Skip to content

Commit a89d4d9

Browse files
nishimotzclaude
nishimotz
and
claude
committed
Update duplicate detection to check all three HTML folders
🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <noreply@anthropic.com>
1 parent 91dc3f0 commit a89d4d9

File tree

1 file changed

+64
-30
lines changed

1 file changed

+64
-30
lines changed

‎.github/workflows/check-duplicates.yml‎

Lines changed: 64 additions & 30 deletions
Original file line numberDiff line numberDiff line change
@@ -31,45 +31,79 @@ jobs:
3131
- name: Check for duplicate filenames
3232
id: check
3333
run: |
34-
echo "Checking for duplicate filenames between HTML_見送り and HTML_未着手 folders..."
34+
echo "Checking for duplicate filenames across HTML, HTML_見送り, and HTML_未着手 folders..."
3535
36-
# Get basenames from both directories
37-
dir1_files=$(find "WAIC-TEST/HTML_見送り" -name "*.md" -exec basename {} \; 2>/dev/null | sort)
38-
dir2_files=$(find "WAIC-TEST/HTML_未着手" -name "*.md" -exec basename {} \; 2>/dev/null | sort)
36+
# Get basenames from all three directories
37+
html_files=$(find "WAIC-TEST/HTML" -name "*.md" ! -name "README.md" -exec basename {} \; 2>/dev/null | sort)
38+
miokuri_files=$(find "WAIC-TEST/HTML_見送り" -name "*.md" -exec basename {} \; 2>/dev/null | sort)
39+
michakushu_files=$(find "WAIC-TEST/HTML_未着手" -name "*.md" -exec basename {} \; 2>/dev/null | sort)
3940
40-
# Find duplicates
41-
duplicates=$(comm -12 <(echo "$dir1_files") <(echo "$dir2_files"))
41+
# Find duplicates and files to delete
42+
issue_body=""
43+
files_to_delete=""
44+
duplicates_found=false
4245
43-
if [ -n "$duplicates" ]; then
44-
echo "❌ Duplicate filenames found:"
45-
echo "$duplicates"
46+
# Check HTML vs HTML_未着手 (delete from HTML_未着手)
47+
html_vs_michakushu=$(comm -12 <(echo "$html_files") <(echo "$michakushu_files"))
48+
if [ -n "$html_vs_michakushu" ]; then
49+
echo "❌ Duplicates found between HTML and HTML_未着手:"
50+
duplicates_found=true
51+
for file in $html_vs_michakushu; do
52+
echo " - $file"
53+
html_path=$(find WAIC-TEST/HTML -name "$file" -exec echo {} \;)
54+
michakushu_path=$(find WAIC-TEST/HTML_未着手 -name "$file" -exec echo {} \;)
55+
echo " 📁 HTML: $html_path"
56+
echo " 📁 HTML_未着手: $michakushu_path (削除対象)"
57+
issue_body="$issue_body- \`$file\`\n - 📁 HTML: \`$html_path\`\n - 📁 HTML_未着手: \`$michakushu_path\` (削除対象)\n\n"
58+
files_to_delete="$files_to_delete$michakushu_path\n"
59+
done
4660
echo ""
47-
echo "Files exist in both HTML_見送り and HTML_未着手:"
48-
49-
# Create detailed list for issue and files to delete
50-
issue_body=""
51-
files_to_delete=""
52-
for file in $duplicates; do
61+
fi
62+
63+
# Check HTML vs HTML_見送り (delete from HTML_見送り)
64+
html_vs_miokuri=$(comm -12 <(echo "$html_files") <(echo "$miokuri_files"))
65+
if [ -n "$html_vs_miokuri" ]; then
66+
echo "❌ Duplicates found between HTML and HTML_見送り:"
67+
duplicates_found=true
68+
for file in $html_vs_miokuri; do
69+
echo " - $file"
70+
html_path=$(find WAIC-TEST/HTML -name "$file" -exec echo {} \;)
71+
miokuri_path=$(find WAIC-TEST/HTML_見送り -name "$file" -exec echo {} \;)
72+
echo " 📁 HTML: $html_path"
73+
echo " 📁 HTML_見送り: $miokuri_path (削除対象)"
74+
issue_body="$issue_body- \`$file\`\n - 📁 HTML: \`$html_path\`\n - 📁 HTML_見送り: \`$miokuri_path\` (削除対象)\n\n"
75+
files_to_delete="$files_to_delete$miokuri_path\n"
76+
done
77+
echo ""
78+
fi
79+
80+
# Check HTML_見送り vs HTML_未着手 (delete from HTML_未着手)
81+
miokuri_vs_michakushu=$(comm -12 <(echo "$miokuri_files") <(echo "$michakushu_files"))
82+
if [ -n "$miokuri_vs_michakushu" ]; then
83+
echo "❌ Duplicates found between HTML_見送り and HTML_未着手:"
84+
duplicates_found=true
85+
for file in $miokuri_vs_michakushu; do
5386
echo " - $file"
5487
miokuri_path=$(find WAIC-TEST/HTML_見送り -name "$file" -exec echo {} \;)
5588
michakushu_path=$(find WAIC-TEST/HTML_未着手 -name "$file" -exec echo {} \;)
56-
echo " 📁 HTML_見送り/$miokuri_path"
57-
echo " 📁 HTML_未着手/$michakushu_path"
58-
issue_body="$issue_body- \`$file\`\n - 📁 \`$miokuri_path\`\n - 📁 \`$michakushu_path\`\n\n"
89+
echo " 📁 HTML_見送り: $miokuri_path"
90+
echo " 📁 HTML_未着手: $michakushu_path (削除対象)"
91+
issue_body="$issue_body- \`$file\`\n - 📁 HTML_見送り: \`$miokuri_path\`\n - 📁 HTML_未着手: \`$michakushu_path\` (削除対象)\n\n"
5992
files_to_delete="$files_to_delete$michakushu_path\n"
6093
done
61-
94+
echo ""
95+
fi
96+
97+
if [ "$duplicates_found" = true ]; then
6298
echo "duplicates_found=true" >> $GITHUB_OUTPUT
6399
echo "duplicate_list<<EOF" >> $GITHUB_OUTPUT
64100
echo -e "$issue_body" >> $GITHUB_OUTPUT
65101
echo "EOF" >> $GITHUB_OUTPUT
66102
echo "files_to_delete<<EOF" >> $GITHUB_OUTPUT
67103
echo -e "$files_to_delete" >> $GITHUB_OUTPUT
68104
echo "EOF" >> $GITHUB_OUTPUT
69-
70-
# Don't fail, proceed to auto-fix
71105
else
72-
echo "✅ No duplicate filenames found between the two directories."
106+
echo "✅ No duplicate filenames found across all directories."
73107
echo "duplicates_found=false" >> $GITHUB_OUTPUT
74108
fi
75109
@@ -83,9 +117,9 @@ jobs:
83117
with:
84118
token: ${{ secrets.GITHUB_TOKEN }}
85119

