| hebasta | e2bd6e4 | 2020-03-03 19:07:20 +0100 | [diff] [blame] | 1 | package de.ids_mannheim.korap.plkexport; |
| 2 | |
| hebasta | e049676 | 2020-03-30 00:11:01 +0200 | [diff] [blame] | 3 | import java.io.IOException; |
| Akron | bb076c7 | 2020-09-28 14:39:15 +0200 | [diff] [blame] | 4 | import java.io.StringWriter; |
| Akron | b87af2f | 2020-09-24 14:38:02 +0200 | [diff] [blame] | 5 | import java.io.InputStream; |
| Akron | 69cd35d | 2020-11-20 13:17:31 +0100 | [diff] [blame] | 6 | import java.lang.Thread; |
| Akron | bb076c7 | 2020-09-28 14:39:15 +0200 | [diff] [blame] | 7 | import java.util.HashMap; |
| Akron | c931cd0 | 2020-09-15 10:54:17 +0200 | [diff] [blame] | 8 | import java.util.Properties; |
| Akron | 69cd35d | 2020-11-20 13:17:31 +0100 | [diff] [blame] | 9 | import java.util.Base64; |
| Akron | a578eb0 | 2020-11-26 12:34:17 +0100 | [diff] [blame] | 10 | import java.util.ResourceBundle; |
| 11 | import java.util.Locale; |
| Akron | 669d0bb | 2020-10-02 13:12:01 +0200 | [diff] [blame] | 12 | import java.util.regex.Matcher; |
| 13 | import java.util.regex.Pattern; |
| hebasta | ff955d6 | 2020-05-27 15:59:14 +0200 | [diff] [blame] | 14 | |
| Marc Kupietz | a2a1797 | 2022-03-27 12:03:48 +0200 | [diff] [blame^] | 15 | import jakarta.ws.rs.BadRequestException; |
| 16 | import jakarta.ws.rs.WebApplicationException; |
| 17 | import jakarta.ws.rs.FormParam; |
| 18 | import jakarta.ws.rs.QueryParam; |
| 19 | import jakarta.ws.rs.PathParam; |
| 20 | import jakarta.ws.rs.POST; |
| 21 | import jakarta.ws.rs.GET; |
| 22 | import jakarta.ws.rs.Path; |
| 23 | import jakarta.ws.rs.Produces; |
| 24 | import jakarta.ws.rs.Consumes; |
| 25 | import jakarta.ws.rs.client.Client; |
| 26 | import jakarta.ws.rs.client.ClientBuilder; |
| 27 | import jakarta.ws.rs.client.WebTarget; |
| 28 | import jakarta.ws.rs.client.Invocation; |
| 29 | import jakarta.ws.rs.core.Context; |
| 30 | import jakarta.ws.rs.core.UriBuilder; |
| 31 | import jakarta.ws.rs.core.MediaType; |
| 32 | import jakarta.ws.rs.core.Response; |
| 33 | import jakarta.ws.rs.core.Response.ResponseBuilder; |
| 34 | import jakarta.ws.rs.core.Response.Status; |
| Akron | af145eb | 2020-11-24 16:55:47 +0100 | [diff] [blame] | 35 | |
| Akron | 3588101 | 2020-11-24 20:05:06 +0100 | [diff] [blame] | 36 | import static org.apache.commons.io.FilenameUtils.getExtension; |
| 37 | |
| Akron | a241a47 | 2021-04-27 18:55:58 +0200 | [diff] [blame] | 38 | import org.tinylog.Logger; |
| 39 | |
| Akron | af145eb | 2020-11-24 16:55:47 +0100 | [diff] [blame] | 40 | import org.glassfish.jersey.media.sse.EventOutput; |
| 41 | import org.glassfish.jersey.media.sse.OutboundEvent; |
| 42 | import org.glassfish.jersey.media.sse.SseFeature; |
| 43 | |
| Marc Kupietz | a2a1797 | 2022-03-27 12:03:48 +0200 | [diff] [blame^] | 44 | import jakarta.servlet.http.Cookie; |
| 45 | import jakarta.servlet.http.HttpServletRequest; |
| Akron | a578eb0 | 2020-11-26 12:34:17 +0100 | [diff] [blame] | 46 | import org.glassfish.jersey.server.ContainerRequest; |
| hebasta | ff955d6 | 2020-05-27 15:59:14 +0200 | [diff] [blame] | 47 | |
| Akron | e34bd4f | 2020-09-18 10:46:01 +0200 | [diff] [blame] | 48 | import static de.ids_mannheim.korap.plkexport.Util.*; |
| 49 | |
| Akron | bb076c7 | 2020-09-28 14:39:15 +0200 | [diff] [blame] | 50 | // Template engine |
| 51 | import freemarker.template.Configuration; |
| 52 | import freemarker.template.Template; |
| 53 | |
| hebasta | fed4194 | 2020-11-02 10:00:36 +0100 | [diff] [blame] | 54 | /** |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 55 | * TODO: |
| Akron | 18c627d | 2020-11-26 22:09:08 +0100 | [diff] [blame] | 56 | * - Rename "is cutted" to "truncated". |
| Akron | 83f3994 | 2021-03-22 16:37:25 +0100 | [diff] [blame] | 57 | * - Restructure: |
| 58 | * - Load the exporter objects as part of |
| 59 | * the MatchAggregator instead of full |
| 60 | * inheritance. |
| Akron | 55def63 | 2020-11-26 16:00:02 +0100 | [diff] [blame] | 61 | * - Switch Cookie mechanism to container req for |
| 62 | * better testing capabilities. |
| 63 | * - Test ExWsConf. |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 64 | * - Do not expect all meta data per match. |
| Akron | 69cd35d | 2020-11-20 13:17:31 +0100 | [diff] [blame] | 65 | * - Upgrade default pageSize to 50. |
| Akron | ea77cb4 | 2020-11-18 14:04:21 +0100 | [diff] [blame] | 66 | * - Add loading marker. |
| Akron | e2db9dd | 2020-11-19 13:35:12 +0100 | [diff] [blame] | 67 | * - Add infos to JsonExporter. |
| Akron | 18c627d | 2020-11-26 22:09:08 +0100 | [diff] [blame] | 68 | * - e.g. q & cq string representation. |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 69 | * - Check pageSize after init (so pageSize is not |
| 70 | * greater than what the server supports). |
| 71 | * - Restrict CORS to meaningful sources. |
| 72 | * - Add arbitrary information for RTF header |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 73 | * - Add information regarding max_exp_limit |
| 74 | * to export form. |
| 75 | * - Maybe set matches from parent window |
| 76 | * (if available) as export default (if |
| 77 | * smaller than max_exp_limit) |
| 78 | * - IDS-internal user should be allowed 100.000 |
| 79 | * matches per export, while external users |
| 80 | * should be limited to 10.000. |
| 81 | * - Add 1000-separator to numbers. |
| Akron | a578eb0 | 2020-11-26 12:34:17 +0100 | [diff] [blame] | 82 | * - Get the list of availables locales based |
| 83 | * on the given ResourceBundle. |
| 84 | * - Check for q/ql definition in JS. |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 85 | * |
| 86 | * IDEAS: |
| Akron | 18c627d | 2020-11-26 22:09:08 +0100 | [diff] [blame] | 87 | * - Support more granular exporter-specific options |
| 88 | * - Create a template mechanism for RTF export. |
