tartrazine/lexers/python.xml

<lexer>
  <config>
    <name>Python</name>
    <alias>python</alias>
    <alias>py</alias>
    <alias>sage</alias>
    <alias>python3</alias>
    <alias>py3</alias>
    <filename>*.py</filename>
    <filename>*.pyi</filename>
    <filename>*.pyw</filename>
    <filename>*.jy</filename>
    <filename>*.sage</filename>
    <filename>*.sc</filename>
    <filename>SConstruct</filename>
    <filename>SConscript</filename>
    <filename>*.bzl</filename>
    <filename>BUCK</filename>
    <filename>BUILD</filename>
    <filename>BUILD.bazel</filename>
    <filename>WORKSPACE</filename>
    <filename>WORKSPACE.bzlmod</filename>
    <filename>WORKSPACE.bazel</filename>
    <filename>MODULE.bazel</filename>
    <filename>REPO.bazel</filename>
    <filename>*.tac</filename>
    <mime_type>text/x-python</mime_type>
    <mime_type>application/x-python</mime_type>
    <mime_type>text/x-python3</mime_type>
    <mime_type>application/x-python3</mime_type>
  </config>
  <rules>
    <state name="numbers">
      <rule pattern="(\d(?:_?\d)*\.(?:\d(?:_?\d)*)?|(?:\d(?:_?\d)*)?\.\d(?:_?\d)*)([eE][+-]?\d(?:_?\d)*)?">
        <token type="LiteralNumberFloat"/>
      </rule>
      <rule pattern="\d(?:_?\d)*[eE][+-]?\d(?:_?\d)*j?">
        <token type="LiteralNumberFloat"/>
      </rule>
      <rule pattern="0[oO](?:_?[0-7])+">
        <token type="LiteralNumberOct"/>
      </rule>
      <rule pattern="0[bB](?:_?[01])+">
        <token type="LiteralNumberBin"/>
      </rule>
      <rule pattern="0[xX](?:_?[a-fA-F0-9])+">
        <token type="LiteralNumberHex"/>
      </rule>
      <rule pattern="\d(?:_?\d)*">
        <token type="LiteralNumberInteger"/>
      </rule>
    </state>
    <state name="expr">
      <rule pattern="(?i)(rf|fr)(&#34;&#34;&#34;)">
        <bygroups>
          <token type="LiteralStringAffix"/>
          <token type="LiteralStringDouble"/>
        </bygroups>
        <combined state="rfstringescape" state="tdqf"/>
      </rule>
      <rule pattern="(?i)(rf|fr)(&#39;&#39;&#39;)">
        <bygroups>
          <token type="LiteralStringAffix"/>
          <token type="LiteralStringSingle"/>
        </bygroups>
        <combined state="rfstringescape" state="tsqf"/>
      </rule>
      <rule pattern="(?i)(rf|fr)(&#34;)">
        <bygroups>
          <token type="LiteralStringAffix"/>
          <token type="LiteralStringDouble"/>
        </bygroups>
        <combined state="rfstringescape" state="dqf"/>
      </rule>
      <rule pattern="(?i)(rf|fr)(&#39;)">
        <bygroups>
          <token type="LiteralStringAffix"/>
          <token type="LiteralStringSingle"/>
        </bygroups>
        <combined state="rfstringescape" state="sqf"/>
      </rule>
      <rule pattern="([fF])(&#34;&#34;&#34;)">
        <bygroups>
          <token type="LiteralStringAffix"/>
          <token type="LiteralStringDouble"/>
        </bygroups>
        <combined state="fstringescape" state="tdqf"/>
      </rule>
      <rule pattern="([fF])(&#39;&#39;&#39;)">
        <bygroups>
          <token type="LiteralStringAffix"/>
          <token type="LiteralStringSingle"/>
        </bygroups>
        <combined state="fstringescape" state="tsqf"/>
      </rule>
      <rule pattern="([fF])(&#34;)">
        <bygroups>
          <token type="LiteralStringAffix"/>
          <token type="LiteralStringDouble"/>
        </bygroups>
