4f70fd7fb8
All checks were successful
Test / rust-fmt-check (pull_request) Successful in 1m18s
Test / frontend-tests (pull_request) Successful in 1m29s
Test / frontend-typecheck (pull_request) Successful in 1m33s
Test / rust-clippy (pull_request) Successful in 3m16s
Test / rust-tests (pull_request) Successful in 4m33s
The PROMPT string contained backtick-quoted text for the Evidence field example. Inside a double-quoted bash string, backticks trigger command substitution, causing 'exact: command not found' at runtime. Fix: build the prompt using a single-quoted heredoc (no shell expansion inside) then splice dynamic values via sed and python3 replace() instead of shell variable interpolation.
371 lines
16 KiB
YAML
371 lines
16 KiB
YAML
name: PR Review Automation
|
|
|
|
on:
|
|
pull_request:
|
|
types: [opened, synchronize, reopened, edited]
|
|
|
|
concurrency:
|
|
group: pr-review-${{ github.event.pull_request.number }}
|
|
cancel-in-progress: true
|
|
|
|
jobs:
|
|
review:
|
|
runs-on: ubuntu-latest
|
|
permissions:
|
|
pull-requests: write
|
|
container:
|
|
image: ubuntu:22.04
|
|
options: --dns 8.8.8.8 --dns 1.1.1.1
|
|
steps:
|
|
- name: Install dependencies
|
|
shell: bash
|
|
run: |
|
|
set -euo pipefail
|
|
apt-get update -qq && apt-get install -y -qq git curl jq
|
|
|
|
- name: Checkout code
|
|
shell: bash
|
|
env:
|
|
REPOSITORY: ${{ github.repository }}
|
|
run: |
|
|
set -euo pipefail
|
|
git init
|
|
git remote add origin "https://gogs.tftsr.com/${REPOSITORY}.git"
|
|
git fetch --depth=1 origin ${{ github.head_ref }}
|
|
git checkout FETCH_HEAD
|
|
|
|
- name: Build review context
|
|
id: context
|
|
shell: bash
|
|
run: |
|
|
set -euo pipefail
|
|
git fetch origin ${{ github.base_ref }}
|
|
|
|
# List changed source files (exclude generated/lock files)
|
|
git diff --name-only origin/${{ github.base_ref }}..HEAD \
|
|
-- ':!Cargo.lock' ':!package-lock.json' ':!*.lock' \
|
|
> /tmp/pr_files.txt
|
|
|
|
FILE_COUNT=$(wc -l < /tmp/pr_files.txt | tr -d ' ')
|
|
echo "files_changed=${FILE_COUNT}" >> $GITHUB_OUTPUT
|
|
|
|
if [ "$FILE_COUNT" -eq 0 ]; then
|
|
echo "No reviewable files changed."
|
|
echo "diff_size=0" >> $GITHUB_OUTPUT
|
|
exit 0
|
|
fi
|
|
|
|
# Build context: full file content for each changed file.
|
|
# Files <= 500 lines: include complete content.
|
|
# Files > 500 lines: include the per-file diff with generous context (±50 lines).
|
|
#
|
|
# Secret scrubbing: match actual credential VALUES only — known API key formats,
|
|
# or keyword="long_quoted_literal" (25+ chars). Never scrub on keyword alone,
|
|
# which would silently delete function signatures, variable declarations, and tests.
|
|
SECRET_PATTERN='AKIA[A-Z0-9]{16}|gh[opsu]_[A-Za-z0-9_]{36,}|xox[baprs]-[0-9]{10,13}-[0-9]{10,13}-[a-zA-Z0-9]{24}|(password|token|api_key|secret)[[:space:]]*=[[:space:]]*["'"'"'][A-Za-z0-9+/_\-!@#]{25,}["'"'"']'
|
|
# Only strip lines that are ENTIRELY a long base64 blob (e.g. PEM cert bodies)
|
|
B64_PATTERN='^[[:space:]]*[A-Za-z0-9+/]{60,}={0,2}[[:space:]]*$'
|
|
|
|
> /tmp/pr_context.txt
|
|
while IFS= read -r file; do
|
|
[ -f "$file" ] || continue
|
|
lines=$(wc -l < "$file" | tr -d ' ')
|
|
printf '\n════════ FILE: %s (%s lines) ════════\n' "$file" "$lines" >> /tmp/pr_context.txt
|
|
if [ "$lines" -le 500 ]; then
|
|
# Full file — model sees the complete implementation
|
|
grep -v -E "$SECRET_PATTERN" "$file" \
|
|
| grep -v -E "$B64_PATTERN" \
|
|
>> /tmp/pr_context.txt || true
|
|
else
|
|
# Large file — emit annotated diff hunks (±50 lines of context each)
|
|
printf '[File too large for full view (%s lines) — showing changed sections only]\n' "$lines" >> /tmp/pr_context.txt
|
|
git diff -U50 origin/${{ github.base_ref }}..HEAD -- "$file" \
|
|
| grep -v -E "$SECRET_PATTERN" \
|
|
| grep -v -E "$B64_PATTERN" \
|
|
>> /tmp/pr_context.txt || true
|
|
fi
|
|
done < /tmp/pr_files.txt
|
|
|
|
TOTAL=$(wc -l < /tmp/pr_context.txt | tr -d ' ')
|
|
echo "diff_size=${TOTAL}" >> $GITHUB_OUTPUT
|
|
|
|
# Cap at 6000 lines so we stay within the model's context window
|
|
if [ "$TOTAL" -gt 6000 ]; then
|
|
head -n 6000 /tmp/pr_context.txt > /tmp/pr_context_capped.txt
|
|
mv /tmp/pr_context_capped.txt /tmp/pr_context.txt
|
|
echo "[CONTEXT TRUNCATED at 6000 lines — ${TOTAL} total]" >> /tmp/pr_context.txt
|
|
fi
|
|
|
|
- name: Build codebase index
|
|
id: index
|
|
if: steps.context.outputs.diff_size != '0'
|
|
shell: bash
|
|
run: |
|
|
set -euo pipefail
|
|
# Build a compact index of everything that EXISTS in this codebase.
|
|
# Included in the prompt so the model cannot invent functions/commands/tables
|
|
# that are not present — any finding referencing something absent from this
|
|
# index is immediately suspect.
|
|
{
|
|
echo "## CODEBASE INDEX"
|
|
echo "These are the ONLY Tauri commands, TypeScript exports, Rust public functions,"
|
|
echo "and database tables that exist in this project. Before raising any finding,"
|
|
echo "confirm that every symbol you cite appears in this list or in the file"
|
|
echo "contents below. If it does not appear in either, your finding is fabricated."
|
|
echo ""
|
|
|
|
echo "### Registered Tauri commands (lib.rs generate_handler![]):"
|
|
grep -oE 'commands::[a-z_]+::[a-z_]+' src-tauri/src/lib.rs 2>/dev/null \
|
|
| sort -u | sed 's/^/ /' || true
|
|
echo ""
|
|
|
|
echo "### TypeScript invoke wrappers (src/lib/tauriCommands.ts):"
|
|
grep -E '^export (const|interface|type) ' src/lib/tauriCommands.ts 2>/dev/null \
|
|
| sed 's/^/ /' || true
|
|
echo ""
|
|
|
|
echo "### Public Rust functions in src-tauri/src/commands/:"
|
|
grep -rh --include='*.rs' '^pub ' src-tauri/src/commands/ 2>/dev/null \
|
|
| grep 'fn ' | sed 's/^/ /' | sort || true
|
|
echo ""
|
|
|
|
echo "### Database tables (src-tauri/src/db/migrations.rs):"
|
|
grep -oE '"[0-9]+_[a-z_]+"' src-tauri/src/db/migrations.rs 2>/dev/null \
|
|
| tr -d '"' | sed 's/^/ /' || true
|
|
echo ""
|
|
} > /tmp/codebase_index.txt
|
|
|
|
INDEX_LINES=$(wc -l < /tmp/codebase_index.txt | tr -d ' ')
|
|
echo "index_lines=${INDEX_LINES}" >> $GITHUB_OUTPUT
|
|
echo "Built codebase index: ${INDEX_LINES} lines"
|
|
|
|
- name: Analyze with LLM
|
|
id: analyze
|
|
if: steps.context.outputs.diff_size != '0'
|
|
shell: bash
|
|
env:
|
|
LITELLM_URL: http://172.0.0.29:11434/v1
|
|
LITELLM_API_KEY: ${{ secrets.OLLAMA_API_KEY }}
|
|
PR_TITLE: ${{ github.event.pull_request.title }}
|
|
PR_NUMBER: ${{ github.event.pull_request.number }}
|
|
run: |
|
|
set -euo pipefail
|
|
CHANGED_FILES=$(tr '\n' ' ' < /tmp/pr_files.txt)
|
|
INDEX=$(cat /tmp/codebase_index.txt)
|
|
CONTEXT=$(cat /tmp/pr_context.txt)
