fix: multi-line and conditional sec blocks in analytical models

[Siel]

Summary

Fix sec block failing with "Result must be length 1" when it contains multiple statements or if-conditionals in analytical (template-based) models.

Example that was failing

mod <- PM_model$new(
  pri = list(
    Ka = ab(0.5, 6),
    Ke = ab(0.1, 1.5),
    v0 = ab(25, 120)
  ),
  cov = list(
    gender = interp("none"),
    wt = interp()
  ),
  sec = function() {
    v = v0 * wt
    if (gender < 1) { v = v0 * 0.8 }
  },
  eqn = function(){ two_comp_bolus },
  out = function(){ Y[1] = X[2]/v },
  err = list(proportional(5, c(0.1, 0.15, 0, 0)))
)

This also failed with multiple assignments (no conditionals):

sec = function() {
  v = v0 * wt
  cl = cl0 * wt
}

Root cause

deparse() on multi-line R expressions (e.g. if-blocks) returns a character vector longer than 1, which broke map_chr() calls that assumed single-element output in the transpiler and template detection code.

Changes

• model_transpiler.R: collapse deparse() output in transpile_analytic_eqn()
• mod_lib.R: same fix in get_found_model() for template detection
• test-Examples.R: regression test for multi-line sec with conditionals

Thanks to Hector Julian Perez Madrid for finding this bug.

[Copilot]

Pull request overview

This PR fixes a failure in analytical (template-based) model handling where deparse() could return a multi-element character vector for multi-line expressions, which then broke purrr::map_chr()/template-detection logic (e.g., “Result must be length 1”).

Changes:

• Collapse multi-line deparse() output into a single string when scanning EQN statements in transpile_analytic_eqn().
• Apply the same deparse() collapsing approach in get_found_model() to make template detection robust to multi-line expressions.
• Add a regression test covering an analytical model with a multi-line sec block containing an if conditional.
• (Unrelated) Large-scale formatting changes in PMutilities.R.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 3 comments.

File	Description
R/model_transpiler.R	Collapses deparse() output during analytic EQN parsing to prevent map_chr() length errors.
R/mod_lib.R	Collapses deparse() output during template detection; switches to map_lgl() for clearer boolean results.
tests/testthat/test-Examples.R	Adds regression test for a multi-line sec block with conditionals in an analytical model.
R/PMutilities.R	Contains extensive formatting-only changes unrelated to the stated PR purpose.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

The new regression test doesn’t currently exercise the code paths changed in get_found_model()/transpile_analytic_eqn() (i.e., deparse() returning a multi-element character vector for a multi-line expression inside the EQN/template-detection scan). As written, the EQN block only contains a single symbol, so the pre-fix code would likely not fail here. Consider adjusting/adding a case where the EQN block contains an if-block or another multi-line statement alongside the template name, so the test would fail without this PR and pass with it.

[Copilot]

Since the bug is specific to analytical/template-based models, it would be good for this test to assert that the model was detected as Analytical (e.g., via mod$model_list$type) before checking sec contents. Otherwise, a future change that accidentally routes this model through the ODE path could let the test pass while no longer covering the intended behavior.

[Copilot]

This PR includes a very large reformatting-only change set in PMutilities.R that isn’t mentioned in the PR description and appears unrelated to the analytical SEC/EQN deparse fix. This makes the review noisy and increases merge-conflict risk; consider reverting these formatting changes or splitting them into a separate PR focused on styling/formatting.

[mhovd]

Apparently the CI is breaking because of spindle ?

Compiling spindle v0.2.5
error[E0597]: `task` does not live long enough
   --> /Users/runner/.cargo/registry/src/index.crates.io-1949cf8c6b5b557f/spindle-0.2.5/src/lib.rs:597:10
    |
594 |                     let task = AssertUnwindSafe(task);
    |                         ---- binding `task` declared here
595 |
596 |                     let task = |thread_id: usize, panic_slot: &AtomicPtr<PanicLoad>| match std::panic::catch_unwind(|| {
    |                                ----------------------------------------------------- value captured here
597 |                         ({ task }.0)(thread_id);
    |                            ^^^^ borrowed value does not live long enough
...
609 |                     let mut task = (&raw const task) as *const TaskInner;
    |                                                         ---------------- type annotation requires that `task` is borrowed for `'static`
...
635 |                 }
    |                 - `task` dropped here while still borrowed
    ```
    
    I will check if I can figure out why.

[mhovd]

We can probably revise this list, many of these are no longer really reserved.

[mhovd]

But that doesn't have to be in this PR.