tqdm學習-一個快速，可擴展的Python和CLI進度條
參考：https://pypi.org/project/tqdm/html
1.安裝：前端
(base) userdeMacBook-Pro:~ user$ conda activate deeplearning (deeplearning) userdeMacBook-Pro:~ user$ conda install -c conda-forge tqdm Collecting package metadata: done Solving environment: done ## Package Plan ## environment location: /anaconda3/envs/deeplearning added / updated specs: - tqdm ... Downloading and Extracting Packages python-1.6 | 3.7 MB | ##################################### | 100% tqdm-4.35.0 | 42 KB | ##################################### | 100% decorator-4.4.0 | 13 KB | ##################################### | 100% ca-certificates-2019 | 143 KB | ##################################### | 100% openssl-1.1.1c | 1.9 MB | ##################################### | 100% Preparing transaction: done Verifying transaction: done Executing transaction: done
使用這個方法安裝好像將我conda的環境的python版本換成了1.6版本，很差：python
(deeplearning) userdeMBP:bin user$ jupyter notebook Traceback (most recent call last): File "/anaconda3/envs/deeplearning/bin/jupyter", line 7, in <module> from jupyter_core.command import main ModuleNotFoundError: No module named 'jupyter_core'
解決，回到原來的狀態：git
(deeplearning) userdeMBP:bin user$ conda uninstall tqdm Collecting package metadata: done Solving environment: done ## Package Plan ## environment location: /anaconda3/envs/deeplearning removed specs: - tqdm The following packages will be REMOVED: tqdm-4.35.0-py_0 Proceed ([y]/n)? y Preparing transaction: done Verifying transaction: done Executing transaction: done (deeplearning) userdeMBP:bin user$ conda update python
The following packages will be UPDATED:
certifi pkgs/main::certifi-2019.3.9-py36_0 --> conda-forge::certifi-2019.9.11-py37_0
python 1.6-0 --> 3.7.3-h93065d6_1
readline pkgs/main::readline-7.0-h1de35cc_5 --> conda-forge::readline-8.0-hcfe32e1_0
sqlite pkgs/main::sqlite-3.27.2-ha441bb4_0 --> conda-forge::sqlite-3.29.0-hb7d70f7_1
tk pkgs/main::tk-8.6.8-ha441bb4_0 --> conda-forge::tk-8.6.9-h2573ce8_1003
wheel anaconda/pkgs/main::wheel-0.33.1-py36~ --> conda-forge::wheel-0.33.6-py37_0
在anaconda上從新安裝下jupyter notebook便可github
換成了下面的下載方法：sql
(base) userdembp:bin user$ conda activate deeplearning3.5 (deeplearning3.5) userdembp:bin user$ pip install -e git+https://github.com/tqdm/tqdm.git@master#egg=tqdm Obtaining tqdm from git+https://github.com/tqdm/tqdm.git@master#egg=tqdm Cloning https://github.com/tqdm/tqdm.git (to revision master) to ./src/tqdm Installing collected packages: tqdm Running setup.py develop for tqdm Successfully installed tqdm You are using pip version 19.0.3, however version 19.2.3 is available. You should consider upgrading via the 'pip install --upgrade pip' command.
2.使用網絡
tqdm是通用的，能以不少種方式使用。下面給出主要的三種方式：多線程
1）基於迭代器的方法：併發
即將tqdm封裝在任意迭代器中app
# conding:utf-8 from tqdm import tqdm import time text = "" for char in tqdm(["a", "b", "c", "d"]): time.sleep(0.25) text = text + char
print(text)
運行返回：
/anaconda3/envs/deeplearning3.5/bin/python3.5 /Users/user/PycharmProjects/new/learning.py 75%|███████▌ | 3/4 [00:00<00:00, 3.97it/s]abcd 100%|██████████| 4/4 [00:01<00:00, 3.96it/s] Process finished with exit code 0
tqdm(range(i))可使用trange(i)替換：
# conding:utf-8 from tqdm import tqdm import time text = 0 for i in tqdm(range(10)): time.sleep(0.25) text += i print(text)
返回：
/anaconda3/envs/deeplearning3.5/bin/python3.5 /Users/user/PycharmProjects/new/learning.py 100%|██████████| 10/10 [00:02<00:00, 3.96it/s] 45 Process finished with exit code 0
等價於：
# conding:utf-8 from tqdm import trange import time text = 0 for i in trange(10): time.sleep(0.25) text += i print(text)
返回：
/anaconda3/envs/deeplearning3.5/bin/python3.5 /Users/user/PycharmProjects/new/learning.py 90%|█████████ | 9/10 [00:02<00:00, 3.96it/s]45 100%|██████████| 10/10 [00:02<00:00, 3.96it/s] Process finished with exit code 0
在循環外面實例化可以實現tqdm()的手動控制：
# conding:utf-8 from tqdm import tqdm import time pbar = tqdm(["a", "b", "c", "d"]) for char in pbar: time.sleep(0.25) pbar.set_description("processing %s" % char)
返回：
/anaconda3/envs/deeplearning3.5/bin/python3.5 /Users/user/PycharmProjects/new/learning.py processing d: 100%|██████████| 4/4 [00:01<00:00, 3.95it/s] Process finished with exit code 0
前四步是processing a,processing b,processing c
2)手動控制
經過使用with語句來實現tqdm()的手動控制：
# conding:utf-8 from tqdm import tqdm import time with tqdm(total=100) as pbar: for i in range(10): time.sleep(0.1) pbar.update(10)
返回：
/anaconda3/envs/deeplearning3.5/bin/python3.5 /Users/user/PycharmProjects/new/learning.py 100%|██████████| 100/100 [00:01<00:00, 97.06it/s] Process finished with exit code 0
若是提供了可選變量total(或者如len()的可迭代函數)，就會顯示預測狀態
with語句也是可選的(你也能夠直接賦值tqdm()到一個變量上，重點就在於你不要忘記了在最後的時候手動del或closr()它)，使用with的好處就是它會在最後自動關閉
# conding:utf-8 from tqdm import tqdm import time pbar = tqdm(total=100) for i in range(10): time.sleep(0.1) pbar.update(10) pbar.close()
返回：
/anaconda3/envs/deeplearning3.5/bin/python3.5 /Users/user/PycharmProjects/new/learning.py 100%|██████████| 100/100 [00:01<00:00, 97.40it/s] Process finished with exit code 0
3）模塊
可能tqdm最優美的使用就是在腳本或命令行中。簡單在管道中插入tqdm（或者命令python -m tqdm），這樣在打印過程到stderr時將傳遞全部stdin到stdout
下面的例子闡述了在當前目錄中計算全部python文件中的行數的例子，而且包含這相應的記時信息：
(deeplearning3.5) userdembp:new user$ time find . -name '*.py' -type f -exec cat \{} \; | wc -l 8 real 0m0.015s user 0m0.004s sys 0m0.009s (deeplearning3.5) userdembp:new user$ time find . -name '*.py' -type f -exec cat \{} \; | tqdm | wc -l 9it [00:00, 49998.33it/s] 8 real 0m0.273s user 0m0.207s sys 0m0.053s (deeplearning3.5) userdembp:new user$
此時該目錄下就只有一個learning.py文件，裏面的代碼爲：
