diff options
Diffstat (limited to 'pygments/lexers/sql.py')
| -rw-r--r-- | pygments/lexers/sql.py | 198 |
1 files changed, 102 insertions, 96 deletions
diff --git a/pygments/lexers/sql.py b/pygments/lexers/sql.py index 7540f079..91e2237c 100644 --- a/pygments/lexers/sql.py +++ b/pygments/lexers/sql.py @@ -40,14 +40,14 @@ import re -from pygments.lexer import Lexer, RegexLexer, do_insertions, bygroups +from pygments.lexer import Lexer, RegexLexer, do_insertions, bygroups, words from pygments.token import Punctuation, \ Text, Comment, Operator, Keyword, Name, String, Number, Generic from pygments.lexers import get_lexer_by_name, ClassNotFound from pygments.util import iteritems from pygments.lexers._postgres_builtins import KEYWORDS, DATATYPES, \ - PSEUDO_TYPES, PLPGSQL_KEYWORDS + PSEUDO_TYPES, PLPGSQL_KEYWORDS __all__ = ['PostgresLexer', 'PlPgsqlLexer', 'PostgresConsoleLexer', @@ -57,6 +57,7 @@ line_re = re.compile('.*?\n') language_re = re.compile(r"\s+LANGUAGE\s+'?(\w+)'?", re.IGNORECASE) + def language_callback(lexer, match): """Parse the content of a $-string using a lexer @@ -102,7 +103,7 @@ class PostgresBase(object): if lang.lower() == 'sql': return get_lexer_by_name('postgresql', **self.options) - tries = [ lang ] + tries = [lang] if lang.startswith('pl'): tries.append(lang[2:]) if lang.endswith('u'): @@ -138,17 +139,17 @@ class PostgresLexer(PostgresBase, RegexLexer): (r'\s+', Text), (r'--.*?\n', Comment.Single), (r'/\*', Comment.Multiline, 'multiline-comments'), - (r'(' + '|'.join([s.replace(" ", "\s+") - for s in DATATYPES + PSEUDO_TYPES]) - + r')\b', Name.Builtin), - (r'(' + '|'.join(KEYWORDS) + r')\b', Keyword), + (r'(' + '|'.join(s.replace(" ", "\s+") + for s in DATATYPES + PSEUDO_TYPES) + + r')\b', Name.Builtin), + (words(KEYWORDS, suffix=r'\b'), Keyword), (r'[+*/<>=~!@#%^&|`?-]+', Operator), (r'::', Operator), # cast (r'\$\d+', Name.Variable), (r'([0-9]*\.[0-9]*|[0-9]+)(e[+-]?[0-9]+)?', Number.Float), (r'[0-9]+', Number.Integer), (r"(E|U&)?'(''|[^'])*'", String.Single), - (r'(U&)?"(""|[^"])*"', String.Name), # quoted identifier + (r'(U&)?"(""|[^"])*"', String.Name), # quoted identifier (r'(?s)(\$[^\$]*\$)(.*?)(\1)', language_callback), (r'[a-z_]\w*', Name), @@ -183,7 +184,7 @@ class PlPgsqlLexer(PostgresBase, RegexLexer): for i, pattern in enumerate(tokens['root']): if pattern[1] == Keyword: tokens['root'][i] = ( - r'(' + '|'.join(KEYWORDS + PLPGSQL_KEYWORDS) + r')\b', + words(KEYWORDS + PLPGSQL_KEYWORDS, suffix=r'\b'), Keyword) del i break @@ -240,11 +241,14 @@ class lookahead(object): def __init__(self, x): self.iter = iter(x) self._nextitem = None + def __iter__(self): return self + def send(self, i): self._nextitem = i return i + def __next__(self): if self._nextitem is not None: ni = self._nextitem @@ -305,12 +309,12 @@ class PostgresConsoleLexer(Lexer): # TODO: better handle multiline comments at the end with # a lexer with an external state? if re_psql_command.match(curcode) \ - or re_end_command.search(curcode): + or re_end_command.search(curcode): break # Emit the combined stream of command and prompt(s) for item in do_insertions(insertions, - sql.get_tokens_unprocessed(curcode)): + sql.get_tokens_unprocessed(curcode)): yield item # Emit the output lines @@ -326,7 +330,7 @@ class PostgresConsoleLexer(Lexer): mmsg = re_message.match(line) if mmsg is not None: if mmsg.group(1).startswith("ERROR") \ - or mmsg.group(1).startswith("FATAL"): + or mmsg.group(1).startswith("FATAL"): out_token = Generic.Error yield (mmsg.start(1), Generic.Strong, mmsg.group(1)) yield (mmsg.start(2), out_token, mmsg.group(2)) @@ -351,91 +355,94 @@ class SqlLexer(RegexLexer): (r'\s+', Text), (r'--.*?\n', Comment.Single), (r'/\*', Comment.Multiline, 'multiline-comments'), - (r'(ABORT|ABS|ABSOLUTE|ACCESS|ADA|ADD|ADMIN|AFTER|AGGREGATE|' - r'ALIAS|ALL|ALLOCATE|ALTER|ANALYSE|ANALYZE|AND|ANY|ARE|AS|' - r'ASC|ASENSITIVE|ASSERTION|ASSIGNMENT|ASYMMETRIC|AT|ATOMIC|' - r'AUTHORIZATION|AVG|BACKWARD|BEFORE|BEGIN|BETWEEN|BITVAR|' - r'BIT_LENGTH|BOTH|BREADTH|BY|C|CACHE|CALL|CALLED|CARDINALITY|' - r'CASCADE|CASCADED|CASE|CAST|CATALOG|CATALOG_NAME|CHAIN|' - r'CHARACTERISTICS|CHARACTER_LENGTH|CHARACTER_SET_CATALOG|' - r'CHARACTER_SET_NAME|CHARACTER_SET_SCHEMA|CHAR_LENGTH|CHECK|' - r'CHECKED|CHECKPOINT|CLASS|CLASS_ORIGIN|CLOB|CLOSE|CLUSTER|' - r'COALSECE|COBOL|COLLATE|COLLATION|COLLATION_CATALOG|' - r'COLLATION_NAME|COLLATION_SCHEMA|COLUMN|COLUMN_NAME|' - r'COMMAND_FUNCTION|COMMAND_FUNCTION_CODE|COMMENT|COMMIT|' - r'COMMITTED|COMPLETION|CONDITION_NUMBER|CONNECT|CONNECTION|' - r'CONNECTION_NAME|CONSTRAINT|CONSTRAINTS|CONSTRAINT_CATALOG|' - r'CONSTRAINT_NAME|CONSTRAINT_SCHEMA|CONSTRUCTOR|CONTAINS|' - r'CONTINUE|CONVERSION|CONVERT|COPY|CORRESPONTING|COUNT|' - r'CREATE|CREATEDB|CREATEUSER|CROSS|CUBE|CURRENT|CURRENT_DATE|' - r'CURRENT_PATH|CURRENT_ROLE|CURRENT_TIME|CURRENT_TIMESTAMP|' - r'CURRENT_USER|CURSOR|CURSOR_NAME|CYCLE|DATA|DATABASE|' - r'DATETIME_INTERVAL_CODE|DATETIME_INTERVAL_PRECISION|DAY|' - r'DEALLOCATE|DECLARE|DEFAULT|DEFAULTS|DEFERRABLE|DEFERRED|' - r'DEFINED|DEFINER|DELETE|DELIMITER|DELIMITERS|DEREF|DESC|' - r'DESCRIBE|DESCRIPTOR|DESTROY|DESTRUCTOR|DETERMINISTIC|' - r'DIAGNOSTICS|DICTIONARY|DISCONNECT|DISPATCH|DISTINCT|DO|' - r'DOMAIN|DROP|DYNAMIC|DYNAMIC_FUNCTION|DYNAMIC_FUNCTION_CODE|' - r'EACH|ELSE|ENCODING|ENCRYPTED|END|END-EXEC|EQUALS|ESCAPE|EVERY|' - r'EXCEPTION|EXCEPT|EXCLUDING|EXCLUSIVE|EXEC|EXECUTE|EXISTING|' - r'EXISTS|EXPLAIN|EXTERNAL|EXTRACT|FALSE|FETCH|FINAL|FIRST|FOR|' - r'FORCE|FOREIGN|FORTRAN|FORWARD|FOUND|FREE|FREEZE|FROM|FULL|' - r'FUNCTION|G|GENERAL|GENERATED|GET|GLOBAL|GO|GOTO|GRANT|GRANTED|' - r'GROUP|GROUPING|HANDLER|HAVING|HIERARCHY|HOLD|HOST|IDENTITY|' - r'IGNORE|ILIKE|IMMEDIATE|IMMUTABLE|IMPLEMENTATION|IMPLICIT|IN|' - r'INCLUDING|INCREMENT|INDEX|INDITCATOR|INFIX|INHERITS|INITIALIZE|' - r'INITIALLY|INNER|INOUT|INPUT|INSENSITIVE|INSERT|INSTANTIABLE|' - r'INSTEAD|INTERSECT|INTO|INVOKER|IS|ISNULL|ISOLATION|ITERATE|JOIN|' - r'KEY|KEY_MEMBER|KEY_TYPE|LANCOMPILER|LANGUAGE|LARGE|LAST|' - r'LATERAL|LEADING|LEFT|LENGTH|LESS|LEVEL|LIKE|LIMIT|LISTEN|LOAD|' - r'LOCAL|LOCALTIME|LOCALTIMESTAMP|LOCATION|LOCATOR|LOCK|LOWER|' - r'MAP|MATCH|MAX|MAXVALUE|MESSAGE_LENGTH|MESSAGE_OCTET_LENGTH|' - r'MESSAGE_TEXT|METHOD|MIN|MINUTE|MINVALUE|MOD|MODE|MODIFIES|' - r'MODIFY|MONTH|MORE|MOVE|MUMPS|NAMES|NATIONAL|NATURAL|NCHAR|' - r'NCLOB|NEW|NEXT|NO|NOCREATEDB|NOCREATEUSER|NONE|NOT|NOTHING|' - r'NOTIFY|NOTNULL|NULL|NULLABLE|NULLIF|OBJECT|OCTET_LENGTH|OF|OFF|' - r'OFFSET|OIDS|OLD|ON|ONLY|OPEN|OPERATION|OPERATOR|OPTION|OPTIONS|' - r'OR|ORDER|ORDINALITY|OUT|OUTER|OUTPUT|OVERLAPS|OVERLAY|OVERRIDING|' - r'OWNER|PAD|PARAMETER|PARAMETERS|PARAMETER_MODE|PARAMATER_NAME|' - r'PARAMATER_ORDINAL_POSITION|PARAMETER_SPECIFIC_CATALOG|' - r'PARAMETER_SPECIFIC_NAME|PARAMATER_SPECIFIC_SCHEMA|PARTIAL|' - r'PASCAL|PENDANT|PLACING|PLI|POSITION|POSTFIX|PRECISION|PREFIX|' - r'PREORDER|PREPARE|PRESERVE|PRIMARY|PRIOR|PRIVILEGES|PROCEDURAL|' - r'PROCEDURE|PUBLIC|READ|READS|RECHECK|RECURSIVE|REF|REFERENCES|' - r'REFERENCING|REINDEX|RELATIVE|RENAME|REPEATABLE|REPLACE|RESET|' - r'RESTART|RESTRICT|RESULT|RETURN|RETURNED_LENGTH|' - r'RETURNED_OCTET_LENGTH|RETURNED_SQLSTATE|RETURNS|REVOKE|RIGHT|' - r'ROLE|ROLLBACK|ROLLUP|ROUTINE|ROUTINE_CATALOG|ROUTINE_NAME|' - r'ROUTINE_SCHEMA|ROW|ROWS|ROW_COUNT|RULE|SAVE_POINT|SCALE|SCHEMA|' - r'SCHEMA_NAME|SCOPE|SCROLL|SEARCH|SECOND|SECURITY|SELECT|SELF|' - r'SENSITIVE|SERIALIZABLE|SERVER_NAME|SESSION|SESSION_USER|SET|' - r'SETOF|SETS|SHARE|SHOW|SIMILAR|SIMPLE|SIZE|SOME|SOURCE|SPACE|' - r'SPECIFIC|SPECIFICTYPE|SPECIFIC_NAME|SQL|SQLCODE|SQLERROR|' - r'SQLEXCEPTION|SQLSTATE|SQLWARNINIG|STABLE|START|STATE|STATEMENT|' - r'STATIC|STATISTICS|STDIN|STDOUT|STORAGE|STRICT|STRUCTURE|STYPE|' - r'SUBCLASS_ORIGIN|SUBLIST|SUBSTRING|SUM|SYMMETRIC|SYSID|SYSTEM|' - r'SYSTEM_USER|TABLE|TABLE_NAME| TEMP|TEMPLATE|TEMPORARY|TERMINATE|' - r'THAN|THEN|TIMESTAMP|TIMEZONE_HOUR|TIMEZONE_MINUTE|TO|TOAST|' - r'TRAILING|TRANSATION|TRANSACTIONS_COMMITTED|' - r'TRANSACTIONS_ROLLED_BACK|TRANSATION_ACTIVE|TRANSFORM|' - r'TRANSFORMS|TRANSLATE|TRANSLATION|TREAT|TRIGGER|TRIGGER_CATALOG|' - r'TRIGGER_NAME|TRIGGER_SCHEMA|TRIM|TRUE|TRUNCATE|TRUSTED|TYPE|' - r'UNCOMMITTED|UNDER|UNENCRYPTED|UNION|UNIQUE|UNKNOWN|UNLISTEN|' - r'UNNAMED|UNNEST|UNTIL|UPDATE|UPPER|USAGE|USER|' - r'USER_DEFINED_TYPE_CATALOG|USER_DEFINED_TYPE_NAME|' - r'USER_DEFINED_TYPE_SCHEMA|USING|VACUUM|VALID|VALIDATOR|VALUES|' - r'VARIABLE|VERBOSE|VERSION|VIEW|VOLATILE|WHEN|WHENEVER|WHERE|' - r'WITH|WITHOUT|WORK|WRITE|YEAR|ZONE)\b', Keyword), - (r'(ARRAY|BIGINT|BINARY|BIT|BLOB|BOOLEAN|CHAR|CHARACTER|DATE|' - r'DEC|DECIMAL|FLOAT|INT|INTEGER|INTERVAL|NUMBER|NUMERIC|REAL|' - r'SERIAL|SMALLINT|VARCHAR|VARYING|INT8|SERIAL8|TEXT)\b', + (words(( + 'ABORT', 'ABS', 'ABSOLUTE', 'ACCESS', 'ADA', 'ADD', 'ADMIN', 'AFTER', 'AGGREGATE', + 'ALIAS', 'ALL', 'ALLOCATE', 'ALTER', 'ANALYSE', 'ANALYZE', 'AND', 'ANY', 'ARE', 'AS', + 'ASC', 'ASENSITIVE', 'ASSERTION', 'ASSIGNMENT', 'ASYMMETRIC', 'AT', 'ATOMIC', + 'AUTHORIZATION', 'AVG', 'BACKWARD', 'BEFORE', 'BEGIN', 'BETWEEN', 'BITVAR', + 'BIT_LENGTH', 'BOTH', 'BREADTH', 'BY', 'C', 'CACHE', 'CALL', 'CALLED', 'CARDINALITY', + 'CASCADE', 'CASCADED', 'CASE', 'CAST', 'CATALOG', 'CATALOG_NAME', 'CHAIN', + 'CHARACTERISTICS', 'CHARACTER_LENGTH', 'CHARACTER_SET_CATALOG', + 'CHARACTER_SET_NAME', 'CHARACTER_SET_SCHEMA', 'CHAR_LENGTH', 'CHECK', + 'CHECKED', 'CHECKPOINT', 'CLASS', 'CLASS_ORIGIN', 'CLOB', 'CLOSE', 'CLUSTER', + 'COALSECE', 'COBOL', 'COLLATE', 'COLLATION', 'COLLATION_CATALOG', + 'COLLATION_NAME', 'COLLATION_SCHEMA', 'COLUMN', 'COLUMN_NAME', + 'COMMAND_FUNCTION', 'COMMAND_FUNCTION_CODE', 'COMMENT', 'COMMIT', + 'COMMITTED', 'COMPLETION', 'CONDITION_NUMBER', 'CONNECT', 'CONNECTION', + 'CONNECTION_NAME', 'CONSTRAINT', 'CONSTRAINTS', 'CONSTRAINT_CATALOG', + 'CONSTRAINT_NAME', 'CONSTRAINT_SCHEMA', 'CONSTRUCTOR', 'CONTAINS', + 'CONTINUE', 'CONVERSION', 'CONVERT', 'COPY', 'CORRESPONTING', 'COUNT', + 'CREATE', 'CREATEDB', 'CREATEUSER', 'CROSS', 'CUBE', 'CURRENT', 'CURRENT_DATE', + 'CURRENT_PATH', 'CURRENT_ROLE', 'CURRENT_TIME', 'CURRENT_TIMESTAMP', + 'CURRENT_USER', 'CURSOR', 'CURSOR_NAME', 'CYCLE', 'DATA', 'DATABASE', + 'DATETIME_INTERVAL_CODE', 'DATETIME_INTERVAL_PRECISION', 'DAY', + 'DEALLOCATE', 'DECLARE', 'DEFAULT', 'DEFAULTS', 'DEFERRABLE', 'DEFERRED', + 'DEFINED', 'DEFINER', 'DELETE', 'DELIMITER', 'DELIMITERS', 'DEREF', 'DESC', + 'DESCRIBE', 'DESCRIPTOR', 'DESTROY', 'DESTRUCTOR', 'DETERMINISTIC', + 'DIAGNOSTICS', 'DICTIONARY', 'DISCONNECT', 'DISPATCH', 'DISTINCT', 'DO', + 'DOMAIN', 'DROP', 'DYNAMIC', 'DYNAMIC_FUNCTION', 'DYNAMIC_FUNCTION_CODE', + 'EACH', 'ELSE', 'ENCODING', 'ENCRYPTED', 'END', 'END-EXEC', 'EQUALS', 'ESCAPE', 'EVERY', + 'EXCEPTION', 'EXCEPT', 'EXCLUDING', 'EXCLUSIVE', 'EXEC', 'EXECUTE', 'EXISTING', + 'EXISTS', 'EXPLAIN', 'EXTERNAL', 'EXTRACT', 'FALSE', 'FETCH', 'FINAL', 'FIRST', 'FOR', + 'FORCE', 'FOREIGN', 'FORTRAN', 'FORWARD', 'FOUND', 'FREE', 'FREEZE', 'FROM', 'FULL', + 'FUNCTION', 'G', 'GENERAL', 'GENERATED', 'GET', 'GLOBAL', 'GO', 'GOTO', 'GRANT', 'GRANTED', + 'GROUP', 'GROUPING', 'HANDLER', 'HAVING', 'HIERARCHY', 'HOLD', 'HOST', 'IDENTITY', + 'IGNORE', 'ILIKE', 'IMMEDIATE', 'IMMUTABLE', 'IMPLEMENTATION', 'IMPLICIT', 'IN', + 'INCLUDING', 'INCREMENT', 'INDEX', 'INDITCATOR', 'INFIX', 'INHERITS', 'INITIALIZE', + 'INITIALLY', 'INNER', 'INOUT', 'INPUT', 'INSENSITIVE', 'INSERT', 'INSTANTIABLE', + 'INSTEAD', 'INTERSECT', 'INTO', 'INVOKER', 'IS', 'ISNULL', 'ISOLATION', 'ITERATE', 'JOIN', + 'KEY', 'KEY_MEMBER', 'KEY_TYPE', 'LANCOMPILER', 'LANGUAGE', 'LARGE', 'LAST', + 'LATERAL', 'LEADING', 'LEFT', 'LENGTH', 'LESS', 'LEVEL', 'LIKE', 'LIMIT', 'LISTEN', 'LOAD', + 'LOCAL', 'LOCALTIME', 'LOCALTIMESTAMP', 'LOCATION', 'LOCATOR', 'LOCK', 'LOWER', + 'MAP', 'MATCH', 'MAX', 'MAXVALUE', 'MESSAGE_LENGTH', 'MESSAGE_OCTET_LENGTH', + 'MESSAGE_TEXT', 'METHOD', 'MIN', 'MINUTE', 'MINVALUE', 'MOD', 'MODE', 'MODIFIES', + 'MODIFY', 'MONTH', 'MORE', 'MOVE', 'MUMPS', 'NAMES', 'NATIONAL', 'NATURAL', 'NCHAR', + 'NCLOB', 'NEW', 'NEXT', 'NO', 'NOCREATEDB', 'NOCREATEUSER', 'NONE', 'NOT', 'NOTHING', + 'NOTIFY', 'NOTNULL', 'NULL', 'NULLABLE', 'NULLIF', 'OBJECT', 'OCTET_LENGTH', 'OF', 'OFF', + 'OFFSET', 'OIDS', 'OLD', 'ON', 'ONLY', 'OPEN', 'OPERATION', 'OPERATOR', 'OPTION', 'OPTIONS', + 'OR', 'ORDER', 'ORDINALITY', 'OUT', 'OUTER', 'OUTPUT', 'OVERLAPS', 'OVERLAY', 'OVERRIDING', + 'OWNER', 'PAD', 'PARAMETER', 'PARAMETERS', 'PARAMETER_MODE', 'PARAMATER_NAME', + 'PARAMATER_ORDINAL_POSITION', 'PARAMETER_SPECIFIC_CATALOG', + 'PARAMETER_SPECIFIC_NAME', 'PARAMATER_SPECIFIC_SCHEMA', 'PARTIAL', + 'PASCAL', 'PENDANT', 'PLACING', 'PLI', 'POSITION', 'POSTFIX', 'PRECISION', 'PREFIX', + 'PREORDER', 'PREPARE', 'PRESERVE', 'PRIMARY', 'PRIOR', 'PRIVILEGES', 'PROCEDURAL', + 'PROCEDURE', 'PUBLIC', 'READ', 'READS', 'RECHECK', 'RECURSIVE', 'REF', 'REFERENCES', + 'REFERENCING', 'REINDEX', 'RELATIVE', 'RENAME', 'REPEATABLE', 'REPLACE', 'RESET', + 'RESTART', 'RESTRICT', 'RESULT', 'RETURN', 'RETURNED_LENGTH', + 'RETURNED_OCTET_LENGTH', 'RETURNED_SQLSTATE', 'RETURNS', 'REVOKE', 'RIGHT', + 'ROLE', 'ROLLBACK', 'ROLLUP', 'ROUTINE', 'ROUTINE_CATALOG', 'ROUTINE_NAME', + 'ROUTINE_SCHEMA', 'ROW', 'ROWS', 'ROW_COUNT', 'RULE', 'SAVE_POINT', 'SCALE', 'SCHEMA', + 'SCHEMA_NAME', 'SCOPE', 'SCROLL', 'SEARCH', 'SECOND', 'SECURITY', 'SELECT', 'SELF', + 'SENSITIVE', 'SERIALIZABLE', 'SERVER_NAME', 'SESSION', 'SESSION_USER', 'SET', + 'SETOF', 'SETS', 'SHARE', 'SHOW', 'SIMILAR', 'SIMPLE', 'SIZE', 'SOME', 'SOURCE', 'SPACE', + 'SPECIFIC', 'SPECIFICTYPE', 'SPECIFIC_NAME', 'SQL', 'SQLCODE', 'SQLERROR', + 'SQLEXCEPTION', 'SQLSTATE', 'SQLWARNINIG', 'STABLE', 'START', 'STATE', 'STATEMENT', + 'STATIC', 'STATISTICS', 'STDIN', 'STDOUT', 'STORAGE', 'STRICT', 'STRUCTURE', 'STYPE', + 'SUBCLASS_ORIGIN', 'SUBLIST', 'SUBSTRING', 'SUM', 'SYMMETRIC', 'SYSID', 'SYSTEM', + 'SYSTEM_USER', 'TABLE', 'TABLE_NAME', ' TEMP', 'TEMPLATE', 'TEMPORARY', 'TERMINATE', + 'THAN', 'THEN', 'TIMESTAMP', 'TIMEZONE_HOUR', 'TIMEZONE_MINUTE', 'TO', 'TOAST', + 'TRAILING', 'TRANSATION', 'TRANSACTIONS_COMMITTED', + 'TRANSACTIONS_ROLLED_BACK', 'TRANSATION_ACTIVE', 'TRANSFORM', + 'TRANSFORMS', 'TRANSLATE', 'TRANSLATION', 'TREAT', 'TRIGGER', 'TRIGGER_CATALOG', + 'TRIGGER_NAME', 'TRIGGER_SCHEMA', 'TRIM', 'TRUE', 'TRUNCATE', 'TRUSTED', 'TYPE', + 'UNCOMMITTED', 'UNDER', 'UNENCRYPTED', 'UNION', 'UNIQUE', 'UNKNOWN', 'UNLISTEN', + 'UNNAMED', 'UNNEST', 'UNTIL', 'UPDATE', 'UPPER', 'USAGE', 'USER', + 'USER_DEFINED_TYPE_CATALOG', 'USER_DEFINED_TYPE_NAME', + 'USER_DEFINED_TYPE_SCHEMA', 'USING', 'VACUUM', 'VALID', 'VALIDATOR', 'VALUES', + 'VARIABLE', 'VERBOSE', 'VERSION', 'VIEW', 'VOLATILE', 'WHEN', 'WHENEVER', 'WHERE', + 'WITH', 'WITHOUT', 'WORK', 'WRITE', 'YEAR', 'ZONE'), suffix=r'\b'), + Keyword), + (words(( + 'ARRAY', 'BIGINT', 'BINARY', 'BIT', 'BLOB', 'BOOLEAN', 'CHAR', 'CHARACTER', 'DATE', + 'DEC', 'DECIMAL', 'FLOAT', 'INT', 'INTEGER', 'INTERVAL', 'NUMBER', 'NUMERIC', 'REAL', + 'SERIAL', 'SMALLINT', 'VARCHAR', 'VARYING', 'INT8', 'SERIAL8', 'TEXT'), suffix=r'\b'), Name.Builtin), (r'[+*/<>=~!@#%^&|`?-]', Operator), (r'[0-9]+', Number.Integer), # TODO: Backslash escapes? (r"'(''|[^'])*'", String.Single), - (r'"(""|[^"])*"', String.Symbol), # not a real string literal in ANSI SQL - (r'[a-z_]\w*', Name), + (r'"(""|[^"])*"', String.Symbol), # not a real string literal in ANSI SQL + (r'[a-z_][\w\$]*', Name), # allow $s in strings for Oracle (r'[;:()\[\],\.]', Punctuation) ], 'multiline-comments': [ @@ -464,10 +471,9 @@ class MySqlLexer(RegexLexer): (r'/\*', Comment.Multiline, 'multiline-comments'), (r'[0-9]+', Number.Integer), (r'[0-9]*\.[0-9]+(e[+-][0-9]+)', Number.Float), - # TODO: add backslash escapes - (r"'(''|[^'])*'", String.Single), - (r'"(""|[^"])*"', String.Double), - (r"`(``|[^`])*`", String.Symbol), + (r"'(\\\\|\\'|''|[^'])*'", String.Single), + (r'"(\\\\|\\"|""|[^"])*"', String.Double), + (r"`(\\\\|\\`|``|[^`])*`", String.Symbol), (r'[+*/<>=~!@#%^&|`?-]', Operator), (r'\b(tinyint|smallint|mediumint|int|integer|bigint|date|' r'datetime|time|bit|bool|tinytext|mediumtext|longtext|text|' |