#!/bin/bash

set -eu

# Copyright 2020 Google Inc. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

# Tool to evaluate the transitive closure of the ninja dependency graph of the
# files and targets a given target depends on.
#
# i.e. the list of things that, if changed, could cause a change to a target.

readonly me=$(basename "${0}")

readonly usage="usage: ${me} {options} target [target...]

Evaluate the transitive closure of files and ninja targets that one or more
targets depend on.

Dependency Options:

  -(no)order_deps   Whether to include order-only dependencies. (Default false)
  -(no)implicit     Whether to include implicit dependencies. (Default true)
  -(no)explicit     Whether to include regular / explicit deps. (Default true)

  -nofollow         Unanchored regular expression. Matching paths and targets
                    always get reported. Their dependencies do not get reported
                    unless first encountered in a 'container' file type.
                    Multiple allowed and combined using '|'.
                    e.g. -nofollow='*.so' not -nofollow='.so$'
                    -nofollow='*.so|*.dex' or -nofollow='*.so' -nofollow='.dex'
                    (Defaults to no matches)
  -container        Unanchored regular expression. Matching file extensions get
                    treated as 'container' files for -nofollow option.
                    Multiple allowed and combines using '|'
                    (Default 'apex|apk|zip|jar|tar|tgz')

Output Options:

  -(no)quiet        Suppresses progress output to stderr and interactive
    alias -(no)q    prompts. By default, when stderr is a tty, progress gets
                    reported to stderr; when both stderr and stdin are tty,
                    the script asks user whether to delete intermediate files.
                    When suppressed or not prompted, script always deletes the
                    temporary / intermediate files.
  -sep=<delim>      Use 'delim' as output field separator between notice
                    checksum and notice filename in notice output.
                    e.g. sep='\\t'
                    (Default space)
  -csv              Shorthand for -sep=','
  -directories=<f>  Output directory names of dependencies to 'f'.
    alias -d        User '/dev/stdout' to send directories to stdout. Defaults
                    to no directory output.
  -notices=<file>   Output license and notice file paths to 'file'.
    alias -n        Use '/dev/stdout' to send notices to stdout. Defaults to no
                    license/notice output.
  -projects=<file>  Output git project names to 'file'. Use '/dev/stdout' to
    alias -p        send projects to stdout. Defaults to no project output.
  -targets=<fils>   Output target dependencies to 'file'. Use '/dev/stdout' to
    alias -t        send targets to stdout.
                    When no directory, notice, project or target output options
                    given, defaults to stdout. Otherwise, defaults to no target
                    output.

At minimum, before running this script, you must first run:
$ source build/envsetup.sh
$ lunch
$ m nothing
to setup the build environment, choose a target platform, and build the ninja
dependency graph.
"

