Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 1 | package main |
| 2 | |
| 3 | import ( |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 4 | "fmt" |
| 5 | "os" |
| 6 | "os/signal" |
| 7 | "strings" |
| 8 | "syscall" |
| 9 | |
Akron | a00d475 | 2025-05-26 17:34:36 +0200 | [diff] [blame] | 10 | "github.com/KorAP/KoralPipe-TermMapper/config" |
Akron | fa55bb2 | 2025-05-26 15:10:42 +0200 | [diff] [blame] | 11 | "github.com/KorAP/KoralPipe-TermMapper/mapper" |
Akron | 1fc750e | 2025-05-26 16:54:18 +0200 | [diff] [blame] | 12 | "github.com/alecthomas/kong" |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 13 | "github.com/gofiber/fiber/v2" |
| 14 | "github.com/rs/zerolog" |
| 15 | "github.com/rs/zerolog/log" |
| 16 | ) |
| 17 | |
Akron | 74e1c07 | 2025-05-26 14:38:25 +0200 | [diff] [blame] | 18 | const ( |
| 19 | maxInputLength = 1024 * 1024 // 1MB |
| 20 | maxParamLength = 1024 // 1KB |
| 21 | ) |
| 22 | |
Akron | a00d475 | 2025-05-26 17:34:36 +0200 | [diff] [blame] | 23 | type appConfig struct { |
Akron | 1fc750e | 2025-05-26 16:54:18 +0200 | [diff] [blame] | 24 | Port int `kong:"short='p',default='8080',help='Port to listen on'"` |
| 25 | Config string `kong:"short='c',required,help='YAML configuration file containing mapping directives'"` |
| 26 | LogLevel string `kong:"short='l',default='info',help='Log level (debug, info, warn, error)'"` |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 27 | } |
| 28 | |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame^] | 29 | // TemplateData holds data for the Kalamar plugin template |
| 30 | type TemplateData struct { |
| 31 | Title string |
| 32 | Version string |
| 33 | Description string |
| 34 | MappingIDs []string |
| 35 | } |
| 36 | |
Akron | a00d475 | 2025-05-26 17:34:36 +0200 | [diff] [blame] | 37 | func parseConfig() *appConfig { |
| 38 | cfg := &appConfig{} |
Akron | 1fc750e | 2025-05-26 16:54:18 +0200 | [diff] [blame] | 39 | ctx := kong.Parse(cfg, |
| 40 | kong.Description("A web service for transforming JSON objects using term mapping rules."), |
| 41 | kong.UsageOnError(), |
| 42 | ) |
| 43 | if ctx.Error != nil { |
| 44 | fmt.Fprintln(os.Stderr, ctx.Error) |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 45 | os.Exit(1) |
| 46 | } |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 47 | return cfg |
| 48 | } |
| 49 | |
| 50 | func setupLogger(level string) { |
| 51 | // Parse log level |
| 52 | lvl, err := zerolog.ParseLevel(strings.ToLower(level)) |
| 53 | if err != nil { |
| 54 | log.Error().Err(err).Str("level", level).Msg("Invalid log level, defaulting to info") |
| 55 | lvl = zerolog.InfoLevel |
| 56 | } |
| 57 | |
| 58 | // Configure zerolog |
| 59 | zerolog.SetGlobalLevel(lvl) |
| 60 | log.Logger = log.Output(zerolog.ConsoleWriter{Out: os.Stderr}) |
| 61 | } |
| 62 | |
| 63 | func main() { |
| 64 | // Parse command line flags |
Akron | 1fc750e | 2025-05-26 16:54:18 +0200 | [diff] [blame] | 65 | cfg := parseConfig() |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 66 | |
| 67 | // Set up logging |
Akron | 1fc750e | 2025-05-26 16:54:18 +0200 | [diff] [blame] | 68 | setupLogger(cfg.LogLevel) |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 69 | |
Akron | a00d475 | 2025-05-26 17:34:36 +0200 | [diff] [blame] | 70 | // Load configuration file |
| 71 | yamlConfig, err := config.LoadConfig(cfg.Config) |
| 72 | if err != nil { |
| 73 | log.Fatal().Err(err).Msg("Failed to load configuration") |
| 74 | } |
| 75 | |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 76 | // Create a new mapper instance |
Akron | a00d475 | 2025-05-26 17:34:36 +0200 | [diff] [blame] | 77 | m, err := mapper.NewMapper(yamlConfig.Lists) |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 78 | if err != nil { |
| 79 | log.Fatal().Err(err).Msg("Failed to create mapper") |
| 80 | } |
| 81 | |
| 82 | // Create fiber app |
| 83 | app := fiber.New(fiber.Config{ |
| 84 | DisableStartupMessage: true, |
Akron | 74e1c07 | 2025-05-26 14:38:25 +0200 | [diff] [blame] | 85 | BodyLimit: maxInputLength, |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 86 | }) |
| 87 | |
| 88 | // Set up routes |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame^] | 89 | setupRoutes(app, m, yamlConfig) |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 90 | |
