kaqing 2.0.171__py3-none-any.whl → 2.0.204__py3-none-any.whl

adam/utils.py

@@ -1,4 +1,4 @@
-from abc import abstractmethod
+from abc import ABC
 from concurrent.futures import Future, ThreadPoolExecutor
 from contextlib import redirect_stdout
 import copy
@@ -12,27 +12,58 @@ from pathlib import Path
 import random
 import string
 import threading
-from typing import Callable, Iterator, TypeVar
+import traceback
+from typing import Callable, Iterator, TypeVar, Union
 from dateutil import parser
 import subprocess
 import sys
 import time
 import click
 import yaml
+from prompt_toolkit.completion import Completer
 
 from . import __version__
 
+T = TypeVar('T')
+
 log_state = threading.local()
 
-class LogConfig:
-    is_debug = lambda: False
-    is_debug_timing = lambda: False
-    is_display_help = True
+class ConfigReadable:
+    def is_debug() -> bool:
+        pass
+
+    def get(self, key: str, default: T) -> T:
+        pass
+
+class ConfigHolder:
+    # the singleton pattern
+    def __new__(cls, *args, **kwargs):
+        if not hasattr(cls, 'instance'): cls.instance = super(ConfigHolder, cls).__new__(cls)
+
+        return cls.instance
+
+    def __init__(self):
+        if not hasattr(self, 'config'):
+            # set by Config
+            self.config: 'ConfigReadable' = None
+            # only for testing
+            self.is_display_help = True
+            # set by ReplSession
+            self.append_command_history = lambda entry: None
+
+NO_SORT = 0
+SORT = 1
+REVERSE_SORT = -1
+
+def tabulize(lines: list[T], fn: Callable[..., T] = None, header: str = None, dashed_line = False, separator = ' ', to: int = 1, sorted: int = NO_SORT):
+    if fn:
+        lines = list(map(fn, lines))
 
-def to_tabular(lines: str, header: str = None, dashed_line = False):
-    return lines_to_tabular(lines.split('\n'), header, dashed_line)
+    if sorted == SORT:
+        lines.sort()
+    elif sorted == REVERSE_SORT:
+        lines.sort(reverse=True)
 
-def lines_to_tabular(lines: list[str], header: str = None, dashed_line = False, separator = ' '):
     maxes = []
     nls = []
 
@@ -63,7 +94,14 @@ def lines_to_tabular(lines: list[str], header: str = None, dashed_line = False,
     for line in lines:
         format_line(line)
 
-    return '\n'.join(nls)
+    table = '\n'.join(nls)
+
+    if to == 1:
+        log(table)
+    elif to == 2:
+        log2(table)
+
+    return table
 
 def convert_seconds(total_seconds_float):
     total_seconds_int = int(total_seconds_float)  # Convert float to integer seconds
@@ -117,7 +155,7 @@ def duration(start_time: float, end_time: float = None):
     d = convert_seconds(end_time - start_time)
     t = []
     if d:
-        t.append(f'{d}h')
+        t.append(f'{d[0]}h')
     if t or d[1]:
         t.append(f'{d[1]}m')
     t.append(f'{d[2]}s')
@@ -142,7 +180,11 @@ def deep_merge_dicts(dict1, dict2):
             merged_dict[key] = deep_merge_dicts(merged_dict[key], value)
         elif key not in merged_dict or value:
             # Otherwise, overwrite or add the value from dict2
-            merged_dict[key] = value
+            if key in merged_dict and isinstance(merged_dict[key], Completer):
+                pass
+                # print('SEAN completer found, ignoring', key, value)
+            else:
+                merged_dict[key] = value
     return merged_dict
 
 def deep_sort_dict(d):
@@ -180,7 +222,7 @@ def get_deep_keys(d, current_path=""):
     return keys
 
 def display_help(replace_arg = False):
-    if not LogConfig.is_display_help:
+    if not ConfigHolder().is_display_help:
         return
 
     args = copy.copy(sys.argv)
@@ -233,7 +275,7 @@ def json_to_csv(json_data: list[dict[any, any]], delimiter: str = ','):
         return None
 
 def log_to_file(config: dict[any, any]):
-    try:
+    with log_exc():
         base = f"/kaqing/logs"
         os.makedirs(base, exist_ok=True)
 
