C extensin for Variable#lax_parse

ext/liquid/liquid_ext.c

@@ -12,4 +12,5 @@ void Init_liquid(void)
 
     init_liquid_tokenizer();
     init_liquid_block();
+    init_liquid_variable();
 }

ext/liquid/liquid_ext.h

@@ -8,6 +8,7 @@
 #include "tokenizer.h"
 #include "block.h"
 #include "utils.h"
+#include "variable.h"
 
 extern VALUE mLiquid;
 extern VALUE cLiquidTemplate, cLiquidTag, cLiquidVariable;

ext/liquid/variable.c

@@ -0,0 +1,179 @@
+#include "liquid_ext.h"
+
+VALUE cLiquidVariable;
+extern VALUE mLiquid;
+
+static void free_variable(void *ptr)
+{
+    struct liquid_variable *variable = ptr;
+    xfree(variable);
+}
+
+static VALUE rb_variable_allocate(VALUE klass)
+{
+    VALUE obj;
+    struct liquid_variable *variable;
+
+    obj = Data_Make_Struct(klass, struct liquid_variable, NULL, free_variable, variable);
+    return obj;
+}
+
+static inline int skip_whitespace(char * str, int len) 
+{
+  int skipped = 0; char * ptr = str;
+  while (skipped < len && isspace(*ptr))
+    {skipped++; ptr++;}
+  return skipped;
+}
+
+static char * get_quoted_fragment(char * str, int len, int * ret_size, int * end_offset, bool colon) 
+{
+  int p = 0; /* Current position in string */
+  int start = -1, end = -1; /* Start and end indices for the found string */
+  char quoted_by = -1; /* Is the current part of string quoted by a single or double quote? If so 
+                          ignore any special chars */
+
+  while (p < len) {
+    
+    switch (str[p]) {
+      case '"': 
+        if (start == -1) {start = p; quoted_by = '"';}
+        else if (str[start] == '"') {end = p; goto quoted_fragment_found;}
+        else if (quoted_by == -1) quoted_by = '"';
+        else if (quoted_by == '"') quoted_by = -1;
+        break;
+      case '\'':
+        if (start == -1) {start = p; quoted_by = '\'';}
+        else if (str[start] == '\'') {end = p; goto quoted_fragment_found;}
+        else if (quoted_by == -1) quoted_by = '\'';
+        else if (quoted_by == '\'') quoted_by = -1;
+        break;
+      case ':': 
+        if (colon)
+          if (start != -1 && quoted_by == -1) {end = p-1; goto quoted_fragment_found;} 
+        else
+          if (start == -1) start = p;
+        break;
+      case '|':
+      case ',':
+      case '\n':
+      case '\r':
+      case '\f':
+      case '\t':
+      case '\v':
+      case ' ': 
+        if (start != -1 && quoted_by == -1) {end = p-1; goto quoted_fragment_found;} 
+        break;
+      default: 
+        if (start == -1) start = p; 
+        break;
+    }
+    p++;
+  }
+  if (p == len && start != -1 && end == -1) end = len-1;
+
+quoted_fragment_found:
+  if (end >= start) {
+    *ret_size = end-start+1;
+    *end_offset = end+1;
+    return &str[start];
+  } else {
+    *ret_size = 0;
+    return NULL;
+  }
+}
+
+static VALUE get_filters(char * str, int len, VALUE self) {
+  VALUE filters_arr = rb_ary_new(); 
+
+  int p = 0; 
+  int ret_size, end_offset; 
+  char * f;
+
+  while(p<len) {
+    if (str[p] == '|') {
+      VALUE filter = rb_ary_new();
+      VALUE f_args = rb_ary_new();
+
+      p += skip_whitespace(&str[p+1], len-p-1);
+      f = get_quoted_fragment(&str[p], len-p, &ret_size, &end_offset, true);
+      p += end_offset;
+
+      if (f) {
+        if (f[ret_size-1] == ':') ret_size--;
+        rb_ary_push(filter, rb_str_new(f, ret_size));
+      }
+
+      /* Check for filter arguments */
+      do {
+        if (p<len) {
+          p += skip_whitespace(&str[p], len-p);
+
+          // printf("\n1. %.*s\n", len-p, &str[p]);
+
+          if (str[p] != '|') {
+            f = get_quoted_fragment(&str[p], len-p, &ret_size, &end_offset, false);
+
+            // printf("\n2. %.*s\n", ret_size, f);
+            
+            p += end_offset;
+            p += skip_whitespace(&str[p], len-p);
+
+            if (str[p] == '|') p--;
+
+            if (f) rb_ary_push(f_args, rb_str_new(f, ret_size));
+          
+          } else p--;
+        }
+
+      } while (str[p] == ',' || str[p] == ':');
+
+      rb_ary_push(filter, f_args);
+
+      /* Add to filters_arr array */
+      rb_ary_push(filters_arr, filter);
+    }
+    p++;
+  }
+  return filters_arr;
+}
+
+static VALUE rb_variable_lax_parse(VALUE self, VALUE m) 
+{
+  char * markup = RSTRING_PTR(m);
+  int markup_len = RSTRING_LEN(m);
+
+  char * cursor = markup; int cursor_pos = 0; 
+  VALUE filters_arr; 
+  int size, end_offset;
+
+  /* Extract name */
+  cursor_pos += skip_whitespace(markup, markup_len); 
+  cursor = markup + cursor_pos;
+  cursor = get_quoted_fragment(cursor, markup_len - cursor_pos, &size, &end_offset, false);
+
+  if (cursor == NULL) {
+    rb_iv_set(self, "@name", Qnil);
+    filters_arr = rb_ary_new(); 
+    rb_iv_set(self, "@filters", filters_arr);
+  }
+  else 
+  {
+    rb_iv_set(self, "@name", rb_str_new(cursor, size));
+
+    /* Extract filters */
+    if (end_offset < markup_len) {
+      cursor = &markup[end_offset];
+      filters_arr = get_filters(cursor, markup_len - end_offset, self);
+      rb_iv_set(self, "@filters", filters_arr);
+    }
+  }
+  return filters_arr;
+}
+
+void init_liquid_variable()
+{
+    cLiquidVariable = rb_define_class_under(mLiquid, "Variable", rb_cObject);
+    rb_define_alloc_func(cLiquidVariable, rb_variable_allocate);
+    rb_define_method(cLiquidVariable, "lax_parse", rb_variable_lax_parse, 1);
+}

ext/liquid/variable.h

@@ -0,0 +1,13 @@
+#ifndef LIQUID_VARIABLE_H
+#define LIQUID_VARIABLE_H
+
+#include <regex.h>
+
+struct liquid_variable {
+    char *markup; long markup_len;
+    char *name; long name_len;
+};
+
+void init_liquid_variable();
+
+#endif

lib/liquid/variable.rb

@@ -11,7 +11,6 @@ module Liquid
   #   {{ user | link }}
   #
   class Variable
-    FilterParser = /(?:#{FilterSeparator}|(?:\s*(?:#{QuotedFragment}|#{ArgumentSeparator})\s*)+)/o
     EasyParse = /\A *(\w+(?:\.\w+)*) *\z/
     attr_accessor :filters, :name, :warnings
 
@@ -35,22 +34,22 @@ module Liquid
       end
     end
 
-    def lax_parse(markup)
+#     def lax_parse(markup)
-      @filters = []
+#       @filters = []
-      if match = markup.match(/\s*(#{QuotedFragment})(.*)/o)
+#       if match = markup.match(/\s*(#{QuotedFragment})(.*)/o)
-        @name = match[1]
+#         @name = match[1]
-        if match[2].match(/#{FilterSeparator}\s*(.*)/o)
+#         if match[2].match(/#{FilterSeparator}\s*(.*)/o)
-          filters = Regexp.last_match(1).scan(FilterParser)
+#           filters = Regexp.last_match(1).scan(FilterParser)
-          filters.each do |f|
+#           filters.each do |f|
-            if matches = f.match(/\s*(\w+)/)
+#             if matches = f.match(/\s*(\w+)/)
-              filtername = matches[1]
+#               filtername = matches[1]
-              filterargs = f.scan(/(?:#{FilterArgumentSeparator}|#{ArgumentSeparator})\s*((?:\w+\s*\:\s*)?#{QuotedFragment})/o).flatten
+#               filterargs = f.scan(/(?:#{FilterArgumentSeparator}|#{ArgumentSeparator})\s*((?:\w+\s*\:\s*)?#{QuotedFragment})/o).flatten
-              @filters << [filtername, filterargs]
+#               @filters << [filtername, filterargs]
-            end
+#             end
-          end
+#           end
-        end
+#         end
-      end
+#       end
-    end
+#     end
 
     def strict_parse(markup)
       # Very simple valid cases

test/liquid/variable_test.rb

@@ -51,11 +51,9 @@ class VariableTest < Test::Unit::TestCase
   end
 
   def test_filter_with_date_parameter
-
     var = Variable.new(%! '2006-06-06' | date: "%m/%d/%Y"!)
     assert_equal "'2006-06-06'", var.name
     assert_equal [["date",["\"%m/%d/%Y\""]]], var.filters
-
   end
 
   def test_filters_without_whitespace
@@ -73,7 +71,7 @@ class VariableTest < Test::Unit::TestCase
   end
 
   def test_symbol
-    var = Variable.new("http://disney.com/logo.gif | image: 'med' ", :error_mode => :lax)
+    var = Variable.new("http://disney.com/logo.gif | image: 'med'", :error_mode => :lax)
     assert_equal "http://disney.com/logo.gif", var.name
     assert_equal [["image",["'med'"]]], var.filters
   end