| 91 | // Start server |
| 92 | go func() { |
Akron | 1fc750e | 2025-05-26 16:54:18 +0200 | [diff] [blame] | 93 | log.Info().Int("port", cfg.Port).Msg("Starting server") |
| 94 | if err := app.Listen(fmt.Sprintf(":%d", cfg.Port)); err != nil { |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 95 | log.Fatal().Err(err).Msg("Server error") |
| 96 | } |
| 97 | }() |
| 98 | |
| 99 | // Wait for interrupt signal |
| 100 | sigChan := make(chan os.Signal, 1) |
| 101 | signal.Notify(sigChan, os.Interrupt, syscall.SIGTERM) |
| 102 | <-sigChan |
| 103 | |
| 104 | // Graceful shutdown |
| 105 | log.Info().Msg("Shutting down server") |
| 106 | if err := app.Shutdown(); err != nil { |
| 107 | log.Error().Err(err).Msg("Error during shutdown") |
| 108 | } |
| 109 | } |
| 110 | |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame^] | 111 | func setupRoutes(app *fiber.App, m *mapper.Mapper, yamlConfig *config.MappingLists) { |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 112 | // Health check endpoint |
| 113 | app.Get("/health", func(c *fiber.Ctx) error { |
| 114 | return c.SendString("OK") |
| 115 | }) |
| 116 | |
| 117 | // Transformation endpoint |
| 118 | app.Post("/:map/query", handleTransform(m)) |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame^] | 119 | |
| 120 | // Kalamar plugin endpoint |
| 121 | app.Get("/kalamarplugin", handleKalamarPlugin(yamlConfig)) |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 122 | } |
| 123 | |
| 124 | func handleTransform(m *mapper.Mapper) fiber.Handler { |
| 125 | return func(c *fiber.Ctx) error { |
| 126 | // Get parameters |
| 127 | mapID := c.Params("map") |
| 128 | dir := c.Query("dir", "atob") |
| 129 | foundryA := c.Query("foundryA", "") |
| 130 | foundryB := c.Query("foundryB", "") |
| 131 | layerA := c.Query("layerA", "") |
| 132 | layerB := c.Query("layerB", "") |
| 133 | |
Akron | 74e1c07 | 2025-05-26 14:38:25 +0200 | [diff] [blame] | 134 | // Validate input parameters |
| 135 | if err := validateInput(mapID, dir, foundryA, foundryB, layerA, layerB, c.Body()); err != nil { |
| 136 | return c.Status(fiber.StatusBadRequest).JSON(fiber.Map{ |
| 137 | "error": err.Error(), |
| 138 | }) |
| 139 | } |
| 140 | |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 141 | // Validate direction |
| 142 | if dir != "atob" && dir != "btoa" { |
| 143 | return c.Status(fiber.StatusBadRequest).JSON(fiber.Map{ |
| 144 | "error": "invalid direction, must be 'atob' or 'btoa'", |
| 145 | }) |
| 146 | } |
| 147 | |
| 148 | // Parse request body |
Akron | 2cbdab5 | 2025-05-23 17:57:10 +0200 | [diff] [blame] | 149 | var jsonData any |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 150 | if err := c.BodyParser(&jsonData); err != nil { |
| 151 | return c.Status(fiber.StatusBadRequest).JSON(fiber.Map{ |
| 152 | "error": "invalid JSON in request body", |
| 153 | }) |
| 154 | } |
| 155 | |
Akron | a1a183f | 2025-05-26 17:47:33 +0200 | [diff] [blame] | 156 | // Parse direction |
| 157 | direction, err := mapper.ParseDirection(dir) |
| 158 | if err != nil { |
| 159 | return c.Status(fiber.StatusBadRequest).JSON(fiber.Map{ |
| 160 | "error": err.Error(), |
| 161 | }) |
| 162 | } |
| 163 | |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 164 | // Apply mappings |
Akron | 7b4984e | 2025-05-26 19:12:20 +0200 | [diff] [blame] | 165 | result, err := m.ApplyQueryMappings(mapID, mapper.MappingOptions{ |
Akron | a1a183f | 2025-05-26 17:47:33 +0200 | [diff] [blame] | 166 | Direction: direction, |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 167 | FoundryA: foundryA, |
| 168 | FoundryB: foundryB, |
| 169 | LayerA: layerA, |
| 170 | LayerB: layerB, |
| 171 | }, jsonData) |
| 172 | |
| 173 | if err != nil { |
| 174 | log.Error().Err(err). |
| 175 | Str("mapID", mapID). |
| 176 | Str("direction", dir). |
| 177 | Msg("Failed to apply mappings") |
| 178 | |
| 179 | return c.Status(fiber.StatusInternalServerError).JSON(fiber.Map{ |
| 180 | "error": err.Error(), |
| 181 | }) |
| 182 | } |
| 183 | |
| 184 | return c.JSON(result) |
| 185 | } |
| 186 | } |
Akron | 74e1c07 | 2025-05-26 14:38:25 +0200 | [diff] [blame] | 187 | |
