Implement and test byte-splitting helper function

bridge/helper/helper.go

@@ -219,6 +219,33 @@ func ClipMessage(text string, length int, clippingMessage string) string {
 	return text
 }
 
+func ClipOrSplitMessage(text string, length int, clippingMessage string, splitMax int) []string {
+	var msgParts []string
+	var remainingText = text
+	// Invariant of this splitting loop: No text is lost (msgParts+remainingText is the original text),
+	// and all parts is guaranteed to satisfy the length requirement.
+	for len(msgParts) < splitMax - 1 && len(remainingText) > length {
+		// Decision: The text needs to be split (again).
+		var chunk string
+		var wasted = 0
+		// The longest UTF-8 encoding of a valid rune is 4 bytes (0xF4 0x8F 0xBF 0xBF, encoding U+10FFFF),
+		// so we should never need to waste 4 or more bytes at a time.
+		for wasted < 4 && wasted < length {
+			chunk = remainingText[:length - wasted]
+			if r, _ := utf8.DecodeLastRuneInString(chunk); r == utf8.RuneError {
+				wasted += 1
+			} else {
+				break
+			}
+		}
+		// Note: At this point, "chunk" might still be invalid, if "text" is very broken.
+		msgParts = append(msgParts, chunk)
+		remainingText = remainingText[len(chunk):]
+	}
+	msgParts = append(msgParts, ClipMessage(remainingText, length, clippingMessage))
+	return msgParts
+}
+
 // ParseMarkdown takes in an input string as markdown and parses it to html
 func ParseMarkdown(input string) string {
 	extensions := parser.HardLineBreak | parser.NoIntraEmphasis | parser.FencedCode

bridge/helper/helper_test.go

@@ -125,3 +125,105 @@ func TestConvertWebPToPNG(t *testing.T) {
 		t.Fail()
 	}
 }
+
+var clippingOrSplittingTestCases = map[string]struct {
+	inputText       string
+	clipSplitLength int
+	clippingMessage string
+	splitMax        int
+	expectedOutput  []string
+}{
+	"Short single-line message, split 3": {
+		inputText:       "short",
+		clipSplitLength: 20,
+		clippingMessage: "?!?!",
+		splitMax:        3,
+		expectedOutput:  []string{"short"},
+	},
+	"Short single-line message, split 1": {
+		inputText:       "short",
+		clipSplitLength: 20,
+		clippingMessage: "?!?!",
+		splitMax:        1,
+		expectedOutput:  []string{"short"},
+	},
+	"Short single-line message, split 0": {
+		// Mainly check that we don't crash.
+		inputText:       "short",
+		clipSplitLength: 20,
+		clippingMessage: "?!?!",
+		splitMax:        0,
+		expectedOutput:  []string{"short"},
+	},
+	"Long single-line message, noclip": {
+		inputText:       "Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.",
+		clipSplitLength: 50,
+		clippingMessage: "?!?!",
+		splitMax:        10,
+		expectedOutput:  []string{
+			"Lorem ipsum dolor sit amet, consectetur adipiscing",
+			" elit, sed do eiusmod tempor incididunt ut labore ",
+			"et dolore magna aliqua.",
+		},
+	},
+	"Long single-line message, noclip tight": {
+		inputText:       "Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.",
+		clipSplitLength: 50,
+		clippingMessage: "?!?!",
+		splitMax:        3,
+		expectedOutput:  []string{
+			"Lorem ipsum dolor sit amet, consectetur adipiscing",
+			" elit, sed do eiusmod tempor incididunt ut labore ",
+			"et dolore magna aliqua.",
+		},
+	},
+	"Long single-line message, clip custom": {
+		inputText:       "Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.",
+		clipSplitLength: 50,
+		clippingMessage: "?!?!",
+		splitMax:        2,
+		expectedOutput:  []string{
+			"Lorem ipsum dolor sit amet, consectetur adipiscing",
+			" elit, sed do eiusmod tempor incididunt ut lab?!?!",
+		},
+	},
+	"Long single-line message, clip built-in": {
+		inputText:       "Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.",
+		clipSplitLength: 50,
+		clippingMessage: "",
+		splitMax:        2,
+		expectedOutput:  []string{
+			"Lorem ipsum dolor sit amet, consectetur adipiscing",
+			" elit, sed do eiusmod tempor inc <clipped message>",
+		},
+	},
+	"Short multi-line message": {
+		inputText:       "I\ncan't\nget\nno\nsatisfaction!",
+		clipSplitLength: 50,
+		clippingMessage: "",
+		splitMax:        2,
+		expectedOutput:  []string{"I\ncan't\nget\nno\nsatisfaction!"},
+	},
+	"Long message containing UTF-8 multi-byte runes": {
+		inputText:       "人人生而自由，在尊嚴和權利上一律平等。 他們都具有理性和良知，應該以兄弟情誼的精神對待彼此。",
+		clipSplitLength: 50,
+		clippingMessage: "",
+		splitMax:        10,
+		expectedOutput:  []string{
+			"人人生而自由，在尊嚴和權利上一律", // Note: only 48 bytes!
+			"平等。 他們都具有理性和良知，應該", // Note: only 49 bytes!
+			"以兄弟情誼的精神對待彼此。",
+		},
+	},
+}
+
+func TestClipOrSplitMessage(t *testing.T) {
+	for testname, testcase := range clippingOrSplittingTestCases {
+		actualOutput := ClipOrSplitMessage(testcase.inputText, testcase.clipSplitLength, testcase.clippingMessage, testcase.splitMax)
+		assert.Equalf(t, testcase.expectedOutput, actualOutput, "'%s' testcase should give expected lines with clipping+splitting.", testname)
+		for _, splitLine := range testcase.expectedOutput {
+			byteLength := len([]byte(splitLine))
+			assert.True(t, byteLength <= testcase.clipSplitLength, "Splitted line '%s' of testcase '%s' should not exceed the maximum byte-length (%d vs. %d).", splitLine, testname, testcase.clipSplitLength, byteLength)
+		}
+	}
+}