Was.OrcSearch – Blame information for rev 3
?pathlinks?
Rev | Author | Line No. | Line |
---|---|---|---|
1 | office | 1 | using System; |
2 | using System.Collections.Generic; |
||
3 | using System.Linq; |
||
4 | using Lucene.Net.Analysis.Standard; |
||
5 | using Lucene.Net.Documents; |
||
6 | using Lucene.Net.Index; |
||
7 | using Lucene.Net.QueryParsers; |
||
8 | using Lucene.Net.Search; |
||
9 | using Lucene.Net.Store; |
||
10 | using Was.OrcSearch.EventArgs; |
||
11 | using Was.OrcSearch.Extensions; |
||
2 | office | 12 | using Was.OrcSearch.Helpers; |
1 | office | 13 | using Was.OrcSearch.Metadata.Interfaces; |
14 | using Was.OrcSearch.Services.Interfaces; |
||
15 | |||
16 | namespace Was.OrcSearch.Services |
||
17 | { |
||
18 | public abstract class SearchServiceBase : ISearchService |
||
19 | { |
||
20 | #region Constructors |
||
21 | |||
22 | protected SearchServiceBase(ISearchQueryService searchQueryService) |
||
23 | { |
||
24 | _searchQueryService = searchQueryService; |
||
25 | } |
||
26 | |||
27 | #endregion |
||
28 | |||
29 | #region Properties |
||
30 | |||
31 | public int IndexedObjectCount |
||
32 | { |
||
33 | get |
||
34 | { |
||
35 | lock (_lockObject) |
||
36 | { |
||
37 | return _indexedObjects.Count; |
||
38 | } |
||
39 | } |
||
40 | } |
||
41 | |||
42 | #endregion |
||
43 | |||
44 | #region Constants |
||
45 | |||
46 | private const string IndexId = "__index_id"; |
||
47 | private static int _currentIndex; |
||
48 | |||
49 | #endregion |
||
50 | |||
51 | #region Fields |
||
52 | |||
53 | private readonly object _lockObject = new object(); |
||
54 | |||
55 | private readonly ISearchQueryService _searchQueryService; |
||
56 | |||
57 | private readonly Dictionary<int, ISearchable> _indexedObjects = new Dictionary<int, ISearchable>(); |
||
58 | private readonly Dictionary<ISearchable, int> _searchableIndexes = new Dictionary<ISearchable, int>(); |
||
59 | |||
60 | private readonly Dictionary<string, ISearchableMetadata> _searchableMetadata = |
||
61 | new Dictionary<string, ISearchableMetadata>(); |
||
62 | |||
63 | private bool _initialized; |
||
64 | |||
65 | private Directory _indexDirectory; |
||
66 | |||
67 | #endregion |
||
68 | |||
69 | #region Events |
||
70 | |||
71 | public event EventHandler<System.EventArgs> Updating; |
||
72 | |||
73 | public event EventHandler<System.EventArgs> Updated; |
||
74 | |||
75 | public event EventHandler<SearchEventArgs> Searching; |
||
76 | |||
77 | public event EventHandler<SearchEventArgs> Searched; |
||
78 | |||
79 | #endregion |
||
80 | |||
81 | #region Methods |
||
82 | |||
83 | public virtual IEnumerable<ISearchableMetadata> GetSearchableMetadata() |
||
84 | { |
||
85 | lock (_lockObject) |
||
86 | { |
||
87 | var searchableMetadata = new List<ISearchableMetadata>(_searchableMetadata.Values); |
||
88 | return searchableMetadata; |
||
89 | } |
||
90 | } |
||
91 | |||
92 | public virtual void AddObjects(IEnumerable<ISearchable> searchables) |
||
93 | { |
||
94 | Initialize(); |
||
95 | |||
96 | Updating?.Invoke(this, null); |
||
97 | |||
98 | lock (_lockObject) |
||
99 | { |
||
100 | using (var analyzer = new StandardAnalyzer(LuceneDefaults.Version)) |
||
101 | { |
||
102 | using (var writer = |
||
103 | new IndexWriter(_indexDirectory, analyzer, IndexWriter.MaxFieldLength.UNLIMITED)) |
||
104 | { |
||
105 | foreach (var searchable in searchables) |
||
106 | { |
||
107 | var index = _currentIndex++; |
||
108 | _indexedObjects.Add(index, searchable); |
||
109 | _searchableIndexes.Add(searchable, index); |
||
110 | |||
111 | var document = new Document(); |
||
112 | document.Add( |
||
113 | new Field(IndexId, index.ToString(), Field.Store.YES, Field.Index.NOT_ANALYZED)); |
||
114 | |||
115 | var metadata = searchable.MetadataCollection; |
||
116 | var searchableMetadatas = metadata.All.OfType<ISearchableMetadata>(); |
||
117 | |||
118 | foreach (var searchableMetadata in searchableMetadatas) |
||
119 | { |
||
120 | var searchableMetadataValue = searchableMetadata.GetValue(searchable.Instance); |
||
2 | office | 121 | // Original: ObjectToStringHelper.ToString(searchableMetadataValue); |
122 | // TODO Support more serializable types. |
||
1 | office | 123 | var searchableMetadataValueAsString = |
2 | office | 124 | string.Join(" ", searchableMetadataValue.Stringify()); |
3 | office | 125 | |
126 | // DEBUG |
||
127 | //Console.WriteLine("String metadata: " + string.Join(" ", searchableMetadataValue.Stringify())); |
||
1 | office | 128 | |
129 | var field = new Field(searchableMetadata.SearchName, searchableMetadataValueAsString, |
||
130 | Field.Store.YES, |
||
131 | searchableMetadata.Analyze ? Field.Index.ANALYZED : Field.Index.NOT_ANALYZED, |
||
3 | office | 132 | Field.TermVector.YES); |
1 | office | 133 | |
134 | document.Add(field); |
||
135 | |||
136 | if (!_searchableMetadata.ContainsKey(searchableMetadata.SearchName)) |
||
137 | _searchableMetadata.Add(searchableMetadata.SearchName, searchableMetadata); |
||
138 | } |
||
139 | |||
140 | writer.AddDocument(document); |
||
141 | } |
||
142 | |||
143 | writer.Optimize(); |
||
144 | writer.Commit(); |
||
145 | } |
||
146 | } |
||
147 | } |
||
148 | |||
149 | Updated?.Invoke(this, null); |
||
150 | } |
||
151 | |||
152 | public virtual void RemoveObjects(IEnumerable<ISearchable> searchables) |
||
153 | { |
||
154 | Initialize(); |
||
155 | |||
156 | lock (_lockObject) |
||
157 | { |
||
158 | Updating?.Invoke(this, null); |
||
159 | |||
160 | using (var analyzer = new StandardAnalyzer(LuceneDefaults.Version)) |
||
161 | { |
||
162 | using (var writer = |
||
163 | new IndexWriter(_indexDirectory, analyzer, IndexWriter.MaxFieldLength.UNLIMITED)) |
||
164 | { |
||
165 | foreach (var searchable in searchables) |
||
166 | { |
||
167 | int index; |
||
168 | if (!_searchableIndexes.TryGetValue(searchable, out index)) continue; |
||
169 | |||
170 | var queryAsText = $"{IndexId}:{index}"; |
||
171 | var parser = new QueryParser(LuceneDefaults.Version, string.Empty, analyzer); |
||
172 | var query = parser.Parse(queryAsText); |
||
173 | |||
174 | writer.DeleteDocuments(query); |
||
175 | |||
176 | _searchableIndexes.Remove(searchable); |
||
177 | _indexedObjects.Remove(index); |
||
178 | } |
||
179 | |||
180 | writer.Optimize(); |
||
181 | writer.Commit(); |
||
182 | } |
||
183 | } |
||
184 | |||
185 | Updated?.Invoke(this, null); |
||
186 | } |
||
187 | } |
||
188 | |||
189 | public void ClearAllObjects() |
||
190 | { |
||
191 | Initialize(); |
||
192 | |||
193 | Updating?.Invoke(this, null); |
||
194 | |||
195 | lock (_lockObject) |
||
196 | { |
||
197 | using (var analyzer = new StandardAnalyzer(LuceneDefaults.Version)) |
||
198 | { |
||
199 | using (var writer = |
||
200 | new IndexWriter(_indexDirectory, analyzer, IndexWriter.MaxFieldLength.UNLIMITED)) |
||
201 | { |
||
202 | _indexedObjects.Clear(); |
||
203 | _searchableMetadata.Clear(); |
||
204 | |||
205 | writer.DeleteAll(); |
||
206 | |||
207 | writer.Optimize(); |
||
208 | writer.Commit(); |
||
209 | } |
||
210 | } |
||
211 | } |
||
212 | } |
||
213 | |||
214 | public virtual IEnumerable<ISearchable> Search(string filter, int maxResults = SearchDefaults.DefaultResults) |
||
215 | { |
||
216 | Initialize(); |
||
217 | |||
218 | var results = new List<ISearchable>(); |
||
219 | |||
