oj-toolkit 0.2.0__tar.gz → 0.2.2__tar.gz

{oj_toolkit-0.2.0 → oj_toolkit-0.2.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: oj-toolkit
-Version: 0.2.0
+Version: 0.2.2
 Summary: Centralized shared utilities for ownjoo-org projects: standardized logging config, type validation, data parsing, console output, and async utilities
 Author-email: Speedy <speedy@ownjoo.org>
 License: MIT
@@ -33,7 +33,7 @@ Requires-Dist: mypy>=1.0; extra == "dev"
 Requires-Dist: pylint>=2.0; extra == "dev"
 Dynamic: license-file
 
-# ownjoo-toolkit
+# oj-toolkit
 
 Centralized shared utilities library for all ownjoo-org projects. Provides battle-tested functions for type validation, data parsing, and progress logging.
 
@@ -187,11 +187,19 @@ fields = dig_many(
 # Bind a path once, reuse it against many objects
 get_name = Digger(path='name', exp=str, default='')
 [get_name(user) for user in data['users']]  # Returns: ['Alice', 'Bob']
+
+# Constrain a string result with a regex (checked via re.fullmatch after conversion)
+mac = dig({'device': {'mac': 'AA:BB:CC:DD:EE:FF'}}, path='device.mac', exp=str,
+          pattern=r'([0-9A-Fa-f]{2}:){5}[0-9A-Fa-f]{2}')
+# Returns: 'AA:BB:CC:DD:EE:FF'
 ```
 
-> **Note:** `dig()`'s default `post_processor` (`validate()`) only accepts a value if you pass
-> `exp=<type>` -- without it, validation always fails and you get `default` back (`None` unless set).
-> Pass `exp=` for a real value, or `post_processor=None` to skip validation and get the raw match.
+> **Note:** if the path resolves to nothing (missing key/index, or `src` doesn't contain it),
+> `dig()`'s default `post_processor` (`validate()`) still runs and returns `default` (`None`
+> unless set) -- a missing path never silently drops `default` in favor of a bare `None`.
+> Passing no `exp=` means "accept any type, no isinstance check" -- not "always fail". Pass
+> `exp=<type>` to add a type check, or `post_processor=None` to bypass validation entirely and
+> get the raw match untouched.
 
 ### Logging Setup
 
@@ -711,18 +719,25 @@ else:
 
 Returns `True` if the terminal likely supports Unicode characters. Used internally to choose between ASCII and Unicode box/table borders.
 
+Checks `NO_COLOR`/`CI`/`TERM` env vars first, then verifies `sys.stdout.encoding` can actually
+encode a box-drawing character -- this catches Windows consoles that default to a codepage
+(`cp1252`, `cp437`, etc.) that can't, even though the platform/terminal is otherwise
+Unicode-capable. Set `PYTHONIOENCODING=utf-8` (or use Windows Terminal with a UTF-8 codepage)
+to get Unicode borders on Windows instead of the ASCII fallback.
+
 ### `parsing` Module
 
-#### `validate(v, exp=None, default=None, converter=None, validator=None, **kwargs)`
+#### `validate(v, exp=None, default=None, converter=None, validator=None, pattern=None, **kwargs)`
 
 Generic validation utility that converts and validates a value.
 
 - **Parameters:**
   - `v` (Any): The value to validate
-  - `exp` (Type): Expected type. Enables automatic converter selection (list → str_to_list, datetime → get_datetime)
-  - `default` (Any): Return this if validation fails. Default: None
-  - `converter` (Callable): Custom converter function. Default: auto-selected based on exp
-  - `validator` (Callable): Custom validator function(result, exp, **kwargs) → bool. Default: isinstance check
+  - `exp` (Type): Expected type. Enables automatic converter selection (list → str_to_list, datetime → get_datetime). If `None`, no isinstance check is performed -- any type passes (see `validator` below)
+  - `default` (Any): Return this if validation fails, or `v` is missing/`None`. Default: None
+  - `converter` (Callable): Custom converter function. Default: auto-selected based on exp. Auto-selection is limited to `list`/`datetime` -- `bool`/`int`/`float`/etc. are pass-through unless you supply your own `converter=`
+  - `validator` (Callable): Custom validator function(result, exp, **kwargs) → bool. Default: isinstance check (`exp=None` always passes; a non-type `exp` is rejected rather than raising)
+  - `pattern` (str | re.Pattern): Optional regex checked via `re.fullmatch` against the result, but only when the result is a `str` -- ignored for other types. Applied after conversion/validation succeed; a mismatch falls back to `default`
   - `**kwargs`: Passed to converter and validator
 
 - **Returns:** Converted and validated value, or default if validation fails
@@ -734,6 +749,16 @@ Generic validation utility that converts and validates a value.
 numbers = validate('1,2,3', exp=list)  # Returns: ['1', '2', '3']
 dt = validate('2024-01-15T10:30:00', exp=datetime)  # Returns: datetime object
 
+# exp=bool/int/float do NOT auto-coerce strings -- pass converter= explicitly for that
+validate('42', exp=int)                    # Returns: None ('42' is a str, not an int)
+validate('42', exp=int, converter=int)     # Returns: 42
+
+# Regex-constrain a string result
+validate('AA:BB:CC:DD:EE:FF', exp=str, pattern=r'([0-9A-Fa-f]{2}:){5}[0-9A-Fa-f]{2}')
+# Returns: 'AA:BB:CC:DD:EE:FF'
+validate('not-a-mac', exp=str, pattern=r'([0-9A-Fa-f]{2}:){5}[0-9A-Fa-f]{2}', default='invalid')
+# Returns: 'invalid'
+
 # Custom converter and validator
 def to_upper(v, *args, **kwargs):
     return str(v).upper()
@@ -795,6 +820,12 @@ get_datetime('01/15/2024 10:30:00', format_str='%m/%d/%Y %H:%M:%S')  # Custom
 Extract and post-process a value from a nested data structure, using [jmespath](https://jmespath.org/)
 under the hood (`jmespath` is a required dependency).
 
+> **Compilation caching:** compiled jmespath expressions are cached in a process-global
+> `lru_cache(maxsize=256)`, keyed by the expression string. A bare `dig()` call already avoids
+> recompiling a repeated path -- this isn't a `Digger`-only optimization. The cache holds up to
+> 256 distinct expressions across the whole process (shared by every `dig()`/`Digger` caller),
+> LRU-evicted beyond that.
+
 - **Parameters:**
   - `src` (Mapping | Sequence): Data structure to navigate
   - `path` (int | str | list | None):
@@ -811,16 +842,18 @@ under the hood (`jmespath` is a required dependency).
     Pass `None` to skip post-processing and get the raw match.
   - `**kwargs`: Passed to `post_processor`
 
-- **Returns:** Post-processed value, or `None` if extraction fails, no candidate path matches, or
-  no post-processor is specified and nothing was found
+- **Returns:** Post-processed value. If extraction fails or no candidate path matches, the
+  post-processor (default `validate()`) still runs and returns `default` (`None` unless set) --
+  unless `post_processor=None`, in which case the raw (possibly missing/`None`) value is returned
+  untouched with no default substitution.
 
 > **Type checkers:** pass `exp=<type>` to get a precise `<type> | None` return type instead of `Any`
 > (`dig(data, path='...', exp=str)` type-checks as `str | None`). Passing `post_processor=None` types
 > as `Any` (raw, unvalidated value). A custom `post_processor` types as that callable's own return type.
 
-> **Passing no `exp`:** the default `validate()` post-processor only accepts a value when you give
-> it `exp=<type>` -- without one, validation always fails and you get `default` back (`None` unless
-> set). Either pass `exp=`, or `post_processor=None` to bypass validation entirely.
+> **Passing no `exp`:** the default `validate()` post-processor treats a missing `exp` as "accept
+> any type" -- it does not fail. Pass `exp=<type>` to add an isinstance check, or
+> `post_processor=None` to bypass validation entirely and get the raw match.
 
 **Example:**
 
@@ -855,6 +888,39 @@ name = dig(data, path=['response.users[0].display_name', 'response.users[0].name
 # Returns: 'Alice'
 ```
 