86-
- name: Delete duplicate files from HTML_未着手
120+
- name: Delete duplicate files
87121
run: |
88-
echo "Deleting duplicate files from HTML_未着手 folder..."
122+
echo "Deleting duplicate files..."
89123
90124
# Delete files listed in files_to_delete
91125
echo -e "${{ needs.check-duplicates.outputs.files_to_delete }}" | while read -r file; do
@@ -99,18 +133,18 @@ jobs:
99133
uses: peter-evans/create-pull-request@v7
100134
with:
101135
token: ${{ secrets.GITHUB_TOKEN }}
102-
commit-message: "🗑️ Remove duplicate files from HTML_未着手 folder [skip ci]"
103-
title: "🗑️ 重複ファイルの自動削除 (HTML_未着手)"
136+
commit-message: "🗑️ Remove duplicate files across HTML folders [skip ci]"
137+
title: "🗑️ 重複ファイルの自動削除"
104138
body: |
105139
## 概要
106-
HTML_見送り と HTML_未着手 フォルダ間で重複するファイルを検出し、HTML_未着手 から自動削除しました。
140+
HTML、HTML_見送り、HTML_未着手 フォルダ間で重複するファイルを検出し、優先度の低いフォルダから自動削除しました。
107141
108142
## 削除されたファイル
109143
${{ needs.check-duplicates.outputs.duplicate_list }}
110144
111-
## 削除理由
112-
- ファイルが HTML_見送り フォルダに既に存在するため
113-
- 重複を避けるために HTML_未着手 から削除
145+
## 削除ルール
146+
- 優先度: HTML > HTML_見送り > HTML_未着手
147+
- 重複が見つかった場合、優先度の低い方から削除
114148
115149
## 確認事項
116150
- [ ] 削除されたファイルの内容が HTML_見送り のファイルと同一であることを確認

0 commit comments

Comments
 (0)