Update to latest DeLFT version

grobid-core/src/main/java/org/grobid/core/jni/DeLFTModel.java

@@ -53,7 +53,7 @@ public InitModel(String modelName, File modelPath) {
         public void run() { 
             Jep jep = JEPThreadPool.getInstance().getJEPInstance(); 
             try { 
-                jep.eval(this.modelName+" = sequenceLabelling.Sequence('" + this.modelName.replace("_", "-") + "')");
+                jep.eval(this.modelName+" = Sequence('" + this.modelName.replace("_", "-") + "')");
                 jep.eval(this.modelName+".load(dir_path='"+modelPath.getAbsolutePath()+"')");
             } catch(JepException e) {
                 LOGGER.error("DeLFT model initialization failed", e);
@@ -182,7 +182,7 @@ public void run() {
                 }
 
                 // init model to be trained
-                jep.eval("model = sequenceLabelling.Sequence('"+this.modelName+
+                jep.eval("model = Sequence('"+this.modelName+
                     "', max_epoch=100, recurrent_dropout=0.50, embeddings_name='glove-840B', use_ELMo="+useELMo+")");
 
                 // actual training

grobid-core/src/main/java/org/grobid/core/jni/JEPThreadPool.java

@@ -84,10 +84,11 @@ public Jep getJEPInstance() {
                 jep.eval("import numpy as np");
                 jep.eval("import keras.backend as K");
                 jep.eval("os.chdir('" + delftPath.getAbsolutePath() + "')");
-                jep.eval("from utilities.Embeddings import Embeddings");
-                jep.eval("import sequenceLabelling");
-                jep.eval("from sequenceLabelling.reader import load_data_and_labels_crf_file");
-                jep.eval("from sequenceLabelling.reader import load_data_crf_string");
+                jep.eval("from delft.utilities.Embeddings import Embeddings");
+                jep.eval("import delft.sequenceLabelling");
+                jep.eval("from delft.sequenceLabelling import Sequence");
+                jep.eval("from delft.sequenceLabelling.reader import load_data_and_labels_crf_file");
+                jep.eval("from delft.sequenceLabelling.reader import load_data_crf_string");
                 jep.eval("from sklearn.model_selection import train_test_split");
             } catch(JepException e) {
                 LOGGER.error("JEP initialization failed", e);

grobid-core/src/main/java/org/grobid/core/visualization/CitationsVisualizer.java

@@ -8,6 +8,7 @@
 import org.apache.pdfbox.cos.COSArray;
 import org.apache.pdfbox.pdmodel.PDDocument;
 import org.apache.pdfbox.pdmodel.PDPage;
+import org.apache.pdfbox.pdmodel.interactive.documentnavigation.destination.PDPageDestination;
 import org.apache.pdfbox.pdmodel.PDPageContentStream;
 import org.apache.pdfbox.pdmodel.common.PDRectangle;
 import org.apache.pdfbox.pdmodel.graphics.color.PDColor;
@@ -209,7 +210,9 @@ private static void annotatePage(PDDocument document,
 			Pair<Integer, Integer> thePlace = dictionary.get(teiId);
 			if (thePlace != null) {
 				PDPageFitWidthDestination destination = new PDPageFitWidthDestination();
-				destination.setPageNumber(thePlace.getA());
+                PDPage pdpage = document.getPage(thePlace.getA());
+                destination.setPage(pdpage);
+				//destination.setPageNumber(thePlace.getA());
 				destination.setTop(thePlace.getB());
                 PDActionGoTo action = new PDActionGoTo();
 				action.setDestination(destination);

grobid-core/src/main/java/org/grobid/core/visualization/FigureTableVisualizer.java

@@ -276,7 +276,7 @@ public static PDDocument annotateFigureAndTables(
         String q = XQueryProcessor.getQueryFromResources("figure-table-coords.xq");
         String tei = teiDoc.getTei();
         if (singleFile) {
-            System.out.println(tei);
+            //System.out.println(tei);
         }
         XQueryProcessor pr = new XQueryProcessor(tei);
         SequenceIterator it = pr.getSequenceIterator(q);
@@ -296,7 +296,7 @@ public static PDDocument annotateFigureAndTables(
 