# conding:utf-8 from tqdm import tqdm import time pbar = tqdm(total=100) for i in range(10): time.sleep(0.1) pbar.update(10) pbar.close()
可見除去空行的確是8行
注意tqdm一般使用的參數也可以指定：
(deeplearning3.5) userdembp:new user$ time find . -name '*.py' -type f -exec cat \{} \; | tqdm -unit loc --unit_scale --total 8 >> out.log 9.00loc [00:00, 33.6kloc/s] real 0m0.383s user 0m0.218s sys 0m0.076s (deeplearning3.5) userdembp:new user$ ls learning.py out.log (deeplearning3.5) userdembp:new user$ cat out.log # conding:utf-8 from tqdm import tqdm import time pbar = tqdm(total=100) for i in range(10): time.sleep(0.1) pbar.update(10) pbar.close()
這裏即將單位換成loc，而後將獲得的內容輸入到out.log文件夾中
4）文檔
class tqdm(): """ 裝飾一個迭代器對象，返回一個表現得就像原來可迭代的迭代器；可是在每次值被請求時就打印一個動態的更新進度條 """ def __init__(self, iterable=None, desc=None, total=None, leave=True, file=None, ncols=None, mininterval=0.1, maxinterval=10.0, miniters=None, ascii=None, disable=False, unit='it', unit_scale=False, dynamic_ncols=False, smoothing=0.3, bar_format=None, initial=0, position=None, postfix=None, unit_divisor=1000):
參數：
- iterable : iterable, optional：使用一個進度條可迭代地去裝飾。留下空白去手動處理更新
- desc : str, optional：進度條的前綴
- total : int, optional：期待的迭代數。若是不指定的話，就等價於len(iterable)。若是設置爲float("inf")或者萬不得已時，只有基本的進程統計會展現出來(無ETA，也無進度條)。若是gui=True，且該參數須要子序列去更新，指定一個初始的大的隨機正整數便可，如int(9e9)
- leave : bool, optional:默認爲True，即在迭代的最後保持進度條的全部蹤影，簡單來講就是會把進度條的最終形態保留下來。若是爲None，則僅在position=0時保留下來，即保留第一個
- file : io.TextIOWrapper or io.StringIO, optional：指定輸出進程信息的地方，默認爲sys.stderr。使用file.write(str)和file.flush()方法。對於encoding編碼，可見write_bytes
- ncols : int, optional:整個輸出信息的寬度。若是指定，將在動態地從新設置進度條的大小來將其保留在這樣的邊界中。若是沒有指定，就會嘗試去使用環境設置的寬度。回退的計量寬度爲10，計數器和統計數據的大小沒有限制
- mininterval : float, optional：顯示更新間隔[默認值:0.1]秒的最小進度，即更新時間
- maxinterval : float, optional：顯示更新間隔[默認值:10]秒的最大進度。通過長時間的顯示更新滯後時會自動調整miniters參數來對應mininterval參數。只有在設置dynamic_miniters=True或顯示線程開啓時才工做
- miniters : int, optional：即在迭代中顯示更新間隔的最小進度，即更新週期。若是設置爲0或dynamic_miniters=True時就會自動調整和mininterval同樣大(CPU效率更高，適合緊湊的循環)。若是設置>0，將跳過特定迭代數的顯示。可以經過調整這個和mininterval參數去得到高效的循環。若是你的進度不穩定，迭代速度有快有慢(網絡、跳過項目等)，您應該將miniter設置爲1。若是設置了該值，mininterval則會自動設爲0
- ascii : bool or str, optional：若是沒有指定或者設置爲False，將會使用unicode編碼（光滑的塊）去填補計量。回退使用的是ASCII字符 」 123456789#」
- disable : bool, optional：是否不使用整個進度條的封裝，默認爲False，即便用封裝。若是設置爲None，則在non-TTY時不使用
- unit : str, optional：用來定義每一個迭代單元的字符串。默認爲
"it"，表示每一個迭代；在下載或解壓時，設爲"B"，表明每一個「塊」。
- unit_scale : bool or int or float, optional：若是設置爲1或者True，迭代數量就會被自動減小或者重置，且將在國際單位制標準後面添加一個度量前綴(kilo、mega等)[默認:False]，其實就是若是迭代數過大，它就會自動在後面加上M、k等字符來表示迭代進度等，好比，在下載進度條的例子中，若是爲
False，數據大小是按照字節顯示，設爲True以後轉換爲Kb、Mb。
- dynamic_ncols : bool, optional：若是設置了，就會不斷地更換環境的ncols參數（容許用於窗口大小重置），默認爲False
- smoothing : float, optional：速度估計的指數移動平均平滑因子(在GUI模式中忽略)。範圍從0(平均速度)到1(當前/瞬時速度)[默認值:0.3]。
- bar_format : str, optional：指定自定義進度欄字符串格式。可能會影響性能。(默認值: ‘{l_bar}{bar}{r_bar}’), l_bar = ’{desc}: {percentage:3.0f}%|’和r_bar = ’| {n_fmt}/{total_fmt} [{elapsed}<{remaining}, ‘ ‘{rate_fmt}{postfix}]’ 可能的var爲: l_bar, bar, r_bar, n, n_fmt, total, total_fmt, percentage, rate, rate_fmt, rate_noinv, rate_noinv_fmt, rate_inv, rate_inv_fmt, elapsed, elapsed_s, remaining, remaining_s, desc, postfix, unit。注意，若是{desc}後面是空的，那麼其後面的「:」將自動刪除。
- initial : int, optional：初始計數器值。在從新啓動進度條時有用[默認值:0]。
- position : int, optional：若是未指定，請指定要自動打印此欄的行偏移量(從0開始)。對於一次管理多個進度條是有用的(如線程)。
- postfix : dict or *, optional：指定要在進度欄末顯示的其餘統計信息。若是可能(dict)，調用set_postfix(**postfix) 。
- unit_divisor : float, optional：默認爲1000，若是unit_scale=True，則忽略它
- write_bytes : bool, optional：若是爲默認值None和file未指定時，字節將會被寫在python2中。若是設置爲True，也是寫成字節。在其餘的狀況下則默認寫成unicode格式
- 額外的CLI可選項：
- delim : chr, optional：分隔字符[默認值:' n ']。使用「0」表示null。注意::在Windows系統中，Python將「n」轉換爲「rn」。
- buf_size : int, optional：指定delim時使用的以字節爲單位的字符串緩衝區大小[默認值:256]。
- bytes : bool, optional：若是爲真，將計數字節，忽略delim參數，並默認unit_scale爲真，unit_divisor爲1024，unit爲' B '。
- manpath : str, optional：安裝tqdm手冊頁的目錄。
- log : str, optional：打印的日誌信息類別，CRITICAL|FATAL|ERROR|WARN(ING)|[default: ‘INFO’]|DEBUG|NOTSET，默認爲INFO
- 返回：
- 裝飾後的迭代器
class tqdm(): def update(self, n=1): """ 手動更新進度條，對流streams有用，好比讀文件 E.g.: >>> t = tqdm(total=filesize) # Initialise >>> for current_buffer in stream: ... ... ... t.update(len(current_buffer)) >>> t.close() 最後一行高度推薦使用，可是若是``t.update()`` 是在``filesize``即將徹底到達和打印時調用的話可能就不須要 Parameters ---------- n : int, optional 添加到迭代內部計數器的增加數[default: 1] """ def close(self): """清除(if leave=False)和關閉進度條""" def clear(self, nomove=False): """清除當前的進度條顯示.""" def refresh(self): """強迫更新該進度條的顯示Force refresh the display of this bar.""" def unpause(self): """從最新運行時間重啓tqdm計時器""" def reset(self, total=None): """ 爲了重複使用，重設爲第0次迭代。考慮和``leave=True``設置一塊兒使用 Parameters ---------- total : int, optional. 用於新進度條的次數. """ def set_description(self, desc=None, refresh=True): """ 設置/修改進度條的描述格式 Parameters ---------- desc : str, optional refresh : bool, optional Forces refresh [default: True]. """ def set_postfix(self, ordered_dict=None, refresh=True, **kwargs): """ 設置/修改後綴(additional stats) with automatic formatting based on datatype. Parameters ---------- ordered_dict : dict or OrderedDict, optional refresh : bool, optional Forces refresh [default: True]. kwargs : dict, optional """ @classmethod def write(cls, s, file=sys.stdout, end="\n"): """經過tqdm打印信息(不覆蓋進度條).""" @property def format_dict(self): """給只讀權限人員訪問的公用API""" def display(self, msg=None, pos=None): """ 使用``self.sp`` 去展現指定``pos``中的``msg``. 當繼承使用時，要考慮重載該函數 e.g.: ``self.some_frontend(**self.format_dict)`` instead of ``self.sp``. Parameters ---------- msg : str, optional. What to display (default: ``repr(self)``). pos : int, optional. Position to ``moveto`` (default: ``abs(self.pos)``). """ def trange(*args, **kwargs): """ tqdm(xrange(*args), **kwargs)函數的縮寫 Python3+版本中使用 range來替換 xrange. """ class tqdm_gui(tqdm): """Experimental GUI version""" def tgrange(*args, **kwargs): """Experimental GUI version of trange""" class tqdm_notebook(tqdm): """Experimental IPython/Jupyter Notebook widget""" def tnrange(*args, **kwargs): """Experimental IPython/Jupyter Notebook widget version of trange"""
