feature: support unicode length validation (#15)

Author: Gerrard-YNWA

lib/jsonschema.lua

@@ -395,6 +395,22 @@ local function to_lua_code(var)
   return code .. "}"
 end
 
+local function utf8_len_func(ctx)
+  return sformat([[function(s)
+      local c, j=0, 1
+      while j <= #s do
+        local cb = %s(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
+        end
+        c = c + 1
+      end
+      return c
+    end]], ctx:libfunc("string.byte"))
+end
+
 generate_validator = function(ctx, schema)
   -- get type informations as they will be necessary anyway
   local datatype = ctx:localvar(sformat('%s(%s)',
@@ -747,15 +763,19 @@ generate_validator = function(ctx, schema)
   if schema.minLength or schema.maxLength or schema.pattern then
     ctx:stmt(sformat('if %s == "string" then', datatype))
     if schema.minLength then
-      ctx:stmt(sformat('  if #%s < %d then', ctx:param(1), schema.minLength))
-      ctx:stmt(sformat('    return false, %s("string too short, expected at least %d, got %%d", #%s)',
-                       ctx:libfunc('string.format'), schema.minLength, ctx:param(1)))
+      ctx:stmt(sformat('  local utf8_len_func = %s', utf8_len_func(ctx)))
+      ctx:stmt(sformat('  local c = utf8_len_func(%s)',ctx:param(1)))
+      ctx:stmt(sformat('  if c < %d then', schema.minLength))
+      ctx:stmt(sformat('    return false, %s("string too short, expected at least %d, got ") ..c',
+                       ctx:libfunc('string.format'), schema.minLength))
       ctx:stmt(        '  end')
     end
     if schema.maxLength then
-      ctx:stmt(sformat('  if #%s > %d then', ctx:param(1), schema.maxLength))
-      ctx:stmt(sformat('    return false, %s("string too long, expected at most %d, got %%d", #%s)',
-                       ctx:libfunc('string.format'), schema.maxLength, ctx:param(1)))
+      ctx:stmt(sformat('  local utf8_len_func = %s', utf8_len_func(ctx)))
+      ctx:stmt(sformat('  local c = utf8_len_func(%s)',ctx:param(1)))
+      ctx:stmt(sformat('  if c > %d then', schema.maxLength))
+      ctx:stmt(sformat('    return false, %s("string too long, expected at most %d, got ") .. c',
+                       ctx:libfunc('string.format'), schema.maxLength))
       ctx:stmt(        '  end')
     end
     if schema.pattern then

t/draft4.lua

@@ -25,12 +25,6 @@ local blacklist = {
   ['regexes are not anchored by default and are case sensitive'] = {
     ['recognized members are accounted for'] = true, -- uses a unsupported pattern construct
   },
-  ['minLength validation'] = {
-    ['one supplementary Unicode code point is not long enough'] = true, -- unicode handling
-  },
-  ['maxLength validation'] = {
-    ['two supplementary Unicode code points is long enough'] = true, -- unicode handling
-  },
   ['required validation'] = {
     ['ignores arrays'] = true
   },

t/draft6.lua

@@ -25,12 +25,6 @@ local blacklist = {
   ['regexes are not anchored by default and are case sensitive'] = {
     ['recognized members are accounted for'] = true, -- uses a unsupported pattern construct
   },
-  ['minLength validation'] = {
-    ['one supplementary Unicode code point is not long enough'] = true, -- unicode handling
-  },
-  ['maxLength validation'] = {
-    ['two supplementary Unicode code points is long enough'] = true, -- unicode handling
-  },
   ['required validation'] = {
     ['ignores arrays'] = true
   },

t/draft7.lua

@@ -25,12 +25,6 @@ local blacklist = {
   ['regexes are not anchored by default and are case sensitive'] = {
     ['recognized members are accounted for'] = true, -- uses a unsupported pattern construct
   },
-  ['minLength validation'] = {
-    ['one supplementary Unicode code point is not long enough'] = true, -- unicode handling
-  },
-  ['maxLength validation'] = {
-    ['two supplementary Unicode code points is long enough'] = true, -- unicode handling
-  },
   ['required validation'] = {
     ['ignores arrays'] = true
   },