Split one git history into two

git-split

#!/bin/bash

usage()
{
    cat <<-END
Split one git history into two.

One of them only contains changes of specified files.
The other contains other changes(or optionally contains full history).

Usage:
    git-split [ -b <to-new-branch> ] [-r] [-m] [-s <prefix> ] <from>..[<to>] -- <fns>

Synoposis:

    Initial:

        * 358aee4 (HEAD) fix css
        |  css/site.css | 13 ++---
        |
        * f869f40 init site
        |  html/index.html    | 12 ++--
        |  html/header.html   | 34 +-----
        |  css/site.css       | 61 +++++++------
        |
        * 82554be fix doc of Build


    To split folder html into a separate branch:

        git-split -m 82554be..HEAD -- html/

    Results in:

        *   948149a Merge commit 'aa76f4de5be5ddbb2908ba145fea5697c7baf728' into HEAD
        |\\
        | * aa76f4d init site
        | |  html/index.html  | 12 ++--
        | |  html/header.html | 34 +-----
        | |
        * | 088e903 fix css
        | |  css/site.css | 13 ++---
        | |
        * | 0b606c2 init site
        | |  css/site.css | 61 +++++++------
        |/
        * 82554be fix doc of Build


Arguments:

    <from> and <to> specifies history range to split.
    E.g. to split the latest 3 commit:

        git-split master~3..master

    "--" must present as a separator.

    <fns> are several file patterns.
    E.g.:

        git-split master~3..master -- src/foo.* deps/config.yaml

Options:

    -b <new-branch>             Create a new branch.
                                When absent, create a temporary branch "tmpbranch".

    -o                          Create an orphan branch instead of one based on <from>.

    -r                          Remove <fns> from original history.
                                Without -r, the original history will not be modified.

    -m                          Merge back action:
                                "" for doing nothing;
                                "merge": merge the new branch to commmit "<to>";
                                "rebase": rebase "<to>" onto the new branch.
                                non empty -m implies -r.

    -s <prefix>                 Strip <prefix> dir segments from paths that are splitted out.
                                E.g.:
                                    git-split -s 3 base..master -- my/foo/dir/bar.txt

                                    removes "my/foo/dir".
END
}



clr()
{
    # clr light read blabla

    local black=0
    local white=7
    local red=1
    local green=2
    local brown=3
    local blue=4
    local purple=5
    local cyan=6

    local light=""
    local color=$1
    shift

    if [ "$color" == "light" ]; then
        light="tput bold; "
        color=$1
        shift
    fi

    local code=$(eval 'echo $'$color)
    local cmd="${light}tput setaf $code"
    local color_str="$(eval "$cmd")"

    echo $color_str"$@""$(tput sgr0)"
}

shlib_init_colors()
{
    Black="$(                   tput setaf 0)"
    BlackBG="$(                 tput setab 0)"
    DarkGrey="$(     tput bold; tput setaf 0)"
    LightGrey="$(               tput setaf 7)"
    LightGreyBG="$(             tput setab 7)"
    White="$(        tput bold; tput setaf 7)"
    Red="$(                     tput setaf 1)"
    RedBG="$(                   tput setab 1)"
    LightRed="$(     tput bold; tput setaf 1)"
    Green="$(                   tput setaf 2)"
    GreenBG="$(                 tput setab 2)"
    LightGreen="$(   tput bold; tput setaf 2)"
    Brown="$(                   tput setaf 3)"
    BrownBG="$(                 tput setab 3)"
    Yellow="$(       tput bold; tput setaf 3)"
    Blue="$(                    tput setaf 4)"
    BlueBG="$(                  tput setab 4)"
    LightBlue="$(    tput bold; tput setaf 4)"
    Purple="$(                  tput setaf 5)"
    PurpleBG="$(                tput setab 5)"
    Pink="$(         tput bold; tput setaf 5)"
    Cyan="$(                    tput setaf 6)"
    CyanBG="$(                  tput setab 6)"
    LightCyan="$(    tput bold; tput setaf 6)"
    NC="$(                      tput sgr0)" # No Color
}