|
|
|
|
# Build the prompt via a single-quoted heredoc so the shell never
|
|
# interprets backticks, dollar signs, or other special characters inside.
|
|
# Variables that must expand ($PR_TITLE etc.) are spliced in by jq --arg,
|
|
# not by shell interpolation, so the prompt text itself is always literal.
|
|
PROMPT_TEMPLATE=$(cat << 'ENDPROMPT'
|
|
You are a senior engineer performing a code review for the following pull request.
|
|
|
|
PR Title: __PR_TITLE__
|
|
Files changed: __CHANGED_FILES__
|
|
|
|
---
|
|
__INDEX__
|
|
---
|
|
|
|
## Changed file contents
|
|
|
|
Each section below contains the COMPLETE, FINAL content of one changed file after
|
|
the PR's changes have been applied. This is the full file — not a diff. For files
|
|
over 500 lines, only the changed sections are shown with surrounding context.
|
|
|
|
---
|
|
__CONTEXT__
|
|
---
|
|
|
|
## Instructions
|
|
|
|
Before raising any finding:
|
|
1. Confirm every symbol (function name, command name, variable) you cite exists in
|
|
the CODEBASE INDEX above or in the file contents above. If it appears in neither,
|
|
discard the finding — it does not exist in this project.
|
|
2. Quote the exact line(s) from the file contents that support the finding.
|
|
3. Confirm the issue is a genuine problem, not intentional design.
|
|
4. If any step fails, discard the finding silently — do not mention it.
|
|
|
|
Do NOT show your reasoning process. Do NOT mention discarded findings.
|
|
Output only confirmed issues.
|
|
|
|
Severity levels:
|
|
- BLOCKER: will fail to compile, corrupt data, or introduce a security vulnerability
|
|
- WARNING: real risk that should be addressed before merge
|
|
- SUGGESTION: minor improvement, acceptable as a follow-up PR
|
|
|
|
Focus on: security bugs, logic errors, data loss, injection vectors, unhandled
|
|
error paths that could silently corrupt state.
|
|
Ignore: style preferences, missing comments, speculative future concerns.
|
|
|
|
## Output format (do not deviate)
|
|
|
|
**Summary** (2-3 sentences: what the PR does and your overall assessment)
|
|
|
|
**Findings**
|
|
- [SEVERITY] file:line -- description
|
|
Evidence: quoted line from the file above
|
|
Fix: concrete suggested change
|
|
|
|
(Write "No findings." if there are none.)
