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 | a8a66ce | 2025-06-05 10:50:17 +0200 | [diff] [blame^] | 24 | Port *int `kong:"short='p',help='Port to listen on'"` |
Akron | e1cff7c | 2025-06-04 18:43:32 +0200 | [diff] [blame] | 25 | Config string `kong:"short='c',help='YAML configuration file containing mapping directives and global settings'"` |
| 26 | Mappings []string `kong:"short='m',help='Individual YAML mapping files to load'"` |
Akron | a8a66ce | 2025-06-05 10:50:17 +0200 | [diff] [blame^] | 27 | LogLevel *string `kong:"short='l',help='Log level (debug, info, warn, error)'"` |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 28 | } |
| 29 | |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 30 | // TemplateData holds data for the Kalamar plugin template |
| 31 | type TemplateData struct { |
| 32 | Title string |
| 33 | Version string |
Akron | fc77b5e | 2025-06-04 11:44:43 +0200 | [diff] [blame] | 34 | Hash string |
| 35 | Date string |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 36 | Description string |
Akron | 06d21f0 | 2025-06-04 14:36:07 +0200 | [diff] [blame] | 37 | Server string |
| 38 | SDK string |
Akron | c376dcc | 2025-06-04 17:00:18 +0200 | [diff] [blame] | 39 | MapID string |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 40 | MappingIDs []string |
| 41 | } |
| 42 | |
Akron | a00d475 | 2025-05-26 17:34:36 +0200 | [diff] [blame] | 43 | func parseConfig() *appConfig { |
| 44 | cfg := &appConfig{} |
Akron | fc77b5e | 2025-06-04 11:44:43 +0200 | [diff] [blame] | 45 | |
| 46 | desc := config.Description |
| 47 | desc += " [" + config.Version + "]" |
| 48 | |
Akron | 1fc750e | 2025-05-26 16:54:18 +0200 | [diff] [blame] | 49 | ctx := kong.Parse(cfg, |
Akron | fc77b5e | 2025-06-04 11:44:43 +0200 | [diff] [blame] | 50 | kong.Description(desc), |
Akron | 1fc750e | 2025-05-26 16:54:18 +0200 | [diff] [blame] | 51 | kong.UsageOnError(), |
| 52 | ) |
| 53 | if ctx.Error != nil { |
| 54 | fmt.Fprintln(os.Stderr, ctx.Error) |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 55 | os.Exit(1) |
| 56 | } |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 57 | return cfg |
| 58 | } |
| 59 | |
| 60 | func setupLogger(level string) { |
| 61 | // Parse log level |
| 62 | lvl, err := zerolog.ParseLevel(strings.ToLower(level)) |
| 63 | if err != nil { |
| 64 | log.Error().Err(err).Str("level", level).Msg("Invalid log level, defaulting to info") |
| 65 | lvl = zerolog.InfoLevel |
| 66 | } |
| 67 | |
| 68 | // Configure zerolog |
| 69 | zerolog.SetGlobalLevel(lvl) |
| 70 | log.Logger = log.Output(zerolog.ConsoleWriter{Out: os.Stderr}) |
| 71 | } |
| 72 | |
| 73 | func main() { |
| 74 | // Parse command line flags |
Akron | 1fc750e | 2025-05-26 16:54:18 +0200 | [diff] [blame] | 75 | cfg := parseConfig() |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 76 | |
Akron | e1cff7c | 2025-06-04 18:43:32 +0200 | [diff] [blame] | 77 | // Validate command line arguments |
| 78 | if cfg.Config == "" && len(cfg.Mappings) == 0 { |
| 79 | log.Fatal().Msg("At least one configuration source must be provided: use -c for main config file or -m for mapping files") |
| 80 | } |
| 81 | |
Akron | e1cff7c | 2025-06-04 18:43:32 +0200 | [diff] [blame] | 82 | // Load configuration from multiple sources |