@@ -248,8 +290,6 @@ def log_to_file(config: dict[any, any]):
                     f.write(config)
             else:
                     f.write(config)
-    except:
-        pass
 
 def copy_config_file(rel_path: str, module: str, suffix: str = '.yaml', show_out = True):
     dir = f'{Path.home()}/.kaqing'
@@ -271,9 +311,31 @@ def is_lambda(func):
     return callable(func) and hasattr(func, '__name__') and func.__name__ == '<lambda>'
 
 def debug(s = None):
-    if LogConfig.is_debug():
+    if ConfigHolder().config.is_debug():
         log2(f'DEBUG {s}')
 
+def debug_complete(s = None):
+    CommandLog.log(f'DEBUG {s}', config=ConfigHolder().config.get('debugs.complete', 'off'))
+
+def debug_trace():
+    if ConfigHolder().config.is_debug():
+    # if LogConfig.is_debug():
+        log2(traceback.format_exc())
+
+def in_docker() -> bool:
+    if os.path.exists('/.dockerenv'):
+        return True
+
+    try:
+        with open('/proc/1/cgroup', 'rt') as f:
+            for line in f:
+                if 'docker' in line or 'lxc' in line:
+                    return True
+    except FileNotFoundError:
+        pass
+
+    return False
+
 class Ing:
     def __init__(self, msg: str, suppress_log=False):
         self.msg = msg
@@ -285,7 +347,7 @@ class Ing:
 
         try:
             if not log_state.ing_cnt:
-                if not self.suppress_log and not LogConfig.is_debug():
+                if not self.suppress_log and not ConfigHolder().config.is_debug():
                     log2(f'{self.msg}...', nl=False)
 
             return None
@@ -295,7 +357,7 @@ class Ing:
     def __exit__(self, exc_type, exc_val, exc_tb):
         log_state.ing_cnt -= 1
         if not log_state.ing_cnt:
-            if not self.suppress_log and not LogConfig.is_debug():
+            if not self.suppress_log and not ConfigHolder().config.is_debug():
                 log2(' OK')
 
         return False
@@ -316,7 +378,7 @@ def ing(msg: str, body: Callable[[], None]=None, suppress_log=False):
     return r
 
 def loggable():
-    return LogConfig.is_debug() or not hasattr(log_state, 'ing_cnt') or not log_state.ing_cnt
+    return ConfigHolder().config and ConfigHolder().config.is_debug() or not hasattr(log_state, 'ing_cnt') or not log_state.ing_cnt
 
 class TimingNode:
     def __init__(self, depth: int, s0: time.time = time.time(), line: str = None):
@@ -344,7 +406,7 @@ class LogTiming:
         self.s0 = s0
 
     def __enter__(self):
-        if not LogConfig.is_debug_timing():
+        if (config := ConfigHolder().config.get('debugs.timings', 'off')) not in ['on', 'file']:
             return
 
         if not hasattr(log_state, 'timings'):
@@ -356,7 +418,7 @@ class LogTiming:
             self.s0 = time.time()
 
     def __exit__(self, exc_type, exc_val, exc_tb):
-        if not LogConfig.is_debug_timing():
+        if (config := ConfigHolder().config.get('debugs.timings', 'off')) not in ['on', 'file']:
             return False
 
         child = log_state.timings
@@ -367,7 +429,9 @@ class LogTiming:
         log_state.timings = self.me
 
         if not self.me.depth:
-            log2(self.me.tree())
+            # log timings finally
+            CommandLog.log(self.me.tree(), config)
+
             log_state.timings = TimingNode(0)
 
         return False
@@ -376,7 +440,7 @@ def log_timing(msg: str, body: Callable[[], None]=None, s0: time.time = None):
     if not s0 and not body:
         return LogTiming(msg, s0=s0)
 
