#!/bin/bash

# Count lines of code in the project in an intelligent way.
#
# Notes/gotchas:
#
#   1. Use temporary files instead of variables to allow easier examination
#      after the script is run, and also because newline handling in shell
#      variables is rather hairy & error prone (e.g., command substitution
#      strips trailing newlines but echo adds a trailing newline, hiding it).
#
#   2. To add/update language definitions, cloc only has “merge, but mine are
#      lower priority” (--read-lang-def) and “use mine only, with no default
#      definitions” (--force-lang-def). We need “merge my definitions with
#      default, but with mine higher priority”, so we emulate it with
#      --force-lang-def, providing all the language definitions we need, some
#      of which are altered.
#
#   3. cloc also does not have a way to define file prefixes (say,
#      Dockerfile.*). So we list all the Dockerfiles we have manually in the
#      language definitions.
#
# [1]: https://github.com/AlDanial/cloc/issues/401

set -e -o pipefail

export LC_ALL=C
cd "$(dirname "$0")"/..

countem () {
    msg=$1
    infile=$2
    outfile=${2}.out
    catfile=${2}.cat
    ignore_expected=${3:-0}
    nolist=$4
    echo
    echo "*** $msg ***"
    #cat "$infile"
    rm -f /tmp/loc.ignore.out
    cloc --force-lang-def=/dev/stdin \
         --categorized="$catfile" \
         --list-file="$infile" \
         --ignored=/tmp/loc.ignore.out \
         <<'EOF' | tee "$outfile"
Bash
    filter remove_matches ^\s*#
    filter remove_inline #.*$
    extension bash
    extension bats
    extension bats.in
    script_exe bash
    3rd_gen_scale 3.81
    end_of_line_continuation \\$
C
    filter rm_comments_in_strings " /* */
    filter rm_comments_in_strings " // 
    filter call_regexp_common C++
    extension c
    extension h
    3rd_gen_scale 0.77
    end_of_line_continuation \\$
conf-like
    filter remove_matches ^\s*#
    filter remove_inline #.*$
    extension rpmlintrc
    extension spec
    filename .dockerignore
    filename .gitattributes
    filename .gitignore
    3rd_gen_scale 1.00
    end_of_line_continuation \\$
Dockerfile
    filter remove_matches ^\s*#
    filter remove_inline #.*$
    extension df
    filename Dockerfile
    filename Dockerfile.argenv
    filename Dockerfile.centos_7ch
    filename Dockerfile.almalinux_9ch
    filename Dockerfile.debian_12ch
    filename Dockerfile.file-quirks
    filename Dockerfile.libfabric
    filename Dockerfile.metadata
    filename Dockerfile.mpich
    filename Dockerfile.nvidia
    filename Dockerfile.ocimanifest
    filename Dockerfile.openmpi
    filename Dockerfile.quick
    3rd_gen_scale 2.00
    end_of_line_continuation \\$
JSON
    filter remove_matches ^\s*$
    extension json
    3rd_gen_scale 2.50
m4
    filter remove_matches ^dnl\s
    filter remove_matches ^\s*#
    filter remove_inline #.*$
    extension ac
    extension m4
    3rd_gen_scale 1.00
Make
    filter remove_matches ^\s*#
    filter remove_inline #.*$
    extension Gnumakefile
    extension Makefile
    extension am
    extension gnumakefile
    extension makefile
    extension mk
    filename Gnumakefile
    filename Makefile
    filename gnumakefile
    filename makefile
    script_exe make
    3rd_gen_scale 2.50
    end_of_line_continuation \\$
Markdown
    filter remove_between_general <!-- -->
    filter remove_between_regex \[(comment|\/\/)?\]\s*:?\s*(<\s*>|#)?\s*\(.*? .*?\)
    extension md
    3rd_gen_scale 1.00
patch
    filter remove_matches ^#
    filter remove_matches ^\-\-\-
    filter remove_matches ^\+\+\+
    filter remove_matches ^\s
    filter remove_matches ^@@
    extension diff
    extension patch
    3rd_gen_scale 1.00
plain text
    filter remove_matches TEXT_HAS_NO_COMMENTS_BUT_A_FILTER_IS_REQUIRED
    extension txt
    filename NOTICE
    filename PERUSEME
    filename README
    filename VERSION
    filename approved-trailing-whitespace
    3rd_gen_scale 1.00
POSIX sh
    filter remove_matches ^\s*#
    filter remove_inline #.*$
    extension sh
    script_exe sh
    3rd_gen_scale 3.81
    end_of_line_continuation \\$
Python
    filter remove_matches ^\s*#
    filter docstring_to_C
    filter call_regexp_common C
    filter remove_inline #.*$
    extension py
    extension py.in
    script_exe python
    script_exe python2.6
    script_exe python2.7
    script_exe python3
    script_exe python3.3
    script_exe python3.4
    script_exe python3.5
    3rd_gen_scale 4.20
    end_of_line_continuation \\$
ReST
    filter remove_between_regex ^\.\. ^[^\s\.]
    extension rst
    3rd_gen_scale 1.50
Ruby
    filter remove_matches ^\s*#
    filter remove_below_above ^=begin ^=end
    filter remove_inline #.*$
    extension rake
    extension rb
    filename Rakefile
    filename rakefile
    filename Vagrantfile
    script_exe ruby
    3rd_gen_scale 4.20
    end_of_line_continuation \\$
VTK
    filter remove_matches ^\s*#
    extension vtk
    3rd_gen_scale 1.00
    end_of_line_continuation \\$
YAML
    filter remove_matches ^\s*#
    filter remove_inline #.*$
    extension yaml
    extension yml
    extension yaml.in
    extension yml.in
    3rd_gen_scale 0.90
EOF
    if [[ -z $nolist ]]; then
        cat "$catfile"
    fi
    # Counting the number of ignored files is a problem because (1) before
    # 1.84, the list of ignored files was empty [1], but (2) starting with
    # 1.98, --quiet is assumed if stdin is not a TTY and there’s no way to get
    # the previous behavior, so you can no longer look at the output. I
    # couldn’t think of anything better than to assume the ≥1.98 behavior,
    # which is probably OK since that came out in Sept. 2019.
    #
    # [1]: https://github.com/AlDanial/cloc/issues/401
    ignore_actual=$(cat /tmp/loc.ignore.out | grep -c . || true)
    if [[ $ignore_expected -ne $ignore_actual ]]; then
        grep -F '(unknown)' "$catfile" || true
        echo
        echo "🚨🚨🚨 cloc ignoring wrong number of files; expected ${ignore_expected}, actual ${ignore_actual} 🚨🚨🚨"
        cat /tmp/loc.ignore.out
        exit 1
    fi
}