| 83 | yamlConfig, err := config.LoadFromSources(cfg.Config, cfg.Mappings) |
Akron | a00d475 | 2025-05-26 17:34:36 +0200 | [diff] [blame] | 84 | if err != nil { |
| 85 | log.Fatal().Err(err).Msg("Failed to load configuration") |
| 86 | } |
| 87 | |
Akron | a8a66ce | 2025-06-05 10:50:17 +0200 | [diff] [blame^] | 88 | finalPort := yamlConfig.Port |
| 89 | finalLogLevel := yamlConfig.LogLevel |
| 90 | |
| 91 | // Use command line values if provided (they override config file) |
| 92 | if cfg.Port != nil { |
| 93 | finalPort = *cfg.Port |
| 94 | } |
| 95 | if cfg.LogLevel != nil { |
| 96 | finalLogLevel = *cfg.LogLevel |
| 97 | } |
| 98 | |
| 99 | // Set up logging with the final log level |
| 100 | setupLogger(finalLogLevel) |
| 101 | |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 102 | // Create a new mapper instance |
Akron | a00d475 | 2025-05-26 17:34:36 +0200 | [diff] [blame] | 103 | m, err := mapper.NewMapper(yamlConfig.Lists) |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 104 | if err != nil { |
| 105 | log.Fatal().Err(err).Msg("Failed to create mapper") |
| 106 | } |
| 107 | |
| 108 | // Create fiber app |
| 109 | app := fiber.New(fiber.Config{ |
| 110 | DisableStartupMessage: true, |
Akron | 74e1c07 | 2025-05-26 14:38:25 +0200 | [diff] [blame] | 111 | BodyLimit: maxInputLength, |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 112 | }) |
| 113 | |
| 114 | // Set up routes |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 115 | setupRoutes(app, m, yamlConfig) |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 116 | |
| 117 | // Start server |
| 118 | go func() { |
Akron | a8a66ce | 2025-06-05 10:50:17 +0200 | [diff] [blame^] | 119 | log.Info().Int("port", finalPort).Msg("Starting server") |
| 120 | if err := app.Listen(fmt.Sprintf(":%d", finalPort)); err != nil { |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 121 | log.Fatal().Err(err).Msg("Server error") |
| 122 | } |
| 123 | }() |
| 124 | |
| 125 | // Wait for interrupt signal |
| 126 | sigChan := make(chan os.Signal, 1) |
| 127 | signal.Notify(sigChan, os.Interrupt, syscall.SIGTERM) |
| 128 | <-sigChan |
| 129 | |
| 130 | // Graceful shutdown |
| 131 | log.Info().Msg("Shutting down server") |
| 132 | if err := app.Shutdown(); err != nil { |
| 133 | log.Error().Err(err).Msg("Error during shutdown") |
| 134 | } |
| 135 | } |
| 136 | |
Akron | 06d21f0 | 2025-06-04 14:36:07 +0200 | [diff] [blame] | 137 | func setupRoutes(app *fiber.App, m *mapper.Mapper, yamlConfig *config.MappingConfig) { |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 138 | // Health check endpoint |
| 139 | app.Get("/health", func(c *fiber.Ctx) error { |
| 140 | return c.SendString("OK") |
| 141 | }) |
| 142 | |
| 143 | // Transformation endpoint |
| 144 | app.Post("/:map/query", handleTransform(m)) |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 145 | |
| 146 | // Kalamar plugin endpoint |
Akron | c471c0a | 2025-06-04 11:56:22 +0200 | [diff] [blame] | 147 | app.Get("/", handleKalamarPlugin(yamlConfig)) |
Akron | c376dcc | 2025-06-04 17:00:18 +0200 | [diff] [blame] | 148 | app.Get("/:map", handleKalamarPlugin(yamlConfig)) |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 149 | } |
