clerk/internal/analyzer/build.go (view raw)
| 1 | package analyzer |
| 2 | |
| 3 | import ( |
| 4 | "fmt" |
| 5 | "sort" |
| 6 | "strings" |
| 7 | |
| 8 | "olexsmir.xyz/clerk/journal" |
| 9 | "olexsmir.xyz/clerk/journal/ast" |
| 10 | ) |
| 11 | |
| 12 | // Build constructs [Analysis] from a flat resolved journal view. |
| 13 | func Build(rj *journal.ResolvedJournal) *Analysis { |
| 14 | fileIndex := make(map[*journal.ParsedFile]int) |
| 15 | for i, pf := range rj.Occurrences { |
| 16 | fileIndex[pf] = i |
| 17 | } |
| 18 | |
| 19 | a := &Analysis{ |
| 20 | Files: rj.Occurrences, |
| 21 | Accounts: make(map[string]*AccountInfo), |
| 22 | Commodities: make(map[string]*CommodityInfo), |
| 23 | Payees: make(map[string]*PayeeInfo), |
| 24 | Tags: make(map[string]*TagInfo), |
| 25 | AccountsByPrefix: make(map[string][]string), |
| 26 | PayeeTemplates: make(map[string][]PostingTemplate), |
| 27 | CommodityDecimalMarks: make(map[string]byte), |
| 28 | TransactionsByKey: make(map[string][]*ast.Transaction), |
| 29 | } |
| 30 | for _, item := range rj.Items { |
| 31 | if item.IsInclude { |
| 32 | continue |
| 33 | } |
| 34 | idx := fileIndex[item.Occurrence] |
| 35 | a.addEntry(idx, item.Occurrence.Ast.Entries[item.EntryIndex]) |
| 36 | } |
| 37 | a.buildPrefixIndex() |
| 38 | a.sortAccountNames() |
| 39 | a.collectPayeeNames() |
| 40 | a.collectDates() |
| 41 | a.collectTags() |
| 42 | return a |
| 43 | } |
| 44 | |
| 45 | func TxDuplicateKey(tx *ast.Transaction) string { |
| 46 | var b strings.Builder |
| 47 | fmt.Fprintf(&b, "%04d-%02d-%02d|", tx.Date.Year, tx.Date.Month, tx.Date.Day) |
| 48 | if tx.Payee != nil { |
| 49 | b.WriteString(tx.Payee.Name) |
| 50 | } |
| 51 | b.WriteByte('|') |
| 52 | for _, p := range tx.Postings { |
| 53 | b.WriteString(p.Account.String()) |
| 54 | b.WriteByte(',') |
| 55 | } |
| 56 | return b.String() |
| 57 | } |
| 58 | |
| 59 | func (a *Analysis) addEntry(fileIndex int, entry ast.Entry) { |
| 60 | switch e := entry.(type) { |
| 61 | case *ast.AccountDirective: |
| 62 | a.addAccountDirective(e) |
| 63 | case *ast.CommodityDirective: |
| 64 | a.addCommodityDirective(e) |
| 65 | case *ast.PayeeDirective: |
| 66 | a.addPayeeDirective(e) |
| 67 | case *ast.TagDirective: |
| 68 | a.addTagDirective(e) |
| 69 | case *ast.Comment: |
| 70 | a.addCommentTags(fileIndex, nil, e) |
| 71 | case *ast.Transaction: |
| 72 | a.Transactions = append(a.Transactions, e) |
| 73 | a.addPostings(fileIndex, e.Postings, &e.Date) |
| 74 | a.addPayee(fileIndex, e.Payee) |
| 75 | a.addPayeeTemplate(e) |
| 76 | a.addCommentTags(fileIndex, &e.Date, e.Comment) |
| 77 | for _, c := range e.HeaderComments { |
| 78 | a.addCommentTags(fileIndex, &e.Date, c) |
| 79 | } |
| 80 | key := TxDuplicateKey(e) |
| 81 | a.TransactionsByKey[key] = append(a.TransactionsByKey[key], e) |
| 82 | case *ast.PeriodicTransaction: |
| 83 | a.PeriodicTransactions = append(a.PeriodicTransactions, e) |
| 84 | a.addPostings(fileIndex, e.Postings, nil) |
| 85 | a.addCommentTags(fileIndex, nil, e.Comment) |
| 86 | for _, c := range e.HeaderComments { |
| 87 | a.addCommentTags(fileIndex, nil, c) |
| 88 | } |
