|
| 1 | +import java.net.URL; |
| 2 | +import java.util.Scanner; |
| 3 | +import java.util.ArrayList; |
| 4 | +import java.util.TreeMap; |
| 5 | +import java.io.IOException; |
| 6 | +import java.io.PrintWriter; |
| 7 | +import java.io.File; |
| 8 | + |
| 9 | +public class DownloadDefinitions { |
| 10 | + public static void main(String[] args) throws IOException { |
| 11 | + try (PrintWriter dest = new PrintWriter(new File("edu/virginia/ged5to7/GedcomDefinitions.java"))) { |
| 12 | + dest.println("/* WARNING: This file is automatically generated and should not be edited by hand */"); |
| 13 | + dest.println("package edu.virginia.ged5to7;"); |
| 14 | + dest.println("import static java.util.Arrays.binarySearch;"); |
| 15 | + dest.println("\n/** A container for the substructure, payload, and enumeration rules from gedcom.io */"); |
| 16 | + dest.println("public class GedcomDefinitions {"); |
| 17 | + |
| 18 | + |
| 19 | + Scanner s; |
| 20 | + char before; |
| 21 | + ArrayList<String> lines; |
| 22 | + |
| 23 | + // enumerations |
| 24 | + s = new Scanner(new URL("https://github.com/FamilySearch/GEDCOM/raw/main/extracted-files/enumerations.tsv").openStream()); |
| 25 | + lines = new ArrayList<String>(); |
| 26 | + while(s.hasNext()) lines.add(s.nextLine()); |
| 27 | + lines.sort(null); |
| 28 | + dest.println(" private static final String[] enumKeys ="); |
| 29 | + before = '{'; |
| 30 | + for(String line : lines) { |
| 31 | + dest.println(" "+before+'"'+line.replaceAll("\t[^\t]*$","\"")); |
| 32 | + before = ','; |
| 33 | + } |
| 34 | + dest.println(" };"); |
| 35 | + dest.println(" private static final String[] enumVals ="); |
| 36 | + before = '{'; |
| 37 | + for(String line : lines) { |
| 38 | + dest.println(" "+before+line.replaceAll(".*\t","\"")+"\""); |
| 39 | + before = ','; |
| 40 | + } |
| 41 | + dest.println(" };"); |
| 42 | + dest.println(" /** Looks up the URI of an enumeration based on the GEDCOM 7 spec"); |
| 43 | + dest.println(" * @param ctx the URI of the containing structure."); |
| 44 | + dest.println(" * use <code>null</code> for an extension."); |
| 45 | + dest.println(" * @param tag the enumeration value"); |
| 46 | + dest.println(" * @return the URI of the enumeration value, or <code>null</code> if unknown"); |
| 47 | + dest.println(" */"); |
| 48 | + dest.println(" public static String enumURI(String ctx, String tag) {"); |
| 49 | + dest.println(" if (ctx == null) {"); |
| 50 | + dest.println(" String val = \"https://gedcom.io/terms/v7/\"+tag;"); |
| 51 | + dest.println(" int idx = binarySearch(enumVals, val);"); |
| 52 | + dest.println(" if (idx < 0) return null;"); |
| 53 | + dest.println(" return enumVals[idx];"); |
| 54 | + dest.println(" } else {"); |
| 55 | + dest.println(" String key = ctx+'\\t'+tag;"); |
| 56 | + dest.println(" int idx = binarySearch(enumKeys, key);"); |
| 57 | + dest.println(" if (idx < 0) return null;"); |
| 58 | + dest.println(" return enumVals[idx];"); |
| 59 | + dest.println(" }"); |
| 60 | + dest.println(" }"); |
| 61 | + |
| 62 | + dest.println(); |
| 63 | + |
| 64 | + // substructures |
| 65 | + s = new Scanner(new URL("https://github.com/FamilySearch/GEDCOM/raw/main/extracted-files/substructures.tsv").openStream()); |
| 66 | + lines = new ArrayList<String>(); |
| 67 | + while(s.hasNext()) lines.add(s.nextLine()); |
| 68 | + lines.sort(null); |
| 69 | + dest.println(" private static final String[] structKeys ="); |
| 70 | + before = '{'; |
| 71 | + for(String line : lines) { |
| 72 | + dest.println(" "+before+'"'+line.replaceAll("\t[^\t]*$","\"")); |
| 73 | + before = ','; |
| 74 | + } |
| 75 | + dest.println(" };"); |
| 76 | + dest.println(" private static final String[] structVals ="); |
| 77 | + before = '{'; |
| 78 | + for(String line : lines) { |
| 79 | + dest.println(" "+before+line.replaceAll(".*\t","\"")+"\""); |
| 80 | + before = ','; |
| 81 | + } |
| 82 | + dest.println(" };"); |
| 83 | + dest.println(" /** Looks up the URI of an structure type based on the GEDCOM 7 spec"); |
| 84 | + dest.println(" * @param ctx the URI of the containing structure type"); |
| 85 | + dest.println(" * use <code>\"\"</code> for a record and <code>null</code> for an extension."); |
| 86 | + dest.println(" * @param tag the tag of the structure"); |
