fix: endless loop in utf8_len while input is an invalid utf8 string (#62)

Author: Gerrard-YNWA

.github/workflows/ci.yml

@@ -17,14 +17,19 @@ jobs:
         uses: actions/checkout@v2
         with:
           submodules: recursive
-      - name: Linux Get dependencies
-        run: |
-          sudo apt install -y build-essential libncurses5-dev libreadline-dev libssl-dev perl lua5.1 liblua5.1-0-dev
-          curl -fsSL https://raw.githubusercontent.com/apache/apisix/master/utils/linux-install-luarocks.sh | sh
 
-      - name: Linux Before install
-        run: |
-          sudo luarocks install luacheck
+      - name: Setup Lua
+        uses: leafo/[email protected]
+        with:
+          luaVersion: "5.1.5"
+
+      - name: Setup Luarocks
+        uses: leafo/gh-actions-luarocks@v4
+        with:
+          luarocksVersion: "3.8.0"
+
+      - name: Linux Get dependencies
+        run: sudo apt install -y build-essential libncurses5-dev libreadline-dev libssl-dev perl
 
       - name: Linux Install
         run: |
@@ -36,6 +41,6 @@ jobs:
 
       - name: Linux Script
         run: |
-          sudo luarocks install rockspec/jsonschema-master-0.rockspec
+          luarocks make rockspec/jsonschema-master-0.rockspec
           export PATH=$OPENRESTY_PREFIX/nginx/sbin:$PATH
           make test

Makefile

@@ -1,8 +1,8 @@
 OR_EXEC ?= $(shell which openresty)
 LUA_JIT_DIR ?= $(shell ${OR_EXEC} -V 2>&1 | grep prefix | grep -Eo 'prefix=(.*)/nginx\s+--' | grep -Eo '/.*/')luajit
 LUAROCKS_VER ?= $(shell luarocks --version | grep -E -o  "luarocks [0-9]+.")
-LUA_PATH ?= "./lib/?.lua;./deps/lib/lua/5.1/?.lua;./deps/share/lua/5.1/?.lua;;"
-LUA_CPATH ?= "./deps/lib/lua/5.1/?.so;;"
+LUA_PATH ?= ./lib/?.lua;./deps/lib/lua/5.1/?.lua;./deps/share/lua/5.1/?.lua;;
+LUA_CPATH ?= ./deps/lib/lua/5.1/?.so;;
 
 
 ### help:         Show Makefile rules.
@@ -27,11 +27,11 @@ endif
 
 ### test:         Run the test case
 test:
-	LUA_PATH=$(LUA_PATH) LUA_CPATH=$(LUA_CPATH) resty t/draft4.lua
-	LUA_PATH=$(LUA_PATH) LUA_CPATH=$(LUA_CPATH) resty t/draft6.lua
-	LUA_PATH=$(LUA_PATH) LUA_CPATH=$(LUA_CPATH) resty t/draft7.lua
-	LUA_PATH=$(LUA_PATH) LUA_CPATH=$(LUA_CPATH) resty t/default.lua
-	LUA_PATH=$(LUA_PATH) LUA_CPATH=$(LUA_CPATH) resty t/200more_variables.lua
+	LUA_PATH="$(LUA_PATH)" LUA_CPATH="$(LUA_CPATH)" resty t/draft4.lua
+	LUA_PATH="$(LUA_PATH)" LUA_CPATH="$(LUA_CPATH)" resty t/draft6.lua
+	LUA_PATH="$(LUA_PATH)" LUA_CPATH="$(LUA_CPATH)" resty t/draft7.lua
+	LUA_PATH="$(LUA_PATH)" LUA_CPATH="$(LUA_CPATH)" resty t/default.lua
+	LUA_PATH="$(LUA_PATH)" LUA_CPATH="$(LUA_CPATH)" resty t/200more_variables.lua
 
 
 ### clean:        Clean the test case

lib/jsonschema.lua

@@ -369,17 +369,60 @@ function validatorlib.tablekind(t)
   return 0 -- mixed array/object
 end
 
-function validatorlib.utf8_len(s)
-  local c, j=0, 1
-  while j <= #s do
-    local cb = str_byte(s, j)
-    if cb >= 0 and cb <= 127 then j = j + 1
-    elseif cb >= 192 and cb <= 223 then j = j + 2
-    elseif cb >= 224 and cb <= 239 then j = j + 3
-    elseif cb >= 240 and cb <= 247 then j = j + 4
+
+local accept_range = {
+  {lo = 0x80, hi = 0xBF},
+  {lo = 0xA0, hi = 0xBF},
+  {lo = 0x80, hi = 0x9F},
+  {lo = 0x90, hi = 0xBF},
+  {lo = 0x80, hi = 0x8F}
+}
+
+function validatorlib.utf8_len(str)
+  local i, n, c = 1, #str, 0
+  local first, byte, left_size, range_idx
+
+  while i <= n do
+    first = str_byte(str, i)
+    if first >= 0x80 and first <= 0xF4 then
+      left_size = 0
+      range_idx = 1
+      if first >= 0xC2 and first <= 0xDF then --2 bytes
+        left_size = 1
+      elseif first >= 0xE0 and first <= 0xEF then --3 bytes
+        left_size = 2
+        if first == 0xE0 then
+          range_idx = 2
+        elseif first == 0xED then
+          range_idx = 3
+        end
+      elseif first >= 0xF0 and first <= 0xF4 then --4 bytes
+        left_size = 3
+        if first == 0xF0 then
+          range_idx = 4
+        elseif first == 0xF4 then
+          range_idx = 5
+        end
+      end
+
+      if i + left_size > n then --invalid
+        left_size = 0
+      end
+
+      for j = 1, left_size do
+        byte = str_byte(str, i + j)
+        if byte < accept_range[range_idx].lo or byte > accept_range[range_idx].hi then --invalid
+          left_size = 0
+          break
+        end
+        range_idx = 1
+      end
+      i = i + left_size
     end
+    i = i + 1
     c = c + 1
   end
+
   return c
 end
 

t/default.lua

@@ -187,3 +187,29 @@ local t = {
 local ok = validator(t)
 assert(ok~=nil, "fail: failed to negative check of int64")
 ngx.say("passed: pass negative check of int64")
+
+----------------------------------------------------- test case 7
+-- check string len
+-- issue #61
+local cases = {
+    {"abcd", 4},
+    {"☺☻☹", 3},
+    {"1,2,3,4", 7},
+    {"\xff", 1},
+    {"\xc2\x80", 1},
+    {"\xe0\x00", 2},
+    {"\xe2\x80a", 3},
+    {"\xed\x80\x80", 1},
+    {"\xf0\x80", 2},
+    {"\xf4\x80", 2},
+}
+
+local schema = {}
+for i, case in ipairs(cases) do
+    schema.minLength = case[2]
+    schema.maxLength = case[2]
+    local validator = jsonschema.generate_validator(schema)
+    local ok, err = validator(case[1])
+    assert(ok, string.format("fail: validate case %d,  err: %s, ", i, err))
+end
+ngx.say("passed: check string len")