        <combined state="fstringescape" state="dqf"/>
      </rule>
      <rule pattern="([fF])(&#39;)">
        <bygroups>
          <token type="LiteralStringAffix"/>
          <token type="LiteralStringSingle"/>
        </bygroups>
        <combined state="fstringescape" state="sqf"/>
      </rule>
      <rule pattern="(?i)(rb|br|r)(&#34;&#34;&#34;)">
        <bygroups>
          <token type="LiteralStringAffix"/>
          <token type="LiteralStringDouble"/>
        </bygroups>
        <push state="tdqs"/>
      </rule>
      <rule pattern="(?i)(rb|br|r)(&#39;&#39;&#39;)">
        <bygroups>
          <token type="LiteralStringAffix"/>
          <token type="LiteralStringSingle"/>
        </bygroups>
        <push state="tsqs"/>
      </rule>
      <rule pattern="(?i)(rb|br|r)(&#34;)">
        <bygroups>
          <token type="LiteralStringAffix"/>
          <token type="LiteralStringDouble"/>
        </bygroups>
        <push state="dqs"/>
      </rule>
      <rule pattern="(?i)(rb|br|r)(&#39;)">
        <bygroups>
          <token type="LiteralStringAffix"/>
          <token type="LiteralStringSingle"/>
        </bygroups>
        <push state="sqs"/>
      </rule>
      <rule pattern="([uUbB]?)(&#34;&#34;&#34;)">
        <bygroups>
          <token type="LiteralStringAffix"/>
          <token type="LiteralStringDouble"/>
        </bygroups>
        <combined state="stringescape" state="tdqs"/>
      </rule>
      <rule pattern="([uUbB]?)(&#39;&#39;&#39;)">
        <bygroups>
          <token type="LiteralStringAffix"/>
          <token type="LiteralStringSingle"/>
        </bygroups>
        <combined state="stringescape" state="tsqs"/>
      </rule>
      <rule pattern="([uUbB]?)(&#34;)">
        <bygroups>
          <token type="LiteralStringAffix"/>
          <token type="LiteralStringDouble"/>
        </bygroups>
        <combined state="stringescape" state="dqs"/>
      </rule>
      <rule pattern="([uUbB]?)(&#39;)">
        <bygroups>
          <token type="LiteralStringAffix"/>
          <token type="LiteralStringSingle"/>
        </bygroups>
        <combined state="stringescape" state="sqs"/>
      </rule>
      <rule pattern="[^\S\n]+">
        <token type="Text"/>
      </rule>
      <rule>
        <include state="numbers"/>
      </rule>
      <rule pattern="!=|==|&lt;&lt;|&gt;&gt;|:=|[-~+/*%=&lt;&gt;&amp;^|.]">
        <token type="Operator"/>
      </rule>
      <rule pattern="[]{}:(),;[]">
        <token type="Punctuation"/>
      </rule>
      <rule pattern="(in|is|and|or|not)\b">
        <token type="OperatorWord"/>
      </rule>
      <rule>
        <include state="expr-keywords"/>
      </rule>
      <rule>
        <include state="builtins"/>
      </rule>
      <rule>
        <include state="magicfuncs"/>
      </rule>
      <rule>
        <include state="magicvars"/>
      </rule>
      <rule>
        <include state="name"/>
      </rule>
    </state>
    <state name="fstrings-double">
      <rule pattern="\}">
        <token type="LiteralStringInterpol"/>
      </rule>
      <rule pattern="\{">
        <token type="LiteralStringInterpol"/>
        <push state="expr-inside-fstring"/>
      </rule>
      <rule pattern="[^\\\&#39;&#34;{}\n]+">
        <token type="LiteralStringDouble"/>
      </rule>
      <rule pattern="[\&#39;&#34;\\]">
        <token type="LiteralStringDouble"/>
      </rule>
    </state>
    <state name="keywords">
      <rule pattern="(yield from|nonlocal|continue|finally|except|lambda|assert|global|return|raise|yield|while|break|await|async|pass|else|elif|with|try|for|del|as|if|match|case)\b">
        <token type="Keyword"/>
      </rule>
      <rule pattern="(False|True|None)\b">
        <token type="KeywordConstant"/>
      </rule>
    </state>
    <state name="dqs">
      <rule pattern="&#34;">
        <token type="LiteralStringDouble"/>
        <pop depth="1"/>
      </rule>
      <rule pattern="\\\\|\\&#34;|\\\n">
        <token type="LiteralStringEscape"/>
      </rule>
      <rule>
        <include state="strings-double"/>
      </rule>
    </state>
    <state name="fromimport">
      <rule pattern="(\s+)(import)\b">
        <bygroups>
          <token type="Text"/>
          <token type="KeywordNamespace"/>
        </bygroups>
        <pop depth="1"/>
      </rule>
      <rule pattern="\.">
        <token type="NameNamespace"/>
      </rule>
      <rule pattern="None\b">
        <token type="NameBuiltinPseudo"/>
        <pop depth="1"/>
      </rule>
      <rule pattern="[_\p{L}][_\p{L}\p{N}]*">
        <token type="NameNamespace"/>
      </rule>
      <rule>
        <pop depth="1"/>
      </rule>
    </state>
    <state name="builtins">
      <rule pattern="(?&lt;!\.)(staticmethod|classmethod|memoryview|__import__|issubclass|isinstance|frozenset|bytearray|enumerate|reversed|property|compile|complex|delattr|hasattr|setattr|globals|getattr|divmod|filter|locals|format|object|sorted|slice|print|bytes|range|input|tuple|round|super|float|eval|list|dict|repr|type|vars|hash|next|bool|open|iter|oct|pow|min|zip|max|map|bin|len|set|any|dir|all|abs|str|sum|chr|int|hex|ord|id)\b">
        <token type="NameBuiltin"/>
      </rule>
      <rule pattern="(?&lt;!\.)(self|Ellipsis|NotImplemented|cls)\b">
        <token type="NameBuiltinPseudo"/>
      </rule>
      <rule pattern="(?&lt;!\.)(PendingDeprecationWarning|ConnectionAbortedError|ConnectionRefusedError|UnicodeTranslateError|ConnectionResetError|ModuleNotFoundError|NotImplementedError|FloatingPointError|StopAsyncIteration|UnicodeDecodeError|DeprecationWarning|UnicodeEncodeError|NotADirectoryError|ProcessLookupError|ZeroDivisionError|IsADirectoryError|FileNotFoundError|UnboundLocalError|KeyboardInterrupt|ChildProcessError|EnvironmentError|IndentationError|InterruptedError|BlockingIOError|ArithmeticError|ConnectionError|BrokenPipeError|FileExistsError|ResourceWarning|PermissionError|RuntimeWarning|ReferenceError|AttributeError|AssertionError|UnicodeWarning|RecursionError|StopIteration|BaseException|OverflowError|SyntaxWarning|FutureWarning|GeneratorExit|ImportWarning|UnicodeError|TimeoutError|WindowsError|RuntimeError|BytesWarning|SystemError|UserWarning|MemoryError|ImportError|LookupError|BufferError|SyntaxError|SystemExit|ValueError|IndexError|NameError|Exception|TypeError|TabError|EOFError|KeyError|VMSError|Warning|OSError|IOError)\b">
        <token type="NameException"/>
      </rule>
    </state>
    <state name="classname">
      <rule pattern="[_\p{L}][_\p{L}\p{N}]*">
        <token type="NameClass"/>
        <pop depth="1"/>
      </rule>
    </state>
    <state name="import">
      <rule pattern="(\s+)(as)(\s+)">
        <bygroups>
          <token type="Text"/>
          <token type="Keyword"/>
          <token type="Text"/>
        </bygroups>
      </rule>
      <rule pattern="\.">
        <token type="NameNamespace"/>
      </rule>
      <rule pattern="[_\p{L}][_\p{L}\p{N}]*">
        <token type="NameNamespace"/>
      </rule>
      <rule pattern="(\s*)(,)(\s*)">
        <bygroups>
          <token type="Text"/>
          <token type="Operator"/>
          <token type="Text"/>
        </bygroups>
      </rule>
      <rule>
        <pop depth="1"/>
      </rule>
    </state>
    <state name="tsqs">
      <rule pattern="&#39;&#39;&#39;">
        <token type="LiteralStringSingle"/>
        <pop depth="1"/>
      </rule>
      <rule>
        <include state="strings-single"/>
      </rule>
      <rule pattern="\n">
        <token type="LiteralStringSingle"/>
      </rule>
    </state>
    <state name="strings-double">
      <rule pattern="%(\(\w+\))?[-#0 +]*([0-9]+|[*])?(\.([0-9]+|[*]))?[hlL]?[E-GXc-giorsaux%]">
        <token type="LiteralStringInterpol"/>
      </rule>
      <rule pattern="\{((\w+)((\.\w+)|(\[[^\]]+\]))*)?(\![sra])?(\:(.?[&lt;&gt;=\^])?[-+ ]?#?0?(\d+)?,?(\.\d+)?[E-GXb-gnosx%]?)?\}">
        <token type="LiteralStringInterpol"/>
      </rule>
      <rule pattern="[^\\\&#39;&#34;%{\n]+">
        <token type="LiteralStringDouble"/>
      </rule>
      <rule pattern="[\&#39;&#34;\\]">
        <token type="LiteralStringDouble"/>
      </rule>
      <rule pattern="%|(\{{1,2})">
        <token type="LiteralStringDouble"/>
      </rule>
    </state>
    <state name="tdqf">
      <rule pattern="&#34;&#34;&#34;">
        <token type="LiteralStringDouble"/>
        <pop depth="1"/>
      </rule>
      <rule>
        <include state="fstrings-double"/>
      </rule>
      <rule pattern="\n">
        <token type="LiteralStringDouble"/>
      </rule>
    </state>
    <state name="expr-inside-fstring-inner">
      <rule pattern="[{([]">
        <token type="Punctuation"/>
        <push state="expr-inside-fstring-inner"/>
      </rule>
      <rule pattern="[])}]">
        <token type="Punctuation"/>
        <pop depth="1"/>
      </rule>
      <rule pattern="\s+">
        <token type="Text"/>
      </rule>
      <rule>
        <include state="expr"/>
      </rule>
    </state>
    <state name="sqs">
      <rule pattern="&#39;">
        <token type="LiteralStringSingle"/>
        <pop depth="1"/>
      </rule>
      <rule pattern="\\\\|\\&#39;|\\\n">
        <token type="LiteralStringEscape"/>
      </rule>
      <rule>
        <include state="strings-single"/>
      </rule>
    </state>
    <state name="funcname">
      <rule>
        <include state="magicfuncs"/>
      </rule>
      <rule pattern="[_\p{L}][_\p{L}\p{N}]*">
        <token type="NameFunction"/>
        <pop depth="1"/>
      </rule>
      <rule>
        <pop depth="1"/>
      </rule>
    </state>
    <state name="expr-keywords">
      <rule pattern="(yield from|async for|lambda|yield|await|else|for|if)\b">
        <token type="Keyword"/>
      </rule>
      <rule pattern="(False|True|None)\b">
        <token type="KeywordConstant"/>
      </rule>
    </state>
    <state name="name">
      <rule pattern="@[_\p{L}][_\p{L}\p{N}]*(\s*\.\s*[_\p{L}][_\p{L}\p{N}]*)*">
        <token type="NameDecorator"/>
      </rule>
      <rule pattern="@">
        <token type="Operator"/>
      </rule>
      <rule pattern="[_\p{L}][_\p{L}\p{N}]*">
        <token type="Name"/>
      </rule>
    </state>
    <state name="magicfuncs">
      <rule pattern="(__instancecheck__|__subclasscheck__|__getattribute__|__length_hint__|__rfloordiv__|__ifloordiv__|__itruediv__|__contains__|__floordiv__|__rtruediv__|__reversed__|__setitem__|__complex__|__rdivmod__|__delattr__|__rmatmul__|__ilshift__|__prepare__|__delitem__|__rrshift__|__imatmul__|__rlshift__|__setattr__|__truediv__|__getitem__|__missing__|__getattr__|__irshift__|__rshift__|__format__|__invert__|__matmul__|__divmod__|__delete__|__aenter__|__lshift__|__await__|__bytes__|__anext__|__aiter__|__aexit__|__round__|__float__|__enter__|__index__|__iadd__|__ipow__|__rpow__|__iter__|__init__|__ixor__|__rmul__|__rmod__|__imul__|__imod__|__iand__|__hash__|__rsub__|__exit__|__rxor__|__bool__|__call__|__rand__|__next__|__radd__|__isub__|__repr__|__set__|__add__|__new__|__neg__|__xor__|__and__|__mul__|__mod__|__sub__|__len__|__str__|__ror__|__ior__|__pos__|__del__|__get__|__dir__|__abs__|__int__|__pow__|__eq__|__gt__|__le__|__lt__|__ne__|__or__|__ge__)\b">
        <token type="NameFunctionMagic"/>
      </rule>
    </state>
    <state name="root">
      <rule pattern="\n">
        <token type="Text"/>
      </rule>
      <rule pattern="^(\s*)([rRuUbB]{,2})(&#34;&#34;&#34;(?:.|\n)*?&#34;&#34;&#34;)">
        <bygroups>
          <token type="Text"/>
          <token type="LiteralStringAffix"/>
          <token type="LiteralStringDoc"/>
        </bygroups>
      </rule>
      <rule pattern="^(\s*)([rRuUbB]{,2})(&#39;&#39;&#39;(?:.|\n)*?&#39;&#39;&#39;)">
        <bygroups>
          <token type="Text"/>
          <token type="LiteralStringAffix"/>
          <token type="LiteralStringDoc"/>
        </bygroups>
      </rule>
      <rule pattern="\A#!.+$">
        <token type="CommentHashbang"/>
      </rule>
      <rule pattern="#.*$">
        <token type="CommentSingle"/>
      </rule>
      <rule pattern="\\\n">
        <token type="Text"/>
      </rule>
      <rule pattern="\\">
        <token type="Text"/>
      </rule>
      <rule>
        <include state="keywords"/>
      </rule>
      <rule pattern="(def)((?:\s|\\\s)+)">
        <bygroups>
          <token type="Keyword"/>
          <token type="Text"/>
        </bygroups>
        <push state="funcname"/>
      </rule>
      <rule pattern="(class)((?:\s|\\\s)+)">
        <bygroups>
          <token type="Keyword"/>
          <token type="Text"/>
        </bygroups>
        <push state="classname"/>
      </rule>
      <rule pattern="(from)((?:\s|\\\s)+)">
        <bygroups>
          <token type="KeywordNamespace"/>
          <token type="Text"/>
        </bygroups>
        <push state="fromimport"/>
      </rule>
      <rule pattern="(import)((?:\s|\\\s)+)">
        <bygroups>
          <token type="KeywordNamespace"/>
          <token type="Text"/>
        </bygroups>
        <push state="import"/>
      </rule>
      <rule>
        <include state="expr"/>
      </rule>
    </state>
    <state name="fstrings-single">
      <rule pattern="\}">
        <token type="LiteralStringInterpol"/>
      </rule>
      <rule pattern="\{">
        <token type="LiteralStringInterpol"/>
        <push state="expr-inside-fstring"/>
      </rule>
      <rule pattern="[^\\\&#39;&#34;{}\n]+">
        <token type="LiteralStringSingle"/>
      </rule>
      <rule pattern="[\&#39;&#34;\\]">
        <token type="LiteralStringSingle"/>
      </rule>
    </state>
    <state name="magicvars">
      <rule pattern="(__annotations__|__kwdefaults__|__qualname__|__objclass__|__defaults__|__closure__|__globals__|__weakref__|__module__|__class__|__bases__|__slots__|__file__|__code__|__name__|__func__|__dict__|__self__|__mro__|__doc__)\b">
        <token type="NameVariableMagic"/>
      </rule>
    </state>
    <state name="tsqf">
      <rule pattern="&#39;&#39;&#39;">
        <token type="LiteralStringSingle"/>
        <pop depth="1"/>
      </rule>
      <rule>
        <include state="fstrings-single"/>
      </rule>
      <rule pattern="\n">
        <token type="LiteralStringSingle"/>
      </rule>
    </state>
    <state name="strings-single">
      <rule pattern="%(\(\w+\))?[-#0 +]*([0-9]+|[*])?(\.([0-9]+|[*]))?[hlL]?[E-GXc-giorsaux%]">
        <token type="LiteralStringInterpol"/>
      </rule>
      <rule pattern="\{((\w+)((\.\w+)|(\[[^\]]+\]))*)?(\![sra])?(\:(.?[&lt;&gt;=\^])?[-+ ]?#?0?(\d+)?,?(\.\d+)?[E-GXb-gnosx%]?)?\}">
        <token type="LiteralStringInterpol"/>
      </rule>
      <rule pattern="[^\\\&#39;&#34;%{\n]+">
        <token type="LiteralStringSingle"/>
      </rule>
      <rule pattern="[\&#39;&#34;\\]">
        <token type="LiteralStringSingle"/>
      </rule>
      <rule pattern="%|(\{{1,2})">
        <token type="LiteralStringSingle"/>
      </rule>
    </state>
    <state name="rfstringescape">
      <rule pattern="\{\{">
        <token type="LiteralStringEscape"/>
      </rule>
      <rule pattern="\}\}">
        <token type="LiteralStringEscape"/>
      </rule>
    </state>
    <state name="sqf">
      <rule pattern="&#39;">
        <token type="LiteralStringSingle"/>
        <pop depth="1"/>
      </rule>
      <rule pattern="\\\\|\\&#39;|\\\n">
        <token type="LiteralStringEscape"/>
      </rule>
      <rule>
        <include state="fstrings-single"/>
      </rule>
    </state>
    <state name="dqf">
      <rule pattern="&#34;">
        <token type="LiteralStringDouble"/>
        <pop depth="1"/>
      </rule>
      <rule pattern="\\\\|\\&#34;|\\\n">
        <token type="LiteralStringEscape"/>
      </rule>
      <rule>
        <include state="fstrings-double"/>
      </rule>
    </state>
    <state name="expr-inside-fstring">
      <rule pattern="[{([]">
        <token type="Punctuation"/>
        <push state="expr-inside-fstring-inner"/>
      </rule>
      <rule pattern="(=\s*)?(\![sraf])?\}">
        <token type="LiteralStringInterpol"/>
        <pop depth="1"/>
      </rule>
      <rule pattern="(=\s*)?(\![sraf])?:">
        <token type="LiteralStringInterpol"/>
        <pop depth="1"/>
      </rule>
      <rule pattern="\s+">
        <token type="Text"/>
      </rule>
      <rule>
        <include state="expr"/>
      </rule>
    </state>
    <state name="tdqs">
      <rule pattern="&#34;&#34;&#34;">
        <token type="LiteralStringDouble"/>
        <pop depth="1"/>
      </rule>
      <rule>
        <include state="strings-double"/>
      </rule>
      <rule pattern="\n">
        <token type="LiteralStringDouble"/>
      </rule>
    </state>
    <state name="fstringescape">
      <rule>
        <include state="rfstringescape"/>
      </rule>
      <rule>
        <include state="stringescape"/>
      </rule>
    </state>
    <state name="stringescape">
      <rule pattern="\\([\\abfnrtv&#34;\&#39;]|\n|N\{.*?\}|u[a-fA-F0-9]{4}|U[a-fA-F0-9]{8}|x[a-fA-F0-9]{2}|[0-7]{1,3})">
        <token type="LiteralStringEscape"/>
      </rule>
    </state>
  </rules>
</lexer>