Refactor Prerecorded API to Abstract REST Internals Away From User

deepgram · Oct 25, 2023 · 4d28e39 · 4d28e39
1 parent 39281f8
commit 4d28e39
Show file tree

Hide file tree

Showing 29 changed files with 1,571 additions and 423 deletions.
diff --git a/examples/prerecorded/main.go b/examples/prerecorded/main.go
@@ -5,13 +5,17 @@
 package main
 
 import (
+	"context"
 	"encoding/json"
 	"fmt"
 	"log"
 	"os"
 	"strings"
 
-	api "github.com/deepgram-devs/deepgram-go-sdk/pkg/api/prerecorded/v1"
+	prettyjson "github.com/hokaccha/go-prettyjson"
+
+	prerecorded "github.com/deepgram-devs/deepgram-go-sdk/pkg/api/prerecorded/v1"
+	interfaces "github.com/deepgram-devs/deepgram-go-sdk/pkg/client/interfaces"
 	client "github.com/deepgram-devs/deepgram-go-sdk/pkg/client/prerecorded"
 )
 
@@ -25,18 +29,21 @@ func main() {
 		os.Exit(1)
 	}
 
-	dg := client.New(deepgramApiKey)
+	// context
+	ctx := context.Background()
 
-	prClient := api.New(dg)
+	c := client.New(deepgramApiKey)
+	dg := prerecorded.New(c)
 
 	filePath := "https://static.deepgram.com/examples/Bueller-Life-moves-pretty-fast.wav"
 	var res interface{}
 	var err error
 
 	if isURL(filePath) {
-		res, err = prClient.PreRecordedFromURL(
-			api.UrlSource{Url: filePath},
-			api.PreRecordedTranscriptionOptions{
+		res, err = dg.FromURL(
+			ctx,
+			filePath,
+			interfaces.PreRecordedTranscriptionOptions{
 				Punctuate:  true,
 				Diarize:    true,
 				Language:   "en-US",
@@ -54,11 +61,10 @@ func main() {
 		}
 		defer file.Close()
 
-		source := api.ReadStreamSource{Stream: file, Mimetype: "YOUR_FILE_MIME_TYPE"}
-
-		res, err = prClient.PreRecordedFromStream(
-			source,
-			api.PreRecordedTranscriptionOptions{
+		res, err = dg.FromStream(
+			ctx,
+			file,
+			interfaces.PreRecordedTranscriptionOptions{
 				Punctuate:  true,
 				Diarize:    true,
 				Language:   "en-US",
@@ -71,13 +77,18 @@ func main() {
 		}
 	}
 
-	jsonStr, err := json.MarshalIndent(res, "", "  ")
+	data, err := json.Marshal(res)
 	if err != nil {
-		fmt.Println("Error marshaling JSON:", err)
+		log.Printf("RecognitionResult json.Marshal failed. Err: %v\n", err)
 		return
 	}
 
-	log.Printf("%s", jsonStr)
+	prettyJson, err := prettyjson.Format(data)
+	if err != nil {
+		log.Printf("prettyjson.Marshal failed. Err: %v\n", err)
+		return
+	}
+	log.Printf("\n\nResult:\n%s\n\n", prettyJson)
 }
 
 // Function to check if a string is a valid URL

diff --git a/examples/streaming/main.go b/examples/streaming/main.go
@@ -34,7 +34,7 @@ func main() {
 		Punctuate: true,
 	}
 
-	dgClient, err := client.NewWithDefaults(ctx, transcriptOptions)
+	dgClient, err := client.NewWithDefaults(ctx, "", transcriptOptions)
 	if err != nil {
 		log.Println("ERROR creating LiveTranscription connection:", err)
 		return

