|
| 1 | +name: Exportera SFS-data till Git Repository |
| 2 | + |
| 3 | +on: |
| 4 | + schedule: |
| 5 | + # Kör varje söndag kl 03:00 UTC (veckovis export) |
| 6 | + - cron: '0 3 * * 0' |
| 7 | + workflow_dispatch: # Tillåter manuell körning |
| 8 | + inputs: |
| 9 | + export_mode: |
| 10 | + description: 'Export-läge: "all" = alla filer, "changed" = bara nya/ändrade sedan senaste veckan' |
| 11 | + required: false |
| 12 | + default: 'changed' |
| 13 | + type: choice |
| 14 | + options: |
| 15 | + - changed |
| 16 | + - all |
| 17 | + years: |
| 18 | + description: 'År-intervall att exportera (t.ex. "2024-2026" eller "2024"). Används bara i "all"-läge.' |
| 19 | + required: false |
| 20 | + type: string |
| 21 | + branch_name: |
| 22 | + description: 'Git branch namn (lämna tom för auto-genererad med datum)' |
| 23 | + required: false |
| 24 | + type: string |
| 25 | + batch_size: |
| 26 | + description: 'Antal filer per batch' |
| 27 | + required: false |
| 28 | + default: '100' |
| 29 | + type: string |
| 30 | + skip_initial: |
| 31 | + description: 'Hoppa över initial commits (bara temporal)' |
| 32 | + required: false |
| 33 | + default: false |
| 34 | + type: boolean |
| 35 | + skip_temporal: |
| 36 | + description: 'Hoppa över temporal commits (bara initial)' |
| 37 | + required: false |
| 38 | + default: false |
| 39 | + type: boolean |
| 40 | + |
| 41 | +permissions: |
| 42 | + contents: read |
| 43 | + |
| 44 | +jobs: |
| 45 | + export-to-git: |
| 46 | + runs-on: ubuntu-latest |
| 47 | + |
| 48 | + steps: |
| 49 | + - name: Checkout workflow-artifact-data branch |
| 50 | + uses: actions/checkout@v4 |
| 51 | + with: |
| 52 | + ref: workflow-artifact-data |
| 53 | + fetch-depth: 0 # Hämta full historik för git operationer |
| 54 | + token: ${{ secrets.GITHUB_TOKEN }} |
| 55 | + |
| 56 | + - name: Set up Python |
| 57 | + uses: actions/setup-python@v4 |
| 58 | + with: |
| 59 | + python-version: '3.11' |
| 60 | + |
| 61 | + - name: Install dependencies |
| 62 | + run: | |
| 63 | + python -m pip install --upgrade pip |
| 64 | + pip install -r requirements.txt |
| 65 | +
|
| 66 | + - name: Verify data directories exist |
| 67 | + run: | |
| 68 | + if [ ! -d "data/sfs_json" ]; then |
| 69 | + echo "❌ Fel: data/sfs_json katalog finns inte" |
| 70 | + exit 1 |
| 71 | + fi |
| 72 | + if [ ! -d "data/md-markers" ]; then |
| 73 | + echo "❌ Fel: data/md-markers katalog finns inte" |
| 74 | + exit 1 |
| 75 | + fi |
| 76 | +
|
| 77 | + JSON_COUNT=$(find data/sfs_json -name "*.json" | wc -l) |
| 78 | + MD_COUNT=$(find data/md-markers -name "*.md" | wc -l) |
| 79 | +
|
| 80 | + echo "✅ Hittade $JSON_COUNT JSON-filer" |
| 81 | + echo "✅ Hittade $MD_COUNT Markdown-filer" |
| 82 | +
|
| 83 | + if [ "$JSON_COUNT" -eq 0 ]; then |
| 84 | + echo "❌ Inga JSON-filer att exportera" |
| 85 | + exit 1 |
| 86 | + fi |
| 87 | +
|
| 88 | + - name: Determine export filter |
| 89 | + id: export_filter |
| 90 | + run: | |
| 91 | + EXPORT_MODE="${{ inputs.export_mode || 'changed' }}" |
