commit eccc8f90971b85cb8ca52f1ccc558a0e4ea41512 · treybastian.com/1brc

+23 -8
src/main/java/dev/morling/onebrc/CreateMeasurements3.java
···

       101
       101
        
                       // Use a 7th-order curve to simulate the name length distribution.

     

       102
       102
        
                       // It gives us mostly short names, but with large outliers.

     

       103
       103
        
                       var nameLen = (int) (yOffset + factor * Math.pow(rnd.nextDouble() - xOffset, power));

     

       104
       104
       -
                       minLen = Integer.min(minLen, nameLen);

     

       105
       105
       -
                       maxLen = Integer.max(maxLen, nameLen);

     

       106
       104
        
                       var count = nameSource.read(buf, 0, nameLen);

     

       107
       105
        
                       if (count == -1) {

     

       108
       106
        
                           throw new Exception("Name source exhausted");

     

       109
       107
        
                       }

     

       110
       110
       -
                       var name = new String(buf, 0, nameLen).trim();

     

       111
       111
       -
                       while (name.length() < nameLen) {

     

       112
       112
       -
                           name += readNonSpace(nameSource);

     

       108
       108
       +
                       var nameBuf = new StringBuilder(nameLen);

     

       109
       109
       +
                       nameBuf.append(buf, 0, nameLen);

     

       110
       110
       +
                       if (Character.isWhitespace(nameBuf.charAt(0))) {

     

       111
       111
       +
                           nameBuf.setCharAt(0, readNonSpace(nameSource));

     

       112
       112
       +
                       }

     

       113
       113
       +
                       if (Character.isWhitespace(nameBuf.charAt(nameBuf.length() - 1))) {

     

       114
       114
       +
                           nameBuf.setCharAt(nameBuf.length() - 1, readNonSpace(nameSource));

     

       113
       115
        
                       }

     

       116
       116
       +
                       var name = nameBuf.toString();

     

       114
       117
        
                       while (names.contains(name)) {

     

       115
       115
       -
                           name = name.substring(1) + readNonSpace(nameSource);

     

       118
       118
       +
                           nameBuf.setCharAt(rnd.nextInt(nameBuf.length()), readNonSpace(nameSource));

     

       119
       119
       +
                           name = nameBuf.toString();

     

       116
       120
        
                       }

     

       117
       117
       -
                       while (name.getBytes(StandardCharsets.UTF_8).length > 100) {

     

       118
       118
       -
                           name = name.substring(0, name.length() - 1);

     

       121
       121
       +
                       int actualLen;

     

       122
       122
       +
                       while (true) {

     

       123
       123
       +
                           actualLen = name.getBytes(StandardCharsets.UTF_8).length;

     

       124
       124
       +
                           if (actualLen <= 100) {

     

       125
       125
       +
                               break;

     

       126
       126
       +
                           }

     

       127
       127
       +
                           nameBuf.deleteCharAt(nameBuf.length() - 1);

     

       128
       128
       +
                           if (Character.isWhitespace(nameBuf.charAt(nameBuf.length() - 1))) {

     

       129
       129
       +
                               nameBuf.setCharAt(nameBuf.length() - 1, readNonSpace(nameSource));

     

       130
       130
       +
                           }

     

       131
       131
       +
                           name = nameBuf.toString();

     

       119
       132
        
                       }

     

       120
       133
        
                       if (name.indexOf(';') != -1) {

     

       121
       134
        
                           throw new Exception("Station name contains a semicolon!");

     

       122
       135
        
                       }

     

       123
       136
        
                       names.add(name);

     

       137
       137
       +
                       minLen = Integer.min(minLen, actualLen);

     

       138
       138
       +
                       maxLen = Integer.max(maxLen, actualLen);

     

       124
       139
        
                       var lat = Float.parseFloat(row.substring(row.indexOf(';') + 1));

     

       125
       140
        
                       // Guesstimate mean temperature using cosine of latitude

     

       126
       141
        
                       var avgTemp = (float) (30 * Math.cos(Math.toRadians(lat))) - 10;