""" # Simple tqdm examples and profiling # Benchmark for i in _range(int(1e8)): pass # Basic demo import tqdm for i in tqdm.trange(int(1e8)): pass # Some decorations import tqdm for i in tqdm.trange(int(1e8), miniters=int(1e6), ascii=True, desc="cool", dynamic_ncols=True): pass # Nested bars from tqdm import trange for i in trange(10): for j in trange(int(1e7), leave=False, unit_scale=True): pass # Experimental GUI demo import tqdm for i in tqdm.tgrange(int(1e8)): pass # Comparison to https://code.google.com/p/python-progressbar/ try: from progressbar.progressbar import ProgressBar except ImportError: pass else: for i in ProgressBar()(_range(int(1e8))): pass # Dynamic miniters benchmark from tqdm import trange for i in trange(int(1e8), miniters=None, mininterval=0.1, smoothing=0): pass # Fixed miniters benchmark from tqdm import trange for i in trange(int(1e8), miniters=4500000, mininterval=0.1, smoothing=0): pass """ from time import sleep from timeit import timeit import re # Simple demo from tqdm import trange for i in trange(16, leave=True): sleep(0.1) # Profiling/overhead tests stmts = filter(None, re.split(r'\n\s*#.*?\n', __doc__)) for s in stmts: print(s.replace('import tqdm\n', '')) print(timeit(stmt='try:\n\t_range = xrange' '\nexcept:\n\t_range = range\n' + s, number=1), 'seconds')
一個個分析：
""" # Simple tqdm examples and profiling # Benchmark for i in _range(int(1e8)): pass """ from time import sleep from timeit import timeit import re # Simple demo from tqdm import trange for i in trange(16, leave=True): sleep(0.1) # Profiling/overhead tests stmts = filter(None, re.split(r'\n\s*#.*?\n', __doc__)) for s in stmts: print(s.replace('import tqdm\n', '')) print(timeit(stmt='try:\n\t_range = xrange' '\nexcept:\n\t_range = range\n' + s, number=1), 'seconds')
返回：
100%|██████████| 16/16 [00:01<00:00, 9.59it/s] # Benchmark for i in _range(int(1e8)): pass 2.328720851000071 seconds
下面都同樣，僅運行一個來分析學習
1）
""" # Simple tqdm examples and profiling # Basic demo import tqdm for i in tqdm.trange(int(1e8)): pass """
返回：
100%|██████████| 16/16 [00:01<00:00, 9.65it/s] 0%| | 301598/100000000 [00:00<00:33, 3015975.69it/s] # Basic demo for i in tqdm.trange(int(1e8)): pass 100%|██████████| 100000000/100000000 [00:18<00:00, 5407942.99it/s] 18.4932925789999 seconds
2）
""" # Simple tqdm examples and profiling # Some decorations import tqdm for i in tqdm.trange(int(1e8), miniters=int(1e6), ascii=True, desc="cool", dynamic_ncols=True): pass """
中間：
返回：
100%|██████████| 16/16 [00:01<00:00, 9.71it/s] cool: 0%| | 0/100000000 [00:00<?, ?it/s] # Some decorations for i in tqdm.trange(int(1e8), miniters=int(1e6), ascii=True, desc="cool", dynamic_ncols=True): pass cool: 100%|##########| 100000000/100000000 [00:17<00:00, 5650878.97it/s] 17.699349756000174 seconds
3）
""" # Simple tqdm examples and profiling # Nested bars from tqdm import trange for i in trange(2): for j in trange(int(1e7), leave=False, unit_scale=True): pass """
返回：
100%|██████████| 16/16 [00:01<00:00, 9.76it/s] 0%| | 0/2 [00:00<?, ?it/s] 0%| | 0.00/10.0M [00:00<?, ?it/s] 3%|▎ | 327k/10.0M [00:00<00:02, 3.27Mit/s] # Nested bars from tqdm import trange for i in trange(2): for j in trange(int(1e7), leave=False, unit_scale=True): pass 7%|▋ | 667k/10.0M [00:00<00:02, 3.31Mit/s] 10%|█ | 1.04M/10.0M [00:00<00:02, 3.43Mit/s] 13%|█▎ | 1.34M/10.0M [00:00<00:02, 3.27Mit/s] 17%|█▋ | 1.70M/10.0M [00:00<00:02, 3.38Mit/s] 21%|██ | 2.11M/10.0M [00:00<00:02, 3.56Mit/s] 25%|██▌ | 2.55M/10.0M [00:00<00:01, 3.77Mit/s] 30%|██▉ | 2.99M/10.0M [00:00<00:01, 3.93Mit/s] 34%|███▍ | 3.44M/10.0M [00:00<00:01, 4.11Mit/s] 39%|███▉ | 3.91M/10.0M [00:01<00:01, 4.27Mit/s] 44%|████▍ | 4.39M/10.0M [00:01<00:01, 4.41Mit/s] 49%|████▉ | 4.88M/10.0M [00:01<00:01, 4.54Mit/s] 54%|█████▍ | 5.38M/10.0M [00:01<00:00, 4.67Mit/s] 59%|█████▊ | 5.87M/10.0M [00:01<00:00, 4.74Mit/s] 64%|██████▎ | 6.35M/10.0M [00:01<00:00, 4.77Mit/s] 68%|██████▊ | 6.83M/10.0M [00:01<00:00, 4.76Mit/s] 73%|███████▎ | 7.31M/10.0M [00:01<00:00, 4.64Mit/s] 78%|███████▊ | 7.77M/10.0M [00:01<00:00, 4.48Mit/s] 82%|████████▏ | 8.22M/10.0M [00:01<00:00, 4.44Mit/s] 87%|████████▋ | 8.71M/10.0M [00:02<00:00, 4.55Mit/s] 92%|█████████▏| 9.20M/10.0M [00:02<00:00, 4.67Mit/s] 97%|█████████▋| 9.68M/10.0M [00:02<00:00, 4.69Mit/s] 50%|█████ | 1/2 [00:02<00:02, 2.30s/it] 0%| | 0.00/10.0M [00:00<?, ?it/s] 3%|▎ | 296k/10.0M [00:00<00:03, 2.96Mit/s] 6%|▌ | 586k/10.0M [00:00<00:03, 2.94Mit/s] 9%|▉ | 896k/10.0M [00:00<00:03, 2.99Mit/s] 12%|█▏ | 1.24M/10.0M [00:00<00:02, 3.10Mit/s] 16%|█▌ | 1.59M/10.0M [00:00<00:02, 3.21Mit/s] 19%|█▉ | 1.94M/10.0M [00:00<00:02, 3.31Mit/s] 23%|██▎ | 2.33M/10.0M [00:00<00:02, 3.46Mit/s] 27%|██▋ | 2.71M/10.0M [00:00<00:02, 3.57Mit/s] 31%|███ | 3.12M/10.0M [00:00<00:01, 3.69Mit/s] 35%|███▌ | 3.53M/10.0M [00:01<00:01, 3.81Mit/s] 39%|███▉ | 3.93M/10.0M [00:01<00:01, 3.88Mit/s] 44%|████▍ | 4.40M/10.0M [00:01<00:01, 4.08Mit/s] 49%|████▉ | 4.88M/10.0M [00:01<00:01, 4.27Mit/s] 54%|█████▍ | 5.38M/10.0M [00:01<00:01, 4.48Mit/s] 59%|█████▊ | 5.87M/10.0M [00:01<00:00, 4.60Mit/s] 64%|██████▍ | 6.38M/10.0M [00:01<00:00, 4.72Mit/s] 69%|██████▉ | 6.90M/10.0M [00:01<00:00, 4.85Mit/s] 74%|███████▍ | 7.41M/10.0M [00:01<00:00, 4.93Mit/s] 79%|███████▉ | 7.92M/10.0M [00:01<00:00, 4.98Mit/s] 84%|████████▍ | 8.42M/10.0M [00:02<00:00, 4.74Mit/s] 89%|████████▉ | 8.90M/10.0M [00:02<00:00, 4.77Mit/s] 94%|█████████▍| 9.40M/10.0M [00:02<00:00, 4.84Mit/s] 99%|█████████▉| 9.93M/10.0M [00:02<00:00, 4.95Mit/s] 100%|██████████| 2/2 [00:04<00:00, 2.32s/it] 4.642330375000256 seconds
