| hebasta | e2bd6e4 | 2020-03-03 19:07:20 +0100 | [diff] [blame] | 1 | package de.ids_mannheim.korap.plkexport; |
| 2 | |
| hebasta | fed4194 | 2020-11-02 10:00:36 +0100 | [diff] [blame] | 3 | import java.io.BufferedWriter; |
| 4 | import java.io.File; |
| 5 | import java.io.FileWriter; |
| hebasta | e049676 | 2020-03-30 00:11:01 +0200 | [diff] [blame] | 6 | import java.io.IOException; |
| Akron | bb076c7 | 2020-09-28 14:39:15 +0200 | [diff] [blame] | 7 | import java.io.StringWriter; |
| Akron | b87af2f | 2020-09-24 14:38:02 +0200 | [diff] [blame] | 8 | import java.lang.Thread; |
| 9 | import java.io.InputStream; |
| hebasta | e049676 | 2020-03-30 00:11:01 +0200 | [diff] [blame] | 10 | import java.net.URLEncoder; |
| Akron | bb076c7 | 2020-09-28 14:39:15 +0200 | [diff] [blame] | 11 | import java.util.HashMap; |
| hebasta | e049676 | 2020-03-30 00:11:01 +0200 | [diff] [blame] | 12 | import java.util.ArrayList; |
| 13 | import java.util.Collection; |
| 14 | import java.util.Iterator; |
| 15 | import java.util.LinkedList; |
| Akron | c931cd0 | 2020-09-15 10:54:17 +0200 | [diff] [blame] | 16 | import java.util.Properties; |
| Akron | 669d0bb | 2020-10-02 13:12:01 +0200 | [diff] [blame] | 17 | import java.util.regex.Matcher; |
| 18 | import java.util.regex.Pattern; |
| 19 | import java.util.Base64; |
| hebasta | ff955d6 | 2020-05-27 15:59:14 +0200 | [diff] [blame] | 20 | |
| 21 | import javax.ws.rs.BadRequestException; |
| Akron | 328f9ea | 2020-09-22 17:11:22 +0200 | [diff] [blame] | 22 | import javax.ws.rs.WebApplicationException; |
| hebasta | e049676 | 2020-03-30 00:11:01 +0200 | [diff] [blame] | 23 | import javax.ws.rs.FormParam; |
| 24 | import javax.ws.rs.POST; |
| Akron | b87af2f | 2020-09-24 14:38:02 +0200 | [diff] [blame] | 25 | import javax.ws.rs.GET; |
| hebasta | e2bd6e4 | 2020-03-03 19:07:20 +0100 | [diff] [blame] | 26 | import javax.ws.rs.Path; |
| 27 | import javax.ws.rs.Produces; |
| hebasta | e049676 | 2020-03-30 00:11:01 +0200 | [diff] [blame] | 28 | import javax.ws.rs.client.Client; |
| 29 | import javax.ws.rs.client.ClientBuilder; |
| 30 | import javax.ws.rs.client.WebTarget; |
| Akron | be6777b | 2020-10-02 11:35:03 +0200 | [diff] [blame] | 31 | import javax.ws.rs.client.Invocation; |
| 32 | import javax.ws.rs.core.Context; |
| Akron | bfd5fb8 | 2020-09-15 14:28:25 +0200 | [diff] [blame] | 33 | import javax.ws.rs.core.UriBuilder; |
| hebasta | e2bd6e4 | 2020-03-03 19:07:20 +0100 | [diff] [blame] | 34 | import javax.ws.rs.core.MediaType; |
| hebasta | e049676 | 2020-03-30 00:11:01 +0200 | [diff] [blame] | 35 | import javax.ws.rs.core.Response; |
| 36 | import javax.ws.rs.core.Response.ResponseBuilder; |
| hebasta | ff955d6 | 2020-05-27 15:59:14 +0200 | [diff] [blame] | 37 | import javax.ws.rs.core.Response.Status; |
| Akron | 669d0bb | 2020-10-02 13:12:01 +0200 | [diff] [blame] | 38 | import javax.servlet.http.Cookie; |
| Akron | 328f9ea | 2020-09-22 17:11:22 +0200 | [diff] [blame] | 39 | import java.net.ConnectException; |
| Akron | 669d0bb | 2020-10-02 13:12:01 +0200 | [diff] [blame] | 40 | import javax.servlet.http.HttpServletRequest; |
| hebasta | ff955d6 | 2020-05-27 15:59:14 +0200 | [diff] [blame] | 41 | |
| Akron | e34bd4f | 2020-09-18 10:46:01 +0200 | [diff] [blame] | 42 | import static de.ids_mannheim.korap.plkexport.Util.*; |