+##### `dig()` Behavior Matrix
+
+Every row below is a real `dig()` call, backed by a passing unit test in
+`test/unit/parsing/test_parsing.py` -- the test name is the proof, not just a claim. This isn't
+every possible combination of kwargs (that cross product is huge and mostly redundant); it's one
+representative call per distinct decision point in `dig()`/`validate()`. `*(unset)*` means the
+kwarg wasn't passed (the function's own default applies); `--` means it isn't relevant to that row.
+
+| `src` | `path` | `exp` | `default` | `post_processor` | `pattern` | `validator` | `pop` | Result | Proven by |
+|---|---|---|---|---|---|---|---|---|---|
+| `{'first': 'a', 'second': ['blah']}` | `'second[0]'` (found) | `str` | *(unset)* | *(unset)* | *(unset)* | *(unset)* | `False` | `'blah'` | `test_should_get_value_from_dict` |
+| `['', ['blah']]` | `'[1][0]'` (found) | `str` | `''` | *(unset)* | *(unset)* | *(unset)* | `False` | `'blah'` (default unused -- value was valid) | `test_should_get_value_from_list` |
+| `{'a': {'b': 'value'}}` | `'x.y.z'` (missing) | `str` | `'default'` | *(unset)* | *(unset)* | *(unset)* | `False` | `'default'` | `test_should_return_default_on_missing_intermediate_key` |
+| `{'items': ['a', 'b', 'c']}` | `'items.invalid_index'` (missing) | `str` | `'default'` | *(unset)* | *(unset)* | *(unset)* | `False` | `'default'` | `test_should_return_default_when_accessing_list_with_string_key` |
+| `{'a': {'b': 'value'}}` | `'x.y.z'` (missing) | `str` | *(unset)* | *(unset)* | *(unset)* | *(unset)* | `False` | `None` (no default set) | `test_should_return_none_for_missing_path_with_exp_and_no_default` |
+| `{'a': {'b': 'c'}}` | `'x.y.z'` (missing) | *(unset)* | *(unset)* | *(unset)* | *(unset)* | *(unset)* | `False` | `None` (`exp=None` means "anything passes"; the found value already is `None`) | `test_should_handle_get_value_with_invalid_path` |
+| `{'key': 'blah'}` | `'key'` (found) | *(unset)* | *(unset)* | *(unset)* | *(unset)* | *(unset)* | `False` | `'blah'` (raw pass-through, no isinstance check) | `test_should_not_crash_when_exp_omitted_from_dig` |
+| `{'a': 'not-an-int'}` | `'a'` (found, wrong type) | `int` | *(unset)* | *(unset)* | *(unset)* | *(unset)* | `False` | `None` | `test_should_return_none_for_dig_result_type_mismatch` |
+| `{'a': 'not-an-int'}` | `'a'` (found, wrong type) | `int` | `-1` | *(unset)* | *(unset)* | *(unset)* | `False` | `-1` | `test_should_return_default_for_dig_result_type_mismatch` |
+| `{'a': 'blah'}` | `'a'` (found, non-type `exp`) | `'str'` | `'fallback'` | *(unset)* | *(unset)* | *(unset)* | `False` | `'fallback'` (garbage `exp` rejected, not raised) | `test_should_return_default_for_dig_with_non_type_expected_type` |
+| `{'key': 'raw_value'}` | `'key'` (found) | -- | -- | `None` | -- | -- | `False` | raw value, unvalidated | `test_should_return_raw_value_when_no_post_processor` |
+| `{'a': 1}` | `'missing'` (missing) | -- | `'default'` (ignored) | `None` | -- | -- | `False` | `None` (`default` is never consulted -- it's `validate()`'s kwarg, and `validate()` never runs) | `test_should_return_none_for_missing_path_with_no_post_processor` |
+| `{'response': {'users': [{'name': 'Alice'}, {'name': 'Bob'}]}}` | `'response.users'` (found) | -- | -- | `len` | -- | -- | `False` | `2` (custom post_processor gets the raw match) | `test_should_use_custom_post_processor_function` |
+| `'blah'` | `None` (src is the value) | `str` | *(unset)* | *(unset)* | *(unset)* | custom lambda | `False` | `'blah'` (custom validator overrides isinstance check) | `test_should_get_value_with_passed_validator` |
+| `'test_string'` | `None` (src is the value) | `str` | *(unset)* | *(unset)* | *(unset)* | *(unset)* | `False` | the src value itself | `test_should_post_process_source_when_path_is_none` |
+| `{'device': {'mac': 'AA:BB:CC:DD:EE:FF'}}` | `'device.mac'` (found) | `str` | *(unset)* | *(unset)* | matching regex | *(unset)* | `False` | matched MAC string | `test_should_validate_dig_result_with_matching_pattern` |
+| `{'device': {'mac': 'not-a-mac'}}` | `'device.mac'` (found) | `str` | `'invalid'` | *(unset)* | non-matching regex | *(unset)* | `False` | `'invalid'` | `test_should_return_default_for_dig_result_with_non_matching_pattern` |
+| `{'a': 1, 'b': 2}` | `'a'` (found) | `int` | *(unset)* | *(unset)* | -- | -- | `True` | `1`, and `'a'` removed from `src` | `test_should_pop_key_from_dict` |
+| `{'items': [{'id': 1}, {'id': 2}]}` | `'items[?id==\`2\`].id'` (found, ambiguous) | `list` | *(unset)* | *(unset)* | -- | -- | `True` (refused) | `[2]`, `src` left unmutated (pop refused + warning logged) | `test_should_refuse_pop_with_ambiguous_jmespath_filter` |
+| `{'user': {'name': 'Alice'}}` | `['user.nickname', 'user.name']` (fallback) | `str` | *(unset)* | *(unset)* | -- | -- | `False` | `'Alice'` (first candidate wins) | `test_should_use_first_matching_path_in_fallback_list` |
+| `{'user': {'name': 'Alice'}}` | `['user.nickname', 'user.alias']` (fallback, none match) | `str` | *(unset)* | *(unset)* | -- | -- | `False` | `None` | `test_should_return_none_when_no_fallback_path_matches` |
+| `{'user': {'name': 'Alice', 'nickname': 'Ali'}}` | `['user.nickname', 'user.name']` (fallback) | `str` | *(unset)* | *(unset)* | -- | -- | `True` | `'Ali'`, only the winning candidate's key removed | `test_should_pop_only_the_winning_fallback_path` |
+
 #### `dig_many(src, paths, **common_kwargs)`
 