| 188 | // validateInput checks if the input parameters are valid |
| 189 | func validateInput(mapID, dir, foundryA, foundryB, layerA, layerB string, body []byte) error { |
Akron | 69d43bf | 2025-05-26 17:09:00 +0200 | [diff] [blame] | 190 | // Define parameter checks |
| 191 | params := []struct { |
Akron | 74e1c07 | 2025-05-26 14:38:25 +0200 | [diff] [blame] | 192 | name string |
| 193 | value string |
| 194 | }{ |
| 195 | {"mapID", mapID}, |
| 196 | {"dir", dir}, |
| 197 | {"foundryA", foundryA}, |
| 198 | {"foundryB", foundryB}, |
| 199 | {"layerA", layerA}, |
| 200 | {"layerB", layerB}, |
Akron | 69d43bf | 2025-05-26 17:09:00 +0200 | [diff] [blame] | 201 | } |
| 202 | |
| 203 | for _, param := range params { |
| 204 | // Check input lengths |
| 205 | if len(param.value) > maxParamLength { |
| 206 | return fmt.Errorf("%s too long (max %d bytes)", param.name, maxParamLength) |
| 207 | } |
| 208 | // Check for invalid characters in parameters |
Akron | 74e1c07 | 2025-05-26 14:38:25 +0200 | [diff] [blame] | 209 | if strings.ContainsAny(param.value, "<>{}[]\\") { |
| 210 | return fmt.Errorf("%s contains invalid characters", param.name) |
| 211 | } |
| 212 | } |
| 213 | |
Akron | 69d43bf | 2025-05-26 17:09:00 +0200 | [diff] [blame] | 214 | if len(body) > maxInputLength { |
| 215 | return fmt.Errorf("request body too large (max %d bytes)", maxInputLength) |
| 216 | } |
| 217 | |
Akron | 74e1c07 | 2025-05-26 14:38:25 +0200 | [diff] [blame] | 218 | return nil |
| 219 | } |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame^] | 220 | |
| 221 | func handleKalamarPlugin(yamlConfig *config.MappingLists) fiber.Handler { |
| 222 | return func(c *fiber.Ctx) error { |
| 223 | // Get list of available mapping IDs |
| 224 | var mappingIDs []string |
| 225 | for _, list := range yamlConfig.Lists { |
| 226 | mappingIDs = append(mappingIDs, list.ID) |
| 227 | } |
| 228 | |
| 229 | // Prepare template data |
| 230 | data := TemplateData{ |
| 231 | Title: "KoralPipe TermMapper - Kalamar Plugin", |
| 232 | Version: "1.0.0", |
| 233 | Description: "A KortalPipe web service for transforming JSON objects using term mapping rules.", |
| 234 | MappingIDs: mappingIDs, |
| 235 | } |
| 236 | |
| 237 | // Generate HTML |
| 238 | html := generateKalamarPluginHTML(data) |
| 239 | |
| 240 | c.Set("Content-Type", "text/html") |
| 241 | return c.SendString(html) |
| 242 | } |
| 243 | } |
| 244 | |
| 245 | // generateKalamarPluginHTML creates the HTML template for the Kalamar plugin page |
| 246 | // This function can be easily modified to change the appearance and content |
| 247 | func generateKalamarPluginHTML(data TemplateData) string { |
| 248 | html := `<!DOCTYPE html> |
| 249 | <html lang="en"> |
| 250 | <head> |
| 251 | <meta charset="UTF-8"> |
| 252 | <meta name="viewport" content="width=device-width, initial-scale=1.0"> |
| 253 | <title>` + data.Title + `</title> |
| 254 | </head> |
| 255 | <body> |
| 256 | <div class="container"> |
| 257 | <h1>` + data.Title + `</h1> |
| 258 | |
| 259 | <div> |
| 260 | <h2>Plugin Information</h2> |
| 261 | <p><strong>Version:</strong> ` + data.Version + `</p> |
| 262 | <p><strong>Description:</strong> ` + data.Description + `</p> |
| 263 | </div> |
| 264 | |
| 265 | <div class="plugin-info"> |
| 266 | <h2>Available API Endpoints</h2> |
| 267 | <div class="api-endpoint"> |
| 268 | <strong>POST</strong> /:map/query?dir=atob&foundryA=&foundryB=&layerA=&layerB= |
| 269 | <br><small>Transform JSON objects using term mapping rules</small> |
| 270 | </div> |
| 271 | <div class="api-endpoint"> |
| 272 | <strong>GET</strong> /health |
| 273 | <br><small>Health check endpoint</small> |
| 274 | </div> |
| 275 | <div class="api-endpoint"> |
| 276 | <strong>GET</strong> /kalamarplugin |
| 277 | <br><small>This entry point for Kalamar integration</small> |
| 278 | </div> |
| 279 | </div> |
| 280 | |
| 281 | <div class="plugin-info"> |
| 282 | <h2>Available Term Mappings</h2> |
| 283 | <ul>` |
| 284 | |
| 285 | for _, id := range data.MappingIDs { |
| 286 | html += ` |
| 287 | <li>` + id + `</li>` |
| 288 | } |
| 289 | |
| 290 | html += ` |
| 291 | </ul> |
| 292 | </div> |
| 293 | </body> |
| 294 | </html>` |
| 295 | |
| 296 | return html |
| 297 | } |