From cda88b5c03f43019faa00387b28a1e0676de6920 Mon Sep 17 00:00:00 2001 From: kumagoro1202 Date: Thu, 26 Mar 2026 23:23:56 +0900 Subject: [PATCH 1/2] feat(nabledge-1.4): add upgrade-check workflow for Nablarch version upgrade impact analysis - Add workflows/upgrade-check.md: 4-step workflow for detecting upgrade impacts using rule-based analysis (R1-R8) and optional LLM-assisted evaluation - Add scripts/upgrade-checker.sh: bash script implementing R1-R8 detection rules covering 88% of affected items across v5/v6 release notes - Update SKILL.md: add upgrade-check workflow entry in Step 0 routing Based on analysis of 83 affected items (v5: 75, v6: 8) from release notes JSON. --- .claude/skills/nabledge-1.4/SKILL.md | 4 + .../nabledge-1.4/scripts/upgrade-checker.sh | 632 ++++++++++++++++++ .../nabledge-1.4/workflows/upgrade-check.md | 143 ++++ 3 files changed, 779 insertions(+) create mode 100755 .claude/skills/nabledge-1.4/scripts/upgrade-checker.sh create mode 100644 .claude/skills/nabledge-1.4/workflows/upgrade-check.md diff --git a/.claude/skills/nabledge-1.4/SKILL.md b/.claude/skills/nabledge-1.4/SKILL.md index cc849cfe3..198999dcf 100644 --- a/.claude/skills/nabledge-1.4/SKILL.md +++ b/.claude/skills/nabledge-1.4/SKILL.md @@ -41,6 +41,10 @@ nabledge-1.4 code-analysis **"code-analysis" argument** (`nabledge-1.4 code-analysis`): - Execute `workflows/code-analysis.md` to analyze user's code +**"upgrade-check" argument** (`nabledge-1.4 upgrade-check [--from ] [--to ] [--project-dir ]`): +- Execute `workflows/upgrade-check.md` to analyze upgrade impact +- This workflow detects affected items between Nablarch versions using rule-based analysis + ## Critical Constraints **Knowledge files only**: Answer using ONLY information in `knowledge/*.json`. DO NOT use external knowledge or LLM training data. diff --git a/.claude/skills/nabledge-1.4/scripts/upgrade-checker.sh b/.claude/skills/nabledge-1.4/scripts/upgrade-checker.sh new file mode 100755 index 000000000..2c803947f --- /dev/null +++ b/.claude/skills/nabledge-1.4/scripts/upgrade-checker.sh @@ -0,0 +1,632 @@ +#!/bin/bash +# upgrade-checker.sh - Rule-based Nablarch upgrade impact checker +# +# Scans release note JSON files between two Nablarch versions and checks +# whether the changes affect the target project. +# +# Rules implemented: +# R1: Sections with "システムへの影響あり" in index[].hints +# R2: Sections with "影響がある変更" in index[].title (supplement to R1) +# R3: Warning/important blocks in section body (> **警告**: / > **重要**:) +# R4: Table rows where the "影響" column value is "あり" +# R5: FQCNs extracted from backticks (com.nablarch.*) +# R6: Config keys extracted from backticks (nablarch.*) +# R7: Artifact IDs extracted from table "修正バージョン"/"アーティファクトID" columns +# R8: Project grep for R5/R6/R7 extracted items +# +# Usage: +# upgrade-checker.sh --release-notes-dir DIR --project-dir DIR --from VERSION --to VERSION +# +# Example: +# upgrade-checker.sh \ +# --release-notes-dir ~/nabledge-dev/tools/knowledge-creator/.cache/v5/knowledge/releases/releases \ +# --project-dir ~/my-nablarch-project \ +# --from 5u13 \ +# --to 5u18 + +set -euo pipefail + +SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)" +SKILL_DIR="$(cd "$SCRIPT_DIR/.." && pwd)" + +# --------------------------------------------------------------------------- +# Argument parsing +# --------------------------------------------------------------------------- +RELEASE_NOTES_DIR="" +PROJECT_DIR="" +FROM_VERSION="" +TO_VERSION="" + +usage() { + cat >&2 <&2; usage ;; + esac +done + +# --------------------------------------------------------------------------- +# Validation +# --------------------------------------------------------------------------- +if [[ -z "$RELEASE_NOTES_DIR" || -z "$PROJECT_DIR" || -z "$FROM_VERSION" || -z "$TO_VERSION" ]]; then + echo "Error: All four arguments are required." >&2 + usage +fi + +if [[ ! -d "$RELEASE_NOTES_DIR" ]]; then + echo "Error: --release-notes-dir does not exist: $RELEASE_NOTES_DIR" >&2 + exit 1 +fi + +if [[ ! -d "$PROJECT_DIR" ]]; then + echo "Error: --project-dir does not exist: $PROJECT_DIR" >&2 + exit 1 +fi + +# Validate version format: digits optionally followed by 'u' + digits (e.g. 5, 5u13, 6u3) +if ! echo "$FROM_VERSION" | grep -qE '^[0-9]+(u[0-9]+)?$'; then + echo "Error: Invalid --from version format: $FROM_VERSION (expected e.g. 5u13, 6, 6u3)" >&2 + usage +fi + +if ! echo "$TO_VERSION" | grep -qE '^[0-9]+(u[0-9]+)?$'; then + echo "Error: Invalid --to version format: $TO_VERSION (expected e.g. 5u13, 6, 6u3)" >&2 + usage +fi + +# Check jq availability +if ! command -v jq &>/dev/null; then + echo "Error: jq is required but not found. Install jq or ensure it is on PATH." >&2 + exit 1 +fi + +# --------------------------------------------------------------------------- +# Version utilities (implemented in Python for robust comparison) +# --------------------------------------------------------------------------- + +# Convert version string "5u13" -> sortable integer 5013, "6" -> 6000, "6u3" -> 6003 +version_to_int() { + python3 -c " +import sys, re +v = sys.argv[1] +m = re.match(r'^(\d+)(?:u(\d+))?$', v) +if not m: + sys.exit(1) +major = int(m.group(1)) +minor = int(m.group(2)) if m.group(2) else 0 +print(major * 1000 + minor) +" "$1" +} + +FROM_INT=$(version_to_int "$FROM_VERSION") +TO_INT=$(version_to_int "$TO_VERSION") + +if [[ "$FROM_INT" -ge "$TO_INT" ]]; then + echo "Error: --from ($FROM_VERSION) must be less than --to ($TO_VERSION)" >&2 + exit 1 +fi + +# --------------------------------------------------------------------------- +# Phase 1: Filter JSON files by version range (exclusive from, inclusive to) +# Files naming pattern: releases-nablarch{major}(u{minor})?-releasenote.json +# We include files where: FROM_INT < file_version_int <= TO_INT +# --------------------------------------------------------------------------- + +SELECTED_FILES=() + +while IFS= read -r -d '' filepath; do + filename="$(basename "$filepath")" + file_ver=$(python3 -c " +import re, sys +m = re.match(r'releases-nablarch(\d+)(u(\d+))?-releasenote\.json', sys.argv[1]) +if not m: + sys.exit(0) +major = int(m.group(1)) +minor = int(m.group(3)) if m.group(3) else 0 +print(major * 1000 + minor) +" "$filename" 2>/dev/null || true) + + if [[ -z "$file_ver" ]]; then + continue # filename does not match the pattern + fi + + if [[ "$file_ver" -gt "$FROM_INT" && "$file_ver" -le "$TO_INT" ]]; then + SELECTED_FILES+=("$filepath") + fi +done < <(find "$RELEASE_NOTES_DIR" -maxdepth 1 -name "releases-nablarch*-releasenote.json" -print0 | sort -z) + +if [[ ${#SELECTED_FILES[@]} -eq 0 ]]; then + echo "# Upgrade Check Results: Nablarch ${FROM_VERSION} → ${TO_VERSION}" + echo "" + echo "No release note JSON files found in the specified version range." + echo "Directory: $RELEASE_NOTES_DIR" + exit 0 +fi + +# --------------------------------------------------------------------------- +# Phase 2 & 3: Apply R1-R8 rules using Python for robust text processing +# --------------------------------------------------------------------------- + +python3 - "$RELEASE_NOTES_DIR" "$PROJECT_DIR" "$FROM_VERSION" "$TO_VERSION" "${SELECTED_FILES[@]}" <<'PYEOF' +import sys +import os +import re +import json +import subprocess + +_, release_notes_dir, project_dir, from_ver, to_ver, *selected_files = sys.argv + +# Sort files by version +def file_to_ver(path): + m = re.search(r'nablarch(\d+)(u(\d+))?