From 6e1f3bd16a680aaecf601e4995a2119267072c99 Mon Sep 17 00:00:00 2001
From: Ozbolt Menegatti <ozbolt.menegatti@gmail.com>
Date: Wed, 8 Jul 2020 21:21:10 +0200
Subject: [PATCH 1/4] saving example_id correctly (see import of corpus
 example)

---
 src/message/ske_messages.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/message/ske_messages.py b/src/message/ske_messages.py
index 6900cdf..1a84b3e 100644
--- a/src/message/ske_messages.py
+++ b/src/message/ske_messages.py
@@ -220,7 +220,7 @@ 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()

From 8cb63eaa1d5d0ef74abe369000f31d5359798862 Mon Sep 17 00:00:00 2001
From: Ozbolt Menegatti <ozbolt.menegatti@gmail.com>
Date: Wed, 8 Jul 2020 21:21:24 +0200
Subject: [PATCH 2/4] mid component should be colocation, not headword

---
 src/message/ske_messages.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/message/ske_messages.py b/src/message/ske_messages.py
index 1a84b3e..0032312 100644
--- a/src/message/ske_messages.py
+++ b/src/message/ske_messages.py
@@ -229,7 +229,7 @@ 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"]

From 5396868601ff0ac5e0a13d9fb361a59b14a04b1c Mon Sep 17 00:00:00 2001
From: Ozbolt Menegatti <ozbolt.menegatti@gmail.com>
Date: Wed, 8 Jul 2020 21:21:24 +0200
Subject: [PATCH 3/4] mid component should be colocation, not headword,

Also, readd structureName and handle unknown structure when importing ske collocation
---
 src/message/ske_messages.py | 13 +++++++++++--
 1 file changed, 11 insertions(+), 2 deletions(-)

diff --git a/src/message/ske_messages.py b/src/message/ske_messages.py
index 0032312..9769f3a 100644
--- a/src/message/ske_messages.py
+++ b/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}
 
@@ -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 = ""

From b17510ffe7f4a1a511f5a0f5240dbc17abb218e3 Mon Sep 17 00:00:00 2001
From: Ozbolt Menegatti <ozbolt.menegatti@gmail.com>
Date: Wed, 8 Jul 2020 21:52:42 +0200
Subject: [PATCH 4/4] improving build, correct phony targets and more correct
 and verbose download_structure_conversions

---
 build/Makefile                          |  3 +--
 build/download_structure_conversions.sh | 10 +++++++---
 2 files changed, 8 insertions(+), 5 deletions(-)

diff --git a/build/Makefile b/build/Makefile
index 5113996..ba3d85d 100644
--- a/build/Makefile
+++ b/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
diff --git a/build/download_structure_conversions.sh b/build/download_structure_conversions.sh
index e313eb9..db8540a 100755
--- a/build/download_structure_conversions.sh
+++ b/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)"