diff options
Diffstat (limited to 'src/mintage')
-rw-r--r-- | src/mintage/parser.go | 297 | ||||
-rw-r--r-- | src/mintage/parser_test.go | 233 |
2 files changed, 530 insertions, 0 deletions
diff --git a/src/mintage/parser.go b/src/mintage/parser.go new file mode 100644 index 0000000..364b6e8 --- /dev/null +++ b/src/mintage/parser.go @@ -0,0 +1,297 @@ +package mintage + +import ( + "bufio" + "fmt" + "io" + "strconv" + "strings" + "time" +) + +type SyntaxError struct { + expected, got string + file string + linenr int +} + +func (e SyntaxError) Error() string { + return fmt.Sprintf("%s:%d: syntax error: expected %s but got %s", + e.file, e.linenr, e.expected, e.got) +} + +type Data struct { + Standard []SRow + Commemorative []CRow +} + +type SRow struct { + Year int + Mintmark string + Mintages [typeCount][denoms]int +} + +type CRow struct { + Year int + Mintmark string + Name string + Mintage [typeCount]int +} + +const ( + TypeCirc = iota + TypeNIFC + TypeProof + typeCount +) + +const ( + Unknown = -iota - 1 + Invalid +) + +const ( + denoms = 8 + ws = " \t" +) + +func Parse(r io.Reader, file string) (Data, error) { + yearsSince := time.Now().Year() - 1999 + 1 + data := Data{ + Standard: make([]SRow, 0, yearsSince), + Commemorative: make([]CRow, 0, yearsSince), + } + + scanner := bufio.NewScanner(r) + for linenr := 1; scanner.Scan(); linenr++ { + line := strings.Trim(scanner.Text(), ws) + if isBlankOrComment(line) { + continue + } + + if len(line) < 4 || !isNumeric(line[:4], false) { + return Data{}, SyntaxError{ + expected: "4-digit year", + got: line, + linenr: linenr, + file: file, + } + } + + var ( + commem bool + mintmark string + ) + year, _ := strconv.Atoi(line[:4]) + line = line[4:] + + if len(line) != 0 { + if strings.ContainsRune(ws, rune(line[0])) { + commem = true + goto out + } + if line[0] != '-' { + return Data{}, SyntaxError{ + expected: "end-of-line or mintmark", + got: line, + linenr: linenr, + file: file, + } + } + + if line = line[1:]; len(line) == 0 { + return Data{}, SyntaxError{ + expected: "mintmark", + got: "end-of-line", + linenr: linenr, + file: file, + } + } + + switch i := strings.IndexAny(line, ws); i { + case 0: + return Data{}, SyntaxError{ + expected: "mintmark", + got: "whitespace", + linenr: linenr, + file: file, + } + case -1: + mintmark = line + default: + mintmark, line = line[:i], line[i:] + commem = true + } + } + out: + + if !commem { + row := SRow{ + Year: year, + Mintmark: mintmark, + } + for i := range row.Mintages { + line = "" + for isBlankOrComment(line) { + if !scanner.Scan() { + return Data{}, SyntaxError{ + expected: "mintage row", + got: "end-of-file", + linenr: linenr, + file: file, + } + } + line = strings.Trim(scanner.Text(), ws) + linenr++ + } + + tokens := strings.FieldsFunc(line, func(r rune) bool { + return strings.ContainsRune(ws, r) + }) + if tokcnt := len(tokens); tokcnt != denoms { + word := "entries" + if tokcnt == 1 { + word = "entry" + } + return Data{}, SyntaxError{ + expected: fmt.Sprintf("%d mintage entries", denoms), + got: fmt.Sprintf("%d %s", tokcnt, word), + linenr: linenr, + file: file, + } + } + + for j, tok := range tokens { + if tok != "?" && !isNumeric(tok, true) { + return Data{}, SyntaxError{ + expected: "numeric mintage figure or ‘?’", + got: tok, + linenr: linenr, + file: file, + } + } + + if tok == "?" { + row.Mintages[i][j] = Unknown + } else { + row.Mintages[i][j] = atoiWithDots(tok) + } + } + } + + data.Standard = append(data.Standard, row) + } else { + row := CRow{ + Year: year, + Mintmark: mintmark, + } + line = strings.TrimLeft(line, ws) + if line[0] != '"' { + return Data{}, SyntaxError{ + expected: "string", + got: line, + linenr: linenr, + file: file, + } + } + + line = line[1:] + switch i := strings.IndexByte(line, '"'); i { + case -1: + return Data{}, SyntaxError{ + expected: "closing quote", + got: "end-of-line", + linenr: linenr, + file: file, + } + case 0: + return