py/modmicropython: Expose repl_autocomplete as python function. - #17011

Code size report:

Reference:  esp32/boards/SEEED_XIAO_ESP32C6: Add new XIAO board definition. [2dc2e30]
Comparison: py/modmicropython: Expose repl_autocomplete as python function. [merge of aeda0fa]
  mpy-cross:    +0 +0.000% 
   bare-arm:    +0 +0.000% 
minimal x86:    +0 +0.000% 
   unix x64:  +296 +0.035% standard[incl +32(data)]
      stm32:  +108 +0.027% PYBV10
      esp32:  +124 +0.007% ESP32_GENERIC[incl +48(data)]
     mimxrt:   +96 +0.025% TEENSY40
        rp2:  +104 +0.011% RPI_PICO_W
       samd:  +104 +0.038% ADAFRUIT_ITSYBITSY_M4_EXPRESS
  qemu rv32:  +116 +0.025% VIRT_RV32

Adding autocomplete for aiorepl is very desirable, and this is looking promising! Here are some of the tests I ran:

>>>import micropython
>>>micropython.repl_autocomplete("impo")  # Should complete 'import'
'rt '
>>> class Foo:
...     def _bar():
...         pass
...     def alpha():
...         pass
>>> micropython.repl_autocomplete("f = Fo") # Should complete Foo
'o'
>>>f = Foo()
>>> micropython.repl_autocomplete("f.") # Should complete alpha (ignoring _bar)
'alpha'

Note that this is correct and consistent with MicroPython's built-in tab completion - but differs in some cases to CPython, at least at v3.12. CPython will also supply parentheses, ie it would return 'o()' to complete f = Fo and 'alpha(' for f.. That's not a problem with this PR, but should be resolved in mp_repl_autocomplete.

I also tested a module with private members and they were also correctly filtered out.

Looks good!

but differs in some cases to CPython, at least at v3.12.

Perhaps that should just be documented as such. I'd be fine with consistent behavior with the MicroPython repl.

Ok I'm really not sure why the unit tests are still failing on some builds; unix standard in particular. The test passes for me locally with the same build.
Are local variable names somehow handled differently in CI that they're not visible to the auto complete function?
I only just found/ looked at the cmdline repl autocomplete unit test, I should rewrite the test for this module to use test cases more similar to that.

Updated the branch with a few fixes and an additional feature:

Fixed a type mismatch in repl_autocomplete — was using ssize_t (not available on MSVC, hence all the Windows CI failures) and <= 0 which incorrectly returned None for zero-length completions. Now uses size_t and compares against (size_t)(-1) matching how readline.c handles the same sentinel. The test also now uses repr() on outputs which avoids the trailing whitespace comparison issue that was likely causing the unix CI failures.

Also added micropython.stdio_mode_raw(enabled) in a second commit. This wraps the existing mp_hal_stdio_mode_raw() / mp_hal_stdio_mode_orig() HAL functions so Python code can control terminal mode directly — needed for aiorepl to properly manage terminal I/O without relying on the native REPL's implicit mode switching. Gated behind MICROPY_PY_MICROPYTHON_STDIO_RAW with a default of 0 in mpconfig.h, enabled on unix.

Added docs entries for both functions in micropython.rst.

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 98.46%. Comparing base (2dc2e30) to head (aeda0fa).
⚠️ Report is 491 commits behind head on master.

@@           Coverage Diff           @@
##           master   #17011   +/-   ##
=======================================
  Coverage   98.46%   98.46%           
=======================================
  Files         176      176           
  Lines       22784    22793    +9     
=======================================
+ Hits        22435    22444    +9     
  Misses        349      349           

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