|
|
|
|
**Verdict**: APPROVE / APPROVE WITH COMMENTS / REQUEST CHANGES
|
|
ENDPROMPT
|
|
)
|
|
|
|
# Splice runtime values into the template using sed so nothing is eval'd
|
|
PROMPT=$(printf '%s' "$PROMPT_TEMPLATE" \
|
|
| sed "s|__PR_TITLE__|${PR_TITLE}|g" \
|
|
| sed "s|__CHANGED_FILES__|${CHANGED_FILES}|g")
|
|
# INDEX and CONTEXT may contain special sed chars — use python for those
|
|
PROMPT=$(python3 -c "
|
|
import sys
|
|
template = sys.stdin.read()
|
|
index = open('/tmp/codebase_index.txt').read()
|
|
context = open('/tmp/pr_context.txt').read()
|
|
print(template.replace('__INDEX__', index).replace('__CONTEXT__', context), end='')
|
|
" <<< "$PROMPT")
|
|
|
|
BODY=$(jq -cn \
|
|
--arg model "qwen3-coder-next" \
|
|
--arg content "$PROMPT" \
|
|
'{model: $model, messages: [{role: "user", content: $content}], stream: false}')
|
|
echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] PR #${PR_NUMBER} - Calling liteLLM API (${#BODY} bytes)..."
|
|
HTTP_CODE=$(curl -s --max-time 300 --connect-timeout 30 \
|
|
--retry 3 --retry-delay 10 --retry-connrefused --retry-max-time 300 \
|
|
-o /tmp/llm_response.json -w "%{http_code}" \
|
|
-X POST "$LITELLM_URL/chat/completions" \
|
|
-H "Authorization: Bearer $LITELLM_API_KEY" \
|
|
-H "Content-Type: application/json" \
|
|
-d "$BODY")
|
|
echo "HTTP status: $HTTP_CODE"
|
|
echo "Response file size: $(wc -c < /tmp/llm_response.json) bytes"
|
|
if [ "$HTTP_CODE" != "200" ]; then
|
|
echo "ERROR: liteLLM returned HTTP $HTTP_CODE"
|
|
cat /tmp/llm_response.json
|
|
exit 1
|
|
fi
|
|
if ! jq empty /tmp/llm_response.json 2>/dev/null; then
|
|
echo "ERROR: Invalid JSON response from liteLLM"
|
|
cat /tmp/llm_response.json
|
|
exit 1
|
|
fi
|
|
REVIEW=$(jq -r '.choices[0].message.content // empty' /tmp/llm_response.json)
|
|
if [ -z "$REVIEW" ]; then
|
|
echo "ERROR: No content in liteLLM response"
|
|
exit 1
|
|
fi
|
|
echo "Review length: ${#REVIEW} chars"
|
|
echo "$REVIEW" > /tmp/pr_review.txt
|
|
|
|
- name: Verify findings against codebase
|
|
if: steps.analyze.outcome == 'success'
|
|
shell: bash
|
|
run: |
|
|
set -euo pipefail
|
|
# For each finding that contains a fenced code block under "Evidence:",
|
|
# grep at least one substantial line of that block against the FULL repository.
|
|
# Searching the full repo (not just changed files) prevents false UNVERIFIED
|
|
# tags when the model correctly quotes unchanged files, while still flagging
|
|
# fabricated code that doesn't exist anywhere in the codebase.
|
|
python3 - << 'PYEOF'
|
|
import re, os, subprocess
|
|
|
|
review = open('/tmp/pr_review.txt').read()
|
|
|
|
# Load ENTIRE tracked repository (all .rs, .ts, .tsx, .yml, .toml, .json files)
|
|
result = subprocess.run(
|
|
['git', 'ls-files', '--',
|
|
'*.rs', '*.ts', '*.tsx', '*.yml', '*.yaml', '*.toml', '*.json', '*.sql'],
|
|
capture_output=True, text=True
|
|
)
|
|
all_tracked = [f.strip() for f in result.stdout.splitlines() if f.strip()]
|
|
|
|
all_content_parts = []
|
|
for path in all_tracked:
|
|
if os.path.isfile(path):
|
|
try:
|
|
all_content_parts.append(open(path).read())
|
|
except Exception:
|
|
pass
|
|
all_content = '\n'.join(all_content_parts)
|
|
|
|
def evidence_exists(block: str) -> bool:
|
|
"""True if ≥1 significant line from the block is found verbatim in changed files."""