screen_width()
{
    local chr="${1--}"
    chr="${chr:0:1}"

    local width=$(tput cols 2||echo 80)
    width="${COLUMNS:-$width}"

    echo $width
}

hr()
{
    # generate a full screen width horizontal ruler
    local width=$(screen_width)

    printf -vl "%${width}s\n" && echo ${l// /$chr};
}

remove_color()
{
    # remove color control chars from stdin or first argument

    local sed=gsed
    which -s $sed || sed=sed

    local s="$1"
    if [ -z "$s" ]; then
        $sed -r "s/\x1B\[([0-9]{1,2}(;[0-9]{1,2})?)?[mGK]//g"
    else
        echo "$s" | remove_color
    fi

}

text_hr()
{
    # generate a full screen width sperator line with text.
    # text_hr "-" "a title"
    # > a title -----------------------------------------
    #
    # variable LR=l|m|r controls alignment

    local chr="$1"
    shift

    local bb="$(echo "$@" | remove_color)"
    local text_len=${#bb}

    local width=$(screen_width)
    let width=width-text_len

    local lr=${LR-m}
    case $lr in
        m)
            let left=width/2
            let right=width-left
            echo "$(printf -vl "%${left}s\n" && echo ${l// /$chr})$@$(printf -vl "%${right}s\n" && echo ${l// /$chr})"
            ;;
        r)

            echo "$(printf -vl "%${width}s\n" && echo ${l// /$chr})$@"
            ;;
        *)
            # l by default
            echo "$@$(printf -vl "%${width}s\n" && echo ${l// /$chr})"
            ;;
    esac

}



SHLIB_LOG_VERBOSE=1
SHLIB_LOG_FORMAT='[$(date +"%Y-%m-%d %H:%M:%S")] $level $title $mes'

die()
{
    err "$@" >&2
    exit 1
}
die_empty()
{
    if test -z "$1"
    then
        shift
        die empty: "$@"
    fi
}

set_verbose()
{
    SHLIB_LOG_VERBOSE=${1-1}
}

log()
{
    local color="$1"
    local title="$2"
    local level="$_LOG_LEVEL"
    shift
    shift

    local mes="$@"
    local NC="$(tput sgr0)"

    if [ -t 1 ]; then
        title="${color}${title}${NC}"
        level="${color}${level}${NC}"
    fi
    eval "echo \"$SHLIB_LOG_FORMAT\""
}
dd()
{
    debug "$@"
}
debug()
{
    if [ ".$SHLIB_LOG_VERBOSE" = ".1" ]; then
        local LightCyan="$(tput bold ; tput setaf 6)"
        _LOG_LEVEL=DEBUG log "$LightCyan" "$@" >&2
    fi
}
info()
{
    local Brown="$(tput setaf 3)"
    _LOG_LEVEL=" INFO" log "$Brown" "$@"
}
ok() {
    local Green="$(tput setaf 2)"
    _LOG_LEVEL="   OK" log "${Green}" "$@"
}
err() {
    local Red="$(tput setaf 1)"
    _LOG_LEVEL="ERROR" log "${Red}" "$@"
}

git_hash()
{
    git rev-parse $1 \
        || die "'git_hash $@'"
}
git_is_merge()
{
    test $(git cat-file -p "$1" | grep "^parent " | wc -l) -gt 1
}
git_parents()
{
    git rev-list --parents -n 1 ${1-HEAD} | { read self parents; echo $parents; }
}
git_rev_list()
{
    # --parents
    # print parent in this form:
    #     <commit> <parent-1> <parent-2> ..

    git rev-list \
        --reverse \
        --topo-order \
        --default HEAD \
        --simplify-merges \
        "$@" \
        || die "'git rev-list $@'"
}
git_tree_hash()
{
    git rev-parse "$1^{tree}"
}
git_ver()
{
    local git_version=$(git --version | awk '{print $NF}')
    local git_version_1=${git_version%%.*}
    local git_version_2=${git_version#*.}
    git_version_2=${git_version_2%%.*}

    printf "%03d%03d" $git_version_1 $git_version_2
}
git_working_root()
{
    git rev-parse --show-toplevel
}

git_rev_exist()
{
    git rev-parse --verify --quiet "$1" >/dev/null
}

git_branch_default_remote()
{
    local branchname=$1
    git config --get branch.${branchname}.remote
}
git_branch_default_upstream_ref()
{
    local branchname=$1
    git config --get branch.${branchname}.merge
}
git_branch_default_upstream()
{
    git rev-parse --abbrev-ref --symbolic-full-name "$1"@{upstream}

    # OR
    # git_branch_default_upstream_ref "$@" | sed 's/^refs\/heads\///'
}
git_branch_exist()
{
    git_rev_exist "refs/heads/$1"
}

git_head_branch()
{
    git symbolic-ref --short HEAD
}

git_commit_date()
{

    # git_commit_date author|commit <ref> [date-format]

    # by default output author-date
    local what_date="%ad"
    if [ "$1" = "commit" ]; then
        # commit date instead of author date
        what_date="%cd"
    fi
    shift

    local ref=$1
    shift

    local fmt="%Y-%m-%d %H:%M:%S"
    if [ "$#" -gt 0 ]; then
        fmt="$1"
    fi
    shift

    git log -n1 --format="$what_date" --date=format:"$fmt" "$ref"
}
git_commit_copy()
{
    # We're going to set some environment vars here, so
    # do it in a subshell to get rid of them safely later
    dd copy_commit "{$1}" "{$2}" "{$3}"
    git log -1 --pretty=format:'%an%n%ae%n%ad%n%cn%n%ce%n%cd%n%s%n%n%b' "$1" |
    (
    read GIT_AUTHOR_NAME
    read GIT_AUTHOR_EMAIL
    read GIT_AUTHOR_DATE
    read GIT_COMMITTER_NAME
    read GIT_COMMITTER_EMAIL
    read GIT_COMMITTER_DATE
    export  GIT_AUTHOR_NAME \
        GIT_AUTHOR_EMAIL \
        GIT_AUTHOR_DATE \
        GIT_COMMITTER_NAME \
        GIT_COMMITTER_EMAIL \
        GIT_COMMITTER_DATE

    # (echo -n "$annotate"; cat ) |

    git commit-tree "$2" $3  # reads the rest of stdin
    ) || die "Can't copy commit $1"
}

git_object_type()
{
    # $0 ref|hash
    # output "commit", "tree" etc
    git cat-file -t "$@" 2>/dev/null
}
git_object_add_by_commit_path()
{
    # add an blob or tree object to target_path in index
    # the object to add is specified by commit and path
    local target_path="$1"
    local src_commit="$2"
    local src_path="$3"

    local src_dir="$(dirname "$src_path")/"
    local src_name="$(basename "$src_path")"
    local src_treeish="$(git rev-parse "$src_commit:$src_dir")"

    git_object_add_by_tree_name "$target_path" "$src_treeish" "$src_name"

}
git_object_add_by_tree_name()
{
    # add an blob or tree object to target_path in index
    local target_path="$1"
    local src_treeish="$2"
    local src_name="$3"

    dd "arg: target_path: ($target_path) src_treeish: ($src_treeish) src_name: ($src_name)"

    local target_dir="$(dirname $target_path)/"
    local target_fn="$(basename $target_path)"
    local treeish

    if [ -z "$src_name" ] || [ "$src_name" = "." ] || [ "$src_name" = "./" ]; then
        treeish="$src_treeish"
    else
        treeish=$(git ls-tree "$src_treeish" "$src_name" | awk '{print $3}')

        if [ -z "$treeish" ]; then
            die "source treeish not found: in tree: ($src_treeish) name: ($src_name)"
        fi
    fi

    dd "hash of object to add is: $treeish"

