cloudflare/pint
Publicmirrored from https://github.com/cloudflare/pintAvailable
internal/discovery/git_branch.go
354lines · modecode
| 1 | package discovery |
| 2 | |
| 3 | import ( |
| 4 | "bytes" |
| 5 | "fmt" |
| 6 | "log/slog" |
| 7 | "slices" |
| 8 | "sort" |
| 9 | "strings" |
| 10 | |
| 11 | "github.com/cloudflare/pint/internal/git" |
| 12 | "github.com/cloudflare/pint/internal/output" |
| 13 | "github.com/cloudflare/pint/internal/parser" |
| 14 | ) |
| 15 | |
| 16 | func NewGitBranchFinder( |
| 17 | gitCmd git.CommandRunner, |
| 18 | filter git.PathFilter, |
| 19 | baseBranch string, |
| 20 | maxCommits int, |
| 21 | ) GitBranchFinder { |
| 22 | return GitBranchFinder{ |
| 23 | gitCmd: gitCmd, |
| 24 | filter: filter, |
| 25 | baseBranch: baseBranch, |
| 26 | maxCommits: maxCommits, |
| 27 | } |
| 28 | } |
| 29 | |
| 30 | type GitBranchFinder struct { |
| 31 | gitCmd git.CommandRunner |
| 32 | baseBranch string |
| 33 | filter git.PathFilter |
| 34 | maxCommits int |
| 35 | } |
| 36 | |
| 37 | func (f GitBranchFinder) Find(allEntries []Entry) (entries []Entry, err error) { |
| 38 | changes, err := git.Changes(f.gitCmd, f.baseBranch, f.filter) |
| 39 | if err != nil { |
| 40 | return nil, err |
| 41 | } |
| 42 | |
| 43 | if totalCommits := countCommits(changes); totalCommits > f.maxCommits { |
| 44 | return nil, fmt.Errorf("number of commits to check (%d) is higher than maxCommits (%d), exiting", totalCommits, f.maxCommits) |
| 45 | } |
| 46 | |
| 47 | shouldSkip, err := f.shouldSkipAllChecks(changes) |
| 48 | if err != nil { |
| 49 | return nil, err |
| 50 | } |
| 51 | if shouldSkip { |
| 52 | return nil, nil |
| 53 | } |
| 54 | |
| 55 | for _, change := range changes { |
| 56 | var entriesBefore, entriesAfter []Entry |
| 57 | entriesBefore, err = readRules( |
| 58 | change.Path.Before.EffectivePath(), |
| 59 | change.Path.Before.Name, |
| 60 | bytes.NewReader(change.Body.Before), |
| 61 | !f.filter.IsRelaxed(change.Path.Before.Name), |
| 62 | ) |
| 63 | if err != nil { |
| 64 | slog.Debug("Cannot read before rules", slog.String("path", change.Path.Before.Name), slog.Any("err", err)) |
| 65 | } |
| 66 | entriesAfter, err = readRules( |
| 67 | change.Path.After.EffectivePath(), |
| 68 | change.Path.After.Name, |
| 69 | bytes.NewReader(change.Body.After), |
| 70 | !f.filter.IsRelaxed(change.Path.After.Name), |
| 71 | ) |
| 72 | if err != nil { |
| 73 | return nil, fmt.Errorf("invalid file syntax: %w", err) |
| 74 | } |
| 75 | |
| 76 | failedEntries := entriesWithPathErrors(entriesAfter) |
| 77 | |
| 78 | slog.Debug( |
| 79 | "Parsing git file change", |
| 80 | slog.Any("commits", change.Commits), |
| 81 | slog.String("before.path", change.Path.Before.Name), |
| 82 | slog.String("before.target", change.Path.Before.SymlinkTarget), |
| 83 | slog.Any("before.type", change.Path.Before.Type), |
| 84 | slog.Int("before.entries", len(entriesBefore)), |
| 85 | slog.String("after.path", change.Path.After.Name), |
| 86 | slog.String("after.target", change.Path.After.SymlinkTarget), |
| 87 | slog.Any("after.type", change.Path.After.Type), |
| 88 | slog.Int("after.entries", len(entriesAfter)), |
| 89 | slog.Any("modifiedLines", change.Body.ModifiedLines), |
| 90 | ) |
| 91 | for _, me := range matchEntries(entriesBefore, entriesAfter) { |
| 92 | switch { |
| 93 | case !me.hasBefore && me.hasAfter: |
| 94 | me.after.State = Added |
| 95 | me.after.ModifiedLines = commonLines(change.Body.ModifiedLines, me.after.ModifiedLines) |
| 96 | slog.Debug( |
| 97 | "Rule added on HEAD branch", |
| 98 | slog.String("name", me.after.Rule.Name()), |
| 99 | slog.String("state", me.after.State.String()), |
| 100 | slog.String("path", me.after.Path.Name), |
| 101 | slog.String("ruleLines", me.after.Rule.Lines.String()), |
| 102 | slog.String("modifiedLines", output.FormatLineRangeString(me.after.ModifiedLines)), |
| 103 | ) |
| 104 | entries = append(entries, me.after) |
| 105 | case me.hasBefore && me.hasAfter: |
| 106 | switch { |
| 107 | case me.isIdentical && !me.wasMoved: |
| 108 | me.after.State = Noop |
| 109 | me.after.ModifiedLines = []int{} |
| 110 | slog.Debug( |
| 111 | "Rule content was not modified on HEAD, identical rule present before", |
| 112 | slog.String("name", me.after.Rule.Name()), |
| 113 | slog.String("lines", me.after.Rule.Lines.String()), |
| 114 | ) |
| 115 | case me.wasMoved: |
| 116 | me.after.State = Moved |
| 117 | me.after.ModifiedLines = git.CountLines(change.Body.After) |
| 118 | slog.Debug( |
| 119 | "Rule content was not modified on HEAD but the file was moved or renamed", |
| 120 | slog.String("name", me.after.Rule.Name()), |
| 121 | slog.String("lines", me.after.Rule.Lines.String()), |
| 122 | ) |
| 123 | default: |
| 124 | me.after.State = Modified |
| 125 | me.after.ModifiedLines = commonLines(change.Body.ModifiedLines, me.after.ModifiedLines) |
| 126 | slog.Debug( |
| 127 | "Rule modified on HEAD branch", |
| 128 | slog.String("name", me.after.Rule.Name()), |
| 129 | slog.String("state", me.after.State.String()), |
| 130 | slog.String("path", me.after.Path.Name), |
| 131 | slog.String("ruleLines", me.after.Rule.Lines.String()), |
| 132 | slog.String("modifiedLines", output.FormatLineRangeString(me.after.ModifiedLines)), |
| 133 | ) |
| 134 | } |
| 135 | entries = append(entries, me.after) |
| 136 | case me.hasBefore && !me.hasAfter && len(failedEntries) == 0: |
| 137 | me.before.State = Removed |
| 138 | ml := commonLines(change.Body.ModifiedLines, me.before.ModifiedLines) |
| 139 | if len(ml) > 0 { |
| 140 | me.before.ModifiedLines = ml |
| 141 | } |
| 142 | slog.Debug( |
| 143 | "Rule removed on HEAD branch", |
| 144 | slog.String("name", me.before.Rule.Name()), |
| 145 | slog.String("state", me.before.State.String()), |
| 146 | slog.String("path", me.before.Path.Name), |
| 147 | slog.String("ruleLines", me.before.Rule.Lines.String()), |
| 148 | slog.String("modifiedLines", output.FormatLineRangeString(me.before.ModifiedLines)), |
| 149 | ) |
| 150 | entries = append(entries, me.before) |
| 151 | case me.hasBefore && !me.hasAfter && len(failedEntries) > 0: |
| 152 | slog.Debug( |
| 153 | "Rule not present on HEAD branch but there are parse errors", |
| 154 | slog.String("name", me.before.Rule.Name()), |
| 155 | slog.String("state", me.before.State.String()), |
| 156 | slog.String("path", me.before.Path.Name), |
| 157 | slog.String("ruleLines", me.before.Rule.Lines.String()), |
