path: root/sib

#!/usr/bin/env bash
# Copyright 2026 Hee-Suk Kim <[email protected]>
# SPDX-License-Identifier: GPL-3.0-only

set -o pipefail

: "${SIB_DIR:=$HOME/.sib}"
export SIB_DIR

readonly TRACE_MAGIC=_SIBTRACE
readonly TRACE_VERSION=v1
readonly DEFAULT_PLM_MODEL=gpt-5.6-luna
readonly DEFAULT_PLM_ENDPOINT=https://api.openai.com/v1/responses
readonly COMMIT_DATE='999999999 +0000'

die() {
    [ $# -eq 0 ] || echo fatal: "$@" >&2
    exit 1
}

lgit() {
    env GIT_AUTHOR_NAME='Sibyl' \
        GIT_COMMITTER_NAME='Sibyl' \
        GIT_AUTHOR_EMAIL='sib@local' \
        GIT_COMMITTER_EMAIL='sib@local' \
        git --git-dir="$SIB_DIR" "$@"
}

cmd_init() {
    [[ $# -eq 0 ]] || die 'no argument is allowed'

    local reinit=
    [[ -f "$SIB_DIR/HEAD" ]] && reinit=true

    mkdir -p "$SIB_DIR"
    lgit init --bare --quiet

    if [[ -z $reinit ]]; then
        echo 'ref: refs/conv/scratch' > "$SIB_DIR/HEAD"
        sib config set sib.model $DEFAULT_PLM_MODEL
        sib config set sib.endpoint $DEFAULT_PLM_ENDPOINT
        echo "Initialized empty Sib store in $SIB_DIR" >&2
    else
        echo "Reinitialized existing Sib store in $SIB_DIR" >&2
    fi
}

# hash and format blob (to mktree format)
# $1: blob name
hf_blob() {
    [ -n "$1" ] || return 1;
    printf '100644 blob %s\t%s\n' $(lgit hash-object -w --stdin) "$1"
}

key_is_valid() {
    [[ $1 =~ ^[A-Za-z_][A-Za-z0-9._-]*$ ]]
}

tf_key() {
    local line key hash extra n=0

    while IFS= read -r line || [[ -n $line ]]; do
        n=$((n + 1))

        IFS=$'\t' read -r hash key <<<"$line"

        [[ -n $key && -n $hash ]] || {
            echo "invalid trace entry at line $n" >&2
            return 1
        }

        key_is_valid "$key" || {
            echo "invalid trace key at line $n: $key" >&2
            return 1
        }

        echo "$key"
    done
}

# chain-trace [-p <chain>] <trace>
cmd_chain_trace() {
    local parent_args=() trace=

    if [[ $# -eq 1 ]]; then
        trace="$1"
    elif [[ $# -eq 3 && "$1" == "-p" ]]; then
        cmd_ls_trace $2 >/dev/null # check if parent is valid
        parent_args=(-p "$2")
        trace="$3"
    else
        die "Usage: chain-trace [-p <chain>] <trace>"
    fi

    # check if trace is valid
    cmd_ls_trace $trace >/dev/null || die trace $trace is invalid

    GIT_AUTHOR_DATE="$COMMIT_DATE" \
    GIT_COMMITTER_DATE="$COMMIT_DATE" \
    lgit commit-tree "${parent_args[@]}" $trace </dev/null
}

cmd_mktrace() {
    [[ $# -eq 0 ]] || die 'no argument is allowed'
    local lines=$(cat) line key hash

    if [[ -n $lines ]]; then
        keys=$(printf '%s\n' "$lines" | tf_key) || die 'invalid trace format'

        printf '%s\n' "$keys" | sort | uniq -d | grep -q . &&
            die 'duplicate key found in input'

        while IFS= read -r key; do
            case "$key" in
                "$TRACE_MAGIC"*) die "reserved key prefix: $key" ;;
            esac
        done <<<"$keys"
    fi