⚠️這裏返回[00:00<00:03, 2.96Mit/s]中前面的00:00表示已用時間，以秒爲單位，因此在1秒前都爲0，<後面的00:03表示剩餘須要花的時間，2.96Mit/s表示速度
刪掉參數unit_scale=True：
返回可見設置時使用M單位簡化數字：
100%|██████████| 16/16 [00:01<00:00, 9.64it/s] 0%| | 0/1 [00:00<?, ?it/s] 0%| | 0/10000000 [00:00<?, ?it/s] 3%|▎ | 277015/10000000 [00:00<00:03, 2770126.22it/s] # Nested bars from tqdm import trange for i in trange(1): for j in trange(int(1e7), leave=False): pass 6%|▌ | 558583/10000000 [00:00<00:03, 2783635.65it/s] 9%|▉ | 895786/10000000 [00:00<00:03, 2937395.56it/s] 13%|█▎ | 1258399/10000000 [00:00<00:02, 3114883.50it/s] 17%|█▋ | 1667463/10000000 [00:00<00:02, 3354857.99it/s] 21%|██ | 2105227/10000000 [00:00<00:02, 3607718.69it/s] 26%|██▌ | 2551042/10000000 [00:00<00:01, 3826706.29it/s] 30%|███ | 3028044/10000000 [00:00<00:01, 4068052.94it/s] 35%|███▌ | 3502923/10000000 [00:00<00:01, 4250850.87it/s] 40%|███▉ | 3975152/10000000 [00:01<00:01, 4382092.05it/s] 44%|████▍ | 4444320/10000000 [00:01<00:01, 4470580.81it/s] 49%|████▉ | 4913953/10000000 [00:01<00:01, 4535991.51it/s] 54%|█████▍ | 5412819/10000000 [00:01<00:00, 4662832.67it/s] 59%|█████▉ | 5903458/10000000 [00:01<00:00, 4733320.55it/s] 64%|██████▍ | 6407480/10000000 [00:01<00:00, 4821381.19it/s] 69%|██████▉ | 6890932/10000000 [00:01<00:00, 4732373.17it/s] 74%|███████▎ | 7365457/10000000 [00:01<00:00, 4589660.71it/s] 78%|███████▊ | 7828722/10000000 [00:01<00:00, 4602468.26it/s] 83%|████████▎ | 8290265/10000000 [00:01<00:00, 4587601.91it/s] 88%|████████▊ | 8799822/10000000 [00:02<00:00, 4729019.14it/s] 93%|█████████▎| 9297792/10000000 [00:02<00:00, 4801530.70it/s] 98%|█████████▊| 9806813/10000000 [00:02<00:00, 4884621.49it/s] 100%|██████████| 1/1 [00:02<00:00, 2.26s/it] 2.264081185000123 seconds
設置leave=True，返回：
100%|██████████| 16/16 [00:01<00:00, 9.66it/s] 0%| | 0/1 [00:00<?, ?it/s] 0%| | 0/10000000 [00:00<?, ?it/s] 3%|▎ | 322630/10000000 [00:00<00:02, 3226280.00it/s] # Nested bars from tqdm import trange for i in trange(1): for j in trange(int(1e7), leave=True): pass 7%|▋ | 672600/10000000 [00:00<00:02, 3303711.10it/s] 11%|█ | 1057128/10000000 [00:00<00:02, 3449449.52it/s] 14%|█▍ | 1427037/10000000 [00:00<00:02, 3520729.15it/s] 18%|█▊ | 1832741/10000000 [00:00<00:02, 3666110.82it/s] 22%|██▏ | 2244986/10000000 [00:00<00:02, 3792038.55it/s] 27%|██▋ | 2691460/10000000 [00:00<00:01, 3971547.54it/s] 32%|███▏ | 3158486/10000000 [00:00<00:01, 4158176.06it/s] 36%|███▋ | 3644204/10000000 [00:00<00:01, 4345786.38it/s] 41%|████ | 4122465/10000000 [00:01<00:01, 4468208.17it/s] 46%|████▌ | 4612764/10000000 [00:01<00:01, 4590323.02it/s] 51%|█████ | 5111718/10000000 [00:01<00:01, 4703203.98it/s] 56%|█████▌ | 5608081/10000000 [00:01<00:00, 4778414.54it/s] 61%|██████ | 6086716/10000000 [00:01<00:00, 4780780.12it/s] 66%|██████▌ | 6564640/10000000 [00:01<00:00, 4667924.67it/s] 70%|███████ | 7031914/10000000 [00:01<00:00, 4532180.69it/s] 75%|███████▌ | 7521866/10000000 [00:01<00:00, 4636454.48it/s] 80%|████████ | 8035404/10000000 [00:01<00:00, 4775645.33it/s] 86%|████████▌ | 8551806/10000000 [00:01<00:00, 4885870.39it/s] 91%|█████████ | 9066446/10000000 [00:02<00:00, 4961214.82it/s] 100%|██████████| 10000000/10000000 [00:02<00:00, 4552905.40it/s][A 100%|██████████| 1/1 [00:02<00:00, 2.20s/it] 2.2038861539999743 seconds
不一樣在於內部迭代保存最終的結果，即：
100%|██████████| 10000000/10000000 [00:02<00:00, 4552905.40it/s]
4）
""" # Simple tqdm examples and profiling # Experimental GUI demo import tqdm for i in tqdm.tgrange(int(1e8)): pass """
返回：
100%|██████████| 16/16 [00:01<00:00, 9.69it/s] # Experimental GUI demo for i in tqdm.tgrange(int(1e8)): pass
19.253753004999908 seconds
圖爲：
5）
""" # Simple tqdm examples and profiling # Comparison to https://code.google.com/p/python-progressbar/ try: from progressbar.progressbar import ProgressBar except ImportError: pass else: for i in ProgressBar()(_range(int(1e8))): pass """
返回：
100%|██████████| 16/16 [00:01<00:00, 9.64it/s] # Comparison to https://code.google.com/p/python-progressbar/ try: from progressbar.progressbar import ProgressBar except ImportError: pass else: for i in ProgressBar()(_range(int(1e8))): pass 0.0015016719999039196 seconds
6）
""" # Simple tqdm examples and profiling # Dynamic miniters benchmark from tqdm import trange for i in trange(int(1e8), miniters=None, mininterval=0.1, smoothing=0): pass """
最終：
100%|██████████| 16/16 [00:01<00:00, 9.65it/s] 0%| | 313552/100000000 [00:00<00:31, 3134484.22it/s] # Dynamic miniters benchmark from tqdm import trange for i in trange(int(1e8), miniters=None, mininterval=0.1, smoothing=0): pass 100%|██████████| 100000000/100000000 [00:16<00:00, 5960460.94it/s] 16.779653078000138 seconds
即mininterval=0.1秒後更新進度欄中的進度
若是設置爲miniters=2, mininterval=0：
""" # Simple tqdm examples and profiling # Dynamic miniters benchmark from tqdm import trange for i in trange(int(1e8), miniters=2, mininterval=0): pass """
返回爲：
0%| | 0/100000000 [00:00<?, ?it/s] 0%| | 2/100000000 [00:00<06:45, 246723.76it/s] 0%| | 4/100000000 [00:00<6:07:28, 4535.36it/s] 0%| | 6/100000000 [00:00<15:44:50, 1763.97it/s] 0%| | 8/100000000 [00:00<19:05:36, 1454.82it/s] ...
