add logformatter back and provide better github log output

.github/workflows/lima.yml

@@ -28,6 +28,12 @@ on:
 
 permissions: {}
 
+# needed for logformatter
+env:
+  PR_NUMBER: ${{ github.event.pull_request.number || '' }}
+  PR_TITLE: ${{ github.event.pull_request.title || ''  }}
+  GITHUB_RUN_ID: ${{ github.run_id }}
+
 jobs:
   lima:
     if: ${{ inputs.if }}
@@ -64,6 +70,10 @@ jobs:
         run: | # zizmor: ignore[template-injection]
           ./hack/ci/ci.sh ${{ inputs.test }} ${{ inputs.mode }} ${{ inputs.priv }} ${{ inputs.distro }}
 
+      - name: Output failure log as GITHUB_STEP_SUMMARY
+        if: failure()
+        run: hack/ci/github_log_summary.py hack/ci/logs/*.html > $GITHUB_STEP_SUMMARY || true
+
       # TODO: figure out how to cache the binaries from the build job to the actual test tasks
       # - name: Upload binaries as artifact
       #   if: ${{ inputs.test == 'build' }}
@@ -77,6 +87,6 @@ jobs:
         if: always() # always collect the log
         uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a #v7.0.1
         with:
-          name: "journal-${{ inputs.test }}-${{ inputs.mode }}-${{ inputs.priv }}-${{ inputs.distro }}.log"
-          path: "./hack/ci/journal.log"
+          name: "${{ inputs.test }}-${{ inputs.mode }}-${{ inputs.priv }}-${{ inputs.distro }}.logs"
+          path: "./hack/ci/logs/*"
           if-no-files-found: ignore

.gitignore

@@ -12,6 +12,7 @@ coverprofile
 /docs/*.[158].gz
 /docs/build/
 /docs/remote
+/hack/ci/logs/
 **.DS_Store
 .gopathok
 .idea*

hack/ci/ci.sh

@@ -20,23 +20,30 @@ IMAGE_URL="https://objectstorage.us-ashburn-1.oraclecloud.com/n/id0lmbbwgcdv/b/p
 
 trap "limactl delete --force $LIMA_VM_NAME" EXIT
 
+echo "::group::Starting VM"
 limactl --yes start --plain --name=$LIMA_VM_NAME --cpus $(nproc) --memory 8 --nested-virt \
     --set ".images=[{\"location\":\"$IMAGE_URL\", \"arch\": \"x86_64\"}]" \
     "$SCRIPT_DIR/template.lima.yml"
 
-limactl copy "$REPO_DIR" $LIMA_VM_NAME:/var/tmp/podman
+limactl shell $LIMA_VM_NAME mkdir -p /var/tmp/podman-container-tools
+
+limactl copy "$REPO_DIR" $LIMA_VM_NAME:/var/tmp/podman-container-tools/podman
+
+echo "::endgroup::"
 
 set +e
 
-limactl shell --workdir /var/tmp/podman $LIMA_VM_NAME ./hack/ci/runner.sh "${@}"
+limactl shell --preserve-env --workdir /var/tmp/podman-container-tools/podman $LIMA_VM_NAME ./hack/ci/runner.sh "${@}"
 rc=$?
 
-limactl shell --workdir /var/tmp/podman $LIMA_VM_NAME sudo ./hack/ci/logcollector.sh journal &> "$SCRIPT_DIR/journal.log"
+echo "::group::Collecting logs"
+limactl copy $LIMA_VM_NAME:/var/tmp/podman-container-tools/podman/hack/ci/logs/ $SCRIPT_DIR/logs
+echo "::endgroup::"
 
