add package logo

README.Rmd

@@ -14,7 +14,7 @@ knitr::opts_chunk$set(
 options(balancing.quiet = TRUE)
 ```
 
-# balancing
+# balancing <a href="https://r-causal.github.io/balancing/"><img src="man/figures/logo.png" align="right" height="138" alt="balancing website" /></a>
 
 <!-- badges: start -->
 [![R-CMD-check](https://github.com/r-causal/balancing/actions/workflows/R-CMD-check.yaml/badge.svg)](https://github.com/r-causal/balancing/actions/workflows/R-CMD-check.yaml)

README.md

@@ -1,7 +1,7 @@
 
 <!-- README.md is generated from README.Rmd. Please edit that file -->
 
-# balancing
+# balancing <a href="https://r-causal.github.io/balancing/"><img src="man/figures/logo.png" align="right" height="138" alt="balancing website" /></a>
 
 <!-- badges: start -->
 

data-raw/logo/.gitignore

@@ -0,0 +1 @@
+exports/

data-raw/logo/README.md

@@ -0,0 +1,13 @@
+# Package logo
+
+The text-free artwork was generated with the built-in image-generation tool.
+`prompt.txt` records the artwork prompt. `sticker.svg` adds editable typography
+and a border clipped to the 1.74 by 2 inch hexagon.
+
+Run `Rscript data-raw/logo/build.R` from the package root to export a 600 dpi
+print PNG and an SVG with outlined text to `exports/`, then install the README
+logo with `usethis::use_logo()`.
+
+The build requires the R packages base64enc, magick, rsvg, systemfonts, and
+usethis, plus Baskerville installed locally.
+The artwork, fonts, and print tools are not package runtime dependencies.

data-raw/logo/build.R

@@ -0,0 +1,48 @@
+# Run from the package root: Rscript data-raw/logo/build.R
+args <- commandArgs(trailingOnly = FALSE)
+script <- sub("^--file=", "", args[grepl("^--file=", args)])
+source_dir <- dirname(normalizePath(script))
+root <- normalizePath(file.path(source_dir, "../.."))
+svg <- paste(readLines(file.path(source_dir, "sticker.svg")), collapse = "\n")
+families <- regmatches(svg, gregexpr('font-family="[^"]+"', svg))[[1]]
+families <- unique(sub('font-family="([^"]+)"', "\\1", families))
+missing <- setdiff(families, systemfonts::system_fonts()$family)
+if (length(missing)) {
+  stop(
+    "Install the source fonts before rebuilding: ",
+    paste(missing, collapse = ", ")
+  )
+}
+art <- base64enc::dataURI(
+  file = file.path(source_dir, "artwork.png"),
+  mime = "image/png"
+)
+svg <- sub("artwork.png", art, svg, fixed = TRUE)
+exports <- file.path(source_dir, "exports")
+dir.create(exports, showWarnings = FALSE)
+print_png <- file.path(exports, "logo.png")
+raster <- rsvg::rsvg_png(charToRaw(svg), width = 1044, height = 1200)
+magick::image_write(
+  magick::image_read(raster),
+  path = print_png,
+  format = "png",
+  density = "600x600"
+)
+
+# Outline the text separately to avoid resampling the embedded artwork.
+text_only <- gsub("<image[^>]*/>", "", svg)
+text_only <- gsub("<rect[^>]*/>", "", text_only)
+text_only <- gsub('<polygon id="sticker-border"[^>]*/>', "", text_only)
+print_svg <- file.path(exports, "logo.svg")
+rsvg::rsvg_svg(charToRaw(text_only), print_svg, width = 1740, height = 2000)
+outlined <- paste(readLines(print_svg), collapse = "\n")
+inner <- sub("(?s).*?<svg[^>]*>", "", outlined, perl = TRUE)
+inner <- sub("</svg>\\s*$", "", inner, perl = TRUE)
+without_text <- gsub("(?s)<text[^>]*>.*?</text>", "", svg, perl = TRUE)
+combined <- sub("</svg>", paste0(inner, "</svg>"), without_text, fixed = TRUE)
+stopifnot(!grepl("<text[ >]", combined))
+writeLines(combined, print_svg)
+
+usethis::proj_set(root)
+options(usethis.overwrite = TRUE)
+usethis::use_logo(print_png)

data-raw/logo/prompt.txt

@@ -0,0 +1 @@
+Use case: stylized-concept. A full-bleed fine-art OIL PAINTING for the balancing R package sticker. Portrait 3:4, no typography anywhere. This must look like a beautiful physical painting, NOT a logo, icon, isolated cutout, clipart, vector, digital mascot or product render. Paint an intimate still life in a quiet artist's studio: an old narrow balance beam made of weathered wood rests perfectly level across a simple dark triangular fulcrum on a worn pale stone table. A squat antique brass calibration weight with a round lifting knob sits at each end, matched size and equal distance so it is truly balanced. Show the objects in subtle three-quarter perspective with convincing depth, but the balance is visibly level. The brass weights have irregular tarnish and warm reflected light, not shiny perfect gold. A large soft window-light shape enters from upper left across a muted smoky teal wall, casting a long painterly shadow behind the arrangement. Rich thick brush marks and palette-knife scrapes, lost and found edges, delicious ochre, blue-black, pale celadon, dirty cream and warm sienna. Like a refined 20th-century European still-life painting reproduced in an art book. Strong expressive composition, asymmetrical lighting and architectural shadows, absolutely no heavy uniform outlines. The image fills the whole rectangular canvas with painted wall and tabletop, NO transparent background, no empty white backdrop. Main balance occupies central width x12%-88%, y30%-63%; lower quarter is painted quiet dark teal tabletop/shadow for real lettering later. No text, numbers, labels, watermark, border, hexagon, writing or decorative symbols.

pkgdown/favicon/site.webmanifest

@@ -0,0 +1,21 @@
+{
+  "name": "",
+  "short_name": "",
+  "icons": [
+    {
+      "src": "/web-app-manifest-192x192.png",
+      "sizes": "192x192",
+      "type": "image/png",
+      "purpose": "maskable"
+    },
+    {
+      "src": "/web-app-manifest-512x512.png",
+      "sizes": "512x512",
+      "type": "image/png",
+      "purpose": "maskable"
+    }
+  ],
+  "theme_color": "#ffffff",
+  "background_color": "#ffffff",
+  "display": "standalone"
+}

tests/testthat/_snaps/ipw.md

@@ -12,7 +12,8 @@
           x2), method = bw_entropy(), estimand = "ate") 
 
       Outcome Model:
-        Call: stats::glm(formula = formula, family = family, data = data, weights = .wts) 
+        Call: stats::glm(formula = y ~ exposure, family = family, data = data, 
+          weights = .wts) 
 
       Marginal estimates:
                      estimate  std.err      z   ci.lower ci.upper conf.level

tests/testthat/test-ipw.R

@@ -701,6 +701,8 @@ test_that("an ipw() result prints for a balancing fit", {
   )
   w <- as.numeric(stats::weights(fit))
   outcome_mod <- fit_outcome(y ~ exposure, data, w, stats::binomial())
+  # R versions differ in whether glm() stores the formula symbol or its value.
+  outcome_mod$call$formula <- quote(y ~ exposure)
 
   result <- ipw(fit, outcome_mod)