    {
        echo -n $TRACE_VERSION | hf_blob $TRACE_MAGIC;

        if [[ -n $lines ]]; then
            while IFS=$'\t' read -r hash key; do
                printf '100644 blob %s\t%s\n' "$hash" "$key"
            done <<<"$lines"
        fi
    } | lgit mktree || die Invalid mktrace format
}

cmd_ls_trace() {
    [[ $# -eq 1 ]] || die 'usage: ls-trace <trace>'

    local tmp keys magic version

    tmp="$(lgit ls-tree "$1" 2>/dev/null)" || die "invalid object name $1"
    tmp=$(awk 'NF==4 { OFS="\t"; print $3, $4 }' <<<"$tmp")

    # Now $tmp is ls-trace format before screening

    keys=$(tf_key <<<"$tmp") || die 'invalid trace format'

    printf '%s\n' "$keys" | sort | uniq -d | grep -q . &&
            die "duplicate key found in trace $1"

    magic=$(awk -F"\t" -v k=$TRACE_MAGIC '$2 == k { print $1 }' <<<"$tmp")
    [[ -z $magic ]] && die 'No magic found on trace'

    # Here it is ensured $magic is valid hash from ls-tree and single
    # line by upper dedup check
    version=$(lgit cat-file blob $magic)
    [[ $version != $TRACE_VERSION ]] && die "invalid trace version: $version"

    # Here, every key starting with $TRACE_MAGIC will be considered as
    # non-key and internal indicator. NF >= 2 is to not print \t for
    # empty input
    awk -F"\t" -v k=$TRACE_MAGIC -v OFS='\t' \
        'NF >= 2 && index($2, k) != 1 { print $1, $2 }' <<<"$tmp" ||
        true # return 0 for empty trace
}

cmd_git() {
    lgit "$@"
}

# hash and format blob (to trace format)
# $1: blob name
tf_blob() {
    local key=$1 hash

    [[ -n $key ]] || return 1
    hash=$(lgit hash-object -w --stdin) || return
    printf '%s\t%s\n' "$hash" "$key"
}

# Input: {"key": "value", "key": "value", ...}
cmd_mktrace_json() {
    [[ $# -eq 0 ]] || die 'no argument is allowed'

    local json="$(cat)"

    # Reject null/array/object/non-object
    jq -e 'type == "object" and all(.[]; type == "string")' >/dev/null <<<"$json" \
        || die 'input must be a JSON object containing only string values'

    tr=$(jq -r 'keys_unsorted[]' <<<"$json" | while IFS= read -r key; do
             [[ -z $key ]] && die empty key detacted on JSON
             jq -j --arg k "$key" '.[$k]' <<<"$json" | tf_blob "$key"
         done) || die
    cmd_mktrace <<<"$tr"
}

# print single trace object to one-line json (jsonl)
cmd_ls_trace_json() {
    [[ $# -eq 1 ]] || die 'usage: ls-trace-json <trace>'

    local entries
    entries=$(cmd_ls_trace "$1") || exit 1
    [ -n "$entries" ] || { echo '{}'; return 0; }
    echo "$entries" | while IFS=$'\t' read -r hash key; do
        lgit cat-file blob "$hash" \
            | jq -Rsc --arg k "$key" '{($k): .}'
    done | jq -cs 'add // {}'
}

cmd_chain_trace_json() {
    local tmp

    tmp="$(cmd_mktrace_json)" || die

    cmd_chain_trace "$@" "$tmp"
}

cmd_chain_jsonl() {
    [[ $# -eq 0 ]] || die 'no argument is allowed'

    local line trace chain parent= n=0
    while IFS= read -r line || [ -n "$line" ]; do
        [ -n "$line" ] || continue          # skip blank lines
        trace=$(printf '%s' "$line" | cmd_mktrace_json) \
            || die "mktrace failed at line $((n+1))"
        if [ -z "$parent" ]; then
            chain=$(cmd_chain_trace "$trace")
        else
            chain=$(cmd_chain_trace -p "$parent" "$trace")
        fi
        parent="$chain"
        n=$((n+1))
    done