-    if not LogConfig.is_debug_timing():
+    if not ConfigHolder().config.get('debugs.timings', False):
         if body:
             return body()
 
@@ -396,7 +460,9 @@ def log_timing(msg: str, body: Callable[[], None]=None, s0: time.time = None):
 
 def timing_log_line(depth: int, msg: str, s0: time.time):
     elapsed = time.time() - s0
-    prefix = '[timings] '
+    offloaded = '-' if threading.current_thread().name.startswith('offload') or threading.current_thread().name.startswith('async') else '+'
+    prefix = f'[{offloaded} timings] '
+
     if depth:
         if elapsed > 0.01:
             prefix = ('  ' * (depth-1)) + '* '
@@ -416,7 +482,90 @@ def wait_log(msg: str):
 def clear_wait_log_flag():
     WaitLog.wait_log_flag = False
 
-T = TypeVar('T')
+def bytes_generator_from_file(file_path, chunk_size=4096):
+    with open(file_path, 'rb') as f:
+        while True:
+            chunk = f.read(chunk_size)
+            if not chunk:
+                break
+            yield chunk
+
+class GeneratorStream(io.RawIOBase):
+    def __init__(self, generator):
+        self._generator = generator
+        self._buffer = b''  # Buffer to store leftover bytes from generator yields
+
+    def readable(self):
+        return True
+
+    def _read_from_generator(self):
+        try:
+            chunk = next(self._generator)
+            if isinstance(chunk, str):
+                chunk = chunk.encode('utf-8')  # Encode if generator yields strings
+            self._buffer += chunk
+        except StopIteration:
+            pass  # Generator exhausted
+
+    def readinto(self, b):
+        # Fill the buffer if necessary
+        while len(self._buffer) < len(b):
+            old_buffer_len = len(self._buffer)
+            self._read_from_generator()
+            if len(self._buffer) == old_buffer_len:  # Generator exhausted and buffer empty
+                break
+
+        bytes_to_read = min(len(b), len(self._buffer))
+        b[:bytes_to_read] = self._buffer[:bytes_to_read]
+        self._buffer = self._buffer[bytes_to_read:]
+        return bytes_to_read
+
+    def read(self, size=-1):
+        if size == -1:  # Read all remaining data
+            while True:
+                old_buffer_len = len(self._buffer)
+                self._read_from_generator()
+                if len(self._buffer) == old_buffer_len:
+                    break
+            data = self._buffer
+            self._buffer = b''
+            return data
+        else:
+            # Ensure enough data in buffer
+            while len(self._buffer) < size:
+                old_buffer_len = len(self._buffer)
+                self._read_from_generator()
+                if len(self._buffer) == old_buffer_len:
+                    break
+
+            data = self._buffer[:size]
+            self._buffer = self._buffer[size:]
+            return data
+
+class LogTrace:
+    def __init__(self, err_msg: Union[str, callable, bool] = None):
+        self.err_msg = err_msg
+
+    def __enter__(self):
+        return None
+
+    def __exit__(self, exc_type, exc_val, exc_tb):
+        if exc_type is not None:
+            if self.err_msg is True:
+                log2(str(exc_val))
+            elif callable(self.err_msg):
+                log2(self.err_msg(exc_val))
+            elif self.err_msg is not False and self.err_msg:
+                log2(self.err_msg)
+
+            if self.err_msg is not False and ConfigHolder().config.is_debug():
+                traceback.print_exception(exc_type, exc_val, exc_tb, file=sys.stderr)
+
+        # swallow exception
+        return True
+
+def log_exc(err_msg: Union[str, callable, bool] = None):
+    return LogTrace(err_msg=err_msg)
 
 class ParallelService:
     def __init__(self, handler: 'ParallelMapHandler'):
@@ -450,37 +599,44 @@ class ParallelService:
         else:
             return iterator
 
