1 package org.unicode.cldr.util; 2 3 import java.util.ArrayList; 4 import java.util.Arrays; 5 import java.util.EnumSet; 6 import java.util.HashMap; 7 import java.util.List; 8 import java.util.Map; 9 import java.util.Map.Entry; 10 import java.util.Set; 11 import java.util.TreeSet; 12 import java.util.regex.Matcher; 13 import java.util.regex.Pattern; 14 15 import org.unicode.cldr.util.PatternPlaceholders.PlaceholderInfo; 16 17 import com.ibm.icu.text.MessageFormat; 18 import com.ibm.icu.util.Output; 19 20 public class PathDescription { 21 22 public enum ErrorHandling { 23 SKIP, CONTINUE 24 } 25 26 // BE sure to sync with the list in xmbSkip! 27 public static final Set<String> EXTRA_LANGUAGES = new TreeSet<>( 28 Arrays 29 .asList( 30 "ach|af|ak|ak|am|ar|az|be|bem|bg|bh|bn|br|bs|ca|chr|ckb|co|crs|cs|cy|da|de|de_AT|de_CH|ee|el|en|en_AU|en_CA|en_GB|en_US|eo|es|es_419|es_ES|et|eu|fa|fi|fil|fo|fr|fr_CA|fr_CH|fy|ga|gaa|gd|gl|gn|gsw|gu|ha|haw|he|hi|hr|ht|hu|hy|ia|id|ig|io|is|it|ja|jv|ka|kg|kk|km|kn|ko|kri|ku|ky|la|lg|ln|lo|loz|lt|lua|lv|mfe|mg|mi|mk|ml|mn|mr|ms|mt|my|nb|ne|nl|nl_BE|nn|nso|ny|nyn|oc|om|or|pa|pcm|pl|ps|pt|pt_BR|pt_PT|qu|rm|rn|ro|ro|ro_MD|ru|rw|sd|si|sk|sl|sn|so|sq|sr|sr_Latn|sr_ME|st|su|sv|sw|ta|te|tg|th|ti|tk|tlh|tn|to|tr|tt|tum|ug|uk|und|ur|uz|vi|wo|xh|yi|yo|zh|zh_Hans|zh_Hant|zh_HK|zu|zxx" 31 .split("|"))); 32 33 private static final Pattern METAZONE_PATTERN = Pattern 34 .compile("//ldml/dates/timeZoneNames/metazone\\[@type=\"([^\"]*)\"]/(.*)/(.*)"); 35 private static final Pattern STAR_ATTRIBUTE_PATTERN = PatternCache.get("=\"([^\"]*)\""); 36 37 private static final StandardCodes STANDARD_CODES = StandardCodes.make(); 38 private static Map<String, String> ZONE2COUNTRY = STANDARD_CODES.getZoneToCounty(); 39 private static RegexLookup<String> pathHandling = new RegexLookup<String>().loadFromFile(PathDescription.class, 40 "data/PathDescription.txt"); 41 42 // set in construction 43 44 private final CLDRFile english; 45 private final Map<String, String> extras; 46 private final ErrorHandling errorHandling; 47 private final Map<String, List<Set<String>>> starredPaths; 48 private final Set<String> allMetazones; 49 50 // used on instance 51 52 private Matcher metazoneMatcher = METAZONE_PATTERN.matcher(""); 53 private String starredPathOutput; 54 private Output<String[]> pathArguments = new Output<>(); 55 private EnumSet<Status> status = EnumSet.noneOf(Status.class); 56 57 public static final String MISSING_DESCRIPTION = "Before translating, please see http://cldr.org/translation."; 58 PathDescription(SupplementalDataInfo supplementalDataInfo, CLDRFile english, Map<String, String> extras, Map<String, List<Set<String>>> starredPaths, ErrorHandling errorHandling)59 public PathDescription(SupplementalDataInfo supplementalDataInfo, 60 CLDRFile english, 61 Map<String, String> extras, 62 Map<String, List<Set<String>>> starredPaths, 63 ErrorHandling errorHandling) { 64 this.english = english; 65 this.extras = extras == null ? new HashMap<>() : extras; 66 this.starredPaths = starredPaths == null ? new HashMap<>() : starredPaths; 67 allMetazones = supplementalDataInfo.getAllMetazones(); 68 this.errorHandling = errorHandling; 69 } 70 getStarredPathOutput()71 public String getStarredPathOutput() { 72 return starredPathOutput; 73 } 74 getStatus()75 public EnumSet<Status> getStatus() { 76 return status; 77 } 78 79 public enum Status { 80 SKIP, NULL_VALUE, EMPTY_CONTENT, NOT_REQUIRED 81 } 82 getRawDescription(String path, String value, Object context)83 public String getRawDescription(String path, String value, Object context) { 84 status.clear(); 85 return pathHandling.get(path, context, pathArguments); 