| 87 | + dest.println(" * @return the URI of the structure type, or <code>null</code> if unknown"); |
| 88 | + dest.println(" */"); |
| 89 | + dest.println(" public static String structURI(String ctx, String tag) {"); |
| 90 | + dest.println(" if (ctx == null) {"); |
| 91 | + dest.println(" String val = \"https://gedcom.io/terms/v7/\"+tag;"); |
| 92 | + dest.println(" int idx = binarySearch(structVals, val);"); |
| 93 | + dest.println(" if (idx < 0) return null;"); |
| 94 | + dest.println(" return structVals[idx];"); |
| 95 | + dest.println(" } else {"); |
| 96 | + dest.println(" String key = ctx+'\\t'+tag;"); |
| 97 | + dest.println(" int idx = binarySearch(structKeys, key);"); |
| 98 | + dest.println(" if (idx < 0) return null;"); |
| 99 | + dest.println(" return structVals[idx];"); |
| 100 | + dest.println(" }"); |
| 101 | + dest.println(" }"); |
| 102 | + |
| 103 | + // structure types -- uses same file and substructures above |
| 104 | + TreeMap<String,String> knownStructs = new TreeMap<String, String>(); |
| 105 | + for(String line : lines) { |
| 106 | + knownStructs.put(line.replaceAll(".*\t",""), line.replaceAll("^[^\t]*\t|\t[^\t]*$","")); |
| 107 | + } |
| 108 | + dest.println(" private static final String[] knownStructs ="); |
| 109 | + before = '{'; |
| 110 | + for(String uri : knownStructs.keySet()) { |
| 111 | + dest.println(" "+before+'"'+uri+'"'); |
| 112 | + before = ','; |
| 113 | + } |
| 114 | + dest.println(" };"); |
| 115 | + dest.println(" private static final String[] uriTag ="); |
| 116 | + before = '{'; |
| 117 | + for(String tag : knownStructs.values()) { |
| 118 | + dest.println(" "+before+'"'+tag+'"'); |
| 119 | + before = ','; |
| 120 | + } |
| 121 | + dest.println(" };"); |
| 122 | + dest.println(" /** Looks up the tag of a structure URIbased on the GEDCOM 7 spec"); |
| 123 | + dest.println(" * @param uri the URI of the structure type"); |
| 124 | + dest.println(" * @return the tag of the structure type, or <code>null</code> if unknown"); |
| 125 | + dest.println(" */"); |
| 126 | + dest.println(" public static String structTag(String uri) {"); |
| 127 | + dest.println(" if (uri == null) return null;"); |
| 128 | + dest.println(" int idx = binarySearch(knownStructs, uri);"); |
| 129 | + dest.println(" if (idx < 0) return null;"); |
| 130 | + dest.println(" return uriTag[idx];"); |
| 131 | + dest.println(" }"); |
| 132 | + |
| 133 | + |
| 134 | + |
| 135 | + dest.println(); |
| 136 | + |
| 137 | + // payloads |
| 138 | + s = new Scanner(new URL("https://github.com/FamilySearch/GEDCOM/raw/main/extracted-files/payloads.tsv").openStream()); |
| 139 | + lines = new ArrayList<String>(); |
| 140 | + while(s.hasNext()) lines.add(s.nextLine()); |
| 141 | + lines.sort(null); |
| 142 | + dest.println(" private static final String[] payloadKeys ="); |
| 143 | + before = '{'; |
| 144 | + for(String line : lines) { |
| 145 | + dest.println(" "+before+'"'+line.replaceAll("\t[^\t]*$","\"")); |
| 146 | + before = ','; |
| 147 | + } |
| 148 | + dest.println(" };"); |
| 149 | + dest.println(" private static final String[] payloadVals ="); |
| 150 | + before = '{'; |
| 151 | + for(String line : lines) { |
| 152 | + dest.println(" "+before+line.replaceAll(".*\t","\"")+"\""); |
| 153 | + before = ','; |
| 154 | + } |
| 155 | + dest.println(" };"); |
| 156 | + dest.println(" /** Looks up the payload type of a structure based on the GEDCOM 7 spec"); |
| 157 | + dest.println(" * @param ctx the URI of the containing structure type"); |
| 158 | + dest.println(" * @return the type code (URI or <code>\"Y|<NULL>\"</code> or <code>\"\"</code> or <code>\"@XREF:</code>tag<code>\"</code>) of the payload type, or <code>null</code> if unknown"); |
| 159 | + dest.println(" */"); |
| 160 | + dest.println(" public static String payloadURI(String ctx) {"); |
| 161 | + dest.println(" if (ctx == null) return null;"); |
| 162 | + dest.println(" int idx = binarySearch(payloadKeys, ctx);"); |
| 163 | + dest.println(" if (idx < 0) return null;"); |
| 164 | + dest.println(" return payloadVals[idx];"); |
| 165 | + dest.println(" }"); |
| 166 | + |
| 167 | + |
| 168 | + dest.println("}"); |
| 169 | + } |
| 170 | + } |
| 171 | +} |
0 commit comments