feat: #42 import confluence html zip

Author: shaunthegeek

README.md

@@ -31,3 +31,27 @@ chmod +x coding.phar
 sudo mv coding.phar /usr/local/bin/coding
 coding list
 ```
+
+## Confluence 导入 CODING Wiki
+
+1. 浏览器访问 Confluence 空间，导出 HTML，获得一个 zip 压缩包。
+
+![image](https://user-images.githubusercontent.com/4971414/127876158-8ab62714-e43f-4e20-8865-f8817f9264e1.png)
+
+2. 浏览器访问 CODING，创建个人令牌
+
+![image](https://user-images.githubusercontent.com/4971414/127877027-68a3f58e-c253-4ba9-b4f9-68b6673582a3.png)
+
+3. 打开命令行，进入 zip 文件所在的目录，执行命令导入：
+
+```shell
+cd ~/Downloads/
+docker run -it -v $(pwd):/root --env CODING_IMPORT_PROVIDER=Confluence \
+  --env CODING_IMPORT_DATA_TYPE=HTML \
+  --env CODING_IMPORT_DATA_PATH=./Confluence-space-export-231543-81.html.zip \
+  --env CODING_TOKEN=foo \
+  ecoding/coding-cli wiki:import
+```
+
+![image](https://user-images.githubusercontent.com/4971414/127878108-f778bfd6-fe7f-49f3-9590-9efd68404df5.png)
+

app/Commands/WikiImportCommand.php

@@ -2,7 +2,6 @@
 
 namespace App\Commands;
 
-use App\Coding;
 use App\Coding\Disk;
 use App\Coding\Wiki;
 use Confluence\Content;
@@ -11,6 +10,7 @@
 use Illuminate\Support\Str;
 use LaravelFans\Confluence\Facades\Confluence;
 use LaravelZero\Framework\Commands\Command;
+use ZipArchive;
 
 class WikiImportCommand extends Command
 {
@@ -26,7 +26,7 @@ class WikiImportCommand extends Command
     protected $signature = 'wiki:import
         {--coding_import_provider= : 数据来源，如 Confluence、MediaWiki}
         {--coding_import_data_type= : 数据类型，如 HTML、API}
-        {--coding_import_data_path= : 空间导出的 HTML 目录，如 ./confluence/space1/}
+        {--coding_import_data_path= : 空间导出的 HTML zip 文件路径，如 ./Confluence-space-export-231543-81.html.zip}
         {--confluence_base_uri= : Confluence API URL，如 http://localhost:8090/confluence/rest/api/}
         {--confluence_username=}
         {--confluence_password=}
@@ -135,15 +135,8 @@ private function handleConfluenceApi(): int
 
     private function handleConfluenceHtml(): int
     {
-        $dataPath = $this->option('coding_import_data_path');
-        if (is_null($dataPath)) {
-            $dataPath = config('coding.import.data_path') ?? trim($this->ask(
-                '空间导出的 HTML 目录',
-                './confluence/space1/'
-            ));
-        }
-        $dataPath = str_ends_with($dataPath, '/index.html') ? substr($dataPath, 0, -10) : Str::finish($dataPath, '/');
-        $filePath = $dataPath . 'index.html';
+        $htmlDir = $this->unzipConfluenceHtml();
+        $filePath = $htmlDir . 'index.html';
         if (!file_exists($filePath)) {
             $this->error("文件不存在：$filePath");
             return 1;
@@ -171,7 +164,7 @@ private function handleConfluenceHtml(): int
             }
             $this->info('发现 ' . count($pages['tree']) . ' 个一级页面');
             $this->info("开始导入 CODING：");
-            $this->uploadConfluencePages($dataPath, $pages['tree'], $pages['titles']);
+            $this->uploadConfluencePages($htmlDir, $pages['tree'], $pages['titles']);
         } catch (\ErrorException $e) {
             $this->error($e->getMessage());
             return 1;
@@ -239,4 +232,31 @@ private function uploadConfluencePages(string $dataPath, array $tree, array $tit
             }
         }
     }
+
+    private function unzipConfluenceHtml(): string
+    {
+        $dataPath = $this->option('coding_import_data_path');
+        if (is_null($dataPath)) {
+            $dataPath = config('coding.import.data_path') ?? trim($this->ask(
+                '空间导出的 HTML zip 文件路径',
+                './confluence/space1.zip'
+            ));
+        }
+
+        if (str_ends_with($dataPath, '.zip')) {
+            $zip = new ZipArchive();
+            $zip->open($dataPath);
+            $tmpDir = sys_get_temp_dir() . '/confluence-' . Str::uuid();
+            mkdir($tmpDir);
+            for ($i = 0; $i < $zip->numFiles; $i++) {
+                // HACK crash when zip include root path /
+                if ($zip->getNameIndex($i) != '/' && $zip->getNameIndex($i) != '__MACOSX/_') {
+                    $zip->extractTo($tmpDir, [$zip->getNameIndex($i)]);
+                }
+            }
+            $zip->close();
+            return $tmpDir . '/' . scandir($tmpDir, 1)[0] . '/';
+        }
