oga 0.2.0 → 0.2.1

data/ext/java/org/liboga/xml/Lexer.rl

@@ -101,6 +101,31 @@ public class Lexer extends RubyObject
         int pe    = data.length;
         int eof   = data.length;
 
+        String id_advance_line        = "advance_line";
+        String id_on_attribute        = "on_attribute";
+        String id_on_attribute_ns     = "on_attribute_ns";
+        String id_on_cdata            = "on_cdata";
+        String id_on_comment          = "on_comment";
+        String id_on_doctype_end      = "on_doctype_end";
+        String id_on_doctype_inline   = "on_doctype_inline";
+        String id_on_doctype_name     = "on_doctype_name";
+        String id_on_doctype_start    = "on_doctype_start";
+        String id_on_doctype_type     = "on_doctype_type";
+        String id_on_element_end      = "on_element_end";
+        String id_on_element_name     = "on_element_name";
+        String id_on_element_ns       = "on_element_ns";
+        String id_on_element_open_end = "on_element_open_end";
+        String id_on_element_start    = "on_element_start";
+        String id_on_proc_ins_end     = "on_proc_ins_end";
+        String id_on_proc_ins_name    = "on_proc_ins_name";
+        String id_on_proc_ins_start   = "on_proc_ins_start";
+        String id_on_string_body      = "on_string_body";
+        String id_on_string_dquote    = "on_string_dquote";
+        String id_on_string_squote    = "on_string_squote";
+        String id_on_text             = "on_text";
+        String id_on_xml_decl_end     = "on_xml_decl_end";
+        String id_on_xml_decl_start   = "on_xml_decl_start";
+
         %% write exec;
 
         this.lines = lines;
@@ -160,6 +185,17 @@ public class Lexer extends RubyObject
 
         this.callMethod(context, "advance_line", lines);
     }
+
+    /**
+     * Returns true if we're in an HTML script tag. See
+     * Oga::XML::Lexer#inside_html_script? for more information.
+     */
+    public Boolean inside_html_script_p()
+    {
+        ThreadContext context = this.runtime.getCurrentContext();
+
+        return this.callMethod(context, "inside_html_script?").isTrue();
+    }
 }
 
 %%{

data/ext/ragel/base_lexer.rl

@@ -28,6 +28,17 @@
     # When you call a method in Ruby make sure that said method is defined as
     # an instance method in the `Oga::XML::Lexer` class.
     #
+    # The name of the callback to invoke should be an identifier starting with
+    # "id_". The identifier should be defined in the associated C and Java code.
+    # In case of C code its value should be a Symbol as a ID object, for Java
+    # it should be a String. For example:
+    #
+    #     ID id_foo = rb_intern("foo");
+    #
+    # And for Java:
+    #
+    #     String id_foo = "foo";
+    #
     # ## Machine Transitions
     #
     # To transition from one machine to another always use `fnext` instead of
@@ -59,7 +70,7 @@
     comment       = comment_start (any* -- comment_end) comment_end;
 
     action start_comment {
-        callback("on_comment", data, encoding, ts + 4, te - 3);
+        callback(id_on_comment, data, encoding, ts + 4, te - 3);
     }
 
     # CDATA
@@ -75,7 +86,7 @@
     cdata       = cdata_start (any* -- cdata_end) cdata_end;
 
     action start_cdata {
-        callback("on_cdata", data, encoding, ts + 9, te - 3);
+        callback(id_on_cdata, data, encoding, ts + 9, te - 3);
     }
 
     # Processing Instructions
@@ -93,8 +104,8 @@
     proc_ins_end   = '?>';
 
     action start_proc_ins {
-        callback_simple("on_proc_ins_start");
-        callback("on_proc_ins_name", data, encoding, ts + 2, te);
+        callback_simple(id_on_proc_ins_start);
+        callback(id_on_proc_ins_name, data, encoding, ts + 2, te);
 
         mark = te;
 
@@ -103,8 +114,8 @@
 
     proc_ins_body := |*
         proc_ins_end => {
-            callback("on_text", data, encoding, mark, ts);
-            callback_simple("on_proc_ins_end");
+            callback(id_on_text, data, encoding, mark, ts);
+            callback_simple(id_on_proc_ins_end);
 
             mark = 0;
 
