|
| 1 | +package cmd |
| 2 | + |
| 3 | +import ( |
| 4 | + "database/sql" |
| 5 | + "encoding/csv" |
| 6 | + "encoding/json" |
| 7 | + "fmt" |
| 8 | + "io" |
| 9 | + "os" |
| 10 | + "runtime" |
| 11 | + "strings" |
| 12 | + |
| 13 | + _ "github.com/marcboeker/go-duckdb" |
| 14 | + "github.com/spf13/cobra" |
| 15 | + "github.com/wesm/msgvault/internal/query" |
| 16 | +) |
| 17 | + |
| 18 | +var queryFormat string |
| 19 | + |
| 20 | +var queryCmd = &cobra.Command{ |
| 21 | + Use: "query [sql]", |
| 22 | + Short: "Run a SQL query against the analytics cache", |
| 23 | + Long: `Run arbitrary SQL against the Parquet analytics cache. |
| 24 | +
|
| 25 | +The following views are available: |
| 26 | + messages, participants, message_recipients, labels, |
| 27 | + message_labels, attachments, conversations, sources |
| 28 | +
|
| 29 | +Convenience views: |
| 30 | + v_messages - messages with resolved sender and labels |
| 31 | + v_senders - per-sender aggregates |
| 32 | + v_domains - per-domain aggregates |
| 33 | + v_labels - label name with message count and size |
| 34 | + v_threads - per-conversation aggregates |
| 35 | +
|
| 36 | +Output formats: |
| 37 | + json - JSON object with columns, rows, row_count (default) |
| 38 | + csv - CSV with header row |
| 39 | + table - Aligned text table |
| 40 | +
|
| 41 | +Examples: |
| 42 | + msgvault query "SELECT from_email, COUNT(*) AS n FROM v_messages GROUP BY 1 ORDER BY 2 DESC LIMIT 10" |
| 43 | + msgvault query --format csv "SELECT * FROM v_senders ORDER BY message_count DESC" |
| 44 | + msgvault query --format table "SELECT name, message_count FROM v_labels"`, |
| 45 | + Args: cobra.ExactArgs(1), |
| 46 | + RunE: func(cmd *cobra.Command, args []string) error { |
| 47 | + dbPath := cfg.DatabaseDSN() |
| 48 | + analyticsDir := cfg.AnalyticsDir() |
| 49 | + |
| 50 | + staleness := cacheNeedsBuild(dbPath, analyticsDir) |
| 51 | + if staleness.NeedsBuild { |
| 52 | + fmt.Fprintf(os.Stderr, |
| 53 | + "Building analytics cache (%s)...\n", |
| 54 | + staleness.Reason) |
| 55 | + result, err := buildCache( |
| 56 | + dbPath, analyticsDir, staleness.FullRebuild, |
| 57 | + ) |
| 58 | + if err != nil { |
| 59 | + return fmt.Errorf("build cache: %w", err) |
| 60 | + } |
| 61 | + if !result.Skipped { |
| 62 | + fmt.Fprintf(os.Stderr, |
| 63 | + "Cached %d messages.\n", |
| 64 | + result.ExportedCount) |
| 65 | + } |
| 66 | + } |
| 67 | + |
| 68 | + if !query.HasCompleteParquetData(analyticsDir) { |
| 69 | + return fmt.Errorf( |
| 70 | + "analytics cache is empty — sync some " + |
| 71 | + "messages first") |
| 72 | + } |
| 73 | + |
| 74 | + return executeQuery( |
| 75 | + analyticsDir, args[0], queryFormat, os.Stdout, |
| 76 | + ) |
| 77 | + }, |
| 78 | +} |
| 79 | + |
| 80 | +// executeQuery opens an in-memory DuckDB, registers views over |
| 81 | +// the Parquet files in analyticsDir, runs the SQL, and writes |
| 82 | +// the results in the requested format. |
| 83 | +func executeQuery( |
| 84 | + analyticsDir, sqlStr, format string, w io.Writer, |
| 85 | +) error { |
| 86 | + db, err := sql.Open("duckdb", "") |
| 87 | + if err != nil { |
| 88 | + return fmt.Errorf("open duckdb: %w", err) |
| 89 | + } |
| 90 | + defer func() { _ = db.Close() }() |
| 91 | + |
| 92 | + db.SetMaxOpenConns(1) |
| 93 | + |
| 94 | + threads := runtime.GOMAXPROCS(0) |
| 95 | + if _, err := db.Exec( |
| 96 | + fmt.Sprintf("SET threads = %d", threads), |
| 97 | + ); err != nil { |
| 98 | + return fmt.Errorf("set threads: %w", err) |
| 99 | + } |
| 100 | + |
| 101 | + if err := query.RegisterViews(db, analyticsDir); err != nil { |
| 102 | + return fmt.Errorf("register views: %w", err) |
| 103 | + } |
| 104 | + |
| 105 | + rows, err := db.Query(sqlStr) |
| 106 | + if err != nil { |
| 107 | + return fmt.Errorf("execute query: %w", err) |
| 108 | + } |
| 109 | + defer func() { _ = rows.Close() }() |
| 110 | + |
| 111 | + cols, err := rows.Columns() |
| 112 | + if err != nil { |
| 113 | + return fmt.Errorf("get columns: %w", err) |
| 114 | + } |
| 115 | + |
| 116 | + var allRows [][]any |
| 117 | + for rows.Next() { |
| 118 | + row, scanErr := scanRow(cols, rows) |
| 119 | + if scanErr != nil { |
| 120 | + return scanErr |
| 121 | + } |
| 122 | + allRows = append(allRows, row) |
| 123 | + } |
| 124 | + if err := rows.Err(); err != nil { |
| 125 | + return fmt.Errorf("iterate rows: %w", err) |
| 126 | + } |
| 127 | + |
| 128 | + switch format { |
| 129 | + case "json": |
| 130 | + return writeJSON(w, cols, allRows) |
| 131 | + case "csv": |
| 132 | + return writeCSV(w, cols, allRows) |
| 133 | + case "table": |
| 134 | + return writeTable(w, cols, allRows) |
| 135 | + default: |
| 136 | + return fmt.Errorf("unknown format %q (use json, csv, or table)", format) |
| 137 | + } |
| 138 | +} |
| 139 | + |
| 140 | +// scanRow scans a single row into a slice of interface{} values, |
| 141 | +// converting []byte to string for clean serialization. |
| 142 | +func scanRow( |
| 143 | + cols []string, rows *sql.Rows, |
| 144 | +) ([]any, error) { |
| 145 | + vals := make([]any, len(cols)) |
| 146 | + ptrs := make([]any, len(cols)) |
| 147 | + for i := range vals { |
| 148 | + ptrs[i] = &vals[i] |
| 149 | + } |
| 150 | + if err := rows.Scan(ptrs...); err != nil { |
| 151 | + return nil, fmt.Errorf("scan row: %w", err) |
| 152 | + } |
| 153 | + for i, v := range vals { |
| 154 | + if b, ok := v.([]byte); ok { |
| 155 | + vals[i] = string(b) |
| 156 | + } |
| 157 | + } |
| 158 | + return vals, nil |
| 159 | +} |
| 160 | + |
| 161 | +func writeJSON( |
| 162 | + w io.Writer, cols []string, rows [][]any, |
| 163 | +) error { |
| 164 | + result := query.QueryResult{ |
| 165 | + Columns: cols, |
| 166 | + Rows: rows, |
| 167 | + RowCount: len(rows), |
| 168 | + } |
| 169 | + enc := json.NewEncoder(w) |
| 170 | + enc.SetIndent("", " ") |
| 171 | + return enc.Encode(result) |
| 172 | +} |
| 173 | + |
| 174 | +// displayVal formats a value for CSV/table output. SQL NULLs |
