HarborForge.PlexumPlugin/internal/kbblock/kbblock.go

// Package kbblock implements the per-session knowledge-base block
// owned by the HarborForge Plexum plugin per Plexum DESIGN-DYNAMIC-
// BLOCK.md §3.3 / §4.4.
//
// The kb-block stores HarborForge KB facts the agent has cached for
// the current session. Each entry renders as
//
//	<kb-fact id=N kb=<code> source=topic:<slug>>content</kb-fact>
//
// (id = HF backend's DB primary key for the fact, NOT a per-session
// monotonic). InsertSeq controls cache-insertion-order rendering per
// §9 #4. Duplicate Add(id, ...) silently no-ops per §9 #4.
//
// Storage location: plugin-managed file at
//
//	<PLEXUM_PROFILE_ROOT>/agents/<agentID>/sessions/<sessionID>/
//	plugins/harbor-forge/kb-block.json
//
// Plexum core's session cleanup (del-session) cascades this directory
// alongside the rest of the session state.
//
// Fade-out (§9 #3 — KB fade params = memory fade params n=5/w=10/m=70)
// is NOT applied in v1. Phase 2 v1 stores raw content + agent
// dynamic-kb-evict is the only eviction path. Fade will be added once
// we have prod data on whether KB facts need it.
package kbblock

import (
	"encoding/json"
	"fmt"
	"os"
	"path/filepath"
	"sort"
	"strings"
	"sync"
)

const FileName = "kb-block.json"
const currentVersion = 1

// Entry is one cached KB fact.
type Entry struct {
	ID                int    `json:"id"`           // HF backend DB primary key
	KBCode            string `json:"kb_code"`      // e.g. "KB-PAYROT"
	SourceTopic       string `json:"source_topic"` // human slug from HF topic, e.g. "debugging"
	Content           string `json:"content"`
	InsertSeq         int    `json:"insert_seq"` // per-session monotonic, for render order
	AddedAtTurn       int    `json:"added_at_turn"`
	LastRefreshAtTurn int    `json:"last_refresh_at_turn"`
}

// Block is the in-memory representation of kb-block.json.
type Block struct {
	mu      sync.Mutex
	path    string
	NextSeq int      `json:"next_seq"`
	Entries []*Entry `json:"entries"`
}

// SessionDir returns the plugin-scoped session subdir under the Plexum
// profile root: <profileRoot>/agents/<agentID>/sessions/<sessionID>/
// plugins/harbor-forge.
func SessionDir(profileRoot, agentID, sessionID string) string {
	return filepath.Join(profileRoot, "agents", agentID, "sessions", sessionID, "plugins", "harbor-forge")
}

// Open loads the block at <SessionDir>/kb-block.json. Missing file →
// empty block (NextSeq=1). Caller is expected to Close-free; Block
// can be re-Opened cheaply per-call (small JSON).
func Open(profileRoot, agentID, sessionID string) (*Block, error) {
	if profileRoot == "" || agentID == "" || sessionID == "" {
		return nil, fmt.Errorf("kbblock.Open: profileRoot/agentID/sessionID all required")
	}
	path := filepath.Join(SessionDir(profileRoot, agentID, sessionID), FileName)
	b := &Block{path: path, NextSeq: 1}
	raw, err := os.ReadFile(path)
	if err != nil {
		if os.IsNotExist(err) {
			return b, nil
		}
		return nil, fmt.Errorf("kbblock: read %s: %w", path, err)
	}
	if len(raw) == 0 {
		return b, nil
	}
	var loaded struct {
		NextSeq int      `json:"next_seq"`
		Entries []*Entry `json:"entries"`
	}
	if err := json.Unmarshal(raw, &loaded); err != nil {
		return nil, fmt.Errorf("kbblock: parse %s: %w", path, err)
	}
	if loaded.NextSeq < 1 {
		loaded.NextSeq = 1
	}
	b.NextSeq = loaded.NextSeq
	b.Entries = loaded.Entries
	return b, nil
}

// Path returns the underlying file path.
func (b *Block) Path() string { return b.path }

// Len returns the entry count.
func (b *Block) Len() int {
	b.mu.Lock()
	defer b.mu.Unlock()
	return len(b.Entries)
}

// Has reports whether a fact with this ID is already cached.
func (b *Block) Has(id int) bool {
	b.mu.Lock()
	defer b.mu.Unlock()
	for _, e := range b.Entries {
		if e.ID == id {
			return true
		}
	}
	return false
}