| 43 | |
| Akron | bb076c7 | 2020-09-28 14:39:15 +0200 | [diff] [blame] | 44 | // Template engine |
| 45 | import freemarker.template.Configuration; |
| 46 | import freemarker.template.Template; |
| 47 | |
| hebasta | fed4194 | 2020-11-02 10:00:36 +0100 | [diff] [blame] | 48 | /** |
| Akron | 232248d | 2020-11-11 18:55:10 +0100 | [diff] [blame] | 49 | * TODO: |
| Akron | 232248d | 2020-11-11 18:55:10 +0100 | [diff] [blame] | 50 | * - Delete the temp file of the export at the end |
| Akron | c1c1824 | 2020-11-18 18:24:12 +0100 | [diff] [blame] | 51 | * - Do not expect all meta data per match. |
| Akron | b95c5a4 | 2020-11-17 13:44:17 +0100 | [diff] [blame] | 52 | * - Add progress mechanism. |
| 53 | * - Add CSV export format. |
| Akron | d2072ee | 2020-11-17 16:12:41 +0100 | [diff] [blame] | 54 | * - Add table layout to RTF information. |
| Akron | ea77cb4 | 2020-11-18 14:04:21 +0100 | [diff] [blame] | 55 | * - Add loading marker. |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 56 | * - Add hitc to form. |
| hebasta | fed4194 | 2020-11-02 10:00:36 +0100 | [diff] [blame] | 57 | */ |
| hebasta | a66693a | 2020-07-19 16:51:28 +0200 | [diff] [blame] | 58 | |
| hebasta | e049676 | 2020-03-30 00:11:01 +0200 | [diff] [blame] | 59 | @Path("/") |
| hebasta | e2bd6e4 | 2020-03-03 19:07:20 +0100 | [diff] [blame] | 60 | public class IdsExportService { |
| 61 | |
| Akron | c931cd0 | 2020-09-15 10:54:17 +0200 | [diff] [blame] | 62 | Properties properties = ExWSConf.properties(null); |
| Akron | b87af2f | 2020-09-24 14:38:02 +0200 | [diff] [blame] | 63 | |
| Akron | bb076c7 | 2020-09-28 14:39:15 +0200 | [diff] [blame] | 64 | private final ClassLoader cl = Thread.currentThread().getContextClassLoader(); |
| 65 | |
| 66 | InputStream is = cl.getResourceAsStream("assets/export.js"); |
| 67 | private final String exportJsStr = streamToString(is); |
| Akron | 7f242bb | 2020-09-28 20:19:11 +0200 | [diff] [blame] | 68 | |
| 69 | Configuration cfg = new Configuration(); |
| 70 | { |
| 71 | cfg.setClassForTemplateLoading(IdsExportService.class, "/assets/templates"); |
| 72 | cfg.setDefaultEncoding("UTF-8"); |
| 73 | } |
| Akron | be6777b | 2020-10-02 11:35:03 +0200 | [diff] [blame] | 74 | |
| 75 | private final static String octets = |
| 76 | "(?:25[0-5]|2[0-4][0-9]|[0-1]?[0-9]{1,2})"; |
| 77 | |
| 78 | private final static String ipre = |
| 79 | octets + "\\." + octets + "\\." + octets + "\\." + octets; |
| Akron | 669d0bb | 2020-10-02 13:12:01 +0200 | [diff] [blame] | 80 | |
| 81 | private static Pattern authrep = Pattern.compile("\"auth\":\"([^\"]+?)\""); |
| 82 | |
| 83 | private final static Base64.Decoder b64Dec = Base64.getDecoder(); |
| 84 | |
| 85 | @Context |
| 86 | private HttpServletRequest req; |
| Akron | c931cd0 | 2020-09-15 10:54:17 +0200 | [diff] [blame] | 87 | |
| Akron | e57937b | 2020-11-17 08:49:31 +0100 | [diff] [blame] | 88 | |
| hebasta | e049676 | 2020-03-30 00:11:01 +0200 | [diff] [blame] | 89 | /** |
| 90 | * WebService calls Kustvakt Search Webservices and returns |
| Akron | eedac91 | 2020-11-16 12:54:42 +0100 | [diff] [blame] | 91 | * response as json (all of the response) and |
| 92 | * as rtf (matches) |
| hebasta | e049676 | 2020-03-30 00:11:01 +0200 | [diff] [blame] | 93 | * |
