perl perl perl

--HG--
branch : pmacs2
This commit is contained in:
moculus 2007-04-01 05:35:10 +00:00
parent 83ef208a0f
commit 5cf1a2843e
2 changed files with 162 additions and 136 deletions

View File

@ -3,23 +3,23 @@ from lex2 import Grammar, ConstantRule, PatternRule, ContextPatternRule, RegionR
class PodGrammar(Grammar): class PodGrammar(Grammar):
rules = [ rules = [
PatternRule( PatternRule(
name=r'header', name=r'entry',
pattern=r'(?<=^=head[1-4]) +.*$', pattern=r'(?<=^=head[1-4]) +.*$',
), ),
PatternRule( PatternRule(
name=r'indent_level', name=r'entry',
pattern=r'(?<=^=over) +.*$', pattern=r'(?<=^=over) +.*$',
), ),
PatternRule( PatternRule(
name=r'item_entry', name=r'entry',
pattern=r'(?<=^=item) +.*$', pattern=r'(?<=^=item) +.*$',
), ),
PatternRule( PatternRule(
name=r'format', name=r'entry',
pattern=r'(?:(?<=^=begin)|(?<=^=end)) +.*$', pattern=r'(?:(?<=^=begin)|(?<=^=end)) +.*$',
), ),
PatternRule( PatternRule(
name=r'encoding_type', name=r'entry',
pattern=r'(?<=^=encoding) +.*$', pattern=r'(?<=^=encoding) +.*$',
), ),
] ]
@ -31,20 +31,20 @@ class StringGrammar(Grammar):
pattern=r'\\.', pattern=r'\\.',
), ),
PatternRule( PatternRule(
name=r'hash_deref', name=r'deref',
pattern=r"\$\$*[A-Za-z0-9_](?:[A-Za-z0-9_]|::)*(?:->{(?:[a-zA-Z_][a-zA-Z_0-9]*|'(?:\\.|[^'\\])*'|\"(\\.|[^\\\"])*\")})+", pattern=r"\$\$*[A-Za-z0-9_](?:[A-Za-z0-9_]|::)*(?:->{\$?(?:[a-zA-Z_][a-zA-Z_0-9]*|'(?:\\.|[^'\\])*'|\"(\\.|[^\\\"])*\")}|->\[\$?[0-9a-zA-Z_]+\])+",
), ),
PatternRule( PatternRule(
name=r'length_scalar', name=r'length',
pattern=r"\$#[A-Za-z0-9_](?:[A-Za-z0-9_]|::)*", pattern=r"\$#[A-Za-z0-9_](?:[A-Za-z0-9_]|::)*",
), ),
ContextPatternRule( ContextPatternRule(
name=r'system_scalar', name=r'scalar',
pattern=r"\$[^A-Za-z0-9 %(delim)s](?![A-Za-z0-9_])", pattern=r"\$[^A-Za-z0-9 %(delim)s](?![A-Za-z0-9_])",
fallback=r"\$[^A-Za-z0-9 ](?![A-Za-z0-9_])", fallback=r"\$[^A-Za-z0-9 ](?![A-Za-z0-9_])",
), ),
PatternRule( PatternRule(
name=r'system_array', name=r'array',
pattern=r"@_", pattern=r"@_",
), ),
PatternRule( PatternRule(
@ -63,12 +63,33 @@ class StringGrammar(Grammar):
class PerlGrammar(Grammar): class PerlGrammar(Grammar):
rules = [ rules = [
# heredocs
RegionRule( RegionRule(
name=r'heredoc', name=r'heredoc1',
start=r"<< *(?P<heredoc>[a-zA-Z0-9_]+) *;", start=r"<<(?P<heredoc>[a-zA-Z0-9_]+) *;",
grammar=StringGrammar(), grammar=StringGrammar(),
end=r'^%(heredoc)s$', end=r'^%(heredoc)s$',
), ),
RegionRule(
name=r'heredoc1',
start=r"<< *\"(?P<heredoc>[a-zA-Z0-9_]+)\" *;",
grammar=StringGrammar(),
end=r'^%(heredoc)s$',
),
RegionRule(
name=r'heredoc2',
start=r"<< *'(?P<heredoc>[a-zA-Z0-9_]+)' *;",
grammar=Grammar(),
end=r'^%(heredoc)s$',
),
RegionRule(
name=r'eval_heredoc',
start=r"<< *`(?P<heredoc>[a-zA-Z0-9_]+)` *;",
grammar=StringGrammar(),
end=r'^%(heredoc)s$',
),
# end block
RegionRule( RegionRule(
name=r'endblock', name=r'endblock',
start=r"^__END__|__DATA__ *$", start=r"^__END__|__DATA__ *$",
@ -112,25 +133,29 @@ class PerlGrammar(Grammar):
pattern=r"(?<!->)(?:STDIN|STDERR|STDOUT|and|cmp|continue|do|else|elsif|eq|eval|foreach|for|if|last|my|next|ne|not|or|our|package|require|return|sub|undef|unless|until|use|while)(?![a-zA-Z0-9_])", pattern=r"(?<!->)(?:STDIN|STDERR|STDOUT|and|cmp|continue|do|else|elsif|eq|eval|foreach|for|if|last|my|next|ne|not|or|our|package|require|return|sub|undef|unless|until|use|while)(?![a-zA-Z0-9_])",
), ),
PatternRule( PatternRule(
name=r'bareword_hash_index', name=r'hash_key',
pattern=r'(?<={)[A-Za-z0-9_]+(?=})', pattern=r'(?<={)[A-Za-z0-9_]+(?=})',
), ),
PatternRule( PatternRule(
name=r'bareword_hash_key', name=r'hash_key',
pattern=r'[A-Za-z0-9_]+(?= *=>)', pattern=r'[A-Za-z0-9_]+(?= *=>)',
), ),
PatternRule( PatternRule(
name=r'length_scalar', name=r'length',
pattern=r"\$#[A-Za-z0-9_](?:[A-Za-z0-9_]|::)*", pattern=r"\$#[A-Za-z0-9_](?:[A-Za-z0-9_]|::)*",
), ),
PatternRule( PatternRule(
name=r'system_scalar', name=r'scalar',
pattern=r"\$[][><ab/'\"_@\?#\$!%^|&*()](?![A-Za-z0-9_])", pattern=r"\$[][><ab/'\"_@\?#\$!%^|&*()](?![A-Za-z0-9_])",
), ),
PatternRule( PatternRule(
name=r'system_array', name=r'array',
pattern=r"@_", pattern=r"@_",
), ),
PatternRule(
name=r'function',
pattern=r"\$\$*[A-Za-z0-9_](?:[A-Za-z0-9_]|::)*(?=-> *\()",
),
PatternRule( PatternRule(
name=r'scalar', name=r'scalar',
pattern=r"\$\$*[A-Za-z0-9_](?:[A-Za-z0-9_]|::)*", pattern=r"\$\$*[A-Za-z0-9_](?:[A-Za-z0-9_]|::)*",
@ -144,41 +169,41 @@ class PerlGrammar(Grammar):
pattern=r"%\$*[A-Za-z_](?:[A-Za-z0-9_]|::)*", pattern=r"%\$*[A-Za-z_](?:[A-Za-z0-9_]|::)*",
), ),
PatternRule( PatternRule(
name=r'dereference', name=r'deref',
pattern=r"[@%\$&\*](?={)", pattern=r"[@%\$&\*](?={)",
), ),
RegionRule( RegionRule(
name=r'quoted_region1', name=r'quoted',
start=r'q[rqwx]? *(?P<delim>[^ #])', start=r'q[rqwx]? *(?P<delim>[^ #])',
grammar=Grammar(), grammar=Grammar(),
end=r'%(delim)s', end=r'%(delim)s',
), ),
RegionRule( RegionRule(
name=r'quoted_region2', name=r'quoted',
start=r'q[rqwx]?#', start=r'q[rqwx]?#',
grammar=Grammar(), grammar=Grammar(),
end=r'#', end=r'#',
), ),
RegionRule( RegionRule(
name=r'bracket_quoted_region1', name=r'quoted',
start=r'q[rqwx]? *\(', start=r'q[rqwx]? *\(',
grammar=Grammar(), grammar=Grammar(),
end=r'\)', end=r'\)',
), ),
RegionRule( RegionRule(
name=r'bracket_quoted_region2', name=r'quoted',
start=r'q[rqwx]? *{', start=r'q[rqwx]? *{',
grammar=Grammar(), grammar=Grammar(),
end=r'}', end=r'}',
), ),
RegionRule( RegionRule(
name=r'bracket_quoted_region3', name=r'quoted',
start=r'q[rqwx]? *<', start=r'q[rqwx]? *<',
grammar=Grammar(), grammar=Grammar(),
end=r'>', end=r'>',
), ),
RegionRule( RegionRule(
name=r'bracket_quoted_region4', name=r'quoted',
start=r'q[rqwx]? *\[', start=r'q[rqwx]? *\[',
grammar=Grammar(), grammar=Grammar(),
end=r'\]', end=r'\]',
@ -186,19 +211,19 @@ class PerlGrammar(Grammar):
# match regexes # match regexes
RegionRule( RegionRule(
name=r'match_regex1', name=r'match',
start=r'(?:(?<==~)|(?<=!~)|(?<=\()) *(?P<delim>/)', start=r'(?:(?<==~)|(?<=!~)|(?<=\()) *(?P<delim>/)',
grammar=StringGrammar(), grammar=StringGrammar(),
end=r'/[a-z]*', end=r'/[a-z]*',
), ),
RegionRule( RegionRule(
name=r'match_regex2', name=r'match',
start=r'm *(?P<delim>[^ #a-zA-Z0-9_])', start=r'm *(?P<delim>[^ #a-zA-Z0-9_])',
grammar=StringGrammar(), grammar=StringGrammar(),
end=r'%(delim)s[a-z]*', end=r'%(delim)s[a-z]*',
), ),
RegionRule( RegionRule(
name=r'match_regex3', name=r'match',
start=r'm(?P<delim>#)', start=r'm(?P<delim>#)',
grammar=StringGrammar(), grammar=StringGrammar(),
end=r'#[a-z]*', end=r'#[a-z]*',
@ -206,7 +231,7 @@ class PerlGrammar(Grammar):
# replace regexes # replace regexes
DualRegionRule( DualRegionRule(
name=r'replace_regex1', name=r'replace',
start=r's *(?P<delim>[^ a-zA-Z0-9_])', start=r's *(?P<delim>[^ a-zA-Z0-9_])',
grammar1=StringGrammar(), grammar1=StringGrammar(),
middle=r'%(delim)s', middle=r'%(delim)s',
@ -214,21 +239,39 @@ class PerlGrammar(Grammar):
end=r'%(delim)s[a-z]*', end=r'%(delim)s[a-z]*',
), ),
DualRegionRule( DualRegionRule(
name=r'replace_regex2', name=r'replace',
start=r's#', start=r's(?P<delim>#)',
grammar1=StringGrammar(), grammar1=StringGrammar(),
middle=r'#', middle=r'#',
grammar2=StringGrammar(), grammar2=StringGrammar(),
end=r'#[a-z]*', end=r'#[a-z]*',
), ),
# translate operator
DualRegionRule(
name=r'translate',
start=r'(?:y|tr) *(?P<delim>[^ a-zA-Z0-9_])',
grammar1=Grammar(),
middle=r'%(delim)s',
grammar2=Grammar(),
end=r'%(delim)s[a-z]*',
),
DualRegionRule(
name=r'translate',
start=r'(?:y|tr)#',
grammar1=Grammar(),
middle=r'#',
grammar2=Grammar(),
end=r'#[a-z]*',
),
# some more basic stuff
PatternRule( PatternRule(
name=r'package', name=r'package',
pattern=r"(?<=package )(?:[a-zA-Z_][a-zA-Z_0-9]*::)*[a-zA-Z_][a-zA-Z_0-9]*", pattern=r"(?<=package )(?:[a-zA-Z_][a-zA-Z_0-9]*::)*[a-zA-Z_][a-zA-Z_0-9]*",
), ),
PatternRule( PatternRule(
name=r'sub', name=r'sub',
#pattern=r"(?<=sub )[a-zA-Z_][a-zA-Z_0-9]*(?= *{)",
pattern=r"(?<=sub )[a-zA-Z_][a-zA-Z_0-9]*", pattern=r"(?<=sub )[a-zA-Z_][a-zA-Z_0-9]*",
), ),
PatternRule( PatternRule(
@ -240,33 +283,27 @@ class PerlGrammar(Grammar):
pattern=r'[a-zA-Z_][a-zA-Z0-9_]*:(?!:)', pattern=r'[a-zA-Z_][a-zA-Z0-9_]*:(?!:)',
), ),
PatternRule( PatternRule(
name=r'instance_method', name=r'method',
pattern=r"(?<=->)[a-zA-Z_][a-zA-Z_0-9]*", pattern=r"(?<=->)[a-zA-Z_][a-zA-Z_0-9]*",
), ),
PatternRule( PatternRule(
name=r'static_method', name=r'function',
pattern=r"&?(?:[a-zA-Z_][a-zA-Z_0-9]*::)+[a-zA-Z_][a-zA-Z_0-9]*", pattern=r"&\$*(?:[a-zA-Z_][a-zA-Z_0-9]*::)*[a-zA-Z_][a-zA-Z_0-9]*",
), ),
PatternRule( PatternRule(
name=r'builtin_method', name=r'function',
pattern=r"(?:[a-zA-Z_][a-zA-Z_0-9]*::)*[a-zA-Z_][a-zA-Z_0-9]*(?= *\()",
),
PatternRule(
name=r'builtin',
pattern=r"(?<!->)&?(?:write|warn|wantarray|waitpid|wait|vec|values|utime|use|untie|unshift|unpack|unlink|undef|umask|ucfirst|uc|truncate|times|time|tied|tie|telldir|tell|syswrite|system|sysseek|sysread|sysopen|syscall|symlink|substr|sub|study|stat|srand|sqrt|sprintf|split|splice|sort|socketpair|socket|sleep|sin|shutdown|shmwrite|shmread|shmget|shmctl|shift|setsockopt|setservent|setpwent|setprotoent|setpriority|setpgrp|setnetent|sethostent|setgrent|send|semop|semget|semctl|select|seekdir|seek|scalar|rmdir|rindex|rewinddir|reverse|return|reset|require|rename|ref|redo|recv|readpipe|readlink|readline|readdir|read|rand|quotemeta|push|prototype|printf|print|pos|pop|pipe|package|pack|our|ord|opendir|open|oct|no|next|my|msgsnd|msgrcv|msgget|msgctl|mkdir|map|lstat|log|lock|localtime|local|listen|link|length|lcfirst|lc|last|kill|keys|join|ioctl|int|index|import|hex|grep|goto|gmtime|glob|getsockopt|getsockname|getservent|getservbyport|getservbyname|getpwuid|getpwnam|getpwent|getprotoent|getprotobynumber|getprotobyname|getpriority|getppid|getpgrp|getpeername|getnetent|getnetbyname|getnetbyaddr|getlogin|gethostent|gethostbyname|gethostbyaddr|getgrnam|getgrgid|getgrent|getc|formline|format|fork|flock|fileno|fcntl|exp|exit|exists|exec|eval|eof|endservent|endpwent|endprotoent|endnetent|endhostent|endgrent|each|dump|do|die|delete|defined|dbmopen|dbmclose|crypt|cos|continue|connect|closedir|close|chroot|chr|chown|chop|chomp|chmod|chdir|caller|bless|binmode|bind|atan2|alarm|accept|abs)(?![a-zA-Z0-9_])", pattern=r"(?<!->)&?(?:write|warn|wantarray|waitpid|wait|vec|values|utime|use|untie|unshift|unpack|unlink|undef|umask|ucfirst|uc|truncate|times|time|tied|tie|telldir|tell|syswrite|system|sysseek|sysread|sysopen|syscall|symlink|substr|sub|study|stat|srand|sqrt|sprintf|split|splice|sort|socketpair|socket|sleep|sin|shutdown|shmwrite|shmread|shmget|shmctl|shift|setsockopt|setservent|setpwent|setprotoent|setpriority|setpgrp|setnetent|sethostent|setgrent|send|semop|semget|semctl|select|seekdir|seek|scalar|rmdir|rindex|rewinddir|reverse|return|reset|require|rename|ref|redo|recv|readpipe|readlink|readline|readdir|read|rand|quotemeta|push|prototype|printf|print|pos|pop|pipe|package|pack|our|ord|opendir|open|oct|no|next|my|msgsnd|msgrcv|msgget|msgctl|mkdir|map|lstat|log|lock|localtime|local|listen|link|length|lcfirst|lc|last|kill|keys|join|ioctl|int|index|import|hex|grep|goto|gmtime|glob|getsockopt|getsockname|getservent|getservbyport|getservbyname|getpwuid|getpwnam|getpwent|getprotoent|getprotobynumber|getprotobyname|getpriority|getppid|getpgrp|getpeername|getnetent|getnetbyname|getnetbyaddr|getlogin|gethostent|gethostbyname|gethostbyaddr|getgrnam|getgrgid|getgrent|getc|formline|format|fork|flock|fileno|fcntl|exp|exit|exists|exec|eval|eof|endservent|endpwent|endprotoent|endnetent|endhostent|endgrent|each|dump|do|die|delete|defined|dbmopen|dbmclose|crypt|cos|continue|connect|closedir|close|chroot|chr|chown|chop|chomp|chmod|chdir|caller|bless|binmode|bind|atan2|alarm|accept|abs)(?![a-zA-Z0-9_])",
), ),
PatternRule( PatternRule(
name=r'method', name=r'class',
pattern=r"&(?:[a-zA-Z_][a-zA-Z_0-9]*::)*[a-zA-Z_][a-zA-Z_0-9]*", pattern=r"(?:[a-zA-Z_][a-zA-Z_0-9]*::)*[a-zA-Z_][a-zA-Z_0-9]*",
), ),
PatternRule(
name=r'ref_method', # nested regions
pattern=r"&\$(?:[a-zA-Z_][a-zA-Z_0-9]*::)*[a-zA-Z_][a-zA-Z_0-9]*",
),
PatternRule(
name=r'bareword_method',
pattern=r"(?:[a-zA-Z_][a-zA-Z_0-9]*::)*[a-zA-Z_][a-zA-Z_0-9]*(?= *\()",
),
#PatternRule(
# name=r'delimiter',
# pattern=r"\(|\)|\[|\]|{|}|,|;|->|=>|=|\?|(?<!:):(?!=:)",
#),
RegionRule( RegionRule(
name=r'paren', name=r'paren',
start=r'\(', start=r'\(',
@ -285,12 +322,14 @@ class PerlGrammar(Grammar):
grammar=None, grammar=None,
end=r'\]', end=r'\]',
), ),
# some basic stuff
PatternRule( PatternRule(
name=r'delimiter', name=r'delimiter',
pattern=r",|;|->|=>|=|\?|(?<!:):(?!=:)", pattern=r",|;|->|=>|=|\?|(?<!:):(?!=:)",
), ),
PatternRule( PatternRule(
name=r'unary_operator', name=r'operator',
pattern=r"\+=|-=|\*=|/=|//=|%=|&=\|\^=|>>=|<<=|\*\*=", pattern=r"\+=|-=|\*=|/=|//=|%=|&=\|\^=|>>=|<<=|\*\*=",
), ),
PatternRule( PatternRule(

127
test3.py
View File

@ -17,99 +17,86 @@ for i in range(0, len(color_list)):
color_dict[color_names[i]] = color_list[i] color_dict[color_names[i]] = color_list[i]
token_colors = { token_colors = {
# basic stuff
'escaped': 'lpurple', 'escaped': 'lpurple',
'null': 'white', 'null': 'white',
'delimiter': 'white', 'delimiter': 'white',
'pod.start': 'lred',
'pod.null': 'lred',
'pod.end': 'lred',
'pod.header': 'lpurple',
'pod.indent_level': 'lpurple',
'pod.item_entry': 'lpurple',
'pod.format': 'lpurple',
'pod.encoding_type': 'lpurple',
'sub': 'lcyan', 'sub': 'lcyan',
'number': 'white', 'number': 'white',
'operator': 'white', 'operator': 'white',
'heredoc': 'lgreen',
'endblock': 'lred', 'endblock': 'lred',
'pod': 'lred', 'keyword': 'lpurple',
'scalar': 'yellow',
'array': 'yellow',
'deref': 'yellow',
'hash': 'yellow',
'hash_key': 'lgreen',
'comment': 'lred', 'comment': 'lred',
#'string1': 'lgreen', 'function': 'lcyan',
'builtin': 'lpurple',
'method': 'lcyan',
'bareword': 'white',
'label': 'lcyan',
'package': 'lcyan',
'class': 'lcyan',
'use': 'lcyan',
'method': 'lcyan',
# heredoc
'heredoc1.start': 'lgreen',
'heredoc1.null': 'lgreen',
'heredoc1.end': 'lgreen',
'heredoc2.start': 'lgreen',
'heredoc2.null': 'lgreen',
'heredoc2.end': 'lgreen',
'eval_heredoc.start': 'lcyan',
'eval_heredoc.null': 'lcyan',
'eval_heredoc.end': 'lcyan',
# pod
'pod.start': 'lred',
'pod.null': 'lred',
'pod.entry': 'lpurple',
'pod.end': 'lred',
# "" strings
'string1.start': 'lgreen', 'string1.start': 'lgreen',
'string1.null': 'lgreen', 'string1.null': 'lgreen',
'string1.escaped': 'lpurple', 'string1.escaped': 'lpurple',
#'string1.scalar': 'yellow', 'string1.deref': 'yellow',
#'string1.system_scalar': 'yellow',
'string1.hash_deref': 'yellow',
#'string1.hash_bareword_index': 'lgreen',
'string1.end': 'lgreen', 'string1.end': 'lgreen',
#'string2': 'lgreen',
# '' strings
'string2.start': 'lgreen', 'string2.start': 'lgreen',
'string2.null': 'lgreen', 'string2.null': 'lgreen',
'string2.end': 'lgreen', 'string2.end': 'lgreen',
# `` strings
'evalstring': 'lcyan', 'evalstring': 'lcyan',
'default_string': 'lgreen',
'keyword': 'lpurple',
'length_scalar': 'yellow',
'system_scalar': 'yellow',
'system_array': 'yellow',
'scalar': 'yellow',
'dereference': 'yellow',
'array': 'yellow',
'hash': 'yellow',
'bareword_hash_index': 'lgreen',
# quoted region # quoted region
'quoted_region1': 'lcyan', 'quoted': 'lcyan',
'quoted_region1.start': 'lcyan', 'quoted.start': 'lcyan',
'quoted_region1.null': 'lcyan', 'quoted.null': 'lcyan',
'quoted_region1.end': 'lcyan', 'quoted.end': 'lcyan',
'quoted_region2': 'lcyan',
'quoted_region2.start': 'lcyan',
'quoted_region2.null': 'lcyan',
'quoted_region2.end': 'lcyan',
# match regex # match regex
'match_regex1.start': 'lcyan', 'match.start': 'lcyan',
'match_regex1.end': 'lcyan', 'match.end': 'lcyan',
'match_regex1.null': 'lcyan', 'match.null': 'lcyan',
'match_regex2.start': 'lcyan',
'match_regex2.end': 'lcyan',
'match_regex2.null': 'lcyan',
'match_regex3.start': 'lcyan',
'match_regex3.end': 'lcyan',
'match_regex3.null': 'lcyan',
# replace regex # replace regex
'replace_regex1.start': 'lcyan', 'replace.start': 'lcyan',
'replace_regex1.middle': 'lcyan', 'replace.middle': 'lcyan',
'replace_regex1.end': 'lcyan', 'replace.end': 'lcyan',
'replace_regex1.null': 'lcyan', 'replace.null': 'lcyan',
'replace_regex2.start': 'lcyan',
'replace_regex2.middle': 'lcyan', # translate regex
'replace_regex2.end': 'lcyan', 'translate.start': 'lpurple',
'replace_regex2.null': 'lcyan', 'translate.middle': 'lpurple',
'translate.end': 'lpurple',
'translate.null': 'lpurple',
#
'bareword_hash_key': 'lgreen',
'interpolated_scalar': 'yellow',
'interpolated_system_scalar': 'yellow',
'interpolated_array': 'yellow',
'interpolated_system_array': 'yellow',
'interpolated_hash': 'yellow',
'label': 'lcyan',
'package': 'lcyan',
'use': 'lcyan',
'method': 'lcyan',
'methodref': 'lcyan',
'method_declaration': 'lcyan',
'instance_method': 'lcyan',
'static_method': 'lcyan',
'builtin_method': 'lpurple',
'bareword_method': 'lcyan',
#'bareword': 'yellow',
'bizzaro': 'lpurple',
} }
paths = sys.argv[1:] paths = sys.argv[1:]