Ré-écriture de l'algo

csv-parser.go

@@ -1,95 +0,0 @@
-package csvparser
-
-import (
-	"strings"
-	"errors"
-)
-
-type EncloserId int
-
-const (
-	None EncloserId = iota
-	DoubleQuotes
-	SingleQuotes
-	RoundBrackets
-	SquareBrackets
-	CurlyBrackets
-)
-
-type Encloser struct {
-	Open  byte
-	Close byte
-}
-
-var EnclosersRunes = map[EncloserId]Encloser{
-	DoubleQuotes:   Encloser{'"', '"'},
-	SingleQuotes:   Encloser{'\'', '\''},
-	RoundBrackets:  Encloser{'(', ')'},
-	SquareBrackets: Encloser{'[', ']'},
-	CurlyBrackets:  Encloser{'{', '}'},
-}
-
-type Parser struct {
-	Enclosers       []EncloserId
-	Delimiter       string
-	Fields		[]string
-	Line		string
-}
-
-func (parser *Parser) Init() {
-	parser.Enclosers = []EncloserId{DoubleQuotes, SquareBrackets}
-	parser.Delimiter = " \t"
-}
-
-func (parser *Parser) ExtractEnclosedFieldValue(endChar byte) error {
-	lineLen := len(parser.Line)
-	for i := 1; i < lineLen; i++ {
-		if parser.Line[i] == endChar && parser.Line[i-1] != '\\' {
-			parser.Fields = append(parser.Fields, parser.Line[0:i])
-      parser.Line = parser.Line[i+1:]
-			return nil
-		}
-	}
-	return errors.New("Encloser close not found")
-}
-
-func (parser *Parser) Parse(CsvLine string) error {
-	var err error = nil
-
-  parser.Fields = make([]string,0)
-	parser.Line = CsvLine
-	for len(parser.Line) > 0 {
-		parser.Line = strings.TrimLeft(parser.Line, parser.Delimiter)
-		parser.Line = strings.TrimRight(parser.Line, parser.Delimiter)
-		if len(parser.Line) == 0 {
-			break
-		}
-		// Search for an encloser
-		encloserId := None
-		for _, id := range parser.Enclosers {
-			if parser.Line[0] == EnclosersRunes[id].Open {
-				encloserId = id
-				break
-			}
-		}
-		if encloserId != None {
-      parser.Line = parser.Line[1:]
-			err = parser.ExtractEnclosedFieldValue(EnclosersRunes[encloserId].Close)
-			if err != nil {
-				return err
-			}
-		} else {
-			nextSpace := strings.IndexAny(parser.Line, parser.Delimiter)
-			if nextSpace != -1 {
-				parser.Fields = append(parser.Fields, parser.Line[:nextSpace])
-				parser.Line = parser.Line[nextSpace:]
-			} else {
-				parser.Fields = append(parser.Fields, parser.Line)
-				parser.Line = ""
-				break
-			}
-
-		}
-	}
-	return nil
-}

csv-parser_test.go

@@ -1,68 +0,0 @@
-package csvparser
-
-import(
-	"testing"
-)
-
-
-func TestParse(t *testing.T) {
-	var parser Parser
-
-  var CsvTestValues = [...]string {
-    "field1 field2 field3", // standard CSV
-    " field1   field2 field3 ", // Space or multiple spaces as delimiters
-    "   field1 field2 field3 ", // Spaces + tabs as delimiters
-    "   \"field1\" field2 field3 ", // Enclosed fields
-    "   \"field1\" field2 [field3] ", // Enclosed fields
-  }
-  var CsvExpectedValues = [...]string {
-    "field1",
-    "field2",
-    "field3",
-  }
-
-  parser.Init()
-  for _, v := range CsvTestValues {
-    err := parser.Parse(v)
-    if err != nil {
-      t.Error("Parse error:", err, " in ", v)
-    }
-    if len(parser.Fields) != len(CsvExpectedValues) {
-      t.Error("Extracted field number does not match expected", parser.Fields)
-    }
-    for i,val := range CsvExpectedValues {
-      if parser.Fields[i] != val {
-        t.Error("Field values do not match", i, " expected ", val, " got ", parser.Fields[i])
-      }
-    }
-  }
-}
-func TestParseWithEscape(t *testing.T) {
-	var parser Parser
-
-  var CsvTestValues = [...]string {
-    "\"\\\"field1 and more\" field2 [\\[field3] ", // Enclosed fields
-  }
-  var CsvExpectedValues = [...]string {
-    "\\\"field1 and more",
-    "field2",
-    "\\[field3",
-  }
-
-
-  parser.Init()
-  for _, v := range CsvTestValues {
-    err := parser.Parse(v)
-    if err != nil {
-      t.Error("Parse error:", err, " in ", v)
-    }
-    if len(parser.Fields) != len(CsvExpectedValues) {
-      t.Error("Extracted field number does not match expected", parser.Fields)
-    }
-    for i,val := range CsvExpectedValues {
-      if parser.Fields[i] != val {
-        t.Error("Field values do not match", i, " expected ", val, " got ", parser.Fields[i])
-      }
-    }
-  }
-}