| 158 | slog.String("modifiedLines", output.FormatLineRangeString(me.before.ModifiedLines)), |
| 159 | ) |
| 160 | default: |
| 161 | slog.Warn( |
| 162 | "Unknown rule state", |
| 163 | slog.String("state", me.before.State.String()), |
| 164 | slog.String("path", me.before.Path.Name), |
| 165 | slog.String("modifiedLines", output.FormatLineRangeString(me.before.ModifiedLines)), |
| 166 | ) |
| 167 | entries = append(entries, me.after) |
| 168 | } |
| 169 | } |
| 170 | } |
| 171 | |
| 172 | symlinks, err := addSymlinkedEntries(entries) |
| 173 | if err != nil { |
| 174 | return nil, err |
| 175 | } |
| 176 | |
| 177 | for _, entry := range symlinks { |
| 178 | if f.filter.IsPathAllowed(entry.Path.Name) { |
| 179 | entries = append(entries, entry) |
| 180 | } |
| 181 | } |
| 182 | |
| 183 | var found bool |
| 184 | for _, entry := range entries { |
| 185 | found = false |
| 186 | if entry.State == Removed { |
| 187 | goto NEXT |
| 188 | } |
| 189 | for i, globEntry := range allEntries { |
| 190 | if entry.Path.Name == globEntry.Path.Name && entry.Rule.IsSame(globEntry.Rule) { |
| 191 | allEntries[i].State = entry.State |
| 192 | allEntries[i].ModifiedLines = entry.ModifiedLines |
| 193 | found = true |
| 194 | break |
| 195 | } |
| 196 | } |
| 197 | NEXT: |
| 198 | if !found { |
| 199 | allEntries = append(allEntries, entry) |
| 200 | } |
| 201 | } |
| 202 | |
| 203 | slog.Debug("Git branch finder completed", slog.Int("count", len(allEntries))) |
| 204 | return allEntries, nil |
| 205 | } |
| 206 | |
| 207 | func (f GitBranchFinder) shouldSkipAllChecks(changes []*git.FileChange) (bool, error) { |
| 208 | commits := map[string]struct{}{} |
| 209 | for _, change := range changes { |
| 210 | for _, commit := range change.Commits { |
| 211 | commits[commit] = struct{}{} |
| 212 | } |
| 213 | } |
| 214 | |
| 215 | for commit := range commits { |
| 216 | msg, err := git.CommitMessage(f.gitCmd, commit) |
| 217 | if err != nil { |
| 218 | return false, fmt.Errorf("failed to get commit message for %s: %w", commit, err) |
| 219 | } |
| 220 | for _, comment := range []string{"[skip ci]", "[no ci]"} { |
| 221 | if strings.Contains(msg, comment) { |
| 222 | slog.Info( |
| 223 | fmt.Sprintf("Found a commit with '%s', skipping all checks", comment), |
| 224 | slog.String("commit", commit)) |
| 225 | return true, nil |
| 226 | } |
| 227 | } |
| 228 | } |
| 229 | |
| 230 | return false, nil |
| 231 | } |
| 232 | |
| 233 | func commonLines(a, b []int) (common []int) { |
| 234 | for _, ai := range a { |
| 235 | if slices.Contains(b, ai) { |
| 236 | common = append(common, ai) |
| 237 | } |
| 238 | } |
| 239 | for _, bi := range b { |
| 240 | if slices.Contains(a, bi) && !slices.Contains(common, bi) { |
| 241 | common = append(common, bi) |
| 242 | } |
| 243 | } |
| 244 | return common |
| 245 | } |
| 246 | |
| 247 | type matchedEntry struct { |
| 248 | before Entry |
| 249 | after Entry |
| 250 | hasBefore bool |
| 251 | hasAfter bool |
| 252 | isIdentical bool |
| 253 | wasMoved bool |
| 254 | } |
| 255 | |
| 256 | func matchEntries(before, after []Entry) (ml []matchedEntry) { |
| 257 | for _, a := range after { |
| 258 | slog.Debug( |
