llparse 0.1.0__tar.gz → 0.1.2__tar.gz

{llparse-0.1.0/llparse.egg-info → llparse-0.1.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: llparse
-Version: 0.1.0
+Version: 0.1.2
 Summary: A Parody of llparse written for writing C Parsers with Python
 Author-email: Vizonex <VizonexBusiness@gmail.com>
 Requires-Python: >=3.9
@@ -9,6 +9,10 @@ License-File: LICENSE
 Dynamic: license-file
 
 # pyllparse
+[![PyPI version](https://badge.fury.io/py/llparse.svg)](https://badge.fury.io/py/llparse)
+[![PyPI - Downloads](https://img.shields.io/pypi/dm/llparse)](https://badge.fury.io/py/llparse)
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
+
 A python parody of the typescript library llparse.
 
 I take no credit for the orginal work done by indutny and I was originally very nervous about making 

llparse-0.1.0/PKG-INFO → llparse-0.1.2/README.md

@@ -1,14 +1,8 @@
-Metadata-Version: 2.4
-Name: llparse
-Version: 0.1.0
-Summary: A Parody of llparse written for writing C Parsers with Python
-Author-email: Vizonex <VizonexBusiness@gmail.com>
-Requires-Python: >=3.9
-Description-Content-Type: text/markdown
-License-File: LICENSE
-Dynamic: license-file
-
 # pyllparse
+[![PyPI version](https://badge.fury.io/py/llparse.svg)](https://badge.fury.io/py/llparse)
+[![PyPI - Downloads](https://img.shields.io/pypi/dm/llparse)](https://badge.fury.io/py/llparse)
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
+
 A python parody of the typescript library llparse.
 
 I take no credit for the orginal work done by indutny and I was originally very nervous about making 

{llparse-0.1.0 → llparse-0.1.2}/llparse/C_compiler.py

@@ -29,32 +29,32 @@ class CCompiler:
         out.append("#include <string.h>")
         out.append("")
         # Seems LLParse was updated from /* UNREACHABLE */ abort(); to a Macro, Intresting...
-        out.append('#ifdef __SSE4_2__')
-        out.append(' #ifdef _MSC_VER')
-        out.append('  #include <nmmintrin.h>')
-        out.append(' #else  /* !_MSC_VER */')
-        out.append('  #include <x86intrin.h>')
-        out.append(' #endif  /* _MSC_VER */')
-        out.append('#endif  /* __SSE4_2__ */')
-        out.append('')
-
-        out.append('#ifdef __ARM_NEON__')
-        out.append(' #include <arm_neon.h>')
-        out.append('#endif  /* __ARM_NEON__ */')
-        out.append('')
-
-        out.append('#ifdef __wasm__')
-        out.append(' #include <wasm_simd128.h>')
-        out.append('#endif  /* __wasm__ */')
-        out.append('')
-
-        out.append('#ifdef _MSC_VER')
-        out.append(' #define ALIGN(n) _declspec(align(n))')
-        out.append(' #define UNREACHABLE __assume(0)')
-        out.append('#else  /* !_MSC_VER */')
-        out.append(' #define ALIGN(n) __attribute__((aligned(n)))')
-        out.append(' #define UNREACHABLE __builtin_unreachable()')
-        out.append('#endif  /* _MSC_VER */')
+        out.append("#ifdef __SSE4_2__")
+        out.append(" #ifdef _MSC_VER")
+        out.append("  #include <nmmintrin.h>")
+        out.append(" #else  /* !_MSC_VER */")
+        out.append("  #include <x86intrin.h>")
+        out.append(" #endif  /* _MSC_VER */")
+        out.append("#endif  /* __SSE4_2__ */")
+        out.append("")
+
+        out.append("#ifdef __ARM_NEON__")
+        out.append(" #include <arm_neon.h>")
+        out.append("#endif  /* __ARM_NEON__ */")
+        out.append("")
+
+        out.append("#ifdef __wasm__")
+        out.append(" #include <wasm_simd128.h>")
+        out.append("#endif  /* __wasm__ */")
+        out.append("")
+
+        out.append("#ifdef _MSC_VER")
+        out.append(" #define ALIGN(n) _declspec(align(n))")
+        out.append(" #define UNREACHABLE __assume(0)")
+        out.append("#else  /* !_MSC_VER */")
+        out.append(" #define ALIGN(n) __attribute__((aligned(n)))")
+        out.append(" #define UNREACHABLE __builtin_unreachable()")
+        out.append("#endif  /* _MSC_VER */")
 
