samples/LuceneInAction/lia/advsearching/SortingExample.py

   1 # ====================================================================
   2 #   Licensed under the Apache License, Version 2.0 (the "License");
   3 #   you may not use this file except in compliance with the License.
   4 #   You may obtain a copy of the License at
   5 #
   6 #       http://www.apache.org/licenses/LICENSE-2.0
   7 #
   8 #   Unless required by applicable law or agreed to in writing, software
   9 #   distributed under the License is distributed on an "AS IS" BASIS,
  10 #   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  11 #   See the License for the specific language governing permissions and
  12 #   limitations under the License.
  13 # ====================================================================
  14
  15 import os
  16
  17 from lucene import \
  18     SimpleFSDirectory, Document, Field, IndexSearcher, StandardAnalyzer, \
  19     MatchAllDocsQuery, Sort, SortField, DecimalFormat, System, File, \
  20     TopFieldCollector, QueryParser, Version, BooleanQuery, BooleanClause
  21
  22
  23 class SortingExample(object):
  24
  25     def __init__(self, directory):
  26
  27         self.directory = directory
  28
  29     def displayResults(self, query, sort):
  30
  31         searcher = IndexSearcher(self.directory, True)
  32
  33         fillFields = False
  34         computeMaxScore = False
  35         docsScoredInOrder = False
  36         computeScores = True
  37
  38         collector = TopFieldCollector.create(sort, 20,
  39                                              fillFields,
  40                                              computeScores,
  41                                              computeMaxScore,
  42                                              docsScoredInOrder)
  43
  44         searcher.search(query, None, collector)
  45         scoreDocs = collector.topDocs().scoreDocs
  46
  47         print "\nResults for:", query, "sorted by", sort
  48         print "Title".rjust(30), "pubmonth".rjust(10), \
  49               "id".center(4), "score".center(15)
  50
  51         scoreFormatter = DecimalFormat("0.######")
  52         for scoreDoc in scoreDocs:
  53             doc = searcher.doc(scoreDoc.doc)
  54             title = doc["title"]
  55             if len(title) > 30:
  56                 title = title[:30]
  57             print title.encode('ascii', 'replace').rjust(30), \
  58                   doc["pubmonth"].rjust(10), \
  59                   str(scoreDoc.doc).center(4), \
  60                   scoreFormatter.format(scoreDoc.score).ljust(12)
  61             print "  ", doc["category"]
  62             # print searcher.explain(query, scoreDoc.doc)
  63
  64         searcher.close()
  65
  66     def main(cls, argv):
  67
  68         allBooks = MatchAllDocsQuery()
  69         parser = QueryParser(Version.LUCENE_CURRENT, "contents",
  70                              StandardAnalyzer(Version.LUCENE_CURRENT))
  71         query = BooleanQuery()
  72         query.add(allBooks, BooleanClause.Occur.SHOULD)
  73         query.add(parser.parse("java OR action"), BooleanClause.Occur.SHOULD)
  74
  75         indexDir = System.getProperty("index.dir")
  76         directory = SimpleFSDirectory(File(indexDir))
  77
  78         example = SortingExample(directory)
  79
  80         example.displayResults(query, Sort.RELEVANCE)
  81         example.displayResults(query, Sort.INDEXORDER)
  82         example.displayResults(query,
  83                                Sort(SortField("category", SortField.STRING)))
  84         example.displayResults(query,
  85                                Sort(SortField("pubmonth", SortField.INT, True)))
  86
  87         example.displayResults(query,
  88                                Sort([SortField("category", SortField.STRING),
  89                                      SortField.FIELD_SCORE,
  90                                      SortField("pubmonth", SortField.INT, True)]))
  91
  92         example.displayResults(query,
  93                                Sort([SortField.FIELD_SCORE,
  94                                      SortField("category", SortField.STRING)]))
  95         directory.close()
  96
  97     main = classmethod(main)