Fix nondeterministic recursive descent

Author: jg-rp

.gitignore

@@ -80,7 +80,6 @@ ENV/
 
 # Dev utils
 dev.py
-benchmark.py
 profile_.py
 
 # Test fixtures

jsonpath_rfc9535/segments.py

@@ -142,9 +142,10 @@ def _children(node: JSONPathNode) -> Iterable[JSONPathNode]:
         while queue:
             _node = queue.popleft()
             yield _node
+            # Visit child nodes now or queue them for later?
+            visit_children = random.choice([True, False])  # noqa: S311
             for child in _children(_node):
-                # Queue the child node or visit it now?
-                if random.choice([True, False]):  # noqa: S311
+                if visit_children:
                     queue.append(child)
                 else:
                     yield child

pyproject.toml

@@ -169,4 +169,4 @@ convention = "google"
 "scripts/__init__.py" = ["D104"]
 "tests/*" = ["D100", "D101", "D104", "D103"]
 "jsonpath_rfc9535/lex.py" = ["E741"]
-"scripts/nondeterministic_descent.py" = ["D103", "T201", "D101", "D102"]
+"scripts/nondeterministic_descent.py" = ["D103", "T201", "D101", "D102", "S311"]

scripts/benchmark.py

@@ -0,0 +1,115 @@
+import json
+import timeit
+from typing import Any
+from typing import Mapping
+from typing import NamedTuple
+from typing import Sequence
+from typing import Union
+
+# ruff: noqa: D100 D101 D103 T201
+
+
+class CTSCase(NamedTuple):
+    query: str
+    data: Union[Sequence[Any], Mapping[str, Any]]
+
+
+def valid_queries() -> Sequence[CTSCase]:
+    with open("tests/cts/cts.json") as fd:
+        data = json.load(fd)
+
+    return [
+        (CTSCase(t["selector"], t["document"]))
+        for t in data["tests"]
+        if not t.get("invalid_selector", False)
+    ]
+
+
+QUERIES = valid_queries()
+
+COMPILE_AND_FIND_SETUP = "from jsonpath_rfc9535 import query"
+
+COMPILE_AND_FIND_STMT = """\
+for path, data in QUERIES:
+    list(query(path, data))"""
+
+COMPILE_AND_FIND_VALUES_STMT = """\
+for path, data in QUERIES:
+    [node.value for node in query(path, data)]"""
+
+JUST_COMPILE_SETUP = "from jsonpath_rfc9535 import compile"
+
+JUST_COMPILE_STMT = """\
+for path, _ in QUERIES:
+    compile(path)"""
+
+JUST_FIND_SETUP = """\
+from jsonpath_rfc9535 import compile
+compiled_queries = [(compile(q), d) for q, d in QUERIES]
+"""
+
+JUST_FIND_STMT = """\
+for path, data in compiled_queries:
+    list(path.query(data))"""
+
+JUST_FIND_VALUES_STMT = """\
+for path, data in compiled_queries:
+    [node.value for node in path.query(data)]"""
+
+
+def benchmark(number: int = 100, best_of: int = 3) -> None:
+    print(f"repeating {len(QUERIES)} queries {number} times, best of {best_of} rounds")
+
+    results = timeit.repeat(
+        COMPILE_AND_FIND_STMT,
+        setup=COMPILE_AND_FIND_SETUP,
+        globals={"QUERIES": QUERIES},
+        number=number,
+        repeat=best_of,
+    )
+
+    print("compile and find".ljust(30), f"\033[92m{min(results):.3f}\033[0m")
+
+    results = timeit.repeat(
+        COMPILE_AND_FIND_VALUES_STMT,
+        setup=COMPILE_AND_FIND_SETUP,
+        globals={"QUERIES": QUERIES},
+        number=number,
+        repeat=best_of,
+    )
+
+    print("compile and find (values)".ljust(30), f"{min(results):.3f}")
+
+    results = timeit.repeat(
+        JUST_COMPILE_STMT,
+        setup=JUST_COMPILE_SETUP,
+        globals={"QUERIES": QUERIES},
+        number=number,
+        repeat=best_of,
+    )
+
+    print("just compile".ljust(30), f"{min(results):.3f}")
+
+    results = timeit.repeat(
+        JUST_FIND_STMT,
+        setup=JUST_FIND_SETUP,
+        globals={"QUERIES": QUERIES},
+        number=number,
+        repeat=best_of,
+    )
+
+    print("just find".ljust(30), f"\033[92m{min(results):.3f}\033[0m")
+
+    results = timeit.repeat(
+        JUST_FIND_VALUES_STMT,
+        setup=JUST_FIND_SETUP,
+        globals={"QUERIES": QUERIES},
+        number=number,
+        repeat=best_of,
+    )
+
+    print("just find (values)".ljust(30), f"{min(results):.3f}")
+
+
+if __name__ == "__main__":
+    benchmark()