| 89 | * - Support CSV separator and quote symbol change. |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 90 | * - Prettify VC in RTF export (maybe similar to |
| 91 | * the visualisation in Kalamar) |
| hebasta | fed4194 | 2020-11-02 10:00:36 +0100 | [diff] [blame] | 92 | */ |
| hebasta | a66693a | 2020-07-19 16:51:28 +0200 | [diff] [blame] | 93 | |
| hebasta | e049676 | 2020-03-30 00:11:01 +0200 | [diff] [blame] | 94 | @Path("/") |
| Akron | 3436079 | 2020-11-20 15:06:00 +0100 | [diff] [blame] | 95 | public class Service { |
| hebasta | e2bd6e4 | 2020-03-03 19:07:20 +0100 | [diff] [blame] | 96 | |
| Akron | 3588101 | 2020-11-24 20:05:06 +0100 | [diff] [blame] | 97 | private Properties prop = ExWSConf.properties(null); |
| Akron | b87af2f | 2020-09-24 14:38:02 +0200 | [diff] [blame] | 98 | |
| Akron | bb076c7 | 2020-09-28 14:39:15 +0200 | [diff] [blame] | 99 | private final ClassLoader cl = Thread.currentThread().getContextClassLoader(); |
| 100 | |
| 101 | InputStream is = cl.getResourceAsStream("assets/export.js"); |
| 102 | private final String exportJsStr = streamToString(is); |
| Akron | 7f242bb | 2020-09-28 20:19:11 +0200 | [diff] [blame] | 103 | |
| Helge | 605cf6d | 2023-03-24 16:04:19 +0100 | [diff] [blame] | 104 | Configuration cfg = new Configuration(Configuration.VERSION_2_3_32); |
| 105 | |
| Akron | 7f242bb | 2020-09-28 20:19:11 +0200 | [diff] [blame] | 106 | { |
| Akron | 3436079 | 2020-11-20 15:06:00 +0100 | [diff] [blame] | 107 | cfg.setClassForTemplateLoading(Service.class, "/assets/templates"); |
| Akron | 7f242bb | 2020-09-28 20:19:11 +0200 | [diff] [blame] | 108 | cfg.setDefaultEncoding("UTF-8"); |
| 109 | } |
| Akron | be6777b | 2020-10-02 11:35:03 +0200 | [diff] [blame] | 110 | |
| 111 | private final static String octets = |
| 112 | "(?:25[0-5]|2[0-4][0-9]|[0-1]?[0-9]{1,2})"; |
| 113 | |
| 114 | private final static String ipre = |
| 115 | octets + "\\." + octets + "\\." + octets + "\\." + octets; |
| Akron | 669d0bb | 2020-10-02 13:12:01 +0200 | [diff] [blame] | 116 | |
| 117 | private static Pattern authrep = Pattern.compile("\"auth\":\"([^\"]+?)\""); |
| 118 | |
| 119 | private final static Base64.Decoder b64Dec = Base64.getDecoder(); |
| 120 | |
| Akron | a241a47 | 2021-04-27 18:55:58 +0200 | [diff] [blame] | 121 | |
| Akron | 669d0bb | 2020-10-02 13:12:01 +0200 | [diff] [blame] | 122 | @Context |
| Akron | a578eb0 | 2020-11-26 12:34:17 +0100 | [diff] [blame] | 123 | private HttpServletRequest servletReq; |
| 124 | |
| 125 | @Context |
| 126 | private ContainerRequest req; |
| Akron | e57937b | 2020-11-17 08:49:31 +0100 | [diff] [blame] | 127 | |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 128 | /* |
| 129 | * Private method to run the export, |
| 130 | * either static or streaming |
| 131 | */ |
| Akron | 3588101 | 2020-11-24 20:05:06 +0100 | [diff] [blame] | 132 | private Exporter export(String fname, |
| Akron | b91c303 | 2020-11-23 19:24:40 +0100 | [diff] [blame] | 133 | String format, |
| 134 | String q, |
| 135 | String cq, |
| 136 | String ql, |
| 137 | String cutoffStr, |
| 138 | int hitc, |
| Akron | af145eb | 2020-11-24 16:55:47 +0100 | [diff] [blame] | 139 | EventOutput eventOutput |
| 140 | ) throws WebApplicationException { |
| Akron | d0b1cfe | 2020-11-20 19:26:52 +0100 | [diff] [blame] | 141 | |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 142 | // These parameters are mandatory |
| Akron | 90799f1 | 2020-09-17 17:10:34 +0200 | [diff] [blame] | 143 | String[][] params = { |
| Akron | 90799f1 | 2020-09-17 17:10:34 +0200 | [diff] [blame] | 144 | { "format", format }, |
| 145 | { "q", q }, |
| 146 | { "ql", ql } |
| 147 | }; |
| hebasta | e049676 | 2020-03-30 00:11:01 +0200 | [diff] [blame] | 148 | |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 149 | // Check that all mandatory parameters are available |
| hebasta | ff955d6 | 2020-05-27 15:59:14 +0200 | [diff] [blame] | 150 | for (int i = 0; i < params.length; i++) { |
| 151 | if (params[i][1] == null || params[i][1].trim().isEmpty()) |
| Akron | b91c303 | 2020-11-23 19:24:40 +0100 | [diff] [blame] | 152 | throw new WebApplicationException( |
| 153 | responseForm(Status.BAD_REQUEST, |
| 154 | "Parameter " + "\"" |
| 155 | + params[i][0] + "\"" + |
| 156 | " is missing or empty")); |
| Akron | 232248d | 2020-11-11 18:55:10 +0100 | [diff] [blame] | 157 | }; |
| Akron | 232248d | 2020-11-11 18:55:10 +0100 | [diff] [blame] | 158 | |
| 159 | // Retrieve cutoff value |
| hebasta | fed4194 | 2020-11-02 10:00:36 +0100 | [diff] [blame] | 160 | boolean cutoff = false; |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 161 | if (cutoffStr != null && ( |
| 162 | cutoffStr.equals("true") || |
| 163 | cutoffStr.equals("1")) |
| 164 | ) { |
| Akron | 232248d | 2020-11-11 18:55:10 +0100 | [diff] [blame] | 165 | cutoff = true; |
| 166 | }; |
| Akron | af145eb | 2020-11-24 16:55:47 +0100 | [diff] [blame] | 167 | |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 168 | // Load configuration values |
| Akron | 69cd35d | 2020-11-20 13:17:31 +0100 | [diff] [blame] | 169 | String scheme = prop.getProperty("api.scheme", "https"); |
| 170 | String port = prop.getProperty("api.port", "8089"); |
| 171 | String host = prop.getProperty("api.host", "localhost"); |
| 172 | String path = prop.getProperty("api.path", ""); |
| Akron | 5405dec | 2020-11-26 20:39:24 +0100 | [diff] [blame] | 173 | String source = prop.getProperty("api.source"); |
| Akron | 69cd35d | 2020-11-20 13:17:31 +0100 | [diff] [blame] | 174 | int pageSize = Integer.parseInt(prop.getProperty("conf.page_size", "5")); |
| 175 | int maxResults = Integer.parseInt(prop.getProperty("conf.max_exp_limit", "10000")); |
| hebasta | a66693a | 2020-07-19 16:51:28 +0200 | [diff] [blame] | 176 | |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 177 | // Adjust the number of requested hits |
