py/parse: Fold anonymous string objects. - #19598

Codecov Report

❌ Patch coverage is 94.28571% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 98.55%. Comparing base (aee96a0) to head (7b4a4a1).

@@            Coverage Diff             @@
##           master   #19598      +/-   ##
==========================================
- Coverage   98.55%   98.55%   -0.01%     
==========================================
  Files         182      182              
  Lines       23316    23351      +35     
  Branches        5        5              
==========================================
+ Hits        22980    23013      +33     
- Misses        335      337       +2     
  Partials        1        1              

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

Code size report:

Reference:  docs/esp32/quickref: Add config for ESP32-WROOM-32UE with KSZ8863 LAN. [aee96a0]
Comparison: py/parse: Fold anonymous string objects. [merge of 7b4a4a1]
  mpy-cross:  +352 +0.092% 
   bare-arm:    +0 +0.000% 
minimal x86:    +0 +0.000% 
   unix x64:  +352 +0.041% standard
      stm32:  +216 +0.053% PYBV10
      esp32:  +308 +0.017% ESP32_GENERIC
     mimxrt:  +224 +0.057% TEENSY40
        rp2:  +256 +0.027% RPI_PICO_W
       samd:  +224 +0.081% ADAFRUIT_ITSYBITSY_M4_EXPRESS
  qemu rv32:  +240 +0.052% VIRT_RV32

Octoprobe PR report

• Duration: 1h 42min, 82487 tests run, 74ms/test
• Tested ports: esp32 esp8266 mimxrt nrf rp2 samd stm32
• Summary Report , Logdirectory

but is it needed?
after all Python syntax allows to write this without the + .

# fmt: off
s1 = "Hello World"
s2 = "Hello" " " "World"
s3 = "Hello" \
     " " \
     "World"
s4 = (
    "Hello"
     " " 
     "World"
)

and they are all bytecode equivalent.

mpy_source_file: /tmp/example.mpy
source_file: /tmp/example.py
header: 4d:06:00:1f
arch: NONE
qstr_table[6]:
    /tmp/example.py
    <module>
    s1
    s2
    s3
    s4
obj_table: ['Hello World']
simple_name: <module>
  raw bytecode: 26 00:0c:01:40:24:24:64:20:23:00:16:02:23:00:16:03:23:00:16:04:23:00:16:05:51:63
  prelude: (1, 0, 0, 0, 0, 0)
  args: []
  line info: 40:24:24:64:20
  23:00       LOAD_CONST_OBJ 'Hello World'
  16:02       STORE_NAME s1
  23:00       LOAD_CONST_OBJ 'Hello World'
  16:03       STORE_NAME s2
  23:00       LOAD_CONST_OBJ 'Hello World'
  16:04       STORE_NAME s3
  23:00       LOAD_CONST_OBJ 'Hello World'
  16:05       STORE_NAME s4
  51          LOAD_CONST_NONE 
  63          RETURN_VALUE 
  children: []

but is it needed? after all Python syntax allows to write this without the +

That's why this is marked as draft :)

I've got loads of code with strings wrapped using + instead of an escaped continuation, I'm not sure if it was PyCharm's (or the default IntelliJ Python language support) default method for breaking up long lines back then, but it was pressing enough for me to actually patch the parser. I've also moved on from IntelliJ stuff, but the patch still remained.

Now, I can probably make this more palatable by sneaking in const() string folding (ie. A = const("hello"); print(A + ", world") becoming A = const("hello"); print("hello, world")), but that will maybe have a larger impact on footprint.

On second thought, maybe this is better to stay out of the main source tree.

Sorry :(

Using adjacent string concatenation is IMO a good solution, rather than + operator.

If anything this could be addressed with a small addition to the docs, saying that the former is optimised but the latter (+) is not.

In the meantime I've added that to the flake8 plugin: https://codeberg.org/agatti/flake8-micropython/commit/cb06cb3084ccf11a38a33667e9a153f3eb040291.

Docs-wise, where should that paragraph be added to, docs/develop/optimizations.rst or docs/reference/speed_python.rst?

Docs-wise, where should that paragraph be added to

I think docs/reference/speed_python.rst is best, maybe under the "RAM allocation" section, or "Strings vs bytes".