mirror of
https://github.com/Heretek-AI/heretek-actions.git
synced 2026-10-12 02:03:46 +00:00
Fix ocr-scan: correct package, add LLM config, fix output handling
Root cause analysis (5-perspective adversarial review): 1. Wrong NPM package — used @open-code-review/cli (doesn't exist), correct is @alibaba-group/open-code-review (verified on npmjs.com) 2. Missing LLM config — OCR_LLM_URL/OCR_LLM_TOKEN are custom env vars OCR doesn't read. Added ocr config set step mapping to a custom provider with the protocol/url/key from workflow secrets. 3. No install verification — previous install step swallowed all errors with 2>/dev/null || true. Now uses npm install -g + ocr scan --help + ocr llm test for fail-fast verification. 4. Wrong output path — scan --format json writes JSON to stdout, not .ocr/result.json. Now captures stdout directly to .ocr/result.json. 5. Missing comments field in envelope — build-envelope.sh didn't handle .comments[] from scan output (only handled .issues and .review_comments). Added .comments extraction with field mapping. Fixes: - .github/workflows/ocr-scan.yml: Split into Install & Configure + Run Scan steps. Correct package, LLM config via custom provider, direct output capture, explicit --model flag. - .github/actions/ocr-envelope/build-envelope.sh: Handle .comments[] from scan output, map to uniform findings format with file/line/ message/severity, handle object-typed .summary field.
This commit is contained in:
John Smith
@@ -23,9 +23,28 @@ WARNING_COUNT=0
|
||||
if [ -f "$OCR_RESULT_FILE" ]; then
|
||||
echo "Reading OCR results from ${OCR_RESULT_FILE}"
|
||||
|
||||
# Parse OCR JSON output — the format varies by OCR version
|
||||
# Common fields: issues, warnings, errors, summary, review_comments
|
||||
FINDINGS=$(cat "$OCR_RESULT_FILE" | jq '[.issues // (.review_comments // (.results // []))] | flatten' 2>/dev/null || echo '[]')
|
||||
# Parse OCR JSON output — format varies by command and version
|
||||
# ocr review: .issues[] or .review_comments[]
|
||||
# ocr scan: .comments[] with {path, content, start_line, end_line, ...}
|
||||
RAW_JSON=$(cat "$OCR_RESULT_FILE")
|
||||
|
||||
# Extract comment-like findings from the various OCR output shapes
|
||||
FINDINGS=$(echo "$RAW_JSON" | jq '
|
||||
if .issues then .issues
|
||||
elif .review_comments then .review_comments
|
||||
elif .comments then
|
||||
# Map scan comments to uniform findings format
|
||||
[.comments[] | {
|
||||
file: .path,
|
||||
line: (.start_line // 0),
|
||||
message: .content,
|
||||
severity: "warning",
|
||||
suggestion: (.suggestion_code // "")
|
||||
}]
|
||||
elif .results then .results
|
||||
else []
|
||||
end | flatten
|
||||
' 2>/dev/null || echo '[]')
|
||||
|
||||
# Count by severity
|
||||
ERROR_COUNT=$(echo "$FINDINGS" | jq '[.[] | select(.severity == "error" or .severity == "critical" or .severity == "high")] | length' 2>/dev/null || echo 0)
|
||||
@@ -42,15 +61,26 @@ if [ -f "$OCR_RESULT_FILE" ]; then
|
||||
fi
|
||||
|
||||
# Build summary from OCR report
|
||||
REVIEW_SUMMARY=$(cat "$OCR_RESULT_FILE" | jq -r '.summary // (.review_summary // "\(.issues | length) issues found")' 2>/dev/null || echo "OCR ${OCR_COMMAND} completed")
|
||||
COMMENT_COUNT=$(echo "$FINDINGS" | jq 'length' 2>/dev/null || echo 0)
|
||||
REVIEW_SUMMARY=$(echo "$RAW_JSON" | jq -r '
|
||||
if .summary | type == "object" then
|
||||
"\(.summary.files_reviewed // "?") file(s) reviewed, \(.summary.comments // 0) comment(s)"
|
||||
elif .summary | type == "string" then .summary
|
||||
elif .review_summary then .review_summary
|
||||
else "\(.issues | length) issues found"
|
||||
end
|
||||
' 2>/dev/null || echo "OCR ${OCR_COMMAND} completed")
|
||||
SUMMARY="${ERROR_COUNT} errors, ${WARNING_COUNT} warnings, ${INFO_COUNT:-0} info — ${REVIEW_SUMMARY}"
|
||||
|
||||
# Build outputs containing the raw OCR data
|
||||