可見每兩個迭代就更新一次
若是同時設置了這兩個參數miniters=2, mininterval=1，以大的設置的時間爲主。以下面的這個設置miniters僅爲2，花的時間少，mininterval爲1秒，因此以1秒間隔顯示爲主，忽略miniters設置：
0%| | 0/100000000 [00:00<?, ?it/s] # Dynamic miniters benchmark from tqdm import trange for i in trange(int(1e8), miniters=2, mininterval=1): pass 3%|▎ | 3127793/100000000 [00:01<00:30, 3127793.00it/s] 6%|▌ | 5865282/100000000 [00:02<00:31, 2999494.96it/s] 9%|▉ | 9147057/100000000 [00:03<00:29, 3078945.08it/s] 12%|█▏ | 12443754/100000000 [00:04<00:27, 3141189.20it/s] ...
注意：返回[00:01<00:30, 3127793.00it/s]中的00:01表示1秒
若是設置爲miniters=4500000, mininterval=0.1，4500000個間隔花的時間更長，因此以迭代數爲主：
0%| | 0/100000000 [00:00<?, ?it/s] # Fixed miniters benchmark from tqdm import trange for i in trange(int(1e8), miniters=4500000, mininterval=0.1, smoothing=0): pass 4%|▍ | 4500000/100000000 [00:00<00:16, 5715229.42it/s] 9%|▉ | 9000000/100000000 [00:01<00:16, 5545180.93it/s] 14%|█▎ | 13500000/100000000 [00:02<00:15, 5635364.34it/s] 18%|█▊ | 18000000/100000000 [00:03<00:14, 5656277.70it/s] 22%|██▎ | 22500000/100000000 [00:03<00:13, 5729412.82it/s] 27%|██▋ | 27000000/100000000 [00:04<00:12, 5779243.27it/s] 32%|███▏ | 31500000/100000000 [00:05<00:11, 5820303.28it/s] 36%|███▌ | 36000000/100000000 [00:06<00:10, 5851854.45it/s] 40%|████ | 40500000/100000000 [00:06<00:10, 5873663.41it/s] 45%|████▌ | 45000000/100000000 [00:07<00:09, 5902913.08it/s]
...
Description and additional stats
定製信息能夠經過設置desc和postfix參數來動態顯示和更新在tqdm進度欄上：
from tqdm import trange from random import random, randint from time import sleep with trange(10) as t: for i in t: # 描述將顯示在左邊 t.set_description('GEN %i' % i) # 後綴將顯示在右邊，根據參數的數據類型自動格式化 t.set_postfix(loss=random(), gen=randint(1,999), str='h', lst=[1, 2]) sleep(0.1) with tqdm(total=10, bar_format="{postfix[0]} {postfix[1][value]:>8.2g}", postfix=["Batch", dict(value=0)]) as t: for i in range(10): sleep(0.1) t.postfix[1]["value"] = i / 2 t.update()
返回：
GEN 9: 100%|██████████| 10/10 [00:01<00:00, 9.42it/s, gen=356, loss=0.806, lst=[1, 2], str=h] Batch 4.5
記得在bar_format字符串中使用{postfix[...]}來指向：
- postfix須要在兼容格式中做爲初始參數傳遞
- 若是postfix是類字典對象，將自動轉換爲一個字符串。爲了防止該行爲，在字典中鍵不是字符串的地方加入一個額外的項，即上面postfix=["Batch", dict(value=0)]中的value=0
額外的bar_format參數也可以經過複寫format_dict參數來定義，進度欄自己能夠用ascii碼修改:
from tqdm import tqdm class TqdmExtraFormat(tqdm): """Provides a `total_time` format parameter""" @property def format_dict(self): d = super(TqdmExtraFormat, self).format_dict
#d["total"]表示總迭代數，d["n"]表示當前爲第幾輪迭代 print(d["elapsed"], d["total"], d["n"]) total_time = d["elapsed"] * (d["total"] or 0) / max(d["n"], 1) d.update(total_time=self.format_interval(total_time) + " in total") return d for i in TqdmExtraFormat( range(10), ascii=" .oO0", bar_format="{total_time}: {percentage:.0f}%|{bar}{r_bar}"): sleep(0.25) print(i)
返回：
00:00 in total: 0%| | 0/10 [00:00<?, ?it/s] 0 10 0 00:02 in total: 10%|0 | 1/10 [00:00<00:02, 3.92it/s] 0 0.2551310062408447 10 1 00:02 in total: 20%|00 | 2/10 [00:00<00:02, 3.92it/s] 1 0.5111739635467529 10 2 00:02 in total: 30%|000 | 3/10 [00:00<00:01, 3.91it/s] 2 0.7674551010131836 10 3 00:02 in total: 40%|0000 | 4/10 [00:01<00:01, 3.92it/s] 3 1.0219080448150635 10 4 00:02 in total: 50%|00000 | 5/10 [00:01<00:01, 3.93it/s] 4 1.2738640308380127 10 5 00:02 in total: 60%|000000 | 6/10 [00:01<00:01, 3.94it/s] 5 1.525794267654419 10 6 00:02 in total: 70%|0000000 | 7/10 [00:01<00:00, 3.94it/s] 6 1.7809131145477295 10 7 00:02 in total: 80%|00000000 | 8/10 [00:02<00:00, 3.94it/s] 7 2.035114049911499 10 8 00:02 in total: 90%|000000000 | 9/10 [00:02<00:00, 3.93it/s] 8 2.2901201248168945 10 9 00:02 in total: 100%|0000000000| 10/10 [00:02<00:00, 3.93it/s] 9 2.543692111968994 10 10 2.544473886489868 10 10
參數ascii=" .oO0"指定使用0來填補進度條
參數bar_format="{total_time}: {percentage:.0f}%|{bar}{r_bar}"定義的值total_time即覆寫format_dict中定義的total_time值，即輸出的00:00 in total；{percentage:.0f}%即表示進度的百分比，小數位值不顯示，設爲0；因此l_bar = {total_time}: {percentage:.0f}%|,{bar}{r_bar}沒有定義，即表示使用默認定義
注意{bar}還支持格式說明符[width][type]
- width:
- 未指定(默認值):自動填充ncols
- 設置的是整數且int >= 0: 即覆蓋ncols邏輯的固定寬度
- 設置的是整數且int < 0: 則使用默認值減去該值
- type:
- a : ascii (覆寫即等價於ascii=True)
- u : unicode(覆寫 ascii=False)
- b : blank(覆寫 ascii=" ")
這意味着可使用如下方法建立具備右對齊文本的固定欄:bar_format="{l_bar}{bar:10}|{bar:-10b}right-justified"
Nested progress bars
tqdm支持嵌套的進度條。這裏有一個例子:
from tqdm import trange from time import sleep for i in trange(4, desc='1st loop'): for j in trange(5, desc='2nd loop'): for k in trange(20, desc='3nd loop', leave=False): sleep(0.01)
返回：
1st loop: 0%| | 0/4 [00:00<?, ?it/s] 2nd loop: 0%| | 0/5 [00:00<?, ?it/s] 3nd loop: 0%| | 0/20 [00:00<?, ?it/s] 3nd loop: 50%|█████ | 10/20 [00:00<00:00, 90.89it/s] 3nd loop: 95%|█████████▌| 19/20 [00:00<00:00, 90.62it/s] 2nd loop: 20%|██ | 1/5 [00:00<00:00, 4.39it/s] 3nd loop: 0%| | 0/20 [00:00<?, ?it/s] 3nd loop: 45%|████▌ | 9/20 [00:00<00:00, 86.21it/s] 3nd loop: 90%|█████████ | 18/20 [00:00<00:00, 85.41it/s] 2nd loop: 40%|████ | 2/5 [00:00<00:00, 4.29it/s] 3nd loop: 0%| | 0/20 [00:00<?, ?it/s] 3nd loop: 50%|█████ | 10/20 [00:00<00:00, 95.65it/s] 3nd loop: 95%|█████████▌| 19/20 [00:00<00:00, 93.32it/s] 2nd loop: 60%|██████ | 3/5 [00:00<00:00, 4.31it/s] 3nd loop: 0%| | 0/20 [00:00<?, ?it/s] 3nd loop: 50%|█████ | 10/20 [00:00<00:00, 92.64it/s] 3nd loop: 95%|█████████▌| 19/20 [00:00<00:00, 91.18it/s] 2nd loop: 80%|████████ | 4/5 [00:00<00:00, 4.32it/s] 3nd loop: 0%| | 0/20 [00:00<?, ?it/s] 3nd loop: 50%|█████ | 10/20 [00:00<00:00, 94.50it/s] 3nd loop: 95%|█████████▌| 19/20 [00:00<00:00, 91.41it/s] 2nd loop: 100%|██████████| 5/5 [00:01<00:00, 4.27it/s] 1st loop: 25%|██▌ | 1/4 [00:01<00:03, 1.17s/it] 2nd loop: 0%| | 0/5 [00:00<?, ?it/s] 3nd loop: 0%| | 0/20 [00:00<?, ?it/s] 3nd loop: 50%|█████ | 10/20 [00:00<00:00, 90.59it/s] 3nd loop: 95%|█████████▌| 19/20 [00:00<00:00, 89.93it/s] 2nd loop: 20%|██ | 1/5 [00:00<00:00, 4.33it/s] 3nd loop: 0%| | 0/20 [00:00<?, ?it/s] 3nd loop: 50%|█████ | 10/20 [00:00<00:00, 91.40it/s] 3nd loop: 95%|█████████▌| 19/20 [00:00<00:00, 90.54it/s] 2nd loop: 40%|████ | 2/5 [00:00<00:00, 4.32it/s] 3nd loop: 0%| | 0/20 [00:00<?, ?it/s] 3nd loop: 50%|█████ | 10/20 [00:00<00:00, 93.82it/s] 3nd loop: 95%|█████████▌| 19/20 [00:00<00:00, 91.93it/s] 2nd loop: 60%|██████ | 3/5 [00:00<00:00, 4.34it/s] 3nd loop: 0%| | 0/20 [00:00<?, ?it/s] 3nd loop: 50%|█████ | 10/20 [00:00<00:00, 96.33it/s] 3nd loop: 100%|██████████| 20/20 [00:00<00:00, 94.51it/s] 2nd loop: 80%|████████ | 4/5 [00:00<00:00, 4.37it/s] 3nd loop: 0%| | 0/20 [00:00<?, ?it/s] 3nd loop: 50%|█████ | 10/20 [00:00<00:00, 93.01it/s] 3nd loop: 100%|██████████| 20/20 [00:00<00:00, 92.52it/s] 2nd loop: 100%|██████████| 5/5 [00:01<00:00, 4.36it/s] 1st loop: 50%|█████ | 2/4 [00:02<00:02, 1.17s/it] 2nd loop: 0%| | 0/5 [00:00<?, ?it/s] 3nd loop: 0%| | 0/20 [00:00<?, ?it/s] 3nd loop: 50%|█████ | 10/20 [00:00<00:00, 92.09it/s] 3nd loop: 95%|█████████▌| 19/20 [00:00<00:00, 90.62it/s] 2nd loop: 20%|██ | 1/5 [00:00<00:00, 4.33it/s] 3nd loop: 0%| | 0/20 [00:00<?, ?it/s] 3nd loop: 50%|█████ | 10/20 [00:00<00:00, 90.28it/s] 3nd loop: 95%|█████████▌| 19/20 [00:00<00:00, 89.40it/s] 2nd loop: 40%|████ | 2/5 [00:00<00:00, 4.31it/s] 3nd loop: 0%| | 0/20 [00:00<?, ?it/s] 3nd loop: 50%|█████ | 10/20 [00:00<00:00, 94.83it/s] 3nd loop: 95%|█████████▌| 19/20 [00:00<00:00, 92.97it/s] 2nd loop: 60%|██████ | 3/5 [00:00<00:00, 4.32it/s] 3nd loop: 0%| | 0/20 [00:00<?, ?it/s] 3nd loop: 50%|█████ | 10/20 [00:00<00:00, 90.53it/s] 3nd loop: 95%|█████████▌| 19/20 [00:00<00:00, 90.02it/s] 2nd loop: 80%|████████ | 4/5 [00:00<00:00, 4.33it/s] 3nd loop: 0%| | 0/20 [00:00<?, ?it/s] 3nd loop: 50%|█████ | 10/20 [00:00<00:00, 91.51it/s] 3nd loop: 95%|█████████▌| 19/20 [00:00<00:00, 89.57it/s] 2nd loop: 100%|██████████| 5/5 [00:01<00:00, 4.29it/s] 1st loop: 75%|███████▌ | 3/4 [00:03<00:01, 1.17s/it] 2nd loop: 0%| | 0/5 [00:00<?, ?it/s] 3nd loop: 0%| | 0/20 [00:00<?, ?it/s] 3nd loop: 50%|█████ | 10/20 [00:00<00:00, 94.66it/s] 3nd loop: 95%|█████████▌| 19/20 [00:00<00:00, 92.52it/s] 2nd loop: 20%|██ | 1/5 [00:00<00:00, 4.41it/s] 3nd loop: 0%| | 0/20 [00:00<?, ?it/s] 3nd loop: 45%|████▌ | 9/20 [00:00<00:00, 88.28it/s] 3nd loop: 95%|█████████▌| 19/20 [00:00<00:00, 88.91it/s] 2nd loop: 40%|████ | 2/5 [00:00<00:00, 4.36it/s] 3nd loop: 0%| | 0/20 [00:00<?, ?it/s] 3nd loop: 45%|████▌ | 9/20 [00:00<00:00, 86.31it/s] 3nd loop: 90%|█████████ | 18/20 [00:00<00:00, 86.96it/s] 2nd loop: 60%|██████ | 3/5 [00:00<00:00, 4.31it/s] 3nd loop: 0%| | 0/20 [00:00<?, ?it/s] 3nd loop: 45%|████▌ | 9/20 [00:00<00:00, 89.50it/s] 3nd loop: 90%|█████████ | 18/20 [00:00<00:00, 87.95it/s] 2nd loop: 80%|████████ | 4/5 [00:00<00:00, 4.27it/s] 3nd loop: 0%| | 0/20 [00:00<?, ?it/s] 3nd loop: 50%|█████ | 10/20 [00:00<00:00, 92.30it/s] 3nd loop: 100%|██████████| 20/20 [00:00<00:00, 91.61it/s] 2nd loop: 100%|██████████| 5/5 [00:01<00:00, 4.26it/s] 1st loop: 100%|██████████| 4/4 [00:04<00:00, 1.17s/it]
參數desc='3nd loop'指定輸出的l_bar的內容，即前綴
在Windows上，若是能夠的話，colorama將用於保持嵌套條在各自的行上。
對於手動控制定位(例如多線程使用)，能夠指定位置=n，其中最外層的欄位n=0，下一欄位n=1，以此類推:
from time import sleep from tqdm import trange, tqdm from multiprocessing import Pool, freeze_support, RLock L = list(range(3)) def progresser(n): interval = 0.001 / (n + 2) total = 5000 # {:<04.2}表示左對齊，數字總共4位，其中小數兩位 text = "#{}, est. {:<04.2}s".format(n, interval * total) for i in trange(total, desc=text, position=n): sleep(interval) if __name__ == '__main__': freeze_support() # for Windows support p = Pool(len(L), # 同時開啓3個進程 # again, for Windows support initializer=tqdm.set_lock, initargs=(RLock(),)) p.map(progresser, L) print("\n" * (len(L) - 2)) #換行
返回：
