From 81d16e96fd29d3395e20609627f2bb4f72604ee3 Mon Sep 17 00:00:00 2001
From: Nick Sweeting <github@sweeting.me>
Date: Tue, 1 Oct 2024 21:43:35 -0700
Subject: [PATCH] fix toml_util circular import in abx

---
 archivebox/abx/archivebox/base_configset.py |   2 +-
 archivebox/abx/archivebox/toml_util.py      | 114 ++++++++++++++++++++
 2 files changed, 115 insertions(+), 1 deletion(-)
 create mode 100644 archivebox/abx/archivebox/toml_util.py

diff --git a/archivebox/abx/archivebox/base_configset.py b/archivebox/abx/archivebox/base_configset.py
index 8a8298a3..4a7e8c13 100644
--- a/archivebox/abx/archivebox/base_configset.py
+++ b/archivebox/abx/archivebox/base_configset.py
@@ -14,7 +14,7 @@ from pydantic_pkgr.base_types import func_takes_args_or_kwargs
 import abx
 
 from .base_hook import BaseHook, HookType
-from archivebox.misc import toml_util
+from . import toml_util
 
 
 PACKAGE_DIR = Path(__file__).resolve().parent.parent
diff --git a/archivebox/abx/archivebox/toml_util.py b/archivebox/abx/archivebox/toml_util.py
new file mode 100644
index 00000000..d4784335
--- /dev/null
+++ b/archivebox/abx/archivebox/toml_util.py
@@ -0,0 +1,114 @@
+from typing import Any, List, Callable
+
+import json
+import ast
+import inspect
+import toml
+import re
+import configparser
+
+from pathlib import Path, PosixPath
+
+from pydantic.json_schema import GenerateJsonSchema
+from pydantic_core import to_jsonable_python
+
+JSONValue = str | bool | int | None | List['JSONValue']
+
+TOML_HEADER = "# Converted from INI to TOML format: https://toml.io/en/\n\n"
+
+def load_ini_value(val: str) -> JSONValue:
+    """Convert lax INI values into strict TOML-compliant (JSON) values"""
+    if val.lower() in ('true', 'yes', '1'):
+        return True
+    if val.lower() in ('false', 'no', '0'):
+        return False
+    if val.isdigit():
+        return int(val)
+
+    try:
+        return ast.literal_eval(val)
+    except Exception:
+        pass
+
+    try:
+        return json.loads(val)
+    except Exception:
+        pass
+    
+    return val
+
+
+def convert(ini_str: str) -> str:
+    """Convert a string of INI config into its TOML equivalent (warning: strips comments)"""
+
+    config = configparser.ConfigParser()
+    config.optionxform = str  # capitalize key names
+    config.read_string(ini_str)
+
+    # Initialize an empty dictionary to store the TOML representation
+    toml_dict = {}
+
+    # Iterate over each section in the INI configuration
+    for section in config.sections():
+        toml_dict[section] = {}
+
+        # Iterate over each key-value pair in the section
+        for key, value in config.items(section):
+            parsed_value = load_ini_value(value)
+
+            # Convert the parsed value to its TOML-compatible JSON representation
+            toml_dict[section.upper()][key.upper()] = json.dumps(parsed_value)
+
+    # Build the TOML string
+    toml_str = TOML_HEADER
+    for section, items in toml_dict.items():
+        toml_str += f"[{section}]\n"
+        for key, value in items.items():
+            toml_str += f"{key} = {value}\n"
+        toml_str += "\n"
+
+    return toml_str.strip()
+
+
+
+class JSONSchemaWithLambdas(GenerateJsonSchema):
+    """
+    Encode lambda functions in default values properly.
+    Usage:
+    >>> json.dumps(value, encoder=JSONSchemaWithLambdas())
+    """
+    def encode_default(self, default: Any) -> Any:
+        config = self._config
+        if isinstance(default, Callable):
+            return '{{lambda ' + inspect.getsource(default).split('=lambda ')[-1].strip()[:-1] + '}}'
+        return to_jsonable_python(
+           default,
+           timedelta_mode=config.ser_json_timedelta,
+           bytes_mode=config.ser_json_bytes,
+           serialize_unknown=True
+        )
+
+    # for computed_field properties render them like this instead:
+    # inspect.getsource(field.wrapped_property.fget).split('def ', 1)[-1].split('\n', 1)[-1].strip().strip('return '),
+
+
+def better_toml_dump_str(val: Any) -> str:
+    try:
+        return toml.encoder._dump_str(val)     # type: ignore
+    except Exception:
+        # if we hit any of toml's numerous encoding bugs,
+        # fall back to using json representation of string
+        return json.dumps(str(val))
+
+class CustomTOMLEncoder(toml.encoder.TomlEncoder):
+    """
+    Custom TomlEncoder to work around https://github.com/uiri/toml's many encoding bugs.
+    More info: https://github.com/fabiocaccamo/python-benedict/issues/439
+    >>> toml.dumps(value, encoder=CustomTOMLEncoder())
+    """
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+        self.dump_funcs[Path] = lambda x: json.dumps(str(x))
+        self.dump_funcs[PosixPath] = lambda x: json.dumps(str(x))
+        self.dump_funcs[str] = better_toml_dump_str
+        self.dump_funcs[re.RegexFlag] = better_toml_dump_str