Design a JSON Merge API with Pluggable Behaviors

Problem Statement

Configuration management systems often need to merge JSON documents (such as base defaults and environment-specific overrides). In this task, you will design and implement a non-mutating JSON merge function that supports pluggable strategies for handling arrays.

Your function should accept a base document, a patch document, and an optional configuration controlling array merge behavior. Neither input should be modified in place.

Merge Rules

1. Objects (dicts): If both values are JSON objects, merge them recursively. Keys existing only in base are retained. Keys existing only in patch are added. Keys present in both are merged recursively.
2. Arrays (lists): If both values are JSON arrays, behavior is determined by array_strategy:
   • "overwrite" (default): The patch array replaces the base array.
   • "append": Elements from patch are appended to the end of base.
3. Type Mismatches & Primitives: If the types of base and patch do not match (e.g., a dict vs. an int) or if both values are primitive types (strings, numbers, booleans, null/None), the patch value replaces the base value.
4. Immutability: Neither base nor patch should be modified. The returned structure should be independent of changes to the inputs.

Interface

from typing import Any, Literal

def merge_json(
    base: Any,
    patch: Any,
    array_strategy: Literal["overwrite", "append"] = "overwrite"
) -> Any:
    pass

Examples

Example 1

base = {
    "device": "telemetry-v1",
    "enabled": True,
    "thresholds": {"speed": 120, "temp": 75},
    "tags": ["active"]
}
patch = {
    "enabled": False,
    "thresholds": {"temp": 80, "pressure": 32},
    "tags": ["verified"]
}

# With array_strategy="overwrite" (default)
result = merge_json(base, patch)
# Output:
# {
#     "device": "telemetry-v1",
#     "enabled": False,
#     "thresholds": {"speed": 120, "temp": 80, "pressure": 32},
#     "tags": ["verified"]
# }

Example 2

# Using the same base and patch with array_strategy="append"
result = merge_json(base, patch, array_strategy="append")
# Output:
# {
#     "device": "telemetry-v1",
#     "enabled": False,
#     "thresholds": {"speed": 120, "temp": 80, "pressure": 32},
#     "tags": ["active", "verified"]
# }

Constraints & Edge Cases

• Input structures consist of valid JSON types: dict, list, str, int, float, bool, and None.
• Maximum nesting depth is <= 100 levels.
• Assume no circular references exist within inputs.
• Ensure None in patch is treated as a valid value that overwrites base (e.g., merge_json({"a": 1}, {"a": None}) results in {"a": None}).