Bash 脚本的单元测试

{"type":"doc","content":[{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"为什么要为 Bash 脚本写单元测试？"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"因为 Bash 脚本通常都是在执行一些与操作系统有关的操作，可能会对运行环境造成一些不可逆的操作，比如修改或者删除文件、升级系统中的软件包等。所以为了确保 Bash 脚本的安全可靠，在生产环境中部署之前一定需要做好足够的测试以确保其行为符合我们的预期。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"如何能够安全可靠的去测试 Bash 脚本呢？有人可能会说我们可以用 Docker 容器。是的，这样做即安全又方便。在容器隔离出来的环境中不用担心脚本会破坏我们的系统，而且也能非常简单的快速重建出一个可用的测试环境。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"不过呢，请考虑以下的几个常见的场景："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"场景一：在执行 Bash 脚本测试前，我们需要需要事先安装好所有在 Bash 脚本中会用到的第三方工具，否则这些测试将会因为命令找不到而执行失败。例如，我们在脚本中使用了 Bazel 这个构建工具。我们必须提前安装并配置好 Bazel，而且不要忘记为了能够正常使用 Bazel 还得需要一个支持使用 Bazel 构建的工程。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"场景二：测试结果的稳定性可能取决于脚本中访问的第三方服务的稳定性。比如，我们在脚本中使用 curl 命令从一个网络服务中获取数据，但这个服务有时候可能会访问失败。有可能是因为网络不稳定导致的，也可能是因为这个服务本身不稳定。再或者如果我们需要第三方服务返回不同的数据以便测试脚本的不同分支逻辑，但我们可能很难去修改这个第三方服务的数据。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"场景三：Bash 脚本的测试用例的执行时间取决于脚本中使用的命令的执行时间。例如，如果我们中脚本中使用了 Gradle 来构建一个工程，由于不同的工程大小 Gradle 的一个构建可能要执行3分钟或者3个小时。这还只是一个测试用例，如果我们还有20个或者100个测试用例呢？我们是否还能在几秒内获得测试报告呢？"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"即使使用了容器来执行 Bash 脚本测试，也一样无法避免上面的几个问题。环境的准备过程可能会随着测试用例的增多而变的繁琐，测试用例的稳定性和执行时长取决于第三方命令和服务的稳定性和执行时长，还可能很难做到使用不同数据来覆盖不同的测试场景。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"对于测试 Bash 脚本来说，我们真正要验证的是 Bash 脚本的执行逻辑。比如在 Bash 脚本中可能会根据传入的参数来组合出内部所调用的命令的选项和参数，我们要验证的是这些选项和参数确实如我们预期的。至于调用的命令在接受了这些选项和参数后由于什么原因而失败，可能我们并不关心这所有的可能原因。因为这会有更多的外部影响因素，比如硬件和网络都是否工作正常、第三方服务是否正常运行、构建工程所需的编译器是否安装并配置妥当、授权和认证信息是否都有效、等等。但对于 Bash 脚本来说，这些外部原因导致的结果就是所调用的命令执行成功或者失败了。所以 Bash 脚本只要关注的是脚本中调用的命令是否能够成功执行，以及命令输出了哪些，并决定随后执行脚本中的哪些不同分支逻辑。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"如果说我们就是想知道这个命令搭配上这些选项参数是否能按我们预期的那样工作呢？很简单，那就单独在命令行里面去执行一下。如果在命令行中也不能按预期的工作，放到 Bash 脚本里面也一样不会按预期的工作。这种错误和 Bash 脚本几乎没什么关系了。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"所以，为了尽量去除影响 Bash 脚本验证的那些外部因素，我们应该考虑为 Bash 脚本编写单元测试，以关注在 Bash 脚本的执行逻辑上。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"什么样的测试才是 Bash 脚本的单元测试？"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"首先，所有存在于 "},{"type":"codeinline","content":[{"type":"text","text":"PATH"}]},{"type":"text","text":" 环境变量的路径中的命令都不应该在单元测试中被执行。对 Bash 脚本来说，被调用的这些命令可以正常运行，有返回值，有输出。但脚本中调用的这些命令都是被模拟出来的，用于模拟对应的真实命令的行为。这样，我们在 Bash 脚本的单元测试中就避免了很大一部分的外部依赖，而且测试的执行速度也不会受到真实命令的影响了。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"其次，每个单元测试用例之间都应该是独立的。这意味着，这些测试用例可以独立执行或者被任意乱序执行，而不会影响验证结果。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"最后，这些测试用例可以在不同的操作系统上执行，且都应该得到相同的验证结果。比如 Bash 脚本中使用了只有 GNU/Linux 上才有的命令，对应的单元测试也可以在 Windows 或者 macOS 上执行，且结果一致。"}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"怎样为 Bash 脚本写单元测试？"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"与其他编程语言一样，Bash 也有多个测试框架，比如 "},{"type":"link","attrs":{"href":"https://github.com/sstephenson/bats","title":""},"content":[{"type":"text","text":"Bats"}]},{"type":"text","text":"、"},{"type":"link","attrs":{"href":"https://github.com/kward/shunit2/","title":""},"content":[{"type":"text","text":"Shunit2"}]},{"type":"text","text":" 等，但这些框架实际上并不能隔离所有 PATH 环境变量中的命令。有一个名为 "},{"type":"link","attrs":{"href":"https://bach.sh","title":""},"content":[{"type":"text","text":"Bach Testing Framework"}]},{"type":"text","text":" 的测试框架是目前唯一一个可以为 Bash 脚本编写真正的单元测试的框架。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"Bach Testing Framework 的最独特的特性就是默认不会执行任何位于 "},{"type":"codeinline","content":[{"type":"text","text":"PATH"}]},{"type":"text","text":" 环境变量中的命令，因此 Bach Testing Framework 非常适用于验证 Bash 脚本的执行逻辑。