@@ -124,7 +135,7 @@
     squote = "'";
 
     action emit_string {
-        callback("on_string_body", data, encoding, ts, te);
+        callback(id_on_string_body, data, encoding, ts, te);
 
         if ( lines > 0 )
         {
@@ -135,13 +146,13 @@
     }
 
     action start_string_squote {
-        callback_simple("on_string_squote");
+        callback_simple(id_on_string_squote);
 
         fcall string_squote;
     }
 
     action start_string_dquote {
-        callback_simple("on_string_dquote");
+        callback_simple(id_on_string_dquote);
 
         fcall string_dquote;
     }
@@ -150,7 +161,7 @@
         ^squote* $count_newlines => emit_string;
 
         squote => {
-            callback_simple("on_string_squote");
+            callback_simple(id_on_string_squote);
 
             fret;
         };
@@ -160,7 +171,7 @@
         ^dquote* $count_newlines => emit_string;
 
         dquote => {
-            callback_simple("on_string_dquote");
+            callback_simple(id_on_string_dquote);
 
             fret;
         };
@@ -179,22 +190,35 @@
     doctype_start = '<!DOCTYPE'i whitespace+;
 
     action start_doctype {
-        callback_simple("on_doctype_start");
+        callback_simple(id_on_doctype_start);
         fnext doctype;
     }
 
+    # Machine for processing inline rules of a doctype.
+    doctype_inline := |*
+        ^']'* $count_newlines => {
+            callback(id_on_doctype_inline, data, encoding, ts, te);
+
+            if ( lines > 0 )
+            {
+                advance_line(lines);
+
+                lines = 0;
+            }
+        };
+
+        ']' => { fnext doctype; };
+    *|;
+
     # Machine for processing doctypes. Doctype values such as the public
     # and system IDs are treated as T_STRING tokens.
     doctype := |*
         'PUBLIC' | 'SYSTEM' => {
-            callback("on_doctype_type", data, encoding, ts, te);
+            callback(id_on_doctype_type, data, encoding, ts, te);
         };
 
-        # Consumes everything between the [ and ]. Due to the use of :> the ]
-        # is not consumed by any+.
-        '[' any+ :> ']' => {
-            callback("on_doctype_inline", data, encoding, ts + 1, te - 1);
-        };
+        # Starts a set of inline doctype rules.
+        '[' => { fnext doctype_inline; };
 
         # Lex the public/system IDs as regular strings.
         squote => start_string_squote;
@@ -205,11 +229,11 @@
         whitespace;
 
         identifier => {
-            callback("on_doctype_name", data, encoding, ts, te);
+            callback(id_on_doctype_name, data, encoding, ts, te);
         };
 
         '>' => {
-            callback_simple("on_doctype_end");
+            callback_simple(id_on_doctype_end);
             fnext main;
         };
     *|;
@@ -222,20 +246,20 @@
     xml_decl_end   = '?>';
 
     action start_xml_decl {
-        callback_simple("on_xml_decl_start");
+        callback_simple(id_on_xml_decl_start);
         fnext xml_decl;
     }
 
     # Machine that processes the contents of an XML declaration tag.
     xml_decl := |*
         xml_decl_end => {
-            callback_simple("on_xml_decl_end");
+            callback_simple(id_on_xml_decl_end);
             fnext main;
         };
 
         # Attributes and their values (e.g. version="1.0").
         identifier => {
-            callback("on_attribute", data, encoding, ts, te);
+            callback(id_on_attribute, data, encoding, ts, te);
         };
 
         squote => start_string_squote;
@@ -257,23 +281,23 @@
     element_end   = '</' identifier (':' identifier)* '>';
 
     action start_element {
-        callback_simple("on_element_start");
+        callback_simple(id_on_element_start);
         fhold;
         fnext element_name;
     }
 
     action close_element {
-        callback_simple("on_element_end");
+        callback_simple(id_on_element_end);
     }
 
     # Machine used for lexing the name/namespace of an element.
     element_name := |*
         identifier ':' => {
-            callback("on_element_ns", data, encoding, ts, te - 1);
+            callback(id_on_element_ns, data, encoding, ts, te - 1);
         };
 
         identifier => {
-            callback("on_element_name", data, encoding, ts, te);
+            callback(id_on_element_name, data, encoding, ts, te);
             fnext element_head;
         };
     *|;
@@ -284,16 +308,16 @@
         whitespace | '=';
 
         newline => {
-            callback_simple("advance_line");
+            callback_simple(id_advance_line);
         };
 
         # Attribute names and namespaces.
         identifier ':' => {
-            callback("on_attribute_ns", data, encoding, ts, te - 1);
+            callback(id_on_attribute_ns, data, encoding, ts, te - 1);
         };
 
         identifier => {
-            callback("on_attribute", data, encoding, ts, te);
+            callback(id_on_attribute, data, encoding, ts, te);
         };
 
