.github/workflows/file_health.sh

   1 #!/usr/bin/env zsh
   2
   3 set -o nounset
   4
   5 # Assumption: file names don't contain `:` (for the `cut` invocation).
   6 # Safe to assume, as such a character in a filename would cause trouble on Windows, a platform we support
   7
   8 # any regression turn this non-zero
   9 regressions=0
  10
  11 # exclusions (these are just grep extended regular expressions to match against paths relative to the root of the repository)
  12 exclusions_nonutf8='(.*(7z|gif|ic(ns|o)|png|qm|zip))'
  13 exclusions_bom='src/base/unicodestrings.h'
  14 exclusions_tw='(*.ts)|src/webui/www/private/scripts/lib/*'
  15 exclusions_trailing_newline='configure'
  16 exclusions_no_lf='(*.ts)|(.*svg)|compile_commands.json|src/webui/www/private/scripts/lib/*'
  17
  18 echo -e "\n*** Detect files not encoded in UTF-8 ***\n"
  19
  20 find . -path ./build -prune -false -o -path ./.git -prune -false -o -type f -exec file --mime {} \; | sort \
  21     | grep -v -e "charset=us-ascii" -e "charset=utf-8" | cut -d ":" -f 1 \
  22     | grep -E -v -e "${exclusions_nonutf8}" \
  23     | tee >(echo -e "--> Files not encoded in UTF-8: found" "$(wc -l < /dev/stdin)" "regression(s)\n") \
  24     | xargs -I my_input -0 bash -c 'echo "my_input"; test "$(echo -n "my_input" | wc -l)" -eq 0'
  25 regressions=$((regressions+$?))
  26
  27 echo -e "\n*** Detect files encoded in UTF-8 with BOM ***\n"
  28
  29 grep --exclude-dir={.git,build} -rIl $'\xEF\xBB\xBF' | sort \
  30     | grep -E -v -e "${exclusions_bom}" \
  31     | tee >(echo -e "--> Files encoded in UTF-8 with BOM: found" "$(wc -l < /dev/stdin)" "regression(s)\n") \
  32     | xargs -I my_input -0 bash -c 'echo "my_input"; test "$(echo -n "my_input" | wc -l)" -eq 0'
  33 regressions=$((regressions+$?))
  34
  35 echo -e "\n*** Detect usage of CR byte ***\n"
  36
  37 grep --exclude-dir={.git,build} -rIlU $'\x0D' | sort \
  38     | tee >(echo -e "--> Usage of CR byte: found" "$(wc -l < /dev/stdin)" "regression(s)\n") \
  39     | xargs -I my_input -0 bash -c 'echo "my_input"; test "$(echo -n "my_input" | wc -l)" -eq 0'
  40 regressions=$((regressions+$?))
  41
  42 echo -e "\n*** Detect trailing whitespace in lines ***\n"
  43
  44 grep --exclude-dir={.git,build} -rIl "[[:blank:]]$" | sort \
  45     | grep -E -v -e "${exclusions_tw}" \
  46     | tee >(echo -e "--> Trailing whitespace in lines: found" "$(wc -l < /dev/stdin)" "regression(s)\n") \
  47     | xargs -I my_input -0 bash -c 'echo "my_input"; test "$(echo -n "my_input" | wc -l)" -eq 0';
  48 regressions=$((regressions+$?))
  49
  50 echo -e "\n*** Detect too many trailing newlines ***\n"
  51
  52 find . -path ./build -prune -false -o -path ./.git -prune -false -o -type f -exec file --mime {} \; | sort \
  53     | grep -e "charset=us-ascii" -e "charset=utf-8" | cut -d ":" -f 1 \
  54     | grep -E -v -e "${exclusions_trailing_newline}" \
  55     | xargs -L1 -I my_input bash -c 'test "$(tail -q -c2 "my_input" | hexdump -C | grep "0a 0a")" && echo "my_input"' \
  56     | tee >(echo -e "--> Too many trailing newlines: found" "$(wc -l < /dev/stdin)" "regression(s)\n") \
  57     | xargs -I my_input -0 bash -c 'echo "my_input"; test "$(echo -n "my_input" | wc -l)" -eq 0'
  58 regressions=$((regressions+$?))
  59
  60 echo -e "\n*** Detect no trailing newline ***\n"
  61
  62 find . -path ./build -prune -false -o -path ./.git -prune -false -o -type f -exec file --mime {} \; | sort \
  63     | grep -e "charset=us-ascii" -e "charset=utf-8" |  cut -d ":" -f 1 \
  64     | grep -E -v -e "${exclusions_no_lf}" \
  65     | xargs -L1 -I my_input bash -c 'test "$(tail -q -c1 "my_input" | hexdump -C | grep "0a")" || echo "my_input"' \
  66     | tee >(echo -e "--> No trailing newline: found" "$(wc -l < /dev/stdin)" "regression(s)\n") \
  67     | xargs -I my_input -0 bash -c 'echo "my_input"; test "$(echo -n "my_input" | wc -l)" -eq 0'
  68 regressions=$((regressions+$?))
  69
  70 echo -e "\n*** Detect translation closing tag in new line ***\n"
  71
  72 grep --exclude-dir={.git,build} -nri "^</translation>" | sort \
  73     | cut -d ":" -f 1,2 \
  74     | tee >(echo -e "--> Translation closing tag in new line: found" "$(wc -l < /dev/stdin)" "regression(s)\n") \
  75     | xargs -I my_input -0 bash -c 'echo "my_input"; test "$(echo -n "my_input" | wc -l)" -eq 0'
  76 regressions=$((regressions+$?))
  77
  78 if [ "$regressions" -ne 0 ]; then
  79     regressions=1
  80     echo "\nFile health regressions found. Please fix them (or add them as exclusions)."
  81 else
  82     echo "All OK, no file health regressions found."
  83 fi
  84
  85 exit $regressions;