 Extract several named fields from `src` in one call.
@@ -889,7 +955,9 @@ fields = dig_many(
 
 A pre-bound, reusable `dig()` call. Validates/compiles the jmespath expression once at
 construction (failing fast on a bad expression instead of on first use), then can be called
-like a function against any number of `src` objects without repeating `path`/`exp`/kwargs.
+like a function against any number of `src` objects without repeating `path`/`exp`/kwargs. A
+string `pattern=` kwarg is likewise pre-compiled to a `re.Pattern` at construction time, so
+repeated calls don't re-compile the regex.
 
 **Example:**
 

{oj_toolkit-0.2.0 → oj_toolkit-0.2.2}/README.md

@@ -1,4 +1,4 @@
-# ownjoo-toolkit
+# oj-toolkit
 
 Centralized shared utilities library for all ownjoo-org projects. Provides battle-tested functions for type validation, data parsing, and progress logging.
 
@@ -152,11 +152,19 @@ fields = dig_many(
 # Bind a path once, reuse it against many objects
 get_name = Digger(path='name', exp=str, default='')
 [get_name(user) for user in data['users']]  # Returns: ['Alice', 'Bob']
+
+# Constrain a string result with a regex (checked via re.fullmatch after conversion)
+mac = dig({'device': {'mac': 'AA:BB:CC:DD:EE:FF'}}, path='device.mac', exp=str,
+          pattern=r'([0-9A-Fa-f]{2}:){5}[0-9A-Fa-f]{2}')
+# Returns: 'AA:BB:CC:DD:EE:FF'
 ```
 
-> **Note:** `dig()`'s default `post_processor` (`validate()`) only accepts a value if you pass
-> `exp=<type>` -- without it, validation always fails and you get `default` back (`None` unless set).
-> Pass `exp=` for a real value, or `post_processor=None` to skip validation and get the raw match.
+> **Note:** if the path resolves to nothing (missing key/index, or `src` doesn't contain it),
+> `dig()`'s default `post_processor` (`validate()`) still runs and returns `default` (`None`
+> unless set) -- a missing path never silently drops `default` in favor of a bare `None`.
+> Passing no `exp=` means "accept any type, no isinstance check" -- not "always fail". Pass
+> `exp=<type>` to add a type check, or `post_processor=None` to bypass validation entirely and
+> get the raw match untouched.
 
 ### Logging Setup
 
@@ -676,18 +684,25 @@ else:
 
 Returns `True` if the terminal likely supports Unicode characters. Used internally to choose between ASCII and Unicode box/table borders.
 
+Checks `NO_COLOR`/`CI`/`TERM` env vars first, then verifies `sys.stdout.encoding` can actually
+encode a box-drawing character -- this catches Windows consoles that default to a codepage
+(`cp1252`, `cp437`, etc.) that can't, even though the platform/terminal is otherwise
+Unicode-capable. Set `PYTHONIOENCODING=utf-8` (or use Windows Terminal with a UTF-8 codepage)
+to get Unicode borders on Windows instead of the ASCII fallback.
+
 ### `parsing` Module
 
-#### `validate(v, exp=None, default=None, converter=None, validator=None, **kwargs)`
+#### `validate(v, exp=None, default=None, converter=None, validator=None, pattern=None, **kwargs)`
 
 Generic validation utility that converts and validates a value.
 
 - **Parameters:**
   - `v` (Any): The value to validate
-  - `exp` (Type): Expected type. Enables automatic converter selection (list → str_to_list, datetime → get_datetime)
-  - `default` (Any): Return this if validation fails. Default: None
-  - `converter` (Callable): Custom converter function. Default: auto-selected based on exp
-  - `validator` (Callable): Custom validator function(result, exp, **kwargs) → bool. Default: isinstance check
+  - `exp` (Type): Expected type. Enables automatic converter selection (list → str_to_list, datetime → get_datetime). If `None`, no isinstance check is performed -- any type passes (see `validator` below)
+  - `default` (Any): Return this if validation fails, or `v` is missing/`None`. Default: None
+  - `converter` (Callable): Custom converter function. Default: auto-selected based on exp. Auto-selection is limited to `list`/`datetime` -- `bool`/`int`/`float`/etc. are pass-through unless you supply your own `converter=`
+  - `validator` (Callable): Custom validator function(result, exp, **kwargs) → bool. Default: isinstance check (`exp=None` always passes; a non-type `exp` is rejected rather than raising)
+  - `pattern` (str | re.Pattern): Optional regex checked via `re.fullmatch` against the result, but only when the result is a `str` -- ignored for other types. Applied after conversion/validation succeed; a mismatch falls back to `default`
   - `**kwargs`: Passed to converter and validator
 
 - **Returns:** Converted and validated value, or default if validation fails
@@ -699,6 +714,16 @@ Generic validation utility that converts and validates a value.
 numbers = validate('1,2,3', exp=list)  # Returns: ['1', '2', '3']
 dt = validate('2024-01-15T10:30:00', exp=datetime)  # Returns: datetime object
 
+# exp=bool/int/float do NOT auto-coerce strings -- pass converter= explicitly for that
+validate('42', exp=int)                    # Returns: None ('42' is a str, not an int)
+validate('42', exp=int, converter=int)     # Returns: 42
+
+# Regex-constrain a string result
+validate('AA:BB:CC:DD:EE:FF', exp=str, pattern=r'([0-9A-Fa-f]{2}:){5}[0-9A-Fa-f]{2}')
+# Returns: 'AA:BB:CC:DD:EE:FF'
+validate('not-a-mac', exp=str, pattern=r'([0-9A-Fa-f]{2}:){5}[0-9A-Fa-f]{2}', default='invalid')
+# Returns: 'invalid'
+
 # Custom converter and validator
 def to_upper(v, *args, **kwargs):
     return str(v).upper()
@@ -760,6 +785,12 @@ get_datetime('01/15/2024 10:30:00', format_str='%m/%d/%Y %H:%M:%S')  # Custom
 Extract and post-process a value from a nested data structure, using [jmespath](https://jmespath.org/)
 under the hood (`jmespath` is a required dependency).
 
+> **Compilation caching:** compiled jmespath expressions are cached in a process-global
+> `lru_cache(maxsize=256)`, keyed by the expression string. A bare `dig()` call already avoids
+> recompiling a repeated path -- this isn't a `Digger`-only optimization. The cache holds up to
+> 256 distinct expressions across the whole process (shared by every `dig()`/`Digger` caller),
+> LRU-evicted beyond that.
+
 - **Parameters:**
   - `src` (Mapping | Sequence): Data structure to navigate
   - `path` (int | str | list | None):
@@ -776,16 +807,18 @@ under the hood (`jmespath` is a required dependency).
     Pass `None` to skip post-processing and get the raw match.
   - `**kwargs`: Passed to `post_processor`
 
-- **Returns:** Post-processed value, or `None` if extraction fails, no candidate path matches, or
-  no post-processor is specified and nothing was found
+- **Returns:** Post-processed value. If extraction fails or no candidate path matches, the
+  post-processor (default `validate()`) still runs and returns `default` (`None` unless set) --
+  unless `post_processor=None`, in which case the raw (possibly missing/`None`) value is returned
+  untouched with no default substitution.
 
 > **Type checkers:** pass `exp=<type>` to get a precise `<type> | None` return type instead of `Any`
 > (`dig(data, path='...', exp=str)` type-checks as `str | None`). Passing `post_processor=None` types
 > as `Any` (raw, unvalidated value). A custom `post_processor` types as that callable's own return type.
 
-> **Passing no `exp`:** the default `validate()` post-processor only accepts a value when you give
-> it `exp=<type>` -- without one, validation always fails and you get `default` back (`None` unless
-> set). Either pass `exp=`, or `post_processor=None` to bypass validation entirely.
+> **Passing no `exp`:** the default `validate()` post-processor treats a missing `exp` as "accept
+> any type" -- it does not fail. Pass `exp=<type>` to add an isinstance check, or
+> `post_processor=None` to bypass validation entirely and get the raw match.
 
 **Example:**
 
@@ -820,6 +853,39 @@ name = dig(data, path=['response.users[0].display_name', 'response.users[0].name
 # Returns: 'Alice'
 ```
 
