[WIP] Fix π parsing in complex values for tests

[Copilot]

Thanks for asking me to work on this. I will get started on it and keep this PR's description up to date as I form a plan and make progress.

Original prompt

Fix π Parsing in Complex Values for PR #3

Problem

The PR #3 has test failures like:

FAILED array_api_tests/test_special_cases.py::test_unary[acosh((real(x_i) is +0 or real(x_i) == -0) and imag(x_i) is +0) -> +0 + πj/2] - AssertionError: out=1.5707963267948966j, but should be +0 + πj/2 [acosh()]

The actual output 1.5707963267948966j is numerically correct (it equals π/2), but the test is failing because parse_complex_value() in array_api_tests/test_special_cases.py (lines 495-523) doesn't correctly handle mathematical expressions like πj/2 in complex values.

Root Cause

The issue is in the parse_complex_value() function and r_complex_value regex (line 697-699):

1. The regex r_complex_value = re.compile(r"([+-]?)([^\s]+)\s*([+-])\s*([^\s]+)\s*j") captures the components but doesn't handle expressions like πj/2 which have a division
2. The parse_complex_value() function directly passes captured strings to parse_value(), but for πj/2 it captures π as the imaginary part string, missing the /2 part

Solution Required

Update the regex and parsing logic in array_api_tests/test_special_cases.py to:

1. Update r_complex_value regex (line 697-699) to properly capture complex numbers with mathematical expressions in the imaginary part, e.g.:

   • +0 + πj/2 should capture real=+0, imag=π/2
   • +0 + 0j should capture real=+0, imag=0
   • NaN + NaN j should capture real=NaN, imag=NaN

2. The regex should handle:

   • Optional spaces around operators
   • Mathematical expressions with / for division
   • The j can be attached or have a space before it

3. The key insight: The existing parse_value() function (lines 138-167) already handles π expressions correctly including π/2. We just need to properly extract and pass the imaginary coefficient to it.

Suggested Fix

Replace the r_complex_value regex around line 697 with:

r_complex_value = re.compile(
    r"([+-]?)([^\s]+)\s*([+-])\s*([\dπ/]+)\s*j"
)

This allows capturing digits, π, and / in the imaginary part, which can then be passed to parse_value() which already knows how to evaluate π/2.

Test Cases to Verify

After the fix, these should parse correctly:

• +0 + πj/2 → complex(0.0, 1.5707963267948966)
• +0 + 0j → complex(0.0, 0.0)
• NaN + NaN j → complex(nan, nan)

Files to Modify

• array_api_tests/test_special_cases.py - Update the r_complex_value regex pattern (around line 697-699)

The existing parse_complex_value() function should work correctly once the regex properly captures the full imaginary expression.

This pull request was created from Copilot chat.

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.

[ev-br]

this should have been a part of #3, so continuing there.