update workflows and code format

sunhailin-Leo · Aug 29, 2023 · 1951a3f · 1951a3f
1 parent ae1ee97
commit 1951a3f
Show file tree

Hide file tree

Showing 7 changed files with 91 additions and 30 deletions.
diff --git a/.github/workflows/lint.yml b/.github/workflows/lint.yml
@@ -11,7 +11,7 @@ jobs:
       - uses: actions/checkout@v3
       - uses: actions/setup-go@v4
         with:
-          go-version: 1.20.x
+          go-version: 1.21.x
       - run: go version
       #- run: diff -u <(echo -n) <(gofmt -d .)
       - name: Run golangci-lint

diff --git a/models/bert/model.go b/models/bert/model.go
@@ -215,12 +215,12 @@ func (m *ModelService) generateHTTPOutputs(
 	for i := range inferOutputs {
 		requestOutputs[i] = HTTPOutput{Name: inferOutputs[i].Name}
 		if _, ok := inferOutputs[i].Parameters[ModelRespBodyOutputBinaryDataKey]; ok {
-			requestOutputs[i].Parameters.BinaryData =
-				inferOutputs[i].Parameters[ModelRespBodyOutputBinaryDataKey].GetBoolParam()
+			requestOutputs[i].Parameters.BinaryData = inferOutputs[i].Parameters[ModelRespBodyOutputBinaryDataKey].
+				GetBoolParam()
 		}
 		if _, ok := inferOutputs[i].Parameters[ModelRespBodyOutputClassificationDataKey]; ok {
-			requestOutputs[i].Parameters.Classification =
-				inferOutputs[i].Parameters[ModelRespBodyOutputClassificationDataKey].GetInt64Param()
+			requestOutputs[i].Parameters.Classification = inferOutputs[i].Parameters[ModelRespBodyOutputClassificationDataKey].
+				GetInt64Param()
 		}
 	}
 	return requestOutputs

diff --git a/models/bert/tokenizer.go b/models/bert/tokenizer.go
@@ -180,12 +180,16 @@ func (t *BaseTokenizer) splitOn(text string, shouldSplit func(rune) bool, includ
 			wordLen := len(word)
 			if wordLen > 0 {
 				words = append(words, StringOffsetsPair{
-					String: string(word), Offsets: OffsetsType{Start: offset - wordLen, End: offset}})
+					String:  string(word),
+					Offsets: OffsetsType{Start: offset - wordLen, End: offset},
+				})
 				word = make([]rune, 0, cap(word))
 			}
 			if includeSplitToken {
 				words = append(words, StringOffsetsPair{
-					String: string(r), Offsets: OffsetsType{Start: offset, End: offset + 1}})
+					String:  string(r),
+					Offsets: OffsetsType{Start: offset, End: offset + 1},
+				})
 			}
 		} else {
 			word = append(word, r)
@@ -196,7 +200,9 @@ func (t *BaseTokenizer) splitOn(text string, shouldSplit func(rune) bool, includ
 	// Don't forget the potential last word
 	if len(word) > 0 {
 		words = append(words, StringOffsetsPair{
-			String: string(word), Offsets: OffsetsType{Start: offset - len(word), End: offset}})
+			String:  string(word),
+			Offsets: OffsetsType{Start: offset - len(word), End: offset},
+		})
 	}
 	return words
 }
@@ -213,12 +219,16 @@ func (t *BaseTokenizer) splitOnChinese(text string, shouldSplit func(rune) bool,
 			wordLen := len(word)
 			if wordLen > 0 {
 				words = append(words, StringOffsetsPair{
-					String: string(word), Offsets: OffsetsType{Start: offset - wordLen, End: offset}})
+					String:  string(word),
+					Offsets: OffsetsType{Start: offset - wordLen, End: offset},
+				})
 				word = make([]rune, 0, cap(word))
 			}
 			if includeSplitToken || includeSplitFunc(r) {
 				words = append(words, StringOffsetsPair{
-					String: string(r), Offsets: OffsetsType{Start: offset, End: offset + 1}})
+					String:  string(r),
+					Offsets: OffsetsType{Start: offset, End: offset + 1},
+				})
 			}
 		} else {
 			word = append(word, r)
@@ -229,7 +239,9 @@ func (t *BaseTokenizer) splitOnChinese(text string, shouldSplit func(rune) bool,
 	// Don't forget the potential last word
 	if len(word) > 0 {
 		words = append(words, StringOffsetsPair{
-			String: string(word), Offsets: OffsetsType{Start: offset - len(word), End: offset}})
+			String:  string(word),
+			Offsets: OffsetsType{Start: offset - len(word), End: offset},
+		})
 	}
 	return words
 }
@@ -275,7 +287,7 @@ func (t *WordPieceTokenizer) TokenizeChineseCharMode(text string) []StringOffset
 }
 
 // WordPieceTokenize
-//transforms the input token in a new slice of words or sub-words units based on the supplied vocabulary.
+// transforms the input token in a new slice of words or sub-words units based on the supplied vocabulary.
 // The resulting tokens preserve the alignment with the portion of the original text they belong to.
 func (t *WordPieceTokenizer) WordPieceTokenize(tokens []StringOffsetsPair) []StringOffsetsPair {
 	outputTokens := make([]StringOffsetsPair, 0)
@@ -308,7 +320,9 @@ func (t *WordPieceTokenizer) WordPieceTokenize(tokens []StringOffsetsPair) []Str
 					found = true
 					curStrToken.String = subStr
 					curStrToken.Offsets = OffsetsType{
-						Start: tokens[i].Offsets.Start + start, End: tokens[i].Offsets.Start + end}
+						Start: tokens[i].Offsets.Start + start,
+						End:   tokens[i].Offsets.Start + end,
+					}
 					break
 				}
 				end--

diff --git a/nvidia_inferenceserver/triton_service_interface.go b/nvidia_inferenceserver/triton_service_interface.go
@@ -525,7 +525,7 @@ func (t *TritonClientService) ModelInferStats(
 
 // ModelLoadWithHTTP Load Model with http
 // modelConfigBody ==>
-//https://github.com/triton-inference-server/server/blob/main/docs/protocol/extension_model_repository.md#examples
+// https://github.com/triton-inference-server/server/blob/main/docs/protocol/extension_model_repository.md#examples
 func (t *TritonClientService) ModelLoadWithHTTP(
 	modelName string, modelConfigBody []byte, timeout time.Duration,
 ) (*RepositoryModelLoadResponse, error) {
@@ -796,7 +796,7 @@ func (t *TritonClientService) GetModelTracingSetting(
 
 // SetModelTracingSetting set model tracing setting.
 // Param: settingMap ==>
-//https://github.com/triton-inference-server/server/blob/main/docs/protocol/extension_trace.md#trace-setting-response-json-object
+// https://github.com/triton-inference-server/server/blob/main/docs/protocol/extension_trace.md#trace-setting-response-json-object
 func (t *TritonClientService) SetModelTracingSetting(
 	modelName string, settingMap map[string]*TraceSettingRequest_SettingValue, timeout time.Duration,
 ) (*TraceSettingResponse, error) {

diff --git a/test/tokenizer_test.go b/test/tokenizer_test.go
@@ -15,8 +15,8 @@ func TestFullTokenizerNotChinese(t *testing.T) {
 	}
 	tokenizer := bert.NewWordPieceTokenizer(voc)
 	tokenResult := tokenizer.Tokenize("นครปฐม เมืองนครปฐม ถนนขาด เลขที่ 69 หมู่ 1 ซ. - - ถ. -")
-	var tokenStrSlice = make([]string, len(tokenResult))
-	var tokenOffsetSlice = make([]bert.OffsetsType, len(tokenResult))
+	tokenStrSlice := make([]string, len(tokenResult))
+	tokenOffsetSlice := make([]bert.OffsetsType, len(tokenResult))
 	for i, token := range tokenResult {
 		tokenStrSlice[i] = token.String
 		tokenOffsetSlice[i] = token.Offsets
@@ -31,19 +31,47 @@ func TestFullTokenizerNotChinese(t *testing.T) {
 	expectedTokenSlice := []string{
 		"น", "##คร", "##ป", "##ฐ", "##ม", "เ", "##มือง", "##น", "##คร", "##ป", "##ฐ", "##ม",
 		"ถ", "##น", "##น", "##ข", "##า", "##ด", "เ", "##ล", "##ข", "##ที่", "69", "ห", "##ม",
-		"##ู่", "1", "ซ", ".", "-", "-", "ถ", ".", "-"}
+		"##ู่", "1", "ซ", ".", "-", "-", "ถ", ".", "-",
+	}
 	if !reflect.DeepEqual(tokenStrSlice, expectedTokenSlice) {
 		t.Errorf("Expected '%v', but got '%v'", expectedTokenSlice, tokenStrSlice)
 	}
 