-releasenote', os.path.basename(path)) + if not m: + return 0 + return int(m.group(1)) * 1000 + (int(m.group(3)) if m.group(3) else 0) + +selected_files = sorted(selected_files, key=file_to_ver) + +# --------------------------------------------------------------------------- +# Rule implementations +# --------------------------------------------------------------------------- + +def extract_file_version(path): + """Extract version string from filename, e.g. '5u13' or '6u3'.""" + m = re.search(r'nablarch(\d+)(u(\d+))?-releasenote', os.path.basename(path)) + if not m: + return "unknown" + major = m.group(1) + minor = m.group(3) + return f"{major}u{minor}" if minor else major + +def r1_hints_match(index_entries): + """R1: Find section IDs where hints contain 'システムへの影響あり'.""" + matched = [] + for entry in index_entries: + hints = entry.get("hints", []) + if any("システムへの影響あり" in h for h in hints): + matched.append(entry["id"]) + return matched + +def r2_title_match(index_entries): + """R2: Find section IDs where title contains '影響がある変更'.""" + matched = [] + for entry in index_entries: + title = entry.get("title", "") + if "影響がある変更" in title: + matched.append(entry["id"]) + return matched + +def r3_warning_blocks(section_text): + """R3: Extract warning/important blocks from section body. + Pattern: lines starting with '> **警告**:' or '> **重要**:' etc. + Returns list of full matched lines. + """ + pattern = re.compile(r'^> \*\*(警告|重要|Warning|Important)\*\*:.*', re.MULTILINE) + return [m.group(0) for m in pattern.finditer(section_text)] + +def r4_affected_table_rows(section_text): + """R4: Extract rows from Markdown tables where the 影響 column is 'あり'. + Column names: 'システムへの影響', '影響', 'システム影響' + Values matching: 'あり' (including 'あり（開発）', 'あり（本番）') + Returns list of matched row strings. + """ + results = [] + lines = section_text.split("\n") + i = 0 + while i < len(lines): + line = lines[i].strip() + # Look for a table header row containing an 影響 column + if "|" in line: + # Find header + header_match = re.match(r'\|(.+)\|', line) + if header_match: + headers = [h.strip() for h in line.split("|")[1:-1]] + # Check if any header matches 影響 column pattern + impact_col_idx = None + for idx, h in enumerate(headers): + # Strip markdown bold/other formatting + h_clean = re.sub(r'\*+', '', h).strip() + if re.search(r'システムへの影響|^影響$|システム影響', h_clean): + impact_col_idx = idx + break + + if impact_col_idx is not None: + # Skip separator row (---|---|...) + i += 1 + if i < len(lines) and re.match(r'\|[\s\-:|]+\|', lines[i]): + i += 1 + # Process data rows + while i < len(lines): + row = lines[i].strip() + if not row.startswith("|"): + break + cells = [c.strip() for c in row.split("|")[1:-1]] + if impact_col_idx < len(cells): + cell = cells[impact_col_idx] + # Remove markdown formatting + cell_clean = re.sub(r'\*+', '', cell).strip() + # Match 'あり' but not 'なし' + if re.match(r'あり', cell_clean): + results.append(row) + i += 1 + continue + i += 1 + return results + +def r5_extract_fqcns(section_text): + """R5: Extract FQCNs from backticks in section body. + Pattern: `com.nablarch.[A-Za-z][A-Za-z0-9._$]*` + """ + pattern = re.compile(r'`(com\.nablarch\.[A-Za-z][A-Za-z0-9._$]*)`') + return list(dict.fromkeys(pattern.findall(section_text))) + +def r6_extract_config_keys(section_text): + """R6: Extract config keys from backticks in section body. + Pattern: `nablarch.[a-z][a-zA-Z0-9._-]*` + """ + pattern = re.compile(r'`(nablarch\.[a-z][a-zA-Z0-9._-]*)`') + return list(dict.fromkeys(pattern.findall(section_text))) + +def r7_extract_artifact_ids(section_text): + """R7: Extract Nablarch artifact IDs from table cells. + Pattern: nablarch-[a-z-]+ + """ + pattern = re.compile(r'\bnablarch-[a-z][a-z-]+\b') + matches = pattern.findall(section_text) + return list(dict.fromkeys(matches)) + +def r8_project_grep(items, project_dir, file_extensions=None): + """R8: Grep the project directory for each item. + Returns dict: {item: [matching_file_paths]} or {item: None} on error. + """ + if file_extensions is None: + file_extensions = [".java", ".xml", ".properties", ".yaml", ".yml"] + + results = {} + for item in items: + try: + cmd = ["grep", "-rl", "--include=" + ",".join(f"*{ext}" for ext in file_extensions), + item, project_dir] + # Use find + xargs approach for better glob handling + find_cmd = ["grep", "-rl", item, project_dir, + "--include=*.java", "--include=*.xml", + "--include=*.properties", "--include=*.yaml", + "--include=*.yml"] + proc = subprocess.run(find_cmd, capture_output=True, text=True, timeout=30) + if proc.returncode == 0: + hits = [h.strip() for h in proc.stdout.strip().split("\n") if h.strip()] + results[item] = hits + elif proc.returncode == 1: + results[item] = [] # no match (not an error) + else: + results[item] = None # grep error + except subprocess.TimeoutExpired: + results[item] = None + except Exception: + results[item] = None + return results + +def r8_pom_grep(artifact_ids, project_dir): + """R8: Grep pom.xml files for artifact IDs.""" + results = {} + for aid in artifact_ids: + try: + find_cmd = ["grep", "-rl", aid, project_dir, "--include=*.xml"] + proc = subprocess.run(find_cmd, capture_output=True, text=True, timeout=30) + if proc.returncode == 0: + hits = [h.strip() for h in proc.stdout.strip().split("\n") if h.strip()] + results[aid] = hits + elif proc.returncode == 1: + results[aid] = [] + else: + results[aid] = None + except subprocess.TimeoutExpired: + results[aid] = None + except Exception: + results[aid] = None + return results + +# --------------------------------------------------------------------------- +# Main processing +# --------------------------------------------------------------------------- + +affected_items = [] # {file_ver, file, section_id, rule, pattern, project_hit} +undecidable_items = [] # {file_ver, file, section_id, reason} +not_affected_count = 0 + +all_fqcns = [] +all_config_keys = [] +all_artifact_ids = [] + +# Store per-file, per-section data for final output +file_section_data = {} # (file_ver, section_id) -> {fqcns, config_keys, artifact_ids, ...} + +for filepath in selected_files: + file_ver = extract_file_version(filepath) + try: + with open(filepath, "r", encoding="utf-8") as f: + data = json.load(f) + except (json.JSONDecodeError, IOError) as e: + undecidable_items.append({ + "file_ver": file_ver, + "file": os.path.basename(filepath), + "section_id": "*", + "reason": f"JSON parse error: {e}" + }) + continue + + index_entries = data.get("index", []) + sections = data.get("sections", {}) + + # Apply R1 and R2 to find affected section IDs + r1_sections = set(r1_hints_match(index_entries)) + r2_sections = set(r2_title_match(index_entries)) + candidate_sections = r1_sections | r2_sections + + # Process each section + for entry in index_entries: + section_id = entry["id"] + section_text = sections.get(section_id, "") + + section_key = (file_ver, section_id) + is_affected_section = section_id in candidate_sections + + # Per-section extracted items + fqcns = r5_extract_fqcns(section_text) + config_keys = r6_extract_config_keys(section_text) + artifact_ids = r7_extract_artifact_ids(section_text) + + all_fqcns.extend(fqcns) + all_config_keys.extend(config_keys) + all_artifact_ids.extend(artifact_ids) + + file_section_data[section_key] = { + "file": os.path.basename(filepath), + "file_ver": file_ver, + "section_id": section_id, + "section_title": entry.get("title", ""), + "fqcns": fqcns, + "config_keys": config_keys, + "artifact_ids": artifact_ids, + } + + if is_affected_section: + # R1 match + if section_id in r1_sections: + affected_items.append({ + "file_ver": file_ver, + "file": os.path.basename(filepath), + "section_id": section_id, + "section_title": entry.get("title", ""), + "rule": "R1 (hints: システムへの影響あり)", + "pattern": "hints contains 'システムへの影響あり'", + "project_hit": "", + "fqcns": fqcns, + "config_keys": config_keys, + "artifact_ids": artifact_ids, + }) + elif section_id in r2_sections: + affected_items.append({ + "file_ver": file_ver, + "file": os.path.basename(filepath), + "section_id": section_id, + "section_title": entry.get("title", ""), + "rule": "R2 (title: 影響がある変更)", + "pattern": "title contains '影響がある変更'", + "project_hit": "", + "fqcns": fqcns, + "config_keys": config_keys, + "artifact_ids": artifact_ids, + }) + else: + # For non-candidate sections, still check R3 and R4 + r3_matches = r3_warning_blocks(section_text) + r4_matches = r4_affected_table_rows(section_text) + + if r3_matches: + affected_items.append({ + "file_ver": file_ver, + "file": os.path.basename(filepath), + "section_id": section_id, + "section_title": entry.get("title", ""), + "rule": "R3 (警告/重要ブロック)", + "pattern": r3_matches[0][:80] if r3_matches else "", + "project_hit": "", + "fqcns": fqcns, + "config_keys": config_keys, + "artifact_ids": artifact_ids, + }) + elif r4_matches: + affected_items.append({ + "file_ver": file_ver, + "file": os.path.basename(filepath), + "section_id": section_id, + "section_title": entry.get("title", ""), + "rule": "R4 (テーブル影響列: あり)", + "pattern": r4_matches[0][:80] if r4_matches else "", + "project_hit": "", + "fqcns": fqcns, + "config_keys": config_keys, + "artifact_ids": artifact_ids, + }) + else: + not_affected_count += 1 + + # Also apply R3/R4 to candidate sections (for completeness) + for section_id in candidate_sections: + section_text = sections.get(section_id, "") + # These are already marked as affected; R3/R4 add detail but don't create new entries + +# Deduplicate all extracted items +all_fqcns = list(dict.fromkeys(all_fqcns)) +all_config_keys = list(dict.fromkeys(all_config_keys)) +all_artifact_ids = list(dict.fromkeys(all_artifact_ids)) + +# --------------------------------------------------------------------------- +# Phase 3: R8 - Project grep +# --------------------------------------------------------------------------- +fqcn_hits = r8_project_grep(all_fqcns, project_dir) if all_fqcns else {} +config_hits = r8_project_grep(all_config_keys, project_dir) if all_config_keys else {} +artifact_hits = r8_pom_grep(all_artifact_ids, project_dir) if all_artifact_ids else {} + +def format_project_hits(item, hits_dict): + """Format project grep results as short paths.""" + hits = hits_dict.get(item) + if hits is None: + return "⚠ grep error" + if not hits: + return "(no match)" + # Shorten paths relative to project_dir + short = [] + for h in hits[:3]: # limit to 3 for readability + try: + rel = os.path.relpath(h, project_dir) + except ValueError: + rel = h + short.append(rel) + suffix = f" (+{len(hits)-3} more)" if len(hits) > 3 else "" + return ", ".join(short) + suffix + +# Enrich affected_items with R8 project hit info +for item in affected_items: + hit_parts = [] + + # Check FQCNs + for fqcn in item.get("fqcns", []): + hits = fqcn_hits.get(fqcn) + if hits: + hit_parts.append(f"{fqcn}: {format_project_hits(fqcn, fqcn_hits)}") + + # Check config keys + for key in item.get("config_keys", []): + hits = config_hits.get(key) + if hits: + hit_parts.append(f"{key}: {format_project_hits(key, config_hits)}") + + # Check artifact IDs + for aid in item.get("artifact_ids", []): + hits = artifact_hits.get(aid) + if hits: + hit_parts.append(f"{aid}: {format_project_hits(aid, artifact_hits)}") + + if hit_parts: + item["project_hit"] = "; ".join(hit_parts[:2]) # limit for table width + if len(hit_parts) > 2: + item["project_hit"] += f" (+{len(hit_parts)-2} more)" + else: + item["project_hit"] = "(no project match)" + +# --------------------------------------------------------------------------- +# Phase 4: Output Markdown report +# --------------------------------------------------------------------------- +total_sections = len(affected_items) + len(undecidable_items) + not_affected_count + +print(f"# Upgrade Check Results: Nablarch {from_ver} → {to_ver}") +print() +print(f"**Release notes analyzed**: {len(selected_files)} file(s)") +print() +print("## Summary") +print() +print(f"- Affected items (rule-based): {len(affected_items)}") +print(f"- Undecidable items: {len(undecidable_items)}") +print(f"- Not affected: {not_affected_count}") +print(f"- Total sections analyzed: {total_sections}") +print() + +# --- Affected Items --- +print("## Affected Items") +print() +if affected_items: + print("| # | Version | Section | Detection Rule | Matched Pattern | Project Hit |") + print("|---|---------|---------|----------------|-----------------|-------------|") + for i, item in enumerate(affected_items, 1): + section_label = f"{item['section_id']} ({item['section_title'][:30]})" if item.get('section_title') else item['section_id'] + pattern_short = item['pattern'][:60] if item['pattern'] else "" + print(f"| {i} | {item['file_ver']} | {section_label} | {item['rule']} | {pattern_short} | {item['project_hit']} |") +else: + print("(No affected items detected by rule-based analysis)") +print() + +# --- R5/R6/R7 Extracted Items --- +if all_fqcns or all_config_keys or all_artifact_ids: + print("## Extracted Technical Items") + print() + + if all_fqcns: + print("### FQCNs (R5)") + print() + print("| FQCN | Project Hit |") + print("|------|-------------|") + for fqcn in all_fqcns: + print(f"| `{fqcn}` | {format_project_hits(fqcn, fqcn_hits)} |") + print() + + if all_config_keys: + print("### Config Keys (R6)") + print() + print("| Config Key | Project Hit |") + print("|------------|-------------|") + for key in all_config_keys: + print(f"| `{key}` | {format_project_hits(key, config_hits)} |") + print() + + if all_artifact_ids: + print("### Artifact IDs (R7 → pom.xml grep)") + print() + print("| Artifact ID | Project Hit |") + print("|-------------|-------------|") + for aid in all_artifact_ids: + print(f"| `{aid}` | {format_project_hits(aid, artifact_hits)} |") + print() + +# --- Undecidable Items --- +print("## Undecidable Items (requires LLM evaluation)") +print() +if undecidable_items: + print("| # | Version | Section | Reason |") + print("|---|---------|---------|--------|") + for i, item in enumerate(undecidable_items, 1): + print(f"| {i} | {item['file_ver']} | {item['section_id']} | {item['reason']} |") +else: + print("(No undecidable items)") +print() + +# --- Not Affected --- +print("## Not Affected") +print() +if not_affected_count > 0: + print(f"{not_affected_count} section(s) had no rule-based matches and are considered not affected.") +else: + print("(All sections were classified by rule-based analysis)") +print() + +# --- Footer --- +print("---") +print() +print("**Notes**:") +print("- 'Affected Items' are sections where R1-R4 detected impact signals.") +print("- 'Project Hit' shows whether R5/R6/R7 extracted items were found in the project source.") +print("- Sections with no R1-R4 signal and no project hits are marked as 'Not Affected'.") +print("- Items requiring human/LLM review (e.g. implicit impacts, conditional changes) are listed under 'Undecidable'.") + +PYEOF diff --git a/.claude/skills/nabledge-1.4/workflows/upgrade-check.md b/.claude/skills/nabledge-1.4/workflows/upgrade-check.md new file mode 100644 index 000000000..2fd8efc59 --- /dev/null +++ b/.claude/skills/nabledge-1.4/workflows/upgrade-check.md @@ -0,0 +1,143 @@ +# Upgrade Check Workflow + +Analyze the impact of upgrading Nablarch from one version to another. Runs a rule-based script to detect affected items and uses LLM evaluation for undecidable cases, then