Add Janet Lexer (#2557)

pygments · Nov 19, 2023 · d92f2cc · d92f2cc
1 parent 446b45c
commit d92f2cc
Show file tree

Hide file tree

Showing 87 changed files with 1,130 additions and 1 deletion.
diff --git a/AUTHORS b/AUTHORS
@@ -46,6 +46,7 @@ Other contributors, listed alphabetically, are:
 * chebee7i -- Python traceback lexer improvements
 * Hiram Chirino -- Scaml and Jade lexers
 * Mauricio Caceres -- SAS and Stata lexers.
+* Michael Camilleri, John Gabriele, sogaiu -- Janet lexer
 * Ian Cooper -- VGL lexer
 * David Corbett -- Inform, Jasmin, JSGF, Snowball, and TADS 3 lexers
 * Leaf Corcoran -- MoonScript lexer

diff --git a/pygments/lexers/_mapping.py b/pygments/lexers/_mapping.py
@@ -234,6 +234,7 @@
     'JMESPathLexer': ('pygments.lexers.jmespath', 'JMESPath', ('jmespath', 'jp'), ('*.jp',), ()),
     'JSLTLexer': ('pygments.lexers.jslt', 'JSLT', ('jslt',), ('*.jslt',), ('text/x-jslt',)),
     'JagsLexer': ('pygments.lexers.modeling', 'JAGS', ('jags',), ('*.jag', '*.bug'), ()),
+    'JanetLexer': ('pygments.lexers.lisp', 'Janet', ('janet',), ('*.janet', '*.jdn'), ('text/x-janet', 'application/x-janet')),
     'JasminLexer': ('pygments.lexers.jvm', 'Jasmin', ('jasmin', 'jasminxt'), ('*.j',), ()),
     'JavaLexer': ('pygments.lexers.jvm', 'Java', ('java',), ('*.java',), ('text/x-java',)),
     'JavascriptDjangoLexer': ('pygments.lexers.templates', 'JavaScript+Django/Jinja', ('javascript+django', 'js+django', 'javascript+jinja', 'js+jinja'), ('*.js.j2', '*.js.jinja2'), ('application/x-javascript+django', 'application/x-javascript+jinja', 'text/x-javascript+django', 'text/x-javascript+jinja', 'text/javascript+django', 'text/javascript+jinja')),

diff --git a/pygments/lexers/lisp.py b/pygments/lexers/lisp.py
@@ -20,7 +20,7 @@
 
 __all__ = ['SchemeLexer', 'CommonLispLexer', 'HyLexer', 'RacketLexer',
            'NewLispLexer', 'EmacsLispLexer', 'ShenLexer', 'CPSALexer',
-           'XtlangLexer', 'FennelLexer']
+           'XtlangLexer', 'FennelLexer', 'JanetLexer']
 
 
 class SchemeLexer(RegexLexer):
@@ -2848,3 +2848,320 @@ class FennelLexer(RegexLexer):
             (r'#', Punctuation),
         ]
     }
+
+
+class JanetLexer(RegexLexer):
+    """A lexer for the Janet programming language.
