This extension is inspired by "Python Quick Print". "Python Quick Print" can quickly print out variables like print(f"==>> a: {a}") on the terminal by using keyboard shortcut Ctrl+Shift+L, while this extension "Quick Python Print" adds some more features:
✨ You can put the cursor at target line, the extension will automatically recognize the variable in the line.
✨ If did not recognize any variable in current line, it will just insert simply print() and move the cursor inside the bracket.
✨ Press Ctrl+Shift+O to print attribute of variable, here is tensor shape by default, e.g. print(f"==>> a.shape: {a.shape}").
✨ Press Ctrl+Shift+T to print function of variable, here is type by default, e.g. print(f"==>> type(a): {type(a)}").
✨ Press Ctrl+Shift+; to insert print("".center(50, "-")) used for printing delimiter line.
✨ Press Ctrl+Shift+/ to comment out or uncomment all print statements inside the selection or in the file.
✨ Press Ctrl+Shift+R to delete all print statements inside the selection or in the file.
✨ Support users to define customized print function like logger.debug, pprint other than origianl print.
✨ Support users to define customized prefix and suffix of print content flexibly in extension settings.
✨ Support printing multiple variables separated by , in one line and multiple variables in multiple lines together.
✨ Able to output colored text in terminal by using python built-in package: termcolor.
If you find a bug or have feature request, you can check out the ChangeLog to see if this bug or feature had been fixed or added in the latest version, if not, please issue it in this extension's GitHub repo.
When I am learning deep learning model, I often want to know how the tensor shape changes along the way, I think this will help me understand how the deep learning model works. Take pytorh for example, these operations view，slice，concat，permute，conv，linear etc all will change the tensor shape. Fortunately, many deep learning framework all have .shape attribute of tensor(as far as I know, there are pytorch, numpy, tensorflow, paddlepaddle, oneflow), so this makes the extension be useful for different deep learning framework users. Afterwards, I add more features to make it also useful for other python developers. I know using debug is a good alternative to get the tensor shape, but I think the print function is still useful in some cases.
Before installation, make sure your vscode version ≥ 1.57.0.
Because since version 0.2.0 which use the "formatted string literals" feature introduced since python 3.6, if your python version < 3.6, you can install previous version (e.g. 0.1.9) by:
This extension is available in the Visual Studio Code Extension Marketplace, you can search "Quick-Python-Print" and install it.
Introduction video (Chinese speech with English subtitles): Youtube | Bilibili (When I made this video, it was version 0.1.1, and there have been many changes in the current version).
This extension only activates within .py and .ipynb files.
For Mac user, the Ctrl should be Cmd.
Note : From version 0.2.0, the inserted code will be a little different, the inserted code will be print(f"==>> a: {a}") instead of print("==>> a: ", a).
Ctrl+Shift+L
Select the variable, then press Ctrl+Shift+L, it will insert print statement at next line:
You can also just put the cursor at the target line, then the extension will automatically recognize the variable that before =, +=, -=, /=, //=, *=, %=:
This extension can also handle some little complicated cases, such as unpacked variables separated by ,, and print multiple variables in multiple lines together:
If you didn't select variable or the extension can't recognize variable, it will simply insert print() and move the cursor inside it.
If the selected content is exactly the content of the entire line, it will treat the selected content as a variable and insert code at the current line :
If the extension recognizes the variable wrongly, you can manually select the variable, or report this issue in GitHub repo.
Ctrl+Shift+O
Ctl+Shift+O is similar to Ctrl+Shift+L except that it will print tensor shape by default:
Ctrl+Shift+T
Ctl+Shift+T is similar to above except that it will print type of variable by default:
Ctrl+Shift+/
Comment out or uncomment the print statement in the selected scope or in the python file:
Ctrl+Shift+Up
Comment out all print statement above the cursor in the python file:
Ctrl+Shift+Down
Comment out all print statement below the cursor in the python file:
Ctrl+Shift+;
Insert a print statement that will print 50 - to form a delimiter like --------------------------------------------------:
Ctrl+Shift+R
Delete all the print statement(include those commented) in the python file or in the selected scope:
Compatibility with different framework
Many deep learning framework all have .shape attribute of tensor:
If any of these shortcuts conflicts with existing shortcut, you may change it in the Keyboard Shortcuts Setting: Press Cmd+P or Ctrl+P and type in >Open Keyboard Shortcuts. Search for the conflict shortcut, then you will find those shortcut using the same keys, then right click one of them, modify or delete keybinding.
This extension has following settings:
You can go to the Extension Settings to modify them to suit your preferences. After modification, it should take effect immediately.
-
How to use with loguru logger or logging logger ?
- see this issue.
-
How to use with pprint ?
- see this issue.
To color the output text in terminal, you need to do these things:
- Go to
Extension Settings, check the5.enable-colored-output-textto be true. And you can select the color you like from the drop-down list. - Add
from termcolor import coloredin the python file - Now Press
Ctrl+Shift+LorCtrl+Shift+OorCtrl+Shift+Twill insert the print statement that can color output text. - Run Python File in Terminal.
I also make a pypi package : printensor to uppack tensors inside list, tuple, dict, generator, then print their tensor shape. After installing and import, you can replace print( with prints( to intergrate it with this extension.
This extension can not handle tensor that cross multiple lines, for example:
a = torch.tensor([[1.0, 2.0, 3.0],
[4.0, 5.0, 6.0]])
You can use Alt + down to move down the inserted print statement.
This extension uses Python 3 syntax, if you're using Python 2 print isn't a function. You can import this behavior from __future__:
from __future__ import print_function
If you find this extension can help you save time, and willing to donate me, I would be very grateful ! ❤
MIT License