| 259 | "Matching HEAD rule", |
| 260 | slog.String("path", a.Path.Name), |
| 261 | slog.String("source", a.Path.SymlinkTarget), |
| 262 | slog.String("name", a.Rule.Name()), |
| 263 | ) |
| 264 | |
| 265 | m := matchedEntry{after: a, hasAfter: true} |
| 266 | beforeSwap := make([]Entry, 0, len(before)) |
| 267 | var matches []Entry |
| 268 | var matched bool |
| 269 | |
| 270 | for _, b := range before { |
| 271 | if !matched && a.Rule.Name() != "" && a.Rule.IsIdentical(b.Rule) { |
| 272 | m.before = b |
| 273 | m.hasBefore = true |
| 274 | m.isIdentical = isEntryIdentical(b, a) |
| 275 | m.wasMoved = a.Path.Name != b.Path.Name |
| 276 | matched = true |
| 277 | slog.Debug( |
| 278 | "Found identical rule on before & after", |
| 279 | slog.Bool("identical", m.isIdentical), |
| 280 | slog.Bool("moved", m.wasMoved), |
| 281 | ) |
| 282 | } else { |
| 283 | beforeSwap = append(beforeSwap, b) |
| 284 | } |
| 285 | } |
| 286 | before = beforeSwap |
| 287 | |
| 288 | if !matched { |
| 289 | before, matches = findRulesByName(before, a.Rule.Name(), a.Rule.Type()) |
| 290 | switch len(matches) { |
| 291 | case 0: |
| 292 | case 1: |
| 293 | m.before = matches[0] |
| 294 | m.hasBefore = true |
| 295 | m.wasMoved = a.Path.Name != matches[0].Path.Name |
| 296 | slog.Debug("Found rule with same name on before & after") |
| 297 | default: |
| 298 | slog.Debug( |
| 299 | "Found multiple rules with same name on before & after", |
| 300 | slog.Int("matches", len(matches)), |
| 301 | ) |
| 302 | before = append(before, matches...) |
| 303 | } |
| 304 | } |
| 305 | |
| 306 | ml = append(ml, m) |
| 307 | } |
| 308 | |
| 309 | for _, b := range before { |
| 310 | ml = append(ml, matchedEntry{before: b, hasBefore: true}) |
| 311 | } |
| 312 | |
| 313 | return ml |
| 314 | } |
| 315 | |
| 316 | func isEntryIdentical(b, a Entry) bool { |
| 317 | if !slices.Equal(sort.StringSlice(b.DisabledChecks), sort.StringSlice(a.DisabledChecks)) { |
| 318 | slog.Debug("List of disabled checks was modified", |
| 319 | slog.Any("before", sort.StringSlice(b.DisabledChecks)), |
| 320 | slog.Any("after", sort.StringSlice(a.DisabledChecks))) |
| 321 | return false |
| 322 | } |
| 323 | return true |
| 324 | } |
| 325 | |
| 326 | func findRulesByName(entries []Entry, name string, typ parser.RuleType) (nomatch, match []Entry) { |
| 327 | for _, entry := range entries { |
| 328 | if entry.PathError == nil && entry.Rule.Type() == typ && entry.Rule.Name() == name { |
| 329 | match = append(match, entry) |
| 330 | } else { |
| 331 | nomatch = append(nomatch, entry) |
| 332 | } |
| 333 | } |
| 334 | return nomatch, match |
| 335 | } |
| 336 | |
| 337 | func countCommits(changes []*git.FileChange) int { |
| 338 | commits := map[string]struct{}{} |
| 339 | for _, change := range changes { |
| 340 | for _, commit := range change.Commits { |
| 341 | commits[commit] = struct{}{} |
| 342 | } |
| 343 | } |
| 344 | return len(commits) |
| 345 | } |
| 346 | |
| 347 | func entriesWithPathErrors(entries []Entry) (match []Entry) { |
| 348 | for _, entry := range entries { |
| 349 | if entry.PathError != nil { |
| 350 | match = append(match, entry) |
| 351 | } |
| 352 | } |
| 353 | return match |
| 354 | } |
| 355 | |