         out.append("")
         out.append(
@@ -83,7 +83,7 @@ class CCompiler:
         out.append("}")
         out.append("")
 
-        # TODO (Vizonex) Make llparse_state_t's Name Optional and alterable incase mixed with 
+        # TODO (Vizonex) Make llparse_state_t's Name Optional and alterable incase mixed with
         # llhttp or another parser
         out.append(f"static llparse_state_t {info.prefix}__run(")
         out.append(f"    {info.prefix}_t* {ARG_STATE},")
@@ -182,9 +182,7 @@ class CCompiler:
             else:
                 # TODO (Vizonex) Merge lines 139 & 140 together in a future update
                 callback = (
-                    f"({info.prefix}__span_cb)"
-                    + ctx.spanCbField(span.index)
-                    + f"({callback})"
+                    f"(({info.prefix}__span_cb)" + ctx.spanCbField(span.index) + f")"
                 )
 
             args = [ctx.stateArg(), posField, f"(const char*) {ctx.endPosArg()}"]

{llparse-0.1.0 → llparse-0.1.2}/llparse/compilator.py

@@ -105,6 +105,9 @@ class Code(Generic[T]):
     def build(self, ctx: "Compilation", out: list[str]):
         pass
 
+    def __hash__(self):
+        return hash(self.ref)
+
 
 class External(Code[_frontend.code.External]):
     def build(self, ctx: "Compilation", out: list[str]):
@@ -371,7 +374,6 @@ class Consume(Node):
         index = ctx.stateField(self.ref.field)
         ty = ctx.getFieldType(self.ref.field)
 
-
         if ty == "i64":
             pass
         elif ty == "i32":
@@ -880,7 +882,7 @@ class Compilation:
     def buildStateEnum(self, out: list[str]):
         # TODO (Vizonex) Give out other names that you could pass as an enum statename
         # this is incase multiple llparse_state_e states are given to compile
-        # example would be mixing llhttp with some other source... 
+        # example would be mixing llhttp with some other source...
         out.append("enum llparse_state_e {")
         out.append(f"  {STATE_ERROR},")
         for stateName in self.stateDict.keys():
@@ -988,8 +990,7 @@ class Compilation:
             out.append(f"{LABEL_PREFIX}{name} : " + "{")
             for line in lines:
                 out.append(f"  {line}")
-            out.append("  /* UNREACHABLE */;")
-            out.append("  abort();")
+            out.append("  UNREACHABLE;")
             out.append("}")
 
     def buildInternalStates(self, out: list[str]):
@@ -1000,8 +1001,7 @@ class Compilation:
             out.append(f"{LABEL_PREFIX}{name}: " + "{")
             for line in lines:
                 out.append(f"  {line}")
-            out.append("  /* UNREACHABLE */;")
-            out.append("  abort();")
+            out.append("  UNREACHABLE;")
             out.append("}")
 
     def addState(self, state: str, lines: list[str]):

{llparse-0.1.0 → llparse-0.1.2}/llparse/cython_builder.py

@@ -2,6 +2,7 @@
 
 from contextlib import contextmanager
 from typing import Optional
+
 from .frontend import IFrontendResult
 from .pyfront.front import Match
 from .pyfront.nodes import Invoke