+
+    .. versionadded:: 2.17.0
+    """
+    name = 'Janet'
+    url =  'https://janet-lang.org/'
+    aliases = ['janet']
+    filenames = ['*.janet', '*.jdn']
+    mimetypes = ['text/x-janet', 'application/x-janet']
+
+    # XXX: gets too slow
+    #flags = re.MULTILINE | re.VERBOSE
+
+    special_forms = (
+        'break', 'def', 'do', 'fn', 'if', 'quote', 'quasiquote', 'splice',
+        'set', 'unquote', 'upscope', 'var', 'while'
+    )
+
+    builtin_macros = (
+        '%=', '*=', '++', '+=', '--', '-=', '->', '->>', '-?>',
+        '-?>>', '/=', 'and', 'as->', 'as-macro', 'as?->',
+        'assert', 'case', 'catseq', 'chr', 'comment', 'compif',
+        'comptime', 'compwhen', 'cond', 'coro', 'def-',
+        'default', 'defdyn', 'defer', 'defmacro', 'defmacro-',
+        'defn', 'defn-', 'delay', 'doc', 'each', 'eachk',
+        'eachp', 'edefer', 'ev/do-thread', 'ev/gather',
+        'ev/spawn', 'ev/spawn-thread', 'ev/with-deadline',
+        'ffi/defbind', 'fiber-fn', 'for', 'forever', 'forv',
+        'generate', 'if-let', 'if-not', 'if-with', 'import',
+        'juxt', 'label', 'let', 'loop', 'match', 'or', 'prompt',
+        'protect', 'repeat', 'seq', 'short-fn', 'tabseq',
+        'toggle', 'tracev', 'try', 'unless', 'use', 'var-',
+        'varfn', 'when', 'when-let', 'when-with', 'with',
+        'with-dyns', 'with-syms', 'with-vars',
+        # obsolete builtin macros
+        'eachy'
+    )
+
+    builtin_functions = (
+        '%', '*', '+', '-', '/', '<', '<=', '=', '>', '>=',
+        'abstract?', 'accumulate', 'accumulate2', 'all',
+        'all-bindings', 'all-dynamics', 'any?', 'apply',
+        'array', 'array/clear', 'array/concat', 'array/ensure',
+        'array/fill', 'array/insert', 'array/new',
+        'array/new-filled', 'array/peek', 'array/pop',
+        'array/push', 'array/remove', 'array/slice',
+        'array/trim', 'array/weak', 'array?', 'asm',
+        'bad-compile', 'bad-parse', 'band', 'blshift', 'bnot',
+        'boolean?', 'bor', 'brshift', 'brushift', 'buffer',
+        'buffer/bit', 'buffer/bit-clear', 'buffer/bit-set',
+        'buffer/bit-toggle', 'buffer/blit', 'buffer/clear',
+        'buffer/fill', 'buffer/format', 'buffer/from-bytes',
+        'buffer/new', 'buffer/new-filled', 'buffer/popn',
+        'buffer/push', 'buffer/push-at', 'buffer/push-byte',
+        'buffer/push-string', 'buffer/push-word',
+        'buffer/slice', 'buffer/trim', 'buffer?', 'bxor',
+        'bytes?', 'cancel', 'cfunction?', 'cli-main', 'cmp',
+        'comp', 'compare', 'compare<', 'compare<=', 'compare=',
+        'compare>', 'compare>=', 'compile', 'complement',
+        'count', 'curenv', 'debug', 'debug/arg-stack',
+        'debug/break', 'debug/fbreak', 'debug/lineage',
+        'debug/stack', 'debug/stacktrace', 'debug/step',
+        'debug/unbreak', 'debug/unfbreak', 'debugger',
+        'debugger-on-status', 'dec', 'deep-not=', 'deep=',
+        'defglobal', 'describe', 'dictionary?', 'disasm',
+        'distinct', 'div', 'doc*', 'doc-format', 'doc-of',
+        'dofile', 'drop', 'drop-until', 'drop-while', 'dyn',
+        'eflush', 'empty?', 'env-lookup', 'eprin', 'eprinf',
+        'eprint', 'eprintf', 'error', 'errorf',
+        'ev/acquire-lock', 'ev/acquire-rlock',
+        'ev/acquire-wlock', 'ev/all-tasks', 'ev/call',
+        'ev/cancel', 'ev/capacity', 'ev/chan', 'ev/chan-close',
+        'ev/chunk', 'ev/close', 'ev/count', 'ev/deadline',
+        'ev/full', 'ev/give', 'ev/give-supervisor', 'ev/go',
+        'ev/lock', 'ev/read', 'ev/release-lock',
+        'ev/release-rlock', 'ev/release-wlock', 'ev/rselect',
+        'ev/rwlock', 'ev/select', 'ev/sleep', 'ev/take',
+        'ev/thread', 'ev/thread-chan', 'ev/write', 'eval',
+        'eval-string', 'even?', 'every?', 'extreme', 'false?',
