tefter/server/importer/simplenote.go

// Package importer reads Simplenote export archives into the store.
package importer

import (
	"archive/zip"
	"crypto/sha256"
	"encoding/json"
	"fmt"
	"io"
	"strings"
	"time"

	"tefter/server/store"
)

// Verified against a real export (2026): the zip contains source/notes.json with
// {activeNotes: [...], trashedNotes: [...]}; each note has id, content,
// creationDate, lastModified (ISO 8601) and optionally tags. There is no
// `markdown` field in current exports.
type snNote struct {
	ID           string   `json:"id"`
	Content      string   `json:"content"`
	CreationDate string   `json:"creationDate"`
	LastModified string   `json:"lastModified"`
	Tags         []string `json:"tags"`
}

type snExport struct {
	ActiveNotes  []snNote `json:"activeNotes"`
	TrashedNotes []snNote `json:"trashedNotes"`
}

type Summary struct {
	Imported int `json:"imported"`
	Skipped  int `json:"skipped"`
	Trashed  int `json:"trashed"`
}

func (s Summary) String() string {
	return fmt.Sprintf("imported %d, skipped %d duplicates, %d trashed", s.Imported, s.Skipped, s.Trashed)
}

func parseTime(iso string, fallback time.Time) int64 {
	if t, err := time.Parse(time.RFC3339, iso); err == nil {
		return t.UnixMilli()
	}
	return fallback.UnixMilli()
}

// ImportZip imports a Simplenote export zip (as a reader) into st. Idempotent:
// notes are deduped by Simplenote id kept in the source_id column.
func ImportZip(st *store.Store, r io.ReaderAt, size int64) (Summary, error) {
	var sum Summary
	zr, err := zip.NewReader(r, size)
	if err != nil {
		return sum, fmt.Errorf("open zip: %w", err)
	}

	var data []byte
	for _, f := range zr.File {
		if f.Name == "source/notes.json" || strings.HasSuffix(f.Name, "/notes.json") || f.Name == "notes.json" {
			rc, err := f.Open()
			if err != nil {
				return sum, err
			}
			data, err = io.ReadAll(rc)
			rc.Close()
			if err != nil {
				return sum, err
			}
			break
		}
	}
	if data == nil {
		return sum, fmt.Errorf("notes.json not found in archive (expected source/notes.json)")
	}

	var exp snExport
	if err := json.Unmarshal(data, &exp); err != nil {
		return sum, fmt.Errorf("parse notes.json: %w", err)
	}

	now := time.Now()
	doImport := func(notes []snNote, deleted bool) error {
		for _, sn := range notes {
			if sn.ID == "" {
				// No id to dedupe on — synthesize one from content+creation date.
				sn.ID = fmt.Sprintf("synth-%x", contentHash(sn))
			}
			n := store.Note{
				Content:    strings.ReplaceAll(sn.Content, "\r\n", "\n"),
				Tags:       sn.Tags,
				CreatedAt:  parseTime(sn.CreationDate, now),
				ModifiedAt: parseTime(sn.LastModified, now),
				Deleted:    deleted,
			}
			inserted, err := st.UpsertImported("simplenote:"+sn.ID, n)
			if err != nil {
				return err
			}
			if inserted {
				if deleted {
					sum.Trashed++
				} else {
					sum.Imported++
				}
			} else {
				sum.Skipped++
			}
		}
		return nil
	}

	if err := doImport(exp.ActiveNotes, false); err != nil {
		return sum, err
	}
	if err := doImport(exp.TrashedNotes, true); err != nil {
		return sum, err
	}
	return sum, nil
}

func contentHash(sn snNote) []byte {
	h := sha256.Sum256([]byte(sn.CreationDate + "\x00" + sn.Content))
	return h[:16]
}