 	expectedTokenOffsetSlice := []bert.OffsetsType{
-		{Start: 0, End: 1}, {Start: 1, End: 3}, {Start: 3, End: 4}, {Start: 4, End: 5}, {Start: 5, End: 6},
-		{Start: 7, End: 8}, {Start: 8, End: 12}, {Start: 12, End: 13}, {Start: 13, End: 15}, {Start: 15, End: 16},
-		{Start: 16, End: 17}, {Start: 17, End: 18}, {Start: 19, End: 20}, {Start: 20, End: 21}, {Start: 21, End: 22},
-		{Start: 22, End: 23}, {Start: 23, End: 24}, {Start: 24, End: 25}, {Start: 26, End: 27}, {Start: 27, End: 28},
-		{Start: 28, End: 29}, {Start: 29, End: 32}, {Start: 33, End: 35}, {Start: 36, End: 37}, {Start: 37, End: 38},
-		{Start: 38, End: 40}, {Start: 41, End: 42}, {Start: 43, End: 44}, {Start: 44, End: 45}, {Start: 46, End: 47},
-		{Start: 48, End: 49}, {Start: 50, End: 51}, {Start: 51, End: 52}, {Start: 53, End: 54},
+		{Start: 0, End: 1},
+		{Start: 1, End: 3},
+		{Start: 3, End: 4},
+		{Start: 4, End: 5},
+		{Start: 5, End: 6},
+		{Start: 7, End: 8},
+		{Start: 8, End: 12},
+		{Start: 12, End: 13},
+		{Start: 13, End: 15},
+		{Start: 15, End: 16},
+		{Start: 16, End: 17},
+		{Start: 17, End: 18},
+		{Start: 19, End: 20},
+		{Start: 20, End: 21},
+		{Start: 21, End: 22},
+		{Start: 22, End: 23},
+		{Start: 23, End: 24},
+		{Start: 24, End: 25},
+		{Start: 26, End: 27},
+		{Start: 27, End: 28},
+		{Start: 28, End: 29},
+		{Start: 29, End: 32},
+		{Start: 33, End: 35},
+		{Start: 36, End: 37},
+		{Start: 37, End: 38},
+		{Start: 38, End: 40},
+		{Start: 41, End: 42},
+		{Start: 43, End: 44},
+		{Start: 44, End: 45},
+		{Start: 46, End: 47},
+		{Start: 48, End: 49},
+		{Start: 50, End: 51},
+		{Start: 51, End: 52},
+		{Start: 53, End: 54},
 	}
 	if !reflect.DeepEqual(tokenOffsetSlice, expectedTokenOffsetSlice) {
 		t.Errorf("Expected '%v', but got '%v'", expectedTokenOffsetSlice, tokenOffsetSlice)
@@ -72,8 +100,8 @@ func TestFullTokenizerChinese(t *testing.T) {
 	}
 	tokenizer := bert.NewWordPieceTokenizer(voc)
 	tokenResult := tokenizer.TokenizeChinese(strings.ToLower("广东省深圳市南山区腾讯大厦"))
-	var tokenStrSlice = make([]string, len(tokenResult))
-	var tokenOffsetSlice = make([]bert.OffsetsType, len(tokenResult))
+	tokenStrSlice := make([]string, len(tokenResult))
+	tokenOffsetSlice := make([]bert.OffsetsType, len(tokenResult))
 	for i, token := range tokenResult {
 		tokenStrSlice[i] = token.String
 		tokenOffsetSlice[i] = token.Offsets
@@ -91,9 +119,19 @@ func TestFullTokenizerChinese(t *testing.T) {
 	}
 
