AutoEMC
为 ProjectE 自动生成安全且准确的 EMC 值。
AutoEMC 会在启动时扫描整合包中的物品注册表和配方,自动填补 ProjectE 缺失的 EMC 值,让你不再受困于满屏“无法学习”物品的知识表。它专为大型厨房 sink 整合包设计,在这些整合包中手动为每个物品定价是不现实的。
功能介绍
- 优先从配方推导 EMC。 AutoEMC 会遍历已加载的配方,在猜测之前先根据已知输入推导出 EMC 值,从而尽可能保持定价的内部一致性。
- 无配方时的智能降级。 对于没有配方路径的物品,采用以下方式进行定价:
- 直接从 ProjectE Integration 随附的数据中提取捆绑的基础值(涵盖 Botania、EnderIO 及其他数十个模组)。
- 基于材质/标签的降级方案(如矿石、锭、粉、方块、原矿形态等),与
forge、mekanism、create 以及其他可配置的标签命名空间进行匹配。
- 以稀有度为基础的降级作为最后手段,并设有反向配方保底机制,确保廉价原材料的价值不会低于由其加工而成的成品。
- 理解带有 NBT 标签的物品。 提供可插拔的 NBT 材质解析器框架(内置支持 Avaritia 奇点等功能),根据物品内嵌的材质对 NBT 变体物品进行定价,而不是将它们笼统地归为一个通用价值。这既可以作为预扫描阶段运行,也可以作为按需解析的实时 ProjectE 转换源。
- 防止经济系统自我崩溃。 多个安全层协同工作:
- 仅限创造模式且无配方的技术性物品被锁定为
1 EMC,以防万一泄露到生存模式中成为无限价值的廉价来源。
- 大波动保护机制会阻止任何单次运行导致已定价物品价值变化超过可配置比例的情况,并将其记录到隔离报告中,而不是静默应用。
- 所有权仲裁器逐项决定 AutoEMC 是继续追踪某个值,还是将其移交给 ProjectE 或整合包作者的手动条目——从而尊重合法的覆盖项,同时防止因 ProjectE 仅仅回显 AutoEMC 的上一个值而导致所有权(及条目)被删除。
- 批量处理与自适应。 即使整合包包含数千种物品,每次扫描阶段也会以自适应大小的小块跨 Tick 运行,避免在启动时冻结服务器。
- 透明化。 每个生成的值都会记录其来源(
baseline、tag-material、rarity-fallback、reverse-recipe-floor、creative-no-recipe 等),并且模组会在每次扫描后输出诊断报告:
generated_entries.json —— AutoEMC 当前拥有的所有条目
underpriced_inputs_report.json —— 相对于其用于制作的物品而言定价过低的非拥有物品
quarantined_drops.json —— 因看起来不安全而被阻止的价值变更
依赖要求
配置亮点
以下内容均可在 autoemc-common.toml 中进行配置:
scanNbtVariants —— 扫描 NBT 标签物品变体(默认开启)
useNbtMaterialResolvers / useLiveNbtMaterialResolution —— 启用基于材质的 NBT 定价,包括预烘焙和/或实时模式
giveCreativeNoRecipeItemsOneEmc —— 切换创造模式无配方物品的安全底线(默认开启)
largeDropProtectionThreshold / maxAllowedDropRatio —— 调整大波动保护阈值
ownershipSmallDifferenceRatio —— AutoEMC 与 ProjectE 之间的差异达到多少被视为“仅是噪音”
tagFallbackNamespaces —— 在材质降级检查哪些标签命名空间
consistencyWarnMinRatio / consistencyWarnMinAbsoluteDelta —— 低价输入报告的阈值
注意事项
- 除非 AutoEMC 确信更改是安全的,否则它永远不会覆盖由 ProjectE 或整合包作者正确定价的值。
- 如果你正在审计整合包的经济系统,
config/autoemc/ 中的调试日志和 JSON 报告是最快查看具体定价内容、方式及原因的方法。
AutoEMC
Automatic, safe EMC value generation for ProjectE.
AutoEMC scans your modpack's item registry and recipes on startup and fills in missing ProjectE EMC values automatically, so you're not stuck with a Table of Existence full of "unlearnable" items. It's built for large kitchen-sink modpacks where manually pricing every item is unrealistic.
What it does
- Solves EMC from recipes first. AutoEMC walks your loaded recipes and derives EMC values from known inputs before ever guessing, so pricing stays internally consistent wherever possible.
- Falls back intelligently when there's no recipe. Items with no recipe path are priced using:
- Bundled baseline values pulled directly from ProjectE Integration's shipped data (Botania, EnderIO, and dozens of other mods).
- Material/tag-based fallback (ore, ingot, dust, block, raw forms, etc.), matched against
forge, mekanism, create, and other configurable tag namespaces.
- Rarity-based fallback as a last resort, with a reverse-recipe floor so cheap raw materials can't end up worth less than the processed goods made from them.
- Understands NBT-tagged items. A pluggable NBT material resolver framework (with built-in support for things like Avaritia singularities) prices NBT-variant items based on their embedded material, instead of lumping them all into one generic value. This can run both as a pre-scan pass and as a live ProjectE conversion source that resolves on demand.
- Protects the economy from itself. Several safety layers work together:
- Creative-only and no-recipe technical items are locked to
1 EMC so they can never become a cheap source of infinite value if they somehow leak into survival.
- A large-drop guard blocks any single-run change that would crash an already-priced item's value by more than a configurable ratio, and logs it to a quarantine report instead of silently applying it.
- An ownership arbiter decides, item by item, whether AutoEMC should keep tracking a value or hand it off to ProjectE/a pack author's manual entry — so legitimate overrides are respected but ProjectE simply echoing AutoEMC's own last value can't cause ownership (and the entry) to be deleted.
- Batched and adaptive. Every scan phase runs in small, adaptively-sized chunks across ticks instead of freezing the server on startup, even on packs with thousands of items.
- Transparent. Every generated value is logged with its source (
baseline, tag-material, rarity-fallback, reverse-recipe-floor, creative-no-recipe, etc.), and the mod writes out diagnostic reports each scan:
generated_entries.json — everything AutoEMC currently owns
underpriced_inputs_report.json — non-owned items that look priced too low relative to what they're used to craft
quarantined_drops.json — value changes that were blocked for looking unsafe
Requirements
- Minecraft 1.16.5, Forge
- ProjectE (required)
- ProjectE Integration (required — AutoEMC's bundled baseline data is sourced from it, and it must be present for those integrations to actually function in-game)
Configuration highlights
All of the following are configurable in autoemc-common.toml:
scanNbtVariants — scan NBT-tagged item variants (on by default)
useNbtMaterialResolvers / useLiveNbtMaterialResolution — enable material-aware NBT pricing, pre-baked and/or live
giveCreativeNoRecipeItemsOneEmc — toggle the creative-item safety floor (on by default)
largeDropProtectionThreshold / maxAllowedDropRatio — tune the large-drop guard
ownershipSmallDifferenceRatio — how much difference between AutoEMC's and ProjectE's value counts as "just noise"
tagFallbackNamespaces — which tag namespaces are checked during material fallback
consistencyWarnMinRatio / consistencyWarnMinAbsoluteDelta — thresholds for the underpriced-input report
Notes
- AutoEMC never overwrites a value that's already correctly priced by ProjectE or a pack author unless it's confident the change is safe.
- If you're auditing a pack's economy, the debug log and JSON reports in
config/autoemc/ are the fastest way to see exactly what was priced, how, and why.
评论 0
登录后参与评分
还没有评论,来抢沙发吧