put pair parsing into internal pkg, add some more unit tests, update fmt specifier for errs

Author: dpaasman00

internal/coreinternal/parseutils/doc.go

@@ -0,0 +1,4 @@
+// Copyright The OpenTelemetry Authors
+// SPDX-License-Identifier: Apache-2.0
+
+package parseutils // import "github.com/open-telemetry/opentelemetry-collector-contrib/internal/coreinternal/parseutils"

internal/coreinternal/parseutils/package_test.go

@@ -0,0 +1,14 @@
+// Copyright The OpenTelemetry Authors
+// SPDX-License-Identifier: Apache-2.0
+
+package parseutils
+
+import (
+	"testing"
+
+	"go.uber.org/goleak"
+)
+
+func TestMain(m *testing.M) {
+	goleak.VerifyTestMain(m)
+}

internal/coreinternal/parseutils/parser.go

@@ -0,0 +1,49 @@
+// Copyright The OpenTelemetry Authors
+// SPDX-License-Identifier: Apache-2.0
+
+package parseutils
+
+import "fmt"
+
+// SplitString will split the input on the delimiter and return the resulting slice while respecting quotes. Outer quotes are stripped.
+// Use in place of `strings.Split` when quotes need to be respected.
+// Requires `delimiter` not be an empty string
+func SplitString(input, delimiter string) ([]string, error) {
+	var result []string
+	current := ""
+	delimiterLength := len(delimiter)
+	quoteChar := "" // "" means we are not in quotes
+
+	for i := 0; i < len(input); i++ {
+		if quoteChar == "" && i+delimiterLength <= len(input) && input[i:i+delimiterLength] == delimiter { // delimiter
+			if current == "" { // leading || trailing delimiter; ignore
+				i += delimiterLength - 1
+				continue
+			}
+			result = append(result, current)
+			current = ""
+			i += delimiterLength - 1
+			continue
+		}
+
+		if quoteChar == "" && (input[i] == '"' || input[i] == '\'') { // start of quote
+			quoteChar = string(input[i])
+			continue
+		}
+		if string(input[i]) == quoteChar { // end of quote
+			quoteChar = ""
+			continue
+		}
+
+		current += string(input[i])
+	}
+
+	if quoteChar != "" { // check for closed quotes
+		return nil, fmt.Errorf("never reached the end of a quoted value")
+	}
+	if current != "" { // avoid adding empty value bc of a trailing delimiter
+		return append(result, current), nil
+	}
+
+	return result, nil
+}