 	expectedTokenOffsetSlice := []bert.OffsetsType{
-		{Start: 0, End: 1}, {Start: 1, End: 2}, {Start: 2, End: 3}, {Start: 3, End: 4}, {Start: 4, End: 5},
-		{Start: 5, End: 6}, {Start: 6, End: 7}, {Start: 7, End: 8}, {Start: 8, End: 9}, {Start: 9, End: 10},
-		{Start: 10, End: 11}, {Start: 11, End: 12}, {Start: 12, End: 13},
+		{Start: 0, End: 1},
+		{Start: 1, End: 2},
+		{Start: 2, End: 3},
+		{Start: 3, End: 4},
+		{Start: 4, End: 5},
+		{Start: 5, End: 6},
+		{Start: 6, End: 7},
+		{Start: 7, End: 8},
+		{Start: 8, End: 9},
+		{Start: 9, End: 10},
+		{Start: 10, End: 11},
+		{Start: 11, End: 12},
+		{Start: 12, End: 13},
 	}
 	if !reflect.DeepEqual(tokenOffsetSlice, expectedTokenOffsetSlice) {
 		t.Errorf("Expected '%v', but got '%v'", expectedTokenOffsetSlice, tokenOffsetSlice)

diff --git a/utils/funcs.go b/utils/funcs.go
@@ -10,6 +10,7 @@ import (
 )
 
 // IsWhitespace checks whether rune c is a BERT whitespace character.
+//
 //go:inline
 func IsWhitespace(c rune) bool {
 	if c <= 0xFF {
@@ -20,6 +21,7 @@ func IsWhitespace(c rune) bool {
 }
 
 // IsControl checks whether rune c is a BERT control character.
+//
 //go:inline
 func IsControl(c rune) bool {
 	switch c {
@@ -35,6 +37,7 @@ func IsControl(c rune) bool {
 }
 
 // IsPunctuation checks whether rune c is a BERT punctuation character.
+//
 //go:inline
 func IsPunctuation(c rune) bool {
 	// return unicode.In(c, utils.Bp, unicode.P)
@@ -43,25 +46,29 @@ func IsPunctuation(c rune) bool {
 }
 
 // IsChinese validates that rune c is in the CJK range according to BERT spec.
+//
 //go:inline
 func IsChinese(c rune) bool {
 	// unicode.Is(unicode.Han, c)
 	return unicode.In(c, BertChineseChar, unicode.P)
 }
 
 // IsChineseOrNumber validates that rune c is in the CJK range according to BERT spec or Number.
+//
 //go:inline
 func IsChineseOrNumber(c rune) bool {
 	return unicode.In(c, BertChineseChar, unicode.P) || unicode.IsNumber(c)
 }
 
 // IsWhiteSpaceOrChinese validates that rune c is whitespace or is Chinese.
+//
 //go:inline
 func IsWhiteSpaceOrChinese(c rune) bool {
 	return IsWhitespace(c) || IsChinese(c)
 }
 
 // IsWhiteSpaceOrChineseOrNumber validates that rune c is whitespace or is Chinese or is Number.
+//
 //go:inline
 func IsWhiteSpaceOrChineseOrNumber(c rune) bool {
 	return IsWhitespace(c) || IsChineseOrNumber(c)

diff --git a/utils/times.go b/utils/times.go
@@ -3,12 +3,14 @@ package utils
 import "time"
 
 // GetNanoTimeFromSys get nano timestamp.
+//
 //go:inline
 func GetNanoTimeFromSys() int64 {
 	return time.Now().UnixNano()
 }
 
 // CalTimeGapWithNS get nano timestamp gap.
+//
 //go:inline
 func CalTimeGapWithNS(begin int64) int64 {
 	return GetNanoTimeFromSys() - begin