| hebasta | ff955d6 | 2020-05-27 15:59:14 +0200 | [diff] [blame] | 94 | * @param fname |
| 95 | * file name |
| 96 | * @param format |
| 97 | * the file format value rtf or json. |
| 98 | * @param q |
| 99 | * the query |
| 100 | * @param ql |
| 101 | * the query language |
| Akron | 9690796 | 2020-11-12 18:47:07 +0100 | [diff] [blame] | 102 | * @param cutoff |
| 103 | * Export more than the first page |
| hebasta | ff955d6 | 2020-05-27 15:59:14 +0200 | [diff] [blame] | 104 | * |
| 105 | * |
| hebasta | e049676 | 2020-03-30 00:11:01 +0200 | [diff] [blame] | 106 | */ |
| 107 | @POST |
| 108 | @Path("export") |
| 109 | @Produces(MediaType.APPLICATION_OCTET_STREAM) |
| Akron | 9690796 | 2020-11-12 18:47:07 +0100 | [diff] [blame] | 110 | public Response export ( |
| Akron | be6777b | 2020-10-02 11:35:03 +0200 | [diff] [blame] | 111 | @FormParam("fname") String fname, |
| 112 | @FormParam("format") String format, |
| 113 | @FormParam("q") String q, |
| Akron | d2072ee | 2020-11-17 16:12:41 +0100 | [diff] [blame] | 114 | @FormParam("cq") String cq, |
| Akron | be6777b | 2020-10-02 11:35:03 +0200 | [diff] [blame] | 115 | @FormParam("ql") String ql, |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 116 | @FormParam("cutoff") String cutoffStr, |
| Akron | b95c5a4 | 2020-11-17 13:44:17 +0100 | [diff] [blame] | 117 | // @FormParam("islimit") String il, |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 118 | @FormParam("hitc") int hitc |
| Akron | be6777b | 2020-10-02 11:35:03 +0200 | [diff] [blame] | 119 | ) throws IOException { |
| Akron | 232248d | 2020-11-11 18:55:10 +0100 | [diff] [blame] | 120 | |
| 121 | // These parameters are required |
| Akron | 90799f1 | 2020-09-17 17:10:34 +0200 | [diff] [blame] | 122 | String[][] params = { |
| Akron | 90799f1 | 2020-09-17 17:10:34 +0200 | [diff] [blame] | 123 | { "format", format }, |
| 124 | { "q", q }, |
| 125 | { "ql", ql } |
| 126 | }; |
| hebasta | e049676 | 2020-03-30 00:11:01 +0200 | [diff] [blame] | 127 | |
| Akron | 232248d | 2020-11-11 18:55:10 +0100 | [diff] [blame] | 128 | // Check that all parameters are available |
| hebasta | ff955d6 | 2020-05-27 15:59:14 +0200 | [diff] [blame] | 129 | for (int i = 0; i < params.length; i++) { |
| 130 | if (params[i][1] == null || params[i][1].trim().isEmpty()) |
| Akron | 90799f1 | 2020-09-17 17:10:34 +0200 | [diff] [blame] | 131 | throw new BadRequestException( |
| 132 | Response |
| 133 | .status(Status.BAD_REQUEST) |
| 134 | .entity("Parameter " + "\"" |
| 135 | + params[i][0] + "\"" + " is missing or empty") |
| 136 | .build()); |
| Akron | 232248d | 2020-11-11 18:55:10 +0100 | [diff] [blame] | 137 | }; |
| Akron | 232248d | 2020-11-11 18:55:10 +0100 | [diff] [blame] | 138 | |
| 139 | // Retrieve cutoff value |
| hebasta | fed4194 | 2020-11-02 10:00:36 +0100 | [diff] [blame] | 140 | boolean cutoff = false; |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 141 | if (cutoffStr != null && ( |
| 142 | cutoffStr.equals("true") || |
| 143 | cutoffStr.equals("1")) |
| 144 | ) { |
| Akron | 232248d | 2020-11-11 18:55:10 +0100 | [diff] [blame] | 145 | cutoff = true; |
| 146 | }; |
| hebasta | fed4194 | 2020-11-02 10:00:36 +0100 | [diff] [blame] | 147 | |
| 148 | ResponseBuilder builder = null; |
| hebasta | e049676 | 2020-03-30 00:11:01 +0200 | [diff] [blame] | 149 | Client client = ClientBuilder.newClient(); |