#0, est. 2.50s: 0%| | 0/5000 [00:00<?, ?it/s] #1, est. 1.70s: 0%| | 0/5000 [00:00<?, ?it/s] #0, est. 2.50s: 4%|▎ | 175/5000 [00:00<00:02, 1743.72it/s] #1, est. 1.70s: 5%|▌ | 258/5000 [00:00<00:01, 2577.38it/s] #0, est. 2.50s: 7%|▋ | 346/5000 [00:00<00:02, 1731.96it/s] #1, est. 1.70s: 10%|█ | 500/5000 [00:00<00:01, 2525.19it/s] #0, est. 2.50s: 10%|█ | 515/5000 [00:00<00:02, 1717.89it/s] #1, est. 1.70s: 15%|█▍ | 729/5000 [00:00<00:01, 2447.84it/s] #0, est. 2.50s: 14%|█▎ | 685/5000 [00:00<00:02, 1712.16it/s] #1, est. 1.70s: 19%|█▉ | 959/5000 [00:00<00:01, 2401.48it/s] #0, est. 2.50s: 17%|█▋ | 855/5000 [00:00<00:02, 1707.39it/s]] #1, est. 1.70s: 24%|██▍ | 1200/5000 [00:00<00:01, 2403.36it/s] #0, est. 2.50s: 21%|██ | 1026/5000 [00:00<00:02, 1706.94it/s] #1, est. 1.70s: 29%|██▊ | 1435/5000 [00:00<00:01, 2385.71it/s] #0, est. 2.50s: 24%|██▍ | 1198/5000 [00:00<00:02, 1709.67it/s] #1, est. 1.70s: 34%|███▎ | 1675/5000 [00:00<00:01, 2389.13it/s] #0, est. 2.50s: 27%|██▋ | 1369/5000 [00:00<00:02, 1707.61it/s] #1, est. 1.70s: 38%|███▊ | 1914/5000 [00:00<00:01, 2387.64it/s] #0, est. 2.50s: 31%|███ | 1541/5000 [00:00<00:02, 1709.89it/s] #1, est. 1.70s: 43%|████▎ | 2153/5000 [00:00<00:01, 2388.28it/s] #0, est. 2.50s: 34%|███▍ | 1712/5000 [00:01<00:01, 1708.18it/s] #1, est. 1.70s: 48%|████▊ | 2392/5000 [00:01<00:01, 2387.33it/s] #0, est. 2.50s: 38%|███▊ | 1882/5000 [00:01<00:01, 1704.82it/s] #1, est. 1.70s: 53%|█████▎ | 2632/5000 [00:01<00:00, 2388.09it/s] #0, est. 2.50s: 41%|████ | 2052/5000 [00:01<00:01, 1701.67it/s] #1, est. 1.70s: 57%|█████▋ | 2867/5000 [00:01<00:00, 2376.26it/s] #0, est. 2.50s: 44%|████▍ | 2224/5000 [00:01<00:01, 1704.14it/s] #1, est. 1.70s: 62%|██████▏ | 3106/5000 [00:01<00:00, 2379.52it/s] #0, est. 2.50s: 48%|████▊ | 2395/5000 [00:01<00:01, 1702.96it/s] #1, est. 1.70s: 67%|██████▋ | 3342/5000 [00:01<00:00, 2364.88it/s] #0, est. 2.50s: 51%|█████▏ | 2566/5000 [00:01<00:01, 1703.43it/s] #1, est. 1.70s: 72%|███████▏ | 3579/5000 [00:01<00:00, 2365.95it/s] #0, est. 2.50s: 55%|█████▍ | 2736/5000 [00:01<00:01, 1702.05it/s] #1, est. 1.70s: 76%|███████▋ | 3817/5000 [00:01<00:00, 2368.33it/s] #2, est. 1.20s: 100%|██████████| 5000/5000 [00:01<00:00, 3076.41it/s] #0, est. 2.50s: 58%|█████▊ | 2906/5000 [00:01<00:01, 1695.95it/s] #0, est. 2.50s: 62%|██████▏ | 3076/5000 [00:01<00:01, 1688.59it/s] #0, est. 2.50s: 65%|██████▍ | 3245/5000 [00:01<00:01, 1687.35it/s] #0, est. 2.50s: 68%|██████▊ | 3414/5000 [00:02<00:00, 1680.56it/s] #0, est. 2.50s: 72%|███████▏ | 3582/5000 [00:02<00:00, 1674.84it/s] #1, est. 1.70s: 100%|██████████| 5000/5000 [00:02<00:00, 2354.49it/s] #0, est. 2.50s: 100%|██████████| 5000/5000 [00:02<00:00, 1672.28it/s]
Hooks and callbacks
tqdm能夠很容易地支持回調/鉤子和手動更新。下面是urllib的一個例子:
urllib.urlretrieve documentation
若是存在，鉤子函數將在網絡鏈接創建時調用一次，以後在讀取每一個塊以後調用一次。
鉤子將傳遞三個參數：到目前爲止傳輸的塊數、塊大小(以字節爲單位)和文件的總大小。
報錯：
module 'urllib' has no attribute 'urlretrieve'
緣由是python2 與python3的urllib不一樣在與python3要加上.request，更改後爲：
import urllib, os from tqdm import tqdm class TqdmUpTo(tqdm): """Provides `update_to(n)` which uses `tqdm.update(delta_n)`.""" def update_to(self, b=1, bsize=1, tsize=None): """ b : int, optional Number of blocks transferred so far [default: 1]. bsize : int, optional Size of each block (in tqdm units) [default: 1]. tsize : int, optional Total size (in tqdm units). If [default: None] remains unchanged. """ if tsize is not None: self.total = tsize self.update(b * bsize - self.n) # will also set self.n = b * bsize eg_link = "https://caspersci.uk.to/matryoshka.zip" with TqdmUpTo(unit='B', unit_scale=True, miniters=1, desc=eg_link.split('/')[-1]) as t: # all optional kwargs urllib.request.urlretrieve(eg_link, filename=os.devnull, reporthook=t.update_to, data=None)
最終爲：
matryoshka.zip: 262kB [00:03, 77.4kB/s]
靈感來自 twine#242。對examples/tqdm_wget.py進行函數替代。
當迭代速度存在較大差別時，建議使用miniter =1(例如，經過不完整的鏈接下載文件)。
Pandas Integration
受廣大要求求，咱們增長了對panda的支持——下面是DataFrame.progress_apply和DataFrameGroupBy.progress_apply的一個例子：
出錯：
ModuleNotFoundError: No module named 'pandas'
解決：
(deeplearning) userdeMacBook-Pro:~ user$ conda install -n deeplearning pandas
更改後例子：
import pandas as pd import numpy as np from tqdm import tqdm df = pd.DataFrame(np.random.randint(0, 100, (100000, 6))) # Register `pandas.progress_apply` and `pandas.Series.map_apply` with `tqdm` # (can use `tqdm_gui`, `tqdm_notebook`, optional kwargs, etc.) tqdm.pandas(desc="my bar!") # Now you can use `progress_apply` instead of `apply` # and `progress_map` instead of `map` df.progress_apply(lambda x: x**2) # can also groupby: # df.groupby(0).progress_apply(lambda x: x**2)
返回：
/anaconda3/envs/deeplearning/bin/src/tqdm/tqdm/_tqdm.py:634: FutureWarning: The Panel class is removed from pandas. Accessing it from the top-level namespace will also be removed in the next version from pandas import Panel my bar!: 100%|██████████| 6/6 [00:00<00:00, 265.34it/s]
數據大小爲(100000, 6)，從[0,100)中隨機取值對其賦初值，而後求平方：
若是您對它的工做原理(以及如何爲您本身的回調修改它)感興趣，請參閱examples文件夾或導入模塊並運行help()。
IPython/Jupyter Integration
IPython/Jupyter經過tqdm_notebook子模塊支持:
出錯：
IntProgress not found. Please update jupyter and ipywidgets.
解決參考https://ipywidgets.readthedocs.io/en/stable/user_install.html
個人操做爲：
(deeplearning) userdeMacBook-Pro:~ user$ conda install -n deeplearning ipywidgets
例子：
from tqdm import tnrange, tqdm_notebook from time import sleep for i in tnrange(3, desc='1st loop'): for j in tqdm_notebook(range(100), desc='2nd loop'): sleep(0.01)
返回：
除了tqdm特性外，子模塊還提供了一個本機Jupyter小部件(兼容IPython v1-v4和Jupyter)，徹底工做的嵌套條和顏色提示(藍色:normal、綠色:completed、紅色:error/interrupt、淡藍色:no ETA);以下顯示
from tqdm import tnrange, tqdm_notebook from time import sleep for i in tqdm_notebook(range(3), desc='1st loop'): for j in tqdm_notebook(range(100), desc='2nd loop'): sleep(0.01)
過程爲：
最終爲：
若是添加參數leave=True:
from tqdm import tnrange, tqdm_notebook from time import sleep for i in tqdm_notebook(range(3), desc='1st loop'): for j in tqdm_notebook(range(100), desc='2nd loop', leave=False): sleep(0.01)
則最後只會留下外層嵌套的結果：
若是中間點擊中止按鈕，則標明紅色：
tqdm也能夠經過使用autonotebook子模塊自動選擇控制檯或筆記本版本:
from tqdm.autonotebook import tqdm tqdm.pandas()
返回：
/anaconda3/envs/deeplearning/bin/src/tqdm/tqdm/autonotebook/__init__.py:18: TqdmExperimentalWarning: Using `tqdm.autonotebook.tqdm` in notebook mode. Use `tqdm.tqdm` instead to force console mode (e.g. in jupyter console) " (e.g. in jupyter console)", TqdmExperimentalWarning)
注意，若是運行在一個筆記本上，這將發出tqdmexperimental警告，由於其不可能區分jupyter notebook和jupyter console。使用auto而不是autonotebook來抑制這個警告。
Custom Integration
能夠繼承tqdm來建立自定義回調(如上面的TqdmUpTo示例)或自定義前端(例如GUIs，如筆記本或繪圖包)。在後一種狀況下要作的有:
- 在def __init__()中調用super().__init__(..., gui=True)來要禁用終端status_printer建立。
- 重定義close(), clear(), display()三個函數
考慮重載display()來使用self.frontend(** .format_dict)而不是self.sp(repr(self))。
Dynamic Monitor/Meter
你能夠用tqdm做爲一個非單調增加的meter。這多是由於n減小(例如CPU使用監視器)或total更改。
一個例子是遞歸搜索文件。total是目前找到的對象數量，n是文件(而不是文件夾)的對象數量:
from tqdm import tqdm import os.path def find_files_recursively(path, show_progress=True): files = [] # total=1 assumes `path` is a file t = tqdm(total=1, unit="file", disable=not show_progress) if not os.path.exists(path): raise IOError("Cannot find:" + path) def append_found_file(f): files.append(f) t.update() def list_found_dir(path): """returns os.listdir(path) assuming os.path.isdir(path)""" listing = os.listdir(path) # subtract 1 since a "file" we found was actually this directory t.total += len(listing) - 1 # fancy way to give info without forcing a refresh t.set_postfix(dir=path[-10:], refresh=False) t.update(0) # may trigger a refresh return listing def recursively_search(path): if os.path.isdir(path): for f in list_found_dir(path): recursively_search(os.path.join(path, f)) else: append_found_file(path) recursively_search(path) t.set_postfix(dir=path) t.close() return files
使用update(0)是讓tqdm決定什麼時候觸發顯示刷新以免控制檯垃圾信息的一種簡便方法。
Writing messages
這是一項正在進行的工做(見#737)。
因爲tqdm使用簡單的打印機制來顯示進度條，因此在打開進度條時，不該該在終端中使用print()編寫任何消息。
爲了在終端中寫入消息而不與tqdm bar顯示發生衝突，提供了.write()方法:
from tqdm import tqdm, trange from time import sleep bar = trange(10) for i in bar: # Print using tqdm class method .write() sleep(0.1) if not (i % 3): tqdm.write("Done task %i" % i) # Can also use bar.write()
返回：
20%|██ | 2/10 [00:00<00:00, 9.48it/s] Done task 0 50%|█████ | 5/10 [00:00<00:00, 9.53it/s] Done task 3 80%|████████ | 8/10 [00:00<00:00, 9.50it/s] Done task 6 100%|██████████| 10/10 [00:01<00:00, 9.56it/s] Done task 9
默認狀況下，這將打印到標準輸出sys.stdout。可是您可使用file參數指定任何相似文件的對象。例如，這能夠用於將寫入的消息重定向到日誌文件或類。
Redirecting writing
若是使用一個能夠將消息打印到控制檯的庫，那麼用tqdm.write()替換print()來編輯庫多是不可取的。在這種狀況下，重定向sys.stdout到tqdm.write()是一個選擇。
重定向sys.stdout，建立一個相似於文件的類，該類將向tqdm.write()寫入任何輸入字符串，並提供參數file=sys.stdout, dynamic_ncols = True。
一個可重用的規範示例以下:
from time import sleep import contextlib import sys from tqdm import tqdm class DummyTqdmFile(object): """Dummy file-like that will write to tqdm""" file = None def __init__(self, file): self.file = file def write(self, x): # Avoid print() second call (useless \n) if len(x.rstrip()) > 0: tqdm.write(x, file=self.file) def flush(self): return getattr(self.file, "flush", lambda: None)() @contextlib.contextmanager def std_out_err_redirect_tqdm(): orig_out_err = sys.stdout, sys.stderr try: sys.stdout, sys.stderr = map(DummyTqdmFile, orig_out_err) yield orig_out_err[0] # Relay exceptions except Exception as exc: raise exc # Always restore sys.stdout/err if necessary finally: sys.stdout, sys.stderr = orig_out_err def some_fun(i): print("Fee, fi, fo,".split()[i]) # Redirect stdout to tqdm.write() (don't forget the `as save_stdout`) with std_out_err_redirect_tqdm() as orig_stdout: # tqdm needs the original stdout # and dynamic_ncols=True to autodetect console width for i in tqdm(range(3), file=orig_stdout, dynamic_ncols=True): sleep(.5) some_fun(i) # After the `with`, printing is restored print("Done!")
返回：
Fee, fi, fo, 100%|██████████| 3/3 [00:01<00:00, 1.97it/s] Done!
Monitoring thread, intervals and miniters
tqdm實現了一些技巧來提升效率和減小開銷。
- 避免沒必要要的頻繁刷新:mininterval定義每次刷新之間要等待多長時間。tqdm老是在後臺更新，但它只顯示每分鐘一次。
- 減小檢查系統時鐘/時間的次數。
- mininterval比miniter更易於配置。一個聰明的調整系統dynamic_miniter將自動調整miniter到適合時間mininterval的迭代量。本質上，tqdm將在沒有實際檢查時間的狀況下檢查是否須要打印。這種行爲仍然能夠經過手動設置miniter來繞過。
然而，考慮一個結合了快速和緩慢迭代的案例。通過幾回快速迭代以後，dynamic_miniter將把miniter設置爲一個很大的數字。當迭代速率隨後減慢時，miniter將保持較大的值，從而下降顯示更新頻率。爲了解決這個問題:
- maxinterval定義顯示刷新之間的最大時間間隔。併發監視線程檢查過時的更新，並在必要時強制執行更新。
監視線程不該該有明顯的開銷，而且默認狀況下保證至少每10秒更新一次。能夠經過設置任何tqdm實例的monitor_interval(即t = tqdm.tqdm(…); t.monitor_interval = 2)直接更改此值。經過設置tqdm.tqdm.monitor_interval = 0能夠在實例化任何tqdm bar以前在應用程序範圍內禁用監視器線程。
相關文章
- 1. python-tqdm進度條
- 2. Python: tqdm 進度條顯示
- 3. tqdm-python實現進度條
- 4. Python tqdm顯示進度條
- 5. Python進度條模塊tqdm實現任務進度可視化
- 6. Python的Tqdm模塊——進度條配置
- 7. Python第三方進度條庫tqdm
- 8. tqdm顯示進度條的時候，輸出多行進度條
- 9. Cli實現一個進度條
- 10. 快速開發一個PHP擴展
- 更多相關文章...
- • Kotlin 擴展 - Kotlin 教程
- • Swift 擴展 - Swift 教程
- • 適用於PHP初學者的學習線路和建議
- • Kotlin學習（一）基本語法
相關標籤/搜索
每日一句
-
每一个你不满意的现在，都有一个你没有努力的曾经。
歡迎關注本站公眾號,獲取更多信息
相關文章