| Akron | 99fda9e | 2020-11-20 13:45:02 +0100 | [diff] [blame] | 178 | if (hitc > 0 && hitc < maxResults) |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 179 | maxResults = hitc; |
| Akron | 99fda9e | 2020-11-20 13:45:02 +0100 | [diff] [blame] | 180 | |
| 181 | // If less than pageSize results are requested - dont't fetch more |
| 182 | if (maxResults < pageSize) |
| 183 | pageSize = maxResults; |
| Akron | b91c303 | 2020-11-23 19:24:40 +0100 | [diff] [blame] | 184 | |
| 185 | ResponseBuilder builder = null; |
| 186 | Client client = ClientBuilder.newClient(); |
| 187 | |
| Akron | e57937b | 2020-11-17 08:49:31 +0100 | [diff] [blame] | 188 | // Create initial search uri |
| Akron | bfd5fb8 | 2020-09-15 14:28:25 +0200 | [diff] [blame] | 189 | UriBuilder uri = UriBuilder.fromPath("/api/v1.0/search") |
| 190 | .host(host) |
| 191 | .port(Integer.parseInt(port)) |
| Akron | 480a908 | 2020-09-16 13:14:09 +0200 | [diff] [blame] | 192 | .scheme(scheme) |
| Akron | bfd5fb8 | 2020-09-15 14:28:25 +0200 | [diff] [blame] | 193 | .queryParam("q", q) |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 194 | .queryParam("context", "40-t,40-t") |
| Akron | bfd5fb8 | 2020-09-15 14:28:25 +0200 | [diff] [blame] | 195 | .queryParam("ql", ql) |
| Akron | 99fda9e | 2020-11-20 13:45:02 +0100 | [diff] [blame] | 196 | .queryParam("count", pageSize) |
| Akron | bfd5fb8 | 2020-09-15 14:28:25 +0200 | [diff] [blame] | 197 | ; |
| Akron | 68f221b | 2020-09-29 23:15:43 +0200 | [diff] [blame] | 198 | |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 199 | // Not yet supported: |
| 200 | // .queryParam("context", "sentence") |
| 201 | |
| Akron | 0bdcab2 | 2020-11-19 18:36:16 +0100 | [diff] [blame] | 202 | if (cq != null && cq.length() > 0) |
| Akron | d2072ee | 2020-11-17 16:12:41 +0100 | [diff] [blame] | 203 | uri = uri.queryParam("cq", cq); |
| Akron | d2072ee | 2020-11-17 16:12:41 +0100 | [diff] [blame] | 204 | |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 205 | if (path != "") |
| Akron | 68f221b | 2020-09-29 23:15:43 +0200 | [diff] [blame] | 206 | uri = uri.path(path); |
| Akron | 69cd35d | 2020-11-20 13:17:31 +0100 | [diff] [blame] | 207 | |
| Akron | be6777b | 2020-10-02 11:35:03 +0200 | [diff] [blame] | 208 | // Get client IP, in case service is behind a proxy |
| Akron | 669d0bb | 2020-10-02 13:12:01 +0200 | [diff] [blame] | 209 | // Get auth (temporarily) via Session riding |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 210 | String xff = "", auth = ""; |
| Akron | a578eb0 | 2020-11-26 12:34:17 +0100 | [diff] [blame] | 211 | if (servletReq != null) { |
| 212 | xff = getClientIP(servletReq.getHeader("X-Forwarded-For")); |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 213 | if (xff == "") |
| Akron | a578eb0 | 2020-11-26 12:34:17 +0100 | [diff] [blame] | 214 | xff = servletReq.getRemoteAddr(); |
| Akron | 669d0bb | 2020-10-02 13:12:01 +0200 | [diff] [blame] | 215 | |
| Akron | a578eb0 | 2020-11-26 12:34:17 +0100 | [diff] [blame] | 216 | auth = authFromCookie(servletReq); |
| Akron | be6777b | 2020-10-02 11:35:03 +0200 | [diff] [blame] | 217 | }; |
| Akron | 669d0bb | 2020-10-02 13:12:01 +0200 | [diff] [blame] | 218 | |
| Akron | 328f9ea | 2020-09-22 17:11:22 +0200 | [diff] [blame] | 219 | String resp; |
| hebasta | fed4194 | 2020-11-02 10:00:36 +0100 | [diff] [blame] | 220 | WebTarget resource; |
| Akron | 696c063 | 2020-11-12 15:30:51 +0100 | [diff] [blame] | 221 | Invocation.Builder reqBuilder; |
| Akron | 69cd35d | 2020-11-20 13:17:31 +0100 | [diff] [blame] | 222 | |
| Akron | 328f9ea | 2020-09-22 17:11:22 +0200 | [diff] [blame] | 223 | try { |
| hebasta | fed4194 | 2020-11-02 10:00:36 +0100 | [diff] [blame] | 224 | resource = client.target(uri.build()); |
| Akron | 696c063 | 2020-11-12 15:30:51 +0100 | [diff] [blame] | 225 | reqBuilder = resource.request(MediaType.APPLICATION_JSON); |
| 226 | resp = authBuilder(reqBuilder, xff, auth).get(String.class); |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 227 | } |
| 228 | |
| 229 | catch (Exception e) { |
| Akron | 328f9ea | 2020-09-22 17:11:22 +0200 | [diff] [blame] | 230 | throw new WebApplicationException( |
| Akron | 7f242bb | 2020-09-28 20:19:11 +0200 | [diff] [blame] | 231 | responseForm(Status.BAD_GATEWAY, "Unable to reach Backend") |
| Akron | 328f9ea | 2020-09-22 17:11:22 +0200 | [diff] [blame] | 232 | ); |
| 233 | } |
| hebasta | e049676 | 2020-03-30 00:11:01 +0200 | [diff] [blame] | 234 | |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 235 | // Get and initialize exporter based on requested format |
| Akron | 3588101 | 2020-11-24 20:05:06 +0100 | [diff] [blame] | 236 | Exporter exp = getExporter(format); |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 237 | exp.setMaxResults(maxResults); |
| Akron | 7412271 | 2020-11-17 09:41:21 +0100 | [diff] [blame] | 238 | exp.setQueryString(q); |
| Akron | d2072ee | 2020-11-17 16:12:41 +0100 | [diff] [blame] | 239 | exp.setCorpusQueryString(cq); |
| Akron | 5405dec | 2020-11-26 20:39:24 +0100 | [diff] [blame] | 240 | if (source != null) |
| 241 | exp.setSource(source); |
| 242 | else |
| 243 | exp.setSource(host, path); |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 244 | |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 245 | // Set filename |
| 246 | if (fname != null) |
| Akron | 7412271 | 2020-11-17 09:41:21 +0100 | [diff] [blame] | 247 | exp.setFileName(fname); |
| Akron | 99fda9e | 2020-11-20 13:45:02 +0100 | [diff] [blame] | 248 | |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 249 | // Set progress mechanism, if passed |
| Akron | 3588101 | 2020-11-24 20:05:06 +0100 | [diff] [blame] | 250 | if (eventOutput != null) { |
| Akron | af145eb | 2020-11-24 16:55:47 +0100 | [diff] [blame] | 251 | exp.setSse(eventOutput); |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 252 | |
| 253 | // Progress requires the creation |
| 254 | // of temporary files |
