From ac12ea8319289a96eb083f25ea27d4bfa723dd85 Mon Sep 17 00:00:00 2001
From: Ghanem <37152329+AbdulrhmnGhanem@users.noreply.github.com>
Date: Fri, 28 Oct 2022 12:38:40 +0200
Subject: [PATCH] processor: handle readme `a` and `img` elements with no URLs
 (#449)

Co-authored-by: Kaspar Emanuel <kaspar.emanuel@gmail.com>
---
 .../src/tasks/processReadme/urlTransformer.ts |  6 +-
 .../integration/test_readme_processing.ts     | 79 +++++++++++++++++++
 2 files changed, 83 insertions(+), 2 deletions(-)
 create mode 100644 processor/test/integration/test_readme_processing.ts

diff --git a/processor/src/tasks/processReadme/urlTransformer.ts b/processor/src/tasks/processReadme/urlTransformer.ts
index ab4cd7904a..bc6af3ea56 100644
--- a/processor/src/tasks/processReadme/urlTransformer.ts
+++ b/processor/src/tasks/processReadme/urlTransformer.ts
@@ -63,8 +63,10 @@ function urlTransformer({ originalUrl, readmeFolder, ownerName, repoName, defaul
     }
 
     function modifyUrl(node, prop: 'href' | 'src', modifier: ReturnType<typeof SrcModifier>) {
-        const newURL = modifier(node.properties[prop])
-        node.properties[prop] = newURL
+        if (node.properties[prop]) {
+            const newURL = modifier(node.properties[prop])
+            node.properties[prop] = newURL
+        }
     }
 }
 
diff --git a/processor/test/integration/test_readme_processing.ts b/processor/test/integration/test_readme_processing.ts
new file mode 100644
index 0000000000..43510d88d1
--- /dev/null
+++ b/processor/test/integration/test_readme_processing.ts
@@ -0,0 +1,79 @@
+import assert from 'assert'
+import cp from 'node:child_process'
+import path from 'node:path'
+import supertest from 'supertest'
+import util from 'node:util'
+
+import { createApp } from '../../src/app.js'
+import { delay } from '../../src/utils.js'
+
+const exec = util.promisify(cp.exec)
+
+const tmpDir = '/data/test/temp/kitspace-processor-test-from-folder-readme-processing'
+const repoDir = path.join(tmpDir, 'repos')
+
+
+describe('readme processing', function () {
+    beforeEach(async function () {
+        await exec(`mkdir -p ${tmpDir}`)
+        await exec(`mkdir -p ${repoDir}`)
+        this.app = createApp(repoDir, { giteaDB: null })
+        this.supertest = supertest(this.app)
+    })
+
+    it("processes readme with invalid <a> tags", async function () {
+
+        /*
+        This project readme
+        * # 8BitmixtapeNEO_ShenzhenReady
+
+        * summary: <a summary for your project> <- This invalid <a>  tag, probably it's meant as a placeholder
+        * site: https://8bitmixtape.github.io/NeoWiki
+        */
+        // first we reset HEAD/master to an exact version of the repo
+        // so future changes of the repo don't affect this test
+        const hash = 'd8f86818c2ad3206abc9b7a0ed26dfdf672dba10'
+        const tmpBare = path.join(tmpDir, '8BitmixtapeNEO_ShenzhenReady.git')
+        await exec(`git clone --bare https://github.com/kitspace-test-repos/8BitmixtapeNEO_ShenzhenReady ${tmpBare}`)
+        await exec(`cd ${tmpBare} && git update-ref HEAD ${hash}`)
+        await exec(
+            `git clone --bare ${tmpBare} ${path.join(repoDir, 'kitspace-test-repos/8BitmixtapeNEO_ShenzhenReady.git')}`,
+        )
+
+        const files = [
+            'kitspace-yaml.json',
+            'readme.html'
+        ]
+
+        for (const f of files) {
+            // at first it may not be processing yet so we get a 404
+            let r = await this.supertest.get(`/status/kitspace-test-repos/8BitmixtapeNEO_ShenzhenReady/HEAD/${f}`)
+            while (r.status === 404) {
+                r = await this.supertest.get(`/status/kitspace-test-repos/8BitmixtapeNEO_ShenzhenReady/HEAD/${f}`)
+                await delay(10)
+            }
+
+            // after a while it should report something
+            assert(r.status === 200)
+            // but it's probably 'in_progress'
+            while (r.body.status === 'in_progress') {
+                r = await this.supertest.get(`/status/kitspace-test-repos/8BitmixtapeNEO_ShenzhenReady/HEAD/${f}`)
+                await delay(10)
+            }
+
+            // at some point it should notice it succeeded
+            assert(r.status === 200)
+            assert(
+                r.body.status === 'done',
+                `expecting body.status to be 'done' but got '${r.body.status
+                }' for ${f}\n${JSON.stringify(r.body, null, 2)}`,
+            )
+        }
+    }
+    )
+
+    afterEach(async function () {
+        await this.app.stop()
+        await exec(`rm -rf ${tmpDir}`)
+    })
+})