| 89 | case *ast.AutomatedTransaction: |
| 90 | a.AutomatedTransactions = append(a.AutomatedTransactions, e) |
| 91 | a.addPostings(fileIndex, e.Postings, nil) |
| 92 | a.addCommentTags(fileIndex, nil, e.Comment) |
| 93 | for _, c := range e.HeaderComments { |
| 94 | a.addCommentTags(fileIndex, nil, c) |
| 95 | } |
| 96 | case *ast.DefaultCommodityDirective: |
| 97 | if e.Amount.Commodity != "" { |
| 98 | mark := e.Amount.QuantityFmt.Decimal |
| 99 | if mark != 0 { |
| 100 | a.CommodityDecimalMarks[e.Amount.Commodity] = mark |
| 101 | } |
| 102 | } |
| 103 | } |
| 104 | |
| 105 | // Every directive-like entry goes into Directives. |
| 106 | switch entry.(type) { |
| 107 | case *ast.AccountDirective, *ast.CommodityDirective, *ast.PayeeDirective, *ast.TagDirective, *ast.IncludeDirective, |
| 108 | *ast.AliasDirective, *ast.YearDirective, *ast.DecimalMarkDirective, *ast.DefaultCommodityDirective, *ast.MarketPriceDirective, |
| 109 | *ast.ConversionDirective, *ast.ApplyDirective, *ast.EndDirective, *ast.CommentBlockDirective, *ast.IgnoredDirective: |
| 110 | a.Directives = append(a.Directives, entry) |
| 111 | } |
| 112 | } |
| 113 | |
| 114 | func (a *Analysis) addAccountDirective(ad *ast.AccountDirective) { |
| 115 | aname := ad.Account.String() |
| 116 | info, ok := a.Accounts[aname] |
| 117 | if !ok { |
| 118 | info = &AccountInfo{} |
| 119 | a.Accounts[aname] = info |
| 120 | } |
| 121 | info.Directives = append(info.Directives, ad) |
| 122 | } |
| 123 | |
| 124 | func (a *Analysis) addPayeeDirective(pd *ast.PayeeDirective) { |
| 125 | info, ok := a.Payees[pd.Name] |
| 126 | if !ok { |
| 127 | info = &PayeeInfo{} |
| 128 | a.Payees[pd.Name] = info |
| 129 | } |
| 130 | info.Directives = append(info.Directives, &ast.Payee{ |
| 131 | Name: pd.Name, |
| 132 | Span: pd.Span, |
| 133 | }) |
| 134 | } |
| 135 | |
| 136 | func (a *Analysis) addTagDirective(td *ast.TagDirective) { |
| 137 | if td.Name == "" { |
| 138 | return |
| 139 | } |
| 140 | info, ok := a.Tags[td.Name] |
| 141 | if !ok { |
| 142 | info = &TagInfo{} |
| 143 | a.Tags[td.Name] = info |
| 144 | } |
| 145 | info.Directives = append(info.Directives, td) |
| 146 | } |
| 147 | |
| 148 | func (a *Analysis) addCommentTags(fileIndex int, date *ast.Date, c *ast.Comment) { |
| 149 | if c == nil { |
| 150 | return |
| 151 | } |
| 152 | for i := range c.Tags { |
| 153 | t := &c.Tags[i] |
| 154 | info, ok := a.Tags[t.Key] |
| 155 | if !ok { |
| 156 | info = &TagInfo{} |
| 157 | a.Tags[t.Key] = info |
| 158 | } |
| 159 | info.Usage = append(info.Usage, TagUsage{FileIndex: fileIndex, Tag: t}) |
| 160 | info.UsedCount++ |
| 161 | if date != nil { |
| 162 | info.LastUsed = maxDate(info.LastUsed, *date) |
| 163 | } |
| 164 | } |
| 165 | } |
| 166 | |
| 167 | func (a *Analysis) addCommodityDirective(cd *ast.CommodityDirective) { |
| 168 | info, ok := a.Commodities[cd.Commodity] |
| 169 | if !ok { |
| 170 | info = &CommodityInfo{} |
| 171 | a.Commodities[cd.Commodity] = info |
| 172 | } |
| 173 | info.Directives = append(info.Directives, cd) |
| 174 | } |
| 175 | |
| 176 | func (a *Analysis) addPayee(fileIndex int, payee *ast.Payee) { |