+##### `dig()` Behavior Matrix
+
+Every row below is a real `dig()` call, backed by a passing unit test in
+`test/unit/parsing/test_parsing.py` -- the test name is the proof, not just a claim. This isn't
+every possible combination of kwargs (that cross product is huge and mostly redundant); it's one
+representative call per distinct decision point in `dig()`/`validate()`. `*(unset)*` means the
+kwarg wasn't passed (the function's own default applies); `--` means it isn't relevant to that row.
+
+| `src` | `path` | `exp` | `default` | `post_processor` | `pattern` | `validator` | `pop` | Result | Proven by |
+|---|---|---|---|---|---|---|---|---|---|
+| `{'first': 'a', 'second': ['blah']}` | `'second[0]'` (found) | `str` | *(unset)* | *(unset)* | *(unset)* | *(unset)* | `False` | `'blah'` | `test_should_get_value_from_dict` |
+| `['', ['blah']]` | `'[1][0]'` (found) | `str` | `''` | *(unset)* | *(unset)* | *(unset)* | `False` | `'blah'` (default unused -- value was valid) | `test_should_get_value_from_list` |
+| `{'a': {'b': 'value'}}` | `'x.y.z'` (missing) | `str` | `'default'` | *(unset)* | *(unset)* | *(unset)* | `False` | `'default'` | `test_should_return_default_on_missing_intermediate_key` |
+| `{'items': ['a', 'b', 'c']}` | `'items.invalid_index'` (missing) | `str` | `'default'` | *(unset)* | *(unset)* | *(unset)* | `False` | `'default'` | `test_should_return_default_when_accessing_list_with_string_key` |
+| `{'a': {'b': 'value'}}` | `'x.y.z'` (missing) | `str` | *(unset)* | *(unset)* | *(unset)* | *(unset)* | `False` | `None` (no default set) | `test_should_return_none_for_missing_path_with_exp_and_no_default` |
+| `{'a': {'b': 'c'}}` | `'x.y.z'` (missing) | *(unset)* | *(unset)* | *(unset)* | *(unset)* | *(unset)* | `False` | `None` (`exp=None` means "anything passes"; the found value already is `None`) | `test_should_handle_get_value_with_invalid_path` |
+| `{'key': 'blah'}` | `'key'` (found) | *(unset)* | *(unset)* | *(unset)* | *(unset)* | *(unset)* | `False` | `'blah'` (raw pass-through, no isinstance check) | `test_should_not_crash_when_exp_omitted_from_dig` |
+| `{'a': 'not-an-int'}` | `'a'` (found, wrong type) | `int` | *(unset)* | *(unset)* | *(unset)* | *(unset)* | `False` | `None` | `test_should_return_none_for_dig_result_type_mismatch` |
+| `{'a': 'not-an-int'}` | `'a'` (found, wrong type) | `int` | `-1` | *(unset)* | *(unset)* | *(unset)* | `False` | `-1` | `test_should_return_default_for_dig_result_type_mismatch` |
+| `{'a': 'blah'}` | `'a'` (found, non-type `exp`) | `'str'` | `'fallback'` | *(unset)* | *(unset)* | *(unset)* | `False` | `'fallback'` (garbage `exp` rejected, not raised) | `test_should_return_default_for_dig_with_non_type_expected_type` |
+| `{'key': 'raw_value'}` | `'key'` (found) | -- | -- | `None` | -- | -- | `False` | raw value, unvalidated | `test_should_return_raw_value_when_no_post_processor` |
+| `{'a': 1}` | `'missing'` (missing) | -- | `'default'` (ignored) | `None` | -- | -- | `False` | `None` (`default` is never consulted -- it's `validate()`'s kwarg, and `validate()` never runs) | `test_should_return_none_for_missing_path_with_no_post_processor` |
+| `{'response': {'users': [{'name': 'Alice'}, {'name': 'Bob'}]}}` | `'response.users'` (found) | -- | -- | `len` | -- | -- | `False` | `2` (custom post_processor gets the raw match) | `test_should_use_custom_post_processor_function` |
+| `'blah'` | `None` (src is the value) | `str` | *(unset)* | *(unset)* | *(unset)* | custom lambda | `False` | `'blah'` (custom validator overrides isinstance check) | `test_should_get_value_with_passed_validator` |
+| `'test_string'` | `None` (src is the value) | `str` | *(unset)* | *(unset)* | *(unset)* | *(unset)* | `False` | the src value itself | `test_should_post_process_source_when_path_is_none` |
+| `{'device': {'mac': 'AA:BB:CC:DD:EE:FF'}}` | `'device.mac'` (found) | `str` | *(unset)* | *(unset)* | matching regex | *(unset)* | `False` | matched MAC string | `test_should_validate_dig_result_with_matching_pattern` |
+| `{'device': {'mac': 'not-a-mac'}}` | `'device.mac'` (found) | `str` | `'invalid'` | *(unset)* | non-matching regex | *(unset)* | `False` | `'invalid'` | `test_should_return_default_for_dig_result_with_non_matching_pattern` |
+| `{'a': 1, 'b': 2}` | `'a'` (found) | `int` | *(unset)* | *(unset)* | -- | -- | `True` | `1`, and `'a'` removed from `src` | `test_should_pop_key_from_dict` |
+| `{'items': [{'id': 1}, {'id': 2}]}` | `'items[?id==\`2\`].id'` (found, ambiguous) | `list` | *(unset)* | *(unset)* | -- | -- | `True` (refused) | `[2]`, `src` left unmutated (pop refused + warning logged) | `test_should_refuse_pop_with_ambiguous_jmespath_filter` |
+| `{'user': {'name': 'Alice'}}` | `['user.nickname', 'user.name']` (fallback) | `str` | *(unset)* | *(unset)* | -- | -- | `False` | `'Alice'` (first candidate wins) | `test_should_use_first_matching_path_in_fallback_list` |
+| `{'user': {'name': 'Alice'}}` | `['user.nickname', 'user.alias']` (fallback, none match) | `str` | *(unset)* | *(unset)* | -- | -- | `False` | `None` | `test_should_return_none_when_no_fallback_path_matches` |
+| `{'user': {'name': 'Alice', 'nickname': 'Ali'}}` | `['user.nickname', 'user.name']` (fallback) | `str` | *(unset)* | *(unset)* | -- | -- | `True` | `'Ali'`, only the winning candidate's key removed | `test_should_pop_only_the_winning_fallback_path` |
+
 #### `dig_many(src, paths, **common_kwargs)`
 