| Akron | 3588101 | 2020-11-24 20:05:06 +0100 | [diff] [blame] | 255 | exp.forceFile(); |
| 256 | }; |
| Akron | d0b1cfe | 2020-11-20 19:26:52 +0100 | [diff] [blame] | 257 | |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 258 | // Initialize export with meta data |
| 259 | // and first matches |
| Akron | acc9f7a | 2020-11-17 17:21:40 +0100 | [diff] [blame] | 260 | try { |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 261 | |
| 262 | // TODO: |
| 263 | // Check return value. |
| Akron | acc9f7a | 2020-11-17 17:21:40 +0100 | [diff] [blame] | 264 | exp.init(resp); |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 265 | } |
| 266 | |
| 267 | catch (Exception e) { |
| Akron | a241a47 | 2021-04-27 18:55:58 +0200 | [diff] [blame] | 268 | Logger.error(e); |
| Akron | 60a1628 | 2021-04-28 20:01:11 +0200 | [diff] [blame] | 269 | String err = e.getMessage(); |
| 270 | if (err == null) { |
| 271 | err = "Unable to initialize export"; |
| 272 | }; |
| 273 | |
| Akron | acc9f7a | 2020-11-17 17:21:40 +0100 | [diff] [blame] | 274 | throw new WebApplicationException( |
| 275 | responseForm( |
| 276 | Status.INTERNAL_SERVER_ERROR, |
| Akron | 60a1628 | 2021-04-28 20:01:11 +0200 | [diff] [blame] | 277 | err |
| Akron | acc9f7a | 2020-11-17 17:21:40 +0100 | [diff] [blame] | 278 | ) |
| 279 | ); |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 280 | }; |
| Akron | d2072ee | 2020-11-17 16:12:41 +0100 | [diff] [blame] | 281 | |
| Akron | 69cd35d | 2020-11-20 13:17:31 +0100 | [diff] [blame] | 282 | // Calculate how many results to fetch |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 283 | int fetchCount = exp.getTotalResults(); |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 284 | if (exp.hasTimeExceeded() || fetchCount > maxResults) |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 285 | fetchCount = maxResults; |
| Akron | d0b1cfe | 2020-11-20 19:26:52 +0100 | [diff] [blame] | 286 | |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 287 | // fetchCount may be different to maxResults now, |
| 288 | // so reset after init (for accurate progress) |
| Akron | b91c303 | 2020-11-23 19:24:40 +0100 | [diff] [blame] | 289 | exp.setMaxResults(fetchCount); |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 290 | |
| Akron | d3df5b0 | 2020-11-12 18:23:17 +0100 | [diff] [blame] | 291 | // If only one page should be exported there is no need |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 292 | // for a temporary export file, unless progress is |
| 293 | // requested. In case all matches are already fetched, |
| 294 | // stop here as well. |
| 295 | if (cutoff || fetchCount <= pageSize) { |
| 296 | |
| Akron | 3588101 | 2020-11-24 20:05:06 +0100 | [diff] [blame] | 297 | try { |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 298 | |
| 299 | // Close all export writers |
| Akron | 3588101 | 2020-11-24 20:05:06 +0100 | [diff] [blame] | 300 | exp.finish(); |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 301 | } |
| 302 | |
| 303 | catch (Exception e) { |
| Akron | a241a47 | 2021-04-27 18:55:58 +0200 | [diff] [blame] | 304 | Logger.error(e); |
| Akron | 60a1628 | 2021-04-28 20:01:11 +0200 | [diff] [blame] | 305 | String err = e.getMessage(); |
| 306 | if (err == null) { |
| 307 | err = "Unable to finish export"; |
| 308 | }; |
| 309 | |
| Akron | 3588101 | 2020-11-24 20:05:06 +0100 | [diff] [blame] | 310 | throw new WebApplicationException( |
| 311 | responseForm( |
| 312 | Status.INTERNAL_SERVER_ERROR, |
| Akron | 60a1628 | 2021-04-28 20:01:11 +0200 | [diff] [blame] | 313 | err |
| Akron | 3588101 | 2020-11-24 20:05:06 +0100 | [diff] [blame] | 314 | ) |
| 315 | ); |
| 316 | }; |
| Akron | b91c303 | 2020-11-23 19:24:40 +0100 | [diff] [blame] | 317 | return exp; |
| Akron | d0b1cfe | 2020-11-20 19:26:52 +0100 | [diff] [blame] | 318 | }; |
| hebasta | e049676 | 2020-03-30 00:11:01 +0200 | [diff] [blame] | 319 | |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 320 | /* |
| 321 | * Page through all results |
| 322 | */ |
| hebasta | a66693a | 2020-07-19 16:51:28 +0200 | [diff] [blame] | 323 | |
| Akron | d0b1cfe | 2020-11-20 19:26:52 +0100 | [diff] [blame] | 324 | // It's not important anymore to get totalResults |
| 325 | uri.queryParam("cutoff", "true"); |
| Akron | d3df5b0 | 2020-11-12 18:23:17 +0100 | [diff] [blame] | 326 | |
| Akron | d0b1cfe | 2020-11-20 19:26:52 +0100 | [diff] [blame] | 327 | // Set offset for paging as a template |
| 328 | uri.queryParam("offset", "{offset}"); |
| Akron | 9690796 | 2020-11-12 18:47:07 +0100 | [diff] [blame] | 329 | |
| Akron | d0b1cfe | 2020-11-20 19:26:52 +0100 | [diff] [blame] | 330 | try { |
| Akron | acc9f7a | 2020-11-17 17:21:40 +0100 | [diff] [blame] | 331 | |
| Akron | d0b1cfe | 2020-11-20 19:26:52 +0100 | [diff] [blame] | 332 | // Iterate over all results |
| 333 | for (int i = pageSize; i <= fetchCount; i+=pageSize) { |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 334 | |
| Akron | d0b1cfe | 2020-11-20 19:26:52 +0100 | [diff] [blame] | 335 | resource = client.target(uri.build(i)); |
| 336 | reqBuilder = resource.request(MediaType.APPLICATION_JSON); |
| 337 | resp = authBuilder(reqBuilder, xff, auth).get(String.class); |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 338 | |
| Akron | d0b1cfe | 2020-11-20 19:26:52 +0100 | [diff] [blame] | 339 | // Stop when no more matches are allowed |
| 340 | if (!exp.appendMatches(resp)) |
| 341 | break; |
| 342 | } |
| Akron | 3588101 | 2020-11-24 20:05:06 +0100 | [diff] [blame] | 343 | |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 344 | // Close all export writers |
| Akron | 3588101 | 2020-11-24 20:05:06 +0100 | [diff] [blame] | 345 | exp.finish(); |
| 346 | |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 347 | } |
| 348 | |
| 349 | catch (Exception e) { |
| Akron | a241a47 | 2021-04-27 18:55:58 +0200 | [diff] [blame] | 350 | Logger.error(e); |