+thread_pools: dict[str, ThreadPoolExecutor] = {}
+thread_pool_lock = threading.Lock()
+
 class ParallelMapHandler:
-    def __init__(self, collection: list, max_workers: int, samples: int = sys.maxsize, msg: str = None, collect = True):
+    def __init__(self, collection: list, workers: int, samples: int = sys.maxsize, msg: str = None, collect = True, name = None):
         self.collection = collection
-        self.max_workers = max_workers
+        self.workers = workers
         self.executor = None
         self.samples = samples
         self.msg = msg
         if msg and msg.startswith('d`'):
-            if LogConfig.is_debug():
+            if ConfigHolder().config.is_debug():
                 self.msg = msg.replace('d`', '', 1)
             else:
                 self.msg = None
         self.collect = collect
+        self.name = name
 
         self.begin = []
         self.end = []
         self.start_time = None
 
     def __enter__(self):
+        self.start_time = None
+
         self.calc_msgs()
 
-        if self.max_workers > 1 and (not self.size() or self.size()) and self.samples == sys.maxsize:
+        if self.workers > 1 and (not self.size() or self.size()) and self.samples == sys.maxsize:
             self.start_time = time.time()
 
-            self.executor = ThreadPoolExecutor(max_workers=self.max_workers)
+            self.executor = self.pool()
+            # self.executor = ThreadPoolExecutor(max_workers=self.workers)
             self.executor.__enter__()
 
         return ParallelService(self)
 
     def __exit__(self, exc_type, exc_val, exc_tb):
-        if self.executor:
+        if not self.name and self.executor:
             self.executor.__exit__(exc_type, exc_val, exc_tb)
 
         if self.end:
@@ -488,6 +644,15 @@ class ParallelMapHandler:
 
         return False
 
+    def pool(self, thread_name_prefix: str = None):
+        if not self.name:
+            return ThreadPoolExecutor(max_workers=self.workers)
+
+        if self.name not in thread_pools:
+            thread_pools[self.name] = ThreadPoolExecutor(max_workers=self.workers, thread_name_prefix=thread_name_prefix)
+
+        return thread_pools[self.name]
+
     def size(self):
         if not self.collection:
             return 0
@@ -498,28 +663,28 @@ class ParallelMapHandler:
         if not self.msg:
             return
 
+        self.begin = []
+        self.end = []
         size = self.size()
-            # return
-
         offloaded = False
         serially = False
         sampling = False
         if size == 0:
             offloaded = True
-            self.msg = self.msg.replace('{size}', '1')
-        elif self.max_workers > 1 and size > 1 and self.samples == sys.maxsize:
-            self.msg = self.msg.replace('{size}', f'{size}')
+            msg = self.msg.replace('{size}', '1')
+        elif self.workers > 1 and size > 1 and self.samples == sys.maxsize:
+            msg = self.msg.replace('{size}', f'{size}')
         elif self.samples < sys.maxsize:
             sampling = True
+            samples = self.samples
             if self.samples > size:
-                self.samples = size
-            self.msg = self.msg.replace('{size}', f'{self.samples}/{size} sample')
+                samples = size
+            msg = self.msg.replace('{size}', f'{samples}/{size} sample')
         else:
             serially = True
-            self.msg = self.msg.replace('{size}', f'{size}')
-            # return
+            msg = self.msg.replace('{size}', f'{size}')
 