 Extract several named fields from `src` in one call.
@@ -854,7 +920,9 @@ fields = dig_many(
 
 A pre-bound, reusable `dig()` call. Validates/compiles the jmespath expression once at
 construction (failing fast on a bad expression instead of on first use), then can be called
-like a function against any number of `src` objects without repeating `path`/`exp`/kwargs.
+like a function against any number of `src` objects without repeating `path`/`exp`/kwargs. A
+string `pattern=` kwarg is likewise pre-compiled to a `re.Pattern` at construction time, so
+repeated calls don't re-compile the regex.
 
 **Example:**
 

{oj_toolkit-0.2.0 → oj_toolkit-0.2.2}/oj_toolkit/console/terminal.py

@@ -48,13 +48,18 @@ def detect_unicode_support() -> bool:
     """Detect if terminal supports Unicode characters.
 
     Checks environment variables and terminal type to determine if the terminal
-    can display Unicode characters (for pretty borders, etc.).
+    can display Unicode characters (for pretty borders, etc.), then verifies
+    sys.stdout's actual encoding can render them.
 
     Environment checks:
     - NO_COLOR: If set, disables colors/Unicode for accessibility
     - TERM: Terminal type (vt100, xterm, etc.)
     - CI: If in CI environment, assume Unicode support
-    - Platform: Windows console has limited Unicode, *nix terminals usually support it
+    - stdout encoding: Windows consoles commonly default to a codepage (cp1252,
+      cp437, etc.) that can't encode box-drawing characters even when the
+      platform/terminal is otherwise capable -- this is checked directly rather
+      than assumed from sys.platform, since PYTHONIOENCODING/chcp/Windows
+      Terminal all change what's actually possible independent of the OS.
 
     Returns:
         True if terminal likely supports Unicode, False otherwise.
@@ -78,20 +83,19 @@ def detect_unicode_support() -> bool:
     if term in ("dumb", "vt100", "vt220"):
         return False
 
-    # Windows console (cmd.exe, PowerShell) has limited Unicode support
-    # but modern versions and Windows Terminal support it
-    if sys.platform == "win32":
-        # Assume modern Windows (10+) or Windows Terminal supports Unicode
-        # Could be more conservative here if needed
+    # Verify stdout can actually encode a representative box-drawing character.
+    # Windows consoles commonly default to a codepage (cp1252, cp437, etc.) that
+    # can't, even when the platform/terminal is otherwise Unicode-capable.
+    try:
+        "─".encode(sys.stdout.encoding)
         return True
-
-    # *nix terminals typically support Unicode
-    if sys.platform in ("linux", "darwin"):
+    except UnicodeEncodeError:
+        return False
+    except (AttributeError, TypeError, LookupError):
+        # stdout has no usable encoding (e.g. a mock/capture stream in tests) --
+        # can't verify either way, so don't penalize it.
         return True
 
-    # Default: attempt Unicode
-    return True
-
 
 def select_style(
     style: str,

{oj_toolkit-0.2.0 → oj_toolkit-0.2.2}/oj_toolkit/parsing/consts.py

@@ -6,7 +6,10 @@ from typing import Callable
 
 # pylint: disable=invalid-name
 DEFAULT_CONVERTER: Callable = lambda value, *args, **kwargs: value
-DEFAULT_VALIDATOR: Callable = lambda value, expected_type, *args, **kwargs: isinstance(value, expected_type)
+DEFAULT_VALIDATOR: Callable = lambda value, expected_type, *args, **kwargs: (
+    expected_type is None
+    or (isinstance(expected_type, (type, tuple)) and isinstance(value, expected_type))
+)
 DEFAULT_SEPARATOR: str = ','
 
 

{oj_toolkit-0.2.0 → oj_toolkit-0.2.2}/oj_toolkit/parsing/types.py

@@ -104,12 +104,13 @@ def get_datetime(
     return result
 
 
-def validate(
+def validate(  # pylint: disable=too-many-arguments,too-many-positional-arguments
         v: Any,
         exp: Type[T] = None,
         default: T | None = None,
         converter: Callable = None,
         validator: Callable | None = DEFAULT_VALIDATOR,
+        pattern: str | re.Pattern | None = None,
         **kwargs
 ) -> T | None:
     """Validate and optionally convert a value with a custom converter and validator.
@@ -117,11 +118,13 @@ def validate(
     This is a generic validation utility that:
     1. Converts the value using a converter function (with automatic selection for common types)
     2. Validates the result using a validator function
-    3. Returns the result if valid, otherwise returns the default value
+    3. Optionally checks the result against a regex pattern (str results only)
+    4. Returns the result if valid, otherwise returns the default value
 
