build/Makefile

@@ -1,5 +1,3 @@
-.PHONY: all release build transcrypt copy $(RES_LESS_FILE) $(RES_PY_FILE) $(RES_HTML_FILE) $(RES_XML_FILE)
-
 SRC_FOLDER=$(CURDIR)/../src
 RES_FOLDER=$(CURDIR)/../res
 
@@ -20,6 +18,7 @@ RES_PY_FILE=$(SRC_FOLDER)/$(PY_FILE)
 RES_HTML_FILE=$(RES_FOLDER)/$(HTML_FILE)
 RES_XML_FILE=$(RES_FOLDER)/$(XML_FILE)
 
+.PHONY: all release build transcrypt copy $(RES_LESS_FILE) $(RES_PY_FILE) $(RES_HTML_FILE) $(RES_XML_FILE) $(CONV_FILE)
 .DEFAULT_GOAL := all
 
 all: BR_FLAGS := $(BR_FLAGS) --debug

build/download_structure_conversions.sh

@@ -3,15 +3,19 @@
 # fail if any command fails
 set -e
 
+OUT_FILE=conversions.csv
+
 if [[ -z "${API_KEY}" ]]; then
   echo "API_KEY not defined, should be api key to CJVT's gitea"
   exit 1
 fi
 
-OUT_FILE=conversions.csv
-TEMP_FILE=$(mktemp)
+if [ -f $OUT_FILE ]; then
+  echo "file for conversion of SKE collocations: '$OUT_FILE' already exists, skipping."
+  exit 0
+fi
 
-rm -r $OUT_FILE
+TEMP_FILE=$(mktemp)
 
 curl -s "https://gitea.cjvt.si/api/v1/repos/generic/data_admin/contents/resources/structure_conversions.csv?token=$API_KEY" -o $TEMP_FILE
 echo "wc: $(wc $TEMP_FILE)"

src/message/ske_messages.py

@@ -58,7 +58,15 @@ class SkeCollocation:
     def __init__(self, data):
         self.word = data.word
         self.frequency = data.count
-        self.structure_name, self.structure_id = convert_structure(data.gramrel)
+        
+        info = convert_structure(data.gramrel)
+        if info is None:
+            self.structure_name = None
+            self.structure_id = None
+        else:
+            self.structure_name = info[0]
+            self.structure_id = info[1]
+            
         
         self.other = {"score": data.score, "cm": data.cm}
 
@@ -220,7 +228,7 @@ class SkeInsert(DataChgClickMessage):
     def _as_corpus_example(self, example):
         new_example = Example()
         new_example.inner = CorpusExample()
-        new_example.inner.other_attributes["exampleId"] = example.s_id
+        new_example.inner.other_attributes["example_id"] = example.s_id
         new_example.inner.cluster = ExampleClusters.first_empty_cluster()
         
         lex_left = ComponentLexeme()
@@ -229,7 +237,7 @@ class SkeInsert(DataChgClickMessage):
         
         lex_mid = ComponentLexeme()
         lex_mid.text = example["mid"]
-        lex_mid.role = "headword"
+        lex_mid.role = "collocation"
         
         lex_right = ComponentLexeme()
         lex_right.text = example["right"]
@@ -243,6 +251,7 @@ class SkeInsert(DataChgClickMessage):
         new_collocation.inner = MultiwordExample()
         
         new_collocation.inner.other_attributes["structure_id"] = example.structure_id
+        new_collocation.inner.other_attributes["structureName"] = example.structure_name
         new_collocation.inner.other_attributes["logDice"] = example.other["score"]
         new_collocation.inner.other_attributes["frequency"] = example.frequency
         new_collocation.inner.type = "collocation"
@@ -253,7 +262,7 @@ class SkeInsert(DataChgClickMessage):
         
         lex_mid = ComponentLexeme()
         lex_mid.text = example.word
-        lex_mid.role = "headword"
+        lex_mid.role = "collocation"
         
         lex_right = ComponentLexeme()
         lex_right.text = ""