Update: styler & check_arg more readable names; __XXX__ custom stylin…

…g; file model for file handler logging; New: color code string list

setup.py

@@ -1,6 +1,6 @@
 from setuptools import setup, find_packages
 
-VERSION = '0.42.12'
+VERSION = '0.42.13'
 DESCRIPTION = 'Machine Learning project startup utilities'
 LONG_DESCRIPTION = 'My commonly used utilities for machine learning projects'
 
@@ -13,7 +13,7 @@
     description=DESCRIPTION,
     long_description=LONG_DESCRIPTION,
     url='https://github.com/StefanHeng/stef-util',
-    download_url='https://github.com/StefanHeng/stef-util/archive/refs/tags/v0.42.12.tar.gz',
+    download_url='https://github.com/StefanHeng/stef-util/archive/refs/tags/v0.42.13.tar.gz',
     packages=find_packages(),
     include_package_data=True,
     install_requires=[

stefutil/concurrency.py

@@ -14,7 +14,7 @@
 from tqdm.contrib import concurrent as tqdm_concurrent
 
 from stefutil.container import group_n
-from stefutil.prettier import ca
+from stefutil.prettier import check_arg as ca
 
 
 __all__ = ['conc_map', 'batched_conc_map', 'conc_yield']
@@ -346,7 +346,7 @@ def _dummy_fn(xx: int = None):
 
 
 if __name__ == '__main__':
-    from stefutil.prettier import rich_progress, rcl
+    from stefutil.prettier import rich_progress, rich_console_log
 
     def try_concurrent_yield():
         # this will gather all results and return
@@ -366,9 +366,9 @@ def check_conc_map():
         n = 10
         # for res in conc_map(fn=_work, args=range(n), with_tqdm=dict(total=n), n_worker=4, mode='process'):
         #     sic(res)
-        from stefutil.prettier import rcl
+        from stefutil.prettier import rich_console_log
         for res in rich_progress(conc_map(fn=_work, args=range(n), mode='process'), total=n):
-            rcl(res)
+            rich_console_log(res)
     # check_conc_map()
 
     def check_conc_yield():
@@ -437,5 +437,5 @@ def check_conc_process_chunk():
             i, _ = i
             # logger.info(f'Process {s.i(i)} terminated')
             lst.append(i)
-        rcl(lst)
+        rich_console_log(lst)
     check_conc_process_chunk()

stefutil/container.py

@@ -7,7 +7,7 @@
 from functools import reduce
 from collections import OrderedDict
 
-from stefutil.prettier import s
+from stefutil.prettier import style as s
 from stefutil.packaging import _use_dl
 
 

stefutil/functional.py

@@ -7,7 +7,7 @@
 from os.path import join as os_join
 from typing import Callable, Union, Any
 
-from stefutil.prettier import ca, date
+from stefutil.prettier import check_arg as ca, date
 
 
 __all__ = ['profile_runtime', 'RecurseLimit']

stefutil/ml.py

@@ -6,7 +6,7 @@
 import math
 from typing import Tuple, Dict, Union, Iterable, Any
 
-from stefutil.prettier import fmt_num, fmt_sizeof, ca
+from stefutil.prettier import fmt_num, fmt_sizeof, check_arg as ca
 from stefutil.packaging import _use_ml, _use_dl
 
 

stefutil/nlp.py

@@ -5,7 +5,7 @@
 from tqdm import tqdm
 
 from stefutil.container import group_n
-from stefutil.prettier import get_logger, s, ca
+from stefutil.prettier import get_logger, style as s, check_arg as ca
 from stefutil.packaging import _use_dl
 
 

stefutil/plot.py

@@ -10,7 +10,7 @@
 from typing import List, Dict, Iterable, Callable, Any, Union
 from dataclasses import dataclass
 
-from stefutil.prettier import s, ca, get_logger, Timer
+from stefutil.prettier import style as s, check_arg as ca, get_logger, Timer
 from stefutil.container import df_col2cat_col
 from stefutil.packaging import installed_packages, _use_plot, _use_ml
 
@@ -37,9 +37,9 @@
 
 
     def set_plot_style():
-        plt.rc('figure', figsize=(16, 9))
-        plt.rc('figure.constrained_layout', use=True)
-        plt.rc('text.latex', preamble='\n'.join([
+        plt.rich_console('figure', figsize=(16, 9))
+        plt.rich_console('figure.constrained_layout', use=True)
+        plt.rich_console('text.latex', preamble='\n'.join([
             r'\usepackage{nicefrac}',
             r'\usepackage{helvet}',
             r'\usepackage{sansmath}',  # render math sans-serif

stefutil/prettier/prettier_debug.py

@@ -14,11 +14,11 @@
 
 
 __all__ = [
-    'MyIceCreamDebugger', 'sic',
-    'rc', 'rcl',
+    'MyIceCreamDebugger', 'icecream',
+    'rich_console', 'rich_console_log',
     '_DEFAULT_ANSI_BACKEND', '_ANSI_REST_ALL',
     'to_rich_markup',
-    'render_nested_ansi_pairs', 'PrettyStyler', 's', 'style',
+    'render_nested_ansi_pairs', 'PrettyStyler', 'style',
 ]
 
 
@@ -42,10 +42,10 @@ def output_width(self, value):
 
 
 # syntactic sugar
-sic = MyIceCreamDebugger()
+sic = icecream = MyIceCreamDebugger()
 
-rc = Console()
-rcl = rc.log
+rich_console = Console()
+rich_console_log = rich_console.log
 
 
 @dataclass
@@ -244,8 +244,8 @@ def __call__(
                 import rich.style
                 style_args['color'] = style_args.pop('fg')
                 style_args['bgcolor'] = style_args.pop('bg')
-                style = rich.style.Style(**style_args)
-                return style.render(text=str(x))  # explicitly convert to str for `False` and `None` styling
+                style_ = rich.style.Style(**style_args)
+                return style_.render(text=str(x))  # explicitly convert to str for `False` and `None` styling
 
             elif self.backend == 'rich-markup':
                 if kwargs != dict():
@@ -270,6 +270,7 @@ class PrettyStyler:
         False: dict(fg='Br', italic=True),
         int: dict(fg='Bc'),
         float: dict(fg='Bc'),
+        'keyword': dict(fg='Bm'),
         str: dict(fg='Bg'),
         'path': dict(fg='m')
     }
@@ -336,6 +337,9 @@ def _get_default_style(x: Union[int, float, bool, str, None]):
             elif isinstance(x, Path) \
                     or (isinstance(x, str) and (os.path.exists(x) or len(x) < 256 and x.count(os.sep) >= 2)):  # heuristics to check for path
                 tp = 'path'
+            # consider `__XXX__` a special keyword
+            elif isinstance(x, str) and x.startswith('__') and x.endswith('__'):
+                tp = 'keyword'
             else:
                 tp = type(x)
             ret = d.get(tp, dict())
@@ -925,3 +929,8 @@ def check_pad():
         }
         print(s.i(d, align_keys=1, indent=1, pad=4))
     # check_pad()
+
+    def check_style_keyword():
+        d = {'__correct__': 1, '__not_named_entity__': 2, '__wrong_boundary__': 3, '__wrong_type__': 4, 'incorrect ': 5}
+        print(s.i(d, color_keys=True))
+    check_style_keyword()

stefutil/prettier/prettier_dl.py

@@ -8,7 +8,8 @@
 
 from tqdm.auto import tqdm
 
-from stefutil.prettier import s, ca
+from stefutil.prettier.prettier_debug import style as s
+from stefutil.prettier.prettier_log import check_arg as ca
 from stefutil.packaging import _use_dl
 
 

stefutil/prettier/prettier_log.py

@@ -3,19 +3,19 @@
 import sys
 import logging
 import datetime
-from typing import List, Dict, Union, Optional, Any, Callable
+from typing import List, Dict, Union, Optional, Any, Callable, Iterable
 from collections import OrderedDict
 
 from stefutil.prettier.prettier import hex2rgb
-from stefutil.prettier.prettier_debug import s, _DEFAULT_ANSI_BACKEND, _ANSI_REST_ALL
+from stefutil.prettier.prettier_debug import style as s, _DEFAULT_ANSI_BACKEND, _ANSI_REST_ALL
 
 
 __all__ = [
     'MyTheme', 'MyFormatter',
     'filter_ansi', 'CleanAnsiFileHandler', 'AnsiFileMap',
     'LOG_STR2LOG_LEVEL', 'get_logging_handler', 'get_logger', 'add_log_handler', 'add_file_handler', 'drop_file_handler',
 
-    'CheckArg', 'ca',
+    'CheckArg', 'check_arg',
 
     'now', 'date',
 ]
@@ -74,6 +74,87 @@ def date():
     return now(for_path=True, fmt='short-date')
 
 
+def _color_code_string(prompt: str = None) -> str:
+    """
+    Color-code a prompt for semantic segmentation by a simple heuristic
+    """
+    # first, split up the prompt into sections
+    ret = ''
+    sep = '\n\n'
+    segs = prompt.split(sep)
+    it = iter(segs)
+    prev = None
+    curr = next(it)
+    assert curr is not None
+    cs = ['b', 'm', 'r', 'y', 'g']
+    i_c = None
+    while curr is not None:
+        # iteratively check, does the difference between current and next segment indicate a new section?
+        # if prev is not None:
+        # declare different is current segment is pretty long, via either char count or line count
+        long_enough = False
+        n_lines = curr.count('\n')
+        n_chars = len(curr)
+        if n_chars > 250:
+            long_enough = True
+        elif n_chars > 150 and n_lines > 0:
+            long_enough = True
+        elif n_lines > 0 and all(len(c) > 60 for c in curr.split('\n')):
+            long_enough = True
+        elif '\n' not in curr and n_chars > 120:
+            long_enough = True
+        elif n_lines > 3:
+            long_enough = True
+        elif '---' in curr or 'Examples:' in curr:
+            long_enough = True
+        if prev is None:
+            i_c = 0
+        elif long_enough:
+            i_c = (i_c + 1) % len(cs)
+        print(long_enough)
+        ret += f'{s.i(curr, fg=cs[i_c])}{sep}'
+
+        prev = curr
+        curr = next(it, None)
+    return ret
+
+
+def print_strings(strings: Union[Callable[[], str], List[str], Iterable[str]], n: int = None) -> List[str]:
+    """
+    color codes a list of strings with heuristics to separate semantic sections
+
+    :param strings: a list of strings or a callable that returns a string
+    :param n: number of strings to print
+    """
+    if isinstance(strings, list):
+        assert n is None
+        prompts = strings
+        n = len(prompts)
+    elif hasattr(strings, '__iter__'):
+        prompts = list(strings)
+        if n is not None:
+            n = min(n, len(prompts))
+    else:
+        n = n or 5
+        prompts = [strings() for _ in range(n)]
+
+        if any(not p for p in prompts):
+            prompts = [p for p in prompts if p]  # filter out empty/None prompts
+            n = len(prompts)
+            assert n > 0
+    assert all(isinstance(p, str) for p in prompts)  # sanity check
+
+    # prompts = [f'Prompt {i}:\n{pl.i(p)}' for i, p in enumerate(prompts, start=1)]
+    prompts = [f'String {i}:\n{_color_code_string(p)}' for i, p in enumerate(prompts, start=1)]
+    if n == 1:
+        print(prompts[0])
+    else:
+        for i in range(n):
+            sep = '\n\n\n' if i != n - 1 else ''
+            print(f'{prompts[i]}{sep}')
+    return prompts
+
+
 class MyTheme:
     """
     Theme based on `sty` and `Atom OneDark`
@@ -306,7 +387,7 @@ def append_ext(file_path: str) -> str:
 
 
 def get_logging_handler(
-        kind: str = 'stdout', file_path: str = None, level: Union[str, int] = 'debug',
+        kind: str = 'stdout', file_path: str = None, level: Union[str, int] = 'debug', file_mode: str = 'a',
         ansi_file_map: Callable[[str], str] = AnsiFileMap.append_ext
 ) -> Union[logging.Handler, List[logging.Handler]]:
     """
@@ -319,6 +400,7 @@ def get_logging_handler(
         If `file+ansi`, both file write handlers w/ and w/o ANSI style filtering
     :param file_path: File path for file logging.
     :param level: Logging level for the handler.
+    :param file_mode: File mode for file logging.
     :param ansi_file_map: Mapping function for the ANSI file handler:
         Returns the mapped file path for ANSI given the original file path.
     """
@@ -353,7 +435,7 @@ def get_logging_handler(
 
             # i.e., when `file-w/-ansi`, use the default file handler - no filter out for the ANSI chars
             cls = CleanAnsiFileHandler if kind == 'file' else logging.FileHandler
-            handler = cls(file_path)
+            handler = cls(file_path, mode=file_mode)
         set_level(handler, level=level)
         handler.setFormatter(MyFormatter(with_color=kind in ['stdout', 'file-w/-ansi']))
         handler.addFilter(HandlerFilter(handler_name=kind))
@@ -468,7 +550,7 @@ def cache_options(self, display_name: str, attr_name: str, options: List[str]):
         setattr(self, attr_name, options)
 
 
-ca = CheckArg()
+ca = check_arg = CheckArg()
 ca.cache_options(  # See `stefutil::plot.py`
     'Bar Plot Orientation', attr_name='bar_orient', options=['v', 'h', 'vertical', 'horizontal']
 )
@@ -573,3 +655,12 @@ def check_filter_ansi():
         print(txt)
         print(filter_ansi(txt))
     # check_filter_ansi()
+
+    def check_print_str():
+        lst = [
+            "a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string"
+            "\n\n\n\n\n\n\n\n\n a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string a long string\n\n\n\n\na long string\n\n\nasdsadasasd",
+            'hello world'
+        ]
+        print_strings(lst)
+    check_print_str()

stefutil/prettier/prettier_progress.py

@@ -11,7 +11,7 @@
 from tqdm.utils import FormatReplace, disp_len, disp_trim
 
 from stefutil.os import rel_path
-from stefutil.prettier.prettier_debug import s, rc
+from stefutil.prettier.prettier_debug import style as s, rich_console
 
 
 __all__ = [
@@ -21,7 +21,7 @@
 
 
 def rich_status(desc: str = None, spinner: str = 'arrow3'):
-    return rc.status(status=desc, spinner=spinner)
+    return rich_console.status(status=desc, spinner=spinner)
 
 
 _I = typing.TypeVar("_I", TextIO, BinaryIO)

stefutil/project.py

@@ -9,7 +9,7 @@
 
 from stefutil.container import get
 from stefutil.os import rel_path
-from stefutil.prettier import now, ca, get_logger, s
+from stefutil.prettier import now, check_arg as ca, get_logger, style as s
 
 
 __all__ = ['SConfig', 'PathUtil']