并且还带来了以下好处："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"简单"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"什么也不用安装。我们就可以执行这些测试。比如可以在一个全新的环境中执行一个调用了大量第三方命令的 Bash 脚本。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"快"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"因为所有的命令都不会被真正执行，所以每一个测试用例的执行都非常快。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"安全"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"因为不会执行任何外部的命令，所以即使因为 Bash 脚本中的某些错误导致执行了一个危险的命令，比如 rm -rf *。Bach 会保证这些危险命令不会被执行。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"与运行环境无关"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"可以在 Windows 上去执行只能工作在 GNU/Linux 上的脚本的测试。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"由于操作系统和 Bash 的一些限制，Bach Testing Framework 无法做到："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"拦截使用绝对路径调用的命令"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"事实上我们应该避免在 Bash 脚本中使用绝对路径，如果不可避免的要使用，我们可以把这个绝对路径抽取为一个变量，或者放入到一个函数中，然后用 @mock API 去模拟这个函数。"}]},{"type":"bulletedlist","content":[{"type":"listitem","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"拦截诸如 >、>>、<< 等等这样的 I/O 重定向"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"是的，无法拦截 I/O 重定向。我们也同样可以把这些重定向操作隔离到一个函数中，然后再模拟这个函数。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"Bach Testing Framework 的使用"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"Bach Testing Framework 需要 Bash v4.3 或更高版本。在 GNU/Linux 上还需要 Coreutils 和 Diffutils，在常用的发行版中都已经默认安装好了。Bach 在 Linux/macOS/Cygwin/Git Bash/FreeBSD 等操作系统或者运行环境中验证通过。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"Bash v4.3+"}]}]},{"type":"listitem","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"Coreutils (GNU/Linux)"}]}]},{"type":"listitem","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"Diffutils (GNU/Linux)"}]}]}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"安装 Bach Testing Framework"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"Bach Testing Framework 的安装很简单，只需要下载 https://github.com/bach-sh/bach/raw/master/bach.sh 到你的项目中，在测试脚本中用 source 命令导入 Bach Testing Framework 的 bach.sh 即可。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"比如："},{"type":"codeinline","content":[{"type":"text","text":"source path/to/bach.sh"}]}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"一个简单的例子"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"与其它的测试框架不同，Bach Testing Framework 的每一个测试用例都是由两个 Bash 函数组成，一个是以 "},{"type":"codeinline","content":[{"type":"text","text":"test-"}]},{"type":"text","text":" 开头的测试执行函数，另一个是同名的以 "},{"type":"codeinline","content":[{"type":"text","text":"-assert"}]},{"type":"text","text":" 结尾的测试验证函数。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"比如在下面的例子中，有两个测试用例，分别是"}]},{"type":"bulletedlist","content":[{"type":"listitem","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"test-rm-rf"}]}]},{"type":"listitem","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"test-rm-your-dot-git"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"一个完整的测试用例："}]},{"type":"codeblock","attrs":{"lang":"shell"},"content":[{"type":"text","text":"#!/usr/bin/env bash\nset -euo pipefail\n\nsource bach.sh # 导入 Bach Testing Framework\n\ntest-rm-rf() {\n # Bach 的标准测试用例是由两个方法组成\n # - test-rm-rf\n # - test-rm-rf-assert\n # 这个方法 `test-rm-rf` 是测试用例的执行\n\n project_log_path=/tmp/project/logs\n sudo rm -rf \"$project_log_ptah/\" # 注意，这里有个笔误！