| 177 | if payee == nil { |
| 178 | return |
| 179 | } |
| 180 | info, ok := a.Payees[payee.Name] |
| 181 | if !ok { |
| 182 | info = &PayeeInfo{} |
| 183 | a.Payees[payee.Name] = info |
| 184 | } |
| 185 | info.Usage = append(info.Usage, PayeeUsage{ |
| 186 | FileIndex: fileIndex, |
| 187 | Payee: payee, |
| 188 | }) |
| 189 | info.UsedCount++ |
| 190 | } |
| 191 | |
| 192 | func (a *Analysis) addPostings(fileIndex int, postings []*ast.Posting, date *ast.Date) { |
| 193 | for _, posting := range postings { |
| 194 | aname := posting.Account.String() |
| 195 | info, ok := a.Accounts[aname] |
| 196 | if !ok { |
| 197 | info = &AccountInfo{} |
| 198 | a.Accounts[aname] = info |
| 199 | } |
| 200 | info.Usages = append(info.Usages, AccountUsage{ |
| 201 | FileIndex: fileIndex, |
| 202 | Posting: posting, |
| 203 | }) |
| 204 | info.UsedCount++ |
| 205 | if date != nil { |
| 206 | info.LastUsed = maxDate(info.LastUsed, *date) |
| 207 | } |
| 208 | |
| 209 | a.addCommodityUsage(fileIndex, posting.Amount, date) |
| 210 | if posting.Cost != nil { |
| 211 | a.addCommodityUsage(fileIndex, &posting.Cost.Amount, date) |
| 212 | } |
| 213 | if posting.Balance != nil { |
| 214 | a.addCommodityUsage(fileIndex, &posting.Balance.Amount, date) |
| 215 | if posting.Balance.Cost != nil { |
| 216 | a.addCommodityUsage(fileIndex, &posting.Balance.Cost.Amount, date) |
| 217 | } |
| 218 | } |
| 219 | |
| 220 | a.addCommentTags(fileIndex, date, posting.Comment) |
| 221 | for i := range posting.Comments { |
| 222 | a.addCommentTags(fileIndex, date, &posting.Comments[i]) |
| 223 | } |
| 224 | |
| 225 | // Collect decimal mark from amount formatting. |
| 226 | if posting.Amount != nil && posting.Amount.Commodity != "" && posting.Amount.QuantityFmt.Decimal != 0 { |
| 227 | if _, ok := a.CommodityDecimalMarks[posting.Amount.Commodity]; !ok { |
| 228 | a.CommodityDecimalMarks[posting.Amount.Commodity] = posting.Amount.QuantityFmt.Decimal |
| 229 | } |
| 230 | } |
| 231 | } |
| 232 | } |
| 233 | |
| 234 | func (a *Analysis) addPayeeTemplate(tx *ast.Transaction) { |
| 235 | payee := "" |
| 236 | if tx.Payee != nil { |
| 237 | payee = tx.Payee.Name |
| 238 | } |
| 239 | if payee == "" { |
| 240 | return |
| 241 | } |
| 242 | |
| 243 | templates := make([]PostingTemplate, len(tx.Postings)) |
| 244 | for i, p := range tx.Postings { |
| 245 | t := PostingTemplate{ |
| 246 | Account: p.Account.String(), |
| 247 | IsInferred: p.Amount == nil, |
| 248 | } |
| 249 | if p.Amount != nil { |
| 250 | t.Amount = p.Amount.Quantity.String() |
| 251 | t.Commodity = p.Amount.Commodity |
| 252 | } |
| 253 | templates[i] = t |
| 254 | } |
| 255 | a.PayeeTemplates[payee] = templates |
| 256 | } |
| 257 | |
| 258 | func (a *Analysis) collectDates() { |
| 259 | seen := make(map[string]bool) |
| 260 | for _, tx := range a.Transactions { |
| 261 | s := formatDate(tx.Date) |
| 262 | if s != "" && !seen[s] { |
| 263 | seen[s] = true |
| 264 | a.Dates = append(a.Dates, tx.Date) |
| 265 | a.DateStrings = append(a.DateStrings, s) |
| 266 | } |
| 267 | } |
| 268 | sort.Slice(a.Dates, func(i, j int) bool { |