internal/coreinternal/parseutils/parser_test.go

@@ -0,0 +1,186 @@
+// Copyright The OpenTelemetry Authors
+// SPDX-License-Identifier: Apache-2.0
+
+package parseutils
+
+import (
+	"fmt"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func Test_SplitString(t *testing.T) {
+	testCases := []struct {
+		name        string
+		input       string
+		delimiter   string
+		expected    []string
+		expectedErr error
+	}{
+		{
+			name:      "simple",
+			input:     "a b c",
+			delimiter: " ",
+			expected: []string{
+				"a",
+				"b",
+				"c",
+			},
+		},
+		{
+			name:      "single quotes",
+			input:     "a 'b c d'",
+			delimiter: " ",
+			expected: []string{
+				"a",
+				"b c d",
+			},
+		},
+		{
+			name:      "double quotes",
+			input:     `a " b c " d`,
+			delimiter: " ",
+			expected: []string{
+				"a",
+				" b c ",
+				"d",
+			},
+		},
+		{
+			name:      "multi-char delimiter",
+			input:     "abc!@!  def !@! g",
+			delimiter: "!@!",
+			expected: []string{
+				"abc",
+				"  def ",
+				" g",
+			},
+		},
+		{
+			name:      "leading and trailing delimiters",
+			input:     "   name=ottl        func=key_value   hello=world ",
+			delimiter: " ",
+			expected: []string{
+				"name=ottl",
+				"func=key_value",
+				"hello=world",
+			},
+		},
+		{
+			name:      "embedded double quotes in single quoted value",
+			input:     `ab c='this is a "co ol" value'`,
+			delimiter: " ",
+			expected: []string{
+				"ab",
+				`c=this is a "co ol" value`,
+			},
+		},
+		{
+			name:      "embedded double quotes end single quoted value",
+			input:     `ab c='this is a "co ol"'`,
+			delimiter: " ",
+			expected: []string{
+				"ab",
+				`c=this is a "co ol"`,
+			},
+		},
+		{
+			name:      "quoted values include whitespace",
+			input:     `name="    ottl " func="  key_ value"`,
+			delimiter: " ",
+			expected: []string{
+				"name=    ottl ",
+				"func=  key_ value",
+			},
+		},
+		{
+			name:      "delimiter longer than input",
+			input:     "abc",
+			delimiter: "aaaa",
+			expected: []string{
+				"abc",
+			},
+		},
+		{
+			name:      "delimiter not found",
+			input:     "a b c",
+			delimiter: "!",
+			expected: []string{
+				"a b c",
+			},
+		},
+		{
+			name: "newlines in input",
+			input: `a
+b
+c`,
+			delimiter: " ",
+			expected: []string{
+				"a\nb\nc",
+			},
+		},
+		{
+			name: "newline delimiter",
+			input: `a b c
+d e f
+g 
+h`,
+			delimiter: "\n",
+			expected: []string{
+				"a b c",
+				"d e f",
+				"g ",
+				"h",
+			},
+		},
+		{
+			name:      "empty input",
+			input:     "",
+			delimiter: " ",
+			expected:  nil,
+		},
+		{
+			name:      "equal input and delimiter",
+			input:     "abc",
+			delimiter: "abc",
+			expected:  nil,
+		},
+		{
+			name:        "unclosed quotes",
+			input:       "a 'b c",
+			delimiter:   " ",
+			expectedErr: fmt.Errorf("never reached the end of a quoted value"),
+		},
+		{
+			name:        "mismatched quotes",
+			input:       `a 'b c' "d '`,
+			delimiter:   " ",
+			expectedErr: fmt.Errorf("never reached the end of a quoted value"),
+		},
+		{
+			name:      "tab delimiters",
+			input:     "a	b	c",
+			delimiter: "\t",
+			expected: []string{
+				"a",
+				"b",
+				"c",
+			},
+		},
+	}
+
+	for _, tc := range testCases {
+		t.Run(tc.name, func(t *testing.T) {
+			result, err := SplitString(tc.input, tc.delimiter)
+
+			if tc.expectedErr == nil {
+				assert.NoError(t, err)
+				assert.Equal(t, tc.expected, result)
+			} else {
+				assert.EqualError(t, err, tc.expectedErr.Error())
+				assert.Nil(t, result)
+			}
+		})
+	}
+}

pkg/ottl/ottlfuncs/README.md

@@ -847,7 +847,7 @@ Examples:
 
 The `ParseKeyValue` Converter returns a `pcommon.Map` that is a result of parsing the target string for key value pairs.
 
-`target` is a Getter that returns a string. `delimiter` is an optional string that is used to split the key and value in a pair, the default is `=`. `pair_delimiter` is an optional string that is used to split key value pairs, the default is white space.
+`target` is a Getter that returns a string. `delimiter` is an optional string that is used to split the key and value in a pair, the default is `=`. `pair_delimiter` is an optional string that is used to split key value pairs, the default is a single space (` `).
 
