phd_research_os/obsidian_export.py

"""
PhD Research OS — Obsidian Export (Phase 4)
=============================================
One-directional export: System → Obsidian vault.
Generates Markdown with YAML frontmatter and wiki-links.
"""

import os
import json
from datetime import datetime
from pathlib import Path

from .db import get_db, search_claims, get_goals_by_priority, from_fixed


class ObsidianExporter:
    """
    Export Research OS data to Obsidian vault as linked Markdown notes.
    
    Structure:
        /vault/
        ├── Claims/
        │   └── CLM_XXXX.md
        ├── Sources/
        │   └── DOI_XXXX.md
        ├── Goals/
        │   └── GOAL_XXXX.md
        └── Dashboard.md
    """
    
    def __init__(self, vault_path: str = "vault", db_path: str = None):
        self.vault_path = Path(vault_path)
        self.db_path = db_path or os.environ.get("RESEARCH_OS_DB", "data/research_os.db")
        
        # Create directories
        for d in ["Claims", "Sources", "Goals", "Conflicts", "Decisions"]:
            (self.vault_path / d).mkdir(parents=True, exist_ok=True)
    
    def export_all(self):
        """Export everything to Obsidian vault. Idempotent — overwrites existing."""
        conn = get_db(self.db_path)
        
        print("Exporting to Obsidian vault...")
        
        # Export claims
        claims = conn.execute("SELECT * FROM claims ORDER BY claim_id").fetchall()
        for claim in claims:
            self._export_claim(dict(claim))
        print(f"  Claims: {len(claims)}")
        
        # Export sources
        sources = conn.execute("SELECT * FROM sources ORDER BY doi").fetchall()
        for source in sources:
            self._export_source(dict(source))
        print(f"  Sources: {len(sources)}")
        
        # Export goals
        goals = conn.execute("SELECT * FROM goals ORDER BY goal_id").fetchall()
        for goal in goals:
            self._export_goal(dict(goal))
        print(f"  Goals: {len(goals)}")
        
        # Export conflicts
        conflicts = conn.execute("SELECT * FROM conflicts ORDER BY conflict_id").fetchall()
        for conflict in conflicts:
            self._export_conflict(dict(conflict))
        print(f"  Conflicts: {len(conflicts)}")
        
        # Export decisions
        decisions = conn.execute("SELECT * FROM decisions ORDER BY decision_id").fetchall()
        for decision in decisions:
            self._export_decision(dict(decision))
        print(f"  Decisions: {len(decisions)}")
        
        # Generate dashboard
        self._export_dashboard(conn)
        print(f"  Dashboard: ✓")
        
        conn.close()
        print(f"\nVault exported to: {self.vault_path}")
    
    def _export_claim(self, claim: dict):
        """Export a single claim as Markdown with YAML frontmatter."""
        cid = claim['claim_id']
        confidence = from_fixed(claim['confidence']) if isinstance(claim['confidence'], int) else claim['confidence']
        missing = json.loads(claim.get('missing_fields', '[]'))
        params = json.loads(claim.get('parameters', '{}'))
        
        content = f"""---
claim_id: {cid}
source_doi: {claim.get('source_doi', 'unknown')}
epistemic_tag: {claim['epistemic_tag']}
confidence: {confidence}
status: {claim['status']}
study_type: {claim.get('study_type', 'unknown')}
is_canonical: {bool(claim.get('is_canonical', 0))}
schema_version: {claim.get('schema_version', '1.0')}
created_at: {claim.get('created_at', '')}
---

## Claim Text

{claim['text']}

## Parameters

{self._format_params(params)}

## Missing Fields

{', '.join(missing) if missing else 'None — claim is complete'}

## Linked Source

{f'[[{claim.get("source_doi", "")}]]' if claim.get('source_doi') else 'No source linked'}
"""
        
        if claim.get('expert_override'):
            override = json.loads(claim['expert_override']) if isinstance(claim['expert_override'], str) else claim['expert_override']
            content += f"""
## Expert Override

- **By:** {override.get('who', 'unknown')}
- **Rationale:** {override.get('rationale', '')}
- **Override ID:** {override.get('override_id', '')}
"""
        
        filepath = self.vault_path / "Claims" / f"{cid}.md"
        filepath.write_text(content)
    
    def _export_source(self, source: dict):
        """Export a source as Markdown."""
        doi = source['doi']
        authors = json.loads(source.get('authors', '[]'))
        
        safe_doi = doi.replace("/", "_").replace(".", "_")
        
        content = f"""---
doi: {doi}
title: "{source.get('title', '')}"
year: {source.get('year', '')}
journal: "{source.get('journal', '')}"
journal_tier: {source.get('journal_tier', '')}
is_canonical: {bool(source.get('is_canonical', 0))}
---

## {source.get('title', 'Untitled')}

**DOI:** [{doi}](https://doi.org/{doi})
**Authors:** {', '.join(authors)}
**Year:** {source.get('year', 'Unknown')}
**Journal:** {source.get('journal', 'Unknown')} (Tier {source.get('journal_tier', '?')})
**Study Type:** {source.get('study_type', 'Unknown')}
"""
        
        filepath = self.vault_path / "Sources" / f"{safe_doi}.md"
        filepath.write_text(content)
    
    def _export_goal(self, goal: dict):
        """Export a goal as Markdown."""
        gid = goal['goal_id']
        linked = json.loads(goal.get('linked_claim_ids', '[]'))
        
        content = f"""---
goal_id: {gid}
priority: {goal['priority']}
status: {goal['status']}
---