| Akron | 60a1628 | 2021-04-28 20:01:11 +0200 | [diff] [blame] | 351 | String err = e.getMessage(); |
| 352 | if (err == null) { |
| 353 | err = "Unable to iterate through results"; |
| 354 | }; |
| Akron | d0b1cfe | 2020-11-20 19:26:52 +0100 | [diff] [blame] | 355 | throw new WebApplicationException( |
| 356 | responseForm( |
| 357 | Status.INTERNAL_SERVER_ERROR, |
| Akron | 60a1628 | 2021-04-28 20:01:11 +0200 | [diff] [blame] | 358 | err |
| Akron | d0b1cfe | 2020-11-20 19:26:52 +0100 | [diff] [blame] | 359 | ) |
| 360 | ); |
| 361 | }; |
| Akron | b91c303 | 2020-11-23 19:24:40 +0100 | [diff] [blame] | 362 | |
| 363 | return exp; |
| 364 | }; |
| 365 | |
| 366 | |
| 367 | /** |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 368 | * WebService that retrieves data from the Kustvakt |
| 369 | * Webservice and returns response in different formats. |
| 370 | * |
| 371 | * Returns an octet stream. |
| Akron | b91c303 | 2020-11-23 19:24:40 +0100 | [diff] [blame] | 372 | * |
| 373 | * @param fname |
| 374 | * file name |
| 375 | * @param format |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 376 | * the file format value |
| Akron | b91c303 | 2020-11-23 19:24:40 +0100 | [diff] [blame] | 377 | * @param q |
| 378 | * the query |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 379 | * @param cq |
| 380 | * the corpus query |
| Akron | b91c303 | 2020-11-23 19:24:40 +0100 | [diff] [blame] | 381 | * @param ql |
| 382 | * the query language |
| 383 | * @param cutoff |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 384 | * Only export the first page |
| 385 | * @param hitc |
| 386 | * Number of matches to fetch |
| Akron | b91c303 | 2020-11-23 19:24:40 +0100 | [diff] [blame] | 387 | */ |
| 388 | @POST |
| 389 | @Path("export") |
| 390 | @Produces(MediaType.APPLICATION_OCTET_STREAM) |
| 391 | public Response staticExport ( |
| 392 | @FormParam("fname") String fname, |
| 393 | @FormParam("format") String format, |
| 394 | @FormParam("q") String q, |
| 395 | @FormParam("cq") String cq, |
| 396 | @FormParam("ql") String ql, |
| 397 | @FormParam("cutoff") String cutoffStr, |
| 398 | @FormParam("hitc") int hitc |
| Akron | b91c303 | 2020-11-23 19:24:40 +0100 | [diff] [blame] | 399 | ) throws IOException { |
| 400 | |
| Akron | af145eb | 2020-11-24 16:55:47 +0100 | [diff] [blame] | 401 | Exporter exp = export(fname, format, q, cq, ql, cutoffStr, hitc, null); |
| Akron | d0b1cfe | 2020-11-20 19:26:52 +0100 | [diff] [blame] | 402 | |
| 403 | return exp.serve().build(); |
| Akron | 9690796 | 2020-11-12 18:47:07 +0100 | [diff] [blame] | 404 | }; |
| Akron | af145eb | 2020-11-24 16:55:47 +0100 | [diff] [blame] | 405 | |
| Akron | 9690796 | 2020-11-12 18:47:07 +0100 | [diff] [blame] | 406 | |
| Akron | b91c303 | 2020-11-23 19:24:40 +0100 | [diff] [blame] | 407 | /** |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 408 | * WebService that retrieves data from the Kustvakt |
| 409 | * Webservice and returns response in different formats. |
| 410 | * |
| 411 | * Returns an event stream. |
| 412 | * |
| 413 | * @param fname |
| 414 | * file name |
| 415 | * @param format |
| 416 | * the file format value |
| 417 | * @param q |
| 418 | * the query |
| 419 | * @param cq |
| 420 | * the corpus query |
| 421 | * @param ql |
| 422 | * the query language |
| 423 | * @param cutoff |
| 424 | * Only export the first page |
| 425 | * @param hitc |
| 426 | * Number of matches to fetch |
| Akron | b91c303 | 2020-11-23 19:24:40 +0100 | [diff] [blame] | 427 | */ |
| Akron | b87af2f | 2020-09-24 14:38:02 +0200 | [diff] [blame] | 428 | @GET |
| Akron | 83f3994 | 2021-03-22 16:37:25 +0100 | [diff] [blame] | 429 | @Path("export") |
| Akron | af145eb | 2020-11-24 16:55:47 +0100 | [diff] [blame] | 430 | @Produces(SseFeature.SERVER_SENT_EVENTS) |
| 431 | @Consumes(SseFeature.SERVER_SENT_EVENTS) |
| 432 | public Response progressExport( |
| Akron | b91c303 | 2020-11-23 19:24:40 +0100 | [diff] [blame] | 433 | @QueryParam("fname") String fname, |
| 434 | @QueryParam("format") String format, |
| 435 | @QueryParam("q") String q, |
| 436 | @QueryParam("cq") String cq, |
| 437 | @QueryParam("ql") String ql, |
| 438 | @QueryParam("cutoff") String cutoffStr, |
| 439 | @QueryParam("hitc") int hitc |
| 440 | ) throws InterruptedException { |
| Akron | d0b1cfe | 2020-11-20 19:26:52 +0100 | [diff] [blame] | 441 | |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 442 | // See |
| 443 | // https://www.baeldung.com/java-ee-jax-rs-sse |
| 444 | // https://www.howopensource.com/2016/01/java-sse-chat-example/ |
| 445 | // https://csetutorials.com/jersey-sse-tutorial.html |
| 446 | // https://eclipse-ee4j.github.io/jersey.github.io/documentation/latest/sse.html |
| Akron | af145eb | 2020-11-24 16:55:47 +0100 | [diff] [blame] | 447 | |
| 448 | final EventOutput eventOutput = new EventOutput(); |
| Akron | d0b1cfe | 2020-11-20 19:26:52 +0100 | [diff] [blame] | 449 | |
| Akron | b91c303 | 2020-11-23 19:24:40 +0100 | [diff] [blame] | 450 | // Send initial event |
| Akron | af145eb | 2020-11-24 16:55:47 +0100 | [diff] [blame] | 451 | if (eventOutput.isClosed()) |
| 452 | return Response.ok("EventSource closed").build(); |
| Akron | b91c303 | 2020-11-23 19:24:40 +0100 | [diff] [blame] | 453 | |
| Marc Kupietz | a2a1797 | 2022-03-27 12:03:48 +0200 | [diff] [blame^] | 454 | Thread t = new Thread( |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 455 | new Runnable() { |
| 456 | |
| Akron | af145eb | 2020-11-24 16:55:47 +0100 | [diff] [blame] | 457 | @Override |
| 458 | public void run() { |
| 459 | final OutboundEvent.Builder eventBuilder = new OutboundEvent.Builder(); |
| 460 | try { |
| 461 | eventBuilder.name("Process"); |
| 462 | eventBuilder.data("init"); |
| 463 | eventOutput.write(eventBuilder.build()); |
