Was.OrcSearch – Blame information for rev 4
?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; |
||
12 | using Was.OrcSearch.Metadata.Interfaces; |
||
4 | office | 13 | using Was.OrcSearch.Services.Extensions; |
1 | office | 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); |
||
4 | office | 121 | |
122 | // DEBUG |
||
123 | //Console.WriteLine("Stringifying: " + searchableMetadataValue); |
||
124 | |||
1 | office | 125 | var searchableMetadataValueAsString = |
2 | office | 126 | string.Join(" ", searchableMetadataValue.Stringify()); |
3 | office | 127 | |
128 | // DEBUG |
||
129 | //Console.WriteLine("String metadata: " + string.Join(" ", searchableMetadataValue.Stringify())); |
||
1 | office | 130 | |
131 | var field = new Field(searchableMetadata.SearchName, searchableMetadataValueAsString, |
||
132 | Field.Store.YES, |
||
133 | searchableMetadata.Analyze ? Field.Index.ANALYZED : Field.Index.NOT_ANALYZED, |
||
4 | office | 134 | Field.TermVector.NO); |
1 | office | 135 | |
136 | document.Add(field); |
||
137 | |||
138 | if (!_searchableMetadata.ContainsKey(searchableMetadata.SearchName)) |
||
139 | _searchableMetadata.Add(searchableMetadata.SearchName, searchableMetadata); |
||
140 | } |
||
141 | |||
142 | writer.AddDocument(document); |
||
143 | } |
||
144 | |||
145 | writer.Optimize(); |
||
146 | writer.Commit(); |
||
147 | } |
||
148 | } |
||
149 | } |
||
150 | |||
151 | Updated?.Invoke(this, null); |
||
152 | } |
||
153 | |||
154 | public virtual void RemoveObjects(IEnumerable<ISearchable> searchables) |
||
155 | { |
||
156 | Initialize(); |
||
157 | |||
158 | lock (_lockObject) |
||
159 | { |
||
160 | Updating?.Invoke(this, null); |
||
161 | |||
162 | using (var analyzer = new StandardAnalyzer(LuceneDefaults.Version)) |
||
163 | { |
||
164 | using (var writer = |
||
165 | new IndexWriter(_indexDirectory, analyzer, IndexWriter.MaxFieldLength.UNLIMITED)) |
||
166 | { |
||
167 | foreach (var searchable in searchables) |
||
168 | { |
||
4 | office | 169 | if (!_searchableIndexes.TryGetValue(searchable, out var index)) continue; |
1 | office | 170 | |
171 | var queryAsText = $"{IndexId}:{index}"; |
||
172 | var parser = new QueryParser(LuceneDefaults.Version, string.Empty, analyzer); |
||
173 | var query = parser.Parse(queryAsText); |
||
174 | |||
175 | writer.DeleteDocuments(query); |
||
176 | |||
177 | _searchableIndexes.Remove(searchable); |
||
178 | _indexedObjects.Remove(index); |
||
179 | } |
||
180 | |||
181 | writer.Optimize(); |
||
182 | writer.Commit(); |
||
183 | } |
||
184 | } |
||
185 | |||
186 | Updated?.Invoke(this, null); |
||
187 | } |
||
188 | } |
||
189 | |||
190 | public void ClearAllObjects() |
||
191 | { |
||
192 | Initialize(); |
||
193 | |||
194 | Updating?.Invoke(this, null); |
||
195 | |||
196 | lock (_lockObject) |
||
197 | { |
||
198 | using (var analyzer = new StandardAnalyzer(LuceneDefaults.Version)) |
||
199 | { |
||
200 | using (var writer = |
||
201 | new IndexWriter(_indexDirectory, analyzer, IndexWriter.MaxFieldLength.UNLIMITED)) |
||
202 | { |
||
203 | _indexedObjects.Clear(); |
||
204 | _searchableMetadata.Clear(); |
||
205 | |||
206 | writer.DeleteAll(); |
||
207 | |||
208 | writer.Optimize(); |
||
209 | writer.Commit(); |
||
210 | } |
||
211 | } |
||
212 | } |
||
213 | } |
||
214 | |||
215 | public virtual IEnumerable<ISearchable> Search(string filter, int maxResults = SearchDefaults.DefaultResults) |
||
216 | { |
||
217 | Initialize(); |
||
218 | |||
219 | var results = new List<ISearchable>(); |
||
220 | |||
221 | lock (_lockObject) |
||
222 | { |
||
223 | try |
||
224 | { |
||
