Make `sed` operations in `remove` only match exact column matches.

In order to make editing more precise, sed regular expressions only
match exact occurrences of the search string within each entry column.

In order to properly handle tab and space separators in a portable
manner, a set of global variables are included that provide strings
of those characters.

hosts

@@ -49,6 +49,17 @@ DEFAULT_COMMAND="${DEFAULT_COMMAND:-list}"
 # The path to the hosts file. This will almost always be /etc/hosts
 HOSTS_PATH="${HOSTS_PATH:-/etc/hosts}"
 
+# Space and tab for regular expressions
+#
+# sed regular expressions have slightly different behaviors dependending on
+# the environment, and POSIX [[:space:]] matches whitespace characters other
+# than just space and tab. These variables provide an easier, portable way to
+# test for just these two characters.
+_TAB_=$'\t'
+_SPACE_=$' '
+_TAB_SPACE_="${_TAB_}${_SPACE_}"
+_TAB_SPACE_CC_="[${_TAB_SPACE_}]"
+
 ###############################################################################
 # Debug
 ###############################################################################
@@ -872,8 +883,16 @@ remove() {
     $_ME help remove
     exit 1
   else
+    # Regular Expression Notes
+    #
+    # - Note double periods in regular expression in order to emulate /.+/,
+    #   which apparently doesn't work properly with all versions of sed.
     local target_records=$(
-      sed -n "s/^\(.*${search_string}.*\)$/\1/p" "${HOSTS_PATH}"
+      sed -n \
+        -e "s/^\(${search_string}[${_TAB_SPACE_}]..*\)$/\1/p" \
+        -e "s/^\(..*[${_TAB_SPACE_}]${search_string}[${_TAB_SPACE_}]..*\)$/\1/p" \
+        -e "s/^\(..*[${_TAB_SPACE_}]${search_string}\)$/\1/p" \
+        "${HOSTS_PATH}"
     )
 
     if [[ -z ${target_records:-} ]]
@@ -901,9 +920,18 @@ remove() {
         esac
       done
     fi
-    # -i '' - in place edit. BSD sed requires extension argument, for GNU
-    # it's optional. More info: http://stackoverflow.com/a/16746032
-    sed -i '' "/^.*${search_string}.*$/d" "${HOSTS_PATH}"
+    # Regular Expression Notes
+    #
+    # - -i '' - in place edit. BSD sed requires extension argument, for GNU
+    #   it's optional. More info: http://stackoverflow.com/a/16746032
+    #
+    # - Note double periods in regular expression in order to emulate /.+/,
+    #   which apparently doesn't work properly with all versions of sed.
+    sed -i '' \
+      -e "/^${search_string}[${_TAB_SPACE_}]..*$/d" \
+      -e "/^..*[${_TAB_SPACE_}]${search_string}[${_TAB_SPACE_}]..*$/d" \
+      -e "/^..*[${_TAB_SPACE_}]${search_string}$/d" \
+      "${HOSTS_PATH}"
     printf "Removed:\n%s\n" "${target_records}"
   fi
 }