| 150 | |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 151 | // Load configuration values |
| 152 | String scheme = properties.getProperty("api.scheme", "https"); |
| 153 | String port = properties.getProperty("api.port", "8089"); |
| 154 | String host = properties.getProperty("api.host", "localhost"); |
| 155 | String path = properties.getProperty("api.path", ""); |
| 156 | int pageSize = Integer.parseInt(properties.getProperty("conf.page_size", "5")); |
| 157 | int maxResults = Integer.parseInt(properties.getProperty("conf.max_exp_limit", "10000")); |
| hebasta | a66693a | 2020-07-19 16:51:28 +0200 | [diff] [blame] | 158 | |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 159 | // Adjust the number of requested hits |
| 160 | if (hitc > 0 && hitc < maxResults) { |
| 161 | maxResults = hitc; |
| 162 | }; |
| 163 | |
| Akron | e57937b | 2020-11-17 08:49:31 +0100 | [diff] [blame] | 164 | // Create initial search uri |
| Akron | bfd5fb8 | 2020-09-15 14:28:25 +0200 | [diff] [blame] | 165 | UriBuilder uri = UriBuilder.fromPath("/api/v1.0/search") |
| 166 | .host(host) |
| 167 | .port(Integer.parseInt(port)) |
| Akron | 480a908 | 2020-09-16 13:14:09 +0200 | [diff] [blame] | 168 | .scheme(scheme) |
| Akron | bfd5fb8 | 2020-09-15 14:28:25 +0200 | [diff] [blame] | 169 | .queryParam("q", q) |
| Akron | 1b20161 | 2020-10-02 17:40:32 +0200 | [diff] [blame] | 170 | // .queryParam("context", "sentence") |
| Akron | 9690796 | 2020-11-12 18:47:07 +0100 | [diff] [blame] | 171 | .queryParam("context", "40-t,40-t") // Not yet supported |
| Akron | bfd5fb8 | 2020-09-15 14:28:25 +0200 | [diff] [blame] | 172 | .queryParam("ql", ql) |
| Akron | bfd5fb8 | 2020-09-15 14:28:25 +0200 | [diff] [blame] | 173 | ; |
| Akron | 68f221b | 2020-09-29 23:15:43 +0200 | [diff] [blame] | 174 | |
| Akron | d2072ee | 2020-11-17 16:12:41 +0100 | [diff] [blame] | 175 | if (cq != null) |
| 176 | uri = uri.queryParam("cq", cq); |
| Akron | d2072ee | 2020-11-17 16:12:41 +0100 | [diff] [blame] | 177 | |
| Akron | 68f221b | 2020-09-29 23:15:43 +0200 | [diff] [blame] | 178 | if (path != "") { |
| 179 | uri = uri.path(path); |
| 180 | }; |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 181 | |
| Akron | 5abeab5 | 2020-11-12 15:08:48 +0100 | [diff] [blame] | 182 | uri = uri.queryParam("count", pageSize); |
| Akron | c931cd0 | 2020-09-15 10:54:17 +0200 | [diff] [blame] | 183 | |
| Akron | be6777b | 2020-10-02 11:35:03 +0200 | [diff] [blame] | 184 | // Get client IP, in case service is behind a proxy |
| 185 | String xff = ""; |
| Akron | 669d0bb | 2020-10-02 13:12:01 +0200 | [diff] [blame] | 186 | // Get auth (temporarily) via Session riding |
| 187 | String auth = ""; |
| Akron | be6777b | 2020-10-02 11:35:03 +0200 | [diff] [blame] | 188 | if (req != null) { |
| 189 | xff = getClientIP(req.getHeader("X-Forwarded-For")); |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 190 | if (xff == "") |
| Akron | be6777b | 2020-10-02 11:35:03 +0200 | [diff] [blame] | 191 | xff = req.getRemoteAddr(); |
| Akron | 669d0bb | 2020-10-02 13:12:01 +0200 | [diff] [blame] | 192 | |
| Akron | 54b30ed | 2020-11-13 10:35:35 +0100 | [diff] [blame] | 193 | auth = authFromCookie(req); |
| Akron | be6777b | 2020-10-02 11:35:03 +0200 | [diff] [blame] | 194 | }; |