generates a structured Markdown report. + +## Input + +- Target project root path +- Upgrade source version (e.g., `5u13`) +- Upgrade target version (e.g., `5u18`) + +## Output + +Markdown impact assessment report presented to the user + +## Steps + +### Step 0: Detect Nablarch version from pom.xml + +**Tool**: Bash + +**Action**: +1. Locate `pom.xml` in the target project (recursive search): + ```bash + find -name "pom.xml" | head -5 + ``` +2. Extract the current Nablarch version from `` or `` elements: + ```bash + grep -E "(nablarch|5u|6u)" /pom.xml | grep -E "version|artifactId" | head -20 + ``` +3. If the version cannot be detected automatically, prompt the user to specify `--from-version` explicitly. + +**Output**: Confirmed Nablarch version range (from/to versions to use in Step 1) + +**Branch**: +- If `pom.xml` is not found → inform user, ask for project root path, HALT workflow +- If version is not detected → ask user to specify the source version manually, then continue + +--- + +### Step 1: Run rule-based filter (upgrade-checker.sh) + +**Tool**: Bash + +**Action**: +1. Determine the release notes JSON directory based on the major version: + - v5: `~/nabledge-dev/tools/knowledge-creator/.cache/v5/knowledge/releases/releases/` + - v6: `~/nabledge-dev/tools/knowledge-creator/.cache/v6/knowledge/releases/releases/` +2. Run the upgrade checker script: + ```bash + bash .claude/skills/nabledge-1.4/scripts/upgrade-checker.sh \ + --release-notes-dir \ + --project-dir \ + --from \ + --to + ``` +3. Capture stdout (JSON format with hit/no-hit/undecidable classification per item) + +**Output**: JSON result listing affected items, no-hit items, and undecidable items + +**Branch**: +- If script exits with non-zero status → display the error message, ask user to verify paths, HALT workflow +- If 0 affected items and 0 undecidable items → inform user that no impact was found, skip to Step 3 + +**Error handling**: +- JSON directory not found: show expected path (`~/.cache/v5/knowledge/releases/releases/`), advise user to run the knowledge-creator tool to populate the cache +- grep errors in script stderr: display stderr output and continue (non-fatal) +- Script not found: verify `.claude/skills/nabledge-1.4/scripts/upgrade-checker.sh` exists + +--- + +### Step 2: LLM-assisted evaluation for undecidable items + +**Tool**: In-memory (LLM evaluation) + +**Action**: +1. For each item marked as `"undecidable"` in Step 1 output: + a. Read the corresponding section content from the release note JSON: + ```bash + bash .claude/skills/nabledge-1.4/scripts/read-sections.sh \ + ":" + ``` + b. Evaluate whether the change affects the target project based on the section text + c. Classify the item as one of: + - `affected` — change clearly applies to this project + - `not_affected` — change does not apply to this project + - `requires_manual_review` — cannot determine automatically; human review needed + +2. For items marked `"affected"` by the rule-based filter: + a. Read the corresponding section content + b. Extract the recommended action or workaround from the section text + +**Output**: Enriched result with LLM classification and recommended actions for each item + +**Branch**: +- If all undecidable items are classified as `not_affected` and the rule-based `affected` list is also empty → inform user that no impact was found, skip to Step 3 (report still generated) + +--- + +### Step 3: Generate impact assessment report + +**Tool**: In-memory (LLM generation) + +**Action**: +1. Compose a Markdown report using the following structure: + +```markdown +# Nablarch Upgrade Impact Assessment +## Target: {from} → {to} +## Summary +- Affected items: N +- Requires manual review: M +- No impact confirmed: K + +## Affected Items +| # | Title | Detection | Match Location | Action | +|---|-------|-----------|----------------|--------| +| 1 | ... | Rule R5 (FQCN match) | src/main/java/Foo.java:42 | ... | + +## Items Requiring Manual Review +| # | Title | Reason | +|---|-------|--------| +| 1 | ... | Conditional impact — verify manually | + +## No Impact Confirmed +The following items were evaluated and confirmed to have no impact on this project: +- ... +``` + +2. Present the report to the user + +**Output**: Markdown impact assessment report + +--- + +## Error handling + +| Scenario | Action | +|----------|--------| +| `pom.xml` not found | Ask user to specify the project root path | +| JSON cache directory not found | Show expected path; advise user to run knowledge-creator to populate cache | +| `upgrade-checker.sh` not found | Verify the scripts directory exists; report path to user | +| `grep` errors in script output | Display stderr and continue (non-fatal) | +| Version range produces no JSON files | Confirm the version range is correct; list available JSON files in the cache directory | From 815b7dce62e566f734803928e21b7b6426064ab3 Mon Sep 17 00:00:00 2001 From: kumagoro1202 Date: Fri, 27 Mar 2026 08:59:05 +0900 Subject: [PATCH 2/2] fix: move upgrade-check to nabledge-5/6 skills (correct target) - Move upgrade-check.md and upgrade-checker.sh from nabledge-1.4 to nabledge-5/6 - Release notes JSON is located in knowledge/releases/releases/ within each skill - Revert nabledge-1.4/SKILL.md to original state (no upgrade-check entry) - Update nabledge-5/SKILL.md and nabledge-6/SKILL.md with upgrade-check workflow entry --- .claude/skills/nabledge-1.4/SKILL.md | 4 - .claude/skills/nabledge-5/SKILL.md | 4 + .../scripts/upgrade-checker.sh | 0 .../workflows/upgrade-check.md | 20 +- .claude/skills/nabledge-6/SKILL.md | 4 + .../nabledge-6/scripts/upgrade-checker.sh | 632 ++++++++++++++++++ .../nabledge-6/workflows/upgrade-check.md | 143 ++++ 7 files changed, 793 insertions(+), 14 deletions(-) rename .claude/skills/{nabledge-1.4 => nabledge-5}/scripts/upgrade-checker.sh (100%) rename .claude/skills/{nabledge-1.4 => nabledge-5}/workflows/upgrade-check.md (84%) create mode 100755 .claude/skills/nabledge-6/scripts/upgrade-checker.sh create mode 100644 .claude/skills/nabledge-6/workflows/upgrade-check.md diff --git a/.claude/skills/nabledge-1.4/SKILL.md b/.claude/skills/nabledge-1.4/SKILL.md index 198999dcf..cc849cfe3 100644 --- a/.claude/skills/nabledge-1.4/SKILL.md +++ b/.claude/skills/nabledge-1.4/SKILL.md @@ -41,10 +41,6 @@ nabledge-1.4 code-analysis **"code-analysis" argument** (`nabledge-1.4 code-analysis`): - Execute `workflows/code-analysis.md` to analyze user's code -**"upgrade-check" argument** (`nabledge-1.4 upgrade-check [--from ] [--to ] [--project-dir ]`): -- Execute `workflows/upgrade-check.md` to analyze upgrade impact -- This workflow detects affected items between Nablarch versions using rule-based analysis - ## Critical Constraints **Knowledge files only**: Answer using ONLY information in `knowledge/*.json`. DO NOT use external knowledge or LLM training data. diff --git a/.claude/skills/nabledge-5/SKILL.md b/.claude/skills/nabledge-5/SKILL.md index c0e338d72..c45b05d98 100644 --- a/.claude/skills/nabledge-5/SKILL.md +++ b/.claude/skills/nabledge-5/SKILL.md @@ -41,6 +41,10 @@ nabledge-5 code-analysis **"code-analysis" argument** (`nabledge-5 code-analysis`): - Execute `workflows/code-analysis.md` to analyze user's code +**"upgrade-check" argument** (`nabledge-5 upgrade-check [--from ] [--to ] [--project-dir ]`): +- Execute `workflows/upgrade-check.md` to analyze upgrade impact +- This workflow detects affected items between Nablarch 5 versions using rule-based analysis + ## Critical Constraints **Knowledge files only**: Answer using ONLY information in `knowledge/*.json`. DO