     Args:
         v: The value to validate.
-        exp: Expected type of the value. If None, no type conversion is attempted.
+        exp: Expected type of the value. If None, no type conversion or isinstance check is
+            attempted (DEFAULT_VALIDATOR passes any value through unchanged).
         default: The value to return if validation fails or v is None. Default: None.
         converter: Custom converter function. If None, one is selected based on exp:
             - exp=list: uses str_to_list
@@ -129,6 +132,9 @@ def validate(
             - otherwise: uses DEFAULT_CONVERTER (pass-through)
         validator: Custom validator function(result, exp, **kwargs) -> bool.
             If None, uses DEFAULT_VALIDATOR (isinstance check).
+        pattern: Optional regex (string or compiled re.Pattern) the result must fully match
+            via re.fullmatch. Only applied when the (converted, validated) result is a str;
+            checked after conversion/validation succeed. On mismatch, returns default.
         **kwargs: Additional arguments passed to converter and validator functions.
 
     Returns:
@@ -142,6 +148,8 @@ def validate(
         0
         >>> validate('a,b,c', exp=list)
         ['a', 'b', 'c']
+        >>> validate('AA:BB:CC:DD:EE:FF', exp=str, pattern=r'([0-9A-Fa-f]{2}:){5}[0-9A-Fa-f]{2}')
+        'AA:BB:CC:DD:EE:FF'
     """
     result: Any = v
     is_valid_result: bool = False
@@ -173,6 +181,11 @@ def validate(
         except Exception as exc_validation:
             logger.debug(f'Failed validation: {validator=}: {exc_validation=}', exc_info=True)
 
+    if is_valid_result and pattern is not None and isinstance(result, str):
+        compiled_pattern = pattern if isinstance(pattern, re.Pattern) else re.compile(pattern)
+        if not compiled_pattern.fullmatch(result):
+            is_valid_result = False
+
     return result if is_valid_result else default
 
 
@@ -192,7 +205,13 @@ _JMESPATH_TERMINAL_RE = re.compile(
 
 @lru_cache(maxsize=256)
 def _compile_jmespath(expr: str) -> Any:
-    """Compile (and cache) a jmespath expression string."""
+    """Compile (and cache) a jmespath expression string.
+
+    This cache is process-global and shared by every dig()/Digger call, keyed on the
+    expression string -- a bare dig() call already avoids recompiling a repeated path, not
+    just Digger. Capped at 256 distinct expressions (LRU-evicted); Digger's up-front compile
+    call at construction populates this same cache rather than maintaining its own.
+    """
     return jmespath.compile(expr)
 
 
@@ -337,10 +356,13 @@ def dig(
     else:
         result = _dig_expr(src, _path_to_expr(path), pop)
 
-    # Apply post-processor if result was found (not None) or if no result
-    if callable(post_processor) and result is not None:
+    # Always run the post-processor -- even on a missing/None result -- so it can apply
+    # its own default handling (validate() already treats v=None as failing an isinstance
+    # check and falls back to kwargs['default']). Skipping post_processor entirely is
+    # reserved for the explicit post_processor=None "give me the raw value" case.
+    if callable(post_processor):
         return post_processor(result, **kwargs)
-    return result  # return found value without post-processing
+    return result
 
 
 def dig_many(
@@ -382,7 +404,8 @@ class Digger:
 
     Validates/compiles the path up front (failing fast on a bad jmespath expression
     instead of on first use) and avoids re-passing the same path/exp/post_processor/kwargs
-    on every call.
+    on every call. A string `pattern=` kwarg is likewise pre-compiled to a re.Pattern here
+    so repeated calls don't re-compile the regex on every invocation.
 
     Example:
         >>> get_name = Digger(path='user.name', exp=str, default='')
@@ -399,6 +422,8 @@ class Digger:
         self.path = path
         self.pop = pop
         self.post_processor = post_processor
+        if isinstance(kwargs.get('pattern'), str):
+            kwargs['pattern'] = re.compile(kwargs['pattern'])
         self.kwargs = kwargs
         for candidate in (path if isinstance(path, (list, tuple)) else [path]):
             if candidate is not None:

{oj_toolkit-0.2.0 → oj_toolkit-0.2.2}/oj_toolkit.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: oj-toolkit
-Version: 0.2.0
+Version: 0.2.2
 Summary: Centralized shared utilities for ownjoo-org projects: standardized logging config, type validation, data parsing, console output, and async utilities
 Author-email: Speedy <speedy@ownjoo.org>
 License: MIT
@@ -33,7 +33,7 @@ Requires-Dist: mypy>=1.0; extra == "dev"
 Requires-Dist: pylint>=2.0; extra == "dev"
 Dynamic: license-file
 
-# ownjoo-toolkit
+# oj-toolkit
 
 Centralized shared utilities library for all ownjoo-org projects. Provides battle-tested functions for type validation, data parsing, and progress logging.
 
