diff --git a/README.md b/README.md
index 48e4f12..8d34bc8 100644
--- a/README.md
+++ b/README.md
@@ -17,3 +17,20 @@ Available languages:
 - [Italian 4.9.4](https://github.com/IQSS/dataverse-docker/blob/master/dataversedock/dataverse-property-files/it-IT/) maintained by [Centro Interdipartimentale UniData](http://www.unidata.unimib.it)
 - [Hungarian, 4.9.4](https://github.com/IQSS/dataverse-docker/tree/master/dataversedock/dataverse-property-files/hu-HU) maintained by [TARKI](http://tarki.hu)
 
+## Generating tools
+
+Bash script `generate.sh` helps to maintains up to date i18n files. It use en_US files to detect all files and keys from latest version et try to fill value for desired language from latest available version for this language.
+
+Usage:
+
+```bash
+bash ./generate.sh fr_CA => generate fr_CA from most recent version
+bash ./generate.sh fr_FR develop => generate fr_FR from develop branch
+```
+
+Process:
+
+1. Create properties files by copying files from en_US, rename then with language code, and removing value inside.
+2. Extract files from most recent version for this language (branch dataverse-vXXX) or from the specified branch
+3. Fill values by searching them in extracted files, in the same filename or in all files
+4. Print "not found" keys (new keys)
\ No newline at end of file
diff --git a/generate.sh b/generate.sh
new file mode 100644
index 0000000..6e80ce7
--- /dev/null
+++ b/generate.sh
@@ -0,0 +1,119 @@
+#!/bin/bash
+
+TARGET_LANG=$1
+BASE_LANG="en_US"
+
+function usage(){
+    echo "Usage: bash $0 LANG [FROM_BRANCH]"
+    echo "Genetate i18n properties files for LANG language, based on keys found en_US/*.properties  and value from LANG from a previous dataverse-v4.XX branch or specific branch (liek develop)"
+    echo "Ex:"
+    echo "  bash $0 fr_CA => generate fr_CA from most recent version"
+    echo "  bash $0 fr_FR develop => generate fr_FR from develop branch"
+    exit 1
+}
+
+
+if [ -z $TARGET_LANG ]; then
+    usage
+fi
+
+if [ ! -d $BASE_LANG ]; then
+    echo "Base language directory is missing: [$BASE_LANG]"
+    usage
+fi
+
+if [ -d $TARGET_LANG ]; then
+    echo "Target language directory already exists. Please delete it manually : rm -r $TARGET_LANG"
+    usage
+fi
+
+FROM_BRANCH=$2
+
+if [ -z $FROM_BRANCH ];then
+    # Detect branch with existing language file
+    for version in $(git branch | grep dataverse | sed "s/.*dataverse-\(.*\)/\\1/g"); do
+        if [[ "$(git show dataverse-$version: | grep -o $TARGET_LANG)" == "$TARGET_LANG" ]]; then
+            FROM_BRANCH=dataverse-$version
+        fi
+    done
+else
+    # Check existing language file
+    if [[ "$(git show $FROM_BRANCH: | grep -o $TARGET_LANG)" != "$TARGET_LANG" ]]; then
+        FROM_BRANCH=""
+    fi
+fi
+
+if [ -z $FROM_BRANCH ]; then
+    echo "Can not find suitable version branch for language $TARGET_LANG"
+    exit 1
+fi
+
+
+echo "=== Generate $TARGET_LANG based on branch $FROM_BRANCH"
+
+# Copie reference language files (en_US)
+cp -r $BASE_LANG $TARGET_LANG
+
+# Force sed to continue parsing when it hits an “invalid” character
+export LANG=C 
+find $TARGET_LANG -name '*.properties' -exec sed -i "s/^\([^=]*=\).*$/\\1/g" {} \;
+
+for file in $TARGET_LANG/* ; do
+    #append new line (while read line bug if no newline at EOF)
+    echo >> $file
+    mv $file ${file/.properties/_$(echo $TARGET_LANG | awk -F '_' '{print $1}').properties}
+done
+
+
+# extract files from original branch
+TMP_DIR=$(mktemp -d )
+for f in $(git show $FROM_BRANCH:$TARGET_LANG | grep properties); do
+    git show $FROM_BRANCH:$TARGET_LANG/$f > $TMP_DIR/$f
+done
+
+# fill files
+for f in $(find $TARGET_LANG -name '*.properties')
+do
+    CURRENT_FILE=$(basename $f)
+    while read originalline
+    do
+        if [[ "$originalline" =~ ^[0-9a-zA-Z].*$ ]]; then
+            originalline=${originalline//u0020/\\u0020}
+            originalline=${originalline//\\/\\\\}
+            originalline=${originalline//&/\\&}
+            targetline=""
+            # First search in same fille
+            if [ -f $TMP_DIR/$CURRENT_FILE ]; then
+                targetline=$(grep -rh "^$originalline" $TMP_DIR/$CURRENT_FILE | head -n 1 )
+                targetline=${targetline//\\/\\\\\\}
+                targetline=${targetline//|/\\|}
+                targetline=${targetline//\\n/\\\\n}
+                targetline=${targetline//u0020/\\u0020}
+                targetline=${targetline//&/\\&}
+            fi
+            if [[ ! -z $targetline ]]; then
+                sed -i "s|^${originalline}$|${targetline}|g" $f
+            else
+                # Second find in all file
+                targetline=$(grep -rh "^$originalline" $TMP_DIR/)
+                targetline=${targetline//\\/\\\\\\}
+                targetline=${targetline//|/\\|}
+                targetline=${targetline//\\n/\\\\n}
+                targetline=${targetline//u0020/\\u0020}
+                targetline=${targetline//&/\\&}
+                targetline_count=$(grep -rh "^$originalline" $TMP_DIR/ | wc -l)
+                if [[ ! -z $targetline && "$targetline_count" -eq "1" ]]; then
+                    sed -i "s|^${originalline}$|${targetline}|g" $f
+                else
+                    echo "Can not translate $f>$originalline"
+                fi
+            fi
+        fi
+    done < $f
+done
+
+find $TARGET_LANG -name '*.properties' -exec sed -i "s/\\\\n/n/g" {} \;
+find $TARGET_LANG -name '*.properties' -exec sed -i "s/\\\\u/u/g" {} \;
+
+
+rm -rf $TMP_DIR
\ No newline at end of file