Add unix/text-to-utf8.sh

1 files changed, 40 insertions, 0 deletions

diff --git a/unix/text-to-utf8.sh b/unix/text-to-utf8.sh
new file mode 100755
index 0000000..b48a6c7
--- /dev/null
+++ b/unix/text-to-utf8.sh
@@ -0,0 +1,40 @@
+#!/bin/sh
+#
+# Convert text file to UTF-8 encoding.
+#
+# Aaron LI
+# 2018-01-17
+#
+
+case "$1" in
+    ""|-h|--help)
+        echo "usage: ${0##*/} <file1> ..."
+        exit 1;
+        ;;
+esac
+
+convert() {
+    from="$1"
+    infile="$2"
+    iconv -f $from -t UTF-8 "$infile" > _tmp.$$
+    if [ $? -eq 0 ]; then
+        mv _tmp.$$ "$infile"
+        echo "converted to UTF-8 from $from"
+    else
+        rm _tmp.$$
+        echo "ERROR: failed conversion from $from"
+    fi
+}
+
+while [ -n "$1" ]; do
+    infile="$1"
+    shift;
+    echo "=== $infile ==="
+    if file "$infile" | grep -q "UTF-8 Unicode"; then
+        echo "already UTF-8; skip it"
+    elif file "$infile" | grep -q "UTF-16 Unicode"; then
+        convert UTF-16 "$infile"
+    else
+        convert GBK "$infile"
+    fi
+done