 # TODO: figure out how to cache the binaries from the build job to the actual test tasks
 # Copy the binaries out of the VM in gh actions so we can upload them as artifact
 # if [[ -n "$GITHUB_ACTIONS" && "$TEST" == build ]]; then
-#    limactl copy $LIMA_VM_NAME:/var/tmp/podman/bin "$REPO_DIR/bin" || die "failed to copy binaries"
+#    limactl copy $LIMA_VM_NAME:/var/tmp/podman-container-tools/podman/bin "$REPO_DIR/bin" || die "failed to copy binaries"
 # fi
 
 exit $rc

hack/ci/github_log_summary.py

@@ -0,0 +1,128 @@
+#!/usr/bin/env python
+
+from html.parser import HTMLParser
+import html
+import sys
+
+class GinkgoLogFilterParser(HTMLParser):
+    def __init__(self):
+        super().__init__()
+        # Stack to keep track of nested elements
+        self.stack = []
+        # Store the raw HTML strings of matching 'tt' elements
+        self.results = []
+
+    def _get_classes(self, attrs):
+        """Helper to extract classes from an attribute list."""
+        for attr, value in attrs:
+            if attr == 'class' and value:
+                return value.split()
+        return []
+
+    def handle_starttag(self, tag, attrs):
+        classes = self._get_classes(attrs)
+        is_tt = 'tt' in classes
+        is_failed = 'log-failed' in classes
+
+        # If we see a 'log-failed' class, flag all 'tt' ancestors in the stack
+        if is_failed:
+            for node in self.stack:
+                if node['is_tt']:
+                    node['keep'] = True
+
+        # Push the new element to the stack
+        self.stack.append({
+            'tag': tag,
+            'text_chunks': [],
+            'is_tt': is_tt,
+            'keep': False
+        })
+
+    def handle_startendtag(self, tag, attrs):
+        # Handle self-closing tags just to check for the failure class
+        classes = self._get_classes(attrs)
+        if 'log-failed' in classes:
+            for node in self.stack:
+                if node['is_tt']:
+                    node['keep'] = True
+
+    def handle_data(self, data):
+        # Capture raw text data
+        if self.stack:
+            self.stack[-1]['text_chunks'].append(data)
+
+    def handle_endtag(self, tag):
+        # Find the matching start tag in the stack
+        for i in reversed(range(len(self.stack))):
+            if self.stack[i]['tag'] == tag:
+                # Pop the matching tag and any unclosed children
+                while len(self.stack) > i:
+                    node = self.stack.pop()
+
+                    # Join all the collected text for this node
+                    node_text = "".join(node['text_chunks'])
+                    # Trim down the spaces for the ginkgo long indentation.
+                    node_text = node_text.removeprefix(' ' * 9)
+
+                    # If this is a 'tt' element and it contains a 'log-failed' child, save the text
+                    if node['is_tt'] and node['keep']:
+                        self.results.append(node_text)
+
+                    # Pass the text of this node up to its parent so we don't lose inner text
+                    if self.stack:
+                        self.stack[-1]['text_chunks'].append(node_text)
+                break
+
+class BatsLogFilterParser(HTMLParser):
+    def __init__(self):
+        super().__init__()
+        # Current Tag which text should be stored
+        self.record = None
+        self.data = ""
+
+    def _get_classes(self, attrs):
+        """Helper to extract classes from an attribute list."""
+        for attr, value in attrs:
+            if attr == 'class' and value:
+                return value.split()
+        return []
+
+    def handle_starttag(self, tag, attrs):
+        classes = self._get_classes(attrs)
+
+        if 'bats-failed' in classes or 'bats-log-failblock' in classes or 'bats-log' in classes:
+            self.record = tag
+
+    def handle_data(self, data):
+        if self.record:
+            self.data += data
+
+    def handle_endtag(self, tag):
+        if self.record == tag:
+            self.data += "\n"
+            self.record = None
+
+
+
+def filter_html_file(file_path):
+    # Read the HTML content
+    with open(file_path, 'r', encoding='utf-8') as f:
+        html_content = f.read()
+
+    if 'int-' in file_path:
+        parser = GinkgoLogFilterParser()
+        parser.feed(html_content)
+        return parser.results
+
+    parser = BatsLogFilterParser()
+    parser.feed(html_content)
+    return [parser.data]
+
+
+# Running the filter
+matching_elements = filter_html_file(sys.argv[1])
+
+for element in matching_elements:
+    print(f"```")
+    print(element)
+    print("```")

hack/ci/lib.sh

@@ -10,6 +10,11 @@ OS_RELEASE_ID="$(
 )"
 OS_REL_VER="$OS_RELEASE_ID-$OS_RELEASE_VER"
 
+TEST_NAME=$(
+    IFS=-
+    echo "${*}"
+)
+
 function die() {
     echo "$1" >&2
     exit 1