| Akron | 669d0bb | 2020-10-02 13:12:01 +0200 | [diff] [blame] | 195 | |
| Akron | 328f9ea | 2020-09-22 17:11:22 +0200 | [diff] [blame] | 196 | String resp; |
| hebasta | fed4194 | 2020-11-02 10:00:36 +0100 | [diff] [blame] | 197 | WebTarget resource; |
| Akron | 696c063 | 2020-11-12 15:30:51 +0100 | [diff] [blame] | 198 | Invocation.Builder reqBuilder; |
| Akron | d2072ee | 2020-11-17 16:12:41 +0100 | [diff] [blame] | 199 | |
| Akron | 328f9ea | 2020-09-22 17:11:22 +0200 | [diff] [blame] | 200 | try { |
| hebasta | fed4194 | 2020-11-02 10:00:36 +0100 | [diff] [blame] | 201 | resource = client.target(uri.build()); |
| Akron | 696c063 | 2020-11-12 15:30:51 +0100 | [diff] [blame] | 202 | reqBuilder = resource.request(MediaType.APPLICATION_JSON); |
| 203 | resp = authBuilder(reqBuilder, xff, auth).get(String.class); |
| Akron | eedac91 | 2020-11-16 12:54:42 +0100 | [diff] [blame] | 204 | |
| Akron | 328f9ea | 2020-09-22 17:11:22 +0200 | [diff] [blame] | 205 | } catch (Exception e) { |
| 206 | throw new WebApplicationException( |
| Akron | 7f242bb | 2020-09-28 20:19:11 +0200 | [diff] [blame] | 207 | responseForm(Status.BAD_GATEWAY, "Unable to reach Backend") |
| Akron | 328f9ea | 2020-09-22 17:11:22 +0200 | [diff] [blame] | 208 | ); |
| 209 | } |
| hebasta | e049676 | 2020-03-30 00:11:01 +0200 | [diff] [blame] | 210 | |
| Akron | eedac91 | 2020-11-16 12:54:42 +0100 | [diff] [blame] | 211 | Exporter exp; |
| 212 | |
| Akron | e57937b | 2020-11-17 08:49:31 +0100 | [diff] [blame] | 213 | // Choose the correct exporter |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 214 | if (format.equals("json")) |
| Akron | eedac91 | 2020-11-16 12:54:42 +0100 | [diff] [blame] | 215 | exp = new JsonExporter(); |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 216 | else |
| Akron | eedac91 | 2020-11-16 12:54:42 +0100 | [diff] [blame] | 217 | exp = new RtfExporter(); |
| Akron | e57937b | 2020-11-17 08:49:31 +0100 | [diff] [blame] | 218 | |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 219 | exp.setMaxResults(maxResults); |
| Akron | 7412271 | 2020-11-17 09:41:21 +0100 | [diff] [blame] | 220 | exp.setQueryString(q); |
| Akron | d2072ee | 2020-11-17 16:12:41 +0100 | [diff] [blame] | 221 | exp.setCorpusQueryString(cq); |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 222 | |
| Akron | 876017d | 2020-11-17 09:19:24 +0100 | [diff] [blame] | 223 | // set filename based on query (if not already set) |
| 224 | if (fname != null) { |
| Akron | 7412271 | 2020-11-17 09:41:21 +0100 | [diff] [blame] | 225 | exp.setFileName(fname); |
| Akron | 876017d | 2020-11-17 09:19:24 +0100 | [diff] [blame] | 226 | }; |
| 227 | |
| Akron | e57937b | 2020-11-17 08:49:31 +0100 | [diff] [blame] | 228 | // Initialize exporter (with meta data and first matches) |
| Akron | acc9f7a | 2020-11-17 17:21:40 +0100 | [diff] [blame] | 229 | try { |
| 230 | exp.init(resp); |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 231 | |
| Akron | acc9f7a | 2020-11-17 17:21:40 +0100 | [diff] [blame] | 232 | } catch (Exception e) { |
| 233 | |
| 234 | throw new WebApplicationException( |
| 235 | responseForm( |
| 236 | Status.INTERNAL_SERVER_ERROR, |
| 237 | e.getMessage() |
| 238 | ) |
| 239 | ); |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 240 | }; |
