Context Navigation

source: trunk/grails-app/services/dbnp/importer/ImporterService.groovy @ 1416

Last change on this file since 1416 was 1416, checked in by t.w.abma@…, 12 years ago
again fixed the importer, persisting seemed to work fine, but Hibernate session was out of sync
Property svn:keywords set to `Date Author Rev`
File size: 23.1 KB

Line
1	/**
2	* Importer service
3	*
4	* The importer service handles the import of tabular, comma delimited and Excel format
5	* based files.
6	*
7	* @package importer
8	* @author t.w.abma@umcutrecht.nl
9	* @since 20100126
10	*
11	* Revision information:
12	* $Rev: 1416 $
13	* $Author: t.w.abma@umcutrecht.nl $
14	* $Date: 2011-01-19 15:01:08 +0000 (wo, 19 jan 2011) $
15	*/
16
17	package dbnp.importer
18	import org.apache.poi.ss.usermodel.*
19	import org.apache.poi.xssf.usermodel.XSSFCell
20
21	import dbnp.studycapturing.TemplateFieldType
22	import dbnp.studycapturing.Template
23	import dbnp.studycapturing.SamplingEvent
24	import dbnp.studycapturing.Study
25	import dbnp.studycapturing.Subject
26	import dbnp.studycapturing.Event
27	import dbnp.studycapturing.Sample
28
29	class ImporterService {
30	def AuthenticationService
31
32	boolean transactional = true
33
34	/**
35	* @param is input stream representing the (workbook) resource
36	* @return high level representation of the workbook
37	*/
38	Workbook getWorkbook(InputStream is) {
39	WorkbookFactory.create(is)
40	}
41
42	/**
43	* @param wb high level representation of the workbook
44	* @param sheetindex sheet to use within the workbook
45	* @return header representation as a MappingColumn hashmap
46	*/
47	def getHeader(Workbook wb, int sheetindex, int headerrow, int datamatrix_start, theEntity=null) {
48	def sheet = wb.getSheetAt(sheetindex)
49	def sheetrow = sheet.getRow(datamatrix_start)
50	//def header = []
51	def header = [:]
52	def df = new DataFormatter()
53	def property = new String()
54
55	//for (Cell c: sheet.getRow(datamatrix_start)) {
56
57	(0..sheetrow.getLastCellNum() -1 ).each { columnindex ->
58
59	//def index = c.getColumnIndex()
60	def datamatrix_celltype = sheet.getRow(datamatrix_start).getCell(columnindex,Row.CREATE_NULL_AS_BLANK).getCellType()
61	def datamatrix_celldata = df.formatCellValue(sheet.getRow(datamatrix_start).getCell(columnindex))
62	def datamatrix_cell = sheet.getRow(datamatrix_start).getCell(columnindex)
63	def headercell = sheet.getRow(headerrow-1+sheet.getFirstRowNum()).getCell(columnindex)
64	def tft = TemplateFieldType.STRING //default templatefield type
65
66	// Check for every celltype, currently redundant code, but possibly this will be
67	// a piece of custom code for every cell type like specific formatting
68
69	switch (datamatrix_celltype) {
70	case Cell.CELL_TYPE_STRING:
71	//parse cell value as double
72	def doubleBoolean = true
73	def fieldtype = TemplateFieldType.STRING
74
75	// is this string perhaps a double?
76	try {
77	formatValue(datamatrix_celldata, TemplateFieldType.DOUBLE)
78	} catch (NumberFormatException nfe) { doubleBoolean = false }
79	finally {
80	if (doubleBoolean) fieldtype = TemplateFieldType.DOUBLE
81	}
82
83	header[columnindex] = new dbnp.importer.MappingColumn(name:df.formatCellValue(headercell),
84	templatefieldtype:fieldtype,
85	index:columnindex,
86	entity:theEntity,
87	property:property);
88
89	break
90	case Cell.CELL_TYPE_NUMERIC:
91	def fieldtype = TemplateFieldType.LONG
92	def doubleBoolean = true
93	def longBoolean = true
94
95	// is this cell really an integer?
96	try {
97	Long.valueOf(datamatrix_celldata)
98	} catch (NumberFormatException nfe) { longBoolean = false }
99	finally {
100	if (longBoolean) fieldtype = TemplateFieldType.LONG
101	}
102
103	// it's not an long, perhaps a double?
104	if (!longBoolean)
105	try {
106	formatValue(datamatrix_celldata, TemplateFieldType.DOUBLE)
107	} catch (NumberFormatException nfe) { doubleBoolean = false }
108	finally {
109	if (doubleBoolean) fieldtype = TemplateFieldType.DOUBLE
110	}
111
112	if (DateUtil.isCellDateFormatted(datamatrix_cell)) fieldtype = TemplateFieldType.DATE
113
114	header[columnindex] = new dbnp.importer.MappingColumn(name:df.formatCellValue(headercell),
115	templatefieldtype:fieldtype,
116	index:columnindex,
117	entity:theEntity,
118	property:property);
119	break
120	case Cell.CELL_TYPE_BLANK:
121	header[columnindex] = new dbnp.importer.MappingColumn(name:df.formatCellValue(headercell),
122	templatefieldtype:TemplateFieldType.STRING,
123	index:columnindex,
124	entity:theEntity,
125	property:property);
126	break
127	default:
128	header[columnindex] = new dbnp.importer.MappingColumn(name:df.formatCellValue(headercell),
129	templatefieldtype:TemplateFieldType.STRING,
130	index:columnindex,
131	entity:theEntity,
132	property:property);
133	break
134	} // end of switch
135	} // end of cell loop
136	return header
137	}
138
139	/**
140	* This method is meant to return a matrix of the rows and columns
141	* used in the preview
142	*
143	* @param wb workbook object
144	* @param sheetindex sheet index used
145	* @param rows amount of rows returned
146	* @return two dimensional array (matrix) of Cell objects
147	*/
148
149	Object[][] getDatamatrix(Workbook wb, header, int sheetindex, int datamatrix_start, int count) {
150	def sheet = wb.getSheetAt(sheetindex)
151	def rows = []
152	def df = new DataFormatter()
153
154	count = (count < sheet.getLastRowNum()) ? count : sheet.getLastRowNum()
155
156	// walk through all rows
157	((datamatrix_start+sheet.getFirstRowNum())..count).each { rowindex ->
158	def row = []
159
160	(0..header.size()-1).each { columnindex ->
161	def c = sheet.getRow(rowindex).getCell(columnindex, Row.CREATE_NULL_AS_BLANK)
162	row.add(c)
163	}
164
165	rows.add(row)
166	}
167
168	return rows
169	}
170
171	/**
172	* This method will move a file to a new location.
173	*
174	* @param file File object to move
175	* @param folderpath folder to move the file to
176	* @param filename (new) filename to give
177	* @return if file has been moved succesful, the new path and filename will be returned, otherwise an empty string will be returned
178	*/
179	def moveFile(File file, String folderpath, String filename) {
180	try {
181	def rnd = ""; //System.currentTimeMillis()
182	file.transferTo(new File(folderpath, rnd+filename))
183	return folderpath + filename
184	} catch(Exception exception) {
185	log.error "File move error, ${exception}"
186	return ""
187	}
188	}
189
190	/**
191	* @return random numeric value
192	*/
193	def random = {
194	return System.currentTimeMillis() + Runtime.runtime.freeMemory()
195	}
196
197	/**
198	* Method to read data from a workbook and to import data into a two dimensional
199	* array
200	*
201	* @param template_id template identifier to use fields from
202	* @param wb POI horrible spreadsheet formatted workbook object
203	* @param mcmap linked hashmap (preserved order) of MappingColumns
204	* @param sheetindex sheet to use when using multiple sheets
205	* @param rowindex first row to start with reading the actual data (NOT the header)
206	* @return two dimensional array containing records (with entities)
207	*
208	* @see dbnp.importer.MappingColumn
209	*/
210	def importData(template_id, Workbook wb, int sheetindex, int rowindex, mcmap) {
211	def sheet = wb.getSheetAt(sheetindex)
212	def template = Template.get(template_id)
213	def table = []
214	def failedcells = [] // list of records
215
216	// walk through all rows and fill the table with records
217	(rowindex..sheet.getLastRowNum()).each { i ->
218	// Create an entity record based on a row read from Excel and store the cells which failed to be mapped
219	def (record, failed) = createRecord(template, sheet.getRow(i), mcmap)
220
221	// Add record with entity and its values to the table
222	table.add(record)
223
224	// If failed cells have been found, add them to the failed cells list
225	if (failed?.importcells?.size()>0) failedcells.add(failed)
226	}
227
228	return [table,failedcells]
229	}
230
231	/** Method to put failed cells back into the datamatrix. Failed cells are cell values
232	* which could not be stored in an entity (e.g. Humu Supiuns in an ontology field).