\n}\ntest-rm-rf-assert() {\n # 这个方法 `test-rm-rf-assert` 是测试用例的验证\n sudo rm -rf / # 这就是真实的将会执行的命令\n # 不要慌！使用 Bach 测试框架不会让这个命令真的执行！\n}\n\ntest-rm-your-dot-git() {\n # 模拟 `find` 命令来查找你的主目录下的所有 `.git` 目录，假设会找到两个目录\n\n @mock find ~ -type d -name .git === @stdout ~/src/your-awesome-project/.git \\\n ~/src/code/.git\n\n # 开始执行！删除你的主目录下的所有 `.git` 目录！\n find ~ -type d -name .git | xargs -- rm -rf\n}\ntest-rm-your-dot-git-assert() {\n # 验证在 `test-rm-your-dot-git` 这个测试执行方法中最终是否会执行以下这个命令。\n\n rm -rf ~/src/your-awesome-project/.git ~/src/code/.git\n}"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"Bach 会分别运行每一个测试用例的两个方法，去验证两个方法中执行的命令及其参数是否是一致的。比如，第一个方法 "},{"type":"codeinline","content":[{"type":"text","text":"test-rm-rf"}]},{"type":"text","text":" 是 Bach 的测试用例的执行，与之对应的测试验证方法就是 "},{"type":"codeinline","content":[{"type":"text","text":"test-rm-rf-assert"}]},{"type":"text","text":" 这个方法"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"在第二个测试用例 "},{"type":"codeinline","content":[{"type":"text","text":"test-rm-your-dot-git"}]},{"type":"text","text":" 中使用了 "},{"type":"codeinline","content":[{"type":"text","text":"@mock"}]},{"type":"text","text":" API 来模拟了命令 "},{"type":"codeinline","content":[{"type":"text","text":"find ~ type d -name .git"}]},{"type":"text","text":" 的行为，这个命令用来找出用户目录下的所有 "},{"type":"codeinline","content":[{"type":"text","text":".git"}]},{"type":"text","text":" 目录。模拟之后，这个命令并不会真的执行，而是利用了 "},{"type":"codeinline","content":[{"type":"text","text":"@stdout"}]},{"type":"text","text":" API 在标准终端上输出了两个虚拟的目录名。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"然后我们就可以执行真正的命令了，将 "},{"type":"codeinline","content":[{"type":"text","text":"find"}]},{"type":"text","text":" 命令的输出结果传递给 "},{"type":"codeinline","content":[{"type":"text","text":"xargs"}]},{"type":"text","text":" 命令，并组合到 "},{"type":"codeinline","content":[{"type":"text","text":"rm -rf"}]},{"type":"text","text":" 命令之后。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"在对应的测试验证函数 "},{"type":"codeinline","content":[{"type":"text","text":"test-rm-your-dot-git-assert"}]},{"type":"text","text":" 里面就验证是 "},{"type":"codeinline","content":[{"type":"text","text":"find ~ -type d -name .git | xargs -- rm -rf"}]},{"type":"text","text":" 的运行结果是否等同于命令 "},{"type":"codeinline","content":[{"type":"text","text":"rm -rf ~/src/your-awesome-project/.git ~/src/code/.git"}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"codeinline","content":[{"type":"text","text":"@mock"}]},{"type":"text","text":" 是 Bach Testing Framework 中很重要的一个 API，利用这个 API 我们就可以模拟 Bash 脚本中所使用的任意命令的行为或者输出。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"比如"}]},{"type":"codeblock","attrs":{"lang":"shell"},"content":[{"type":"text","text":"@mock curl --silent google.com === \\\n @stdout \"baidu.com\""}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"模拟了命令 "},{"type":"codeinline","content":[{"type":"text","text":"curl --silent google.com"}]},{"type":"text","text":" 的执行结果是输出 "},{"type":"codeinline","content":[{"type":"text","text":"baidu.com"}]},{"type":"text","text":"。在真实的正常场景下，我们是无法做到访问 "},{"type":"codeinline","content":[{"type":"text","text":"google.com"}]},{"type":"text","text":" 得到的是 "},{"type":"codeinline","content":[{"type":"text","text":"baidu.com"}]},{"type":"text","text":"。这样模拟之后就可以用来验证 Bash 脚本中处理一个命令不同响应时的行为了。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"codeinline","content":[{"type":"text","text":"@mock"}]},{"type":"text","text":" API 甚至还支持更复杂的行为模拟，我们可以自定义一个复杂的模拟逻辑，比如："}]},{"type":"codeblock","attrs":{"lang":"shell"},"content":[{"type":"text","text":"\n@mock ls <