pyramid formats are introduced

[Artur-man]

No description provided.

[Artur-man]

@claude review

[claude]

Claude finished @Artur-man's task in 1m 53s —— View job

---

Claude's Review

• Gather context (diff, repo guidelines)
• Understand the request (code review)
• Review changed files
• Post review feedback

Summary

Good change — centralizing pyramid/BioFormats extensions into a single .PYRAMID_FORMATS vector (R/utils.R:207-220) is much easier to maintain and extend than the old inline regex in R/ImageArray.R.

Issues found

1. Unescaped . in extensions creates a false-positive-prone regex — R/ImageArray.R:495-496

   pyramid_formats <- paste(paste0(.PYRAMID_FORMATS, "$"), collapse = "|")
   if (grepl(pyramid_formats, image)) {

   Each entry in .PYRAMID_FORMATS (e.g. ".ome.tif") contains a literal . that grepl treats as "match any character" since the string isn't escaped or passed with fixed = TRUE. That means a filename like genome.tif or sample_xafi will also match .ome.tif$ / .afi$ respectively, since . in the pattern absorbs any single character (e.g. n before ome.tif, or x before afi). This bug pre-dates this PR (the original inline pattern had the same issue for .ome.tiff/.ome.tif/.qptiff/.qptif), but this PR both extends it to more extensions (.svs, .afi, .ndpi, .ndpis, .ome.tf2, .ome.tf8, .ome.btf) and centralizes the list — a good opportunity to fix it now, e.g.:

   pyramid_formats <- paste0(gsub("\\.", "\\\\.", .PYRAMID_FORMATS), "$", collapse = "|")

   or build the pattern with fixed = TRUE per-extension and combine with any(vapply(...)).

2. Case sensitivity — some of the newly added formats (e.g. .SVS, .NDPI) commonly appear with uppercase extensions on Windows-originated scanner exports. grepl() here is case-sensitive by default; consider ignore.case = TRUE unless this is intentionally strict.

3. Vignette typo — vignettes/ImageArray.Rmd:244: "(e.g. qptiff, svs, )." has a dangling comma before the closing parenthesis (should be "(e.g. qptiff, svs).").

4. Uncertain comments — R/utils.R:218-219 has # CODEX ? next to .qptiff/.qptif, suggesting the format attribution wasn't confirmed. Worth double-checking (qptiff/qptif are actually PerkinElmer/Akoya Vectra "QPTIFF" format, used by CODEX/PhenoCycler pipelines but not CODEX-specific) and tidying the comment before merge.

Fix this →

No changes were implemented (review only, as requested).