| 150 | |
| 151 | func handleTransform(m *mapper.Mapper) fiber.Handler { |
| 152 | return func(c *fiber.Ctx) error { |
| 153 | // Get parameters |
| 154 | mapID := c.Params("map") |
| 155 | dir := c.Query("dir", "atob") |
| 156 | foundryA := c.Query("foundryA", "") |
| 157 | foundryB := c.Query("foundryB", "") |
| 158 | layerA := c.Query("layerA", "") |
| 159 | layerB := c.Query("layerB", "") |
| 160 | |
Akron | 74e1c07 | 2025-05-26 14:38:25 +0200 | [diff] [blame] | 161 | // Validate input parameters |
| 162 | if err := validateInput(mapID, dir, foundryA, foundryB, layerA, layerB, c.Body()); err != nil { |
| 163 | return c.Status(fiber.StatusBadRequest).JSON(fiber.Map{ |
| 164 | "error": err.Error(), |
| 165 | }) |
| 166 | } |
| 167 | |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 168 | // Validate direction |
| 169 | if dir != "atob" && dir != "btoa" { |
| 170 | return c.Status(fiber.StatusBadRequest).JSON(fiber.Map{ |
| 171 | "error": "invalid direction, must be 'atob' or 'btoa'", |
| 172 | }) |
| 173 | } |
| 174 | |
| 175 | // Parse request body |
Akron | 2cbdab5 | 2025-05-23 17:57:10 +0200 | [diff] [blame] | 176 | var jsonData any |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 177 | if err := c.BodyParser(&jsonData); err != nil { |
| 178 | return c.Status(fiber.StatusBadRequest).JSON(fiber.Map{ |
| 179 | "error": "invalid JSON in request body", |
| 180 | }) |
| 181 | } |
| 182 | |
Akron | a1a183f | 2025-05-26 17:47:33 +0200 | [diff] [blame] | 183 | // Parse direction |
| 184 | direction, err := mapper.ParseDirection(dir) |
| 185 | if err != nil { |
| 186 | return c.Status(fiber.StatusBadRequest).JSON(fiber.Map{ |
| 187 | "error": err.Error(), |
| 188 | }) |
| 189 | } |
| 190 | |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 191 | // Apply mappings |
Akron | 7b4984e | 2025-05-26 19:12:20 +0200 | [diff] [blame] | 192 | result, err := m.ApplyQueryMappings(mapID, mapper.MappingOptions{ |
Akron | a1a183f | 2025-05-26 17:47:33 +0200 | [diff] [blame] | 193 | Direction: direction, |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 194 | FoundryA: foundryA, |
| 195 | FoundryB: foundryB, |
| 196 | LayerA: layerA, |
| 197 | LayerB: layerB, |
| 198 | }, jsonData) |
| 199 | |
| 200 | if err != nil { |
| 201 | log.Error().Err(err). |
| 202 | Str("mapID", mapID). |
| 203 | Str("direction", dir). |
| 204 | Msg("Failed to apply mappings") |
| 205 | |
| 206 | return c.Status(fiber.StatusInternalServerError).JSON(fiber.Map{ |
| 207 | "error": err.Error(), |
| 208 | }) |
| 209 | } |
| 210 | |
| 211 | return c.JSON(result) |
| 212 | } |
| 213 | } |
Akron | 74e1c07 | 2025-05-26 14:38:25 +0200 | [diff] [blame] | 214 | |
| 215 | // validateInput checks if the input parameters are valid |
| 216 | func validateInput(mapID, dir, foundryA, foundryB, layerA, layerB string, body []byte) error { |
Akron | 69d43bf | 2025-05-26 17:09:00 +0200 | [diff] [blame] | 217 | // Define parameter checks |
| 218 | params := []struct { |
Akron | 74e1c07 | 2025-05-26 14:38:25 +0200 | [diff] [blame] | 219 | name string |
| 220 | value string |
| 221 | }{ |
| 222 | {"mapID", mapID}, |
| 223 | {"dir", dir}, |
| 224 | {"foundryA", foundryA}, |