         //VISUALIZING "IMAGE" elements from pdf2xml
         if (visualizePdf2xmlImages) {
-            q = XQueryProcessor.getQueryFromResources("figure-coords-pdf2xml.xq");
+            q = XQueryProcessor.getQueryFromResources("figure-coords-pdfalto.xq");
 
             pr = new XQueryProcessor(xmlFile);
             it = pr.getSequenceIterator(q);

grobid-core/src/main/resources/xq/figure-coords-pdfalto.xq

@@ -0,0 +1,2 @@
+for $i in //Illustration
+return string-join(($i/ancestor::PAGE/@number, $i/@HPOS, $i/@VPOS, $i/@WIDTH, $i/@HEIGHT), ',')

grobid-home/models/affiliation-address/config.json

@@ -1,20 +1,19 @@
 {
-    "fold_number": 1,
     "case_vocab_size": 8,
-    "use_ELMo": false,
-    "char_vocab_size": 152,
-    "model_name": "affiliation-address",
-    "dropout": 0.5,
-    "num_char_lstm_units": 25,
-    "num_word_lstm_units": 100,
-    "use_char_feature": true,
     "word_embedding_size": 300,
-    "model_type": "BidLSTM_CRF",
+    "fold_number": 1,
     "embeddings_name": "glove-840B",
-    "use_crf": true,
+    "num_char_lstm_units": 25,
     "max_char_length": 30,
-    "recurrent_dropout": 0.5,
+    "recurrent_dropout": 0.25,
+    "model_name": "affiliation-address",
+    "num_word_lstm_units": 100,
     "char_embedding_size": 25,
+    "char_vocab_size": 149,
+    "batch_size": 20,
+    "use_char_feature": true,
+    "model_type": "BidLSTM_CRF",
     "case_embedding_size": 5,
-    "batch_size": 20
+    "dropout": 0.5,
+    "use_crf": true
 }

grobid-home/models/citation/config.json

@@ -16,4 +16,4 @@
     "use_char_feature": true,
     "embeddings_name": "glove-840B",
     "word_embedding_size": 300
-}
+}

grobid-home/models/date/config.json

@@ -17,4 +17,4 @@
     "case_embedding_size": 5,
     "num_char_lstm_units": 25,
     "dropout": 0.5
-}
+}

grobid-home/models/figure/config.json

@@ -17,4 +17,4 @@
     "case_embedding_size": 5,
     "use_char_feature": true,
     "max_char_length": 30
-}
+}

grobid-home/models/header/config.json

@@ -17,4 +17,4 @@
     "word_embedding_size": 300,
     "embeddings_name": "glove-840B",
     "use_char_feature": true
-}
+}

grobid-home/models/name-citation/config.json

@@ -1,20 +1,20 @@
 {
-    "case_vocab_size": 8,
+    "max_char_length": 30,
     "fold_number": 1,
+    "use_crf": true,
+    "batch_size": 20,
+    "case_embedding_size": 5,
+    "num_char_lstm_units": 25,
     "embeddings_name": "glove-840B",
+    "model_name": "name-citation",
+    "model_type": "BidLSTM_CRF",
+    "recurrent_dropout": 0.5,
+    "char_vocab_size": 84,
     "use_char_feature": true,
-    "max_char_length": 30,
     "word_embedding_size": 300,
-    "case_embedding_size": 5,
-    "batch_size": 20,
-    "recurrent_dropout": 0.5,
+    "case_vocab_size": 8,
     "use_ELMo": false,
-    "model_name": "name-citation",
     "num_word_lstm_units": 100,
-    "use_crf": true,
-    "num_char_lstm_units": 25,
-    "model_type": "BidLSTM_CRF",
-    "char_vocab_size": 84,
     "char_embedding_size": 25,
     "dropout": 0.5
 }

grobid-home/models/name-header/config.json

@@ -17,4 +17,4 @@
     "char_vocab_size": 161,
     "use_char_feature": true,
     "dropout": 0.5
-}
+}

grobid-home/models/reference-segmenter/config.json

@@ -17,4 +17,4 @@
     "char_embedding_size": 25,
     "dropout": 0.5,
     "use_ELMo": false
-}
+}

grobid-home/models/table/config.json

@@ -3,7 +3,7 @@
     "embeddings_name": "glove-840B",
     "dropout": 0.5,
     "case_embedding_size": 5,
-    "model_name": "table",
+    "model_name": "grobid-table",
     "model_type": "BidLSTM_CRF",
     "num_word_lstm_units": 100,
     "char_vocab_size": 109,
@@ -17,4 +17,4 @@
     "recurrent_dropout": 0.5,
     "word_embedding_size": 300,
     "use_crf": true
-}
+}

grobid-service/src/main/java/org/grobid/service/process/GrobidRestProcessFiles.java

@@ -767,8 +767,12 @@ protected PDDocument annotate(File originFile,
         PDDocument outputDocument = null;
         // list of TEI elements that should come with coordinates
         List<String> elementWithCoords = new ArrayList<>();
-        elementWithCoords.add("ref");
-        elementWithCoords.add("biblStruct");
+        if (type == GrobidRestUtils.Annotation.CITATION) {
+            elementWithCoords.add("ref");
+            elementWithCoords.add("biblStruct");
+        } else if (type == GrobidRestUtils.Annotation.FIGURE) {
+            elementWithCoords.add("figure");
+        }
 
         GrobidAnalysisConfig config = new GrobidAnalysisConfig
             .GrobidAnalysisConfigBuilder()
@@ -777,18 +781,25 @@ protected PDDocument annotate(File originFile,
             .generateTeiCoordinates(elementWithCoords)
             .build();
 
-        Document teiDoc = engine.fullTextToTEIDoc(originFile, config);
+        DocumentSource documentSource = 
+            DocumentSource.fromPdf(originFile, config.getStartPage(), config.getEndPage(), true, true, false);
+
+        Document teiDoc = engine.fullTextToTEIDoc(documentSource, config);
+
+        documentSource = 
+            DocumentSource.fromPdf(originFile, config.getStartPage(), config.getEndPage(), true, true, false);
 
         PDDocument document = PDDocument.load(originFile);
         //If no pages, skip the document
         if (document.getNumberOfPages() > 0) {
-            DocumentSource documentSource = teiDoc.getDocumentSource();
             outputDocument = dispatchProcessing(type, document, documentSource, teiDoc);
         } else {
             throw new RuntimeException("Cannot identify any pages in the input document. " +
                 "The document cannot be annotated. Please check whether the document is valid or the logs.");
         }
 
+        documentSource.close(true, true, false);
+
         return outputDocument;
     }
 

grobid-trainer/resources/dataset/citation/corpus/citations.xml

@@ -3505,5 +3505,9 @@
 		<bibl><author>A. Oliva, A. Torralba</author>, "<title level="a">Modeling the shape of the scene: A holistic representation of the spatial envelope</title>", <title level="j">International journal of computer vision</title>, <date>2001</date>.</bibl>
 		<bibl><author>Boubakri, Narjess, Ghoul, Sadok El, and Saffar, Walid</author>, <date>2012</date>, <title level="a">Cash Holdings of Politically Connected Firms</title>, <note type="report">Working Paper</note>, <orgName>Hong Kong Polytechnic University</orgName>.</bibl>
 		<bibl><author>Jensen, M.C., Meckling, W.H.</author>, <date>1978</date>. <title level="a">Can the corporation survive?</title> <title level="j">Financial Analysts Journal</title>, <biblScope type="vol">34</biblScope>: <biblScope type="pp">31-37</biblScope>.</bibl>
+		<bibl><title level="m">Pdflib tet</title>. <ptr type="web">http://www.pdflib.com/products/tet/</ptr>. Accessed: <date>2015-05-12</date>.</bibl>
+		<bibl><title level="m">Poppler</title>. <ptr type="web">http://poppler.freedesktop.org</ptr>. Accessed: <date>2015-05-12</date>.</bibl>
+		<bibl><title level="m">Tabula</title>. <ptr type="web">http://tabula.technology</ptr>. Accessed: <date>2015-05-13</date>.</bibl>
+
 </listBibl>
 </tei>