| 464 | Exporter exp = export( |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 465 | fname, format, q, cq, ql, cutoffStr, hitc, eventOutput |
| Akron | af145eb | 2020-11-24 16:55:47 +0100 | [diff] [blame] | 466 | ); |
| Akron | 3588101 | 2020-11-24 20:05:06 +0100 | [diff] [blame] | 467 | |
| Akron | af145eb | 2020-11-24 16:55:47 +0100 | [diff] [blame] | 468 | if (eventOutput.isClosed()) |
| 469 | return; |
| Akron | ba3ea11 | 2020-11-24 22:40:18 +0100 | [diff] [blame] | 470 | |
| Akron | af145eb | 2020-11-24 16:55:47 +0100 | [diff] [blame] | 471 | eventBuilder.name("Relocate"); |
| Akron | ba3ea11 | 2020-11-24 22:40:18 +0100 | [diff] [blame] | 472 | eventBuilder.data(exp.getExportID() + ";" + exp.getFileName()); |
| Akron | af145eb | 2020-11-24 16:55:47 +0100 | [diff] [blame] | 473 | eventOutput.write(eventBuilder.build()); |
| Akron | ba3ea11 | 2020-11-24 22:40:18 +0100 | [diff] [blame] | 474 | |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 475 | } |
| 476 | |
| 477 | catch (Exception e) { |
| Akron | af145eb | 2020-11-24 16:55:47 +0100 | [diff] [blame] | 478 | try { |
| 479 | if (eventOutput.isClosed()) |
| 480 | return; |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 481 | |
| Akron | af145eb | 2020-11-24 16:55:47 +0100 | [diff] [blame] | 482 | eventBuilder.name("Error"); |
| 483 | eventBuilder.data(e.getMessage()); |
| 484 | eventOutput.write(eventBuilder.build()); |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 485 | } |
| 486 | |
| 487 | catch (IOException ioe) { |
| Akron | a241a47 | 2021-04-27 18:55:58 +0200 | [diff] [blame] | 488 | Logger.error(ioe); |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 489 | throw new RuntimeException( |
| 490 | "Error when writing event output.", ioe |
| 491 | ); |
| Akron | af145eb | 2020-11-24 16:55:47 +0100 | [diff] [blame] | 492 | }; |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 493 | } |
| 494 | |
| 495 | finally { |
| Akron | af145eb | 2020-11-24 16:55:47 +0100 | [diff] [blame] | 496 | try { |
| 497 | if (eventOutput.isClosed()) |
| 498 | return; |
| Akron | d0b1cfe | 2020-11-20 19:26:52 +0100 | [diff] [blame] | 499 | |
| Akron | af145eb | 2020-11-24 16:55:47 +0100 | [diff] [blame] | 500 | eventBuilder.name("Process"); |
| 501 | eventBuilder.data("done"); |
| 502 | eventOutput.write(eventBuilder.build()); |
| 503 | eventOutput.close(); |
| Marc Kupietz | a2a1797 | 2022-03-27 12:03:48 +0200 | [diff] [blame^] | 504 | } |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 505 | |
| 506 | catch (IOException ioClose) { |
| Akron | a241a47 | 2021-04-27 18:55:58 +0200 | [diff] [blame] | 507 | Logger.error(ioClose); |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 508 | throw new RuntimeException( |
| 509 | "Error when closing the event output.", ioClose |
| 510 | ); |
| Akron | af145eb | 2020-11-24 16:55:47 +0100 | [diff] [blame] | 511 | } |
| 512 | }; |
| 513 | return; |
| 514 | } |
| Marc Kupietz | a2a1797 | 2022-03-27 12:03:48 +0200 | [diff] [blame^] | 515 | }); |
| 516 | t.start(); |
| 517 | // t.join(); |
| Akron | af145eb | 2020-11-24 16:55:47 +0100 | [diff] [blame] | 518 | |
| Akron | 1cbe798 | 2021-04-27 17:18:30 +0200 | [diff] [blame] | 519 | String origin = prop.getProperty("server.origin","*"); |
| 520 | if (servletReq != null) { |
| 521 | // This is temporary to allow for session riding |
| 522 | origin = servletReq.getHeader("Origin"); |
| 523 | }; |
| 524 | |
| Marc Kupietz | a2a1797 | 2022-03-27 12:03:48 +0200 | [diff] [blame^] | 525 | return Response.ok(eventOutput, String.valueOf(SseFeature.SERVER_SENT_EVENTS_TYPE)) |
| Akron | 1cbe798 | 2021-04-27 17:18:30 +0200 | [diff] [blame] | 526 | .header("Access-Control-Allow-Origin", origin) |
| 527 | .header("Access-Control-Allow-Credentials", "true") |
| 528 | .header("Vary","Origin") |
| Akron | af145eb | 2020-11-24 16:55:47 +0100 | [diff] [blame] | 529 | .build(); |
| Akron | d0b1cfe | 2020-11-20 19:26:52 +0100 | [diff] [blame] | 530 | }; |
| 531 | |
| Akron | 3588101 | 2020-11-24 20:05:06 +0100 | [diff] [blame] | 532 | |
| 533 | /** |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 534 | * Relocation target to which the event |
| Akron | 3588101 | 2020-11-24 20:05:06 +0100 | [diff] [blame] | 535 | * stream points to. |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 536 | * |
| 537 | * Returns an octet stream. |
| 538 | * |
| 539 | * @param fname |
| 540 | * file name |
| 541 | * @param file |
| 542 | * the file to fetch |
| Akron | 3588101 | 2020-11-24 20:05:06 +0100 | [diff] [blame] | 543 | */ |
| 544 | @GET |
| 545 | @Path("export/{file}") |
| 546 | @Produces(MediaType.APPLICATION_OCTET_STREAM) |
| 547 | public Response fileExport( |
| 548 | @PathParam("file") String fileStr, |
| 549 | @QueryParam("fname") String fname |
| 550 | ) { |
| 551 | |
| 552 | String format = getExtension(fileStr); |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 553 | |
| Akron | 3588101 | 2020-11-24 20:05:06 +0100 | [diff] [blame] | 554 | // Get exporter object |
| 555 | Exporter exp = getExporter(format); |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 556 | |
| 557 | if (fname != null) |
| Akron | 3588101 | 2020-11-24 20:05:06 +0100 | [diff] [blame] | 558 | exp.setFileName(fname); |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 559 | |
| Akron | 3588101 | 2020-11-24 20:05:06 +0100 | [diff] [blame] | 560 | exp.setFile(fileStr); |
| 561 | |
| 562 | // Return without init |
| 563 | return exp.serve().build(); |
| 564 | }; |
| 565 | |
| Akron | d0b1cfe | 2020-11-20 19:26:52 +0100 | [diff] [blame] | 566 | |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 567 | /** |
| 568 | * The export form. |
| 569 | * |
| 570 | * Returns a HTML file. |
| 571 | */ |
| Akron | d0b1cfe | 2020-11-20 19:26:52 +0100 | [diff] [blame] | 572 | @GET |
| Akron | b87af2f | 2020-09-24 14:38:02 +0200 | [diff] [blame] | 573 | @Path("export") |
