SchemaSpy vs SchemaCrawler - Which Database Documentation Tool is Right for You?

SchemaSpy 与 SchemaCrawler —— 哪款数据库文档工具更适合你？

Both SchemaSpy and SchemaCrawler are free, open-source tools for documenting and analysing relational databases over JDBC. Both have been around for over 20 years. Both can generate entity-relationship diagrams. Yet the two tools are more different than they look. Disclosure: I work on SchemaCrawler, so take this with appropriate scepticism. I have tried to represent SchemaSpy fairly. SchemaSpy 和 SchemaCrawler 都是免费的开源工具，用于通过 JDBC 对关系型数据库进行文档记录和分析。两者都已经存在超过 20 年，且都能生成实体关系图（ERD）。然而，这两款工具的差异远比表面看起来要大。披露：我本人参与 SchemaCrawler 的开发，因此请带着审慎的态度阅读本文。我已尽力客观地呈现 SchemaSpy 的特点。

What SchemaSpy Does Best

SchemaSpy 的核心优势

SchemaSpy’s primary strength is its interactive HTML report. After a single run, you get a navigable website: clickable table pages, hyperlinked foreign keys, anomaly reports, and embedded ER diagrams for every table. It is exactly the kind of output you hand to a non-technical stakeholder, a consultant, or a new team member who needs to understand the data model quickly. SchemaSpy also detects implied relationships - potential foreign keys that are not formally declared in the schema. It provides an orphan table page that surfaces tables with no relationships. These are genuinely useful for legacy databases. If your goal is a shareable, browsable report that looks great in a browser, SchemaSpy delivers. SchemaSpy 的主要优势在于其交互式 HTML 报告。只需运行一次，你就能得到一个可导航的网站：包含可点击的表页面、超链接外键、异常报告以及每个表的嵌入式 ER 图。这正是那种适合交付给非技术利益相关者、顾问或需要快速理解数据模型的新团队成员的输出格式。SchemaSpy 还能检测隐含关系——即那些在模式中未正式声明的潜在外键。它提供了一个“孤立表”页面，用于展示没有任何关系的表。这些功能对于遗留数据库非常有用。如果你的目标是生成一份易于分享、可在浏览器中浏览且外观精美的报告，SchemaSpy 是不二之选。

What SchemaCrawler Does Best

SchemaCrawler 的核心优势

SchemaCrawler’s strength is everything a developer needs before and after the report: searching, diffing, linting, scripting, and integration. SchemaCrawler 的优势在于涵盖了开发者在报告生成前后所需的一切功能：搜索、差异对比、代码检查（Linting）、脚本编写以及集成。

• Diff-able text output: SchemaCrawler’s “schema” command produces clean, structured text output - not HTML. Run it against production and staging, diff the outputs in git, and see exactly what changed. This is the foundation of schema change tracking in CI/CD. 可对比的文本输出： SchemaCrawler 的 “schema” 命令生成的是整洁、结构化的文本输出，而非 HTML。将其分别在生产环境和测试环境运行，通过 git 对比输出结果，即可精确查看变更内容。这是 CI/CD 中模式变更追踪的基础。
• Schema lint: The “lint” command catches design problems automatically: missing primary keys, nullable columns in unique constraints, redundant indices, tables with no relationships, and more. No SchemaSpy equivalent exists. 模式检查（Lint）： “lint” 命令可自动捕获设计问题：如缺失主键、唯一约束中包含可空列、冗余索引、无关联表等。SchemaSpy 没有对应的功能。
• Grep - regex search across the entire schema: --grep-tables and --grep-columns let you search all tables, columns, stored procedures, triggers, and foreign keys by regular expression. Find every column referencing a concept across a 500-table database in a single command. Combine it with --parents and --children to pull the related tables automatically. Grep - 全模式正则搜索： 通过 --grep-tables 和 --grep-columns，你可以使用正则表达式搜索所有的表、列、存储过程、触发器和外键。只需一条命令，即可在拥有 500 张表的数据库中找到所有引用某个概念的列。结合 --parents 和 --children 参数，还能自动提取相关的表。
• Multiple output formats: Text, HTML, JSON, CSV, Markdown, and ER diagrams (via Graphviz). The Markdown output is useful for documentation-as-code; the JSON output is useful for tooling. 多种输出格式： 支持文本、HTML、JSON、CSV、Markdown 以及 ER 图（通过 Graphviz）。Markdown 输出适用于“文档即代码”模式；JSON 输出则便于工具集成。
• Schema extension with PlantUML and dbdiagram.io: SchemaCrawler can generate output in PlantUML and dbdiagram.io formats directly from your live database. This means you can start from what is actually in the database and then edit the diagram to model proposed additions or changes - something neither SchemaSpy nor most ERD tools support directly. 通过 PlantUML 和 dbdiagram.io 进行模式扩展： SchemaCrawler 可以直接从实时数据库生成 PlantUML 和 dbdiagram.io 格式的输出。这意味着你可以基于数据库现状进行建模，并编辑图表以规划新增或变更——这是 SchemaSpy 和大多数 ERD 工具无法直接支持的。
• Scripting - Python, JavaScript, Groovy, Ruby: --command=script runs a script against live schema metadata. Generate custom reports, validate naming conventions, transform output - without writing a Java application. 脚本支持（Python, JavaScript, Groovy, Ruby）： --command=script 允许针对实时模式元数据运行脚本。无需编写 Java 应用程序，即可生成自定义报告、验证命名规范或转换输出格式。
• Full Java API: SchemaCrawler is a JDBC metadata API. Embed it in a Java application and work with tables, columns, indexes, foreign keys, and routines as Java objects. SchemaSpy has no public API. 完整的 Java API： SchemaCrawler 本质上是一个 JDBC 元数据 API。你可以将其嵌入 Java 应用程序中，将表、列、索引、外键和存储过程作为 Java 对象进行操作。SchemaSpy 没有公开的 API。
• GitHub Actions integration: There is an official SchemaCrawler GitHub Action in the marketplace. Run lint, diff, and schema documentation generation as part of any CI/CD workflow. SchemaSpy has no equivalent. GitHub Actions 集成： 市场中提供了官方的 SchemaCrawler GitHub Action。你可以将 lint、diff 和模式文档生成作为 CI/CD 工作流的一部分运行。SchemaSpy 没有类似功能。