## {goal['description']}

**Priority:** {goal['priority']}
**Status:** {goal['status']}

## Linked Claims

{chr(10).join(f'- [[{cid}]]' for cid in linked) if linked else 'No claims linked yet'}
"""
        
        filepath = self.vault_path / "Goals" / f"{gid}.md"
        filepath.write_text(content)
    
    def _export_conflict(self, conflict: dict):
        """Export a conflict as Markdown."""
        content = f"""---
conflict_id: {conflict['conflict_id']}
conflict_type: {conflict['conflict_type']}
resolution_status: {conflict['resolution_status']}
hypothesis_confidence: {conflict.get('hypothesis_confidence', 'low')}
---

## Conflict: [[{conflict['claim_a_id']}]] vs [[{conflict['claim_b_id']}]]

**Type:** {conflict['conflict_type']}
**Status:** {conflict['resolution_status']}

## Generated Hypothesis

{conflict.get('generated_hypothesis', 'No hypothesis generated')}

⚠️ **Hypothesis confidence: LOW** — Requires human review

## Key Differences

{chr(10).join(f'- {d}' for d in json.loads(conflict.get('key_differences', '[]')))}
"""
        
        filepath = self.vault_path / "Conflicts" / f"{conflict['conflict_id']}.md"
        filepath.write_text(content)
    
    def _export_decision(self, decision: dict):
        """Export a decision as Markdown."""
        info_gain = from_fixed(decision['expected_information_gain']) if isinstance(decision['expected_information_gain'], int) else decision['expected_information_gain']
        linked = json.loads(decision.get('linked_claim_ids', '[]'))
        
        content = f"""---
decision_id: {decision['decision_id']}
action: {decision['recommended_action']}
priority: {decision.get('priority', 'medium')}
status: {decision['status']}
info_gain: {info_gain}
---

## {decision.get('action_description', 'No description')}

**Action:** {decision['recommended_action']}
**Priority:** {decision.get('priority', 'medium')}
**Status:** {decision['status']}
**Expected Info Gain:** {info_gain}
**Estimated Effort:** {decision.get('estimated_effort', 'Unknown')}

## Linked Goal

{f'[[{decision.get("linked_goal_id", "")}]]' if decision.get('linked_goal_id') else 'No goal linked'}

## Linked Claims

{chr(10).join(f'- [[{cid}]]' for cid in linked) if linked else 'None'}
"""
        
        if decision.get('why_not_log'):
            content += f"""
## Why Not Log (Rejection Reason)

{decision['why_not_log']}
"""
        
        filepath = self.vault_path / "Decisions" / f"{decision['decision_id']}.md"
        filepath.write_text(content)
    
    def _export_dashboard(self, conn):
        """Generate Dashboard.md with system summary."""
        # Gather stats
        claim_count = conn.execute("SELECT COUNT(*) FROM claims").fetchone()[0]
        source_count = conn.execute("SELECT COUNT(*) FROM sources").fetchone()[0]
        goal_count = conn.execute("SELECT COUNT(*) FROM goals WHERE status = 'Active'").fetchone()[0]
        conflict_count = conn.execute("SELECT COUNT(*) FROM conflicts WHERE resolution_status = 'Unresolved'").fetchone()[0]
        incomplete_count = conn.execute("SELECT COUNT(*) FROM claims WHERE status = 'Incomplete'").fetchone()[0]
        override_count = conn.execute("SELECT COUNT(*) FROM overrides").fetchone()[0]
        
        # Epistemic distribution
        epist = conn.execute(
            "SELECT epistemic_tag, COUNT(*) as cnt FROM claims GROUP BY epistemic_tag"
        ).fetchall()
        
        # Recent sources
        recent = conn.execute(
            "SELECT doi, title FROM sources ORDER BY created_at DESC LIMIT 10"
        ).fetchall()
        
        # High-confidence claims
        high_conf = conn.execute(
            "SELECT claim_id, text, confidence FROM claims ORDER BY confidence DESC LIMIT 10"
        ).fetchall()
        
        now = datetime.now().strftime("%Y-%m-%d %H:%M")
        
        content = f"""---
type: dashboard
updated: {now}
---

# PhD Research OS Dashboard

*Last updated: {now}*

## Overview

| Metric | Count |
|--------|-------|
| Total Claims | {claim_count} |
| Sources | {source_count} |
| Active Goals | {goal_count} |
| Unresolved Conflicts | {conflict_count} |
| Incomplete Claims | {incomplete_count} |
| Expert Overrides | {override_count} |

## Epistemic Distribution

| Tag | Count |
|-----|-------|
{chr(10).join(f'| {dict(e)["epistemic_tag"]} | {dict(e)["cnt"]} |' for e in epist)}

## Recent Ingestions

{chr(10).join(f'- [{dict(s)["title"][:60]}...]({dict(s)["doi"]})' for s in recent)}

## Top Confidence Claims

{chr(10).join(f'- [[{dict(c)["claim_id"]}]] ({from_fixed(dict(c)["confidence"]):.3f}): {dict(c)["text"][:80]}...' for c in high_conf)}

## ⚠️ Attention Required

### Incomplete Claims ({incomplete_count})
{'See Claims/ folder — filter by status: Incomplete' if incomplete_count else '✅ All claims complete'}

### Unresolved Conflicts ({conflict_count})
{'See Conflicts/ folder for items requiring human review' if conflict_count else '✅ No unresolved conflicts'}
"""
        
        filepath = self.vault_path / "Dashboard.md"
        filepath.write_text(content)
    
    def _format_params(self, params: dict) -> str:
        if not params:
            return "No parameters recorded"
        return "\n".join(f"- **{k}:** {v}" for k, v in params.items())