225 | Searching?.Invoke(this, new SearchEventArgs(filter, results)); |
||
226 | |||
227 | Query finalQuery = null; |
||
228 | |||
229 | // Note: There are two issues with using regex here |
||
230 | // 1. Lucene uses lower case interpretation of each string for indexing. |
||
231 | // That means in regular expression we can use only lower case characters |
||
232 | // 2. escape sequences do not work. Not sure why |
||
233 | // |
||
234 | // In order to fix (1), we have to force Lucene to index differently. Probably we need to have two |
||
235 | // versions if indeces. One for regular search and another for regex |
||
236 | //var regexString = filter.ExtractRegexString(); |
||
237 | //if (!string.IsNullOrWhiteSpace(regexString)) |
||
238 | //{ |
||
239 | // var searchableMetadatas = GetSearchableMetadata(); |
||
240 | |||
241 | // var booleanQuery = new BooleanQuery(); |
||
242 | // foreach (var searchableMetadata in searchableMetadatas) |
||
243 | // { |
||
244 | // var query = new RegexQuery(new Term(searchableMetadata.SearchName, regexString)); |
||
245 | // var booleanClause = new BooleanClause(query, Occur.SHOULD); |
||
246 | |||
247 | // booleanQuery.Add(booleanClause); |
||
248 | // } |
||
249 | |||
250 | // if (booleanQuery.Any()) |
||
251 | // { |
||
252 | // finalQuery = booleanQuery; |
||
253 | // } |
||
254 | //} |
||
255 | |||
256 | // ReSharper disable once ConditionIsAlwaysTrueOrFalse |
||
257 | if (finalQuery == null && filter.IsValidOrcSearchFilter()) |
||
258 | using (var analyzer = new StandardAnalyzer(LuceneDefaults.Version)) |
||
259 | { |
||
260 | var queryAsText = _searchQueryService.GetSearchQuery(filter, GetSearchableMetadata()); |
||
261 | |||
262 | var parser = new QueryParser(LuceneDefaults.Version, string.Empty, analyzer); |
||
263 | finalQuery = parser.Parse(queryAsText); |
||
264 | } |
||
265 | |||
266 | if (finalQuery != null) |
||
267 | using (var searcher = new IndexSearcher(_indexDirectory)) |
||
268 | { |
||
269 | var search = searcher.Search(finalQuery, maxResults); |
||
270 | foreach (var scoreDoc in search.ScoreDocs) |
||
271 | { |
||
272 | var score = scoreDoc.Score; |
||
273 | var docId = scoreDoc.Doc; |
||
274 | var doc = searcher.Doc(docId); |
||
275 | |||
276 | var index = int.Parse(doc.Get(IndexId)); |
||
277 | results.Add(_indexedObjects[index]); |
||
278 | } |
||
279 | } |
||
280 | } |
||
4 | office | 281 | catch (ParseException) |
1 | office | 282 | { |
283 | //Log.Warning(ex, "Failed to parse search pattern"); |
||
4 | office | 284 | throw; |
1 | office | 285 | } |
4 | office | 286 | catch (Exception) |
1 | office | 287 | { |
288 | //Log.Error(ex, "An error occurred while searching, returning default results"); |
||
4 | office | 289 | throw; |
1 | office | 290 | } |
291 | finally |
||
292 | { |
||
293 | Searched?.Invoke(this, new SearchEventArgs(filter, results)); |
||
294 | } |
||
295 | } |
||
296 | |||
297 | return results; |
||
298 | } |
||
299 | |||
300 | private void Initialize() |
||
301 | { |
||
302 | if (_initialized) return; |
||
303 | |||
304 | _initialized = true; |
||
305 | |||
306 | _indexDirectory = GetDirectory(); |
||
307 | |||
308 | // Required to create empty index, which is required for our reader |
||
309 | using (var analyzer = new StandardAnalyzer(LuceneDefaults.Version)) |
||
310 | { |
||
311 | using (var indexWriter = |
||
312 | new IndexWriter(_indexDirectory, analyzer, IndexWriter.MaxFieldLength.UNLIMITED)) |
||
313 | { |
||
314 | indexWriter.Commit(); |
||
315 | } |
||
316 | } |
||
317 | } |
||
318 | |||
319 | protected abstract Directory GetDirectory(); |
||
320 | |||
321 | #endregion |
||
322 | } |
||
323 | } |