csvparser.go

@@ -0,0 +1,91 @@
+package csvparser
+
+import (
+	"fmt"
+	"strings"
+	"unicode/utf8"
+)
+
+type CsvParser struct {
+	enclosers []string
+	delimiter rune
+	fields    []string
+}
+
+func (p *CsvParser) Initialize(delimiter string, enclosers []string, lineFormat string) error {
+	if utf8.RuneCountInString(delimiter) != 1 {
+		return fmt.Errorf("delimiter shoud be one character")
+	}
+	p.enclosers = make([]string, 0)
+	for _, encloser := range enclosers {
+		if utf8.RuneCountInString(encloser) != 2 {
+			return fmt.Errorf("encolser should have to characters")
+		}
+	}
+	p.enclosers = enclosers
+	p.delimiter = []rune(delimiter)[0]
+	for _, pair := range enclosers {
+		if utf8.RuneCountInString(pair) != 2 {
+			return fmt.Errorf("encoloser should contain two characters: %s", pair)
+		}
+	}
+	// line format is in the form of: field1 field2 ignore ...
+	// if field name is ignore, it is parsed but not retained
+	p.fields = strings.Split(lineFormat, " ")
+
+	return nil
+}
+
+func (p *CsvParser) Parse(line string) (map[string]string, error) {
+	currentFieldIndex := 0
+	valueStart := 0
+	escape := false
+  enclosed := false
+
+	ret := make(map[string]string)
+	indexMax := len(line) - 1
+  maxFieldIndex := len(p.fields) - 1
+  delimiter := p.delimiter
+	for index, r := range line {
+    if r == '\\' {
+      // Check if EOL before continue
+      escape=true
+      continue
+    }
+    if escape {
+      escape=false
+      continue
+    }
+    if r == delimiter {
+      ret[p.fields[currentFieldIndex]] = line[valueStart : index]
+      currentFieldIndex++
+      if currentFieldIndex > maxFieldIndex {
+        break
+      }
+      valueStart = index + 1
+      if enclosed {
+        enclosed=false
+        delimiter = p.delimiter
+        // Omit next delimiter
+        escape=true
+      }
+      continue
+    }
+    if index >= indexMax {
+			ret[p.fields[currentFieldIndex]] = line[valueStart:]
+      continue
+    }
+
+    for _, encloser := range p.enclosers {
+      runes := []rune(encloser)
+      if r == runes[0] {
+        // opening encloser
+        enclosed = true
+        delimiter = runes[1]
+        valueStart++
+        break
+      }
+    }
+  }
+	return ret, nil
+}

csvparser_test.go

@@ -0,0 +1,66 @@
+package csvparser
+
+import (
+	"fmt"
+	"testing"
+)
+
+func TestCorrectLines(t *testing.T) {
+	var csvParser CsvParser
+	csvParser.Initialize(" ", []string{"\"\"", "[]"}, "firstname lastname complete_name")
+	line := ""
+	fmt.Println("parsing:", line)
+	event, err := csvParser.Parse(line)
+	if err != nil {
+		t.Fatalf("Parsing of empty line failed %v %v", err, event)
+	}
+	fmt.Println("event is", event)
+
+	line = "John"
+	fmt.Println("parsing:", line)
+	event, err = csvParser.Parse(line)
+	if err != nil {
+		t.Fatalf("Parsing of empty line failed %v %v", err, event)
+	}
+	fmt.Println("event is", event)
+
+	line = "John Doe"
+	fmt.Println("parsing:", line)
+	event, err = csvParser.Parse(line)
+	if err != nil {
+		t.Fatalf("Parsing of empty line failed %v %v", err, event)
+	}
+	fmt.Println("event is", event)
+
+	line = "John \"John Doe\" Doe"
+	fmt.Println("parsing:", line)
+	event, err = csvParser.Parse(line)
+	if err != nil {
+		t.Fatalf("Parsing of empty line failed %v %v", err, event)
+	}
+	fmt.Println("event is", event)
+
+	line = "John Doe \"John Doe\""
+	fmt.Println("parsing:", line)
+	event, err = csvParser.Parse(line)
+	if err != nil {
+		t.Fatalf("Parsing of empty line failed %v %v", err, event)
+	}
+	fmt.Println("event is", event)
+
+	line = "John Doe \"John Doe\"\\\\"
+	fmt.Println("parsing:", line)
+	event, err = csvParser.Parse(line)
+	if err != nil {
+		t.Fatalf("Parsing of empty line failed %v %v", err, event)
+	}
+	fmt.Println("event is", event)
+
+	line = "John Doe I don't know him"
+	fmt.Println("parsing:", line)
+	event, err = csvParser.Parse(line)
+	if err != nil {
+		t.Fatalf("Parsing of empty line failed %v %v", err, event)
+	}
+	fmt.Println("event is", event)
+}

go.mod

@@ -1,3 +1,4 @@
-module git.orange.fruit.ovh/laurentu/csv-parser
+module git.passke.org/laurentu/csv-parser
 
-go 1.17
+toolchain go1.22.1
+go 1.22