Text Editor: correct number highlighting #98971

New Issue

jon denning · 2022-06-18T13:22:17+02:00

jon denning commented

2022-06-18 13:22:17 +02:00

Blender's tokenizer does not match CPython's (source ) when it comes to tokenizing numbers.

This task is related to D14739 and #72964. This is a relatively easy issue to fix, but (...thinking out loud on topics WAY outside the scope of this task...) I wonder if it makes sense to take another look at Blender's tokenizer and parsing code. Right now, Blender only tokenizes and then marks up the text in order to offer some basic highlighting for Python only. Also, many of the text editor features (ex: Find) are extremely simple. Could Blender's community benefit from either 1) a more extensive built-in editor or 2) improved integration features for using an external editor? There are a few submissions on Right-Click Select that could be solved/added with an improved parser. Perhaps this would be a good GSoC project?

The following is some example code showing where the two tokenizers agree and where they disagree, and below that is a screenshot showing how the text editor renders the example code.

# https://github.com/python/cpython/blob/3.10/Parser/tokenizer.c

### the good ###

# valid syntax, correctly highlighted

1000    # standard int
1.000   # standard float
1_000   # using _ as separator
1_0_0
0x300   # hex
0x4aA
0o123   # octal
0O123   # (can use capital O)
0b000   # binary
1e400   # scientific notation (exponent)
1E400
1 + 2j  # imaginary numbers
1 - 2j
0x1e20  # hex with scientific notation

# invalid syntax, correctly highlighted

0oo0123  # double o
0bb01    # double b
0xx12    # double x
0x42AXY  # hex with non-hex digits
1e       # incomplete scientific notation (no exponent)
0x1e0x1  # exponent must be in decimal (not hex)


### the bad ###

# valid syntax, wrong highlighting
1e-40   # - is op
1e+42   # e, + is op
1.

# invalid syntax, wrong highlighting
1_     # end with _
1__00  # double _
2jj    # double j
1.0.0  # extra .
1e2e2  # extra e
0B1e2  # cannot have e with 0b
0o1e2  # cannot have e with 0o

Blender's tokenizer does not match CPython's ([source ](https://github.com/python/cpython/blob/3.10/Parser/tokenizer.c)) when it comes to tokenizing numbers. This task is related to [D14739](https://archive.blender.org/developer/D14739) and #72964. This is a relatively easy issue to fix, but (...thinking out loud on topics WAY outside the scope of this task...) I wonder if it makes sense to take another look at Blender's tokenizer and parsing code. Right now, Blender only tokenizes and then marks up the text in order to offer some basic highlighting for Python only. Also, many of the text editor features (ex: Find) are extremely simple. Could Blender's community benefit from either 1) a more extensive built-in editor or 2) improved integration features for using an external editor? There are a few submissions on [Right-Click Select ](https://blender.community/c/rightclickselect) that could be solved/added with an improved parser. Perhaps this would be a good GSoC project? The following is some example code showing where the two tokenizers agree and where they disagree, and below that is a screenshot showing how the text editor renders the example code. ``` # https://github.com/python/cpython/blob/3.10/Parser/tokenizer.c ### the good ### # valid syntax, correctly highlighted 1000 # standard int 1.000 # standard float 1_000 # using _ as separator 1_0_0 0x300 # hex 0x4aA 0o123 # octal 0O123 # (can use capital O) 0b000 # binary 1e400 # scientific notation (exponent) 1E400 1 + 2j # imaginary numbers 1 - 2j 0x1e20 # hex with scientific notation # invalid syntax, correctly highlighted 0oo0123 # double o 0bb01 # double b 0xx12 # double x 0x42AXY # hex with non-hex digits 1e # incomplete scientific notation (no exponent) 0x1e0x1 # exponent must be in decimal (not hex) ### the bad ### # valid syntax, wrong highlighting 1e-40 # - is op 1e+42 # e, + is op 1. # invalid syntax, wrong highlighting 1_ # end with _ 1__00 # double _ 2jj # double j 1.0.0 # extra . 1e2e2 # extra e 0B1e2 # cannot have e with 0b 0o1e2 # cannot have e with 0o ``` ![image.png](https://archive.blender.org/developer/F13188564/image.png)

jon denning commented

2022-06-18 13:22:17 +02:00

Added subscriber: @gfxcoder

Philipp Oeser removed the

Download

What's New

Blender Studio

Manual

Developers Blog

Documentation

Benchmark

Blender Conference

Development Fund

One-time Donations

Text Editor: correct number highlighting #98971