feat(#18): import de extratos OFX e CSV — parser, dedup, preview e confirmação

Parser OFX tolerante ao formato SGML de bancos brasileiros (sem closing tags, BOM, timezone suffix). Parser CSV com separador de campo/decimal configurável. Dedup por FITID (OFX) ou date+amount+desc normalizado. Fluxo: POST /preview → revisão na UI → POST /confirm salva e loga. Migration 002 adiciona external_id com unique index. Co-Authored-By: Claude Sonnet 4.6 <[email protected]>
2026-05-26 20:17:46 -03:00
parent d6782d51a5
commit 83dff2de78
15 changed files with 1065 additions and 3 deletions
@@ -0,0 +1,114 @@
+package parser
+
+import (
+	"bufio"
+	"io"
+	"math"
+	"strconv"
+	"strings"
+	"time"
+
+	"financeiro-carvalho/internal/model"
+)
+
+// ParseOFX parses OFX SGML files as exported by Brazilian banks.
+// It is tolerant of the non-strict SGML format (no closing tags, BOM, etc.).
+func ParseOFX(r io.Reader) ([]model.ImportRow, error) {
+	scanner := bufio.NewScanner(r)
+	scanner.Buffer(make([]byte, 1024*1024), 1024*1024)
+
+	var rows []model.ImportRow
+	var cur *model.ImportRow
+
+	for scanner.Scan() {
+		line := strings.TrimSpace(scanner.Text())
+		line = strings.TrimPrefix(line, "\xef\xbb\xbf") // strip UTF-8 BOM if present
+
+		if line == "" {
+			continue
+		}
+
+		tag, value := splitOFXLine(line)
+
+		switch tag {
+		case "STMTTRN":
+			if cur != nil {
+				rows = append(rows, *cur)
+			}
+			cur = &model.ImportRow{}
+		case "/STMTTRN":
+			if cur != nil {
+				rows = append(rows, *cur)
+				cur = nil
+			}
+		case "DTPOSTED":
+			if cur != nil {
+				cur.Date = parseOFXDate(value)
+			}
+		case "TRNAMT":
+			if cur != nil {
+				amt, _ := strconv.ParseFloat(strings.ReplaceAll(value, ",", "."), 64)
+				cur.Amount = math.Abs(amt)
+				if amt < 0 {
+					cur.Type = "expense"
+				} else {
+					cur.Type = "income"
+				}
+			}
+		case "FITID":
+			if cur != nil {
+				cur.ExternalID = value
+			}
+		case "MEMO":
+			if cur != nil && cur.Description == "" {
+				cur.Description = cleanDescription(value)
+			}
+		case "NAME":
+			if cur != nil {
+				cur.Description = cleanDescription(value)
+			}
+		}
+	}
+	if cur != nil {
+		rows = append(rows, *cur)
+	}
+	return rows, scanner.Err()
+}
+
+// splitOFXLine handles both "<TAG>value" and "TAG:value" formats.
+func splitOFXLine(line string) (tag, value string) {
+	if strings.HasPrefix(line, "<") {
+		end := strings.Index(line, ">")
+		if end < 0 {
+			return "", ""
+		}
+		tag = strings.ToUpper(line[1:end])
+		value = strings.TrimSpace(line[end+1:])
+		return
+	}
+	// header lines like "OFXHEADER:100" — ignore
+	return "", ""
+}
+
+// parseOFXDate handles formats: 20060102, 20060102120000, 20060102120000[-3:BRT]
+func parseOFXDate(s string) string {
+	s = strings.SplitN(s, "[", 2)[0] // strip timezone suffix
+	s = strings.TrimSpace(s)
+	formats := []string{"20060102150405", "20060102"}
+	for _, f := range formats {
+		if len(s) >= len(f) {
+			t, err := time.Parse(f, s[:len(f)])
+			if err == nil {
+				return t.Format("2006-01-02")
+			}
+		}
+	}
+	return s
+}
+
+func cleanDescription(s string) string {
+	s = strings.TrimSpace(s)
+	// remove duplicate whitespace
+	fields := strings.Fields(s)
+	return strings.Join(fields, " ")
+}