| 92 | + echo "📋 Export-läge: $EXPORT_MODE" |
| 93 | +
|
| 94 | + if [ "$EXPORT_MODE" = "all" ]; then |
| 95 | + # Exportera alla filer (eller filtrera på år om specificerat) |
| 96 | + if [ -n "${{ inputs.years }}" ]; then |
| 97 | + echo "📅 Exporterar alla filer för år: ${{ inputs.years }}" |
| 98 | + echo "filter=${{ inputs.years }}" >> $GITHUB_OUTPUT |
| 99 | + echo "filter_type=years" >> $GITHUB_OUTPUT |
| 100 | + else |
| 101 | + echo "📚 Exporterar ALLA filer (ingen filtrering)" |
| 102 | + echo "filter=" >> $GITHUB_OUTPUT |
| 103 | + echo "filter_type=none" >> $GITHUB_OUTPUT |
| 104 | + fi |
| 105 | + else |
| 106 | + # Exportera bara nya/ändrade filer sedan senaste veckan |
| 107 | + SINCE_DATE=$(date -u -d '7 days ago' +'%Y-%m-%d') |
| 108 | + echo "🔍 Letar efter filer ändrade sedan: $SINCE_DATE" |
| 109 | +
|
| 110 | + # Hämta lista på ändrade JSON-filer från git log |
| 111 | + CHANGED_FILES=$(git log --since="$SINCE_DATE" --name-only --pretty=format: --diff-filter=AM -- data/sfs_json/ | grep -E '\.json$' | sort -u || true) |
| 112 | +
|
| 113 | + if [ -z "$CHANGED_FILES" ]; then |
| 114 | + echo "⚠️ Inga nya/ändrade filer hittades sedan $SINCE_DATE" |
| 115 | + echo "changed_count=0" >> $GITHUB_OUTPUT |
| 116 | + echo "filter=" >> $GITHUB_OUTPUT |
| 117 | + echo "filter_type=none" >> $GITHUB_OUTPUT |
| 118 | + exit 0 |
| 119 | + else |
| 120 | + # Räkna antal ändrade filer |
| 121 | + CHANGED_COUNT=$(echo "$CHANGED_FILES" | wc -l | tr -d ' ') |
| 122 | + echo "✅ Hittade $CHANGED_COUNT nya/ändrade filer" |
| 123 | +
|
| 124 | + # Extrahera beteckningar från filnamn (sfs-YYYY-NNN.json -> YYYY:NNN) |
| 125 | + BETECKNINGAR="" |
| 126 | + while IFS= read -r file; do |
| 127 | + if [[ $file =~ sfs-([0-9]{4})-([0-9]+)\.json ]]; then |
| 128 | + BETECKNING="${BASH_REMATCH[1]}:${BASH_REMATCH[2]}" |
| 129 | + if [ -z "$BETECKNINGAR" ]; then |
| 130 | + BETECKNINGAR="$BETECKNING" |
| 131 | + else |
| 132 | + BETECKNINGAR="$BETECKNINGAR,$BETECKNING" |
| 133 | + fi |
| 134 | + fi |
| 135 | + done <<< "$CHANGED_FILES" |
| 136 | +
|
| 137 | + echo "📋 Filter för ändrade dokument: $BETECKNINGAR" |
| 138 | + echo "changed_count=$CHANGED_COUNT" >> $GITHUB_OUTPUT |
| 139 | + echo "filter=$BETECKNINGAR" >> $GITHUB_OUTPUT |
| 140 | + echo "filter_type=beteckningar" >> $GITHUB_OUTPUT |
| 141 | +
|
| 142 | + # Spara lista för debugging |
| 143 | + echo "" |
| 144 | + echo "Ändrade filer:" |
| 145 | + echo "$CHANGED_FILES" |
| 146 | + fi |
| 147 | + fi |
| 148 | +
|
| 149 | + - name: Determine branch name |
| 150 | + id: branch |
| 151 | + run: | |
| 152 | + if [ -n "${{ inputs.branch_name }}" ]; then |
| 153 | + BRANCH_NAME="${{ inputs.branch_name }}" |
| 154 | + else |
| 155 | + # Auto-generera branch namn med datum |
| 156 | + BRANCH_NAME="export-$(date +'%Y-%m-%d')" |
| 157 | + fi |
| 158 | + echo "branch_name=${BRANCH_NAME}" >> $GITHUB_OUTPUT |
| 159 | + echo "📋 Använder branch: ${BRANCH_NAME}" |
| 160 | +
|
| 161 | + - name: Export to Git repository |
| 162 | + # Skippa om inga ändrade filer i "changed" mode |