233	* Empty corrections should not be stored
234	*
235	* @param datamatrix two dimensional array containing entities and possibly also failed cells
236	* @param failedcells list with maps of failed cells in [mappingcolumn, cell] format
237	* @param correctedcells map of corrected cells in [cellhashcode, value] format
238	**/
239	def saveCorrectedCells(datamatrix, failedcells, correctedcells) {
240
241	// Loop through all failed cells (stored as
242	failedcells.each { record ->
243	record.value.importcells.each { cell ->
244
245	// Get the corrected value
246	def correctedvalue = correctedcells.find { it.key.toInteger() == cell.getIdentifier()}.value
247
248	// Find the record in the table which the mappingcolumn belongs to
249	def tablerecord = datamatrix.find { it.hashCode() == record.key }
250
251	// Loop through all entities in the record and correct them if necessary
252	tablerecord.each { rec ->
253	rec.each { entity ->
254	try {
255	// Update the entity field
256	entity.setFieldValue(cell.mappingcolumn.property, correctedvalue)
257	//println "Adjusted " + cell.mappingcolumn.property + " to " + correctedvalue
258	}
259	catch (Exception e) {
260	//println "Could not map corrected ontology: " + cell.mappingcolumn.property + " to " + correctedvalue
261	}
262	}
263	} // end of table record
264	} // end of cell record
265	} // end of failedlist
266	}
267
268	/**
269	* Method to store a matrix containing the entities in a record like structure. Every row in the table
270	* contains one or more entity objects (which contain fields with values). So actually a row represents
271	* a record with fields from one or more different entities.
272	*
273	* @param study entity Study
274	* @param datamatrix two dimensional array containing entities with values read from Excel file
275	*/
276	def saveDatamatrix(Study study, datamatrix) {
277	def validatedSuccesfully = 0
278	def entitystored = null
279
280	// Study passed? Sync data
281	if (study!=null) study.refresh()
282
283	//study.subjects.each { it.refresh() }
284
285	// go through the data matrix, read every record and validate the entity and try to persist it
286	datamatrix.each { record ->
287	record.each { entity ->
288	switch (entity.getClass()) {
289	case Study : log.info "Persisting Study `" + entity + "`: "
290	entity.owner = AuthenticationService.getLoggedInUser()
291	persistEntity(entity)
292	break
293	case Subject : log.info "Persisting Subject `" + entity + "`: "
294
295	// is the current entity not already in the database?
296	//entitystored = isEntityStored(entity)
297
298	// this entity is new, so add it to the study
299	//if (entitystored==null)
300	study.addToSubjects(entity)
301
302	/*else { // existing entity, so update it
303	updateEntity(entitystored, entity)
304	updatedentities.add(entity)
305	}*/
306
307	break
308	case Event : log.info "Persisting Event `" + entity + "`: "
309	study.addToEvents(entity)
310	break
311	case Sample : log.info "Persisting Sample `" + entity +"`: "
312
313	// is this sample validatable (sample name unique for example?)
314	study.addToSamples(entity)
315
316	break
317	case SamplingEvent: log.info "Persisting SamplingEvent `" + entity + "`: "
318	study.addToSamplingEvents(entity)
319	break
320	default : log.info "Skipping persisting of `" + entity.getclass() +"`"
321	break
322	} // end switch
323	} // end record
324	} // end datamatrix
325
326
327	if (study!=null) study.save(failOnError:true)
328	//persistEntity(study)
329
330	//return [validatedSuccesfully, updatedentities, failedtopersist]
331	//return [0,0,0]
332	}
333
334	/**
335	* Check whether an entity already exist. A unique field in the entity is
336	* used to check whether the instantiated entity (read from Excel) is new.