    [ "$n" -gt 0 ] || die 'empty input'
    echo "$chain"
}

cmd_ls_chain() {
    [[ $# -eq 1 ]] || die 'usage: ls-chain <chain-ish>'

    lgit cat-file commit "$1" | awk '
        /^$/ { exit }                          # stop at blank line (msg starts)
        $1 == "tree"   { print "trace " $2 }
        $1 == "parent" { print "parent " $2 }
    '
}

tf_value() {
    cut -d' ' -f2
}

trace_of_chain() {
    cmd_ls_chain "$1" | grep '^trace ' | tf_value
}

cmd_rev_list() {
    lgit rev-list --reverse "$@"
}

cmd_rev_jsonl() {
    [[ $# -eq 1 ]] || die 'usage: rev-jsonl <chain-ish>'

    local commits
    commits=$(cmd_rev_list "$1") || die "rev-list failed for $1"
    while read -r commit; do
        local trace
        trace=$(trace_of_chain "$commit") || die "no trace for $commit"
        cmd_ls_trace_json "$trace"
    done <<< "$commits"
}

cmd_rev_parse() {
    [ $# -eq 1 ] || die 'usage: rev-parse <chain-ish>'

    # Refuse to read refs/conv/HEAD over HEAD.
    # To read it, use rev-parse conv/HEAD
    if [[ $1 != "HEAD" ]] &&
           lgit rev-parse --quiet --verify --end-of-options \
                "refs/conv/$1^{commit}" 2>/dev/null; then
        return 0
    fi

    lgit rev-parse --verify --end-of-options "$1^{commit}"
}

#
# Refs
#

cmd_update_ref() {
    lgit update-ref --create-reflog "$@"
}

cmd_show_ref() {
    [ $# -eq 1 ] || die 'usage: show-ref <ref>'
    lgit show-ref --verify --hash "$1"
}

cmd_ls_refs() {
    local prefix=${1:-refs/}
    lgit for-each-ref --format='%(refname) %(objectname)' "$prefix"
}

cmd_symbolic_ref() {
    lgit symbolic-ref "$@"
}

#
# dispatcher
#

usage() {
    cat <<EOF >&2
usage: sib <ask|log|show|switch|save> [-h ...]
       sib edit <chain-ish>
       sib status
       sib init
EOF
}

cmd_config() {
    lgit config "$@"
}

[[ $# -lt 1 ]] && { usage; die; }

cmd="$1"; shift

if [[ $cmd != "init" ]]; then
    lgit rev-parse --resolve-git-dir "$SIB_DIR" &>/dev/null \
        || die "$SIB_DIR: not a sib storage: run \`sib init'"
fi

# We symlink /usr/bin/sib -> ../libexec/sib-core
# script_dir need to be realpath ../libexec/sib-core
script_dir="$(dirname -- "$(readlink -f -- "${BASH_SOURCE[0]}")")"


cmd_version() {
    echo sib version $(cat $script_dir/VERSION)
}

export PATH="$script_dir:$PATH"

# Priority: Env > config > Default
: "${PLM_MODEL:=$(cmd_config get sib.model 2>/dev/null)}"
: "${PLM_MODEL:=$DEFAULT_PLM_MODEL}"

: "${PLM_ENDPOINT:=$(cmd_config get sib.endpoint 2>/dev/null)}"
: "${PLM_ENDPOINT:=$DEFAULT_PLM_ENDPOINT}"

export PLM_MODEL PLM_ENDPOINT

fn="cmd_${cmd//-/_}"
if declare -F "$fn" > /dev/null; then
    "$fn" "$@"
elif command -v "sib-$cmd" >/dev/null; then
    exec "sib-$cmd" "$@"
else
    echo sib: unknown command $cmd >&2
    echo >&2
    usage
    die
fi