| 574 | @Produces(MediaType.TEXT_HTML) |
| 575 | public Response exportHTML () { |
| Akron | 7f242bb | 2020-09-28 20:19:11 +0200 | [diff] [blame] | 576 | return responseForm(); |
| 577 | }; |
| Akron | bb076c7 | 2020-09-28 14:39:15 +0200 | [diff] [blame] | 578 | |
| Akron | 7f242bb | 2020-09-28 20:19:11 +0200 | [diff] [blame] | 579 | |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 580 | /** |
| 581 | * The export script. |
| 582 | * |
| 583 | * Returns a static JavaScript file. |
| 584 | */ |
| Akron | 9690796 | 2020-11-12 18:47:07 +0100 | [diff] [blame] | 585 | @GET |
| 586 | @Path("export.js") |
| 587 | @Produces("application/javascript") |
| 588 | public Response exportJavascript () { |
| 589 | return Response |
| 590 | .ok(exportJsStr, "application/javascript") |
| 591 | .build(); |
| 592 | }; |
| Akron | 3588101 | 2020-11-24 20:05:06 +0100 | [diff] [blame] | 593 | |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 594 | |
| 595 | /* |
| 596 | * Get exporter object by format |
| 597 | */ |
| Akron | 3588101 | 2020-11-24 20:05:06 +0100 | [diff] [blame] | 598 | private Exporter getExporter (String format) { |
| 599 | // Choose the correct exporter |
| 600 | if (format.equals("json")) |
| 601 | return new JsonExporter(); |
| 602 | else if (format.equals("csv")) |
| 603 | return new CsvExporter(); |
| 604 | |
| 605 | return new RtfExporter(); |
| 606 | }; |
| 607 | |
| Akron | 7f242bb | 2020-09-28 20:19:11 +0200 | [diff] [blame] | 608 | |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 609 | /* |
| 610 | * Decorate request with auth headers |
| 611 | */ |
| 612 | private Invocation.Builder authBuilder ( |
| 613 | Invocation.Builder reqBuilder, |
| 614 | String xff, |
| 615 | String auth |
| 616 | ) { |
| 617 | |
| 618 | if (xff != "") |
| Akron | 696c063 | 2020-11-12 15:30:51 +0100 | [diff] [blame] | 619 | reqBuilder = reqBuilder.header("X-Forwarded-For", xff); |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 620 | |
| 621 | if (auth != "") |
| Akron | 696c063 | 2020-11-12 15:30:51 +0100 | [diff] [blame] | 622 | reqBuilder = reqBuilder.header("Authorization", auth); |
| Akron | 696c063 | 2020-11-12 15:30:51 +0100 | [diff] [blame] | 623 | |
| 624 | return reqBuilder; |
| 625 | }; |
| 626 | |
| Akron | 9690796 | 2020-11-12 18:47:07 +0100 | [diff] [blame] | 627 | |
| Akron | 984fe8f | 2020-11-25 15:21:37 +0100 | [diff] [blame] | 628 | /* |
| 629 | * Get authorization token from cookie |
| 630 | */ |
| Akron | 54b30ed | 2020-11-13 10:35:35 +0100 | [diff] [blame] | 631 | private String authFromCookie (HttpServletRequest r) { |
| 632 | |
| 633 | // This is a temporary solution using session riding - only |
| 634 | // valid for the time being |
| 635 | Cookie[] cookies = r.getCookies(); |
| Akron | 69cd35d | 2020-11-20 13:17:31 +0100 | [diff] [blame] | 636 | |
| 637 | if (cookies == null) |
| 638 | return ""; |
| 639 | |
| Akron | b4d3b2a | 2023-11-30 13:55:21 +0100 | [diff] [blame] | 640 | String cookieName = prop.getProperty("cookie.name", ""); |
| Akron | 54b30ed | 2020-11-13 10:35:35 +0100 | [diff] [blame] | 641 | |
| 642 | // Iterate through all cookies for a Kalamar session |
| 643 | for (int i = 0; i < cookies.length; i++) { |
| Akron | 1cbe798 | 2021-04-27 17:18:30 +0200 | [diff] [blame] | 644 | |
| Akron | b4d3b2a | 2023-11-30 13:55:21 +0100 | [diff] [blame] | 645 | // Check the valid name and ignore irrelevant cookies |
| 646 | if (cookieName == "") { |
| 647 | if (!cookies[i].getName().equals("kalamar")) { |
| 648 | continue; |
| 649 | } |
| 650 | } else if (!cookies[i].getName().equals(cookieName)) { |
| Akron | 54b30ed | 2020-11-13 10:35:35 +0100 | [diff] [blame] | 651 | continue; |
| Akron | b4d3b2a | 2023-11-30 13:55:21 +0100 | [diff] [blame] | 652 | }; |
| Akron | 54b30ed | 2020-11-13 10:35:35 +0100 | [diff] [blame] | 653 | |
| 654 | // Get the value |
| 655 | String b64 = cookies[i].getValue(); |
| 656 | String[] b64Parts = b64.split("--", 2); |
| 657 | if (b64Parts.length == 2) { |
| 658 | // Read the payload |
| 659 | String payload = new String(b64Dec.decode(b64Parts[0])); |
| 660 | if (payload != "") { |
| 661 | Matcher m = authrep.matcher(payload); |
| 662 | if (m.find()) { |
| 663 | return m.group(1); |
| 664 | }; |
| 665 | }; |
| 666 | }; |
| 667 | }; |
| Akron | 54b30ed | 2020-11-13 10:35:35 +0100 | [diff] [blame] | 668 | return ""; |
| 669 | }; |
| 670 | |
| 671 | |
| Akron | 9690796 | 2020-11-12 18:47:07 +0100 | [diff] [blame] | 672 | /* |
| 673 | * Response with form template. |
| 674 | */ |
| 675 | private Response responseForm () { |
| 676 | return responseForm(null, null); |
| Akron | 54b30ed | 2020-11-13 10:35:35 +0100 | [diff] [blame] | 677 | }; |
| Akron | 9690796 | 2020-11-12 18:47:07 +0100 | [diff] [blame] | 678 | |
| 679 | |
| 680 | /* |
| 681 | * Response with form template. |
| 682 | * |
| 683 | * Accepts an error code and message. |
| 684 | */ |
| Akron | 7f242bb | 2020-09-28 20:19:11 +0200 | [diff] [blame] | 685 | private Response responseForm (Status code, String msg) { |
| Akron | bb076c7 | 2020-09-28 14:39:15 +0200 | [diff] [blame] | 686 | StringWriter out = new StringWriter(); |
| 687 | HashMap<String, Object> templateData = new HashMap<String, Object>(); |
| 688 | |
| Akron | 69cd35d | 2020-11-20 13:17:31 +0100 | [diff] [blame] | 689 | // Build uri for assets |
| 690 | String scheme = prop.getProperty("asset.scheme", "https"); |
| 691 | String port = prop.getProperty("asset.port", ""); |
| 692 | String host = prop.getProperty("asset.host", "korap.ids-mannheim.de"); |
| 693 | String path = prop.getProperty("asset.path", ""); |
| Akron | a77f2f0 | 2020-11-26 16:37:48 +0100 | [diff] [blame] | 694 | String defaultHitc = prop.getProperty("conf.default_hitc", "100"); |
| 695 | int maxHitc = Integer.parseInt(prop.getProperty("conf.max_exp_limit", "10000")); |
| Akron | bb076c7 | 2020-09-28 14:39:15 +0200 | [diff] [blame] | 696 | |