220 | lock (_lockObject) |
||
221 | { |
||
222 | try |
||
223 | { |
||
224 | Searching?.Invoke(this, new SearchEventArgs(filter, results)); |
||
225 | |||
226 | Query finalQuery = null; |
||
227 | |||
228 | // Note: There are two issues with using regex here |
||
229 | // 1. Lucene uses lower case interpretation of each string for indexing. |
||
230 | // That means in regular expression we can use only lower case characters |
||
231 | // 2. escape sequences do not work. Not sure why |
||
232 | // |
||
233 | // In order to fix (1), we have to force Lucene to index differently. Probably we need to have two |
||
234 | // versions if indeces. One for regular search and another for regex |
||
235 | //var regexString = filter.ExtractRegexString(); |
||
236 | //if (!string.IsNullOrWhiteSpace(regexString)) |
||
237 | //{ |
||
238 | // var searchableMetadatas = GetSearchableMetadata(); |
||
239 | |||
240 | // var booleanQuery = new BooleanQuery(); |
||
241 | // foreach (var searchableMetadata in searchableMetadatas) |
||
242 | // { |
||
243 | // var query = new RegexQuery(new Term(searchableMetadata.SearchName, regexString)); |
||
244 | // var booleanClause = new BooleanClause(query, Occur.SHOULD); |
||
245 | |||
246 | // booleanQuery.Add(booleanClause); |
||
247 | // } |
||
248 | |||
249 | // if (booleanQuery.Any()) |
||
250 | // { |
||
251 | // finalQuery = booleanQuery; |
||
252 | // } |
||
253 | //} |
||
254 | |||
255 | // ReSharper disable once ConditionIsAlwaysTrueOrFalse |
||
256 | if (finalQuery == null && filter.IsValidOrcSearchFilter()) |
||
257 | using (var analyzer = new StandardAnalyzer(LuceneDefaults.Version)) |
||
258 | { |
||
259 | var queryAsText = _searchQueryService.GetSearchQuery(filter, GetSearchableMetadata()); |
||
260 | |||
261 | var parser = new QueryParser(LuceneDefaults.Version, string.Empty, analyzer); |
||
262 | finalQuery = parser.Parse(queryAsText); |
||
263 | } |
||
264 | |||
265 | if (finalQuery != null) |
||
266 | using (var searcher = new IndexSearcher(_indexDirectory)) |
||
267 | { |
||
268 | var search = searcher.Search(finalQuery, maxResults); |
||
269 | foreach (var scoreDoc in search.ScoreDocs) |
||
270 | { |
||
271 | var score = scoreDoc.Score; |
||
272 | var docId = scoreDoc.Doc; |
||
273 | var doc = searcher.Doc(docId); |
||
274 | |||
275 | var index = int.Parse(doc.Get(IndexId)); |
||
276 | results.Add(_indexedObjects[index]); |
||
277 | } |
||
278 | } |
||
279 | } |
||
280 | catch (ParseException ex) |
||
281 | { |
||
282 | //Log.Warning(ex, "Failed to parse search pattern"); |
||
283 | throw ex; |
||
284 | } |
||
285 | catch (Exception ex) |
||
286 | { |
||
287 | //Log.Error(ex, "An error occurred while searching, returning default results"); |
||
288 | throw ex; |
||
289 | } |
||
290 | finally |
||
291 | { |
||
292 | Searched?.Invoke(this, new SearchEventArgs(filter, results)); |
||
293 | } |
||
294 | } |
||
295 | |||
296 | return results; |
||
297 | } |
||
298 | |||
299 | private void Initialize() |
||
300 | { |
||
301 | if (_initialized) return; |
||
302 | |||
303 | _initialized = true; |
||
304 | |||
305 | _indexDirectory = GetDirectory(); |
||
306 | |||
307 | // Required to create empty index, which is required for our reader |
||
308 | using (var analyzer = new StandardAnalyzer(LuceneDefaults.Version)) |
||
309 | { |
||
310 | using (var indexWriter = |
||
311 | new IndexWriter(_indexDirectory, analyzer, IndexWriter.MaxFieldLength.UNLIMITED)) |
||
312 | { |
||
313 | indexWriter.Commit(); |
||
314 | } |
||
315 | } |
||
316 | } |
||
317 | |||
318 | protected abstract Directory GetDirectory(); |
||
319 | |||
320 | #endregion |
||
321 | } |
||
322 | } |