    if [ "$(git_object_type $treeish)" = "blob" ]; then
        # the treeish imported is a file, not a dir
        # first create a wrapper tree or replace its containing tree

        dd "object to add is blob"

        local dir_treeish
        local target_dir_treeish="$(git rev-parse "HEAD:$target_dir")"
        if [ -n "target_dir_treeish" ]; then
            dir_treeish="$(git rev-parse "HEAD:$target_dir")"
            dd "target dir presents: $target_dir"

        else
            dd "target dir absent"
            dir_treeish=""
        fi

        treeish=$(git_tree_add_blob "$dir_treeish" "$target_fn" $src_treeish $src_name) || die create wrapper treeish
        target_path="$target_dir"

        dd "wrapper treeish: $treeish"
        dd "target_path set to: $target_path"
    else
        dd "object to add is tree"
    fi

    git_treeish_add_to_prefix "$target_path" "$treeish"
}

git_treeish_add_to_prefix()
{
    local target_path="$1"
    local treeish="$2"

    dd treeish content:
    git ls-tree $treeish

    git rm "$target_path" -r --cached || dd removing target "$target_path"

    if [ "$target_path" = "./" ]; then
        git read-tree "$treeish" \
            || die "read-tree $target_path $treeish"
    else
        git read-tree --prefix="$target_path" "$treeish" \
            || die "read-tree $target_path $treeish"
    fi
}

git_tree_add_tree()
{
    # output new tree hash in stdout
    # treeish can be empty
    local treeish="$1"
    local target_fn="$2"
    local item_hash="$3"
    local item_name="$4"

    {
        if [ -n "$treeish" ]; then
            git ls-tree "$treeish" \
                | fgrep -v "	$item_name"
        fi

        cat "040000 tree $item_hash	$target_fn"
    } | git mktree
}
git_tree_add_blob()
{
    # output new tree hash in stdout
    # treeish can be empty
    local treeish="$1"
    local target_fn="$2"
    local blob_treeish="$3"
    local blob_name="$4"

    {
        if [ -n "$treeish" ]; then
            git ls-tree "$treeish" \
                | fgrep -v "	$target_fn"
        fi

        git ls-tree "$blob_treeish" "$blob_name" \
            | awk -v target_fn="$target_fn" -F"	" '{print $1"	"target_fn}'
    } | git mktree
}

git_workdir_save()
{
    local index_hash=$(git write-tree)

    # add modified file to index and read index tree
    git add -u
    local working_hash=$(git write-tree)

    # restore index tree
    git read-tree $index_hash

    echo $index_hash $working_hash
}
git_workdir_load()
{
    local index_hash=$1
    local working_hash=$2

    git_object_type $index_hash || die "invalid index hash: $index_hash"
    git_object_type $working_hash || die "invalid workdir hash: $working_hash"

    # First create a temp commit to restore working tree.
    #
    # git-read-index to index and git-reset does not work because deleted file in
    # index does not apply to working tree.
    #
    # But there is an issue with this:
    #   git checkout --orphan br
    #   git_workdir_load
    # would fails, because ORIG_HEAD is not a commit.

    local working_commit=$(echo "x" | git commit-tree $working_hash) || die get working commit
    git reset --hard $working_commit || die reset to tmp commit
    git reset --soft ORIG_HEAD || die reset to ORIG_HEAD
    git read-tree $index_hash || die "load saved index tree from $index_hash"
}
git_workdir_is_clean()
{
    local untracked="$1"
    if [ "$untracked" == "untracked" ]; then
        [ -z "$(git status --porcelain)" ]
    else
        [ -z "$(git status --porcelain --untracked-files=no)" ]
    fi
}

git_copy_commit()
{
    git_commit_copy "$@"
}

git_diff_ln_new()
{
    # output changed line number of a file: <from> <end>; inclusive:
    # 27 28
    # 307 309
    # 350 350
    #
    # Usage:
    #
    #   diff working tree with HEAD:
    #       git_diff_ln_new HEAD -- <fn>
    #
    #   diff working tree with staged:
    #       git_diff_ln_new -- <fn>
    #
    #   diff staged(cached) with HEAD:
    #       git_diff_ln_new --cached -- <fn>
    #
    # in git-diff output:
    # for add lines:
    # @@ -53 +72,8
    #
    # for remove lines:
    # @@ -155 +179,0

    git diff -U0 "$@" \
        | grep '^@@' \
        | awk '{

    # @@ -155 +179,0
    # $1 $2   $3

    l = $3
    gsub("^+", "", l)

    # add default offset: ",1"
    split(l",1", x, ",")

    # inclusive line range:
    x[2] = x[1] + x[2] - 1

    # line remove format: @@ -155, +179,0
    # do need to output line range for removed.
    if (x[2] >= x[1]) {
        print x[1] " " x[2]
    }

}'
}




os_detect()
{
    local os
    case $(uname -s) in
        Linux)
            os=linux ;;
        *[bB][sS][dD])
            os=bsd ;;
        Darwin)
            os=mac ;;
        *)
            os=unix ;;
    esac
    echo $os
}

mac_ac_power_connection()
{
    #  Connected: (Yes|No)
    system_profiler SPPowerDataType \
        | sed '1,/^ *AC Charger Information:/d' \
        | grep Connected:
}


mac_power()
{

    # $0 is-battery          exit code 0 if using battery.
    # $0 is-ac-power         exit code 0 if using ac power.

    local cmd="$1"
    local os=$(os_detect)

    if [ "$os" != "mac" ]; then
        err "not mac but: $os"
        return 1
    fi

    case $cmd in

        is-battery)
            mac_ac_power_connection | grep -q No
            ;;

        is-ac-power)
            mac_ac_power_connection | grep -q Yes
            ;;

        *)
            err "invalid cmd: $cmd"
            return 1
            ;;
    esac
}

fn_match()
{
    # $0 a.txt *.txt
    case "$1" in
        $2)
            return 0
            ;;
    esac
    return 1
}

new_branch=tmpbranch
merge=""
remove=0
orphan=0
to_strip=0

while getopts "b:m:p:s:orh" opname; do
    case $opname in
        b)
            new_branch=$OPTARG
            ;;
        m)
            merge=$OPTARG
            remove=1
            ;;
        o)
            orphan=1
            ;;
        r)
            remove=1
            ;;
        s)
            to_strip="$OPTARG"
            ;;
        h)
            usage
            exit 0
            ;;
        [?])
            echo "unknown option: ($opname)"
            usage
            exit 1
            ;;
    esac
done
shift $(($OPTIND - 1))

if [ ".$2" != ".--" ]; then
    usage
    exit 1
fi

range=$1
shift
shift
fns="$@"

base=${range%..*}
if [ ".$base" = ".$range" ]; then
    usage
    exit 1
fi

tip=${range#*..}
if [ ".$tip" = "." ]; then
    tip=HEAD
fi

head=$(git symbolic-ref --short HEAD 2>/dev/null)
if [ ".$head" = "." ]; then
    head=$(git rev-parse HEAD)
fi

tip_hash=$(git_hash $tip)
new_hash=
merged_hash=


extract()
{
    local parent=$1
    local tree=
    local commit=

    for c in $(git log --reverse $base..$tip_hash --format="%H" -- $fns); do

        info "== to pick $fns from:"
        git --no-pager log --color --stat --decorate --pretty=oneline --abbrev-commit -M -n1 $c
        info "== END =="

        # mes=$(git log -n1 --format="%B" $c)

        # force prefix when diff, since .gitconfig could hide prefix
        git diff --src-prefix="" --dst-prefix="" --binary $c~ $c -- $fns | git apply -p $to_strip --index

        info "== after patch $c =="
        git status --short
        info "== END =="

