lint.R: move preprocessing to a special function

The preprocessing functions were too unwieldy to leave them on the
command line. This makes it possible to run any lint check directly from
the command line before committing code.

Co-authored-by: Michael Chirico <chiricom@google.com>

Author: aitap

.ci/README.md

@@ -55,25 +55,15 @@ Base R runner for the manual (non-`lintr`) lint checks to be run from GitHub Act
 1. Path to the directory containing files defining the linters. A linter is a function that accepts one argument (typically the path to the file) and signals an error if it fails the lint check.
 2. Path to the directory containing files to check.
 3. A regular expression matching the files to check.
-4. An R expression evaluating to a function that processes a single file path into a form understood by the lint functions. Additionally, the function may return `NULL` to indicate that the file must be skipped.
+
+One of the files in the linter directory may define the `.preprocess` function, which must accept one file path and return a value that other linter functions will understand. The function may also return `NULL` to indicate that the file must be skipped.
 
 Example command lines:
 
 ```sh
-# C linters expect a three-element named list, not the file path
-Rscript .ci/lint.R .ci/linters/c src '[.][ch]$' '
-  function (f) list(
-    c_obj = f, lines = readLines(f),
-    preprocessed = system2("gcc", shQuote(c("-fpreprocessed", "-E", f)), stdout = TRUE, stderr = FALSE)
-  )
-'
-# po linters must skip files that don't differ from master
-Rscript .ci/lint.R .ci/linters/po po '[.]po$' '
-  function (f) {
-    diff_v_master = system2("git", c("diff", "master", f), stdout=TRUE)
-    if (length(diff_v_master)) f
-  }
-'
+Rscript .ci/lint.R .ci/linters/c src '[.][ch]$'
+Rscript .ci/lint.R .ci/linters/po po '[.]po$'
+Rscript .ci/lint.R .ci/linters/md . '[.]R?md$'
 ```
 
 ## GitLab Open Source Program

.ci/lint.R

@@ -7,13 +7,12 @@ if (identical(args, '--help')) {
     'KIND must name the directory containing the *.R files defining the linter functions.',
     'WHERE must name the directory containing the files to lint, e.g. "po", or "src".',
     "WHAT must contain the regular expression matching the files to lint, e.g., '[.]po$', or '[.][ch]$'.",
-    'PREPROCESS is an optional R function that accepts a file path and returns an R object accepted by the lint functions, or NULL to skip the file.'
   ))
   q('no')
 }
-stopifnot(`Invalid arguments, see .ci/lint.R --help` = length(args) %in% 3:4)
+stopifnot(`Invalid arguments, see .ci/lint.R --help` = length(args) == 3)
 
-linter_env = new.env()
+linter_env = list2env(list(.preprocess = identity))
 for (f in list.files(args[[1]], full.names=TRUE)) sys.source(f, linter_env)
 if (!length(ls(linter_env))) stop(
   "No linters found after sourcing files in ", dQuote(args[[1]])
@@ -23,8 +22,7 @@ sources = list.files(args[[2]], pattern = args[[3]], full.names = TRUE, recursiv
 if (!length(sources)) stop(
   "No files to lint found in directory ", dQuote(args[[2]]), " for mask ", dQuote(args[[3]])
 )
-preprocess = if (length(args) == 4) eval(str2lang(args[[4]])) else identity
-sources = Filter(Negate(is.null), lapply(setNames(nm = sources), preprocess))
+sources = Filter(Negate(is.null), lapply(setNames(nm = sources), linter_env$.preprocess))
 
 okay = TRUE
 for (src in names(sources))

.ci/linters/c/00preprocess.R

@@ -0,0 +1,7 @@
+.preprocess = function (f) list(
+  c_obj = f, lines = readLines(f),
+  preprocessed = system2(
+    "gcc", shQuote(c("-fpreprocessed", "-E", f)),
+    stdout = TRUE, stderr = FALSE
+  )
+)

.ci/linters/po/00preprocess.R

@@ -0,0 +1,4 @@
+.preprocess = function (f) {
+  diff_v_master = system2('git', c('diff', 'master', f), stdout=TRUE)
+  if (length(diff_v_master)) f
+}

.github/workflows/code-quality.yaml

@@ -40,15 +40,7 @@ jobs:
         run: /usr/bin/sudo apt-get -y install coccinelle
       - name: Lint
         run: |
-          Rscript .ci/lint.R .ci/linters/c src '[.][ch]$' '
-            function (f) list(
-              c_obj = f, lines = readLines(f),
-              preprocessed = system2(
-                "gcc", shQuote(c("-fpreprocessed", "-E", f)),
-                stdout = TRUE, stderr = FALSE
-              )
-            )
-          '
+          Rscript .ci/lint.R .ci/linters/c src '[.][ch]$'
   lint-po:
     runs-on: ubuntu-latest
     steps:
@@ -61,12 +53,7 @@ jobs:
         # until a translation is added or the blank/fuzzy translations removed. We'd
         # rather only have the failure on one PR, then ignore these files later.
         run: |
-          Rscript .ci/lint.R .ci/linters/po po '[.]po$' "
-            function (f) {
-              diff_v_master = system2('git', c('diff', 'master', f), stdout=TRUE)
-              if (length(diff_v_master)) f
-            }
-          "
+          Rscript .ci/lint.R .ci/linters/po po '[.]po$'
   lint-md:
     runs-on: ubuntu-latest
     steps: