added directories to upload dataset (#24)

* added upload-dataset make target

* added dataset directories to upload

Makefile

@@ -26,6 +26,9 @@ SAMPLE_PDF_URL = https://cdn.elifesciences.org/articles/32671/elife-32671-v2.pdf
 # Specify the location where to copy the model to
 CLOUD_MODELS_PATH =
 
+# Specify the location where to copy the dataset to
+CLOUD_DATATSET_PATH =
+
 NOT_SLOW_PYTEST_ARGS = -m 'not slow'
 
 ARGS =
@@ -117,6 +120,12 @@ generate-grobid-training-data:
 		"$(DATASET_DIR)"
 
 
+upload-dataset:
+	$(RUN) upload-dataset.sh \
+		"${DATASET_DIR}" \
+		"$(CLOUD_DATATSET_PATH)"
+
+
 copy-raw-header-training-data-to-tei:
 	$(RUN) bash -c '\
 		mkdir -p "$(DATASET_DIR)/header/corpus/tei" && \

scripts/upload-dataset.sh

@@ -45,6 +45,33 @@ sub_dirs=(
     "header/corpus/tei"
     "header/corpus/tei-raw"
     "header/corpus/tei-auto"
+    "fulltext/corpus/raw"
+    "fulltext/corpus/tei"
+    "fulltext/corpus/tei-raw"
+    "fulltext/corpus/tei-auto"
+    "figure/corpus/raw"
+    "figure/corpus/tei"
+    "figure/corpus/tei-raw"
+    "figure/corpus/tei-auto"
+    "reference-segmenter/corpus/raw"
+    "reference-segmenter/corpus/tei"
+    "reference-segmenter/corpus/tei-raw"
+    "reference-segmenter/corpus/tei-auto"
+    "affiliation-address/corpus"
+    "affiliation-address/corpus-raw"
+    "affiliation-address/corpus-auto"
+    "citation/corpus"
+    "citation/corpus-raw"
+    "citation/corpus-auto"
+    "name/citation/corpus"
+    "name/citation/corpus-raw"
+    "name/citation/corpus-auto"
+    "name/header/corpus"
+    "name/header/corpus-raw"
+    "name/header/corpus-auto"
+    "date/corpus"
+    "date/corpus-raw"
+    "date/corpus-auto"
     "xml"
 ) 
 for sub_dir in "${sub_dirs[@]}"; do