337	* If the entity is found in the database it will be returned as is.
338	*
339	* @param entity entity object like a Study, Subject, Sample et cetera
340	* @return entity if found, otherwise null
341	*/
342	def isEntityStored(entity) {
343	switch (entity.getClass()) {
344	case Study : return Study.findByCode(entity.code)
345	break
346	case Subject : return Subject.findByParentAndName(entity.parent, entity.name)
347	break
348	case Event : break
349	case Sample :
350	break
351	case SamplingEvent : break
352	default : // unknown entity
353	return null
354	}
355	}
356
357	/**
358	* Find the entity and update the fields. The entity is an instance
359	* read from Excel. This method looks in the database for the entity
360	* having the same identifier. If it has found the same entity
361	* already in the database, it will update the record.
362	*
363	* @param entitystored existing record in the database to update
364	* @param entity entity read from Excel
365	*/
366	def updateEntity(entitystored, entity) {
367	switch (entity.getClass()) {
368	case Study : break
369	case Subject : entitystored.properties = entity.properties
370	entitystored.save()
371	break
372	case Event : break
373	case Sample : break
374	case SamplingEvent : break
375	default : // unknown entity
376	return null
377	}
378	}
379
380	/**
381	* Method to persist entities into the database
382	* Checks whether entity already exists (based on identifier column 'name')
383	*
384	* @param entity entity object like Study, Subject, Protocol et cetera
385	*
386	*/
387	boolean persistEntity(entity) {
388	log.info ".import wizard persisting ${entity}"
389
390	try {
391	entity.save(flush:true)
392	return true
393
394	} catch (Exception e) {
395	def session = sessionFactory.currentSession
396	session.setFlushMode(org.hibernate.FlushMode.MANUAL)
397	log.error ".import wizard, failed to save entity:\n" + org.apache.commons.lang.exception.ExceptionUtils.getRootCauseMessage(e)
398	}
399
400	return true
401	}
402
403	/**
404	* This method creates a record (array) containing entities with values
405	*
406	* @param template_id template identifier
407	* @param excelrow POI based Excel row containing the cells
408	* @param mcmap map containing MappingColumn objects
409	* @return list of entities and list of failed cells
410	*/
411	def createRecord(template, Row excelrow, mcmap) {
412	def df = new DataFormatter()
413	def tft = TemplateFieldType
414	def record = [] // list of entities and the read values
415	def failed = new ImportRecord() // map with entity identifier and failed mappingcolumn
416
417	// Initialize all possible entities with the chosen template
418	def study = new Study(template: template)
419	def subject = new Subject(template: template)
420	def samplingEvent = new SamplingEvent(template: template)
421	def event = new Event(template: template)
422	def sample = new Sample(template: template)
423
424	// Go through the Excel row cell by cell
425	for (Cell cell: excelrow) {
426	// get the MappingColumn information of the current cell
427	def mc = mcmap[cell.getColumnIndex()]
428	def value
429
430	// Check if column must be imported
431	if (mc!=null) if (!mc.dontimport) {
432	try {
433	value = formatValue(df.formatCellValue(cell), mc.templatefieldtype)
434	} catch (NumberFormatException nfe) {
435	value = ""
436	}
437
438	try {
439	// which entity does the current cell (field) belong to?
440	switch (mc.entity) {
441	case Study: // does the entity already exist in the record? If not make it so.
