Fix auto schema version detection

Fix and add unit tests for the following cases:
- when the component is not harvested or partially harvested
- update status once upon complete, to better handle multiple versions of harvest tool results

tools/integration/lib/harvestResultFetcher.js

@@ -14,12 +14,13 @@ class HarvestResultFetcher {
   async fetchToolVersions(tools = defaultTools) {
     const listHarvestResultApi = `${this.apiBaseUrl}/harvest/${this._coordinates}?form=list`
     const harvestResultUrls = await this._fetch(listHarvestResultApi).then(r => r.json())
-    return tools.flatMap(tool =>
-      harvestResultUrls
+    return tools.flatMap(tool => {
+      const found = harvestResultUrls
         .filter(url => url.includes(`/${tool}/`))
         .map(url => url.substring(`${this._coordinates}/${tool}/`.length))
         .map(version => [tool, version])
-    )
+      return found.length ? found : [[tool, '']]
+    })
   }
 
   async _pollForCompletion(poller) {
@@ -54,13 +55,14 @@ class HarvestResultFetcher {
   async isHarvestComplete(toolVersions, startTime, statuses = new Map()) {
     const harvestChecks = (toolVersions || []).map(async ([tool, toolVersion]) => {
       const completed = statuses.get(tool)?.completed || (await this.isHarvestedbyTool(tool, toolVersion, startTime))
-      if (completed) statuses.set(tool, { toolVersion, completed })
+      if (completed && !statuses.get(tool)) statuses.set(tool, { toolVersion, completed })
       return tool
     })
     return Promise.all(harvestChecks).then(tools => tools.every(tool => statuses.get(tool)?.completed))
   }
 
   async isHarvestedbyTool(tool, toolVersion, startTime = 0) {
+    if (!tool || !toolVersion) return false
     const harvested = await this.fetchHarvestResult(tool, toolVersion)
     if (!harvested._metadata) return false
     const fetchedAt = new Date(harvested._metadata.fetchedAt)

tools/integration/test/lib/harvesterResultFetcherTest.js

@@ -49,6 +49,11 @@ describe('HarvestResultFetcher', function () {
       strictEqual(result, true)
     })
 
+    it('should return false when tool version is not provided', async function () {
+      const result = await resultMonitor.isHarvestedbyTool('licensee', '')
+      strictEqual(result, false)
+    })
+
     it('should detect when component is completely harvested', async function () {
       sinon.stub(resultMonitor, 'fetchHarvestResult').resolves(metadata())
       const result = await resultMonitor.isHarvestComplete(defaultToolChecks)
@@ -146,6 +151,17 @@ describe('HarvestResultFetcher', function () {
         ['scancode', '30.3.0']
       ])
     })
+
+    it('should handle no result for tools', async function () {
+      const harvestResults = []
+      fetchStub.resolves(new Response(JSON.stringify(harvestResults)))
+      const toolVersions = await resultMonitor.fetchToolVersions()
+      deepStrictEqual(toolVersions, [
+        ['licensee', ''],
+        ['reuse', ''],
+        ['scancode', '']
+      ])
+    })
   })
 
   describe('pollForToolVersionsComplete', function () {
@@ -180,6 +196,64 @@ describe('HarvestResultFetcher', function () {
         ['scancode', '30.3.0']
       ])
     })
+
+    it('should handle partially harvested results correctly', async function () {
+      sinon
+        .stub(resultMonitor, 'fetchToolVersions')
+        .onFirstCall()
+        .resolves([
+          ['licensee', '9.14.0'],
+          ['reuse', '3.2.2'],
+          ['scancode', '']
+        ])
+        .onSecondCall()
+        .resolves([
+          ['licensee', '9.14.0'],
+          ['reuse', '3.2.1'],
+          ['reuse', '3.2.2'],
+          ['scancode', '30.3.0']
+        ])
+      sinon
+        .stub(resultMonitor, 'isHarvestedbyTool')
+        .withArgs('licensee', '9.14.0')
+        .resolves(true)
+        .withArgs('reuse', '3.2.1')
+        .resolves(true)
+        .withArgs('scancode', '30.3.0')
+        .resolves(true)
+      const toolVersions = await resultMonitor.pollForToolVersionsComplete(poller, Date.now())
+      deepStrictEqual(toolVersions, [
+        ['licensee', '9.14.0'],
+        ['reuse', '3.2.1'],
+        ['scancode', '30.3.0']
+      ])
+    })
+
+    it('should handle tool results with multiple schema versions correctly', async function () {
+      sinon.stub(resultMonitor, 'fetchToolVersions').resolves([
+        ['licensee', '9.14.0'],
+        ['reuse', '3.2.1'],
+        ['reuse', '3.2.2'],
+        ['scancode', '30.3.0']
+      ])
+      sinon
+        .stub(resultMonitor, 'isHarvestedbyTool')
+        .withArgs('licensee', '9.14.0')
+        .onFirstCall()
+        .resolves(false)
+        .onSecondCall()
+        .resolves(true)
+        .withArgs('reuse', '3.2.1')
+        .resolves(true)
+        .withArgs('scancode', '30.3.0')
+        .resolves(true)
+      const toolVersions = await resultMonitor.pollForToolVersionsComplete(poller, Date.now())
+      deepStrictEqual(toolVersions, [
+        ['reuse', '3.2.1'],
+        ['scancode', '30.3.0'],
+        ['licensee', '9.14.0']
+      ])
+    })
   })
 })