added utf8 character statistics counter. utf8_count

1 files changed, 45 insertions, 0 deletions

diff --git a/utf8_count/utf8_count.lua b/utf8_count/utf8_count.lua
new file mode 100644
index 0000000..a196e9a
--- /dev/null
+++ b/utf8_count/utf8_count.lua
@@ -0,0 +1,45 @@
+#!/usr/bin/lua
+
+--check if version is supported
+--why 5.3? utf8 and 64bit support
+if (_VERSION ~= "Lua 5.3" ) then
+	print "Wrong lua versions"
+	print "Supported 5.3 only"
+	os.exit(1)
+end
+
+--open file
+utf8_file = io.open( "test/teabook.txt", "r" )
+if ( utf8_file == nil) then
+	print "Couldnot open file"
+	os.exit(1)
+end
+
+--read contect in utf8
+--l = utf8_file:read("*l")
+hier_table = {}
+for line in utf8_file:lines() do
+	for p,c in utf8.codes( line ) do
+		if hier_table[c] == nil then
+			hier_table[c] = 1
+		else
+			hier_table[c] = hier_table[c] + 1
+		end
+	end
+end
+
+hier_table_sorted = {}
+for k,v in pairs(hier_table) do
+	table.insert( hier_table_sorted, {ch=k,val=v})
+end
+
+function cmpa( a,b )
+	if a.val ~= nil and b.val ~= nil then 
+		return a.val > b.val
+	end
+end
+table.sort( hier_table_sorted, cmpa )
+for k,v in ipairs(hier_table_sorted) do
+	print(utf8.char(v.ch), "=",hier_table[v.ch] )
+end
+