| 225 | {"foundryB", foundryB}, |
| 226 | {"layerA", layerA}, |
| 227 | {"layerB", layerB}, |
Akron | 69d43bf | 2025-05-26 17:09:00 +0200 | [diff] [blame] | 228 | } |
| 229 | |
| 230 | for _, param := range params { |
| 231 | // Check input lengths |
| 232 | if len(param.value) > maxParamLength { |
| 233 | return fmt.Errorf("%s too long (max %d bytes)", param.name, maxParamLength) |
| 234 | } |
| 235 | // Check for invalid characters in parameters |
Akron | 74e1c07 | 2025-05-26 14:38:25 +0200 | [diff] [blame] | 236 | if strings.ContainsAny(param.value, "<>{}[]\\") { |
| 237 | return fmt.Errorf("%s contains invalid characters", param.name) |
| 238 | } |
| 239 | } |
| 240 | |
Akron | 69d43bf | 2025-05-26 17:09:00 +0200 | [diff] [blame] | 241 | if len(body) > maxInputLength { |
| 242 | return fmt.Errorf("request body too large (max %d bytes)", maxInputLength) |
| 243 | } |
| 244 | |
Akron | 74e1c07 | 2025-05-26 14:38:25 +0200 | [diff] [blame] | 245 | return nil |
| 246 | } |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 247 | |
Akron | 06d21f0 | 2025-06-04 14:36:07 +0200 | [diff] [blame] | 248 | func handleKalamarPlugin(yamlConfig *config.MappingConfig) fiber.Handler { |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 249 | return func(c *fiber.Ctx) error { |
Akron | c376dcc | 2025-06-04 17:00:18 +0200 | [diff] [blame] | 250 | mapID := c.Params("map") |
| 251 | |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 252 | // Get list of available mapping IDs |
| 253 | var mappingIDs []string |
| 254 | for _, list := range yamlConfig.Lists { |
| 255 | mappingIDs = append(mappingIDs, list.ID) |
| 256 | } |
| 257 | |
Akron | 06d21f0 | 2025-06-04 14:36:07 +0200 | [diff] [blame] | 258 | // Use values from config (defaults are already applied during parsing) |
| 259 | server := yamlConfig.Server |
| 260 | sdk := yamlConfig.SDK |
| 261 | |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 262 | // Prepare template data |
| 263 | data := TemplateData{ |
Akron | fc77b5e | 2025-06-04 11:44:43 +0200 | [diff] [blame] | 264 | Title: config.Title, |
| 265 | Version: config.Version, |
| 266 | Hash: config.Buildhash, |
| 267 | Date: config.Buildtime, |
| 268 | Description: config.Description, |
Akron | 06d21f0 | 2025-06-04 14:36:07 +0200 | [diff] [blame] | 269 | Server: server, |
| 270 | SDK: sdk, |
Akron | c376dcc | 2025-06-04 17:00:18 +0200 | [diff] [blame] | 271 | MapID: mapID, |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 272 | MappingIDs: mappingIDs, |
| 273 | } |
| 274 | |
| 275 | // Generate HTML |
| 276 | html := generateKalamarPluginHTML(data) |
| 277 | |
| 278 | c.Set("Content-Type", "text/html") |
| 279 | return c.SendString(html) |
| 280 | } |
| 281 | } |
| 282 | |
| 283 | // generateKalamarPluginHTML creates the HTML template for the Kalamar plugin page |
| 284 | // This function can be easily modified to change the appearance and content |
| 285 | func generateKalamarPluginHTML(data TemplateData) string { |
| 286 | html := `<!DOCTYPE html> |
| 287 | <html lang="en"> |
| 288 | <head> |
| 289 | <meta charset="UTF-8"> |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 290 | <title>` + data.Title + `</title> |
Akron | 06d21f0 | 2025-06-04 14:36:07 +0200 | [diff] [blame] | 291 | <script src="` + data.SDK + `" |
| 292 | data-server="` + data.Server + `"></script> |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 293 | </head> |
| 294 | <body> |
| 295 | <div class="container"> |
| 296 | <h1>` + data.Title + `</h1> |
Akron | c376dcc | 2025-06-04 17:00:18 +0200 | [diff] [blame] | 297 | <p>` + data.Description + `</p>` |
| 298 | |
| 299 | if data.MapID != "" { |
| 300 | html += `<p>Map ID: ` + data.MapID + `</p>` |
| 301 | } |
| 302 | |
| 303 | html += ` <h2>Plugin Information</h2> |
Akron | c471c0a | 2025-06-04 11:56:22 +0200 | [diff] [blame] | 304 | <p><strong>Version:</strong> <tt>` + data.Version + `</tt></p> |
| 305 | <p><strong>Build Date:</strong> <tt>` + data.Date + `</tt></p> |
| 306 | <p><strong>Build Hash:</strong> <tt>` + data.Hash + `</tt></p> |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 307 | |
Akron | c471c0a | 2025-06-04 11:56:22 +0200 | [diff] [blame] | 308 | <h2>Available API Endpoints</h2> |
| 309 | <dl> |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 310 | |
Akron | c376dcc | 2025-06-04 17:00:18 +0200 | [diff] [blame] | 311 | <dt><tt><strong>GET</strong> /:map</tt></dt> |
| 312 | <dd><small>Kalamar integration</small></dd> |
Akron | e1cff7c | 2025-06-04 18:43:32 +0200 | [diff] [blame] | 313 | |
| 314 | <dt><tt><strong>POST</strong> /:map/query</tt></dt> |
Akron | c376dcc | 2025-06-04 17:00:18 +0200 | [diff] [blame] | 315 | <dd><small>Transform JSON query objects using term mapping rules</small></dd> |
| 316 | |
Akron | c471c0a | 2025-06-04 11:56:22 +0200 | [diff] [blame] | 317 | </dl> |
Akron | c376dcc | 2025-06-04 17:00:18 +0200 | [diff] [blame] | 318 | |
| 319 | <h2>Available Term Mappings</h2> |
Akron | c471c0a | 2025-06-04 11:56:22 +0200 | [diff] [blame] | 320 | <ul>` |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 321 | |
| 322 | for _, id := range data.MappingIDs { |
| 323 | html += ` |
| 324 | <li>` + id + `</li>` |
| 325 | } |
| 326 | |
| 327 | html += ` |
Akron | c376dcc | 2025-06-04 17:00:18 +0200 | [diff] [blame] | 328 | </ul>` |
Akron | 06d21f0 | 2025-06-04 14:36:07 +0200 | [diff] [blame] | 329 | |
Akron | c376dcc | 2025-06-04 17:00:18 +0200 | [diff] [blame] | 330 | if data.MapID != "" { |
| 331 | html += ` <script> |
Akron | 06d21f0 | 2025-06-04 14:36:07 +0200 | [diff] [blame] | 332 | <!-- activates/deactivates Mapper. --> |
| 333 | |
| 334 | let data = { |
| 335 | 'action' : 'pipe', |
Akron | c376dcc | 2025-06-04 17:00:18 +0200 | [diff] [blame] | 336 | 'service' : 'https://korap.ids-mannheim.de/plugin/termmapper/` + data.MapID + `/query' |
Akron | 06d21f0 | 2025-06-04 14:36:07 +0200 | [diff] [blame] | 337 | }; |
| 338 | |
| 339 | function pluginit (p) { |
| 340 | p.onMessage = function(msg) { |
| 341 | if (msg.key == 'termmapper') { |
| 342 | if (msg.value) { |
| 343 | data['job'] = 'add'; |
| 344 | } |
| 345 | else { |
| 346 | data['job'] = 'del'; |
| 347 | }; |
| 348 | KorAPlugin.sendMsg(data); |
| 349 | }; |
| 350 | }; |
| 351 | }; |
Akron | c376dcc | 2025-06-04 17:00:18 +0200 | [diff] [blame] | 352 | </script>` |
| 353 | } |
| 354 | |
| 355 | html += ` </body> |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 356 | </html>` |
| 357 | |
| 358 | return html |
| 359 | } |