refactor: enhance extraction query handling; add support for JSON column extraction and wildcard patterns

internal/app/db-wrapper/mssql.go

@@ -6,8 +6,11 @@ import (
 	"fmt"
 	"strings"
 
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/config"
 	dbdialects "git.ksdemosapps.com/kylesoda/go-migrate/internal/app/db-wrapper/db_dialects"
+	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/models"
 	mssql "github.com/microsoft/go-mssqldb"
+	"github.com/sirupsen/logrus"
 )
 
 func init() {
@@ -177,25 +180,53 @@ func (mw *mssqlDbWrapper) SaveMassive(ctx context.Context, schema string, table
 	return rowsAffected, nil
 }
 
-func (mw *mssqlDbWrapper) QueryFromObject(ctx context.Context, q ExtractionQuery) (RowsResult, error) {
+func buildExtractQueryMssql(q ExtractionQuery) (string, error) {
 	var sbQuery strings.Builder
 
 	sbQuery.WriteString("SELECT ")
 
-	if len(q.Columns) == 0 {
-		sbQuery.WriteString("*")
-	} else {
-		for i, col := range q.Columns {
-			fmt.Fprintf(&sbQuery, "[%s]", col.Name())
+	hasRegularColumns := len(q.Columns) > 0
+	hasJsonColumns := len(q.FromJsonColumns) > 0
 
+	resolvedJson := make(map[string][]config.FromJsonItem, len(q.FromJsonColumns))
+	if hasJsonColumns {
+		for _, jsonConfig := range q.FromJsonColumns {
+			actualColumnName, err := findColumnByPattern(q.Columns, jsonConfig.Column)
+			if err != nil {
+				return "", err
+			}
+			resolvedJson[actualColumnName] = append(resolvedJson[actualColumnName], jsonConfig)
+		}
+	}
+
+	selectParts := make([]string, 0, len(q.Columns)+len(q.FromJsonColumns))
+	if hasRegularColumns {
+		for _, col := range q.Columns {
+			jsonConfigs, isJsonColumn := resolvedJson[col.Name()]
+			if isJsonColumn {
+				for _, jsonConfig := range jsonConfigs {
+					jsonPath := buildJsonPathMssql(jsonConfig.Field)
+					jsonExpr := fmt.Sprintf("JSON_VALUE([%s], '%s') AS [%s]", col.Name(), jsonPath, col.Name())
+					selectParts = append(selectParts, jsonExpr)
+				}
+				continue
+			}
+
+			colExpr := fmt.Sprintf("[%s]", col.Name())
 			switch col.Type() {
 			case "GEOMETRY":
-				fmt.Fprintf(&sbQuery, ".STAsBinary() AS [%s]", col.Name())
+				colExpr = fmt.Sprintf("[%s].STAsBinary() AS [%s]", col.Name(), col.Name())
 			}
+			selectParts = append(selectParts, colExpr)
+		}
+	} else if !hasJsonColumns {
+		selectParts = append(selectParts, "*")
+	}
 
-			if i < len(q.Columns)-1 {
-				sbQuery.WriteString(", ")
-			}
+	for i, part := range selectParts {
+		sbQuery.WriteString(part)
+		if i < len(selectParts)-1 {
+			sbQuery.WriteString(", ")
 		}
 	}
 
@@ -231,9 +262,39 @@ func (mw *mssqlDbWrapper) QueryFromObject(ctx context.Context, q ExtractionQuery
 
 	fmt.Fprintf(&sbQuery, " ORDER BY [%s] ASC", q.PrimaryKey)
 
-	queryString := sbQuery.String()
+	return sbQuery.String(), nil
+}
 
-	// logrus.Debugf("Query: %s", queryString)
+func findColumnByPattern(columns []models.ColumnType, pattern string) (string, error) {
+	if pattern == "" {
+		return "", fmt.Errorf("column pattern cannot be empty")
+	}
+
+	if before, ok := strings.CutSuffix(pattern, "*"); ok {
+		prefix := before
+		for _, col := range columns {
+			if strings.HasPrefix(col.Name(), prefix) {
+				return col.Name(), nil
+			}
+		}
+		return "", fmt.Errorf("no column found matching pattern '%s'", pattern)
+	}
+
+	for _, col := range columns {
+		if col.Name() == pattern {
+			return col.Name(), nil
+		}
+	}
+	return "", fmt.Errorf("column '%s' not found in table columns", pattern)
+}
+
+func (mw *mssqlDbWrapper) QueryFromObject(ctx context.Context, q ExtractionQuery) (RowsResult, error) {
+	queryString, err := buildExtractQueryMssql(q)
+	if err != nil {
+		return nil, err
+	}
+
+	logrus.Debugf("Query: %s", queryString)
 
 	var queryArgs []any
 
@@ -247,3 +308,11 @@ func (mw *mssqlDbWrapper) QueryFromObject(ctx context.Context, q ExtractionQuery
 
 	return mw.Query(ctx, queryString, queryArgs...)
 }
+
+func buildJsonPathMssql(field string) string {
+	if len(field) > 0 && field[0] == '.' {
+		field = field[1:]
+	}
+
+	return "$." + field
+}