+        return str_ends_with($dataPath, '/index.html') ? substr($dataPath, 0, -10) : Str::finish($dataPath, '/');
+    }
 }

composer.lock

@@ -133,14 +133,14 @@ public function testHandleConfluenceHtmlFileNotExist()
         $this->artisan('wiki:import')
             ->expectsQuestion('数据来源？', 'Confluence')
             ->expectsQuestion('数据类型？', 'HTML')
-            ->expectsQuestion('空间导出的 HTML 目录', '~/Downloads/')
+            ->expectsQuestion('空间导出的 HTML zip 文件路径', '~/Downloads/')
             ->expectsOutput('文件不存在：~/Downloads/index.html')
             ->assertExitCode(1);
 
         $this->artisan('wiki:import')
             ->expectsQuestion('数据来源？', 'Confluence')
             ->expectsQuestion('数据类型？', 'HTML')
-            ->expectsQuestion('空间导出的 HTML 目录', '~/Downloads/index.html')
+            ->expectsQuestion('空间导出的 HTML zip 文件路径', '~/Downloads/index.html')
             ->expectsOutput('文件不存在：~/Downloads/index.html')
             ->assertExitCode(1);
     }
@@ -177,7 +177,7 @@ public function testHandleConfluenceHtmlSuccess()
         $this->artisan('wiki:import')
             ->expectsQuestion('数据来源？', 'Confluence')
             ->expectsQuestion('数据类型？', 'HTML')
-            ->expectsQuestion('空间导出的 HTML 目录', $this->dataDir . 'confluence/space1/')
+            ->expectsQuestion('空间导出的 HTML zip 文件路径', $this->dataDir . 'confluence/space1/')
             ->expectsOutput('空间名称：空间 1')
             ->expectsOutput('空间标识：space1')
             ->expectsOutput('发现 2 个一级页面')
@@ -210,4 +210,59 @@ public function testAskNothing()
             ->expectsOutput('文件不存在：/dev/null/index.html')
             ->assertExitCode(1);
     }
+
+    public function testHandleConfluenceHtmlZipSuccess()
+    {
+        $codingToken = $this->faker->md5;
+        config(['coding.token' => $codingToken]);
+        $codingTeamDomain = $this->faker->domainWord;
+        config(['coding.team_domain' => $codingTeamDomain]);
+        $codingProjectUri = $this->faker->slug;
+        config(['coding.project_uri' => $codingProjectUri]);
+
+        // 注意：不能使用 partialMock
+        // https://laracasts.com/discuss/channels/testing/this-partialmock-doesnt-call-the-constructor
+        $mock = \Mockery::mock(Wiki::class, [])->makePartial();
+        $this->instance(Wiki::class, $mock);
+
+        $mock->shouldReceive('createWikiByUploadZip')->times(5)->andReturn(json_decode(
+            file_get_contents($this->dataDir . 'coding/' . 'CreateWikiByZipResponse.json'),
+            true
+        )['Response']);
+        $mock->shouldReceive('getImportJobStatus')->times(5)->andReturn(json_decode(
+            file_get_contents($this->dataDir . 'coding/' . 'DescribeImportJobStatusResponse.json'),
+            true
+        )['Response']['Data']);
+        $mock->shouldReceive('updateWikiTitle')->times(5)->andReturn(true);
+
+
+        $mockDisk = \Mockery::mock(Disk::class, [])->makePartial();
+        $this->instance(Disk::class, $mockDisk);
+        $mockDisk->shouldReceive('uploadAttachments')->times(5)->andReturn([]);
+
+        $this->artisan('wiki:import')
+            ->expectsQuestion('数据来源？', 'Confluence')
+            ->expectsQuestion('数据类型？', 'HTML')
+            ->expectsQuestion(
+                '空间导出的 HTML zip 文件路径',
+                $this->dataDir . 'confluence/Confluence-space-export-231543-81.html.zip'
+            )
+            ->expectsOutput('空间名称：空间 1')
+            ->expectsOutput('空间标识：space1')
+            ->expectsOutput('发现 1 个一级页面')
+            ->expectsOutput("开始导入 CODING：")
+            ->expectsOutput('标题：空间 1 Home')
+            ->expectsOutput('上传成功，正在处理，任务 ID：a12353fa-f45b-4af2-83db-666bf9f66615')
+            ->expectsOutput('发现 2 个子页面')
+            ->expectsOutput('标题：hello world')
+            ->expectsOutput('上传成功，正在处理，任务 ID：a12353fa-f45b-4af2-83db-666bf9f66615')
+            ->expectsOutput('发现 2 个子页面')
+            ->expectsOutput('标题：hello')
+            ->expectsOutput('上传成功，正在处理，任务 ID：a12353fa-f45b-4af2-83db-666bf9f66615')
+            ->expectsOutput('标题：world')
+            ->expectsOutput('上传成功，正在处理，任务 ID：a12353fa-f45b-4af2-83db-666bf9f66615')
+            ->expectsOutput('标题：你好世界')
+            ->expectsOutput('上传成功，正在处理，任务 ID：a12353fa-f45b-4af2-83db-666bf9f66615')
+            ->assertExitCode(0);
+    }
 }