unidecode for mvregex

files/mvregex

@@ -11,6 +11,7 @@ function helpexit() {
     echo '    -f to replace match [arg1] with format [arg2], ex: -f "[0-9]\+" "%04d"'
     echo '    -p to replace problematic characters [^\w()[]-.] with [arg1]'
     echo '    -a to replace non-alphanumeric(+dot) characters [^\w.] with [arg1]'
+    echo '    -u Unidecode to nearest ascii representation'
     
     exit
 }
@@ -41,6 +42,10 @@ do  [[ "${!i}" = "-h" ]] && helpexit
             MODE=printf
             continue
         }
+    [[ "${!i}" = "-u" ]] && {
+            MODE=unidecode
+            continue
+        }
     [[ "${!i}" = "-p" ]] && {
             SRC='[^]\[0-9a-zA-Z_.()-]'
             continue
@@ -122,6 +127,35 @@ if [ "$MODE" = "printf" ]; then
     exit
 fi
 
+
+if [ "$MODE" = "unidecode" ]; then
+    set -e
+    function decode() {
+      python3 -c "
+import sys, unidecode
+sys.stdout.write(unidecode.unidecode(sys.argv[1]))
+      " "$1"
+    }
+    for file in $( eval $LS ); do [[ ${#file} -gt ${LONGEST} ]] && LONGEST=${#file}; done
+    printf "%-${LONGEST}s\t|%s\n" Matching Replaced
+    for file in $( eval $LS ); do
+       target_file=$( decode "$file" )
+       if [[ ! "$file" = "$target_file" ]]; then
+          printf "%-${LONGEST}s\t|%s\n" "$file" "$target_file"
+          c=$(( $c + 1 ))
+       fi
+    done
+    echo $c' matches. Sure?'
+    read i
+    for file in $( eval $LS ); do
+       target_file=$( decode "$file" )
+       if [[ ! "$file" = "$target_file" ]]; then
+         mv -iv -- "$file" "$target_file"
+       fi
+    done
+    exit
+fi
+
 ## Normal regex replace 
 [[ -z "$SRC" ]] && helpexit
 for file in $( eval $LS | grep -- "$SRC"); do [[ ${#file} -gt ${LONGEST} ]] && LONGEST=${#file}; done