| Akron | d2072ee | 2020-11-17 16:12:41 +0100 | [diff] [blame] | 241 | |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 242 | |
| 243 | /* |
| 244 | * Calculate how many results to fetch |
| 245 | */ |
| 246 | int fetchCount = exp.getTotalResults(); |
| 247 | if (exp.hasTimeExceeded() || fetchCount > maxResults) { |
| 248 | fetchCount = maxResults; |
| 249 | }; |
| 250 | |
| 251 | // The first page was already enough |
| 252 | if (fetchCount <= pageSize) { |
| 253 | cutoff = true; |
| 254 | }; |
| 255 | |
| Akron | d3df5b0 | 2020-11-12 18:23:17 +0100 | [diff] [blame] | 256 | // If only one page should be exported there is no need |
| 257 | // for a temporary export file |
| 258 | if (cutoff) { |
| Akron | eedac91 | 2020-11-16 12:54:42 +0100 | [diff] [blame] | 259 | builder = exp.serve(); |
| hebasta | e049676 | 2020-03-30 00:11:01 +0200 | [diff] [blame] | 260 | } |
| 261 | |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 262 | // Page through all results |
| hebasta | e049676 | 2020-03-30 00:11:01 +0200 | [diff] [blame] | 263 | else { |
| hebasta | a66693a | 2020-07-19 16:51:28 +0200 | [diff] [blame] | 264 | |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 265 | // It's not important anymore to get totalResults |
| 266 | uri.queryParam("cutoff", "true"); |
| Akron | d3df5b0 | 2020-11-12 18:23:17 +0100 | [diff] [blame] | 267 | |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 268 | // Set offset for paging as a template |
| Akron | d3df5b0 | 2020-11-12 18:23:17 +0100 | [diff] [blame] | 269 | uri.queryParam("offset", "{offset}"); |
| Akron | 9690796 | 2020-11-12 18:47:07 +0100 | [diff] [blame] | 270 | |
| Akron | acc9f7a | 2020-11-17 17:21:40 +0100 | [diff] [blame] | 271 | try { |
| 272 | |
| 273 | // Iterate over all results |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 274 | for (int i = pageSize; i <= fetchCount; i+=pageSize) { |
| 275 | resource = client.target(uri.build(i)); |
| Akron | acc9f7a | 2020-11-17 17:21:40 +0100 | [diff] [blame] | 276 | reqBuilder = resource.request(MediaType.APPLICATION_JSON); |
| 277 | resp = authBuilder(reqBuilder, xff, auth).get(String.class); |
| Akron | 62d90a3 | 2020-11-18 20:45:38 +0100 | [diff] [blame] | 278 | |
| 279 | // Stop when no more matches are allowed |
| 280 | if (!exp.appendMatches(resp)) |
| 281 | break; |
| Akron | acc9f7a | 2020-11-17 17:21:40 +0100 | [diff] [blame] | 282 | } |
| 283 | } catch (Exception e) { |
| 284 | throw new WebApplicationException( |
| 285 | responseForm( |
| 286 | Status.INTERNAL_SERVER_ERROR, |
| 287 | e.getMessage() |
| 288 | ) |
| Akron | d3df5b0 | 2020-11-12 18:23:17 +0100 | [diff] [blame] | 289 | ); |
| Akron | acc9f7a | 2020-11-17 17:21:40 +0100 | [diff] [blame] | 290 | }; |
| hebasta | fed4194 | 2020-11-02 10:00:36 +0100 | [diff] [blame] | 291 | |
| Akron | e57937b | 2020-11-17 08:49:31 +0100 | [diff] [blame] | 292 | builder = exp.serve(); |
| hebasta | fed4194 | 2020-11-02 10:00:36 +0100 | [diff] [blame] | 293 | }; |
| Akron | 2c0184d | 2020-09-17 18:22:03 +0200 | [diff] [blame] | 294 | |
| Akron | d3df5b0 | 2020-11-12 18:23:17 +0100 | [diff] [blame] | 295 | return builder.build(); |
| Akron | 9690796 | 2020-11-12 18:47:07 +0100 | [diff] [blame] | 296 | }; |
| 297 | |
| hebasta | e2bd6e4 | 2020-03-03 19:07:20 +0100 | [diff] [blame] | 298 | |