NOT use external knowledge or LLM training data. diff --git a/.claude/skills/nabledge-1.4/scripts/upgrade-checker.sh b/.claude/skills/nabledge-5/scripts/upgrade-checker.sh similarity index 100% rename from .claude/skills/nabledge-1.4/scripts/upgrade-checker.sh rename to .claude/skills/nabledge-5/scripts/upgrade-checker.sh diff --git a/.claude/skills/nabledge-1.4/workflows/upgrade-check.md b/.claude/skills/nabledge-5/workflows/upgrade-check.md similarity index 84% rename from .claude/skills/nabledge-1.4/workflows/upgrade-check.md rename to .claude/skills/nabledge-5/workflows/upgrade-check.md index 2fd8efc59..4f2bfdd08 100644 --- a/.claude/skills/nabledge-1.4/workflows/upgrade-check.md +++ b/.claude/skills/nabledge-5/workflows/upgrade-check.md @@ -42,13 +42,13 @@ Markdown impact assessment report presented to the user **Tool**: Bash **Action**: -1. Determine the release notes JSON directory based on the major version: - - v5: `~/nabledge-dev/tools/knowledge-creator/.cache/v5/knowledge/releases/releases/` - - v6: `~/nabledge-dev/tools/knowledge-creator/.cache/v6/knowledge/releases/releases/` +1. Determine the release notes JSON directory: + - Default path (Nablarch 5): `knowledge/releases/releases/` (within nabledge-5 skill directory) + - To resolve the absolute path, use `$SKILL_DIR/knowledge/releases/releases/` where `SKILL_DIR` is the nabledge-5 skill root 2. Run the upgrade checker script: ```bash - bash .claude/skills/nabledge-1.4/scripts/upgrade-checker.sh \ - --release-notes-dir \ + bash .claude/skills/nabledge-5/scripts/upgrade-checker.sh \ + --release-notes-dir .claude/skills/nabledge-5/knowledge/releases/releases \ --project-dir \ --from \ --to @@ -62,9 +62,9 @@ Markdown impact assessment report presented to the user - If 0 affected items and 0 undecidable items → inform user that no impact was found, skip to Step 3 **Error handling**: -- JSON directory not found: show expected path (`~/.cache/v5/knowledge/releases/releases/`), advise user to run the knowledge-creator tool to populate the cache +- JSON directory not found: show expected path (`knowledge/releases/releases/`), advise user to verify the nabledge-5 skill is properly installed - grep errors in script stderr: display stderr output and continue (non-fatal) -- Script not found: verify `.claude/skills/nabledge-1.4/scripts/upgrade-checker.sh` exists +- Script not found: verify `.claude/skills/nabledge-5/scripts/upgrade-checker.sh` exists --- @@ -76,7 +76,7 @@ Markdown impact assessment report presented to the user 1. For each item marked as `"undecidable"` in Step 1 output: a. Read the corresponding section content from the release note JSON: ```bash - bash .claude/skills/nabledge-1.4/scripts/read-sections.sh \ + bash .claude/skills/nabledge-5/scripts/read-sections.sh \ ":" ``` b. Evaluate whether the change affects the target project based on the section text @@ -137,7 +137,7 @@ The following items were evaluated and confirmed to have no impact on this proje | Scenario | Action | |----------|--------| | `pom.xml` not found | Ask user to specify the project root path | -| JSON cache directory not found | Show expected path; advise user to run knowledge-creator to populate cache | +| JSON cache directory not found | Show expected path; advise user to verify nabledge-5 skill installation | | `upgrade-checker.sh` not found | Verify the scripts directory exists; report path to user | | `grep` errors in script output | Display stderr and continue (non-fatal) | -| Version range produces no JSON files | Confirm the version range is correct; list available JSON files in the cache directory | +| Version range produces no JSON files | Confirm the version range is correct; list available JSON files in the releases directory | diff --git a/.claude/skills/nabledge-6/SKILL.md b/.claude/skills/nabledge-6/SKILL.md index 1cdaf6af1..abe0822b7 100644 --- a/.claude/skills/nabledge-6/SKILL.md +++ b/.claude/skills/nabledge-6/SKILL.md @@ -41,6 +41,10 @@ nabledge-6 code-analysis **"code-analysis" argument** (`nabledge-6 code-analysis`): - Execute `workflows/code-analysis.md` to analyze user's code +**"upgrade-check" argument** (`nabledge-6 upgrade-check [--from ] [--to ] [--project-dir ]`): +- Execute `workflows/upgrade-check.md` to analyze upgrade impact +- This workflow detects affected items between Nablarch 6 versions using rule-based analysis + ## Critical Constraints **Knowledge files only**: Answer using ONLY information in `knowledge/*.json`. DO NOT use external knowledge or LLM training data. diff --git a/.claude/skills/nabledge-6/scripts/upgrade-checker.sh b/.claude/skills/nabledge-6/scripts/upgrade-checker.sh new file mode 100755 index 000000000..2c803947f --- /dev/null +++ b/.claude/skills/nabledge-6/scripts/upgrade-checker.sh @@ -0,0 +1,632 @@ +#!/bin/bash +# upgrade-checker.sh - Rule-based Nablarch upgrade impact checker +# +# Scans release note JSON files between two Nablarch versions and checks +# whether the changes affect the target project. +# +# Rules implemented: +# R1: Sections with "システムへの影響あり" in index[].hints +# R2: Sections with "影響がある変更" in index[].title (supplement to R1) +# R3: Warning/important blocks in section body (> **警告**: / > **重要**:) +# R4: Table rows where the "影響" column value is "あり" +# R5: FQCNs extracted from backticks (com.nablarch.*) +# R6: Config keys extracted from backticks (nablarch.*) +# R7: Artifact IDs extracted from table "修正バージョン"/"アーティファクトID" columns +# R8: Project grep for R5/R6/R7 extracted items +# +# Usage: +# upgrade-checker.sh --release-notes-dir DIR --project-dir DIR --from VERSION --to VERSION +# +# Example: +# upgrade-checker.sh \ +# --release-notes-dir ~/nabledge-dev/tools/knowledge-creator/.cache/v5/knowledge/releases/releases \ +# --project-dir ~/my-nablarch-project \ +# --from 5u13 \ +# --to 5u18 + +set -euo pipefail + +SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)" +SKILL_DIR="$(cd "$SCRIPT_DIR/.." && pwd)" + +# --------------------------------------------------------------------------- +# Argument parsing +# --------------------------------------------------------------------------- +RELEASE_NOTES_DIR="" +PROJECT_DIR="" +FROM_VERSION="" +TO_VERSION="" + +usage() { + cat >&2 <&2; usage ;; + esac +done + +# --------------------------------------------------------------------------- +# Validation +# --------------------------------------------------------------------------- +if [[ -z "$RELEASE_NOTES_DIR" || -z "$PROJECT_DIR" || -z "$FROM_VERSION" || -z "$TO_VERSION" ]]; then + echo "Error: All four arguments are required." >&2 + usage +fi + +if [[ ! -d "$RELEASE_NOTES_DIR" ]]; then + echo "Error: --release-notes-dir does not exist: $RELEASE_NOTES_DIR" >&2 + exit 1 +fi + +if [[ ! -d "$PROJECT_DIR" ]]; then + echo "Error: --project-dir does not exist: $PROJECT_DIR" >&2 + exit 1 +fi + +# Validate version format: digits optionally followed by 'u' + digits (e.g. 5, 5u13, 6u3) +if ! echo "$FROM_VERSION" | grep -qE '^[0-9]+(u[0-9]+)?$'; then + echo "Error: Invalid --from version format: $FROM_VERSION (expected e.g. 5u13, 6, 6u3)" >&2 + usage +fi + +if ! echo "$TO_VERSION" | grep -qE '^[0-9]+(u[0-9]+)?$'; then + echo "Error: Invalid --to version format: $TO_VERSION (expected e.g. 5u13, 6, 6u3)" >&2 + usage +fi + +# Check jq availability +if ! command -v jq &>/dev/null; then + echo "Error: jq is required but not found. Install jq or ensure it is on PATH." >&2 + exit 1 +fi + +# --------------------------------------------------------------------------- +# Version utilities (implemented in Python for robust comparison) +# --------------------------------------------------------------------------- + +# Convert version string "5u13" -> sortable integer 5013, "6" -> 6000, "6u3" -> 6003 +version_to_int() { + python3 -c " +import sys, re +v = sys.argv[1] +m = re.match(r'^(\d+)(?:u(\d+))?