| 269 | return a.Dates[i].Compare(a.Dates[j]) < 0 |
| 270 | }) |
| 271 | sort.Strings(a.DateStrings) |
| 272 | } |
| 273 | |
| 274 | func (a *Analysis) collectTags() { |
| 275 | names := make([]string, 0, len(a.Tags)) |
| 276 | values := make(map[string]bool) |
| 277 | for name, info := range a.Tags { |
| 278 | names = append(names, name) |
| 279 | seen := make(map[string]bool) |
| 280 | for _, u := range info.Usage { |
| 281 | if u.Tag.Value == "" { |
| 282 | continue |
| 283 | } |
| 284 | seen[u.Tag.Value] = true |
| 285 | values[u.Tag.Value] = true |
| 286 | } |
| 287 | if len(seen) > 0 { |
| 288 | info.Values = sortedKeys(seen) |
| 289 | } |
| 290 | } |
| 291 | sort.Strings(names) |
| 292 | a.TagNames = names |
| 293 | a.TagValues = sortedKeys(values) |
| 294 | } |
| 295 | |
| 296 | func sortedKeys(m map[string]bool) []string { |
| 297 | out := make([]string, 0, len(m)) |
| 298 | for k := range m { |
| 299 | out = append(out, k) |
| 300 | } |
| 301 | sort.Strings(out) |
| 302 | return out |
| 303 | } |
| 304 | |
| 305 | func formatDate(d ast.Date) string { |
| 306 | if d.Year == 0 && d.Month == 0 && d.Day == 0 { |
| 307 | return "" |
| 308 | } |
| 309 | if d.Month < 1 || d.Month > 12 || d.Day < 1 || d.Day > 31 { |
| 310 | return "" |
| 311 | } |
| 312 | return fmt.Sprintf("%04d-%02d-%02d", d.Year, d.Month, d.Day) |
| 313 | } |
| 314 | |
| 315 | func (a *Analysis) addCommodityUsage(fileIndex int, am *ast.Amount, date *ast.Date) { |
| 316 | if am == nil || am.Commodity == "" { |
| 317 | return |
| 318 | } |
| 319 | info, ok := a.Commodities[am.Commodity] |
| 320 | if !ok { |
| 321 | info = &CommodityInfo{} |
| 322 | a.Commodities[am.Commodity] = info |
| 323 | } |
| 324 | info.Usages = append(info.Usages, CommodityUsage{ |
| 325 | FileIndex: fileIndex, |
| 326 | Amount: am, |
| 327 | }) |
| 328 | info.UsedCount++ |
| 329 | if date != nil { |
| 330 | info.LastUsed = maxDate(info.LastUsed, *date) |
| 331 | } |
| 332 | } |
| 333 | |
| 334 | func (a *Analysis) buildPrefixIndex() { |
| 335 | for name := range a.Accounts { |
| 336 | parts := strings.Split(name, ":") |
| 337 | for i := 1; i < len(parts); i++ { |
| 338 | prefix := strings.Join(parts[:i], ":") + ":" |
| 339 | a.AccountsByPrefix[prefix] = append(a.AccountsByPrefix[prefix], name) |
| 340 | } |
| 341 | } |
| 342 | } |
| 343 | |
| 344 | func (a *Analysis) collectPayeeNames() { |
| 345 | names := make([]string, 0, len(a.Payees)) |
| 346 | for name := range a.Payees { |
| 347 | names = append(names, name) |
| 348 | } |
| 349 | sort.Strings(names) |
| 350 | a.PayeeNames = names |
| 351 | } |
| 352 | |
| 353 | func (a *Analysis) sortAccountNames() { |
| 354 | names := make([]string, 0, len(a.Accounts)) |
| 355 | for name := range a.Accounts { |
| 356 | names = append(names, name) |
| 357 | } |
| 358 | sort.Strings(names) |
| 359 | a.AccountNames = names |
| 360 | } |
| 361 | |
| 362 | func maxDate(a, b ast.Date) ast.Date { |
| 363 | if a.Year > b.Year || |
| 364 | (a.Year == b.Year && (a.Month > b.Month || |
| 365 | (a.Year == b.Year && a.Month == b.Month && a.Day > b.Day))) { |
| 366 | return a |
| 367 | } |
| 368 | return b |
| 369 | } |