AGENT_OUTPUTS=$(cat "$OCR_RESULT_FILE" | jq '{ocr_command: $cmd, issue_count: (.issues | length)}' --arg cmd "$OCR_COMMAND" 2>/dev/null || echo '{}')
|
||||
AGENT_OUTPUTS=$(echo "$RAW_JSON" | jq '{ocr_command: $cmd, comment_count: $count}' \
|
||||
--arg cmd "$OCR_COMMAND" \
|
||||
--argjson count "$COMMENT_COUNT" \
|
||||
2>/dev/null || echo '{}')
|
||||
export AGENT_OUTPUTS
|
||||
|
||||
# Only include findings if there are any
|
||||
if [ "$(echo "$FINDINGS" | jq 'length' 2>/dev/null || echo 0)" -gt 0 ]; then
|
||||
if [ "$COMMENT_COUNT" -gt 0 ]; then
|
||||
export AGENT_FINDINGS="$FINDINGS"
|
||||
fi
|
||||
|
||||
|
||||
@@ -57,34 +57,72 @@ jobs:
|
||||
with:
|
||||
fetch-depth: "0"
|
||||
|
||||
- name: "Run OCR Scan"
|
||||
id: scan
|
||||
- name: "Install and Configure OCR"
|
||||
id: setup
|
||||
shell: bash
|
||||
env:
|
||||
OCR_LLM_URL: ${{ secrets.ocr_llm_url }}
|
||||
OCR_LLM_TOKEN: ${{ secrets.ocr_llm_token }}
|
||||
OCR_LLM_MODEL: ${{ secrets.ocr_llm_model || inputs.ocr-llm-model || vars.OCR_LLM_MODEL || 'claude-sonnet-4-20250514' }}
|
||||
OCR_USE_ANTHROPIC: ${{ inputs.ocr-use-anthropic || vars.OCR_LLM_USE_ANTHROPIC || 'true' }}
|
||||
run: |
|
||||
# Install OCR CLI globally (puts ocr on PATH)
|
||||
npm install -g @alibaba-group/open-code-review
|
||||
|
||||
echo "✅ ocr installed: $(ocr --version 2>&1 || true)"
|
||||
|
||||
# Verify scan subcommand exists
|
||||
if ! ocr scan --help >/dev/null 2>&1; then
|
||||
echo "❌ 'ocr scan' subcommand not found"
|
||||
exit 1
|
||||
fi
|
||||
|
||||
# Configure LLM provider from workflow secrets
|
||||
PROVIDER="ci-custom"
|
||||
ocr config set provider "${PROVIDER}"
|
||||
ocr config set "custom_providers.${PROVIDER}.url" "${OCR_LLM_URL}"
|
||||
ocr config set "custom_providers.${PROVIDER}.api_key" "${OCR_LLM_TOKEN}"
|
||||
ocr config set "custom_providers.${PROVIDER}.model" "${OCR_LLM_MODEL}"
|
||||
|
||||
if [ "${OCR_USE_ANTHROPIC}" = "true" ]; then
|
||||
ocr config set "custom_providers.${PROVIDER}.protocol" "anthropic"
|
||||
else
|
||||
ocr config set "custom_providers.${PROVIDER}.protocol" "openai"
|
||||
fi
|
||||
|
||||
# Test LLM connectivity
|
||||
echo "🔌 Testing LLM connection..."
|
||||
ocr llm test
|
||||
|
||||
- name: "Run OCR Scan"
|
||||
id: scan
|
||||
shell: bash
|
||||
env:
|
||||
OCR_LLM_MODEL: ${{ secrets.ocr_llm_model || inputs.ocr-llm-model || vars.OCR_LLM_MODEL || 'claude-sonnet-4-20250514' }}
|
||||
SCAN_PATH: ${{ inputs.scan-path || github.event.inputs.scan-path || '.' }}
|
||||
run: |
|
||||
mkdir -p .ocr
|
||||
|
||||
# Run full repo scan via npx (guarantees latest OCR CLI version)
|
||||
echo "🔍 Scanning: ${SCAN_PATH}"
|
||||
npx @open-code-review/cli@latest scan \
|
||||
|
||||
# Run full repo scan — JSON output goes to stdout with --audience agent
|
||||
ocr scan \
|
||||
--path "${SCAN_PATH}" \
|
||||
--format json \
|
||||
--audience agent \
|
||||
2>&1 | tee .ocr/scan-output.txt
|
||||
--model "${OCR_LLM_MODEL}" \
|
||||
> .ocr/result.json 2>.ocr/scan-stderr.txt
|
||||
|
||||
# Check for result.json output (OCR writes here)
|
||||
if [ -f ".ocr/result.json" ]; then
|
||||
echo "OCR results saved to .ocr/result.json"
|
||||
elif [ -f ".ocr/report.json" ]; then
|
||||
echo "OCR results saved to .ocr/report.json"
|
||||
# Show stderr (if any) for debugging
|
||||
if [ -s .ocr/scan-stderr.txt ]; then
|
||||
cat .ocr/scan-stderr.txt >&2
|
||||
fi
|
||||
|
||||
if [ -s .ocr/result.json ]; then
|
||||
echo "✅ OCR results saved to .ocr/result.json"
|
||||
else
|
||||
# Try to extract from stdout
|
||||
tail -c 100000 .ocr/scan-output.txt | grep -oP '\{.*\}' | head -1 > .ocr/result.json 2>/dev/null || true
|
||||
echo "⚠️ No JSON output from OCR scan — scan produced no results or failed"
|
||||
rm -f .ocr/result.json
|
||||
fi
|
||||
|
||||
- name: "Build Agent Envelope"
|
||||
|
||||
Reference in New Issue
Block a user