new action: countLinesWithAt (both txt/csv & parquet) (#5)

Signed-off-by: Hadi <112569860+anotherhadi@users.noreply.github.com>

leak-utils/leak-utils/main.go

@@ -33,6 +33,7 @@ func main() {
 		"mergeFiles",
 		"deleteFirstLines",
 		"deleteLastLines",
+		"countLinesWithAt",
 		"removeUrlSchemeFromUlp",
 	}
 
@@ -162,6 +163,24 @@ func main() {
 			log.Fatal("Failed to remove last lines", "error", err)
 		}
 		return
+	case "countLinesWithAt":
+		var inputFile *string = flag.StringP("input", "i", "", "Input file")
+		var noColors *bool = flag.Bool("no-colors", false, "Remove all colors")
+		var debug *bool = flag.Bool("debug", false, "Debug mode")
+		flag.Parse()
+		if *inputFile == "" {
+			log.Fatal("Input file are required")
+		}
+		if *noColors {
+			settings.DisableColors()
+		}
+		lu.Debug = *debug
+		countAt, countLine, err := misc.CountLinesWithAt(lu, *inputFile)
+		if err != nil {
+			log.Fatal("Failed to count @", "error", err)
+		}
+		fmt.Println(settings.Base.Render("There are"), settings.Accent.Render(fmt.Sprintf("%d", countAt)), settings.Base.Render("lines with @ out of"), settings.Accent.Render(fmt.Sprintf("%d", countLine)), settings.Base.Render("lines in"), settings.Accent.Render(*inputFile))
+		return
 	case "removeUrlSchemeFromUlp":
 		var inputFile *string = flag.StringP("input", "i", "", "Input Parquet file")
 		var noColors *bool = flag.Bool("no-colors", false, "Remove all colors")

leak-utils/misc/countLinesWithAt.go

@@ -0,0 +1,73 @@
+package misc
+
+import (
+	"bufio"
+	"fmt"
+	"os"
+	"strings"
+
+	"github.com/anotherhadi/eleakxir/leak-utils/parquet"
+	"github.com/anotherhadi/eleakxir/leak-utils/settings"
+)
+
+// Count the line with "@" in a file
+func CountLinesWithAt(lu settings.LeakUtils, inputFile string) (nAt, nLines int, err error) {
+	if strings.HasSuffix(inputFile, ".parquet") {
+		return countRowsWithAtInParquet(lu, inputFile)
+	}
+
+	in, err := os.Open(inputFile)
+	if err != nil {
+		return 0, 0, err
+	}
+	defer in.Close()
+
+	scanner := bufio.NewScanner(in)
+	countAt := 0
+	countLine := 0
+	for scanner.Scan() {
+		line := scanner.Text()
+		if strings.Contains(line, "@") {
+			countAt++
+		}
+		countLine++
+	}
+
+	if err := scanner.Err(); err != nil {
+		return 0, 0, err
+	}
+
+	return countAt, countLine, nil
+}
+
+func countRowsWithAtInParquet(lu settings.LeakUtils, inputFile string) (nAt, nLine int, err error) {
+	cols, err := parquet.GetColumns(lu.Db, inputFile)
+	if err != nil {
+		return 0, 0, err
+	}
+	if len(cols) == 0 {
+		return 0, 0, nil
+	}
+
+	whereParts := []string{}
+	for _, col := range cols {
+		whereParts = append(whereParts, fmt.Sprintf("%s LIKE '%%@%%'", col))
+	}
+	whereClause := strings.Join(whereParts, " OR ")
+
+	query := fmt.Sprintf("SELECT COUNT(*) FROM read_parquet('%s') WHERE %s", inputFile, whereClause)
+	var countAt int
+	err = lu.Db.QueryRow(query).Scan(&countAt)
+	if err != nil {
+		return 0, 0, err
+	}
+
+	query = fmt.Sprintf("SELECT COUNT(*) FROM read_parquet('%s')", inputFile)
+	var countLine int
+	err = lu.Db.QueryRow(query).Scan(&countLine)
+	if err != nil {
+		return 0, 0, err
+	}
+
+	return countAt, countLine, nil
+}

leak-utils/parquet/parquet.go

@@ -252,7 +252,7 @@ func GetParquet(db *sql.DB, inputFile string) (parquet *Parquet, err error) {
 	parquet = &Parquet{}
 	parquet.Filepath = inputFile
 
-	parquet.Columns, err = getColumns(db, inputFile)
+	parquet.Columns, err = GetColumns(db, inputFile)
 	if err != nil {
 		return
 	}

leak-utils/parquet/utils.go

@@ -7,8 +7,8 @@ import (
 	"strings"
 )
 
-// getColumns retrieves the column names from the Parquet file.
+// GetColumns retrieves the column names from the Parquet file.
-func getColumns(db *sql.DB, filepath string) ([]string, error) {
+func GetColumns(db *sql.DB, filepath string) ([]string, error) {
 	// Create a view from the parquet file
 	query := fmt.Sprintf("CREATE OR REPLACE VIEW parquet_view AS SELECT * FROM read_parquet('%s')", filepath)
 	_, err := db.Exec(query)