@@ -187,11 +187,19 @@ fields = dig_many(
 # Bind a path once, reuse it against many objects
 get_name = Digger(path='name', exp=str, default='')
 [get_name(user) for user in data['users']]  # Returns: ['Alice', 'Bob']
+
+# Constrain a string result with a regex (checked via re.fullmatch after conversion)
+mac = dig({'device': {'mac': 'AA:BB:CC:DD:EE:FF'}}, path='device.mac', exp=str,
+          pattern=r'([0-9A-Fa-f]{2}:){5}[0-9A-Fa-f]{2}')
+# Returns: 'AA:BB:CC:DD:EE:FF'
 ```
 
-> **Note:** `dig()`'s default `post_processor` (`validate()`) only accepts a value if you pass
-> `exp=<type>` -- without it, validation always fails and you get `default` back (`None` unless set).
-> Pass `exp=` for a real value, or `post_processor=None` to skip validation and get the raw match.
+> **Note:** if the path resolves to nothing (missing key/index, or `src` doesn't contain it),
+> `dig()`'s default `post_processor` (`validate()`) still runs and returns `default` (`None`
+> unless set) -- a missing path never silently drops `default` in favor of a bare `None`.
+> Passing no `exp=` means "accept any type, no isinstance check" -- not "always fail". Pass
+> `exp=<type>` to add a type check, or `post_processor=None` to bypass validation entirely and
+> get the raw match untouched.
 
 ### Logging Setup
 
@@ -711,18 +719,25 @@ else:
 
 Returns `True` if the terminal likely supports Unicode characters. Used internally to choose between ASCII and Unicode box/table borders.
 
+Checks `NO_COLOR`/`CI`/`TERM` env vars first, then verifies `sys.stdout.encoding` can actually
+encode a box-drawing character -- this catches Windows consoles that default to a codepage
+(`cp1252`, `cp437`, etc.) that can't, even though the platform/terminal is otherwise
+Unicode-capable. Set `PYTHONIOENCODING=utf-8` (or use Windows Terminal with a UTF-8 codepage)
+to get Unicode borders on Windows instead of the ASCII fallback.
+
 ### `parsing` Module
 
-#### `validate(v, exp=None, default=None, converter=None, validator=None, **kwargs)`
+#### `validate(v, exp=None, default=None, converter=None, validator=None, pattern=None, **kwargs)`
 
 Generic validation utility that converts and validates a value.
 
 - **Parameters:**
   - `v` (Any): The value to validate
-  - `exp` (Type): Expected type. Enables automatic converter selection (list → str_to_list, datetime → get_datetime)
-  - `default` (Any): Return this if validation fails. Default: None
-  - `converter` (Callable): Custom converter function. Default: auto-selected based on exp
-  - `validator` (Callable): Custom validator function(result, exp, **kwargs) → bool. Default: isinstance check
+  - `exp` (Type): Expected type. Enables automatic converter selection (list → str_to_list, datetime → get_datetime). If `None`, no isinstance check is performed -- any type passes (see `validator` below)
+  - `default` (Any): Return this if validation fails, or `v` is missing/`None`. Default: None
+  - `converter` (Callable): Custom converter function. Default: auto-selected based on exp. Auto-selection is limited to `list`/`datetime` -- `bool`/`int`/`float`/etc. are pass-through unless you supply your own `converter=`
+  - `validator` (Callable): Custom validator function(result, exp, **kwargs) → bool. Default: isinstance check (`exp=None` always passes; a non-type `exp` is rejected rather than raising)
+  - `pattern` (str | re.Pattern): Optional regex checked via `re.fullmatch` against the result, but only when the result is a `str` -- ignored for other types. Applied after conversion/validation succeed; a mismatch falls back to `default`
   - `**kwargs`: Passed to converter and validator
 
 - **Returns:** Converted and validated value, or default if validation fails
@@ -734,6 +749,16 @@ Generic validation utility that converts and validates a value.
 numbers = validate('1,2,3', exp=list)  # Returns: ['1', '2', '3']
 dt = validate('2024-01-15T10:30:00', exp=datetime)  # Returns: datetime object
 
+# exp=bool/int/float do NOT auto-coerce strings -- pass converter= explicitly for that
+validate('42', exp=int)                    # Returns: None ('42' is a str, not an int)
+validate('42', exp=int, converter=int)     # Returns: 42
+
+# Regex-constrain a string result
+validate('AA:BB:CC:DD:EE:FF', exp=str, pattern=r'([0-9A-Fa-f]{2}:){5}[0-9A-Fa-f]{2}')
+# Returns: 'AA:BB:CC:DD:EE:FF'
+validate('not-a-mac', exp=str, pattern=r'([0-9A-Fa-f]{2}:){5}[0-9A-Fa-f]{2}', default='invalid')
+# Returns: 'invalid'
+
 # Custom converter and validator
 def to_upper(v, *args, **kwargs):
     return str(v).upper()
@@ -795,6 +820,12 @@ get_datetime('01/15/2024 10:30:00', format_str='%m/%d/%Y %H:%M:%S')  # Custom
 Extract and post-process a value from a nested data structure, using [jmespath](https://jmespath.org/)
 under the hood (`jmespath` is a required dependency).
 
+> **Compilation caching:** compiled jmespath expressions are cached in a process-global
+> `lru_cache(maxsize=256)`, keyed by the expression string. A bare `dig()` call already avoids
+> recompiling a repeated path -- this isn't a `Digger`-only optimization. The cache holds up to
+> 256 distinct expressions across the whole process (shared by every `dig()`/`Digger` caller),
+> LRU-evicted beyond that.
+
 - **Parameters:**
   - `src` (Mapping | Sequence): Data structure to navigate
   - `path` (int | str | list | None):
@@ -811,16 +842,18 @@ under the hood (`jmespath` is a required dependency).
     Pass `None` to skip post-processing and get the raw match.
   - `**kwargs`: Passed to `post_processor`
 
-- **Returns:** Post-processed value, or `None` if extraction fails, no candidate path matches, or
-  no post-processor is specified and nothing was found
+- **Returns:** Post-processed value. If extraction fails or no candidate path matches, the
+  post-processor (default `validate()`) still runs and returns `default` (`None` unless set) --
+  unless `post_processor=None`, in which case the raw (possibly missing/`None`) value is returned
+  untouched with no default substitution.
 
 > **Type checkers:** pass `exp=<type>` to get a precise `<type> | None` return type instead of `Any`
 > (`dig(data, path='...', exp=str)` type-checks as `str | None`). Passing `post_processor=None` types
 > as `Any` (raw, unvalidated value). A custom `post_processor` types as that callable's own return type.
 
-> **Passing no `exp`:** the default `validate()` post-processor only accepts a value when you give
-> it `exp=<type>` -- without one, validation always fails and you get `default` back (`None` unless
-> set). Either pass `exp=`, or `post_processor=None` to bypass validation entirely.
+> **Passing no `exp`:** the default `validate()` post-processor treats a missing `exp` as "accept
+> any type" -- it does not fail. Pass `exp=<type>` to add an isinstance check, or
+> `post_processor=None` to bypass validation entirely and get the raw match.
 
 **Example:**
 
@@ -855,6 +888,39 @@ name = dig(data, path=['response.users[0].display_name', 'response.users[0].name
 # Returns: 'Alice'
 ```
 