| 175 | +// become empty strings; other values use fmt.Sprintf. |
| 176 | +func displayVal(v any) string { |
| 177 | + if v == nil { |
| 178 | + return "" |
| 179 | + } |
| 180 | + return fmt.Sprintf("%v", v) |
| 181 | +} |
| 182 | + |
| 183 | +func writeCSV( |
| 184 | + w io.Writer, cols []string, rows [][]any, |
| 185 | +) error { |
| 186 | + cw := csv.NewWriter(w) |
| 187 | + |
| 188 | + if err := cw.Write(cols); err != nil { |
| 189 | + return fmt.Errorf("write csv header: %w", err) |
| 190 | + } |
| 191 | + |
| 192 | + for _, row := range rows { |
| 193 | + record := make([]string, len(row)) |
| 194 | + for i, v := range row { |
| 195 | + record[i] = displayVal(v) |
| 196 | + } |
| 197 | + if err := cw.Write(record); err != nil { |
| 198 | + return fmt.Errorf("write csv row: %w", err) |
| 199 | + } |
| 200 | + } |
| 201 | + |
| 202 | + cw.Flush() |
| 203 | + return cw.Error() |
| 204 | +} |
| 205 | + |
| 206 | +func writeTable( |
| 207 | + w io.Writer, cols []string, rows [][]any, |
| 208 | +) error { |
| 209 | + // Convert all values to strings for width calculation |
| 210 | + strRows := make([][]string, len(rows)) |
| 211 | + for i, row := range rows { |
| 212 | + strRows[i] = make([]string, len(row)) |
| 213 | + for j, v := range row { |
| 214 | + strRows[i][j] = displayVal(v) |
| 215 | + } |
| 216 | + } |
| 217 | + |
| 218 | + // Calculate column widths (min = header length) |
| 219 | + widths := make([]int, len(cols)) |
| 220 | + for i, col := range cols { |
| 221 | + widths[i] = len(col) |
| 222 | + } |
| 223 | + for _, row := range strRows { |
| 224 | + for i, val := range row { |
| 225 | + if len(val) > widths[i] { |
| 226 | + widths[i] = len(val) |
| 227 | + } |
| 228 | + } |
| 229 | + } |
| 230 | + |
| 231 | + // Print header |
| 232 | + for i, col := range cols { |
| 233 | + if i > 0 { |
| 234 | + _, _ = fmt.Fprint(w, " ") |
| 235 | + } |
| 236 | + _, _ = fmt.Fprintf(w, "%-*s", widths[i], col) |
| 237 | + } |
| 238 | + _, _ = fmt.Fprintln(w) |
| 239 | + |
| 240 | + // Print separator |
| 241 | + for i, width := range widths { |
| 242 | + if i > 0 { |
| 243 | + _, _ = fmt.Fprint(w, " ") |
| 244 | + } |
| 245 | + _, _ = fmt.Fprint(w, strings.Repeat("-", width)) |
| 246 | + } |
| 247 | + _, _ = fmt.Fprintln(w) |
| 248 | + |
| 249 | + // Print rows |
| 250 | + for _, row := range strRows { |
| 251 | + for i, val := range row { |
| 252 | + if i > 0 { |
| 253 | + _, _ = fmt.Fprint(w, " ") |
| 254 | + } |
| 255 | + _, _ = fmt.Fprintf(w, "%-*s", widths[i], val) |
| 256 | + } |
| 257 | + _, _ = fmt.Fprintln(w) |
| 258 | + } |
| 259 | + |
| 260 | + // Print row count |
| 261 | + _, _ = fmt.Fprintf(w, "(%d rows)\n", len(rows)) |
| 262 | + return nil |
| 263 | +} |
| 264 | + |
| 265 | +func init() { |
| 266 | + rootCmd.AddCommand(queryCmd) |
| 267 | + queryCmd.Flags().StringVar( |
| 268 | + &queryFormat, "format", "json", |
| 269 | + "Output format: json, csv, or table", |
| 270 | + ) |
| 271 | +} |
0 commit comments