Feature Comparison

功能对比

Capability	SchemaCrawler	SchemaSpy
Interactive HTML report	✅	✅
Clickable navigation between tables	✅	✅
ER diagrams	✅	✅
Diff-able text output	✅	❌
Schema lint / design checks	✅	❌
Grep / regex search across schema	✅	❌
Markdown, JSON, CSV output	✅	❌
PlantUML and dbdiagram.io output	✅	❌
Scripting (Python, JS, Groovy)	✅	❌
Java API	✅	❌
GitHub Actions integration	✅	❌
Implied relationship detection	✅	✅
Orphan table detection	✅	✅

Decision Guide

决策指南

Choose SchemaSpy if… 选择 SchemaSpy，如果……

• Your primary output is a shareable, interactive HTML report for non-technical stakeholders. 你的主要需求是为非技术利益相关者提供一份可分享的交互式 HTML 报告。
• You want clickable navigation between related tables out of the box. 你希望开箱即用，实现表与表之间的点击导航。
• You need implied/virtual foreign key detection for a legacy schema with missing FK declarations. 你需要为缺少外键声明的遗留模式检测隐含/虚拟外键。

Choose SchemaCrawler if… 选择 SchemaCrawler，如果……

• You need to track schema changes in version control - diff text output between environments. 你需要通过版本控制追踪模式变更——对比不同环境下的文本输出。
• You want to catch design problems automatically - schema lint in CI. 你希望在 CI 中自动捕获设计问题——即模式 Lint。
• You need to search across a large schema - find all tables or columns matching a pattern. 你需要在大型模式中进行搜索——查找所有符合特定模式的表或列。
• You are building schema checks into a CI/CD pipeline - GitHub Actions integration. 你正在将模式检查构建到 CI/CD 流水线中——利用 GitHub Actions 集成。
• You need output in Markdown, JSON, or CSV as well as HTML. 除了 HTML，你还需要 Markdown、JSON 或 CSV 格式的输出。
• You want to model future schema designs in PlantUML or dbdiagram.io, starting from your live database. 你希望基于实时数据库，使用 PlantUML 或 dbdiagram.io 对未来的模式设计进行建模。
• You want to write scripts that process schema metadata programmatically. 你希望编写脚本以编程方式处理模式元数据。
• You are building a Java application that needs database metadata as objects. 你正在构建一个需要将数据库元数据作为对象处理的 Java 应用程序。

Can You Use Both?

可以两者兼用吗？

Yes. They serve genuinely different workflows. Use SchemaSpy to generate the stakeholder-facing HTML report. Use SchemaCrawler for diff, lint, and grep in your development and CI/CD workflow. The two tools are not competitors - they complement each other. 可以。它们服务于完全不同的工作流。使用 SchemaSpy 生成面向利益相关者的 HTML 报告；在开发和 CI/CD 工作流中使用 SchemaCrawler 进行 diff、lint 和 grep。这两款工具并非竞争对手，而是互补关系。

Try SchemaCrawler at schemacrawler.com. The source is at github.com/schemacrawler/SchemaCrawler. 访问 schemacrawler.com 尝试 SchemaCrawler。源码位于 github.com/schemacrawler/SchemaCrawler。