| Akron | b87af2f | 2020-09-24 14:38:02 +0200 | [diff] [blame] | 299 | @GET |
| 300 | @Path("export") |
| 301 | @Produces(MediaType.TEXT_HTML) |
| 302 | public Response exportHTML () { |
| Akron | 7f242bb | 2020-09-28 20:19:11 +0200 | [diff] [blame] | 303 | return responseForm(); |
| 304 | }; |
| Akron | bb076c7 | 2020-09-28 14:39:15 +0200 | [diff] [blame] | 305 | |
| Akron | 7f242bb | 2020-09-28 20:19:11 +0200 | [diff] [blame] | 306 | |
| Akron | 9690796 | 2020-11-12 18:47:07 +0100 | [diff] [blame] | 307 | @GET |
| 308 | @Path("export.js") |
| 309 | @Produces("application/javascript") |
| 310 | public Response exportJavascript () { |
| 311 | return Response |
| 312 | .ok(exportJsStr, "application/javascript") |
| 313 | .build(); |
| 314 | }; |
| 315 | |
| Akron | 7f242bb | 2020-09-28 20:19:11 +0200 | [diff] [blame] | 316 | |
| Akron | 696c063 | 2020-11-12 15:30:51 +0100 | [diff] [blame] | 317 | // Decorate request with auth headers |
| Akron | 54b30ed | 2020-11-13 10:35:35 +0100 | [diff] [blame] | 318 | private Invocation.Builder authBuilder (Invocation.Builder reqBuilder, |
| 319 | String xff, |
| 320 | String auth) { |
| Akron | 696c063 | 2020-11-12 15:30:51 +0100 | [diff] [blame] | 321 | if (xff != "") { |
| 322 | reqBuilder = reqBuilder.header("X-Forwarded-For", xff); |
| 323 | }; |
| 324 | if (auth != "") { |
| 325 | reqBuilder = reqBuilder.header("Authorization", auth); |
| 326 | }; |
| 327 | |
| 328 | return reqBuilder; |
| 329 | }; |
| 330 | |
| Akron | 9690796 | 2020-11-12 18:47:07 +0100 | [diff] [blame] | 331 | |
| Akron | 54b30ed | 2020-11-13 10:35:35 +0100 | [diff] [blame] | 332 | // Get authorization token from cookie |
| 333 | private String authFromCookie (HttpServletRequest r) { |
| 334 | |
| 335 | // This is a temporary solution using session riding - only |
| 336 | // valid for the time being |
| 337 | Cookie[] cookies = r.getCookies(); |
| 338 | String cookiePath = properties.getProperty("cookie.path", ""); |
| 339 | |
| 340 | // Iterate through all cookies for a Kalamar session |
| 341 | for (int i = 0; i < cookies.length; i++) { |
| 342 | |
| 343 | // Check the valid path |
| 344 | if (cookiePath != "" && cookies[i].getPath() != cookiePath) { |
| 345 | continue; |
| 346 | }; |
| 347 | |
| 348 | // Ignore irrelevant cookies |
| 349 | if (!cookies[i].getName().matches("^kalamar(-.+?)?$")) { |
| 350 | continue; |
| 351 | }; |
| 352 | |
| 353 | // Get the value |
| 354 | String b64 = cookies[i].getValue(); |
| 355 | String[] b64Parts = b64.split("--", 2); |
| 356 | if (b64Parts.length == 2) { |
| 357 | // Read the payload |
| 358 | String payload = new String(b64Dec.decode(b64Parts[0])); |
| 359 | if (payload != "") { |
| 360 | Matcher m = authrep.matcher(payload); |
| 361 | if (m.find()) { |
| 362 | return m.group(1); |
| 363 | }; |
| 364 | }; |
| 365 | }; |
| 366 | }; |
| 367 | |
| 368 | return ""; |
| 369 | }; |
| 370 | |
| 371 | |
| Akron | 9690796 | 2020-11-12 18:47:07 +0100 | [diff] [blame] | 372 | /* |
| 373 | * Response with form template. |
| 374 | */ |
| 375 | private Response responseForm () { |
| 376 | return responseForm(null, null); |
| Akron | 54b30ed | 2020-11-13 10:35:35 +0100 | [diff] [blame] | 377 | }; |
| Akron | 9690796 | 2020-11-12 18:47:07 +0100 | [diff] [blame] | 378 | |
| 379 | |
| 380 | /* |
| 381 | * Response with form template. |
| 382 | * |
| 383 | * Accepts an error code and message. |
| 384 | */ |
| Akron | 7f242bb | 2020-09-28 20:19:11 +0200 | [diff] [blame] | 385 | private Response responseForm (Status code, String msg) { |
| Akron | bb076c7 | 2020-09-28 14:39:15 +0200 | [diff] [blame] | 386 | StringWriter out = new StringWriter(); |
| 387 | HashMap<String, Object> templateData = new HashMap<String, Object>(); |
| 388 | |
| 389 | String scheme = properties.getProperty("asset.scheme", "https"); |
| 390 | String port = properties.getProperty("asset.port", ""); |
| 391 | String host = properties.getProperty("asset.host", "korap.ids-mannheim.de"); |
| Akron | 68f221b | 2020-09-29 23:15:43 +0200 | [diff] [blame] | 392 | String path = properties.getProperty("asset.path", ""); |
| Akron | bb076c7 | 2020-09-28 14:39:15 +0200 | [diff] [blame] | 393 | |
| 394 | UriBuilder uri = UriBuilder.fromPath("") |
| 395 | .host(host) |
| Akron | 68f221b | 2020-09-29 23:15:43 +0200 | [diff] [blame] | 396 | .scheme(scheme); |
| 397 | |
| 398 | if (path != "") { |
| 399 | uri = uri.path(path); |
| 400 | }; |
| Akron | bb076c7 | 2020-09-28 14:39:15 +0200 | [diff] [blame] | 401 | |
| 402 | if (port != "") { |
| 403 | uri = uri.port(Integer.parseInt(port)); |
| Akron | 68f221b | 2020-09-29 23:15:43 +0200 | [diff] [blame] | 404 | }; |
| Akron | bb076c7 | 2020-09-28 14:39:15 +0200 | [diff] [blame] | 405 | |
| 406 | templateData.put("assetPath", uri.build()); |
| 407 | |
| Akron | 7f242bb | 2020-09-28 20:19:11 +0200 | [diff] [blame] | 408 | if (code != null) { |
| 409 | templateData.put("code", code.getStatusCode()); |
| 410 | templateData.put("msg", msg); |
| 411 | }; |
| 412 | |
| 413 | // Generate template |
| Akron | bb076c7 | 2020-09-28 14:39:15 +0200 | [diff] [blame] | 414 | try { |
| 415 | Template template = cfg.getTemplate("export.ftl"); |
| 416 | template.process(templateData, out); |
| 417 | } |
| 418 | catch (Exception e) { |
| 419 | return Response |
| 420 | .ok(new String("Template not found")) |
| 421 | .status(Status.INTERNAL_SERVER_ERROR) |
| 422 | .build(); |
| Akron | 54b30ed | 2020-11-13 10:35:35 +0100 | [diff] [blame] | 423 | }; |
| hebasta | e2bd6e4 | 2020-03-03 19:07:20 +0100 | [diff] [blame] | 424 | |
| Akron | 7f242bb | 2020-09-28 20:19:11 +0200 | [diff] [blame] | 425 | ResponseBuilder resp = Response.ok(out.toString(), "text/html"); |
| 426 | |
| 427 | if (code != null) { |
| 428 | resp = resp.status(code); |
| 429 | }; |
| 430 | |
| 431 | return resp.build(); |
| Akron | eedac91 | 2020-11-16 12:54:42 +0100 | [diff] [blame] | 432 | }; |
| Akron | be6777b | 2020-10-02 11:35:03 +0200 | [diff] [blame] | 433 | |
| 434 | |
| 435 | /* |
| 436 | * This function is a simplification of |
| 437 | * Mojolicious::Plugin::ClientIP |
| 438 | */ |
| 439 | protected static String getClientIP (String xff) { |
| 440 | if (xff == null) { |
| 441 | return ""; |
| 442 | }; |
| 443 | |
| 444 | String[] ips = xff.split("\\s*,\\s*"); |
| 445 | |
| 446 | for (int i = ips.length - 1; i >= 0; i--){ |
| 447 | if (ips[i].matches(ipre)) { |
| 448 | return ips[i]; |
| 449 | }; |
| 450 | }; |
| 451 | |
| 452 | return ""; |
| Akron | 232248d | 2020-11-11 18:55:10 +0100 | [diff] [blame] | 453 | }; |
| Akron | 232248d | 2020-11-11 18:55:10 +0100 | [diff] [blame] | 454 | }; |