| 163 | + if: steps.export_filter.outputs.filter_type != 'none' || steps.export_filter.outputs.changed_count != '0' |
| 164 | + env: |
| 165 | + GIT_GITHUB_PAT: ${{ secrets.GIT_GITHUB_PAT }} |
| 166 | + PYTHONPATH: ${{ github.workspace }} |
| 167 | + run: | |
| 168 | + # Bygg kommando med parametrar |
| 169 | + CMD="python exporters/git/batch_export_to_git.py" |
| 170 | + CMD="$CMD --input data/sfs_json" |
| 171 | + CMD="$CMD --markers-dir data/md-markers" |
| 172 | + CMD="$CMD --branch ${{ steps.branch.outputs.branch_name }}" |
| 173 | + CMD="$CMD --batch-size ${{ inputs.batch_size || '100' }}" |
| 174 | + CMD="$CMD --verbose" |
| 175 | +
|
| 176 | + # Lägg till filter baserat på export-läge |
| 177 | + FILTER="${{ steps.export_filter.outputs.filter }}" |
| 178 | + FILTER_TYPE="${{ steps.export_filter.outputs.filter_type }}" |
| 179 | +
|
| 180 | + if [ -n "$FILTER" ]; then |
| 181 | + if [ "$FILTER_TYPE" = "years" ]; then |
| 182 | + CMD="$CMD --years $FILTER" |
| 183 | + echo "📅 Filtrerar för år: $FILTER" |
| 184 | + elif [ "$FILTER_TYPE" = "beteckningar" ]; then |
| 185 | + CMD="$CMD --filter $FILTER" |
| 186 | + echo "📋 Filtrerar för beteckningar: $FILTER" |
| 187 | + fi |
| 188 | + fi |
| 189 | +
|
| 190 | + # Lägg till skip flags om specificerade |
| 191 | + if [ "${{ inputs.skip_initial }}" = "true" ]; then |
| 192 | + CMD="$CMD --skip-initial" |
| 193 | + echo "⏭️ Hoppar över initial commits" |
| 194 | + fi |
| 195 | +
|
| 196 | + if [ "${{ inputs.skip_temporal }}" = "true" ]; then |
| 197 | + CMD="$CMD --skip-temporal" |
| 198 | + echo "⏭️ Hoppar över temporal commits" |
| 199 | + fi |
| 200 | +
|
| 201 | + echo "Kör: $CMD" |
| 202 | + echo "" |
| 203 | +
|
| 204 | + # Kör export |
| 205 | + eval $CMD |
| 206 | +
|
| 207 | + - name: No files to export |
| 208 | + if: steps.export_filter.outputs.changed_count == '0' && inputs.export_mode == 'changed' |
| 209 | + run: | |
| 210 | + echo "ℹ️ Inga nya/ändrade filer att exportera sedan senaste veckan" |
| 211 | + echo "Körningen avslutas utan fel" |
| 212 | +
|
| 213 | + - name: Export summary |
| 214 | + if: success() |
| 215 | + run: | |
| 216 | + echo "✅ Git export slutförd!" |
| 217 | + echo "Branch: ${{ steps.branch.outputs.branch_name }}" |
| 218 | + echo "Target repo: se-lex/sfs (konfigurerat via GIT_TARGET_REPO env var)" |
| 219 | + echo "" |
| 220 | + echo "Nästa steg:" |
| 221 | + echo "1. Gå till se-lex/sfs repository" |
| 222 | + echo "2. Skapa Pull Request från branch '${{ steps.branch.outputs.branch_name }}'" |
| 223 | + echo "3. Granska och merga till main" |
| 224 | +
|
| 225 | + - name: Export failed |
| 226 | + if: failure() |
| 227 | + run: | |
| 228 | + echo "❌ Git export misslyckades!" |
| 229 | + echo "Kontrollera:" |
| 230 | + echo "1. GIT_GITHUB_PAT secret är korrekt konfigurerad" |
| 231 | + echo "2. PAT har skrivrättigheter till target repository" |
| 232 | + echo "3. Data finns i workflow-artifact-data branch" |
| 233 | + echo "4. Loggar ovan för feldetaljer" |
0 commit comments