|
|
for raw in block.splitlines():
|
|
line = raw.lstrip('+-').strip()
|
|
# Skip blank, very short, pure-comment, or diff-header lines
|
|
if len(line) < 20:
|
|
continue
|
|
if line.startswith(('//','#','/*','*','Fix:','Evidence:','---','+++')):
|
|
continue
|
|
if line in all_content:
|
|
return True
|
|
return False
|
|
|
|
# Split on finding markers; re-join after optional tagging
|
|
severity_re = re.compile(r'\[(BLOCKER|WARNING|SUGGESTION)\]')
|
|
|
|
def tag_if_unverified(finding_text: str) -> str:
|
|
code_match = re.search(r'```[^\n]*\n(.*?)```', finding_text, re.DOTALL)
|
|
if code_match and not evidence_exists(code_match.group(1)):
|
|
# Replace first severity tag with a prefixed version
|
|
return severity_re.sub(
|
|
lambda m: f'[{m.group(1)} — ⚠️ UNVERIFIED: evidence not found in PR files]',
|
|
finding_text, count=1
|
|
)
|
|
return finding_text
|
|
|
|
# Split review into preamble + individual finding blocks
|
|
# Each block starts at a severity marker line
|
|
parts = re.split(r'(?=^\[(?:BLOCKER|WARNING|SUGGESTION)\])', review, flags=re.MULTILINE)
|
|
result = parts[0] # preamble (Summary, etc.)
|
|
for block in parts[1:]:
|
|
result += tag_if_unverified(block)
|
|
|
|
open('/tmp/pr_review.txt', 'w').write(result)
|
|
print(f"Verification complete — {len(parts)-1} finding(s) checked.")
|
|
PYEOF
|
|
|
|
- name: Post review comment
|
|
if: always() && steps.context.outputs.diff_size != '0'
|
|
shell: bash
|
|
env:
|
|
TF_TOKEN: ${{ secrets.TFT_GITEA_TOKEN }}
|
|
PR_NUMBER: ${{ github.event.pull_request.number }}
|
|
REPOSITORY: ${{ github.repository }}
|
|
run: |
|
|
set -euo pipefail
|
|
if [ -z "${TF_TOKEN:-}" ]; then
|
|
echo "ERROR: TFT_GITEA_TOKEN secret is not set"
|
|
exit 1
|
|
fi
|
|
if [ -f "/tmp/pr_review.txt" ] && [ -s "/tmp/pr_review.txt" ]; then
|
|
REVIEW_BODY=$(head -c 65536 /tmp/pr_review.txt)
|
|
BODY=$(jq -n \
|
|
--arg body "Automated PR Review (qwen3-coder-next via liteLLM):\n\n${REVIEW_BODY}" \
|
|
'{body: $body, event: "COMMENT"}')
|
|
else
|
|
BODY=$(jq -n \
|
|
'{body: "Automated PR Review could not be completed - LLM analysis failed or produced no output.", event: "COMMENT"}')
|
|
fi
|
|
HTTP_CODE=$(curl -s --max-time 30 --connect-timeout 10 \
|
|
-o /tmp/review_post_response.json -w "%{http_code}" \
|
|
-X POST "https://gogs.tftsr.com/api/v1/repos/${REPOSITORY}/pulls/${PR_NUMBER}/reviews" \
|
|
-H "Authorization: Bearer $TF_TOKEN" \
|
|
-H "Content-Type: application/json" \
|
|
-d "$BODY")
|
|
echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] Post review HTTP status: $HTTP_CODE"
|
|
if [ "$HTTP_CODE" != "200" ] && [ "$HTTP_CODE" != "201" ]; then
|
|
echo "ERROR: Failed to post review (HTTP $HTTP_CODE)"
|
|
cat /tmp/review_post_response.json
|
|
exit 1
|
|
fi
|
|
|
|
- name: Cleanup
|
|
if: always()
|
|
shell: bash
|
|
run: rm -f /tmp/pr_diff.txt /tmp/pr_context.txt /tmp/codebase_index.txt /tmp/llm_response.json /tmp/pr_review.txt /tmp/review_post_response.json /tmp/pr_files.txt
|