function die() { echo -e "${*}" >&2; exit 2; }

# Reads one input target per line from stdin; outputs (isnotice target) tuples.
#
# output target is a ninja target that the input target depends on
# isnotice in {0,1} with 1 for output targets believed to be license or notice
function getDeps() {
    (tr '\n' '\0' | xargs -0 -r "${ninja_bin}" -f "${ninja_file}" -t query) \
    | awk -v include_order="${include_order_deps}" \
        -v include_implicit="${include_implicit_deps}" \
        -v include_explicit="${include_deps}" \
        -v containers="${container_types}" \
    '
      BEGIN {
        ininput = 0
        isnotice = 0
        currFileName = ""
        currExt = ""
      }
      $1 == "outputs:" {
        ininput = 0
      }
      ininput == 0 && $0 ~ /^\S\S*:$/ {
        isnotice = ($0 ~ /.*NOTICE.*[.]txt:$/)
        currFileName = gensub(/^.*[/]([^/]*)[:]$/, "\\1", "g")
        currExt = gensub(/^.*[.]([^./]*)[:]$/, "\\1", "g")
      }
      ininput != 0 && $1 !~ /^[|][|]?/ {
        if (include_explicit == "true") {
          fileName = gensub(/^.*[/]([^/]*)$/, "\\1", "g")
          print ( \
              (isnotice && $0 !~ /^\s*build[/]soong[/]scripts[/]/) \
              || $0 ~ /NOTICE|LICEN[CS]E/ \
              || $0 ~ /(notice|licen[cs]e)[.]txt/ \
          )" "(fileName == currFileName||currExt ~ "^(" containers ")$")" "gensub(/^\s*/, "", "g")
        }
      }
      ininput != 0 && $1 == "|" {
        if (include_implicit == "true") {
          fileName = gensub(/^.*[/]([^/]*)$/, "\\1", "g")
          $1 = ""
          print ( \
              (isnotice && $0 !~ /^\s*build[/]soong[/]scripts[/]/) \
              || $0 ~ /NOTICE|LICEN[CS]E/ \
              || $0 ~ /(notice|licen[cs]e)[.]txt/ \
          )" "(fileName == currFileName||currExt ~ "^(" containers ")$")" "gensub(/^\s*/, "", "g")
        }
      }
      ininput != 0 && $1 == "||" {
        if (include_order == "true") {
          fileName = gensub(/^.*[/]([^/]*)$/, "\\1", "g")
          $1 = ""
          print ( \
              (isnotice && $0 !~ /^\s*build[/]soong[/]scripts[/]/) \
              || $0 ~ /NOTICE|LICEN[CS]E/ \
              || $0 ~ /(notice|licen[cs]e)[.]txt/ \
          )" "(fileName == currFileName||currExt ~ "^(" containers ")$")" "gensub(/^\s*/, "", "g")
        }
      }
      $1 == "input:" {
        ininput = 1
      }
    '
}

# Reads one input directory per line from stdin; outputs unique git projects.
function getProjects() {
    while read d; do
        while [ "${d}" != '.' ] && [ "${d}" != '/' ]; do
            if [ -d "${d}/.git/" ]; then
                echo "${d}"
                break
            fi
            d=$(dirname "${d}")
        done
    done | sort -u
}


if [ -z "${ANDROID_BUILD_TOP}" ]; then
    die "${me}: Run 'lunch' to configure the build environment"
fi

if [ -z "${TARGET_PRODUCT}" ]; then
    die "${me}: Run 'lunch' to configure the build environment"
fi

readonly ninja_file="${ANDROID_BUILD_TOP}/out/combined-${TARGET_PRODUCT}.ninja"
if [ ! -f "${ninja_file}" ]; then
    die "${me}: Run 'm nothing' to build the dependency graph"
fi

readonly ninja_bin="${ANDROID_BUILD_TOP}/prebuilts/build-tools/linux-x86/bin/ninja"
if [ ! -x "${ninja_bin}" ]; then
    die "${me}: Cannot find ninja executable expected at ${ninja_bin}"
fi


# parse the command-line

declare -a targets # one or more targets to evaluate

include_order_deps=false    # whether to trace through || "order dependencies"
include_implicit_deps=true  # whether to trace through | "implicit deps"
include_deps=true           # whether to trace through regular explicit deps
quiet=false                 # whether to suppress progress

projects_out=''             # where to output the list of projects
directories_out=''          # where to output the list of directories
targets_out=''              # where to output the list of targets/source files
notices_out=''              # where to output the list of license/notice files

sep=" "                     # separator between md5sum and notice filename

nofollow=''                 # regularexp must fully match targets to skip

container_types=''          # regularexp must full match file extension
                            # defaults to 'apex|apk|zip|jar|tar|tgz' below.