         # Attribute values.
@@ -302,13 +326,23 @@
 
         # We're done with the open tag of the element.
         '>' => {
-            callback_simple("on_element_open_end");
-            fnext main;
+            callback_simple(id_on_element_open_end);
+
+            if ( inside_html_script_p() )
+            {
+                mark = ts + 1;
+
+                fnext script_text;
+            }
+            else
+            {
+                fnext main;
+            }
         };
 
         # Self closing tags.
         '/>' => {
-            callback_simple("on_element_end");
+            callback_simple(id_on_element_end);
             fnext main;
         };
     *|;
@@ -337,7 +371,7 @@
 
     text := |*
         terminate_text | allowed_text => {
-            callback("on_text", data, encoding, ts, te);
+            callback(id_on_text, data, encoding, ts, te);
 
             if ( lines > 0 )
             {
@@ -351,7 +385,7 @@
 
         # Text followed by a special tag, such as "foo<!--"
         allowed_text %{ mark = p; } terminate_text => {
-            callback("on_text", data, encoding, ts, mark);
+            callback(id_on_text, data, encoding, ts, mark);
 
             p    = mark - 1;
             mark = 0;
@@ -367,6 +401,30 @@
         };
     *|;
 
+    # <script> tags in HTML can contain basically anything except for the
+    # literal "</script>". As a result of this we can't use the regular text
+    # machine.
+    script_text := |*
+        '</script>' => {
+            callback(id_on_text, data, encoding, mark, ts);
+
+            mark = 0;
+
+            if ( lines > 0 )
+            {
+                advance_line(lines);
+
+                lines = 0;
+            }
+
+            callback_simple(id_on_element_end);
+
+            fnext main;
+        };
+
+        any $count_newlines;
+    *|;
+
     # The main machine aka the entry point of Ragel.
     main := |*
         doctype_start  => start_doctype;

data/lib/oga/css/parser.rb

@@ -301,38 +301,40 @@ module Oga
 ##### State transition tables begin ###
 
 racc_action_table = [
-    13,    41,    19,    43,    23,    13,    44,    45,    26,    25,
-    13,    40,    19,    13,    23,     9,    10,    11,    64,    48,
-     9,    10,    11,    20,    21,     9,    10,    11,    63,    62,
-    49,    65,    66,    20,    21,    13,    35,    19,    13,    23,
-    19,    24,    23,    19,    68,    23,    19,    68,    23,    68,
-     9,    10,    11,     9,    10,    11,    68,    68,    20,    21,
-    68,    20,    21,    13,    20,    21,    13,    20,    21,    19,
-    74,    23,    19,    75,    23,    19,    62,    23,     9,    10,
-    11,     9,    10,    11,    50,    51,    52,    53,    54,    55,
-    20,    21,    77,    20,    21,    62,    20,    21,    62 ]
+    13,    24,    19,    13,    23,    19,    25,    23,    26,    35,
+    13,    40,    19,    41,    23,     9,    10,    11,     9,    10,
+    11,    43,    44,    20,    21,    45,    20,    21,    13,    48,
+    19,    49,    23,    20,    21,    68,    68,    68,    13,    68,
+    19,    68,    23,     9,    10,    11,    64,    19,    68,    23,
+    74,    20,    21,     9,    10,    11,    63,    62,    75,    65,
+    66,    20,    21,    19,    13,    23,    62,    77,    20,    21,
+    19,    13,    23,    19,    62,    23,    62,   nil,   nil,     9,
+    10,    11,    13,   nil,    20,    21,     9,    10,    11,   nil,
+   nil,    20,    21,   nil,    20,    21,   nil,     9,    10,    11,
+    50,    51,    52,    53,    54,    55 ]
 
 racc_action_check = [
-     0,    21,     0,    22,     0,    11,    23,    24,     4,     3,
-    43,    20,    43,    19,    43,     0,     0,     0,    43,    35,
-    11,    11,    11,     0,     0,    43,    43,    43,    43,    43,
-    36,    43,    43,    43,    43,    25,    13,    25,    26,    25,
-    26,     1,    26,     5,    50,     5,     6,    51,     6,    52,
-    25,    25,    25,    26,    26,    26,    53,    54,    25,    25,
-    55,    26,    26,    10,     5,     5,     9,     6,     6,     7,
-    56,     7,    28,    57,    28,    29,    63,    29,    10,    10,
-    10,     9,     9,     9,    37,    37,    37,    37,    37,    37,
-     7,     7,    64,    28,    28,    75,    29,    29,    77 ]
+     0,     1,     0,    25,     0,    25,     3,    25,     4,    13,
+    19,    20,     5,    21,     5,     0,     0,     0,    25,    25,
+    25,    22,    23,     0,     0,    24,    25,    25,    26,    35,
+    26,    36,    26,     5,     5,    50,    51,    52,    43,    53,
+    43,    54,    43,    26,    26,    26,    43,     6,    55,     6,
+    56,    26,    26,    43,    43,    43,    43,    43,    57,    43,
+    43,    43,    43,     7,     9,     7,    63,    64,     6,     6,
+    28,    10,    28,    29,    75,    29,    77,   nil,   nil,     9,
+     9,     9,    11,   nil,     7,     7,    10,    10,    10,   nil,
+   nil,    28,    28,   nil,    29,    29,   nil,    11,    11,    11,
+    37,    37,    37,    37,    37,    37 ]
 
 racc_action_pointer = [
-    -2,    41,   nil,     2,     1,    39,    42,    65,   nil,    64,
-    61,     3,   nil,    33,   nil,   nil,   nil,   nil,   nil,    11,
-     9,    -1,    -5,     4,     7,    33,    36,   nil,    68,    71,
-   nil,   nil,   nil,   nil,   nil,    17,    25,    73,   nil,   nil,
-   nil,   nil,   nil,     8,   nil,   nil,   nil,   nil,   nil,   nil,
-    22,    25,    27,    34,    35,    38,    61,    53,   nil,   nil,
-   nil,   nil,   nil,    55,    72,   nil,   nil,   nil,   nil,   nil,
-   nil,   nil,   nil,   nil,   nil,    74,   nil,    77,   nil,   nil ]
+    -2,     1,   nil,    -1,     1,     8,    43,    59,   nil,    62,
+    69,    80,   nil,     6,   nil,   nil,   nil,   nil,   nil,     8,
+     9,    11,    13,    20,    25,     1,    26,   nil,    66,    69,
+   nil,   nil,   nil,   nil,   nil,    27,    26,    89,   nil,   nil,
+   nil,   nil,   nil,    36,   nil,   nil,   nil,   nil,   nil,   nil,
+    13,    14,    15,    17,    19,    26,    41,    38,   nil,   nil,
+   nil,   nil,   nil,    45,    47,   nil,   nil,   nil,   nil,   nil,
+   nil,   nil,   nil,   nil,   nil,    53,   nil,    55,   nil,   nil ]
 
 racc_action_default = [
     -2,   -59,    -1,    -3,    -4,    -7,    -8,   -10,   -12,   -16,
@@ -346,22 +348,22 @@ racc_action_default = [
 
 racc_goto_table = [
      3,    57,    28,    29,    27,    67,    69,    70,    71,    72,
-    73,    31,    31,    31,    32,    32,    32,    30,    33,    34,
-    36,    76,     4,    39,    37,    46,    47,    27,    27,    38,
-     1,    42,    56,    78,     2,    79,    58,    59,    60,   nil,
+    73,    32,    32,    32,    31,    31,    31,    30,    33,    34,
+     1,    76,     2,     4,    39,    46,    47,    27,    27,    36,
+    37,    38,    42,    78,    56,    79,    58,    59,    60,   nil,
    nil,   nil,   nil,    61 ]
 
 racc_goto_check = [
      3,    23,     5,     5,    11,    19,    19,    19,    19,    19,
-    19,     8,     8,     8,     7,     7,     7,     9,     9,     9,
-    16,    23,     4,    10,    17,     3,     3,    11,    11,    18,
-     1,    21,    22,    23,     2,    23,    24,    25,    26,   nil,
+    19,     7,     7,     7,     8,     8,     8,     9,     9,     9,
+     1,    23,     2,     4,    10,     3,     3,    11,    11,    16,
+    17,    18,    21,    23,    22,    23,    24,    25,    26,   nil,
    nil,   nil,   nil,     3 ]
 
 racc_goto_pointer = [
-   nil,    30,    34,     0,    22,    -4,   nil,     5,     2,     8,
-     4,    -1,   nil,   nil,   nil,   nil,     1,     5,    10,   -45,
-   nil,     9,   -11,   -42,    -7,    -6,    -5 ]
+   nil,    20,    22,     0,    23,    -4,   nil,     2,     5,     8,
+     5,    -1,   nil,   nil,   nil,   nil,    10,    11,    12,   -45,
+   nil,    10,    -9,   -42,    -7,    -6,    -5 ]
 
 racc_goto_default = [
    nil,   nil,   nil,   nil,   nil,     5,     6,     7,     8,   nil,

data/lib/oga/version.rb

@@ -1,3 +1,3 @@
 module Oga
-  VERSION = '0.2.0'
+  VERSION = '0.2.1'
 end # Oga