| 697 | UriBuilder uri = UriBuilder.fromPath("") |
| 698 | .host(host) |
| Akron | 68f221b | 2020-09-29 23:15:43 +0200 | [diff] [blame] | 699 | .scheme(scheme); |
| 700 | |
| Akron | 69cd35d | 2020-11-20 13:17:31 +0100 | [diff] [blame] | 701 | if (path != "") |
| Akron | 68f221b | 2020-09-29 23:15:43 +0200 | [diff] [blame] | 702 | uri = uri.path(path); |
| Akron | bb076c7 | 2020-09-28 14:39:15 +0200 | [diff] [blame] | 703 | |
| Akron | 69cd35d | 2020-11-20 13:17:31 +0100 | [diff] [blame] | 704 | if (port != "") |
| Akron | bb076c7 | 2020-09-28 14:39:15 +0200 | [diff] [blame] | 705 | uri = uri.port(Integer.parseInt(port)); |
| Akron | bb076c7 | 2020-09-28 14:39:15 +0200 | [diff] [blame] | 706 | |
| 707 | templateData.put("assetPath", uri.build()); |
| Akron | a77f2f0 | 2020-11-26 16:37:48 +0100 | [diff] [blame] | 708 | templateData.put("defaultHitc", defaultHitc); |
| 709 | templateData.put("maxHitc", maxHitc); |
| Akron | bb076c7 | 2020-09-28 14:39:15 +0200 | [diff] [blame] | 710 | |
| Akron | 69cd35d | 2020-11-20 13:17:31 +0100 | [diff] [blame] | 711 | // There is an error code to pass |
| Akron | 7f242bb | 2020-09-28 20:19:11 +0200 | [diff] [blame] | 712 | if (code != null) { |
| 713 | templateData.put("code", code.getStatusCode()); |
| 714 | templateData.put("msg", msg); |
| 715 | }; |
| 716 | |
| Akron | a578eb0 | 2020-11-26 12:34:17 +0100 | [diff] [blame] | 717 | try { |
| 718 | templateData.put("dict", this.getDictionary()); |
| 719 | |
| 720 | } catch (Exception e) { |
| Akron | a241a47 | 2021-04-27 18:55:58 +0200 | [diff] [blame] | 721 | Logger.error(e); |
| Akron | a578eb0 | 2020-11-26 12:34:17 +0100 | [diff] [blame] | 722 | return Response |
| 723 | .ok(new String("Dictionary not found")) |
| 724 | .status(Status.INTERNAL_SERVER_ERROR) |
| 725 | .build(); |
| 726 | }; |
| 727 | |
| Akron | 7f242bb | 2020-09-28 20:19:11 +0200 | [diff] [blame] | 728 | // Generate template |
| Akron | bb076c7 | 2020-09-28 14:39:15 +0200 | [diff] [blame] | 729 | try { |
| 730 | Template template = cfg.getTemplate("export.ftl"); |
| Marc Kupietz | 8852588 | 2022-03-26 22:28:36 +0100 | [diff] [blame] | 731 | template.setLocale(getPreferredSupportedLocale()); |
| Akron | bb076c7 | 2020-09-28 14:39:15 +0200 | [diff] [blame] | 732 | template.process(templateData, out); |
| 733 | } |
| Akron | 69cd35d | 2020-11-20 13:17:31 +0100 | [diff] [blame] | 734 | |
| 735 | // Unable to find template |
| Akron | bb076c7 | 2020-09-28 14:39:15 +0200 | [diff] [blame] | 736 | catch (Exception e) { |
| Akron | a241a47 | 2021-04-27 18:55:58 +0200 | [diff] [blame] | 737 | Logger.error(e); |
| Akron | bb076c7 | 2020-09-28 14:39:15 +0200 | [diff] [blame] | 738 | return Response |
| 739 | .ok(new String("Template not found")) |
| 740 | .status(Status.INTERNAL_SERVER_ERROR) |
| 741 | .build(); |
| Akron | 54b30ed | 2020-11-13 10:35:35 +0100 | [diff] [blame] | 742 | }; |
| hebasta | e2bd6e4 | 2020-03-03 19:07:20 +0100 | [diff] [blame] | 743 | |
| Akron | 7f242bb | 2020-09-28 20:19:11 +0200 | [diff] [blame] | 744 | ResponseBuilder resp = Response.ok(out.toString(), "text/html"); |
| 745 | |
| Akron | 69cd35d | 2020-11-20 13:17:31 +0100 | [diff] [blame] | 746 | if (code != null) |
| Akron | 7f242bb | 2020-09-28 20:19:11 +0200 | [diff] [blame] | 747 | resp = resp.status(code); |
| Akron | 7f242bb | 2020-09-28 20:19:11 +0200 | [diff] [blame] | 748 | |
| 749 | return resp.build(); |
| Akron | eedac91 | 2020-11-16 12:54:42 +0100 | [diff] [blame] | 750 | }; |
| Akron | be6777b | 2020-10-02 11:35:03 +0200 | [diff] [blame] | 751 | |
| 752 | |
| 753 | /* |
| Akron | 69cd35d | 2020-11-20 13:17:31 +0100 | [diff] [blame] | 754 | * Get the origin user IP. |
| 755 | * |
| Akron | be6777b | 2020-10-02 11:35:03 +0200 | [diff] [blame] | 756 | * This function is a simplification of |
| 757 | * Mojolicious::Plugin::ClientIP |
| 758 | */ |
| 759 | protected static String getClientIP (String xff) { |
| 760 | if (xff == null) { |
| 761 | return ""; |
| 762 | }; |
| 763 | |
| 764 | String[] ips = xff.split("\\s*,\\s*"); |
| 765 | |
| Akron | 69cd35d | 2020-11-20 13:17:31 +0100 | [diff] [blame] | 766 | for (int i = ips.length - 1; i >= 0; i--) { |
| 767 | if (ips[i].matches(ipre)) |
| Akron | be6777b | 2020-10-02 11:35:03 +0200 | [diff] [blame] | 768 | return ips[i]; |
| Akron | be6777b | 2020-10-02 11:35:03 +0200 | [diff] [blame] | 769 | }; |
| 770 | |
| 771 | return ""; |
| Akron | 232248d | 2020-11-11 18:55:10 +0100 | [diff] [blame] | 772 | }; |
| Akron | a578eb0 | 2020-11-26 12:34:17 +0100 | [diff] [blame] | 773 | |
| 774 | |
| 775 | /* |
| 776 | * Load dictionary for a chosen locale as a resource bundle |
| 777 | */ |
| Marc Kupietz | 8852588 | 2022-03-26 22:28:36 +0100 | [diff] [blame] | 778 | private ResourceBundle getDictionary() throws IOException { |
| 779 | return ResourceBundle.getBundle( |
| 780 | "locales/export", getPreferredSupportedLocale() |
| 781 | ); |
| 782 | } |
| Akron | a578eb0 | 2020-11-26 12:34:17 +0100 | [diff] [blame] | 783 | |
| Marc Kupietz | 8852588 | 2022-03-26 22:28:36 +0100 | [diff] [blame] | 784 | private Locale getPreferredSupportedLocale() throws IOException { |
| 785 | Locale fallback = new Locale("en"); |
| Akron | a578eb0 | 2020-11-26 12:34:17 +0100 | [diff] [blame] | 786 | |
| 787 | if (req != null) { |
| Akron | a578eb0 | 2020-11-26 12:34:17 +0100 | [diff] [blame] | 788 | for (Locale l : req.getAcceptableLanguages()) { |
| 789 | switch (l.getLanguage()) { |
| Marc Kupietz | 8852588 | 2022-03-26 22:28:36 +0100 | [diff] [blame] | 790 | case "de": |
| 791 | return (l); |
| 792 | case "en": |
| 793 | return (l); |
| 794 | } |
| 795 | } |
| 796 | } |
| Akron | a578eb0 | 2020-11-26 12:34:17 +0100 | [diff] [blame] | 797 | |
| Marc Kupietz | 8852588 | 2022-03-26 22:28:36 +0100 | [diff] [blame] | 798 | return fallback; |
| 799 | } |
| 800 | |
| 801 | } |