 For example, the following target `"k1=v1 k2=v2 k3=v3"` will use default delimiters and be parsed into the following map:
 ```

pkg/ottl/ottlfuncs/func_parse_key_value.go

@@ -8,6 +8,7 @@ import (
 	"fmt"
 	"strings"
 
+	"github.com/open-telemetry/opentelemetry-collector-contrib/internal/coreinternal/parseutils"
 	"github.com/open-telemetry/opentelemetry-collector-contrib/pkg/ottl"
 	"go.opentelemetry.io/collector/pdata/pcommon"
 )
@@ -34,17 +35,17 @@ func createParseKeyValueFunction[K any](_ ottl.FunctionContext, oArgs ottl.Argum
 
 func parseKeyValue[K any](target ottl.StringGetter[K], d ottl.Optional[string], p ottl.Optional[string]) (ottl.ExprFunc[K], error) {
 	delimiter := "="
-	if !d.IsEmpty() {
+	if !d.IsEmpty() && d.Get() != "" {
 		delimiter = d.Get()
 	}
 
 	pairDelimiter := " "
-	if !p.IsEmpty() {
+	if !p.IsEmpty() && p.Get() != "" {
 		pairDelimiter = p.Get()
 	}
 
 	if pairDelimiter == delimiter {
-		return nil, fmt.Errorf("pair delimiter \"%s\" cannot be equal to delimiter \"%s\"", pairDelimiter, delimiter)
+		return nil, fmt.Errorf("pair delimiter %q cannot be equal to delimiter %q", pairDelimiter, delimiter)
 	}
 
 	return func(ctx context.Context, tCtx K) (any, error) {
@@ -57,16 +58,16 @@ func parseKeyValue[K any](target ottl.StringGetter[K], d ottl.Optional[string],
 			return nil, fmt.Errorf("cannot parse from empty target")
 		}
 
-		pairs, err := splitPairs(source, pairDelimiter)
+		pairs, err := parseutils.SplitString(source, pairDelimiter)
 		if err != nil {
-			return nil, fmt.Errorf("splitting pairs failed: %w", err)
+			return nil, fmt.Errorf("splitting source %q into pairs failed: %w", source, err)
 		}
 
 		parsed := make(map[string]any)
 		for _, p := range pairs {
 			pair := strings.SplitN(p, delimiter, 2)
 			if len(pair) != 2 {
-				return nil, fmt.Errorf("cannot split '%s' into 2 items, got %d", p, len(pair))
+				return nil, fmt.Errorf("cannot split %q into 2 items, got %d item(s)", p, len(pair))
 			}
 			key := strings.TrimSpace(pair[0])
 			value := strings.TrimSpace(pair[1])
@@ -78,44 +79,3 @@ func parseKeyValue[K any](target ottl.StringGetter[K], d ottl.Optional[string],
 		return result, err
 	}, nil
 }
-
-// splitPairs will split the input on the pairDelimiter and return the resulting slice.
-// `strings.Split` is not used because it does not respect quotes and will split if the delimiter appears in a quoted value
-func splitPairs(input, pairDelimiter string) ([]string, error) {
-	var result []string
-	currentPair := ""
-	delimiterLength := len(pairDelimiter)
-	quoteChar := "" // "" means we are not in quotes
-
-	for i := 0; i < len(input); i++ {
-		if quoteChar == "" && i+delimiterLength <= len(input) && input[i:i+delimiterLength] == pairDelimiter { // delimiter
-			if currentPair == "" { // leading || trailing delimiter; ignore
-				continue
-			}
-			result = append(result, currentPair)
-			currentPair = ""
-			i += delimiterLength - 1
-			continue
-		}
-
-		if quoteChar == "" && (input[i] == '"' || input[i] == '\'') { // start of quote
-			quoteChar = string(input[i])
-			continue
-		}
-		if string(input[i]) == quoteChar { // end of quote
-			quoteChar = ""
-			continue
-		}
-
-		currentPair += string(input[i])
-	}
-
-	if quoteChar != "" { // check for closed quotes
-		return nil, fmt.Errorf("never reached end of a quoted value")
-	}
-	if currentPair != "" { // avoid adding empty value bc of a trailing delimiter
-		return append(result, currentPair), nil
-	}
-
-	return result, nil
-}