// Add appends a new entry. Duplicate ID → silent no-op, returns nil.
// Does NOT persist; caller must Save.
func (b *Block) Add(id int, kbCode, sourceTopic, content string, atTurn int) *Entry {
	b.mu.Lock()
	defer b.mu.Unlock()
	for _, e := range b.Entries {
		if e.ID == id {
			return nil
		}
	}
	e := &Entry{
		ID:                id,
		KBCode:            kbCode,
		SourceTopic:       sourceTopic,
		Content:           content,
		InsertSeq:         b.NextSeq,
		AddedAtTurn:       atTurn,
		LastRefreshAtTurn: atTurn,
	}
	b.NextSeq++
	b.Entries = append(b.Entries, e)
	return e
}

// Remove drops entries by ID. Returns the IDs actually removed.
func (b *Block) Remove(ids ...int) []int {
	b.mu.Lock()
	defer b.mu.Unlock()
	want := map[int]bool{}
	for _, id := range ids {
		want[id] = true
	}
	kept := b.Entries[:0]
	var removed []int
	for _, e := range b.Entries {
		if want[e.ID] {
			removed = append(removed, e.ID)
			continue
		}
		kept = append(kept, e)
	}
	b.Entries = kept
	return removed
}

// Lookup returns the entry with the given ID, or nil.
func (b *Block) Lookup(id int) *Entry {
	b.mu.Lock()
	defer b.mu.Unlock()
	for _, e := range b.Entries {
		if e.ID == id {
			return e
		}
	}
	return nil
}

// AllIDs returns the cached fact IDs sorted by InsertSeq ascending.
func (b *Block) AllIDs() []int {
	b.mu.Lock()
	defer b.mu.Unlock()
	out := make([]*Entry, len(b.Entries))
	copy(out, b.Entries)
	sort.Slice(out, func(i, j int) bool { return out[i].InsertSeq < out[j].InsertSeq })
	ids := make([]int, len(out))
	for i, e := range out {
		ids[i] = e.ID
	}
	return ids
}

// Save atomically writes (tmp+rename, 0600). Empty block with no prior
// file = no-op. Parent dirs auto-created.
func (b *Block) Save() error {
	b.mu.Lock()
	defer b.mu.Unlock()
	if len(b.Entries) == 0 {
		if _, err := os.Stat(b.path); os.IsNotExist(err) {
			return nil
		}
	}
	if err := os.MkdirAll(filepath.Dir(b.path), 0o755); err != nil {
		return err
	}
	payload := struct {
		Version int      `json:"version"`
		NextSeq int      `json:"next_seq"`
		Entries []*Entry `json:"entries"`
	}{Version: currentVersion, NextSeq: b.NextSeq, Entries: b.Entries}
	data, err := json.MarshalIndent(payload, "", "  ")
	if err != nil {
		return err
	}
	tmp := b.path + ".tmp"
	if err := os.WriteFile(tmp, data, 0o600); err != nil {
		return fmt.Errorf("kbblock: write tmp: %w", err)
	}
	return os.Rename(tmp, b.path)
}

// Render returns the inner body of the <kb-block> subblock — a flat
// sequence of `<kb-fact id=N kb=<code> source=topic:<slug>>\n<content>
// \n</kb-fact>` separated by single blank lines, ordered by InsertSeq
// (DESIGN-DYNAMIC-BLOCK.md §9 #4). Empty block returns "".
//
// The plugin's DynamicBlockProvider.RenderDynamicSubblock returns this
// to the Plexum host, which wraps in <kb-block>...</kb-block>.
func (b *Block) Render() string {
	b.mu.Lock()
	defer b.mu.Unlock()
	if len(b.Entries) == 0 {
		return ""
	}
	ordered := make([]*Entry, len(b.Entries))
	copy(ordered, b.Entries)
	sort.Slice(ordered, func(i, j int) bool { return ordered[i].InsertSeq < ordered[j].InsertSeq })
	var sb strings.Builder
	for i, e := range ordered {
		if i > 0 {
			sb.WriteByte('\n')
		}
		fmt.Fprintf(&sb, "<kb-fact id=%d kb=%s", e.ID, e.KBCode)
		if e.SourceTopic != "" {
			fmt.Fprintf(&sb, " source=topic:%s", e.SourceTopic)
		}
		sb.WriteString(">\n")
		sb.WriteString(e.Content)
		if !strings.HasSuffix(e.Content, "\n") {
			sb.WriteByte('\n')
		}
		sb.WriteString("</kb-fact>\n")
	}
	return sb.String()
}