validator/cpp/htmlparser/data/jsongrammar.txt - main - githubmirror/amphtml

validator/cpp/htmlparser/data/jsongrammar.txt
147 строк6 KB
Allan Banaag
Sync for validator cpp engine and cpp htmlparser (#39797)
06 фев 2024, 01:17
06 фев 2024, 01:176e751a4
# JSON grammar. https://www.json.org/

# Begin.
$ '{' DICT ROOT_DICT;
$ '[' ARRAY_FIRST_ITEM ROOT_ARRAY;
$ '"' STRING ROOT_STRING;
$ '-' NUMBER_AFTER_HYPHEN ROOT_NUMBER;
$ "1..9" NUMBER ROOT_NUMBER;
$ '0' NUMBER_AFTER_ZERO ROOT_NUMBER_AFTER_ZERO;
$ 't' TRUE_1  ROOT_BOOL_TRUE;
$ 'f' FALSE_1 ROOT_BOOL_FALSE;
$ 'n' NULL_1 ROOT_NULL_VAL;
# Ignore whitespace chars.
$ "\t\n\r " $;

# true literal (case sensitive, lowercase).
TRUE_1 'r' TRUE_2;
TRUE_2 'u' TRUE_3;
TRUE_3 'e' POP;

# false literal (case sensitive, lowercase).
FALSE_1 'a' FALSE_2;
FALSE_2 'l' FALSE_3;
FALSE_3 's' FALSE_4;
FALSE_4 'e' POP;

# null literal (case sensitive, lowercase).
NULL_1 'u' NULL_2;
NULL_2 'l' NULL_3;
NULL_3 'l' POP;

# String.
# https://www.json.org/img/string.png
STRING .* STRING;
# Stop string mode at closing double quotes.
# STRING_ESCAPE ignores any escaped double quotes.
STRING '"' POP;

# Escape char followed by \. Allowed chars are:
# " - Quotation mark.
# \ - Reverse Solidus.
# / - Solidus.
# b - Backspace.
# f - Formfeed.
# n - Linefeed.
# r - Carriage return.
# t = Horizontal tab.
STRING '\' STRING_ESCAPE;
STRING_ESCAPE "\"\/bfnrt" STRING;
# Next four characters followed by \u
STRING_ESCAPE 'u' UNICODE_HEX_CHAR_FIRST;
UNICODE_HEX_CHAR_FIRST "0..9a..fA..F" UNICODE_HEX_CHAR_SECOND;
UNICODE_HEX_CHAR_SECOND "0..9a..fA..F" UNICODE_HEX_CHAR_THIRD;
UNICODE_HEX_CHAR_THIRD "0..9a..fA..F" UNICODE_HEX_CHAR_LAST;
UNICODE_HEX_CHAR_LAST "0..9a..fA..F" STRING;

# Number.
# https://www.json.org/img/number.png
# 1-9
NUMBER "0..9" NUMBER;
NUMBER '.' NUMBER_AFTER_PERIOD FLOATING_POINT_T;
NUMBER "\t\n\r " POP;
NUMBER ",}]" SHIFT;

NUMBER_AFTER_PERIOD "0..9" NUMBER_AFTER_PERIOD_DIGIT;
NUMBER_AFTER_PERIOD_DIGIT "0..9" NUMBER_AFTER_PERIOD_DIGIT;
NUMBER_AFTER_PERIOD_DIGIT "eE" NUMBER_AFTER_E_NOTATION;
NUMBER_AFTER_PERIOD_DIGIT ",}]\t\n\r " SHIFT;

NUMBER_AFTER_ZERO '.' NUMBER_AFTER_PERIOD FLOATING_POINT_T;
# Even though 0e+3 is invalid syntax (zero must be followed by period), but
# browsers accept it as valid and convert it to 0.
# As per spec (https://www.json.org/img/number.png) this is invalid, but we
# are considering it as OK.
# Commenting the following line will treat 0e+3 number value as invalid json.
NUMBER_AFTER_ZERO "eE" NUMBER_AFTER_E_NOTATION;
NUMBER_AFTER_ZERO ",}]\t\n\r " SHIFT;

NUMBER_AFTER_HYPHEN "0..9" NUMBER;
NUMBER_AFTER_HYPHEN '0' NUMBER_AFTER_ZERO;

NUMBER_AFTER_E_NOTATION "+-" NUMBER_AFTER_PLUS_MINUS_AFTER_E_NOTATION;
NUMBER_AFTER_E_NOTATION "0..9" NUMBER_AFTER_E_DIGIT_NOTATION;

NUMBER_AFTER_E_DIGIT_NOTATION "0..9" NUMBER_AFTER_E_DIGIT_NOTATION;
NUMBER_AFTER_PLUS_MINUS_AFTER_E_NOTATION "0..9" NUMBER_AFTER_PLUS_MINUS_AFTER_E_NOTATION;
NUMBER_AFTER_E_DIGIT_NOTATION ",}]\t\n\r " SHIFT;
NUMBER_AFTER_PLUS_MINUS_AFTER_E_NOTATION " \t\n\r" POP;
NUMBER_AFTER_PLUS_MINUS_AFTER_E_NOTATION ",}]" SHIFT;

# Arrays.
# https://www.json.org/img/array.png
ARRAY_FIRST_ITEM ']' POP ARRAY_END;
ARRAY_FIRST_ITEM '{' DICT|PUSH|ARRAY_SEPARATOR DICT_T;
ARRAY_FIRST_ITEM '[' ARRAY_FIRST_ITEM|PUSH|ARRAY_SEPARATOR ARRAY_T;
ARRAY_FIRST_ITEM '"' STRING|PUSH|ARRAY_SEPARATOR STRING_T;
ARRAY_FIRST_ITEM '-' NUMBER_AFTER_HYPHEN|PUSH|ARRAY_SEPARATOR NUMBER_T;
ARRAY_FIRST_ITEM '0' NUMBER_AFTER_ZERO|PUSH|ARRAY_SEPARATOR NUMBER_T;
ARRAY_FIRST_ITEM "1..9" NUMBER|PUSH|ARRAY_SEPARATOR NUMBER_T;
ARRAY_FIRST_ITEM 't' TRUE_1|PUSH|ARRAY_SEPARATOR TRUE_T;
ARRAY_FIRST_ITEM 'f' FALSE_1|PUSH|ARRAY_SEPARATOR FALSE_T;
ARRAY_FIRST_ITEM 'n' NULL_1|PUSH|ARRAY_SEPARATOR NULL_T;
ARRAY_FIRST_ITEM ']' POP ARRAY_END;
# Ignore whitespace chars between [ and first item.
ARRAY_FIRST_ITEM "\t\n\r " ARRAY_FIRST_ITEM;
ARRAY_SEPARATOR ',' ARRAY_NEXT_ITEM ARRAY_VAL_END;
# Ignore whitespce chars between first item and comma.
ARRAY_SEPARATOR "\t\n\r " ARRAY_SEPARATOR;
ARRAY_SEPARATOR ']' POP ARRAY_END;
ARRAY_NEXT_ITEM '{' DICT|PUSH|ARRAY_SEPARATOR DICT_T;
ARRAY_NEXT_ITEM '[' ARRAY_FIRST_ITEM|PUSH|ARRAY_SEPARATOR ARRAY_T;
ARRAY_NEXT_ITEM '"' STRING|PUSH|ARRAY_SEPARATOR STRING_T;
ARRAY_NEXT_ITEM '-' NUMBER_AFTER_HYPHEN|PUSH|ARRAY_SEPARATOR NUMBER_T;
ARRAY_NEXT_ITEM '0' NUMBER_AFTER_ZERO|PUSH|ARRAY_SEPARATOR NUMBER_T;
ARRAY_NEXT_ITEM "1..9" NUMBER|PUSH|ARRAY_SEPARATOR NUMBER_T;
ARRAY_NEXT_ITEM 't' TRUE_1|PUSH|ARRAY_SEPARATOR TRUE_T;
ARRAY_NEXT_ITEM 'f' FALSE_1|PUSH|ARRAY_SEPARATOR FALSE_T;
ARRAY_NEXT_ITEM 'n' NULL_1|PUSH|ARRAY_SEPARATOR NULL_T;
# Ignore whitespace chars between comma and next item or ].
ARRAY_NEXT_ITEM "\t\n\r " ARRAY_NEXT_ITEM;

# Dictionary.
# https://www.json.org/img/object.png
DICT '"' STRING|PUSH|DICT_KEY DICT_KEY_BEGIN;
DICT_KEY_BEGIN_QUOTE '"' STRING|PUSH|DICT_KEY DICT_KEY_BEGIN;
DICT_KEY_BEGIN_QUOTE "\t\n\r " DICT_KEY_BEGIN_QUOTE;
# Ignore whitespace between { and first key.
DICT "\t\n\r " DICT;
DICT '}' POP DICT_END;
DICT_KEY ':' DICT_VALUE DICT_KEY_END;
# Ignore whitespace betwen key and colon.
DICT_KEY "\t\n\r " DICT_KEY;
DICT_VALUE '"' STRING|PUSH|DICT_END_OR_SEPARATOR STRING_T;
DICT_VALUE 't' TRUE_1|PUSH|DICT_END_OR_SEPARATOR TRUE_T;
DICT_VALUE 'f' FALSE_1|PUSH|DICT_END_OR_SEPARATOR FALSE_T;
DICT_VALUE 'n' NULL_1|PUSH|DICT_END_OR_SEPARATOR NULL_T;
# Ignore whitespace between colon and value.
DICT_VALUE "\t\n\r " DICT_VALUE;
DICT_VALUE '0' NUMBER_AFTER_ZERO|PUSH|DICT_END_OR_SEPARATOR NUMBER_T;
DICT_VALUE '-' NUMBER_AFTER_HYPHEN|PUSH|DICT_END_OR_SEPARATOR NUMBER_T;
DICT_VALUE "1..9" NUMBER|PUSH|DICT_END_OR_SEPARATOR NUMBER_T;
DICT_VALUE '[' ARRAY_FIRST_ITEM|PUSH|DICT_END_OR_SEPARATOR ARRAY_T;
DICT_VALUE '{' DICT|PUSH|DICT_END_OR_SEPARATOR DICT_T;
# Ignore whitespace chars between comma and next value.
DICT_END_OR_SEPARATOR "\t\r\n " DICT_END_OR_SEPARATOR;
DICT_END_OR_SEPARATOR ',' DICT_KEY_BEGIN_QUOTE DICT_VAL_END;
DICT_END_OR_SEPARATOR '}' POP DICT_END;