+        'ffi/align', 'ffi/call', 'ffi/calling-conventions',
+        'ffi/close', 'ffi/context', 'ffi/free', 'ffi/jitfn',
+        'ffi/lookup', 'ffi/malloc', 'ffi/native',
+        'ffi/pointer-buffer', 'ffi/pointer-cfunction',
+        'ffi/read', 'ffi/signature', 'ffi/size', 'ffi/struct',
+        'ffi/trampoline', 'ffi/write', 'fiber/can-resume?',
+        'fiber/current', 'fiber/getenv', 'fiber/last-value',
+        'fiber/maxstack', 'fiber/new', 'fiber/root',
+        'fiber/setenv', 'fiber/setmaxstack', 'fiber/status',
+        'fiber?', 'file/close', 'file/flush', 'file/lines',
+        'file/open', 'file/read', 'file/seek', 'file/tell',
+        'file/temp', 'file/write', 'filter', 'find',
+        'find-index', 'first', 'flatten', 'flatten-into',
+        'flush', 'flycheck', 'freeze', 'frequencies',
+        'from-pairs', 'function?', 'gccollect', 'gcinterval',
+        'gcsetinterval', 'gensym', 'get', 'get-in', 'getline',
+        'getproto', 'group-by', 'has-key?', 'has-value?',
+        'hash', 'idempotent?', 'identity', 'import*', 'in',
+        'inc', 'index-of', 'indexed?', 'int/s64',
+        'int/to-bytes', 'int/to-number', 'int/u64', 'int?',
+        'interleave', 'interpose', 'invert', 'juxt*', 'keep',
+        'keep-syntax', 'keep-syntax!', 'keys', 'keyword',
+        'keyword/slice', 'keyword?', 'kvs', 'last', 'length',
+        'lengthable?', 'load-image', 'macex', 'macex1',
+        'maclintf', 'make-env', 'make-image', 'map', 'mapcat',
+        'marshal', 'math/abs', 'math/acos', 'math/acosh',
+        'math/asin', 'math/asinh', 'math/atan', 'math/atan2',
+        'math/atanh', 'math/cbrt', 'math/ceil', 'math/cos',
+        'math/cosh', 'math/erf', 'math/erfc', 'math/exp',
+        'math/exp2', 'math/expm1', 'math/floor', 'math/gamma',
+        'math/gcd', 'math/hypot', 'math/lcm', 'math/log',
+        'math/log-gamma', 'math/log10', 'math/log1p',
+        'math/log2', 'math/next', 'math/pow', 'math/random',
+        'math/rng', 'math/rng-buffer', 'math/rng-int',
+        'math/rng-uniform', 'math/round', 'math/seedrandom',
+        'math/sin', 'math/sinh', 'math/sqrt', 'math/tan',
+        'math/tanh', 'math/trunc', 'max', 'max-of', 'mean',
+        'memcmp', 'merge', 'merge-into', 'merge-module', 'min',
+        'min-of', 'mod', 'module/add-paths',
+        'module/expand-path', 'module/find', 'module/value',
+        'nan?', 'nat?', 'native', 'neg?', 'net/accept',
+        'net/accept-loop', 'net/address', 'net/address-unpack',
+        'net/chunk', 'net/close', 'net/connect', 'net/flush',
+        'net/listen', 'net/localname', 'net/peername',
+        'net/read', 'net/recv-from', 'net/send-to',
+        'net/server', 'net/setsockopt', 'net/shutdown',
+        'net/write', 'next', 'nil?', 'not', 'not=', 'number?',
+        'odd?', 'one?', 'os/arch', 'os/cd', 'os/chmod',
+        'os/clock', 'os/compiler', 'os/cpu-count',
+        'os/cryptorand', 'os/cwd', 'os/date', 'os/dir',
+        'os/environ', 'os/execute', 'os/exit', 'os/getenv',
+        'os/isatty', 'os/link', 'os/lstat', 'os/mkdir',
+        'os/mktime', 'os/open', 'os/perm-int', 'os/perm-string',
+        'os/pipe', 'os/posix-exec', 'os/posix-fork',
+        'os/proc-close', 'os/proc-kill', 'os/proc-wait',
+        'os/readlink', 'os/realpath', 'os/rename', 'os/rm',
+        'os/rmdir', 'os/setenv', 'os/shell', 'os/sigaction',
+        'os/sleep', 'os/spawn', 'os/stat', 'os/strftime',
+        'os/symlink', 'os/time', 'os/touch', 'os/umask',
+        'os/which', 'pairs', 'parse', 'parse-all',
+        'parser/byte', 'parser/clone', 'parser/consume',
+        'parser/eof', 'parser/error', 'parser/flush',
+        'parser/has-more', 'parser/insert', 'parser/new',
+        'parser/produce', 'parser/state', 'parser/status',
+        'parser/where', 'partial', 'partition', 'partition-by',
+        'peg/compile', 'peg/find', 'peg/find-all', 'peg/match',
+        'peg/replace', 'peg/replace-all', 'pos?', 'postwalk',
+        'pp', 'prewalk', 'prin', 'prinf', 'print', 'printf',
+        'product', 'propagate', 'put', 'put-in', 'quit',
+        'range', 'reduce', 'reduce2', 'repl', 'require',
+        'resume', 'return', 'reverse', 'reverse!',
+        'run-context', 'sandbox', 'scan-number', 'setdyn',
+        'signal', 'slice', 'slurp', 'some', 'sort', 'sort-by',
+        'sorted', 'sorted-by', 'spit', 'string',
+        'string/ascii-lower', 'string/ascii-upper',
+        'string/bytes', 'string/check-set', 'string/find',
+        'string/find-all', 'string/format', 'string/from-bytes',
+        'string/has-prefix?', 'string/has-suffix?',
+        'string/join', 'string/repeat', 'string/replace',
+        'string/replace-all', 'string/reverse', 'string/slice',
+        'string/split', 'string/trim', 'string/triml',
+        'string/trimr', 'string?', 'struct', 'struct/getproto',
+        'struct/proto-flatten', 'struct/to-table',
+        'struct/with-proto', 'struct?', 'sum', 'symbol',
+        'symbol/slice', 'symbol?', 'table', 'table/clear',
+        'table/clone', 'table/getproto', 'table/new',
+        'table/proto-flatten', 'table/rawget', 'table/setproto',
+        'table/to-struct', 'table/weak', 'table/weak-keys',
+        'table/weak-values', 'table?', 'take', 'take-until',
+        'take-while', 'thaw', 'trace', 'true?', 'truthy?',
+        'tuple', 'tuple/brackets', 'tuple/setmap',
+        'tuple/slice', 'tuple/sourcemap', 'tuple/type',
+        'tuple?', 'type', 'unmarshal', 'untrace', 'update',
+        'update-in', 'values', 'varglobal', 'walk',
+        'warn-compile', 'xprin', 'xprinf', 'xprint', 'xprintf',
+        'yield', 'zero?', 'zipcoll',
+        # obsolete builtin functions
+        'tarray/buffer', 'tarray/copy-bytes', 'tarray/length',
+        'tarray/new', 'tarray/properties', 'tarray/slice',
+        'tarray/swap-bytes', 'thread/close', 'thread/current',
+        'thread/exit', 'thread/new', 'thread/receive',
+        'thread/send'
+    )
+
+    builtin_variables = (
+        'debugger-env', 'default-peg-grammar', 'janet/build',
+        'janet/config-bits', 'janet/version', 'load-image-dict',
+        'make-image-dict', 'math/-inf', 'math/e', 'math/inf',
+        'math/int-max', 'math/int-min', 'math/int32-max',
+        'math/int32-min', 'math/nan', 'math/pi', 'module/cache',
+        'module/loaders', 'module/loading', 'module/paths',
+        'root-env', 'stderr', 'stdin', 'stdout'
+    )
+
+    constants = (
+        'false', 'nil', 'true'
+    )
+
+    # XXX: this form not usable to pass to `suffix=`
+    #_token_end = r'''
+    #  (?=            # followed by one of:
+    #      \s           # whitespace
+    #    | \#           # comment
+    #    | [)\]]        # end delimiters
+    #    | $            # end of file
+    #  )
+    #'''
+
+    # ...so, express it like this
+    _token_end = r'(?=\s|#|[)\]]|$)'
+
+    _first_char = r'[a-zA-Z!$%&*+\-./<=>?@^_]'
+    _rest_char = rf'([0-9:]|{_first_char})'
+
+    valid_name = rf'{_first_char}({_rest_char})*'
+
+    _radix_unit = r'[0-9a-zA-Z][0-9a-zA-Z_]*'
+
+    # exponent marker, optional sign, one or more alphanumeric
+    _radix_exp = r'&[+-]?[0-9a-zA-Z]+'
+
+    # 2af3__bee_
+    _hex_unit = r'[0-9a-fA-F][0-9a-fA-F_]*'
+
+    # 12_000__
+    _dec_unit = r'[0-9][0-9_]*'
+
+    # E-23
+    # lower or uppercase e, optional sign, one or more digits
+    _dec_exp = r'[eE][+-]?[0-9]+'
+
+    tokens = {
+        'root': [
+            (r'#.*$', Comment.Single),
+
+            (r'\s+', Whitespace),
+
+            # radix number
+            (rf'''(?x)
+                  [+-]? [0-9]{{1,2}} r {_radix_unit} \. ({_radix_unit})?
+                  ({_radix_exp})?
+               ''',
+             Number),
+
+            (rf'''(?x)
+                  [+-]? [0-9]{{1,2}} r (\.)? {_radix_unit}
+                  ({_radix_exp})?
+               ''',
+             Number),
+
+            # hex number
+            (rf'(?x) [+-]? 0x {_hex_unit} \. ({_hex_unit})?',
+             Number.Hex),
+
+            (rf'(?x) [+-]? 0x (\.)? {_hex_unit}',
+             Number.Hex),
+
+            # decimal number
+            (rf'(?x) [+-]? {_dec_unit} \. ({_dec_unit})? ({_dec_exp})?',
+             Number.Float),
+
+            (rf'(?x) [+-]? (\.)? {_dec_unit} ({_dec_exp})?',
+             Number.Float),
+
+            # strings and buffers
+            (r'@?"', String, 'string'),
+
+            # long-strings and long-buffers
+            #
+            #   non-empty content enclosed by a pair of n-backticks
+            #   with optional leading @
+            (r'@?(`+)(.|\n)+?\1', String),
+
+            # things that hang out on front
+            #
+            #   ' ~ , ; |
+            (r"['~,;|]", Operator),
+
+            # collection delimiters
+            #
+            #   @( ( )
+            #   @[ [ ]
+            #   @{ { }
+            (r'@?[(\[{]|[)\]}]', Punctuation),
+
+            # constants
+            (words(constants, suffix=_token_end), Keyword.Constants),
+
+            # keywords
+            (rf'(:({_rest_char})+|:)', Name.Constant),
+
+            # symbols
+            (words(builtin_variables, suffix=_token_end),
+             Name.Variable.Global),
+
+            (words(special_forms, prefix=r'(?<=\()', suffix=_token_end),
+             Keyword.Reserved),
+
+            (words(builtin_macros, prefix=r'(?<=\()', suffix=_token_end),
+             Name.Builtin),
+
+            (words(builtin_functions, prefix=r'(?<=\()', suffix=_token_end),
+             Name.Function),
+
+            # other symbols
+            (valid_name, Name.Variable),
+        ],
+        'string': [
+            (r'\\(u[0-9a-fA-F]{4}|U[0-9a-fA-F]{6})', String.Escape),
+            (r'\\x[0-9a-fA-F]{2}', String.Escape),
+            (r'\\.', String.Escape),
+            (r'"', String, '#pop'),
+            (r'[^\\"]+', String),
+        ]
+    }
diff --git a/tests/snippets/janet/bool_lit-false.txt b/tests/snippets/janet/bool_lit-false.txt
@@ -0,0 +1,6 @@
+---input---
+false
+
+---tokens---
+'false'       Keyword.Constants
+'\n'          Text.Whitespace
diff --git a/tests/snippets/janet/bool_lit-true.txt b/tests/snippets/janet/bool_lit-true.txt
@@ -0,0 +1,6 @@
+---input---
+true
+
+---tokens---
+'true'        Keyword.Constants
+'\n'          Text.Whitespace
diff --git a/tests/snippets/janet/buf_lit-multiline.txt b/tests/snippets/janet/buf_lit-multiline.txt
@@ -0,0 +1,9 @@
+---input---
+@"this is the first line
+and what is this one?"
+
+---tokens---
+'@"'          Literal.String
+'this is the first line\nand what is this one?' Literal.String
+'"'           Literal.String
+'\n'          Text.Whitespace
diff --git a/tests/snippets/janet/buf_lit-simple.txt b/tests/snippets/janet/buf_lit-simple.txt
@@ -0,0 +1,8 @@
+---input---
+@"good bye"
+
+---tokens---
+'@"'          Literal.String
+'good bye'    Literal.String
+'"'           Literal.String
+'\n'          Text.Whitespace
diff --git a/tests/snippets/janet/buf_lit-with-escape.txt b/tests/snippets/janet/buf_lit-with-escape.txt
@@ -0,0 +1,12 @@
+---input---
+@"ant\fbee\rcougar"
+
+---tokens---
+'@"'          Literal.String
+'ant'         Literal.String
+'\\f'         Literal.String.Escape
+'bee'         Literal.String
+'\\r'         Literal.String.Escape
+'cougar'      Literal.String
+'"'           Literal.String
+'\n'          Text.Whitespace