Revision 48226
Added by Eri Katsari almost 7 years ago
SourceBuildNoCacheMapper.java | ||
---|---|---|
76 | 76 |
} |
77 | 77 |
|
78 | 78 |
String property = fields[1]; |
79 |
String value = fields[2]; |
|
80 |
List<String> blockingKeys = Blocking.tokenBlocking(value, stopWordsMap); |
|
81 |
for (String blockingKey : blockingKeys) { |
|
82 |
//Write BlockingKey, RecordID to output |
|
83 |
context.write(new Text(blockingKey), new Text(id + FIELD_DELIM + result.toString())); |
|
84 |
context.getCounter(SOURCE_BUILD_COUNTERS.BLOCKING_KEYS).increment(1); |
|
79 |
if (!property.contains("name") && !property.contains("label")) { |
|
80 |
String value = fields[2]; |
|
81 |
List<String> blockingKeys = Blocking.tokenBlocking(value, stopWordsMap); |
|
82 |
for (String blockingKey : blockingKeys) { |
|
83 |
//Write BlockingKey, RecordID to output |
|
84 |
context.write(new Text(blockingKey), new Text(id + FIELD_DELIM + result.toString())); |
|
85 |
context.getCounter(SOURCE_BUILD_COUNTERS.BLOCKING_KEYS).increment(1); |
|
86 |
} |
|
85 | 87 |
} |
86 | 88 |
} |
87 | 89 |
} |
Also available in: Unified diff
Changed token blocking to accept year as a token