$', v) +if not m: + sys.exit(1) +major = int(m.group(1)) +minor = int(m.group(2)) if m.group(2) else 0 +print(major * 1000 + minor) +" "$1" +} + +FROM_INT=$(version_to_int "$FROM_VERSION") +TO_INT=$(version_to_int "$TO_VERSION") + +if [[ "$FROM_INT" -ge "$TO_INT" ]]; then + echo "Error: --from ($FROM_VERSION) must be less than --to ($TO_VERSION)" >&2 + exit 1 +fi + +# --------------------------------------------------------------------------- +# Phase 1: Filter JSON files by version range (exclusive from, inclusive to) +# Files naming pattern: releases-nablarch{major}(u{minor})?-releasenote.json +# We include files where: FROM_INT < file_version_int <= TO_INT +# --------------------------------------------------------------------------- + +SELECTED_FILES=() + +while IFS= read -r -d '' filepath; do + filename="$(basename "$filepath")" + file_ver=$(python3 -c " +import re, sys +m = re.match(r'releases-nablarch(\d+)(u(\d+))?-releasenote\.json', sys.argv[1]) +if not m: + sys.exit(0) +major = int(m.group(1)) +minor = int(m.group(3)) if m.group(3) else 0 +print(major * 1000 + minor) +" "$filename" 2>/dev/null || true) + + if [[ -z "$file_ver" ]]; then + continue # filename does not match the pattern + fi + + if [[ "$file_ver" -gt "$FROM_INT" && "$file_ver" -le "$TO_INT" ]]; then + SELECTED_FILES+=("$filepath") + fi +done < <(find "$RELEASE_NOTES_DIR" -maxdepth 1 -name "releases-nablarch*-releasenote.json" -print0 | sort -z) + +if [[ ${#SELECTED_FILES[@]} -eq 0 ]]; then + echo "# Upgrade Check Results: Nablarch ${FROM_VERSION} → ${TO_VERSION}" + echo "" + echo "No release note JSON files found in the specified version range." + echo "Directory: $RELEASE_NOTES_DIR" + exit 0 +fi + +# --------------------------------------------------------------------------- +# Phase 2 & 3: Apply R1-R8 rules using Python for robust text processing +# --------------------------------------------------------------------------- + +python3 - "$RELEASE_NOTES_DIR" "$PROJECT_DIR" "$FROM_VERSION" "$TO_VERSION" "${SELECTED_FILES[@]}" <<'PYEOF' +import sys +import os +import re +import json +import subprocess + +_, release_notes_dir, project_dir, from_ver, to_ver, *selected_files = sys.argv + +# Sort files by version +def file_to_ver(path): + m = re.search(r'nablarch(\d+)(u(\d+))?-releasenote', os.path.basename(path)) + if not m: + return 0 + return int(m.group(1)) * 1000 + (int(m.group(3)) if m.group(3) else 0) + +selected_files = sorted(selected_files, key=file_to_ver) + +# --------------------------------------------------------------------------- +# Rule implementations +# --------------------------------------------------------------------------- + +def extract_file_version(path): + """Extract version string from filename, e.g. '5u13' or '6u3'.""" + m = re.search(r'nablarch(\d+)(u(\d+))?-releasenote', os.path.basename(path)) + if not m: + return "unknown" + major = m.group(1) + minor = m.group(3) + return f"{major}u{minor}" if minor else major + +def r1_hints_match(index_entries): + """R1: Find section IDs where hints contain 'システムへの影響あり'.""" + matched = [] + for entry in index_entries: + hints = entry.get("hints", []) + if any("システムへの影響あり" in h for h in hints): + matched.append(entry["id"]) + return matched + +def r2_title_match(index_entries): + """R2: Find section IDs where title contains '影響がある変更'.""" + matched = [] + for entry in index_entries: + title = entry.get("title", "") + if "影響がある変更" in title: + matched.append(entry["id"]) + return matched + +def r3_warning_blocks(section_text): + """R3: Extract warning/important blocks from section body. + Pattern: lines starting with '> **警告**:' or '> **重要**:' etc. + Returns list of full matched lines. + """ + pattern = re.compile(r'^> \*\*(警告|重要|Warning|Important)\*\*:.*', re.MULTILINE) + return [m.group(0) for m in pattern.finditer(section_text)] + +def r4_affected_table_rows(section_text): + """R4: Extract rows from Markdown tables where the 影響 column is 'あり'. + Column names: 'システムへの影響', '影響', 'システム影響' + Values matching: 'あり' (including 'あり（開発）', 'あり（本番）') + Returns list of matched row strings. + """ + results = [] + lines = section_text.split("\n") + i = 0 + while i < len(lines): + line = lines[i].strip() + # Look for a table header row containing an 影響 column + if "|" in line: + # Find header + header_match = re.match(r'\|(.+)\|', line) + if header_match: + headers = [h.strip() for h in line.split("|")[1:-1]] + # Check if any header matches 影響 column pattern + impact_col_idx = None + for idx, h in enumerate(headers): + # Strip markdown bold/other formatting + h_clean = re.sub(r'\*+', '', h).strip() + if re.search(r'システムへの影響|^影響$|システム影響', h_clean): + impact_col_idx = idx + break + + if impact_col_idx is not None: + # Skip separator row (---|---|...) + i += 1 + if i < len(lines) and re.match(r'\|[\s\-:|]+\|', lines[i]): + i += 1 + # Process data rows + while i < len(lines): + row = lines[i].strip() + if not row.startswith("|"): + break + cells = [c.strip() for c in row.split("|")[1:-1]] + if impact_col_idx < len(cells): + cell = cells[impact_col_idx] + # Remove markdown formatting + cell_clean = re.sub(r'\*+', '', cell).strip() + # Match 'あり' but not 'なし' + if re.match(r'あり', cell_clean): + results.append(row) + i += 1 + continue + i += 1 + return results + +def r5_extract_fqcns(section_text): + """R5: Extract FQCNs from backticks in section body. + Pattern: `com.nablarch.[A-Za-z][A-Za-z0-9._$]*` + """ + pattern = re.compile(r'`(com\.nablarch\.[A-Za-z][A-Za-z0-9._$]*)`') + return list(dict.fromkeys(pattern.findall(section_text))) + +def r6_extract_config_keys(section_text): + """R6: Extract config keys from backticks in section body. + Pattern: `nablarch.[a-z][a-zA-Z0-9._-]*` + """ + pattern = re.compile(r'`(nablarch\.[a-z][a-zA-Z0-9._-]*)`') + return list(dict.fromkeys(pattern.findall(section_text))) + +def r7_extract_artifact_ids(section_text): + """R7: Extract Nablarch artifact IDs from table cells. + Pattern: nablarch-[a-z-]+ + """ + pattern = re.compile(r'\bnablarch-[a-z][a-z-]+\b') + matches = pattern.findall(section_text) + return list(dict.fromkeys(matches)) + +def r8_project_grep(items, project_dir, file_extensions=None): + """R8: Grep the project directory for each item. + Returns dict: {item: [matching_file_paths]} or {item: None} on error. + """ + if file_extensions is None: + file_extensions = [".java", ".xml", ".properties", ".yaml", ".yml"] + + results = {} + for item in items: + try: + cmd = ["grep", "-rl", "--include=" + ",".join(f"*{ext}" for ext in file_extensions), + item, project_dir] + # Use find + xargs approach for better glob handling + find_cmd = ["grep", "-rl", item, project_dir, + "--include=*.java", "--include=*.xml", + "--include=*.properties", "--include=*.yaml", + "--include=*.yml"] + proc = subprocess.run(find_cmd, capture_output=True, text=True, timeout=30) + if proc.returncode == 0: + hits = [h.strip() for h in proc.stdout.strip().split("\n") if h.strip()] + results[item] = hits + elif proc.returncode == 1: + results[item] = [] # no match (not an error) + else: + results[item] = None # grep error + except subprocess.TimeoutExpired: + results[item] = None + except Exception: + results[item] = None + return results + +def r8_pom_grep(artifact_ids, project_dir): + """R8: Grep pom.xml files for artifact IDs.""" + results = {} + for aid in artifact_ids: + try: + find_cmd = ["grep", "-rl", aid, project_dir, "--include=*.xml"] + proc = subprocess.run(find_cmd, capture_output=True, text=True, timeout=30) + if proc.returncode == 0: + hits = [h.strip() for h in proc.stdout.strip().split("\n") if h.strip()] + results[aid] = hits + elif proc.returncode == 1: + results[aid] = [] + else: + results[aid] = None + except subprocess.TimeoutExpired: + results[aid] = None + except Exception: + results[aid] = None + return results + +# --------------------------------------------------------------------------- +# Main processing +# --------------------------------------------------------------------------- + +affected_items = [] # {file_ver, file, section_id, rule, pattern, project_hit} +undecidable_items = [] # {file_ver, file, section_id, reason} +not_affected_count = 0 + +all_fqcns = [] +all_config_keys = [] +all_artifact_ids = [] + +# Store per-file, per-section data for final output +file_section_data = {} # (file_ver, section_id) -> {fqcns, config_keys, artifact_ids, ...