python - glsl 解析器使用 pyparsing 给出 AttributeErrors

Question

我正在尝试更新此 glsl-parser它使用旧的 pyparsing 版本和 python2.x 到 python3.x 和最新的 pyparsing 版本(2.1.9 atm)。

我不知道哪个 pyparsing 版本使用了原始源代码，但它一定很旧，因为在阅读 pyparsing news 后仍在使用 keepOriginalText 辅助方法。我看到这条评论 Removed keepOriginalText helper method，它在很久以前就被弃用了。由 originalTextFor 取代。

无论如何，这里是使用python3.5.1 & pyparsing==2.1.9移植的第一次尝试:

# -*- coding: utf-8 -*-
# -----------------------------------------------------------------------------
# Copyright (c) 2014, Nicolas P. Rougier
# Distributed under the (new) BSD License. See LICENSE.txt for more info.
# -----------------------------------------------------------------------------
import pyparsing

keywords = ("attribute const uniform varying break continue do for while"
            "if else"
            "in out inout"
            "float int void bool true false"
            "lowp mediump highp precision invariant"
            "discard return"
            "mat2 mat3 mat4"
            "vec2 vec3 vec4 ivec2 ivec3 ivec4 bvec2 bvec3 bvec4 sampler2D samplerCube"
            "struct")
reserved = ("asm"
            "class union enum typedef template this packed"
            "goto switch default"
            "inline noinline volatile public static extern external"
            "interface flat long short double half fixed unsigned superp"
            "input output"
            "hvec2 hvec3 hvec4 dvec2 dvec3 dvec4 fvec2 fvec3 fvec4 sampler1D sampler3D"
            "sampler1DShadow sampler2DShadow"
            "sampler2DRect sampler3DRect sampler2DRectShadow"
            "sizeof cast"
            "namespace using")
precision = "lowp mediump high"
storage = "const uniform attribute varying"


# Tokens
# ----------------------------------
LPAREN = pyparsing.Literal("(").suppress()
RPAREN = pyparsing.Literal(")").suppress()
LBRACK = pyparsing.Literal("[").suppress()
RBRACK = pyparsing.Literal("]").suppress()
LBRACE = pyparsing.Literal("{").suppress()
RBRACE = pyparsing.Literal("}").suppress()
IDENTIFIER = pyparsing.Word(pyparsing.alphas + '_', pyparsing.alphanums + '_')
TYPE = pyparsing.Word(pyparsing.alphas + '_', pyparsing.alphanums + "_")
END = pyparsing.Literal(";").suppress()
INT = pyparsing.Word(pyparsing.nums)
FLOAT = pyparsing.Regex(
    '[+-]?(((\d+\.\d*)|(\d*\.\d+))([eE][-+]?\d+)?)|(\d*[eE][+-]?\d+)')
STORAGE = pyparsing.Regex('|'.join(storage.split(' ')))
PRECISION = pyparsing.Regex('|'.join(precision.split(' ')))
STRUCT = pyparsing.Literal("struct").suppress()


# ------------------------
def get_prototypes(code):
    """
    Get all function declarations

    Code example
    ------------

    mediump vec3 function_1(vec4);
    vec3 function_2(float a, float b);
    """

    PARAMETER = pyparsing.Group(pyparsing.Optional(PRECISION).setResultsName("precision") +
                                TYPE.setResultsName("type") +
                                pyparsing.Optional(IDENTIFIER).setResultsName("name"))
    PARAMETERS = pyparsing.delimitedList(PARAMETER).setResultsName(
        "arg", listAllMatches=True)
    PROTOTYPE = (pyparsing.Optional(PRECISION).setResultsName("precision") +
                 TYPE.setResultsName("type") +
                 IDENTIFIER.setResultsName("name") +
                 LPAREN + pyparsing.Optional(PARAMETERS).setResultsName("args") + RPAREN +
                 END)
    PROTOTYPE.ignore(pyparsing.cStyleComment)

    for (token, start, end) in PROTOTYPE.scanString(code):
        print(token.precision, token.type, token.name, '(',)
        for arg in token.args:
            print(arg.precision, arg.type, arg.name, ',',)
        print(')')


# ------------------------
def get_functions(code):
    """
    Get all function definitions

    Code example
    ------------

    mediump vec3 compute_normal(vec4);
    """

    PARAMETER = pyparsing.Group(pyparsing.Optional(PRECISION).setResultsName("precision") +
                                TYPE.setResultsName("type") +
                                pyparsing.Optional(IDENTIFIER).setResultsName("name"))
    PARAMETERS = pyparsing.delimitedList(PARAMETER).setResultsName(
        "arg", listAllMatches=True)
    FUNCTION = (pyparsing.Optional(PRECISION).setResultsName("precision") +
                TYPE.setResultsName("type") +
                IDENTIFIER.setResultsName("name") +
                LPAREN + pyparsing.Optional(PARAMETERS).setResultsName("args") + RPAREN +
                pyparsing.nestedExpr("{", "}").setParseAction(pyparsing.originalTextFor).setResultsName("code"))
    FUNCTION.ignore(pyparsing.cStyleComment)

    for (token, start, end) in FUNCTION.scanString(code):
        print(token.precision, token.type, token.name, '(',)
        for arg in token.args:
            print(arg.precision, arg.type, arg.name, ',',)
        print(') { ... }')

        # print token.code
        # print code[start:end]


# ------------------------
def get_version(code):
    """
    Get shader version (if specified)

    Code example
    ------------

    #version 120
    """

    VERSION = (
        pyparsing.Literal("#") + pyparsing.Keyword("version")).suppress() + INT
    for (token, start, end) in VERSION.scanString(code):
        version = token[0]
        # print code[start:end]
    return version

# ------------------------


def get_declarations(code):
    """
    Get all declarations prefixed with a storage qualifier.

    Code example
    ------------

    uniform lowp vec4 fg_color = vec4(1),
                      bg_color = vec4(vec3(0),1);
    """

    # Callable expression
    EXPRESSION = pyparsing.Forward()
    ARG = pyparsing.Group(EXPRESSION) | IDENTIFIER | FLOAT | INT
    ARGS = pyparsing.delimitedList(ARG)
    EXPRESSION << IDENTIFIER + \
        pyparsing.Group(LPAREN + pyparsing.Optional(ARGS) + RPAREN)

    # Value
    VALUE = (EXPRESSION | pyparsing.Word(pyparsing.alphanums + "_()+-/*")
             ).setParseAction(pyparsing.originalTextFor)

    # Single declaration
    VARIABLE = (IDENTIFIER.setResultsName("name") +
                pyparsing.Optional(LBRACK +
                                   (INT | IDENTIFIER).setResultsName("size")
                                   + RBRACK) +
                pyparsing.Optional(pyparsing.Literal("=").suppress() + VALUE.setResultsName("value")))

    # Several declarations at once
    DECLARATION = (STORAGE.setResultsName("storage") +
                   pyparsing.Optional(PRECISION).setResultsName("precision") +
                   TYPE.setResultsName("type") +
                   pyparsing.delimitedList(VARIABLE.setResultsName("variable", listAllMatches=True)) +
                   END)
    DECLARATION.ignore(pyparsing.cStyleComment)

    for (tokens, start, end) in DECLARATION.scanString(code):
        for token in tokens.variable:
            print(tokens.storage, tokens.precision, tokens.type,)
            print(token.name, token.size)


# ------------------------
def get_definitions(code):
    """
    Get all structure definitions and associated declarations.

    Code example
    ------------

    uniform struct Light {
        vec4 position;
        vec3 color;
    } light0, light1;
    """

    # Single declaration
    DECLARATION = pyparsing.Group(IDENTIFIER.setResultsName("name") +
                                  pyparsing.Optional(LBRACK +
                                                     (INT | IDENTIFIER).setResultsName("size") +
                                                     RBRACK))
    # Several declarations at once
    DECLARATIONS = (pyparsing.Optional(PRECISION) +
                    TYPE +
                    pyparsing.delimitedList(DECLARATION) +
                    END)

    # Definition + declarations
    DEFINITION = (STRUCT +
                  IDENTIFIER.setResultsName("name") +
                  LBRACE + pyparsing.OneOrMore(DECLARATIONS).setResultsName('content') + RBRACE +
                  pyparsing.Optional(pyparsing.delimitedList(DECLARATION.setResultsName("declarations", listAllMatches=True))) +
                  END)
    DEFINITION.ignore(pyparsing.cStyleComment)

    for (tokens, start, end) in DEFINITION.scanString(code):
        for token in tokens.declarations:
            print(tokens.name, token.name)
            # print tokens.content


# ----------------
def resolve(code):
    """
    Expand const and preprocessor definitions in order to get constant values.

    Return the transformed code
    """

    constants = {}

    DEFINITION = (pyparsing.Literal("#") + pyparsing.Literal("define") +
                  IDENTIFIER.setResultsName("name") +
                  pyparsing.restOfLine.setResultsName("value"))

    VALUE = pyparsing.Word(pyparsing.alphanums + "_()+-/*")
    DECLARATION = (pyparsing.Literal("const") +
                   TYPE.setResultsName("type") +
                   IDENTIFIER.setResultsName("name") +
                   pyparsing.Literal("=") +
                   VALUE.setResultsName("value") +
                   pyparsing.Literal(";"))
    REFERENCE = pyparsing.Forward()

    def process_definition(s, l, t):
        value = REFERENCE.transformString(t.value)
        constants[t.name] = value
        REFERENCE << pyparsing.MatchFirst(
            map(pyparsing.Keyword, constants.keys()))
        return "#define " + t.name + " " + value

    def process_declaration(s, l, t):
        value = REFERENCE.transformString(t.value)
        constants[t.name] = value
        REFERENCE << pyparsing.MatchFirst(
            map(pyparsing.Keyword, constants.keys()))
        return "const " + t.type + " " + t.name + "=" + value + ";"

    def process_reference(s, l, t):
        return constants[t[0]]

    REFERENCE.setParseAction(process_reference)
    DEFINITION.setParseAction(process_definition)
    DECLARATION.setParseAction(process_declaration)
    EXPANDER = REFERENCE | DEFINITION | DECLARATION

    code = EXPANDER.transformString(code)
    for key, val in constants.items():
        constants[key] = eval(val)

    return code, constants


# -----------------------------------------------------------------------------
if __name__ == '__main__':

    code = """
#version 120

#define A (1)
const int B=(A+2);
#define C (B+3)
const int D=C+4;

uniform float array[D];

struct Point {
    vec4 position;
    float size;
};

uniform struct Light {
    vec4 position;
    vec3 color;
} light0, light1;

const float PI = 3.14159265358979323846264;
const float SQRT_2 = 1.4142135623730951;

uniform vec4 fg_color = vec4(1),
             bg_color = vec4(vec3(0),1);

mediump vec3 compute_normal(vec4 position, vec3 orientation);
vec3 /* */ compute_light(vec4, vec3, float intensity)
{
   vec3 hello;
   vec3 hello;
}

"""
code, _ = resolve(code)
print("GLSL version: %s\n" % get_version(code))

get_definitions(code)
get_declarations(code)
get_prototypes(code)
get_functions(code)

# code = """
# #if A
# #if B
# #if C
# #endif
# #endif
# #endif
# """

# IF = (pyparsing.Literal('#') + (pyparsing.Keyword('if') | pyparsing.Keyword('ifdef') | pyparsing.Keyword('ifndef')))
# ENDIF = (pyparsing.Literal('#') + pyparsing.Keyword('endif'))
# MACRO = (IF + pyparsing.restOfLine() +
#          SkipTo(ENDIF, include=True)).setParseAction(pyparsing.originalTextFor)
# for (tokens, start, end) in MACRO.scanString(code):
#     print tokens

当您尝试运行上面的 mcve 代码时，您将得到:

GLSL version: 120

('Light', 'light0')
('Light', 'light1')
d:\virtual_envs\py2711\lib\site-packages\pyparsing.py:3536: SyntaxWarning: Cannot combine element of type <type 'int'> with ParserElement
  matchExpr = locMarker("_original_start") + expr + endlocMarker("_original_end")
d:\virtual_envs\py2711\lib\site-packages\pyparsing.py:3536: SyntaxWarning: Cannot combine element of type <type 'NoneType'> with ParserElement
  matchExpr = locMarker("_original_start") + expr + endlocMarker("_original_end")
Traceback (most recent call last):
  File "D:\sources\personal\python\pyqt\pyshaders\gui\glsl-parser.py", line 311, in <module>
    get_declarations(code)
  File "D:\sources\personal\python\pyqt\pyshaders\gui\glsl-parser.py", line 173, in get_declarations
    for (tokens, start, end) in DECLARATION.scanString(code):
  File "d:\virtual_envs\py2711\lib\site-packages\pyparsing.py", line 1258, in scanString
    nextLoc,tokens = parseFn( instring, preloc, callPreParse=False )
  File "d:\virtual_envs\py2711\lib\site-packages\pyparsing.py", line 1084, in _parseNoCache
    loc,tokens = self.parseImpl( instring, preloc, doActions )
  File "d:\virtual_envs\py2711\lib\site-packages\pyparsing.py", line 2576, in parseImpl
    loc, exprtokens = e._parse( instring, loc, doActions )
  File "d:\virtual_envs\py2711\lib\site-packages\pyparsing.py", line 1084, in _parseNoCache
    loc,tokens = self.parseImpl( instring, preloc, doActions )
  File "d:\virtual_envs\py2711\lib\site-packages\pyparsing.py", line 2576, in parseImpl
    loc, exprtokens = e._parse( instring, loc, doActions )
  File "d:\virtual_envs\py2711\lib\site-packages\pyparsing.py", line 1084, in _parseNoCache
    loc,tokens = self.parseImpl( instring, preloc, doActions )
  File "d:\virtual_envs\py2711\lib\site-packages\pyparsing.py", line 3038, in parseImpl
    loc, tokens = self.expr._parse( instring, loc, doActions, callPreParse=False )
  File "d:\virtual_envs\py2711\lib\site-packages\pyparsing.py", line 1084, in _parseNoCache
    loc,tokens = self.parseImpl( instring, preloc, doActions )
  File "d:\virtual_envs\py2711\lib\site-packages\pyparsing.py", line 2576, in parseImpl
    loc, exprtokens = e._parse( instring, loc, doActions )
  File "d:\virtual_envs\py2711\lib\site-packages\pyparsing.py", line 1110, in _parseNoCache
    tokens = fn( instring, tokensStart, retTokens )
  File "d:\virtual_envs\py2711\lib\site-packages\pyparsing.py", line 831, in wrapper
    ret = func(*args[limit[0]:])
  File "d:\virtual_envs\py2711\lib\site-packages\pyparsing.py", line 3542, in originalTextFor
    matchExpr.setParseAction(extractText)
AttributeError: 'NoneType' object has no attribute 'setParseAction'

我还在学习pyparsing，这里有什么问题？

Answer 1

originalTextFor 不是解析 Action ，而是将解析 Action 附加到已定义表达式的辅助方法。在您的示例中，它用在两个地方:

# Value
VALUE = (EXPRESSION | pyparsing.Word(pyparsing.alphanums + "_()+-/*")
         ).setParseAction(pyparsing.originalTextFor)

FUNCTION = (pyparsing.Optional(PRECISION).setResultsName("precision") +
            TYPE.setResultsName("type") +
            IDENTIFIER.setResultsName("name") +
            LPAREN + pyparsing.Optional(PARAMETERS).setResultsName("args") + RPAREN +
            pyparsing.nestedExpr("{", "}").setParseAction(pyparsing.originalTextFor).setResultsName("code"))

将这些更改为:

# Value
VALUE = pyparsing.originalTextFor(EXPRESSION |
                                  pyparsing.Word(pyparsing.alphanums + "_()+-/*"))

FUNCTION = (pyparsing.Optional(PRECISION).setResultsName("precision") +
            TYPE.setResultsName("type") +
            IDENTIFIER.setResultsName("name") +
            LPAREN + pyparsing.Optional(PARAMETERS).setResultsName("args") + RPAREN +
            pyparsing.originalTextFor(pyparsing.nestedExpr("{", "}")).setResultsName("code"))

您可能会发现 setResultsName 的最新 pyparsing 形式看起来更简洁一些，但旧形式仍然可以正常工作:

FUNCTION = (pyparsing.Optional(PRECISION)("precision") +
            TYPE("type") +
            IDENTIFIER("name") +
            LPAREN + pyparsing.Optional(PARAMETERS)("args") + RPAREN +
            pyparsing.originalTextFor(pyparsing.nestedExpr("{", "}"))("code"))

如果您进行此更改，那些使用 listAllMatches 的地方将通过向名称参数添加“*”来处理:

pyparsing.delimitedList(VARIABLE("variable*"))