scripts/nondeterministic_descent.py

@@ -2,6 +2,7 @@
 
 from __future__ import annotations
 
+import json
 import os
 import random
 import sys
@@ -69,6 +70,27 @@ def _visit(node: AuxNode, depth: int = 0) -> None:
         _visit(root)
         return root
 
+    @staticmethod
+    def collections(data: JSONLikeData) -> AuxNode:
+        def _visit(node: AuxNode, depth: int = 0) -> None:
+            if isinstance(node.value, dict):
+                for val in node.value.values():
+                    if isinstance(val, (list, dict)):
+                        _node = AuxNode(depth + 1, val)
+                        _visit(_node, depth + 1)
+                        node.children.append(_node)
+
+            elif isinstance(node.value, list):
+                for val in node.value:
+                    if isinstance(val, (list, dict)):
+                        _node = AuxNode(depth + 1, val)
+                        _visit(_node, depth + 1)
+                        node.children.append(_node)
+
+        root = AuxNode(0, data)
+        _visit(root)
+        return root
+
 
 def pptree(
     node: AuxNode,
@@ -115,13 +137,14 @@ def nondeterministic_visit(root: AuxNode) -> Iterable[AuxNode]:
     while queue:
         _node = queue.popleft()
         yield _node
+        # Visit child nodes now or queue them for later?
+        visit_children = random.choice([True, False])
         for child in _node.children:
-            # Queue the child node or visit it now?
-            if random.choice([True, False]):  # noqa: S311
-                queue.append(child)
-            else:
+            if visit_children:
                 yield child
                 queue.extend(child.children)
+            else:
+                queue.append(child)
 
 
 def get_perms(root: AuxNode) -> List[Tuple[AuxNode, ...]]:
@@ -130,22 +153,43 @@ def get_perms(root: AuxNode) -> List[Tuple[AuxNode, ...]]:
     return sorted(perms, key=lambda t: str(t))
 
 
-def pp_json_path_data(data: JSONLikeData) -> None:
+def pp_json_path_perms(data: JSONLikeData) -> None:
+    print("Input data")
+    print(f"\033[92m{data}\033[0m")
     aux_tree = AuxNode.from_(data)
+    print("\nTree view")
     pptree(aux_tree)
 
-    print("\nPre order\n")
+    print("\nPre order")
     print(", ".join(str(n) for n in pre_order_visit(aux_tree)))
 
-    print("\nLevel order\n")
+    print("\nLevel order")
     print(", ".join(str(n) for n in breadth_first_visit(aux_tree)))
 
-    print("\nNondeterministic order\n")
+    print("\nNondeterministic order")
+    for perm in get_perms(aux_tree):
+        print(", ".join(str(node) for node in perm))
+
+    print("\n---\n\nCollections only")
+    aux_tree = AuxNode.collections(data)
+    pptree(aux_tree)
+
+    print("\nPre order")
+    print(", ".join(str(n) for n in pre_order_visit(aux_tree)))
+
+    print("\nLevel order")
+    print(", ".join(str(n) for n in breadth_first_visit(aux_tree)))
+
+    print("\nNondeterministic order")
     for perm in get_perms(aux_tree):
         print(", ".join(str(node) for node in perm))
 
 
 if __name__ == "__main__":
-    # basic, descendant segment, name shorthand
-    data = {"o": [{"a": "b"}, {"a": "c"}]}
-    pp_json_path_data(data)
+    if len(sys.argv) < 2:  # noqa: PLR2004
+        print("error: no data to process")
+        print(f"usage: {sys.argv[0]} <JSON string>")
+        sys.exit(1)
+
+    data = json.loads(sys.argv[1])
+    pp_json_path_perms(data)