Data{}, SyntaxError{ + expected: "commemorated event", + got: "empty string", + linenr: linenr, + file: file, + } + default: + row.Name, line = line[:i], line[i+1:] + } + + if len(line) != 0 { + return Data{}, SyntaxError{ + expected: "end-of-line", + got: line, + linenr: linenr, + file: file, + } + } + + for isBlankOrComment(line) { + if !scanner.Scan() { + return Data{}, SyntaxError{ + expected: "mintage row", + got: "end-of-file", + linenr: linenr, + file: file, + } + } + line = strings.Trim(scanner.Text(), ws) + linenr++ + } + + tokens := strings.FieldsFunc(line, func(r rune) bool { + return strings.ContainsRune(ws, r) + }) + if tokcnt := len(tokens); tokcnt != typeCount { + word := "entries" + if tokcnt == 1 { + word = "entry" + } + return Data{}, SyntaxError{ + expected: fmt.Sprintf("%d mintage entries", typeCount), + got: fmt.Sprintf("%d %s", tokcnt, word), + linenr: linenr, + file: file, + } + } + + for i, tok := range tokens { + if tok == "?" { + row.Mintage[i] = Unknown + } else { + row.Mintage[i] = atoiWithDots(tok) + } + } + + data.Commemorative = append(data.Commemorative, row) + } + } + + return data, nil +} + +func isNumeric(s string, dot bool) bool { + for _, ch := range s { + switch ch { + case '0', '1', '2', '3', '4', '5', '6', '7', '8', '9': + default: + if ch != '.' || !dot { + return false + } + } + } + return true +} + +func atoiWithDots(s string) int { + n := 0 + for _, ch := range s { + if ch == '.' { + continue + } + n = n*10 + int(ch) - '0' + } + return n +} + +func isBlankOrComment(s string) bool { + return len(s) == 0 || s[0] == '#' +} diff --git a/src/mintage/parser_test.go b/src/mintage/parser_test.go new file mode 100644 index 0000000..76e0f01 --- /dev/null +++ b/src/mintage/parser_test.go @@ -0,0 +1,233 @@ +package mintage + +import ( + "bytes" + "errors" + "testing" +) + +func TestParserComplete(t *testing.T) { + data, err := Parse(bytes.NewBuffer([]byte(` + 2020 + 1000 1001 1002 1003 1004 1005 1006 1007 + 1100 1101 1102 1103 1104 1105 1106 1107 + 1200 1201 1202 1203 1204 1205 1206 1207 + 2021-KNM + 2.000 ? 2002 2003 2004 2005 2006 2007 + 2.100 ? 2102 2103 2104 2105 2106 2107 + 2.200 ? 2202 2203 2204 2205 2206 2207 + 2021-MdP + 3000 3001 3002 3003 3004 3005 3006 3007 + 3100 3101 3102 3103 3104 3105 3106 3107 + 3200 3201 3202 3203 3204 3205 3206 3207 + 2022 + 4000 4001 4.002 4003 4004 4005 4006 4007 + 4100 4101 4.102 4103 4104 4105 4106 4107 + 4200 4201 4.202 4203 4204 4205 4206 4207 + + 2009 "10th Anniversary of Economic and Monetary Union" + 1000 2000 3000 + 2022-⋆ "35th Anniversary of the Erasmus Programme" + 1001 ? 3001 + `)), "-") + + if err != nil { + t.Fatalf(`Expected err=nil; got "%s"`, err) + } + + for i, row := range data.Standard { + for k := TypeCirc; k <= TypeProof; k++ { + for j, col := range row.Mintages[k] { + n := 1000*(i+1) + 100*k + j + if i == 1 && j == 1 { + n = Unknown + } + if col != n { + t.Fatalf("Expected data.Standard[%d].Mintages[%d][%d]=%d; got %d", + i, k, j, col, n) + } + } + } + } + + for i, row := range data.Commemorative { + for k := TypeCirc; k <= TypeProof; k++ { + n := 1000*(k+1) + i + if i == 1 && k == 1 { + n = Unknown + } + if row.Mintage[k] != n { + t.Fatalf("Expected row.Mintage[%d]=%d; got %d", + k, n, row.Mintage[k]) + } + } + } + + if len(data.Standard) != 4 { + t.Fatalf("Expected len(data.Standard)=2; got %d", len(data.Standard)) + } + if len(data.Commemorative) != 2 { + t.Fatalf("Expected len(data.Commemorative)=2; got %d", len(data.Commemorative)) + } + + for i, x := range [...]struct { + year int + mintmark, name string + }{ + {2009, "", "10th Anniversary of Economic and Monetary Union"}, + {2022, "⋆", "35th Anniversary of the Erasmus Programme"}, + } { + if data.Commemorative[i].Year != x.year { + t.Fatalf("Expected data.Commemorative[%d].Year=%d; got %d", + i, x.year, data.Commemorative[i].Year) + } + if data.Commemorative[i].Mintmark != x.mintmark { + t.Fatalf(`Expected data.Commemorative[%d].Mintmark="%s"; got "%s"`, + i, x.mintmark, data.Commemorative[i].Mintmark) + } + if data.Commemorative[i].Name != x.name { + t.Fatalf(`Expected data.Commemorative[%d].Name="%s"; got "%s"`, + i, x.name, data.Commemorative[i].Name) + } + } +} + +func TestParserMintmarks(t *testing.T) { + data, err := Parse(bytes.NewBuffer([]byte(` + 2020 + 1000 1001 1002 1003 1004 1005 1006 1007 + 1100 1101 1102 1103 1104 1105 1106 1107 + 1200 1201 1202 1203 1204 1205 1206 1207 + 2021-KNM + 2.000 ? 2002 2003 2004 2005 2006 2007 + 2.100 ? 2102 2103 2104 2105 2106 2107 + 2.200 ? 2202 2203 2204 2205 2206 2207 + 2021-MdP + 3000 3001 3002 3003 3004 3005 3006 3007 + 3100 3101 3102 3103 3104 3105 3106 3107 + 3200 3201 3202 3203 3204 3205 3206 3207 + 2022 + 4000 4001 4.002 4003 4004 4005 4006 4007 + 4100 4101 4.102 4103 4104 4105 4106 4107 + 4200 4201 4.202 4203 4204 4205 4206 4207 + `)), "-") + + if err != nil { + t.Fatalf(`Expected err=nil; got "%s"`, err) + } + + for i, row := range data.Standard { + for j, col := range row.Mintages[TypeCirc] { + n := 1000*(i+1) + j + if i == 1 && j == 1 { + n = Unknown + } + if col != n { + t.Fatalf("Expected data.Standard[%d].Mintages[TypeCirc][%d]=%d; got %d", + i, j, col, n) + } + } + } + + for i, y := range [...]int{2020, 2021, 2021, 2022} { + if data.Standard[i].Year != y { + t.Fatalf("Expected data.Standard[%d].Year=%d; got %d", + i, y, data.Standard[i].Year) + } + } + + for i, m := range [...]string{"", "KNM", "MdP", ""} { + if data.Standard[i].Mintmark != m { + t.Fatalf(`Expected data.Standard[%d].Mintmark="%s"; got "%s"`, + i, m, data.Standard[i].Mintmark) + } + } +} + +func TestParserNoYear(t *testing.T) { + _, err := Parse(bytes.NewBuffer([]byte(` + 1.000 1001 1002 1003 1004 1005 1006 1007 + 1.100 1101 1102 1103 1104 1105 1106 1107 + 1.200 1201 1202 1203 1204 1205 1206 1207 + 2021 + 2000 ? 2002 2003 2004 2005 2006 2007 + 2100 ? 2102 2103 2104 2105 2106 2107 + 2200 ? 2202 2203 2204 2205 2206 2207 + `)), "-") + + var sErr SyntaxError + if !errors.As(err, &sErr) { + t.Fatalf("Expected err=SyntaxError; got %s", err) + } +} + +func TestParserBadToken(t *testing.T) { + _, err := Parse(bytes.NewBuffer([]byte(` + 2020 + 1.000 1001 1002 1003 1004 1005 1006 1007 + 1.100 1101 1102 1103 1104 1105 1106 1107 + 1.200 1201 1202 1203 1204 1205 1206 1207 + 2021 Naughty! + 2000 ? 2002 2003 2004 2005 2006 2007 + 2100 ? 2102 2103 2104 2105 2106 2107 + 2200 ? 2202 2203 2204 2205 2206 2207 + `)), "-") + + var sErr SyntaxError + if !errors.As(err, &sErr) { + t.Fatalf("Expected err=SyntaxError; got %s", err) + } +} + +func TestParserShortRow(t *testing.T) { + _, err := Parse(bytes.NewBuffer([]byte(` + 2020 + 1.000 1001 1002 1003 1004 1005 1006 1007 + 1.100 1101 1102 1103 1104 1105 1106 1107 + 1.200 1201 1202 1203 1204 1205 1206 1207 + 2021 + 2000 ? 2002 2003 2004 2005 2006 2007 + 2100 ? 2102 2103 2104 2105 2106 + 2200 ? 2202 2203 2204 2205 2206 2207 + `)), "-") + + var sErr SyntaxError + if !errors.As(err, &sErr) { + t.Fatalf("Expected err=SyntaxError; got %s", err) + } +} + +func TestParserLongRow(t *testing.T) { + _, err := Parse(bytes.NewBuffer([]byte(` + 2020 + 1.000 1001 1002 1003 1004 1005 1006 1007 + 1.100 1101 1102 1103 1104 1105 1106 1107 + 1.200 1201 1202 1203 1204 1205 1206 1207 + 2021 + 2000 ? 2002 2003 2004 2005 2006 2007 + 2100 ? 2102 2103 2104 2105 2106 2107 2108 + 2200 ? 2202 2203 2204 2205 2206 2207 + `)), "-") + + var sErr SyntaxError + if !errors.As(err, &sErr) { + t.Fatalf("Expected err=SyntaxError; got %s", err) + } +} + +func TestParserMissingRow(t *testing.T) { + _, err := Parse(bytes.NewBuffer([]byte(` + 2020 + 1.000 1001 1002 1003 1004 1005 1006 1007 + 1.100 1101 1102 1103 1104 1105 1106 1107 + 1.200 1201 1202 1203 1204 1205 1206 1207 + 2021 + 2000 ? 2002 2003 2004 2005 2006 2007 + 2200 ? 2202 2203 2204 2205 2206 2207 + `)), "-") + + var sErr SyntaxError + if !errors.As(err, &sErr) { + t.Fatalf("Expected err=SyntaxError; got %s", err) + } +} |