use_stdin=false             # whether to read targets from stdin i.e. target -

while [ $# -gt 0 ]; do
    case "${1:-}" in
      -)
        use_stdin=true
      ;;
      -*)
        flag=$(expr "${1}" : '^-*\(.*\)$')
        case "${flag:-}" in
          order_deps)
            include_order_deps=true;;
          noorder_deps)
            include_order_deps=false;;
          implicit)
            include_implicit_deps=true;;
          noimplicit)
            include_implicit_deps=false;;
          explicit)
            include_deps=true;;
          noexplicit)
            include_deps=false;;
          csv)
            sep=",";;
          sep)
            sep="${2?"${usage}"}"; shift;;
          sep=)
            sep=$(expr "${flag}" : '^sep=\(.*\)$');;
          q) ;&
          quiet)
            quiet=true;;
          noq) ;&
          noquiet)
            quiet=false;;
          nofollow)
            case "${nofollow}" in
              '')
                nofollow="${2?"${usage}"}";;
              *)
                nofollow="${nofollow}|${2?"${usage}"}";;
            esac
            shift
          ;;
          nofollow=*)
            case "${nofollow}" in
              '')
                nofollow=$(expr "${flag}" : '^nofollow=\(.*\)$');;
              *)
                nofollow="${nofollow}|"$(expr "${flag}" : '^nofollow=\(.*\)$');;
            esac
          ;;
          container)
            container_types="${container_types}|${2?"${usage}"}";;
          container=*)
            container_types="${container_types}|"$(expr "${flag}" : '^container=\(.*\)$');;
          p) ;&
          projects)
            projects_out="${2?"${usage}"}"; shift;;
          p=*) ;&
          projects=*)
            projects_out=$(expr "${flag}" : '^.*=\(.*\)$');;
          d) ;&
          directores)
            directories_out="${2?"${usage}"}"; shift;;
          d=*) ;&
          directories=*)
            directories_out=$(expr "${flag}" : '^.*=\(.*\)$');;
          t) ;&
          targets)
            targets_out="${2?"${usage}"}"; shift;;
          t=*) ;&
          targets=)
            targets_out=$(expr "${flag}" : '^.*=\(.*\)$');;
          n) ;&
          notices)
            notices_out="${2?"${usage}"}"; shift;;
          n=*) ;&
          notices=)
            notices_out=$(expr "${flag}" : '^.*=\(.*\)$');;
          *)
            die "${usage}\n\nUnknown flag ${1}";;
        esac
      ;;
      *)
        targets+=("${1:-}")
      ;;
    esac
    shift
done


# fail fast if command-line arguments are invalid

if [ ! -v targets[0] ] && ! ${use_stdin}; then
    die "${usage}\n\nNo target specified."
fi

if [ -z "${projects_out}" ] \
  && [ -z "${directories_out}" ] \
  && [ -z "${targets_out}" ] \
  && [ -z "${notices_out}" ]
then
    targets_out='/dev/stdout'
fi

if [ -z "${container_types}" ]; then
  container_types='apex|apk|zip|jar|tar|tgz'
fi

# showProgress when stderr is a tty
if [ -t 2 ] && ! ${quiet}; then
    showProgress=true
else
    showProgress=false
fi

# interactive when both stderr and stdin are tty
if ${showProgress} && [ -t 0 ]; then
    interactive=true
else
    interactive=false
fi


readonly tmpFiles=$(mktemp -d "${TMPDIR}.tdeps.XXXXXXXXX")
if [ -z "${tmpFiles}" ]; then
    die "${me}: unable to create temporary directory"
fi

# The deps files contain unique (isnotice target) tuples where
# isnotice in {0,1} with 1 when ninja target 'target' is a license or notice.
readonly oldDeps="${tmpFiles}/old"
readonly newDeps="${tmpFiles}/new"
readonly allDeps="${tmpFiles}/all"

if ${use_stdin}; then # start deps by reading 1 target per line from stdin
  awk '
    NF > 0 {
      print ( \
          $0 ~ /NOTICE|LICEN[CS]E/ \
          || $0 ~ /(notice|licen[cs]e)[.]txt/ \
      )" "gensub(/\s*$/, "", "g", gensub(/^\s*/, "", "g"))
    }
  ' > "${newDeps}"
else # start with no deps by clearing file
  : > "${newDeps}"
fi

# extend deps by appending targets from command-line
for idx in "${!targets[*]}"; do
    isnotice='0'
    case "${targets[${idx}]}" in
      *NOTICE*) ;&
      *LICEN[CS]E*) ;&
      *notice.txt) ;&
      *licen[cs]e.txt)
        isnotice='1';;
    esac
    echo "${isnotice} 1 ${targets[${idx}]}" >> "${newDeps}"
done

# remove duplicates and start with new, old and all the same
sort -u < "${newDeps}" > "${allDeps}"
cp "${allDeps}" "${newDeps}"
cp "${allDeps}" "${oldDeps}"

# report depth of dependenciens when showProgress
depth=0

# 1st iteration always unfiltered
filter='cat'
while [ $(wc -l < "${newDeps}") -gt 0 ]; do
    if ${showProgress}; then
        echo "depth ${depth} has "$(wc -l < "${newDeps}")" targets" >&2
        depth=$(expr ${depth} + 1)
    fi
    ( # recalculate dependencies by combining unique inputs of new deps w. old
        set +e
        sh -c "${filter}" < "${newDeps}" | cut -d\  -f3- | getDeps
        set -e
        cat "${oldDeps}"
    ) | sort -u > "${allDeps}"
    # recalculate new dependencies as net additions to old dependencies
    set +e
    diff "${oldDeps}" "${allDeps}" --old-line-format='' --new-line-format='%L' \
      --unchanged-line-format='' > "${newDeps}"
    set -e
    # apply filters on subsequent iterations
    case "${nofollow}" in
      '')
        filter='cat';;
      *)
        filter="egrep -v '^[01] 0 (${nofollow})$'"
      ;;
    esac
    # recalculate old dependencies for next iteration
    cp "${allDeps}" "${oldDeps}"
done

# found all deps -- clean up last iteration of old and new
rm -f "${oldDeps}"
rm -f "${newDeps}"

if ${showProgress}; then
    echo $(wc -l < "${allDeps}")" targets" >&2
fi

if [ -n "${targets_out}" ]; then
    cut -d\  -f3- "${allDeps}" | sort -u > "${targets_out}"
fi

if [ -n "${directories_out}" ] \
  || [ -n "${projects_out}" ] \
  || [ -n "${notices_out}" ]
then
    readonly allDirs="${tmpFiles}/dirs"
    (
        cut -d\  -f3- "${allDeps}" | tr '\n' '\0' | xargs -0 dirname
    ) | sort -u > "${allDirs}"
    if ${showProgress}; then
        echo $(wc -l < "${allDirs}")" directories" >&2
    fi

    case "${directories_out}" in
      '')        : do nothing;;
      *)
        cat "${allDirs}" > "${directories_out}"
      ;;
    esac
fi

if [ -n "${projects_out}" ] \
  || [ -n "${notices_out}" ]
then
    readonly allProj="${tmpFiles}/projects"
    set +e
    egrep -v '^out[/]' "${allDirs}" | getProjects > "${allProj}"
    set -e
    if ${showProgress}; then
        echo $(wc -l < "${allProj}")" projects" >&2
    fi

    case "${projects_out}" in
      '')        : do nothing;;
      *)
        cat "${allProj}" > "${projects_out}"
      ;;
    esac
fi

case "${notices_out}" in
  '')        : do nothing;;
  *)
    readonly allNotice="${tmpFiles}/notices"
    set +e
    egrep '^1' "${allDeps}" | cut -d\  -f3- | egrep -v '^out/' > "${allNotice}"
    set -e
    cat "${allProj}" | while read proj; do
        for f in LICENSE LICENCE NOTICE license.txt notice.txt; do
            if [ -f "${proj}/${f}" ]; then
                echo "${proj}/${f}"
            fi
        done
    done >> "${allNotice}"
    if ${showProgress}; then
      echo $(cat "${allNotice}" | sort -u | wc -l)" notice targets" >&2
    fi
    readonly hashedNotice="${tmpFiles}/hashednotices"
    ( # md5sum outputs checksum space indicator(space or *) filename newline
        set +e
        sort -u "${allNotice}" | tr '\n' '\0' | xargs -0 -r md5sum 2>/dev/null
        set -e
      # use sed to replace space and indicator with separator
    ) > "${hashedNotice}"
    if ${showProgress}; then
        echo $(cut -d\  -f2- "${hashedNotice}" | sort -u | wc -l)" notice files" >&2
        echo $(cut -d\  -f1 "${hashedNotice}" | sort -u | wc -l)" distinct notices" >&2
    fi
    sed 's/^\([^ ]*\) [* ]/\1'"${sep}"'/g' "${hashedNotice}" | sort > "${notices_out}"
  ;;
esac

if ${interactive}; then
    echo -n "$(date '+%F %-k:%M:%S') Delete ${tmpFiles} ? [n] " >&2
    read answer
    case "${answer}" in [yY]*) rm -fr "${tmpFiles}";; esac
else
    rm -fr "${tmpFiles}"
fi