        # manually commit index

        tree=$(git write-tree)

        dd "tree to commit: $tree"
        git ls-tree -r $tree
        dd "end of tree"

        dd parent: $parent

        commit=$(git_copy_commit "$c" "$tree" "-p $parent")
        dd "new commit:" $commit

        git reset --hard $commit --

        parent=$commit
    done
    return 0
}

build_new_branch()
{
    info "== build_new_branch =="

    git checkout --orphan $new_branch || die checkout --orpahn $new_branch

    if [ "$orphan" = "1" ]; then
        git reset --hard
        git commit --allow-empty -m "init-root" || die commit init-root
    else
        info reset $new_branch to $base
        git reset --hard $base --
    fi

    git clean -dxf || die clean -dxf

    info "HEAD hash: $(git_hash HEAD)"

    extract $(git_hash HEAD) \
        && new_hash=$(git_hash HEAD) \
        && \
        if [ "$new_branch" = "tmpbranch" ]; then
            git checkout --detach HEAD \
                && git branch -d tmpbranch
        else
            :
        fi \
        && info "== new branch $new_branch hash is $new_hash == "
}

discard_changes()
{
    info "== discard_changes =="

    # git checkout would fail if one of the files does not exists.
    # thus newly added file impedes the whole process.
    # git checkout ab12345 -- nonexistent

    # local script='for f in '"$fns"'; do git rm -r -f $f; git checkout '"$base"' -- $f; done'
    local script='for f in '"$fns"'; do echo "== $f =="; git rm -r -f $f; git status -- $f >&2; git checkout '"$base"' -- $f; git status -- $f >&2; done'

    git checkout --detach $tip_hash \
        && info "== discard from $tip ==" \
        && git filter-branch -f --prune-empty --tree-filter "$script" $base..HEAD \
        && [ $(git log $base..HEAD -- "$fns" | wc -l | awk '{print $1}') = 0 ] \
        && info "== confirmed that no changes through $base..HEAD"
    git status --short
    info "== END =="
    info "== history =="
    git --no-pager log --color --stat --decorate --pretty=oneline --abbrev-commit -M $base..HEAD
    info "== END =="
}

merge_or_rebase()
{
    local action=$1

    if [ "$action" == "merge" ]; then
        info "== merge splitted branch back to $tip ==" \
            && \
            if [ ".$new_branch" = "tmpbranch" ]; then
                git merge --commit --no-edit --no-ff $new_hash
            else
                git merge --commit --no-edit --no-ff $new_branch
            fi \
                && merged_hash=$(git_hash HEAD) \
                && info "merged hash: $merged_hash"

    elif [ "$action" == "rebase" ]; then

        info "== rebase $tip onto $new_branch ($new_hash) =="

        git rebase $new_hash || return 1
        merged_hash=$(git_hash HEAD)

        info "rebased hash: $merged_hash"
    else
        die "invalid action: ($action)"
    fi
}

is_same_as_original()
{
    info "== diff with original $tip =="
    git diff $tip || return 1
    ok "Current HEAD is same as $tip"
}

update_tip_ref()
{
    info "remove tmp branch: refs/original/HEAD"
    git update-ref -d refs/original/HEAD

    # if it is a reference(branch, tag), not hash or HEAD

    if [ ".$tip" = ".HEAD" ]; then
        info "git update-ref $tip $merged_hash"
        git update-ref $tip $merged_hash \
            && info "== $tip set to $merged_hash"
    fi

    if git rev-parse refs/heads/$tip >/dev/null 2>/dev/null;  then
        info "git update-ref refs/heads/$tip $merged_hash"
        git update-ref refs/heads/$tip $merged_hash \
            && info "== $tip set to $merged_hash"
    fi

}


info "split $fns from $range"

build_new_branch ||  die build_new_branch

if [ "$remove" = "0" ]; then
    exit 0;
fi

discard_changes || die discard_changes

if [ "$merge" = "" ]; then
    exit 0
fi

merge_or_rebase "$merge" || die merge_or_rebase

is_same_as_original || die is_same_as_original

update_tip_ref || die update_tip_ref

git checkout $tip >/dev/null 2>/dev/null || die git checkout $tip

info "== Done =="