build.sh

#!/usr/bin/env bash
#
# 构建 pgvector-zh 镜像：
#   1. 自动下载 SCWS / zhparser 源码到 downloads/（做 SHA256 校验；文件已存在且校验通过则跳过下载）
#   2. 执行 docker build
#
# 可用环境变量覆盖默认值（覆盖下载地址时，请同步更新对应 SHA256，或将 SHA256 置空跳过校验）：
#   IMAGE_NAME   镜像名，默认 pgvector-zh
#   IMAGE_TAG    镜像标签，默认 0.8.2-pg18
#   SCWS_URL     SCWS 源码包下载地址
#   SCWS_SHA256  SCWS 源码包校验值
#   ZHPARSER_URL     zhparser 源码包下载地址
#   ZHPARSER_SHA256  zhparser 源码包校验值

set -euo pipefail

cd "$(dirname "$0")"

readonly SCWS_VERSION="1.2.3"
readonly SCWS_SHA256="${SCWS_SHA256:-60d50ac3dc42cff3c0b16cb1cfee47d8cb8c8baa142a58bc62854477b81f1af5}"
readonly SCWS_URL="${SCWS_URL:-https://www.xunsearch.com/scws/down/scws-${SCWS_VERSION}.tar.bz2}"

# zhparser 固定在 amutu/zhparser master 分支 2025-10-21 的快照
readonly ZHPARSER_COMMIT="02c08378643739a07bdb0a3f64e249d4da7a9d99"
readonly ZHPARSER_SHA256="${ZHPARSER_SHA256:-195845c477ae209fa706efcb807696730344edd88ad3233c28511cc289aa23c5}"
readonly ZHPARSER_URL="${ZHPARSER_URL:-https://github.com/amutu/zhparser/archive/${ZHPARSER_COMMIT}.zip}"

readonly IMAGE_NAME="${IMAGE_NAME:-pgvector-zh}"
readonly IMAGE_TAG="${IMAGE_TAG:-0.8.2-pg18}"

readonly DOWNLOAD_DIR="downloads"

# verify_sha256 <file> <expected>：兼容 macOS（shasum）与 Linux（sha256sum）
verify_sha256() {
    local file="$1" expected="$2" actual=""
    if command -v sha256sum >/dev/null 2>&1; then
        actual="$(sha256sum "$file" | awk '{print $1}')"
    else
        actual="$(shasum -a 256 "$file" | awk '{print $1}')"
    fi
    [[ "$actual" == "$expected" ]]
}

# fetch <url> <output> <sha256>（sha256 为空时跳过校验）
fetch() {
    local url="$1" out="$2" sha256="$3"

    if [[ -f "$out" ]]; then
        if [[ -z "$sha256" ]] || verify_sha256 "$out" "$sha256"; then
            echo "== 已存在且校验通过，跳过下载: $out"
            return
        fi
        echo "== 已存在但校验不通过，重新下载: $out"
        rm -f "$out"
    fi

    echo "== 下载: $url"
    if ! curl --fail --location --retry 3 --connect-timeout 15 -o "$out" "$url"; then
        echo "下载失败: $url" >&2
        echo "提示: 网络受限环境可设置 https_proxy 等标准代理环境变量后重试" >&2
        exit 1
    fi

    if [[ -n "$sha256" ]] && ! verify_sha256 "$out" "$sha256"; then
        echo "SHA256 校验失败: $out" >&2
        exit 1
    fi
}

mkdir -p "$DOWNLOAD_DIR"

fetch "$SCWS_URL" "$DOWNLOAD_DIR/scws-${SCWS_VERSION}.tar.bz2" "$SCWS_SHA256"
fetch "$ZHPARSER_URL" "$DOWNLOAD_DIR/zhparser.zip" "$ZHPARSER_SHA256"

echo "== 构建镜像: ${IMAGE_NAME}:${IMAGE_TAG}"
docker build --tag "${IMAGE_NAME}:${IMAGE_TAG}" .

echo "== 完成: ${IMAGE_NAME}:${IMAGE_TAG}"