86 } 87 getDescription(String path, String value, Level level, Object context)88 public String getDescription(String path, String value, Level level, Object context) { 89 status.clear(); 90 91 String description = pathHandling.get(path, context, pathArguments); 92 if (description == null) { 93 description = MISSING_DESCRIPTION; 94 } else if ("SKIP".equals(description)) { 95 status.add(Status.SKIP); 96 if (errorHandling == ErrorHandling.SKIP) { 97 return null; 98 } 99 } 100 101 // String localeWhereFound = english.getSourceLocaleID(path, status); 102 // if (!status.pathWhereFound.equals(path)) { 103 // reasonsToPaths.put("alias", path + " " + value); 104 // continue; 105 // } 106 if (value == null) { // a count item? 107 String xpath = extras.get(path); 108 if (xpath != null) { 109 value = english.getStringValue(xpath); 110 } else if (path.contains("/metazone")) { 111 if (metazoneMatcher.reset(path).matches()) { 112 String name = metazoneMatcher.group(1); 113 String type = metazoneMatcher.group(3); 114 value = name.replace('_', ' ') 115 + (type.equals("generic") ? "" : type.equals("daylight") ? " Summer" : " Winter") + " Time"; 116 // System.out.println("Missing: " + path + " : " + value); 117 } 118 } 119 if (value == null) { 120 status.add(Status.NULL_VALUE); 121 if (errorHandling == ErrorHandling.SKIP) { 122 return null; 123 } 124 } 125 } 126 if (value != null && value.length() == 0) { 127 status.add(Status.EMPTY_CONTENT); 128 if (errorHandling == ErrorHandling.SKIP) { 129 return null; 130 } 131 } 132 // if (GenerateXMB.contentMatcher != null && !GenerateXMB.contentMatcher.reset(value).find()) { 133 // PathDescription.addSkipReasons(reasonsToPaths, "content-parameter", level, path, value); 134 // return null; 135 // } 136 137 List<String> attributes = addStarredInfo(starredPaths, path); 138 139 // In special cases, only use if there is a root value (languageNames, ... 140 if (description.startsWith("ROOT")) { 141 int typeEnd = description.indexOf(';'); 142 String type = description.substring(4, typeEnd).trim(); 143 description = description.substring(typeEnd + 1).trim(); 144 145 boolean isMetazone = type.equals("metazone"); 146 String code = attributes.get(0); 147 boolean isRootCode = isRootCode(code, allMetazones, type, isMetazone); 148 if (!isRootCode) { 149 status.add(Status.NOT_REQUIRED); 150 if (errorHandling == ErrorHandling.SKIP) { 151 return null; 152 } 153 } 154 if (isMetazone) { 155 XPathParts parts = XPathParts.getFrozenInstance(path); 156 String daylightType = parts.getElement(-1); 157 daylightType = daylightType.equals("daylight") ? "summer" : daylightType.equals("standard") ? "winter" 158 : daylightType; 159 String length = parts.getElement(-2); 160 length = length.equals("long") ? "" : "abbreviated "; 161 code = code + ", " + length + daylightType + " form"; 162 } else if (type.equals("timezone")) { 163 String country = ZONE2COUNTRY.get(code); 164 int lastSlash = code.lastIndexOf('/'); 165 String codeName = lastSlash < 0 ? code : code.substring(lastSlash + 1).replace('_', ' '); 166 167 boolean found = false; 168 if ("001".equals(country)) { 169 code = "the timezone “" + codeName + "”"; 170 found = true; 171 } else if (country != null) { 172 String countryName = english.getName("territory", country); 173 if (countryName != null) { 174 if (!codeName.equals(countryName)) { 175 code = "the city “" + codeName + "” (in " + countryName + ")"; 176 } else { 177 code = "the country “" + codeName + "”"; 178 } 179 found = true; 180 } 181 } 182 if (!found) { 183 System.out.println("Missing country for timezone " + code); 184 } 185 } 186 description = MessageFormat.format(MessageFormat.autoQuoteApostrophe(description), new Object[] { code }); 187 } else if (path.contains("exemplarCity")) { 188 String regionCode = ZONE2COUNTRY.get(attributes.get(0)); 189 String englishRegionName = english.getName(CLDRFile.TERRITORY_NAME, regionCode); 190 description = MessageFormat.format(MessageFormat.autoQuoteApostrophe(description), 191 new Object[] { englishRegionName }); 192 } else if (description != MISSING_DESCRIPTION) { 193 description = MessageFormat.format(MessageFormat.autoQuoteApostrophe(description), 194 (Object[]) pathArguments.value); 195 } 196 197 return description; 198 } 199 200 /** 201 * Creates an escaped HTML string of placeholder information. 202 * 203 * @param path 204 * the xpath to specify placeholder information for 205 * @return a HTML string, or an empty string if there was no placeholder information 206 */ 207 public String getPlaceholderDescription(String path) { 208 Map<String, PlaceholderInfo> placeholders = PatternPlaceholders.getInstance().get(path); 209 if (placeholders != null && placeholders.size() > 0) { 210 StringBuffer buffer = new StringBuffer(); 211 buffer.append("<table>"); 212 buffer.append("<tr><th>Placeholder</th><th>Meaning</th><th>Example</th></tr>"); 213 for (Entry<String, PlaceholderInfo> entry : placeholders.entrySet()) { 214 PlaceholderInfo info = entry.getValue(); 215 buffer.append("<tr>"); 216 buffer.append("<td>").append(entry.getKey()).append("</td>"); 217 buffer.append("<td>").append(info.name).append("</td>"); 218 buffer.append("<td>").append(info.example).append("</td>"); 219 buffer.append("</tr>"); 220 } 221 buffer.append("</table>"); 222 return buffer.toString(); 223 } 224 return ""; 225 } 226 isRootCode(String code, Set<String> allMetazones, String type, boolean isMetazone)227 private static boolean isRootCode(String code, Set<String> allMetazones, String type, boolean isMetazone) { 228 Set<String> codes = isMetazone ? allMetazones 229 : type.equals("timezone") ? STANDARD_CODES.getCanonicalTimeZones() 230 : STANDARD_CODES.getSurveyToolDisplayCodes(type); 231 // end 232 boolean isRootCode = codes.contains(code) || code.contains("_"); 233 if (!isRootCode && type.equals("language") 234 && EXTRA_LANGUAGES.contains(code)) { 235 isRootCode = true; 236 } 237 return isRootCode; 238 } 239 addStarredInfo(Map<String, List<Set<String>>> starredPaths, String path)240 private List<String> addStarredInfo(Map<String, List<Set<String>>> starredPaths, String path) { 241 Matcher starAttributeMatcher = STAR_ATTRIBUTE_PATTERN.matcher(path); 242 StringBuilder starredPath = new StringBuilder(); 243 List<String> attributes = new ArrayList<>(); 244 int lastEnd = 0; 245 while (starAttributeMatcher.find()) { 246 int start = starAttributeMatcher.start(1); 247 int end = starAttributeMatcher.end(1); 248 starredPath.append(path.substring(lastEnd, start)); 249 starredPath.append(".*"); 250 251 attributes.add(path.substring(start, end)); 252 lastEnd = end; 253 } 254 starredPath.append(path.substring(lastEnd)); 255 String starredPathString = starredPath.toString().intern(); 256 starredPathOutput = starredPathString; 257 258 List<Set<String>> attributeList = starredPaths.get(starredPathString); 259 if (attributeList == null) { 260 starredPaths.put(starredPathString, attributeList = new ArrayList<>()); 261 } 262 int i = 0; 263 for (String attribute : attributes) { 264 if (attributeList.size() <= i) { 265 TreeSet<String> subset = new TreeSet<>(); 266 subset.add(attribute); 267 attributeList.add(subset); 268 } else { 269 Set<String> subset = attributeList.get(i); 270 subset.add(attribute); 271 } 272 ++i; 273 } 274 return attributes; 275 } 276 } 277