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