} + +for filepath in selected_files: + file_ver = extract_file_version(filepath) + try: + with open(filepath, "r", encoding="utf-8") as f: + data = json.load(f) + except (json.JSONDecodeError, IOError) as e: + undecidable_items.append({ + "file_ver": file_ver, + "file": os.path.basename(filepath), + "section_id": "*", + "reason": f"JSON parse error: {e}" + }) + continue + + index_entries = data.get("index", []) + sections = data.get("sections", {}) + + # Apply R1 and R2 to find affected section IDs + r1_sections = set(r1_hints_match(index_entries)) + r2_sections = set(r2_title_match(index_entries)) + candidate_sections = r1_sections | r2_sections + + # Process each section + for entry in index_entries: + section_id = entry["id"] + section_text = sections.get(section_id, "") + + section_key = (file_ver, section_id) + is_affected_section = section_id in candidate_sections + + # Per-section extracted items + fqcns = r5_extract_fqcns(section_text) + config_keys = r6_extract_config_keys(section_text) + artifact_ids = r7_extract_artifact_ids(section_text) + + all_fqcns.extend(fqcns) + all_config_keys.extend(config_keys) + all_artifact_ids.extend(artifact_ids) + + file_section_data[section_key] = { + "file": os.path.basename(filepath), + "file_ver": file_ver, + "section_id": section_id, + "section_title": entry.get("title", ""), + "fqcns": fqcns, + "config_keys": config_keys, + "artifact_ids": artifact_ids, + } + + if is_affected_section: + # R1 match + if section_id in r1_sections: + affected_items.append({ + "file_ver": file_ver, + "file": os.path.basename(filepath), + "section_id": section_id, + "section_title": entry.get("title", ""), + "rule": "R1 (hints: システムへの影響あり)", + "pattern": "hints contains 'システムへの影響あり'", + "project_hit": "", + "fqcns": fqcns, + "config_keys": config_keys, + "artifact_ids": artifact_ids, + }) + elif section_id in r2_sections: + affected_items.append({ + "file_ver": file_ver, + "file": os.path.basename(filepath), + "section_id": section_id, + "section_title": entry.get("title", ""), + "rule": "R2 (title: 影響がある変更)", + "pattern": "title contains '影響がある変更'", + "project_hit": "", + "fqcns": fqcns, + "config_keys": config_keys, + "artifact_ids": artifact_ids, + }) + else: + # For non-candidate sections, still check R3 and R4 + r3_matches = r3_warning_blocks(section_text) + r4_matches = r4_affected_table_rows(section_text) + + if r3_matches: + affected_items.append({ + "file_ver": file_ver, + "file": os.path.basename(filepath), + "section_id": section_id, + "section_title": entry.get("title", ""), + "rule": "R3 (警告/重要ブロック)", + "pattern": r3_matches[0][:80] if r3_matches else "", + "project_hit": "", + "fqcns": fqcns, + "config_keys": config_keys, + "artifact_ids": artifact_ids, + }) + elif r4_matches: + affected_items.append({ + "file_ver": file_ver, + "file": os.path.basename(filepath), + "section_id": section_id, + "section_title": entry.get("title", ""), + "rule": "R4 (テーブル影響列: あり)", + "pattern": r4_matches[0][:80] if r4_matches else "", + "project_hit": "", + "fqcns": fqcns, + "config_keys": config_keys, + "artifact_ids": artifact_ids, + }) + else: + not_affected_count += 1 + + # Also apply R3/R4 to candidate sections (for completeness) + for section_id in candidate_sections: + section_text = sections.get(section_id, "") + # These are already marked as affected; R3/R4 add detail but don't create new entries + +# Deduplicate all extracted items +all_fqcns = list(dict.fromkeys(all_fqcns)) +all_config_keys = list(dict.fromkeys(all_config_keys)) +all_artifact_ids = list(dict.fromkeys(all_artifact_ids)) + +# --------------------------------------------------------------------------- +# Phase 3: R8 - Project grep +# --------------------------------------------------------------------------- +fqcn_hits = r8_project_grep(all_fqcns, project_dir) if all_fqcns else {} +config_hits = r8_project_grep(all_config_keys, project_dir) if all_config_keys else {} +artifact_hits = r8_pom_grep(all_artifact_ids, project_dir) if all_artifact_ids else {} + +def format_project_hits(item, hits_dict): + """Format project grep results as short paths.""" + hits = hits_dict.get(item) + if hits is None: + return "⚠ grep error" + if not hits: + return "(no match)" + # Shorten paths relative to project_dir + short = [] + for h in hits[:3]: # limit to 3 for readability + try: + rel = os.path.relpath(h, project_dir) + except ValueError: + rel = h + short.append(rel) + suffix = f" (+{len(hits)-3} more)" if len(hits) > 3 else "" + return ", ".join(short) + suffix + +# Enrich affected_items with R8 project hit info +for item in affected_items: + hit_parts = [] + + # Check FQCNs + for fqcn in item.get("fqcns", []): + hits = fqcn_hits.get(fqcn) + if hits: + hit_parts.append(f"{fqcn}: {format_project_hits(fqcn, fqcn_hits)}") + + # Check config keys + for key in item.get("config_keys", []): + hits = config_hits.get(key) + if hits: + hit_parts.append(f"{key}: {format_project_hits(key, config_hits)}") + + # Check artifact IDs + for aid in item.get("artifact_ids", []): + hits = artifact_hits.get(aid) + if hits: + hit_parts.append(f"{aid}: {format_project_hits(aid, artifact_hits)}") + + if hit_parts: + item["project_hit"] = "; ".join(hit_parts[:2]) # limit for table width + if len(hit_parts) > 2: + item["project_hit"] += f" (+{len(hit_parts)-2} more)" + else: + item["project_hit"] = "(no project match)" + +# --------------------------------------------------------------------------- +# Phase 4: Output Markdown report +# --------------------------------------------------------------------------- +total_sections = len(affected_items) + len(undecidable_items) + not_affected_count + +print(f"# Upgrade Check Results: Nablarch {from_ver} → {to_ver}") +print() +print(f"**Release notes analyzed**: {len(selected_files)} file(s)") +print() +print("## Summary") +print() +print(f"- Affected items (rule-based): {len(affected_items)}") +print(f"- Undecidable items: {len(undecidable_items)}") +print(f"- Not affected: {not_affected_count}") +print(f"- Total sections analyzed: {total_sections}") +print() + +# --- Affected Items --- +print("## Affected Items") +print() +if affected_items: + print("| # | Version | Section | Detection Rule | Matched Pattern | Project Hit |") + print("|---|---------|---------|----------------|-----------------|-------------|") + for i, item in enumerate(affected_items, 1): + section_label = f"{item['section_id']} ({item['section_title'][:30]})" if item.get('section_title') else item['section_id'] + pattern_short = item['pattern'][:60] if item['pattern'] else "" + print(f"| {i} | {item['file_ver']} | {section_label} | {item['rule']} | {pattern_short} | {item['project_hit']} |") +else: + print("(No affected items detected by rule-based analysis)") +print() + +# --- R5/R6/R7 Extracted Items --- +if all_fqcns or all_config_keys or all_artifact_ids: + print("## Extracted Technical Items") + print() + + if all_fqcns: + print("### FQCNs (R5)") + print() + print("| FQCN | Project Hit |") + print("|------|-------------|") + for fqcn in all_fqcns: + print(f"| `{fqcn}` | {format_project_hits(fqcn, fqcn_hits)} |") + print() + + if all_config_keys: + print("### Config Keys (R6)") + print() + print("| Config Key | Project Hit |") + print("|------------|-------------|") + for key in all_config_keys: + print(f"| `{key}` | {format_project_hits(key, config_hits)} |") + print() + + if all_artifact_ids: + print("### Artifact IDs (R7 → pom.xml grep)") + print() + print("| Artifact ID | Project Hit |") + print("|-------------|-------------|") + for aid in all_artifact_ids: + print(f"| `{aid}` | {format_project_hits(aid, artifact_hits)} |") + print() + +# --- Undecidable Items --- +print("## Undecidable Items (requires LLM evaluation)") +print() +if undecidable_items: + print("| # | Version | Section | Reason |") + print("|---|---------|---------|--------|") + for i, item in enumerate(undecidable_items, 1): + print(f"| {i} | {item['file_ver']} | {item['section_id']} | {item['reason']} |") +else: + print("(No undecidable items)") +print() + +# --- Not Affected --- +print("## Not Affected") +print() +if not_affected_count > 0: + print(f"{not_affected_count} section(s) had no rule-based matches and are considered not affected.") +else: + print("(All sections were classified by rule-based analysis)") +print() + +# --- Footer --- +print("---") +print() +print("**Notes**:") +print("- 'Affected Items' are sections where R1-R4 detected impact signals.") +print("- 'Project Hit' shows whether R5/R6/R7 extracted items were found in the project source.") +print("- Sections with no R1-R4 signal and no project hits are marked as 'Not Affected'.") +print("- Items requiring human/LLM review (e.g. implicit impacts, conditional changes) are listed under 'Undecidable'.") + +PYEOF diff --git a/.claude/skills/nabledge-6/workflows/upgrade-check.md b/.claude/skills/nabledge-6/workflows/upgrade-check.md new file mode 100644 index 000000000..6a771cf19 --- /dev/null +++ b/.claude/skills/nabledge-6/workflows/upgrade-check.md @@ -0,0 +1,143 @@ +# Upgrade Check Workflow + +Analyze the impact of upgrading Nablarch from one version to another. Runs a rule-based script to detect affected items and uses LLM evaluation for undecidable cases, then generates a structured Markdown report. + +## Input + +- Target project root path +- Upgrade source version (e.g., `6`) +- Upgrade target version (e.g., `6u3`) + +## Output + +Markdown impact assessment report presented to the user + +## Steps + +### Step 0: Detect Nablarch version from pom.xml + +**Tool**: Bash + +**Action**: +1. Locate `pom.xml` in the target project (recursive search): + ```bash + find -name "pom.xml" | head -5 + ``` +2. Extract the current Nablarch version from `` or `` elements: + ```bash + grep -E "(nablarch|5u|6u)" /pom.xml | grep -E "version|artifactId" | head -20 + ``` +3. If the version cannot be detected automatically, prompt the user to specify `--from-version` explicitly. + +**Output**: Confirmed Nablarch version range (from/to versions to use in Step 1) + +**Branch**: +- If `pom.xml` is not found → inform user, ask for project root path, HALT workflow +- If version is not detected → ask user to specify the source version manually, then continue + +--- + +### Step 1: Run rule-based filter (upgrade-checker.sh) + +**Tool**: Bash + +**Action**: +1. Determine the release notes JSON directory: + - Default path (Nablarch 6): `knowledge/releases/releases/` (within nabledge-6 skill directory) + - To resolve the absolute path, use `$SKILL_DIR/knowledge/releases/releases/` where `SKILL_DIR` is the nabledge-6 skill root +2. Run the upgrade checker script: + ```bash + bash .claude/skills/nabledge-6/scripts/upgrade-checker.sh \ + --release-notes-dir .claude/skills/nabledge-6/knowledge/releases/releases \ + --project-dir \ + --from \ + --to + ``` +3. Capture stdout (JSON format with hit/no-hit/undecidable classification per item) + +**Output**: JSON result listing affected items, no-hit items, and undecidable items + +**Branch**: +- If script exits with non-zero status → display the error message, ask user to verify paths, HALT workflow +- If 0 affected items and 0 undecidable items → inform user that no impact was found, skip to Step 3 + +**Error handling**: +- JSON directory not found: show expected path (`knowledge/releases/releases/`), advise user to verify the nabledge-6 skill is properly installed +- grep errors in script stderr: display stderr output and continue (non-fatal) +- Script not found: verify `.claude/skills/nabledge-6/scripts/upgrade-checker.sh` exists + +--- + +### Step 2: LLM-assisted evaluation for undecidable items + +**Tool**: In-memory (LLM evaluation) + +**Action**: +1. For each item marked as `"undecidable"` in Step 1 output: + a. Read the corresponding section content from the release note JSON: + ```bash + bash .claude/skills/nabledge-6/scripts/read-sections.sh \ + ":" + ``` + b. Evaluate whether the change affects the target project based on the section text + c. Classify the item as one of: + - `affected` — change clearly applies to this project + - `not_affected` — change does not apply to this project + - `requires_manual_review` — cannot determine automatically; human review needed + +2. For items marked `"affected"` by the rule-based filter: + a. Read the corresponding section content + b. Extract the recommended action or workaround from the section text + +**Output**: Enriched result with LLM classification and recommended actions for each item + +**Branch**: +- If all undecidable items are classified as `not_affected` and the rule-based `affected` list is also empty → inform user that no impact was found, skip to Step 3 (report still generated) + +--- + +### Step 3: Generate impact assessment report + +**Tool**: In-memory (LLM generation) + +**Action**: +1. Compose a Markdown report using the following structure: + +```markdown +# Nablarch Upgrade Impact Assessment +## Target: {from} → {to} +## Summary +- Affected items: N +- Requires manual review: M +- No impact confirmed: K + +## Affected Items +| # | Title | Detection | Match Location | Action | +|---|-------|-----------|----------------|--------| +| 1 | ... | Rule R5 (FQCN match) | src/main/java/Foo.java:42 | ... | + +## Items Requiring Manual Review +| # | Title | Reason | +|---|-------|--------| +| 1 | ... | Conditional impact — verify manually | + +## No Impact Confirmed +The following items were evaluated and confirmed to have no impact on this project: +- ... +``` + +2. Present the report to the user + +**Output**: Markdown impact assessment report + +--- + +## Error handling + +| Scenario | Action | +|----------|--------| +| `pom.xml` not found | Ask user to specify the project root path | +| JSON cache directory not found | Show expected path; advise user to verify nabledge-6 skill installation | +| `upgrade-checker.sh` not found | Verify the scripts directory exists; report path to user | +| `grep` errors in script output | Display stderr and continue (non-fatal) | +| Version range produces no JSON files | Confirm the version range is correct; list available JSON files in the releases directory |