Révision 3752

TXM/trunk/bundles/org.txm.conllu.core/src/org/txm/conllu/core/function/ImportCoNLLUAnnotations.java (revision 3752)
24 24
25 25 
	public static final String[] UD_PROPERTY_NAMES = { "id", "form", "lemma", "upos", "xpos", "feats", "head", "deprel", "deps", "misc" };
26 26
27 
	public static int _importAnnotations(File coonluFile, MainCorpus mainCorpus, String propertiesPrefix, String textId, Boolean normalize_word_ids, Set<String> headPropertiesToProject, Set<String> depsPropertiesToProject, Set<String> udPropertiesToImport) throws IOException, XMLStreamException {
27 
	public static int _importAnnotations(File coonluFile, MainCorpus mainCorpus,
28 
			String propertiesPrefix, String textId,
29 
			Boolean normalize_word_ids,
30 
			Boolean importCommentProperties,
31 
			Set<String> headPropertiesToProject,
32 
			Set<String> depsPropertiesToProject,
33 
			Set<String> udPropertiesToImport) throws IOException, XMLStreamException {
28 34
29 35 
		if (textId == null || textId.length() == 0) { // no text id provided, using the conllu file name
30 36 
			textId = coonluFile.getName().substring(0, coonluFile.getName().length() - 7);
......
118 124 
			}
119 125
120 126 
			if (sent_id != null) {
121 
				properties.put("#"+propertiesPrefix+"sentid", sent_id);
127 
				if (importCommentProperties) properties.put("#"+propertiesPrefix+"sentid", sent_id);
122 128 
				sent_id = ""; // reset value for next sentence
123 129 
			}
124 130 
			else {
125 
				properties.put("#"+propertiesPrefix+"sentid", "");
131 
				if (importCommentProperties) properties.put("#"+propertiesPrefix+"sentid", "");
126 132 
			}
127 133
128 134 
			if (newdoc_id != null) {
129 
				properties.put("#"+propertiesPrefix+"newdocid", newdoc_id);
135 
				if (importCommentProperties) properties.put("#"+propertiesPrefix+"newdocid", newdoc_id);
130 136 
				newdoc_id = null; // reset value for next sentence
131 137 
			}
132 138 
			else {
133 
				properties.put("#"+propertiesPrefix+"newdocid", "");
139 
				if (importCommentProperties) properties.put("#"+propertiesPrefix+"newdocid", "");
134 140 
			}
135 141
136 142 
			if (newpar_id != null) {
137 
				properties.put("#"+propertiesPrefix+"newparid", newpar_id);
143 
				if (importCommentProperties) properties.put("#"+propertiesPrefix+"newparid", newpar_id);
138 144 
				newpar_id = null; // reset value for next sentence
139 145 
			}
140 146 
			else {
141 
				properties.put("#"+propertiesPrefix+"newparid", "");
147 
				if (importCommentProperties) properties.put("#"+propertiesPrefix+"newparid", "");
142 148 
			}
143 149
144 150 
			if (wId == null) {
TXM/trunk/bundles/org.txm.conllu.rcp/src/org/txm/conllu/rcp/commands/ImportCoNLLUAnnotationsFromFile.java (revision 3752)
73 73 
	@Option(name = "normalize_word_ids", usage = "if set the CQP properties are replaced", widget = "Boolean", required = true, def = "false")
74 74 
	Boolean normalize_word_ids;
75 75
76 
	@Option(name = "import_conll_comment_properties", usage = "if set sentid, newdocid and parid will be imported in CQP properties", widget = "Boolean", required = true, def = "false")
77 
	Boolean import_conll_comment_properties;
78
76 79 
	@Option(name = "headPropertiesToProject", usage = "to create the headXYZ properties", widget = "StringArrayMultiple", metaVar="form	lemma	upos	xpos	feats	head	deprel	deps	misc", required = true, def = "deprel,upos")
77 80 
	String headPropertiesToProject;
78 81
......
117 120 
				return null;
118 121 
			}
119 122
120 
			return ImportCoNLLUAnnotationsFromDirectory.importAnnotationsFromCoNLLUFile(mainCorpus, conlluFile, propertiesPrefix, textId, normalize_word_ids,
123 
			return ImportCoNLLUAnnotationsFromDirectory.importAnnotationsFromCoNLLUFile(mainCorpus, conlluFile, propertiesPrefix, textId, normalize_word_ids, import_conll_comment_properties,
121 124 
					new HashSet<String>(Arrays.asList(headPropertiesToProject.split(","))),
122 125 
					new HashSet<String>(Arrays.asList(depsPropertiesToProject.split(","))),
123 126 
					new HashSet<String>(Arrays.asList(udPropertiesToImport.split(","))));
TXM/trunk/bundles/org.txm.conllu.rcp/src/org/txm/conllu/rcp/commands/ImportCoNLLUAnnotationsFromDirectory.java (revision 3752)
80 80
81 81 
	@Option(name = "normalize_word_ids", usage = "if set the CQP properties are replaced", widget = "Boolean", required = true, def = "false")
82 82 
	Boolean normalize_word_ids;
83
84 
	@Option(name = "import_conll_comment_properties", usage = "if set sentid, newdocid and parid will be imported in CQP properties", widget = "Boolean", required = true, def = "false")
85 
	Boolean import_conll_comment_properties;
83 86
84 87 
	@Option(name = "headPropertiesToProject", usage = "to create the head-XYZ properties from the word head", widget = "StringArrayMultiple", metaVar="form	lemma	upos	xpos	feats	head	deprel	deps	misc", required = true, def = "deprel,upos")
85 88 
	String headPropertiesToProject;
......