+##### `dig()` Behavior Matrix
+
+Every row below is a real `dig()` call, backed by a passing unit test in
+`test/unit/parsing/test_parsing.py` -- the test name is the proof, not just a claim. This isn't
+every possible combination of kwargs (that cross product is huge and mostly redundant); it's one
+representative call per distinct decision point in `dig()`/`validate()`. `*(unset)*` means the
+kwarg wasn't passed (the function's own default applies); `--` means it isn't relevant to that row.
+
+| `src` | `path` | `exp` | `default` | `post_processor` | `pattern` | `validator` | `pop` | Result | Proven by |
+|---|---|---|---|---|---|---|---|---|---|
+| `{'first': 'a', 'second': ['blah']}` | `'second[0]'` (found) | `str` | *(unset)* | *(unset)* | *(unset)* | *(unset)* | `False` | `'blah'` | `test_should_get_value_from_dict` |
+| `['', ['blah']]` | `'[1][0]'` (found) | `str` | `''` | *(unset)* | *(unset)* | *(unset)* | `False` | `'blah'` (default unused -- value was valid) | `test_should_get_value_from_list` |
+| `{'a': {'b': 'value'}}` | `'x.y.z'` (missing) | `str` | `'default'` | *(unset)* | *(unset)* | *(unset)* | `False` | `'default'` | `test_should_return_default_on_missing_intermediate_key` |
+| `{'items': ['a', 'b', 'c']}` | `'items.invalid_index'` (missing) | `str` | `'default'` | *(unset)* | *(unset)* | *(unset)* | `False` | `'default'` | `test_should_return_default_when_accessing_list_with_string_key` |
+| `{'a': {'b': 'value'}}` | `'x.y.z'` (missing) | `str` | *(unset)* | *(unset)* | *(unset)* | *(unset)* | `False` | `None` (no default set) | `test_should_return_none_for_missing_path_with_exp_and_no_default` |
+| `{'a': {'b': 'c'}}` | `'x.y.z'` (missing) | *(unset)* | *(unset)* | *(unset)* | *(unset)* | *(unset)* | `False` | `None` (`exp=None` means "anything passes"; the found value already is `None`) | `test_should_handle_get_value_with_invalid_path` |
+| `{'key': 'blah'}` | `'key'` (found) | *(unset)* | *(unset)* | *(unset)* | *(unset)* | *(unset)* | `False` | `'blah'` (raw pass-through, no isinstance check) | `test_should_not_crash_when_exp_omitted_from_dig` |
+| `{'a': 'not-an-int'}` | `'a'` (found, wrong type) | `int` | *(unset)* | *(unset)* | *(unset)* | *(unset)* | `False` | `None` | `test_should_return_none_for_dig_result_type_mismatch` |
+| `{'a': 'not-an-int'}` | `'a'` (found, wrong type) | `int` | `-1` | *(unset)* | *(unset)* | *(unset)* | `False` | `-1` | `test_should_return_default_for_dig_result_type_mismatch` |
+| `{'a': 'blah'}` | `'a'` (found, non-type `exp`) | `'str'` | `'fallback'` | *(unset)* | *(unset)* | *(unset)* | `False` | `'fallback'` (garbage `exp` rejected, not raised) | `test_should_return_default_for_dig_with_non_type_expected_type` |
+| `{'key': 'raw_value'}` | `'key'` (found) | -- | -- | `None` | -- | -- | `False` | raw value, unvalidated | `test_should_return_raw_value_when_no_post_processor` |
+| `{'a': 1}` | `'missing'` (missing) | -- | `'default'` (ignored) | `None` | -- | -- | `False` | `None` (`default` is never consulted -- it's `validate()`'s kwarg, and `validate()` never runs) | `test_should_return_none_for_missing_path_with_no_post_processor` |
+| `{'response': {'users': [{'name': 'Alice'}, {'name': 'Bob'}]}}` | `'response.users'` (found) | -- | -- | `len` | -- | -- | `False` | `2` (custom post_processor gets the raw match) | `test_should_use_custom_post_processor_function` |
+| `'blah'` | `None` (src is the value) | `str` | *(unset)* | *(unset)* | *(unset)* | custom lambda | `False` | `'blah'` (custom validator overrides isinstance check) | `test_should_get_value_with_passed_validator` |
+| `'test_string'` | `None` (src is the value) | `str` | *(unset)* | *(unset)* | *(unset)* | *(unset)* | `False` | the src value itself | `test_should_post_process_source_when_path_is_none` |
+| `{'device': {'mac': 'AA:BB:CC:DD:EE:FF'}}` | `'device.mac'` (found) | `str` | *(unset)* | *(unset)* | matching regex | *(unset)* | `False` | matched MAC string | `test_should_validate_dig_result_with_matching_pattern` |
+| `{'device': {'mac': 'not-a-mac'}}` | `'device.mac'` (found) | `str` | `'invalid'` | *(unset)* | non-matching regex | *(unset)* | `False` | `'invalid'` | `test_should_return_default_for_dig_result_with_non_matching_pattern` |
+| `{'a': 1, 'b': 2}` | `'a'` (found) | `int` | *(unset)* | *(unset)* | -- | -- | `True` | `1`, and `'a'` removed from `src` | `test_should_pop_key_from_dict` |
+| `{'items': [{'id': 1}, {'id': 2}]}` | `'items[?id==\`2\`].id'` (found, ambiguous) | `list` | *(unset)* | *(unset)* | -- | -- | `True` (refused) | `[2]`, `src` left unmutated (pop refused + warning logged) | `test_should_refuse_pop_with_ambiguous_jmespath_filter` |
+| `{'user': {'name': 'Alice'}}` | `['user.nickname', 'user.name']` (fallback) | `str` | *(unset)* | *(unset)* | -- | -- | `False` | `'Alice'` (first candidate wins) | `test_should_use_first_matching_path_in_fallback_list` |
+| `{'user': {'name': 'Alice'}}` | `['user.nickname', 'user.alias']` (fallback, none match) | `str` | *(unset)* | *(unset)* | -- | -- | `False` | `None` | `test_should_return_none_when_no_fallback_path_matches` |
+| `{'user': {'name': 'Alice', 'nickname': 'Ali'}}` | `['user.nickname', 'user.name']` (fallback) | `str` | *(unset)* | *(unset)* | -- | -- | `True` | `'Ali'`, only the winning candidate's key removed | `test_should_pop_only_the_winning_fallback_path` |
+
 #### `dig_many(src, paths, **common_kwargs)`
 
 Extract several named fields from `src` in one call.
@@ -889,7 +955,9 @@ fields = dig_many(
 
 A pre-bound, reusable `dig()` call. Validates/compiles the jmespath expression once at
 construction (failing fast on a bad expression instead of on first use), then can be called
-like a function against any number of `src` objects without repeating `path`/`exp`/kwargs.
+like a function against any number of `src` objects without repeating `path`/`exp`/kwargs. A
+string `pattern=` kwarg is likewise pre-compiled to a `re.Pattern` at construction time, so
+repeated calls don't re-compile the regex.
 
 **Example:**
 

{oj_toolkit-0.2.0 → oj_toolkit-0.2.2}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "oj-toolkit"
-version = "0.2.0"
+version = "0.2.2"
 description = "Centralized shared utilities for ownjoo-org projects: standardized logging config, type validation, data parsing, console output, and async utilities"
 authors = [
     { name = "Speedy", email = "speedy@ownjoo.org" }