Experimental parser: add everything_else class

tools/lexer_generator/code_generator.jinja

 #include "lexer/even-more-experimental-scanner.h"
 
 {%- macro do_key(key) -%}
   {%- for r in key -%}
     {%- if not loop.first %} || {% endif -%}
     {%- if r[0] == 'LATIN_1' -%}
       {%- if r[1][0] == r[1][1] -%}
         yych == {{r[1][0]}}
       {%- elif r[1][0] == 0 -%}
         yych <= {{r[1][1]}}
       {%- elif r[1][1] == 255 and encoding == 'latin1'-%}
         yych >= {{r[1][0]}}
       {%- else -%}
         ({{r[1][0]}} <= yych && yych <= {{r[1][1]}})
       {%- endif -%}
     {%- elif r[0] == 'CLASS' -%}
       {%- if r[1] == 'eos' -%}
         (yych == 0 && cursor_ >= buffer_end_)
       {%- elif r[1] == 'zero' -%}
         (yych == 0 && cursor_ < buffer_end_)
-      {%- elif r[1] == 'byte_order_mark' and encoding == 'utf16'-%}
-        (yych == 0xfffe || yych == 0xfeff)
-      {%- elif r[1] == 'non_latin_1_whitespace' and encoding == 'utf16'-%}
-        {# FIXME: Add and use unicode_cache_->InNonAsciiWhitespace #}
-        (yych > 255 && unicode_cache_->IsWhiteSpace(yych))
-      {%- elif r[1] == 'non_latin_1_letter' and encoding == 'utf16'-%}
-        {# FIXME: Add and use unicode_cache_->InNonAsciiLetter #}
-        (yych > 255 &&  unicode_cache_->IsLetter(yych))
-      {%- elif r[1] == 'non_latin1_identifier_part_not_letter' and encoding == 'utf16'-%}
-        (yych > 255 &&  unicode_cache_->IsIdentifierPartNotLetter(yych))
-      {%- elif r[1] == 'non_latin1_line_terminator' and encoding == 'utf16'-%}
-        (yych > 255 &&  unicode_cache_->IsLineTerminator(yych))
+      {%- elif encoding == 'latin1' -%}
+        false /* {{r[1]}} */
+      {%- elif encoding == 'utf16' -%}
+        {%- if r[1] == 'byte_order_mark' -%}
+          (yych == 0xfffe || yych == 0xfeff)
+        {%- elif r[1] == 'non_latin_1_whitespace' -%}
+          {# FIXME: Add and use unicode_cache_->InNonAsciiWhitespace #}
+          (yych > 255 && unicode_cache_->IsWhiteSpace(yych))
+        {%- elif r[1] == 'non_latin_1_letter' -%}
+          {# FIXME: Add and use unicode_cache_->InNonAsciiLetter #}
+          (yych > 255 &&  unicode_cache_->IsLetter(yych))
+        {%- elif r[1] == 'non_latin_1_identifier_part_not_letter' -%}
+          (yych > 255 &&  unicode_cache_->IsIdentifierPartNotLetter(yych))
+        {%- elif r[1] == 'non_latin_1_line_terminator' -%}
+          (yych > 255 &&  unicode_cache_->IsLineTerminator(yych))
+        {%- elif r[1] == 'non_latin_1_everything_else' -%}
+          {# FIXME: Optimize this away #}
+          (yych > 255 &&
+           !unicode_cache_->IsWhiteSpace(yych) &&
+           !unicode_cache_->IsLetter(yych) &&
+           !unicode_cache_->IsIdentifierPartNotLetter(yych) &&
+           !unicode_cache_->IsLineTerminator(yych))
+        {%- else %}
+          uncompilable code for {{encoding}} {{r[0]}} {{r[1]}}
+        {%- endif -%}
       {%- else -%}
-        false /* {{r[1]}} */
+        uncompilable code for {{encoding}} {{r[0]}} {{r[1]}}
       {%- endif -%}
     {%- else -%}
-      false
+      uncompilable code for {{encoding}} {{r[0]}} {{r[1]}}
     {%- endif -%}
   {%- endfor -%}
 {%- endmacro -%}
 
 
 {% macro dispatch_action(type, value) -%}
   {%- if type == 'code' %}
     {{value}}
   {% elif type == 'terminate' %}
     PUSH_EOS();
@@ skipping 71 matching lines @@
     {% endfor -%}
     }
   {%- endif -%}
 
   {%- for key, transition_state_id in state.transitions %}
     if ({{do_key(key)}}) {
       {{ do_transition(transition_state_id) }}
     }
   {% endfor -%}
 
+  {%- for key, transition_state_id in state['deferred_transitions'] %}
+    if ({{do_key(key)}}) { // deferred transition
+      {{ do_transition(transition_state_id) }}
+    }
+  {% endfor -%}
+
   {%- set match_action = state.match_action -%}
 
   {%- if match_action %}
     {{ dispatch_action(match_action[0], match_action[1]) }}
     goto code_start;
   {% else %}
     goto default_action;
   {%- endif %}
 
 {%- endmacro %}
@@ skipping 67 matching lines @@
 {% endif -%}
     {{dispatch_action(default_action[0], default_action[1])}}
   FORWARD();
   goto code_start;
 
   fell_through:
   CHECK(false);
 }
 } }