114 117 
		MainCorpus mainCorpus = corpus.getMainCorpus();
115 118
116 119 
		try {
117 
			HashSet<String> test = new HashSet<>();
118 
			for (String p : ImportCoNLLUAnnotations.UD_PROPERTY_NAMES) {
120 
			HashSet<String> test = new HashSet<>(); // will contains the CQP properties to update
121 
			for (String p : udPropertiesToImport.split(",")) { // test the properties to import
119 122 
				if (mainCorpus.getProperty(propertiesPrefix + p) != null) {
120 123 
					test.add(propertiesPrefix + p);
121 124 
				}
......
125 128 
				return null;
126 129 
			}
127 130
128 
			return importAnnotations(mainCorpus, conlluDirectory, propertiesPrefix, normalize_word_ids,
131 
			return importAnnotations(mainCorpus, conlluDirectory, propertiesPrefix, normalize_word_ids, import_conll_comment_properties,
129 132 
					new HashSet<String>(Arrays.asList(headPropertiesToProject.split(","))),
130 133 
					new HashSet<String>(Arrays.asList(depsPropertiesToProject.split(","))),
131 134 
					new HashSet<String>(Arrays.asList(udPropertiesToImport.split(","))));
......
151 154 
	 * @throws IOException
152 155 
	 * @throws XMLStreamException
153 156 
	 */
154 
	public static int importAnnotations(MainCorpus mainCorpus, File conlluDirectory, String propertiesPrefix, Boolean normalizeWordIds,
157 
	public static int importAnnotations(MainCorpus mainCorpus, File conlluDirectory, String propertiesPrefix, Boolean normalizeWordIds, Boolean importCommentProperties,
155 158 
			Set<String> headPropertiesToProject, Set<String> depsPropertiesToProject, Set<String> udPropertiesToImport) throws IOException, CqiServerError, CqiClientException, XMLStreamException {
156 159
157 160 
		Log.info(TXMCoreMessages.bind("Importing CONLL-U annotations of {0} in {1} using the ''{2}'' prefix...", conlluDirectory, mainCorpus, propertiesPrefix));
......
174 177
175 178 
		for (File coonluFile : files) {
176 179 
			cpb.tick();
177 
			nWordsInserted += ImportCoNLLUAnnotations._importAnnotations(coonluFile, mainCorpus, propertiesPrefix, null, normalizeWordIds, headPropertiesToProject, depsPropertiesToProject, udPropertiesToImport);
180 
			nWordsInserted += ImportCoNLLUAnnotations._importAnnotations(coonluFile, mainCorpus, propertiesPrefix, null, normalizeWordIds, importCommentProperties, headPropertiesToProject, depsPropertiesToProject, udPropertiesToImport);
178 181 
			nTextProcessed++;
179 182 
		}
180 183 
		cpb.done();
......
216 219 
	 * @throws IOException
217 220 
	 * @throws XMLStreamException
218 221 
	 */
219 
	public static int importAnnotationsFromCoNLLUFile(MainCorpus mainCorpus, File conlluFile, String propertiesPrefix, String textId, Boolean normalize_word_ids,
222 
	public static int importAnnotationsFromCoNLLUFile(MainCorpus mainCorpus, File conlluFile, String propertiesPrefix, String textId, Boolean normalize_word_ids, Boolean importCommentProperties,
220 223 
			Set<String> headPropertiesToProject, Set<String> depsPropertiesToProject, Set<String> udPropertiesToImport) throws IOException,
221 224 
			CqiServerError, CqiClientException, XMLStreamException {
222 225 
		Log.info(TXMCoreMessages.bind("Importing CONLL-u annotations of {0} in {1} using the ''{2}'' prefix...", conlluFile, mainCorpus, propertiesPrefix));
223 226
224 
		int nWordsInserted = ImportCoNLLUAnnotations._importAnnotations(conlluFile, mainCorpus, propertiesPrefix, textId, normalize_word_ids, headPropertiesToProject, depsPropertiesToProject, udPropertiesToImport);
227 
		int nWordsInserted = ImportCoNLLUAnnotations._importAnnotations(conlluFile, mainCorpus, propertiesPrefix, textId, normalize_word_ids, importCommentProperties, headPropertiesToProject, depsPropertiesToProject, udPropertiesToImport);
225 228
226 229 
		if (nWordsInserted == 0) {
227 230 
			Log.warning("** No annotation imported. Aborting.");

Formats disponibles : Unified diff