{llparse-0.1.0 → llparse-0.1.2}/llparse/dot.py

@@ -208,6 +208,3 @@ class Dot:
 
     def escape(self, value: str):
         return "'" + value.replace("\\", "\\$1").replace('"', "\\$1") + "'"
-
-
-# TODO FIX ALL BUFFERS BACK TO STRINGS!

llparse-0.1.2/llparse/errors.py

@@ -0,0 +1,2 @@
+class Error(Exception):
+    pass

{llparse-0.1.0 → llparse-0.1.2}/llparse/frontend.py

@@ -4,6 +4,7 @@ from typing import Literal, Optional, Union
 from .enumerator import Enumerator
 from .pybuilder import LoopChecker
 from .pybuilder import builder as source
+
 # from pyfront.namespace import code, node , transform
 from .pyfront import namespace as _frontend
 from .pyfront.front import Identifier, IWrap, SpanField
@@ -17,13 +18,10 @@ DEFAULT_MIN_TABLE_SIZE = 32
 DEFAULT_MAX_TABLE_WIDTH = 4
 
 
-
 WrappedNode = IWrap[_frontend.node.Node]
 WrappedCode = IWrap[_frontend.code.Code]
 
 
-
-
 @dataclass
 class ITableLookupTarget:
     trie: TrieEmpty
@@ -309,18 +307,18 @@ class Frontend:
         bailout = False
         for child in trie.children:
             if isinstance(child.node, TrieEmpty):
-                print(
-                    'non-leaf trie child of "%s" prevents table allocation' % node.name
-                )
+                # print(
+                #     'non-leaf trie child of "%s" prevents table allocation' % node.name
+                # )
                 bailout = False
                 continue
 
             empty: TrieEmpty = child.node
             if getattr(empty, "value", None) is None:
-                print(
-                    'value passing trie leaf of "%s" prevents table allocation'
-                    % node.name
-                )
+                # print(
+                #     'value passing trie leaf of "%s" prevents table allocation'
+                #     % node.name
+                # )
                 bailout = False
                 continue
 

{llparse-0.1.0 → llparse-0.1.2}/llparse/llparse.py

@@ -4,8 +4,13 @@ from dataclasses import dataclass
 from typing import Optional
 
 from .C_compiler import CCompiler
-from .frontend import (DEFAULT_MAX_TABLE_WIDTH, DEFAULT_MIN_TABLE_SIZE,
-                       Frontend, IImplementation, source)
+from .frontend import (
+    DEFAULT_MAX_TABLE_WIDTH,
+    DEFAULT_MIN_TABLE_SIZE,
+    Frontend,
+    IImplementation,
+    source,
+)
 from .header import HeaderBuilder
 
 

{llparse-0.1.0 → llparse-0.1.2}/llparse/pybuilder/builder.py

@@ -1,6 +1,6 @@
 from typing import Literal, Optional, Union
 
-import pyparse.pybuilder.main_code as code
+from ..pybuilder import main_code as code
 
 # from pydot import graph_from_dot_data
 

{llparse-0.1.0 → llparse-0.1.2}/llparse/pybuilder/loopchecker.py

@@ -1,7 +1,13 @@
+import logging
 from typing import Any, Literal, Union
 
+from ..errors import Error
 from ..pybuilder.main_code import Node
 
+logger = logging.getLogger("llparse.pybuilder.loopchecker")
+logger.setLevel(logging.INFO)
+
+
 MAX_VALUE = 256
 WORD_SIZE = 32
 SIZE = MAX_VALUE // WORD_SIZE
@@ -19,17 +25,20 @@ assert MAX_VALUE % WORD_SIZE == 0
 
 class Lattice:
     def __init__(
-        self, value: Union[Any, list[int], Literal["empty"], Literal["any"]]
+        self, value: Union[Any, list[int], bytes, Literal["empty"], Literal["any"]]
     ) -> None:
         self.value = value
         self.words: list[int] = []
 