if [[ -e ./Makefile.in ]]; then
    echo '🚨🚨🚨 Makefile.in seems to exist 🚨🚨🚨'
    echo 'did you "./autogen.sh --clean --rm-lark"?'
    exit 1
fi

min_cloc=1.84
if    ! command -v cloc > /dev/null \
   || [[ $(  printf '%s\n%s\n' "$min_cloc" "$(cloc --version)" \
           | sort -V | head -1 ) != "$min_cloc" ]]; then
    echo "🚨🚨🚨 no cloc version ≥ $min_cloc 🚨🚨🚨"
    echo 'did you try a better operating system?'
    exit 1
fi

# program (Charliecloud itself)
find ./bin ./lib -type f -a \( \
        -name '*.c' \
     -o -name '*.h' \
     -o -name '*.py' \
     -o -name '*.py.in' \
     -o -name '*.sh' \
     -o -path './bin/ch-*' \) | grep -Fv ./bin/ch-test | sort > /tmp/loc.program
countem "PROGRAM" /tmp/loc.program

# test suite
find ./examples ./test -type f -a \( \
        -name '*.bash' \
     -o -name '*.bats' \
     -o -name '*.bats.in' \
     -o -name '*.c' \
     -o -name '*.df' \
     -o -name '*.json' \
     -o -name '*.patch' \
     -o -name '*.py' \
     -o -name '*.py.in' \
     -o -name '*.sh' \
     -o -name '*.vtk' \
     -o -name '*.yml' \
     -o -name '*.yml.in' \
     -o -name 'Build.*' \
     -o -name 'Dockerfile.*' \
     -o -name .dockerignore \
     -o -name Build \
     -o -name Dockerfile \
     -o -name Makefile \
     -o -name README \
     -o -path ./examples/chtest/printns \
     -o -path ./examples/spack/packages.yaml \
     -o -path ./examples/spack/libfuse.patch \
     -o -path ./test/fixtures/README \
     -o -path ./test/gitlab.com/build-images \
     -o -path ./test/gitlab.com/expand-matrices \
     -o -path ./test/gitlab.com/reviews-email \
     -o -path ./test/approved-trailing-whitespace \
     -o -path ./test/common.bash \
     -o -path ./test/doctest-auto \
     -o -path ./test/old-storage \
     -o -path ./test/sotest/files_inferrable.txt \
     -o -path ./test/whiteout \) | sort > /tmp/loc.test
echo ./bin/ch-test >> /tmp/loc.test
countem "TEST SUITE & EXAMPLES" /tmp/loc.test

# documentation
find . -type f -a \( \
         -path './doc/*.rst' \
      -o -path './doc/releases/*.md' \
      -o -path ./NOTICE \
      -o -path ./README.rst \
      -o -path ./doc/cdi-cray-libfabric.json \
      -o -path ./doc/cdi-nvidia.json \
      -o -path ./doc/conf.py \
      -o -path ./doc/make-deps-overview \
      -o -path ./doc/man/README \
      -o -path ./doc/publish \) | sort > /tmp/loc.doc
countem "DOCUMENTATION" /tmp/loc.doc

# build system
find . -type f -a \( \
        -name Makefile.am \
     -o -path ./autogen.sh \
     -o -path ./configure.ac \
     -o -path ./misc/mkdir-p \
     -o -path ./misc/version \
     -o -path ./misc/m4/README \) | sort > /tmp/loc.build
countem "BUILD SYSTEM" /tmp/loc.build

# packaging code
find ./packaging \( -name Makefile.am \) -prune -o -type f \
     -print | sort > /tmp/loc.packaging
countem "PACKAGING" /tmp/loc.packaging

# misc
find . -type f -a \( \
        -path ./.gitattributes \
     -o -path ./.gitignore \
     -o -path ./VERSION \
     -o -path ./misc/branches-tidy \
     -o -path ./misc/gdb-backtrace \
     -o -path ./misc/grep \
     -o -path ./misc/loc \
     -o -path ./misc/tar-upload \) | sort > /tmp/loc.misc
countem "MISCELLANEOUS" /tmp/loc.misc

# ignored - this includes binaries and files we distribute but didn’t write
find . -type f -a \( \
        -name 'core' \
     -o -name 'a.out' \
     -o -name '*.ico' \
     -o -name '.gitmodules' \
     -o -name 'file_' \
     -o -name '*.png' \
     -o -name '*.tar.gz' \
     -o -name 'file[A-Z0-9y]*' \
     -o -name 's_dir?' \
     -o -name 's_file?' \
     -o -path './.vscode/*' \
     -o -path './misc/m4/*.m4' \
     -o -path './packaging/debian/*' \
     -o -path ./LICENSE \
     -o -path ./test/fixtures/empty-file \) | sort > /tmp/loc.ignored
echo
echo "*** IGNORED ***"
cat /tmp/loc.ignored

# everything
find . \(   -path ./.git \
         -o -path ./doc/html \
         -o -name .DS_Store \
         -o -name ._.DS_Store \
         -o -name __pycache__ \) \
         -prune -o -type f -print \
  | sort > /tmp/loc.all
cat /tmp/loc.{all,ignored} | sort | uniq -u > /tmp/loc.total
countem "TOTAL" /tmp/loc.total 0 nolist

# test for files we forgot
  cat /tmp/loc.{program,test,doc,build,packaging,misc,ignored,all} \
| sort | uniq -c | (grep -Ev '^\s*2' || true) > /tmp/loc.extra
if [[ -s /tmp/loc.extra ]]; then
    echo
    echo 'UNKNOWN FILES'
    cat /tmp/loc.extra
    echo
    echo '🚨🚨🚨 unknown files found 🚨🚨🚨'
    echo 'hint: did you forget to add new file(s) to misc/loc?'
    exit 1
fi

# generate loc.rst
cmd='s/^SUM:.* ([0-9]+)$/\1/p'
cat <<EOF > doc/_loc.rst
.. Do not edit this file — it’s auto-generated.

We pride ourselves on keeping Charliecloud lightweight and simple. The lines
of code as of version $(cat VERSION) is:

.. list-table::

   * - Program itself
     - $(sed -En "${cmd}" /tmp/loc.program.out)
   * - Test suite & examples
     - $(sed -En "${cmd}" /tmp/loc.test.out)
   * - Documentation
     - $(sed -En "${cmd}" /tmp/loc.doc.out)
   * - Build system
     - $(sed -En "${cmd}" /tmp/loc.build.out)
   * - Packaging
     - $(sed -En "${cmd}" /tmp/loc.packaging.out)
   * - Miscellaneous
     - $(sed -En "${cmd}" /tmp/loc.misc.out)
   * - Total
     - $(sed -En "${cmd}" /tmp/loc.total.out)

These include code only, excluding blank lines and comments. They were counted
using \`cloc <https://github.com/AlDanial/cloc>\`_ version $(cloc --version).
We typically quote the "Program itself" number when describing the size of
Charliecloud. (Please do not quote the size in Priedhorsky and Randles 2017,
as that number is very out of date.)

EOF
echo
cat doc/_loc.rst