-        for token in self.msg.split(' '):
+        for token in msg.split(' '):
             if '|' in token:
                 self.begin.append(token.split('|')[0])
                 if not sampling and not serially and not offloaded:
@@ -534,7 +699,21 @@ class ParallelMapHandler:
         elif sampling or serially:
             log2(f'{" ".join(self.begin)} serially...')
         else:
-            log2(f'{" ".join(self.begin)} with {self.max_workers} workers...')
+            log2(f'{" ".join(self.begin)} with {self.workers} workers...')
+
+# parallelizers: dict[str, ParallelMapHandler] = {}
+# parallelizer_lock = threading.Lock()
+
+def parallelize(collection: list, workers: int = 0, samples = sys.maxsize, msg: str = None, collect = True, name = None):
+    return ParallelMapHandler(collection, workers, samples = samples, msg = msg, collect = collect, name = name)
+    # if not name:
+    #     return ParallelMapHandler(collection, workers, samples = samples, msg = msg, collect = collect)
+
+    # with parallelizer_lock:
+    #     if name not in parallelizers:
+    #         parallelizers[name] = ParallelMapHandler(collection, workers, samples = samples, msg = msg, collect = collect, name = name)
+
+    # return parallelizers[name]
 
 class OffloadService:
     def __init__(self, handler: 'OffloadHandler'):
@@ -553,22 +732,24 @@ class OffloadService:
             return future
 
 class OffloadHandler(ParallelMapHandler):
-    def __init__(self, max_workers: int, msg: str = None):
-        super().__init__(None, max_workers, msg=msg, collect=False )
+    def __init__(self, max_workers: int, msg: str = None, name: str = None):
+        super().__init__(None, max_workers, msg=msg, collect=False, name=f'offload-{name}')
 
     def __enter__(self):
+        self.start_time = None
         self.calc_msgs()
 
-        if self.max_workers > 1 and (not self.size() or self.size()) and self.samples == sys.maxsize:
+        if self.workers > 1:
             self.start_time = time.time()
 
-            self.executor = ThreadPoolExecutor(max_workers=self.max_workers)
+            self.executor = self.pool(thread_name_prefix='offload')
+            # self.executor = ThreadPoolExecutor(max_workers=self.workers, thread_name_prefix='offload')
             self.executor.__enter__()
 
         return OffloadService(self)
 
     def __exit__(self, exc_type, exc_val, exc_tb):
-        if self.executor:
+        if not self.name and self.executor:
             self.executor.__exit__(exc_type, exc_val, exc_tb)
 
         if self.end:
@@ -576,38 +757,33 @@ class OffloadHandler(ParallelMapHandler):
 
         return False
 
-    def size(self):
-        if not self.collection:
-            return 0
-
-        return len(self.collection)
-
     def calc_msgs(self):
         if not self.msg:
             return
 
+        self.begin = []
+        self.end = []
         size = self.size()
-            # return
 
         offloaded = False
         serially = False
         sampling = False
         if size == 0:
             offloaded = True
-            self.msg = self.msg.replace('{size}', '1')
-        elif self.max_workers > 1 and size > 1 and self.samples == sys.maxsize:
-            self.msg = self.msg.replace('{size}', f'{size}')
+            msg = self.msg.replace('{size}', '1')
+        elif self.workers > 1 and size > 1 and self.samples == sys.maxsize:
+            msg = self.msg.replace('{size}', f'{size}')
         elif self.samples < sys.maxsize:
             sampling = True
-            if self.samples > size:
-                self.samples = size
-            self.msg = self.msg.replace('{size}', f'{self.samples}/{size} sample')
+            samples = self.samples
+            if samples > size:
+                samples = size
+            msg = self.msg.replace('{size}', f'{samples}/{size} sample')
         else:
             serially = True
-            self.msg = self.msg.replace('{size}', f'{size}')
-            # return
+            msg = self.msg.replace('{size}', f'{size}')
 
-        for token in self.msg.split(' '):
+        for token in msg.split(' '):
             if '|' in token:
                 self.begin.append(token.split('|')[0])
                 if not sampling and not serially and not offloaded:
@@ -622,58 +798,76 @@ class OffloadHandler(ParallelMapHandler):
         elif sampling or serially:
             log2(f'{" ".join(self.begin)} serially...')
         else:
-            log2(f'{" ".join(self.begin)} with {self.max_workers} workers...')
+            log2(f'{" ".join(self.begin)} with {self.workers} workers...')
 
