First commit on scripts branch

Makefile

@@ -13,10 +13,11 @@ SSJ_FILE = "$(MAKE_ROOT)/data/ssj_file_link"
 # KRES_FOLDER = "$(MAKE_ROOT)/data/samples/kres_xml"
 # KRES_FOLDER = "$(MAKE_ROOT)/data/kres_xml_folder_link"
 KRES_FOLDER = "/home/kristjan/kres_data/payload/kres_xml"
+GIGAFIDA_FOLDER = "/home/lukakrsnik/cjvt-valency/data_all/giga_orig"
 # KRES_SRL_FOLDER = "$(MAKE_ROOT)/data/samples/kres_srl_json"
 # KRES_SRL_FOLDER = "$(MAKE_ROOT)/data/kres_json_folder_link"
 KRES_SRL_FOLDER = "/home/kristjan/kres_data/payload/kres_json"
-
+GIGAFIDA_SRL_FOLDER = "/home/lukakrsnik/cjvt-valency/data_all/final_json"
 # This file comes with the source code. Make sure you unpack it and name it right.
 SSKJ_WORDLIST = "$(MAKE_ROOT)/data/wordlist.json"
 SSKJ_JSON = "$(MAKE_ROOT)/data/sskj_senses.json"
@@ -24,8 +25,8 @@ SSKJ_JSON = "$(MAKE_ROOT)/data/sskj_senses.json"
 # for pre-generation the index of all headwords and functors
 APPINDEX_PATH = "$(MAKE_ROOT)/data/appindex.json"
 
-# OUTPUT = "db"
-OUTPUT = "file"
+OUTPUT = "db"
+# OUTPUT = "file"
 OUTDIR = "/project/data"  # if you're running this in docker, make sure to mount the volume
 DBADDR = "0.0.0.0:27017"  # don't use localhost
 
@@ -33,7 +34,7 @@ DBADDR = "0.0.0.0:27017"  # don't use localhost
 # create it from env.default
 include env.local
 
-N_CORES = 3
+N_CORES = 4
 # insert kres files into database in chunks, for fewer connections
 KRES_CHUNK_SIZE = 30
 
@@ -93,7 +94,14 @@ fill-database-kres: data/samples
 		--chunk-size $(KRES_CHUNK_SIZE) \
 		--cores $(N_CORES)
 
-
+fill-database-gigafida: data/samples
+	python3 src/pkg/cjvt-corpusparser/corpusparser/main.py --kres-folder $(GIGAFIDA_FOLDER) \
+		--corpus="gigafida" \
+		--ssj-file $(SSJ_FILE) --kres-srl-folder $(GIGAFIDA_SRL_FOLDER) \
+		--output $(OUTPUT) --outdir $(OUTDIR) --dbaddr $(DBADDR) \
+		--dbuser $(DB_USR_USER) --dbpass $(DB_USR_PASS) \
+		--chunk-size $(KRES_CHUNK_SIZE) \
+		--cores $(N_CORES)
 
 ## Frontend