         # allocate space by filling in data with zeros...
-
-        for _ in range(SIZE):
-            self.words.append(0)
-
-        if len(value) > 1:
+        if value != "any":
+            for _ in range(SIZE):
+                self.words.append(0)
+        else:
+            for _ in range(SIZE):
+                self.words.append(WORD_FILL)
+
+        if isinstance(value, (list, bytes)):
             for single in value:
                 self.add(single)
 
@@ -69,6 +78,9 @@ class Lattice:
             result.words[i] = self.words[i] & other.words[i]
         return result
 
+    def __repr__(self):
+        return f"<Lattice {', '.join(f'{k}: {v!r}' for k, v in self.__dict__.items())}>"
+
     def subtract(self, other: "Lattice") -> "Lattice":
         result = Lattice("empty")
         for i in range(SIZE):
@@ -76,7 +88,13 @@ class Lattice:
         return result
 
     def isEqual(self, other: "Lattice"):
-        return True if (self.value == other.value) else False
+        if self.toJSON() == other.toJSON():
+            return True
+        else:
+            for i in range(SIZE):
+                if self.words[i] != other.words[i]:
+                    return False
+        return True
 
     def toJSON(self):
         isEmpty = True
@@ -111,7 +129,9 @@ class Reachability:
             if otherwise:
                 queue.append(otherwise.node)
 
-        return list(res)
+        # Reverse the order so that we always
+        # throw an error on bad configurations...
+        return res
 
 
 EMPTY_VALUE = Lattice("empty")
@@ -124,38 +144,45 @@ class LoopChecker:
         self.terminatedCache: dict[Node, Lattice] = {}
 
     def clear(self, nodes: list[Node]):
-        self.lattice.update({(node, EMPTY_VALUE) for node in nodes})
+        for node in nodes:
+            self.lattice[node] = EMPTY_VALUE
 
     def check(self, root: Node):
         r = Reachability()
         nodes = r.build(root)
+
         for node in nodes:
             self.clear(nodes)
+            logger.debug("checking loops starting from %s" % node.name)
             self.lattice[node] = ANY_VALUE
+            # we must eliminate randomness so that error always throw
             changed: set[Node] = set([root])
-            while len(changed) != 0:
-                next = set()
-                for changedNode in next:
-                    self.propagate(changedNode, next)
-                changed = next
 
-        self.visit(root, list(changed))
+            while changed:
+                logger.debug("changed %s" % [n.name for n in changed])
+                _next = set()
+                for changedNode in changed:
+                    self.propagate(changedNode, _next)
+                changed = _next
+            logger.debug("lattice stabilized")
+            self.visit(root, [])
 
     def propagate(self, node: Node, changed: set[Node]):
         value = self.lattice[node]
-        terminated: "Lattice" = self.terminate(node, value, changed)
-        if terminated.value == EMPTY_VALUE.value:
+        terminated = self.terminate(node)
+        logger.debug("propagate(%r), initial value %r" % (node.name, value.toJSON()))
+        if not terminated.isEqual(EMPTY_VALUE):
+            logger.debug("node %s terminates %r" % (node.name, terminated.toJSON()))
             value = value.subtract(terminated)
             if value.isEqual(EMPTY_VALUE):
                 return
 
-        keysbyTarget: dict[Node, Lattice] = dict()
+        keysbyTarget: dict[Node, Lattice] = {}
 
         for edge in node.getAllEdges():
             if not edge.noAdvance:
                 continue
 
-            targetValue: Lattice
             if keysbyTarget.get(edge.node):
                 targetValue = keysbyTarget[edge.node]
             else:
@@ -170,9 +197,14 @@ class LoopChecker:
                     continue
 
                 targetValue = targetValue.union(edgeValue)
+
             keysbyTarget[edge.node] = targetValue
 
         for child, childValue in keysbyTarget.items():
+            logger.debug(
+                "node %r propagates %r to %r"
+                % (node.name, childValue.toJSON(), child.name)
+            )
             self.update(child, childValue, changed)
         # FINISHED!
 
@@ -182,13 +214,13 @@ class LoopChecker:
             return False
         self.lattice[node] = newValue
         changed.add(node)
+        return True
 
-    def terminate(self, node: Node, value: Lattice, changed: set[Node]):
-        if self.terminatedCache.get(node):
+    def terminate(self, node: Node):
+        if node in self.terminatedCache:
             return self.terminatedCache[node]
 
         terminated: list[int] = []
-
         for edge in node.getAllEdges():
             if edge.noAdvance:
                 continue
@@ -204,43 +236,52 @@ class LoopChecker:
 
     def visit(self, node: Node, path: list[Node]):
         value = self.lattice[node]
+        logger.debug("enter %s, value is %s" % (node.name, value.toJSON()))
+
         terminated = (
-            self.terminatedCache[node]
-            if self.terminatedCache.get(node)
-            else EMPTY_VALUE
+            EMPTY_VALUE
+            if node not in self.terminatedCache
+            else self.terminatedCache[node]
         )
-        if terminated.isEqual(EMPTY_VALUE):
+
+        if not terminated.isEqual(EMPTY_VALUE):
+            logger.debug(f"subtract terminated {terminated}")
             value = value.subtract(terminated)
             if value.isEqual(EMPTY_VALUE):
+                logger.debug("terminated everything")
                 return
 
         for edge in node.getAllEdges():
-            if edge.noAdvance:
+            if not edge.noAdvance:
                 continue
             edgeValue = value
-            if not (not edge.key or isinstance(edge.key, int)):
+            if edge.key is None or isinstance(edge.key, int):
+                pass
+            else:
                 edgeValue = edgeValue.intersect(Lattice([edge.key[0]]))
 
             if edgeValue.isEqual(EMPTY_VALUE):
+                # logger.debug(edge.node.name + " not recursive")
                 continue
 
-            def indexOf(path: list, obj):
-                try:
-                    return path.index(obj)
-                except Exception:
-                    return -1
+            def indexOf(path: list[Node], obj: Node) -> int:
+                for o in path:
+                    if o.name == obj.name:
+                        return 0
+                return -1
 
-            if indexOf(path, node) != -1:
-                if len(path) == 0:
-                    raise Exception(
-                        f'Detected a loop in "{edge.node.name}" though : {edge.node.name}'
+            if indexOf(path, edge.node) != -1:
+                if len(path) == 1:
+                    raise Error(
+                        f'Detected loop in "{edge.node.name}" through "{edge.node.name}"'
                     )
 
-                raise Exception(
+                raise Error(
                     'Detected loop in "'
                     + edge.node.name
                     + '" through chain '
                     + (" -> ").join(['"' + name.name + '"' for name in path])
                 )
 
-            self.visit(edge.node, path.extend([edge.node]))
+            self.visit(edge.node, path + [edge.node])
+        logger.debug("leave %s" % node.name)

{llparse-0.1.0 → llparse-0.1.2}/llparse/pybuilder/main_code.py

@@ -1,6 +1,16 @@
 import re
+import sys
 from dataclasses import dataclass
-from typing import Optional, TypeVar, Union
+from typing import Callable, Literal, Optional, TypeVar, Union
+
+if sys.version_info < (3, 10):
+    from typing_extensions import ParamSpec
+else:
+    from typing import ParamSpec
+
+_P = ParamSpec("_P")
+_T = TypeVar("_T")
+
 
 Signature = ["match", "value"]
 
@@ -18,7 +28,7 @@ def toBuffer(value: Union[str, int]):
 # TODO Add text validataion...
 
 
-def validate_text(init):
+def validate_text(init: Callable[_P, _T]) -> Callable[_P, _T]:
     def is_valid(args, kwargs):
         if kwargs.get("field"):
             field = kwargs["field"]
@@ -32,15 +42,20 @@ def validate_text(init):
 
 
 class Code:
-    def __init__(self, signature: str, name: str) -> None:
-        assert signature in Signature
+    def __init__(self, signature: Literal["match", "value"], name: str) -> None:
+        assert signature in Signature, "Invalid signature %s" % signature
 
         self.signature = signature
         self.name = name
 
+    def __hash__(self):
+        return hash(self.signature + self.name)
+
 
 class Field(Code):
-    def __init__(self, signature: str, name: str, field: str) -> None:
+    def __init__(
+        self, signature: Literal["match", "value"], name: str, field: str
+    ) -> None:
         self.field = field
         # if re.search(r"[//\s\\]+",field):
         #     raise TypeError(f"Can\'t access internal field from user code because the field: {name} conatins invalid characters")
@@ -48,11 +63,10 @@ class Field(Code):
 
 
 class FieldValue(Field):
-    # NOTE I Added for typehinting Here as it refuses to show up on the ide I'm using (Vizonex)
-
-    def __init__(self, signature: str, name: str, field: str, value: int) -> None:
+    def __init__(
+        self, signature: Literal["match", "value"], name: str, field: str, value: int
+    ) -> None:
         self.value = value
-        self.field = field
         super().__init__(signature, name, field)
 
 
@@ -131,6 +145,10 @@ class Node:
         self.otherwiseEdge: Optional["Edge"] = None
         self.privEdges: list["Edge"] = []
 
+    def key(self):
+        """reversed for sorting to prevent python from creating artificial randomness"""
+        return self.name
+
     def __hash__(self) -> int:
         return hash(self.name)
 
@@ -288,7 +306,7 @@ class Edge:
         self.node = node
         self.noAdvance = noAdvance
 
-        self.key = key
+        self.key = key.encode() if isinstance(key, str) else key
         self.value = value
 
         # Validation...

{llparse-0.1.0 → llparse-0.1.2}/llparse/pybuilder/parsemap.py

@@ -1,5 +1,4 @@
-
-from main_code import Edge, Node
+from ..pybuilder.main_code import Edge, Node
 
 
 class ParserMap:
@@ -9,9 +8,7 @@ class ParserMap:
     def Jsonize(self):
         queue = [self.root]
         seen: set[Node] = set()
-
-
-        while len(queue) != 0:
+        while queue:
             node = queue.pop()
 
             if node in seen:
@@ -33,5 +30,4 @@ class ParserMap:
             return {}
         data = edge.__dict__
         data["node"] = edge.node
-        # print(data["node"].name)
         return data

{llparse-0.1.0 → llparse-0.1.2}/llparse/pyfront/front.py

@@ -10,12 +10,12 @@ T = TypeVar("T")
 Signature = TypeVar("Signature", bytes, str)
 
 
-@dataclass
+@dataclass(unsafe_hash=True)
 class IWrap(Generic[T]):
     ref: T
 
-    def __hash__(self) -> int:
-        return hash(self.ref)
+    # def __hash__(self) -> int:
+    #     return hash(self.ref)
 
 
 def toCacheKey(value: Union[int, bool]) -> str:
@@ -33,6 +33,9 @@ class Code:
     cacheKey: str
     name: str
 
+    def __hash__(self):
+        return hash(self.cacheKey)
+
 
 class External(Code):
     """Inherits from the `Code` class as a subclass of `Code`"""
@@ -47,6 +50,9 @@ class Field(Code):
 
     field: str
 
+    def __hash__(self):
+        return hash(self.cacheKey)
+
 
 class FieldValueError(Exception):
     """FieldValue `value` must be integer"""
@@ -167,6 +173,9 @@ class IUniqueName:
     name: str
     originalName: str
 
+    def __hash__(self):
+        return hash(self.originalName)
+
 
 @dataclass
 class Identifier:

llparse-0.1.2/llparse/pyfront/namespace.py

@@ -0,0 +1,3 @@
+from ..pyfront import front as code
+from ..pyfront import nodes as node
+from ..pyfront import transform as transform