-# class ParallelMapHandler(ParallelHandler):
-#     def __enter__(self):
-#         self.calc_msgs()
+# offloaders: dict[str, OffloadHandler] = {}
+# offloaders_lock = threading.Lock()
 
-#         if self.max_workers > 1 and (not self.size() or self.size()) and self.samples == sys.maxsize:
-#             self.start_time = time.time()
+def offload(max_workers: int = 3, msg: str = None, name: str = None):
+    return OffloadHandler(max_workers, msg = msg, name = name)
+    # if not name:
+    #     return OffloadHandler(max_workers, msg = msg)
 
-#             self.executor = ThreadPoolExecutor(max_workers=self.max_workers)
-#             self.executor.__enter__()
+    # with offloaders_lock:
+    #     if name not in offloaders:
+    #         offloaders[name] = OffloadHandler(max_workers, msg = msg, name = name)
 
-#         if self.collection:
-#             return self.map
-#         else:
-#             return self.submit
+    # return offloaders[name]
 
-#     def map(self, fn: Callable[..., T]) -> Iterator[T]:
-#         iterator = None
-#         if self.executor:
-#             iterator = self.executor.map(fn, self.collection)
-#         elif self.samples < sys.maxsize:
-#             samples = []
+def kaqing_log_file_name(suffix = 'log'):
+    return f"{ConfigHolder().config.get('log-prefix', '/tmp/qing')}-{datetime.now().strftime('%d%H%M%S')}.{suffix}"
 
-#             for elem in self.collection:
-#                 if not self.samples:
-#                     break
+class LogFileHandler:
+    def __init__(self, suffix = 'log'):
+        self.suffix = suffix
 
-#                 samples.append(fn(elem))
-#                 self.samples -= 1
+    def __enter__(self):
+        self.f = open(kaqing_log_file_name(), 'w')
+        self.f.__enter__()
 
-#             iterator = iter(samples)
-#         else:
-#             iterator = map(fn, self.collection)
+        return self.f
 
-#         if self.collect:
-#             return list(iterator)
-#         else:
-#             return iterator
+    def __exit__(self, exc_type, exc_val, exc_tb):
+        self.f.__exit__(exc_type, exc_val, exc_tb)
 
-#     def submit(self, fn: Callable[..., T], /, *args, **kwargs) -> Future[T]:
-#         if self.executor:
-#             return self.executor.submit(fn, *args, **kwargs)
-#         else:
-#             future = Future()
+        if ConfigHolder().append_command_history:
+            ConfigHolder().append_command_history(f':sh cat {self.f.name}')
 
-#             future.set_result(fn(*args, **kwargs))
+        return False
+
+def kaqing_log_file(suffix = 'log'):
+    return LogFileHandler(suffix = suffix)
+
+class CommandLog:
+    log_file = None
 
-#             return future
+    def log(line: str, config: str = 'off'):
+        if config == 'file':
+            if not CommandLog.log_file:
+                try:
+                    CommandLog.log_file = open(kaqing_log_file_name(suffix='cmd.log'), 'w')
+                except:
+                    pass
 
-def parallelize(collection: list, max_workers: int = 0, samples = sys.maxsize, msg: str = None, collect = True):
-    return ParallelMapHandler(collection, max_workers, samples = samples, msg = msg, collect = collect)
+            try:
+                CommandLog.log_file.write(line + '\n')
+            except:
+                pass
+        elif config == 'on':
+            log2(line)
 
-def offload(max_workers: int = 3, msg: str = None):
-    return OffloadHandler(max_workers, msg = msg)
+    def close_log_file():
+        if CommandLog.log_file:
+            try:
+                CommandLog.log_file.close()
+            except:
+                pass
+
+            if ConfigHolder().append_command_history:
+                ConfigHolder().append_command_history(f':sh cat {CommandLog.log_file.name}')
+
+            CommandLog.log_file = None
+
+class ExecResult(ABC):
+    def exit_code(self) -> int:
+        pass