/[suikacvs]/markup/html/whatpm/Whatpm/HTML.pm.src
Suika

Diff of /markup/html/whatpm/Whatpm/HTML.pm.src

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 1.186 by wakaba, Sat Sep 20 07:54:47 2008 UTC revision 1.187 by wakaba, Sat Sep 20 09:28:31 2008 UTC
# Line 966  sub _initialize_tokenizer ($) { Line 966  sub _initialize_tokenizer ($) {
966  ## TODO: Polytheistic slash SHOULD NOT be used. (Applied only to atheists.)  ## TODO: Polytheistic slash SHOULD NOT be used. (Applied only to atheists.)
967  ## (This requirement was dropped from HTML5 spec, unfortunately.)  ## (This requirement was dropped from HTML5 spec, unfortunately.)
968    
969    my $is_space = {
970      0x0009 => 1, # CHARACTER TABULATION (HT)
971      0x000A => 1, # LINE FEED (LF)
972      #0x000B => 0, # LINE TABULATION (VT)
973      0x000C => 1, # FORM FEED (FF)
974      #0x000D => 1, # CARRIAGE RETURN (CR)
975      0x0020 => 1, # SPACE (SP)
976    };
977    
978  sub _get_next_token ($) {  sub _get_next_token ($) {
979    my $self = shift;    my $self = shift;
980    
# Line 1336  sub _get_next_token ($) { Line 1345  sub _get_next_token ($) {
1345            redo A;            redo A;
1346          }          }
1347        } else { # after "<{tag-name}"        } else { # after "<{tag-name}"
1348          unless ({          unless ($is_space->{$self->{nc}} or
1349                   0x0009 => 1, # HT                  {
                  0x000A => 1, # LF  
                  0x000B => 1, # VT  
                  0x000C => 1, # FF  
                  0x0020 => 1, # SP  
1350                   0x003E => 1, # >                   0x003E => 1, # >
1351                   0x002F => 1, # /                   0x002F => 1, # /
1352                   -1 => 1, # EOF                   -1 => 1, # EOF
# Line 1368  sub _get_next_token ($) { Line 1373  sub _get_next_token ($) {
1373          }          }
1374        }        }
1375      } elsif ($self->{state} == TAG_NAME_STATE) {      } elsif ($self->{state} == TAG_NAME_STATE) {
1376        if ($self->{nc} == 0x0009 or # HT        if ($is_space->{$self->{nc}}) {
           $self->{nc} == 0x000A or # LF  
           $self->{nc} == 0x000B or # VT  
           $self->{nc} == 0x000C or # FF  
           $self->{nc} == 0x0020) { # SP  
1377          !!!cp (34);          !!!cp (34);
1378          $self->{state} = BEFORE_ATTRIBUTE_NAME_STATE;          $self->{state} = BEFORE_ATTRIBUTE_NAME_STATE;
1379          !!!next-input-character;          !!!next-input-character;
# Line 1444  sub _get_next_token ($) { Line 1445  sub _get_next_token ($) {
1445          redo A;          redo A;
1446        }        }
1447      } elsif ($self->{state} == BEFORE_ATTRIBUTE_NAME_STATE) {      } elsif ($self->{state} == BEFORE_ATTRIBUTE_NAME_STATE) {
1448        if ($self->{nc} == 0x0009 or # HT        if ($is_space->{$self->{nc}}) {
           $self->{nc} == 0x000A or # LF  
           $self->{nc} == 0x000B or # VT  
           $self->{nc} == 0x000C or # FF  
           $self->{nc} == 0x0020) { # SP  
1449          !!!cp (45);          !!!cp (45);
1450          ## Stay in the state          ## Stay in the state
1451          !!!next-input-character;          !!!next-input-character;
# Line 1544  sub _get_next_token ($) { Line 1541  sub _get_next_token ($) {
1541          }          }
1542        }; # $before_leave        }; # $before_leave
1543    
1544        if ($self->{nc} == 0x0009 or # HT        if ($is_space->{$self->{nc}}) {
           $self->{nc} == 0x000A or # LF  
           $self->{nc} == 0x000B or # VT  
           $self->{nc} == 0x000C or # FF  
           $self->{nc} == 0x0020) { # SP  
1545          !!!cp (59);          !!!cp (59);
1546          $before_leave->();          $before_leave->();
1547          $self->{state} = AFTER_ATTRIBUTE_NAME_STATE;          $self->{state} = AFTER_ATTRIBUTE_NAME_STATE;
# Line 1631  sub _get_next_token ($) { Line 1624  sub _get_next_token ($) {
1624          redo A;          redo A;
1625        }        }
1626      } elsif ($self->{state} == AFTER_ATTRIBUTE_NAME_STATE) {      } elsif ($self->{state} == AFTER_ATTRIBUTE_NAME_STATE) {
1627        if ($self->{nc} == 0x0009 or # HT        if ($is_space->{$self->{nc}}) {
           $self->{nc} == 0x000A or # LF  
           $self->{nc} == 0x000B or # VT  
           $self->{nc} == 0x000C or # FF  
           $self->{nc} == 0x0020) { # SP  
1628          !!!cp (71);          !!!cp (71);
1629          ## Stay in the state          ## Stay in the state
1630          !!!next-input-character;          !!!next-input-character;
# Line 1722  sub _get_next_token ($) { Line 1711  sub _get_next_token ($) {
1711          redo A;                  redo A;        
1712        }        }
1713      } elsif ($self->{state} == BEFORE_ATTRIBUTE_VALUE_STATE) {      } elsif ($self->{state} == BEFORE_ATTRIBUTE_VALUE_STATE) {
1714        if ($self->{nc} == 0x0009 or # HT        if ($is_space->{$self->{nc}}) {
           $self->{nc} == 0x000A or # LF  
           $self->{nc} == 0x000B or # VT  
           $self->{nc} == 0x000C or # FF  
           $self->{nc} == 0x0020) { # SP        
1715          !!!cp (83);          !!!cp (83);
1716          ## Stay in the state          ## Stay in the state
1717          !!!next-input-character;          !!!next-input-character;
# Line 1907  sub _get_next_token ($) { Line 1892  sub _get_next_token ($) {
1892          redo A;          redo A;
1893        }        }
1894      } elsif ($self->{state} == ATTRIBUTE_VALUE_UNQUOTED_STATE) {      } elsif ($self->{state} == ATTRIBUTE_VALUE_UNQUOTED_STATE) {
1895        if ($self->{nc} == 0x0009 or # HT        if ($is_space->{$self->{nc}}) {
           $self->{nc} == 0x000A or # LF  
           $self->{nc} == 0x000B or # HT  
           $self->{nc} == 0x000C or # FF  
           $self->{nc} == 0x0020) { # SP  
1896          !!!cp (107);          !!!cp (107);
1897          $self->{state} = BEFORE_ATTRIBUTE_NAME_STATE;          $self->{state} = BEFORE_ATTRIBUTE_NAME_STATE;
1898          !!!next-input-character;          !!!next-input-character;
# Line 1993  sub _get_next_token ($) { Line 1974  sub _get_next_token ($) {
1974          redo A;          redo A;
1975        }        }
1976      } elsif ($self->{state} == AFTER_ATTRIBUTE_VALUE_QUOTED_STATE) {      } elsif ($self->{state} == AFTER_ATTRIBUTE_VALUE_QUOTED_STATE) {
1977        if ($self->{nc} == 0x0009 or # HT        if ($is_space->{$self->{nc}}) {
           $self->{nc} == 0x000A or # LF  
           $self->{nc} == 0x000B or # VT  
           $self->{nc} == 0x000C or # FF  
           $self->{nc} == 0x0020) { # SP  
1978          !!!cp (118);          !!!cp (118);
1979          $self->{state} = BEFORE_ATTRIBUTE_NAME_STATE;          $self->{state} = BEFORE_ATTRIBUTE_NAME_STATE;
1980          !!!next-input-character;          !!!next-input-character;
# Line 2438  sub _get_next_token ($) { Line 2415  sub _get_next_token ($) {
2415          redo A;          redo A;
2416        }        }
2417      } elsif ($self->{state} == DOCTYPE_STATE) {      } elsif ($self->{state} == DOCTYPE_STATE) {
2418        if ($self->{nc} == 0x0009 or # HT        if ($is_space->{$self->{nc}}) {
           $self->{nc} == 0x000A or # LF  
           $self->{nc} == 0x000B or # VT  
           $self->{nc} == 0x000C or # FF  
           $self->{nc} == 0x0020) { # SP  
2419          !!!cp (155);          !!!cp (155);
2420          $self->{state} = BEFORE_DOCTYPE_NAME_STATE;          $self->{state} = BEFORE_DOCTYPE_NAME_STATE;
2421          !!!next-input-character;          !!!next-input-character;
# Line 2455  sub _get_next_token ($) { Line 2428  sub _get_next_token ($) {
2428          redo A;          redo A;
2429        }        }
2430      } elsif ($self->{state} == BEFORE_DOCTYPE_NAME_STATE) {      } elsif ($self->{state} == BEFORE_DOCTYPE_NAME_STATE) {
2431        if ($self->{nc} == 0x0009 or # HT        if ($is_space->{$self->{nc}}) {
           $self->{nc} == 0x000A or # LF  
           $self->{nc} == 0x000B or # VT  
           $self->{nc} == 0x000C or # FF  
           $self->{nc} == 0x0020) { # SP  
2432          !!!cp (157);          !!!cp (157);
2433          ## Stay in the state          ## Stay in the state
2434          !!!next-input-character;          !!!next-input-character;
# Line 2493  sub _get_next_token ($) { Line 2462  sub _get_next_token ($) {
2462        }        }
2463      } elsif ($self->{state} == DOCTYPE_NAME_STATE) {      } elsif ($self->{state} == DOCTYPE_NAME_STATE) {
2464  ## ISSUE: Redundant "First," in the spec.  ## ISSUE: Redundant "First," in the spec.
2465        if ($self->{nc} == 0x0009 or # HT        if ($is_space->{$self->{nc}}) {
           $self->{nc} == 0x000A or # LF  
           $self->{nc} == 0x000B or # VT  
           $self->{nc} == 0x000C or # FF  
           $self->{nc} == 0x0020) { # SP  
2466          !!!cp (161);          !!!cp (161);
2467          $self->{state} = AFTER_DOCTYPE_NAME_STATE;          $self->{state} = AFTER_DOCTYPE_NAME_STATE;
2468          !!!next-input-character;          !!!next-input-character;
# Line 2529  sub _get_next_token ($) { Line 2494  sub _get_next_token ($) {
2494          redo A;          redo A;
2495        }        }
2496      } elsif ($self->{state} == AFTER_DOCTYPE_NAME_STATE) {      } elsif ($self->{state} == AFTER_DOCTYPE_NAME_STATE) {
2497        if ($self->{nc} == 0x0009 or # HT        if ($is_space->{$self->{nc}}) {
           $self->{nc} == 0x000A or # LF  
           $self->{nc} == 0x000B or # VT  
           $self->{nc} == 0x000C or # FF  
           $self->{nc} == 0x0020) { # SP  
2498          !!!cp (165);          !!!cp (165);
2499          ## Stay in the state          ## Stay in the state
2500          !!!next-input-character;          !!!next-input-character;
# Line 2656  sub _get_next_token ($) { Line 2617  sub _get_next_token ($) {
2617          redo A;          redo A;
2618        }        }
2619      } elsif ($self->{state} == BEFORE_DOCTYPE_PUBLIC_IDENTIFIER_STATE) {      } elsif ($self->{state} == BEFORE_DOCTYPE_PUBLIC_IDENTIFIER_STATE) {
2620        if ({        if ($is_space->{$self->{nc}}) {
             0x0009 => 1, 0x000A => 1, 0x000B => 1, 0x000C => 1, 0x0020 => 1,  
             #0x000D => 1, # HT, LF, VT, FF, SP, CR  
           }->{$self->{nc}}) {  
2621          !!!cp (181);          !!!cp (181);
2622          ## Stay in the state          ## Stay in the state
2623          !!!next-input-character;          !!!next-input-character;
# Line 2786  sub _get_next_token ($) { Line 2744  sub _get_next_token ($) {
2744          redo A;          redo A;
2745        }        }
2746      } elsif ($self->{state} == AFTER_DOCTYPE_PUBLIC_IDENTIFIER_STATE) {      } elsif ($self->{state} == AFTER_DOCTYPE_PUBLIC_IDENTIFIER_STATE) {
2747        if ({        if ($is_space->{$self->{nc}}) {
             0x0009 => 1, 0x000A => 1, 0x000B => 1, 0x000C => 1, 0x0020 => 1,  
             #0x000D => 1, # HT, LF, VT, FF, SP, CR  
           }->{$self->{nc}}) {  
2748          !!!cp (195);          !!!cp (195);
2749          ## Stay in the state          ## Stay in the state
2750          !!!next-input-character;          !!!next-input-character;
# Line 2835  sub _get_next_token ($) { Line 2790  sub _get_next_token ($) {
2790          redo A;          redo A;
2791        }        }
2792      } elsif ($self->{state} == BEFORE_DOCTYPE_SYSTEM_IDENTIFIER_STATE) {      } elsif ($self->{state} == BEFORE_DOCTYPE_SYSTEM_IDENTIFIER_STATE) {
2793        if ({        if ($is_space->{$self->{nc}}) {
             0x0009 => 1, 0x000A => 1, 0x000B => 1, 0x000C => 1, 0x0020 => 1,  
             #0x000D => 1, # HT, LF, VT, FF, SP, CR  
           }->{$self->{nc}}) {  
2794          !!!cp (201);          !!!cp (201);
2795          ## Stay in the state          ## Stay in the state
2796          !!!next-input-character;          !!!next-input-character;
# Line 2964  sub _get_next_token ($) { Line 2916  sub _get_next_token ($) {
2916          redo A;          redo A;
2917        }        }
2918      } elsif ($self->{state} == AFTER_DOCTYPE_SYSTEM_IDENTIFIER_STATE) {      } elsif ($self->{state} == AFTER_DOCTYPE_SYSTEM_IDENTIFIER_STATE) {
2919        if ({        if ($is_space->{$self->{nc}}) {
             0x0009 => 1, 0x000A => 1, 0x000B => 1, 0x000C => 1, 0x0020 => 1,  
             #0x000D => 1, # HT, LF, VT, FF, SP, CR  
           }->{$self->{nc}}) {  
2920          !!!cp (215);          !!!cp (215);
2921          ## Stay in the state          ## Stay in the state
2922          !!!next-input-character;          !!!next-input-character;
# Line 3099  sub _get_next_token ($) { Line 3048  sub _get_next_token ($) {
3048          redo A;          redo A;
3049        }        }
3050      } elsif ($self->{state} == ENTITY_STATE) {      } elsif ($self->{state} == ENTITY_STATE) {
3051        if ({        if ($is_space->{$self->{nc}} or
3052          0x0009 => 1, 0x000A => 1, 0x000B => 1, 0x000C => 1, # HT, LF, VT, FF,            {
3053          0x0020 => 1, 0x003C => 1, 0x0026 => 1, -1 => 1, # SP, <, &              0x003C => 1, 0x0026 => 1, -1 => 1, # <, &
3054          $self->{entity_add} => 1,              $self->{entity_add} => 1,
3055        }->{$self->{nc}}) {            }->{$self->{nc}}) {
3056          !!!cp (1001);          !!!cp (1001);
3057          ## Don't consume          ## Don't consume
3058          ## No error          ## No error
Legend:
Removed from v.1.186  
changed lines
  Added in v.1.187
[email protected]
 ViewVC Help
Powered by ViewVC 1.1.24