from test_sessions: 11 pairs, avg ratio 8.13

scripts/linter_runner.py

@@ -1,530 +0,0 @@
-#!/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()

scripts/session_pair_harvester.py

@@ -22,114 +22,95 @@ import sys
 from pathlib import Path
 from typing import Optional
 
+from session_reader import extract_conversation, read_session
+
 
 def compute_hash(text: str) -> str:
     """Content hash for deduplication."""
     return hashlib.sha256(text.encode()).hexdigest()[:16]
 
 
-def extract_pairs_from_session(session_data: dict, min_ratio: float = 1.5,
+def extract_pairs_from_conversation(conversation: list, session_id: str, model: str,
+                                min_ratio: float = 1.5,
                                 min_response_words: int = 20) -> list:
-    """Extract terse→rich pairs from a single session object."""
+    """Extract terse→rich pairs from a normalized conversation."""
     pairs = []
-    conversations = session_data.get("conversations", [])
-    session_id = session_data.get("id", "unknown")
-    model = session_data.get("model", "unknown")
-
     seen_hashes = set()
 
-    for i, msg in enumerate(conversations):
-        # Look for assistant/gpt responses
-        if msg.get("from") not in ("gpt", "assistant"):
+    for i, msg in enumerate(conversation):
+        # Look for assistant responses
+        if msg.get('role') != 'assistant':
             continue
 
-        response_text = msg.get("value", "")
+        response_text = msg.get('content', '')
         if not response_text or len(response_text.split()) < min_response_words:
             continue
 
-        # Find the preceding human message
+        # Find the preceding user message
         prompt_text = ""
         for j in range(i - 1, -1, -1):
-            if conversations[j].get("from") == "human":
-                prompt_text = conversations[j].get("value", "")
+            if conversation[j].get('role') == 'user':
+                prompt_text = conversation[j].get('content', '')
                 break
 
         if not prompt_text:
             continue
 
         # Filter: skip tool results, system messages embedded as human
-        if prompt_text.startswith("{") and "output" in prompt_text[:100]:
-            continue  # likely a tool result
-        if prompt_text.startswith("# SOUL.md") or prompt_text.startswith("You are"):
-            continue  # system prompt leak
+        if prompt_text.startswith('{') and 'output' in prompt_text[:100]:
+            continue
+        if prompt_text.startswith('# SOUL.md') or prompt_text.startswith('You are'):
+            continue
 
         # Quality filters
         prompt_words = len(prompt_text.split())
         response_words = len(response_text.split())
 
-        # Must have meaningful length ratio
         if prompt_words == 0 or response_words == 0:
             continue
         ratio = response_words / prompt_words
         if ratio < min_ratio:
             continue
 
-        # Skip responses that are mostly code
-        code_blocks = response_text.count("```")
-        if code_blocks >= 4 and len(response_text.replace("```", "").strip()) < 50:
+        code_blocks = response_text.count('```')
+        if code_blocks >= 4 and len(response_text.replace('```', '').strip()) < 50:
             continue
 
-        # Skip responses with tool call artifacts
-        if "tool_call" in response_text[:100] or "function_call" in response_text[:100]:
+        if 'tool_call' in response_text[:100] or 'function_call' in response_text[:100]:
             continue
 
-        # Deduplicate by content hash
         content_hash = compute_hash(prompt_text + response_text[:200])
         if content_hash in seen_hashes:
             continue
         seen_hashes.add(content_hash)
 
-        # Clean up response: remove markdown headers if too many
         clean_response = response_text
 
         pairs.append({
-            "terse": prompt_text.strip(),
-            "rich": clean_response.strip(),
-            "source": session_id,
-            "model": model,
-            "prompt_words": prompt_words,
-            "response_words": response_words,
-            "ratio": round(ratio, 2),
+            'terse': prompt_text.strip(),
+            'rich': clean_response.strip(),
+            'source': session_id,
+            'model': model,
+            'prompt_words': prompt_words,
+            'response_words': response_words,
+            'ratio': round(ratio, 2),
         })
 
     return pairs
 
 
-def extract_from_jsonl_file(filepath: str, **kwargs) -> list:
-    """Extract pairs from a session JSONL file."""
-    pairs = []
-    path = Path(filepath)
 
-    if not path.exists():
-        print(f"Warning: {filepath} not found", file=sys.stderr)
-        return pairs
-
-    content = path.read_text()
-    lines = content.strip().split("\n")
-
-    for line in lines:
-        line = line.strip()
-        if not line:
-            continue
-        try:
-            session = json.loads(line)
-        except json.JSONDecodeError:
-            continue
-
-        session_pairs = extract_pairs_from_session(session, **kwargs)
-        pairs.extend(session_pairs)
-
-    return pairs
+def extract_from_jsonl_file(path: str, **kwargs) -> list:
+    """Read a session file and extract training pairs using normalized conversation."""
+    session_messages = read_session(path)
+    if not session_messages:
+        return []
+    conversation = extract_conversation(session_messages)
+    # Derive session_id and model from first real message metadata
+    first_msg = next((m for m in session_messages if m.get('role') or m.get('from')), {})
+    session_id = first_msg.get('meta_session_id', Path(path).name)
+    model = first_msg.get('model', 'unknown')
+    return extract_pairs_from_conversation(conversation, session_id, model, **kwargs)
 
 
 def deduplicate_pairs(pairs: list) -> list:

tests/test_linter_runner.py

@@ -1,222 +0,0 @@
-#!/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")
-

tests/test_session_pair_harvester.py

@@ -0,0 +1,118 @@
+"""
+Tests for session_pair_harvester — training pair extraction from sessions.
+"""
+
+import json
+import tempfile
+import unittest
+from pathlib import Path
+
+import sys
+from pathlib import Path
+sys.path.insert(0, str(Path(__file__).parent.parent / "scripts"))
+from session_pair_harvester import (
+    extract_pairs_from_conversation,
+    extract_from_jsonl_file,
+    deduplicate_pairs,
+    compute_hash,
+)
+
+
+class TestSessionPairHarvester(unittest.TestCase):
+    def test_compute_hash_consistent(self):
+        h1 = compute_hash("hello world")
+        h2 = compute_hash("hello world")
+        self.assertEqual(h1, h2)
+        self.assertEqual(len(h1), 16)
+
+    def test_extract_simple_qa_pair(self):
+        """A simple user→assistant exchange produces one pair."""
+        conversation = [
+            {"role": "user", "content": "What is the capital of France?"},
+            {"role": "assistant", "content": "The capital of France is Paris. It is a major European city renowned for its art, fashion, gastronomy, cultural heritage, and historical significance. The city attracts millions of tourists annually."},
+        ]
+        pairs = extract_pairs_from_conversation(conversation, "test_session", "test-model")
+        self.assertEqual(len(pairs), 1)
+        self.assertEqual(pairs[0]["terse"], "What is the capital of France?")
+        self.assertIn("Paris", pairs[0]["rich"])
+        self.assertEqual(pairs[0]["source"], "test_session")
+
+    def test_min_ratio_filter(self):
+        """Very short responses are filtered out."""
+        conversation = [
+            {"role": "user", "content": "Yes"},
+            {"role": "assistant", "content": "No."},
+        ]
+        # Default min_ratio = 1.5, min_words = 20 for response
+        pairs = extract_pairs_from_conversation(conversation, "s", "m", min_response_words=3)
+        self.assertEqual(len(pairs), 0)
+
+    def test_min_words_filter(self):
+        """Assistant responses below min word count are skipped."""
+        conversation = [
+            {"role": "user", "content": "Explain the project architecture in detail"},
+            {"role": "assistant", "content": "OK."},
+        ]
+        pairs = extract_pairs_from_conversation(conversation, "s", "m", min_response_words=5)
+        self.assertEqual(len(pairs), 0)
+
+    def test_skip_non_assistant_messages(self):
+        """System and tool messages are ignored."""
+        conversation = [
+            {"role": "system", "content": "You are a helpful assistant."},
+            {"role": "user", "content": "Hello"},
+            {"role": "assistant", "content": "Hi there! How can I help you today?"},
+        ]
+        pairs = extract_pairs_from_conversation(conversation, "s", "m", min_response_words=3)
+        self.assertEqual(len(pairs), 1)
+        self.assertEqual(pairs[0]["terse"], "Hello")
+
+    def test_multiple_pairs_from_one_session(self):
+        """A conversation with several Q&A turns yields multiple pairs."""
+        conversation = [
+            {"role": "user", "content": "First question?"},
+            {"role": "assistant", "content": "Here is a detailed and comprehensive answer that thoroughly explores multiple aspects of the subject. It provides background context and practical implications for the reader."},
+            {"role": "user", "content": "Second?"},
+            {"role": "assistant", "content": "Another comprehensive response with detailed examples. This includes practical code blocks and thorough explanations to ensure deep understanding of the topic at hand."},
+        ]
+        pairs = extract_pairs_from_conversation(conversation, "s", "m", min_ratio=1.0)
+        self.assertEqual(len(pairs), 2)
+
+    def test_deduplication_removes_duplicates(self):
+        """Identical pairs across sessions are deduplicated."""
+        pairs = [
+            {"terse": "q1", "rich": "a1", "source": "s1", "model": "m"},
+            {"terse": "q1", "rich": "a1", "source": "s2", "model": "m"},
+            {"terse": "q2", "rich": "a2", "source": "s1", "model": "m"},
+        ]
+        unique = deduplicate_pairs(pairs)
+        self.assertEqual(len(unique), 2)
+        sources = {p["source"] for p in unique}
+        # First unique pair can be from either s1 or s2
+        self.assertIn("s1", sources)
+
+    def test_integration_with_test_sessions(self):
+        """Harvester finds pairs in real test session files."""
+        repo_root = Path(__file__).parent.parent
+        test_sessions_dir = repo_root / "test_sessions"
+        if not test_sessions_dir.exists():
+            self.skipTest("test_sessions not found")
+
+        pairs = []
+        for jsonl_file in sorted(test_sessions_dir.glob("*.jsonl")):
+            pairs.extend(extract_from_jsonl_file(str(jsonl_file)))
+
+        self.assertGreater(len(pairs), 0, "Should extract at least one pair from test_sessions")
+        for p in pairs:
+            self.assertIn("terse", p)
+            self.assertIn("rich", p)
+            self.assertIn("source", p)
+            self.assertIn("model", p)
+            # Verify content exists
+            self.assertGreater(len(p["terse"]), 0)
+            self.assertGreater(len(p["rich"]), 0)
+
+
+if __name__ == "__main__":
+    unittest.main()
+