442	(record.any {it.getClass() == mc.entity}) ? 0 : record.add(study)
443	study.setFieldValue(mc.property, value)
444	break
445	case Subject: (record.any {it.getClass() == mc.entity}) ? 0 : record.add(subject)
446	subject.setFieldValue(mc.property, value)
447	break
448	case SamplingEvent: (record.any {it.getClass() == mc.entity}) ? 0 : record.add(samplingEvent)
449	samplingEvent.setFieldValue(mc.property, value)
450	break
451	case Event: (record.any {it.getClass() == mc.entity}) ? 0 : record.add(event)
452	event.setFieldValue(mc.property, value)
453	break
454	case Sample: (record.any {it.getClass() == mc.entity}) ? 0 : record.add(sample)
455	sample.setFieldValue(mc.property, value)
456	break
457	case Object: // don't import
458	break
459	} // end switch
460	} catch (Exception iae) {
461	log.error ".import wizard error could not set property `" + mc.property + "` to value `" + value + "`"
462	// store the mapping column and value which failed
463	def identifier
464
465	switch (mc.entity) {
466	case Study: identifier = study.getIdentifier()
467	break
468	case Subject: identifier = subject.getIdentifier()
469	break
470	case SamplingEvent: identifier = samplingEvent.getIdentifier()
471	break
472	case Event: identifier = event.getIdentifier()
473	break
474	case Sample: identifier = sample.getIdentifier()
475	break
476	case Object: // don't import
477	break
478	}
479
480	def mcInstance = new MappingColumn()
481	mcInstance.properties = mc.properties
482	failed.addToImportcells(new ImportCell(mappingcolumn:mcInstance, value:value, entityidentifier:identifier))
483	}
484	} // end
485	} // end for
486	// a failed column means that using the entity.setFieldValue() threw an exception
487	return [record, failed]
488	}
489
490	/**
491	* Method to parse a value conform a specific type
492	* @param value string containing the value
493	* @return object corresponding to the TemplateFieldType
494	*/
495	def formatValue(String value, TemplateFieldType type) throws NumberFormatException {
496	switch (type) {
497	case TemplateFieldType.STRING : return value.trim()
498	case TemplateFieldType.TEXT : return value.trim()
499	case TemplateFieldType.LONG : return (long) Double.valueOf(value)
500	//case TemplateFieldType.FLOAT : return Float.valueOf(value.replace(",","."));
501	case TemplateFieldType.DOUBLE : return Double.valueOf(value.replace(",","."));
502	case TemplateFieldType.STRINGLIST : return value.trim()
503	case TemplateFieldType.ONTOLOGYTERM : return value.trim()
504	case TemplateFieldType.DATE : return value
505	default : return value
506	}
507	}
508
509	// classes for fuzzy string matching
510	// <FUZZY MATCHING>
511	static def similarity(l_seq, r_seq, degree=2) {
512	def l_histo = countNgramFrequency(l_seq, degree)
513	def r_histo = countNgramFrequency(r_seq, degree)
514
515	dotProduct(l_histo, r_histo) /
516	Math.sqrt(dotProduct(l_histo, l_histo) *
517	dotProduct(r_histo, r_histo))
518	}
519
520	static def countNgramFrequency(sequence, degree) {
521	def histo = [:]
522	def items = sequence.size()
523
524	for (int i = 0; i + degree <= items; i++)
525	{
526	def gram = sequence[i..<(i + degree)]
527	histo[gram] = 1 + histo.get(gram, 0)
528	}
529	histo
530	}
531
532	static def dotProduct(l_histo, r_histo) {
533	def sum = 0
534	l_histo.each { key, value ->
535	sum = sum + l_histo[key] * r_histo.get(key, 0)
536	}
537	sum
538	}
539
540	static def stringSimilarity (l_str, r_str, degree=2) {
541
542	similarity(l_str.toString().toLowerCase().toCharArray(),
543	r_str.toString().toLowerCase().toCharArray(),
544	degree)
545	}
546
547	static def mostSimilar(pattern, candidates, threshold=0) {
548	def topScore = 0
549	def bestFit = null
550
551	candidates.each { candidate ->
552	def score = stringSimilarity(pattern, candidate)
553	if (score > topScore) {
554	topScore = score
555	bestFit = candidate
556	}
557	}
558
559	if (topScore < threshold)
560	bestFit = null
561
562	bestFit
563	}
564	// </FUZZY MATCHING>
565
566	}

Note: See TracBrowser for help on using the repository browser.

Download in other formats: