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 | e1cff7c | 2025-06-04 18:43:32 +0200 | [diff] [blame^] | 24 | Port int `kong:"short='p',default='8080',help='Port to listen on'"` |
| 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'"` |
| 27 | 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] | 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 | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 82 | // Set up logging |
Akron | 1fc750e | 2025-05-26 16:54:18 +0200 | [diff] [blame] | 83 | setupLogger(cfg.LogLevel) |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 84 | |
Akron | e1cff7c | 2025-06-04 18:43:32 +0200 | [diff] [blame^] | 85 | // Load configuration from multiple sources |
| 86 | yamlConfig, err := config.LoadFromSources(cfg.Config, cfg.Mappings) |
Akron | a00d475 | 2025-05-26 17:34:36 +0200 | [diff] [blame] | 87 | if err != nil { |
| 88 | log.Fatal().Err(err).Msg("Failed to load configuration") |
| 89 | } |
| 90 | |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 91 | // Create a new mapper instance |
Akron | a00d475 | 2025-05-26 17:34:36 +0200 | [diff] [blame] | 92 | m, err := mapper.NewMapper(yamlConfig.Lists) |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 93 | if err != nil { |
| 94 | log.Fatal().Err(err).Msg("Failed to create mapper") |
| 95 | } |
| 96 | |
| 97 | // Create fiber app |
| 98 | app := fiber.New(fiber.Config{ |
| 99 | DisableStartupMessage: true, |
Akron | 74e1c07 | 2025-05-26 14:38:25 +0200 | [diff] [blame] | 100 | BodyLimit: maxInputLength, |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 101 | }) |
| 102 | |
| 103 | // Set up routes |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 104 | setupRoutes(app, m, yamlConfig) |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 105 | |
| 106 | // Start server |
| 107 | go func() { |
Akron | 1fc750e | 2025-05-26 16:54:18 +0200 | [diff] [blame] | 108 | log.Info().Int("port", cfg.Port).Msg("Starting server") |
| 109 | if err := app.Listen(fmt.Sprintf(":%d", cfg.Port)); err != nil { |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 110 | log.Fatal().Err(err).Msg("Server error") |
| 111 | } |
| 112 | }() |
| 113 | |
| 114 | // Wait for interrupt signal |
| 115 | sigChan := make(chan os.Signal, 1) |
| 116 | signal.Notify(sigChan, os.Interrupt, syscall.SIGTERM) |
| 117 | <-sigChan |
| 118 | |
| 119 | // Graceful shutdown |
| 120 | log.Info().Msg("Shutting down server") |
| 121 | if err := app.Shutdown(); err != nil { |
| 122 | log.Error().Err(err).Msg("Error during shutdown") |
| 123 | } |
| 124 | } |
| 125 | |
Akron | 06d21f0 | 2025-06-04 14:36:07 +0200 | [diff] [blame] | 126 | func setupRoutes(app *fiber.App, m *mapper.Mapper, yamlConfig *config.MappingConfig) { |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 127 | // Health check endpoint |
| 128 | app.Get("/health", func(c *fiber.Ctx) error { |
| 129 | return c.SendString("OK") |
| 130 | }) |
| 131 | |
| 132 | // Transformation endpoint |
| 133 | app.Post("/:map/query", handleTransform(m)) |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 134 | |
| 135 | // Kalamar plugin endpoint |
Akron | c471c0a | 2025-06-04 11:56:22 +0200 | [diff] [blame] | 136 | app.Get("/", handleKalamarPlugin(yamlConfig)) |
Akron | c376dcc | 2025-06-04 17:00:18 +0200 | [diff] [blame] | 137 | app.Get("/:map", handleKalamarPlugin(yamlConfig)) |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 138 | } |
| 139 | |
| 140 | func handleTransform(m *mapper.Mapper) fiber.Handler { |
| 141 | return func(c *fiber.Ctx) error { |
| 142 | // Get parameters |
| 143 | mapID := c.Params("map") |
| 144 | dir := c.Query("dir", "atob") |
| 145 | foundryA := c.Query("foundryA", "") |
| 146 | foundryB := c.Query("foundryB", "") |
| 147 | layerA := c.Query("layerA", "") |
| 148 | layerB := c.Query("layerB", "") |
| 149 | |
Akron | 74e1c07 | 2025-05-26 14:38:25 +0200 | [diff] [blame] | 150 | // Validate input parameters |
| 151 | if err := validateInput(mapID, dir, foundryA, foundryB, layerA, layerB, c.Body()); err != nil { |
| 152 | return c.Status(fiber.StatusBadRequest).JSON(fiber.Map{ |
| 153 | "error": err.Error(), |
| 154 | }) |
| 155 | } |
| 156 | |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 157 | // Validate direction |
| 158 | if dir != "atob" && dir != "btoa" { |
| 159 | return c.Status(fiber.StatusBadRequest).JSON(fiber.Map{ |
| 160 | "error": "invalid direction, must be 'atob' or 'btoa'", |
| 161 | }) |
| 162 | } |
| 163 | |
| 164 | // Parse request body |
Akron | 2cbdab5 | 2025-05-23 17:57:10 +0200 | [diff] [blame] | 165 | var jsonData any |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 166 | if err := c.BodyParser(&jsonData); err != nil { |
| 167 | return c.Status(fiber.StatusBadRequest).JSON(fiber.Map{ |
| 168 | "error": "invalid JSON in request body", |
| 169 | }) |
| 170 | } |
| 171 | |
Akron | a1a183f | 2025-05-26 17:47:33 +0200 | [diff] [blame] | 172 | // Parse direction |
| 173 | direction, err := mapper.ParseDirection(dir) |
| 174 | if err != nil { |
| 175 | return c.Status(fiber.StatusBadRequest).JSON(fiber.Map{ |
| 176 | "error": err.Error(), |
| 177 | }) |
| 178 | } |
| 179 | |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 180 | // Apply mappings |
Akron | 7b4984e | 2025-05-26 19:12:20 +0200 | [diff] [blame] | 181 | result, err := m.ApplyQueryMappings(mapID, mapper.MappingOptions{ |
Akron | a1a183f | 2025-05-26 17:47:33 +0200 | [diff] [blame] | 182 | Direction: direction, |
Akron | 49ceeb4 | 2025-05-23 17:46:01 +0200 | [diff] [blame] | 183 | FoundryA: foundryA, |
| 184 | FoundryB: foundryB, |
| 185 | LayerA: layerA, |
| 186 | LayerB: layerB, |
| 187 | }, jsonData) |
| 188 | |
| 189 | if err != nil { |
| 190 | log.Error().Err(err). |
| 191 | Str("mapID", mapID). |
| 192 | Str("direction", dir). |
| 193 | Msg("Failed to apply mappings") |
| 194 | |
| 195 | return c.Status(fiber.StatusInternalServerError).JSON(fiber.Map{ |
| 196 | "error": err.Error(), |
| 197 | }) |
| 198 | } |
| 199 | |
| 200 | return c.JSON(result) |
| 201 | } |
| 202 | } |
Akron | 74e1c07 | 2025-05-26 14:38:25 +0200 | [diff] [blame] | 203 | |
| 204 | // validateInput checks if the input parameters are valid |
| 205 | func validateInput(mapID, dir, foundryA, foundryB, layerA, layerB string, body []byte) error { |
Akron | 69d43bf | 2025-05-26 17:09:00 +0200 | [diff] [blame] | 206 | // Define parameter checks |
| 207 | params := []struct { |
Akron | 74e1c07 | 2025-05-26 14:38:25 +0200 | [diff] [blame] | 208 | name string |
| 209 | value string |
| 210 | }{ |
| 211 | {"mapID", mapID}, |
| 212 | {"dir", dir}, |
| 213 | {"foundryA", foundryA}, |
| 214 | {"foundryB", foundryB}, |
| 215 | {"layerA", layerA}, |
| 216 | {"layerB", layerB}, |
Akron | 69d43bf | 2025-05-26 17:09:00 +0200 | [diff] [blame] | 217 | } |
| 218 | |
| 219 | for _, param := range params { |
| 220 | // Check input lengths |
| 221 | if len(param.value) > maxParamLength { |
| 222 | return fmt.Errorf("%s too long (max %d bytes)", param.name, maxParamLength) |
| 223 | } |
| 224 | // Check for invalid characters in parameters |
Akron | 74e1c07 | 2025-05-26 14:38:25 +0200 | [diff] [blame] | 225 | if strings.ContainsAny(param.value, "<>{}[]\\") { |
| 226 | return fmt.Errorf("%s contains invalid characters", param.name) |
| 227 | } |
| 228 | } |
| 229 | |
Akron | 69d43bf | 2025-05-26 17:09:00 +0200 | [diff] [blame] | 230 | if len(body) > maxInputLength { |
| 231 | return fmt.Errorf("request body too large (max %d bytes)", maxInputLength) |
| 232 | } |
| 233 | |
Akron | 74e1c07 | 2025-05-26 14:38:25 +0200 | [diff] [blame] | 234 | return nil |
| 235 | } |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 236 | |
Akron | 06d21f0 | 2025-06-04 14:36:07 +0200 | [diff] [blame] | 237 | func handleKalamarPlugin(yamlConfig *config.MappingConfig) fiber.Handler { |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 238 | return func(c *fiber.Ctx) error { |
Akron | c376dcc | 2025-06-04 17:00:18 +0200 | [diff] [blame] | 239 | mapID := c.Params("map") |
| 240 | |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 241 | // Get list of available mapping IDs |
| 242 | var mappingIDs []string |
| 243 | for _, list := range yamlConfig.Lists { |
| 244 | mappingIDs = append(mappingIDs, list.ID) |
| 245 | } |
| 246 | |
Akron | 06d21f0 | 2025-06-04 14:36:07 +0200 | [diff] [blame] | 247 | // Use values from config (defaults are already applied during parsing) |
| 248 | server := yamlConfig.Server |
| 249 | sdk := yamlConfig.SDK |
| 250 | |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 251 | // Prepare template data |
| 252 | data := TemplateData{ |
Akron | fc77b5e | 2025-06-04 11:44:43 +0200 | [diff] [blame] | 253 | Title: config.Title, |
| 254 | Version: config.Version, |
| 255 | Hash: config.Buildhash, |
| 256 | Date: config.Buildtime, |
| 257 | Description: config.Description, |
Akron | 06d21f0 | 2025-06-04 14:36:07 +0200 | [diff] [blame] | 258 | Server: server, |
| 259 | SDK: sdk, |
Akron | c376dcc | 2025-06-04 17:00:18 +0200 | [diff] [blame] | 260 | MapID: mapID, |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 261 | MappingIDs: mappingIDs, |
| 262 | } |
| 263 | |
| 264 | // Generate HTML |
| 265 | html := generateKalamarPluginHTML(data) |
| 266 | |
| 267 | c.Set("Content-Type", "text/html") |
| 268 | return c.SendString(html) |
| 269 | } |
| 270 | } |
| 271 | |
| 272 | // generateKalamarPluginHTML creates the HTML template for the Kalamar plugin page |
| 273 | // This function can be easily modified to change the appearance and content |
| 274 | func generateKalamarPluginHTML(data TemplateData) string { |
| 275 | html := `<!DOCTYPE html> |
| 276 | <html lang="en"> |
| 277 | <head> |
| 278 | <meta charset="UTF-8"> |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 279 | <title>` + data.Title + `</title> |
Akron | 06d21f0 | 2025-06-04 14:36:07 +0200 | [diff] [blame] | 280 | <script src="` + data.SDK + `" |
| 281 | data-server="` + data.Server + `"></script> |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 282 | </head> |
| 283 | <body> |
| 284 | <div class="container"> |
| 285 | <h1>` + data.Title + `</h1> |
Akron | c376dcc | 2025-06-04 17:00:18 +0200 | [diff] [blame] | 286 | <p>` + data.Description + `</p>` |
| 287 | |
| 288 | if data.MapID != "" { |
| 289 | html += `<p>Map ID: ` + data.MapID + `</p>` |
| 290 | } |
| 291 | |
| 292 | html += ` <h2>Plugin Information</h2> |
Akron | c471c0a | 2025-06-04 11:56:22 +0200 | [diff] [blame] | 293 | <p><strong>Version:</strong> <tt>` + data.Version + `</tt></p> |
| 294 | <p><strong>Build Date:</strong> <tt>` + data.Date + `</tt></p> |
| 295 | <p><strong>Build Hash:</strong> <tt>` + data.Hash + `</tt></p> |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 296 | |
Akron | c471c0a | 2025-06-04 11:56:22 +0200 | [diff] [blame] | 297 | <h2>Available API Endpoints</h2> |
| 298 | <dl> |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 299 | |
Akron | c376dcc | 2025-06-04 17:00:18 +0200 | [diff] [blame] | 300 | <dt><tt><strong>GET</strong> /:map</tt></dt> |
| 301 | <dd><small>Kalamar integration</small></dd> |
Akron | e1cff7c | 2025-06-04 18:43:32 +0200 | [diff] [blame^] | 302 | |
| 303 | <dt><tt><strong>POST</strong> /:map/query</tt></dt> |
Akron | c376dcc | 2025-06-04 17:00:18 +0200 | [diff] [blame] | 304 | <dd><small>Transform JSON query objects using term mapping rules</small></dd> |
| 305 | |
Akron | c471c0a | 2025-06-04 11:56:22 +0200 | [diff] [blame] | 306 | </dl> |
Akron | c376dcc | 2025-06-04 17:00:18 +0200 | [diff] [blame] | 307 | |
| 308 | <h2>Available Term Mappings</h2> |
Akron | c471c0a | 2025-06-04 11:56:22 +0200 | [diff] [blame] | 309 | <ul>` |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 310 | |
| 311 | for _, id := range data.MappingIDs { |
| 312 | html += ` |
| 313 | <li>` + id + `</li>` |
| 314 | } |
| 315 | |
| 316 | html += ` |
Akron | c376dcc | 2025-06-04 17:00:18 +0200 | [diff] [blame] | 317 | </ul>` |
Akron | 06d21f0 | 2025-06-04 14:36:07 +0200 | [diff] [blame] | 318 | |
Akron | c376dcc | 2025-06-04 17:00:18 +0200 | [diff] [blame] | 319 | if data.MapID != "" { |
| 320 | html += ` <script> |
Akron | 06d21f0 | 2025-06-04 14:36:07 +0200 | [diff] [blame] | 321 | <!-- activates/deactivates Mapper. --> |
| 322 | |
| 323 | let data = { |
| 324 | 'action' : 'pipe', |
Akron | c376dcc | 2025-06-04 17:00:18 +0200 | [diff] [blame] | 325 | 'service' : 'https://korap.ids-mannheim.de/plugin/termmapper/` + data.MapID + `/query' |
Akron | 06d21f0 | 2025-06-04 14:36:07 +0200 | [diff] [blame] | 326 | }; |
| 327 | |
| 328 | function pluginit (p) { |
| 329 | p.onMessage = function(msg) { |
| 330 | if (msg.key == 'termmapper') { |
| 331 | if (msg.value) { |
| 332 | data['job'] = 'add'; |
| 333 | } |
| 334 | else { |
| 335 | data['job'] = 'del'; |
| 336 | }; |
| 337 | KorAPlugin.sendMsg(data); |
| 338 | }; |
| 339 | }; |
| 340 | }; |
Akron | c376dcc | 2025-06-04 17:00:18 +0200 | [diff] [blame] | 341 | </script>` |
| 342 | } |
| 343 | |
| 344 | html += ` </body> |
Akron | 40aaa63 | 2025-06-03 17:57:52 +0200 | [diff] [blame] | 345 | </html>` |
| 346 | |
| 347 | return html |
| 348 | } |