diff --git a/pkg/api/prerecorded/v1/constants.go b/pkg/api/prerecorded/v1/constants.go
@@ -0,0 +1,17 @@
+// Copyright 2023 Deepgram SDK contributors. All Rights Reserved.
+// Use of this source code is governed by a MIT license that can be found in the LICENSE file.
+// SPDX-License-Identifier: MIT
+
+package prerecorded
+
+import (
+	"errors"
+)
+
+var (
+	// ErrInvalidInput required input was not found
+	ErrInvalidInput = errors.New("required input was not found")
+
+	// ErrInvalidURIExtension couldn't find a period to indicate a file extension
+	ErrInvalidURIExtension = errors.New("couldn't find a period to indicate a file extension")
+)
diff --git a/pkg/api/prerecorded/v1/interfaces/interfaces.go b/pkg/api/prerecorded/v1/interfaces/interfaces.go
@@ -0,0 +1,62 @@
+// Copyright 2023 Deepgram SDK contributors. All Rights Reserved.
+// Use of this source code is governed by a MIT license that can be found in the LICENSE file.
+// SPDX-License-Identifier: MIT
+
+package interfaces
+
+import (
+	"encoding/json"
+	"errors"
+	"fmt"
+	"net/http"
+	"strings"
+)
+
+func (resp *PreRecordedResponse) ToWebVTT() (string, error) {
+	if resp.Results.Utterances == nil {
+		return "", errors.New("this function requires a transcript that was generated with the utterances feature")
+	}
+
+	vtt := "WEBVTT\n\n"
+
+	vtt += "NOTE\nTranscription provided by Deepgram\nRequest ID: " + resp.Metadata.RequestId + "\nCreated: " + resp.Metadata.Created + "\n\n"
+
+	for i, utterance := range resp.Results.Utterances {
+		utterance := utterance
+		start := SecondsToTimestamp(utterance.Start)
+		end := SecondsToTimestamp(utterance.End)
+		vtt += fmt.Sprintf("%d\n%s --> %s\n%s\n\n", i+1, start, end, utterance.Transcript)
+	}
+	return vtt, nil
+}
+
+func (resp *PreRecordedResponse) ToSRT() (string, error) {
+	if resp.Results.Utterances == nil {
+		return "", errors.New("this function requires a transcript that was generated with the utterances feature")
+	}
+
+	srt := ""
+
+	for i, utterance := range resp.Results.Utterances {
+		utterance := utterance
+		start := SecondsToTimestamp(utterance.Start)
+		end := SecondsToTimestamp(utterance.End)
+		end = strings.ReplaceAll(end, ".", ",")
+		srt += fmt.Sprintf("%d\n%s --> %s\n%s\n\n", i+1, start, end, utterance.Transcript)
+
+	}
+	return srt, nil
+}
+
+func SecondsToTimestamp(seconds float64) string {
+	hours := int(seconds / 3600)
+	minutes := int((seconds - float64(hours*3600)) / 60)
+	seconds = seconds - float64(hours*3600) - float64(minutes*60)
+	return fmt.Sprintf("%02d:%02d:%02.3f", hours, minutes, seconds)
+}
+
+func GetJson(resp *http.Response, target interface{}) error {
+	defer resp.Body.Close()
+
+	return json.NewDecoder(resp.Body).Decode(target)
+}
diff --git a/pkg/api/prerecorded/v1/interfaces/types.go b/pkg/api/prerecorded/v1/interfaces/types.go
@@ -0,0 +1,140 @@
+// Copyright 2023 Deepgram SDK contributors. All Rights Reserved.
+// Use of this source code is governed by a MIT license that can be found in the LICENSE file.
+// SPDX-License-Identifier: MIT
+
+package interfaces
+
+// share/common structs
+type Metadata struct {
+	TransactionKey string   `json:"transaction_key"`
+	RequestId      string   `json:"request_id"`
+	Sha256         string   `json:"sha256"`
+	Created        string   `json:"created"`
+	Duration       float64  `json:"duration"`
+	Channels       int      `json:"channels"`
+	Models         []string `json:"models"`
+	ModelInfo      map[string]struct {
+		Name    string `json:"name"`
+		Version string `json:"version"`
+		Arch    string `json:"arch"`
+	} `json:"model_info"`
+	Warnings []*Warning `json:"warnings,omitempty"`
+}
+
+type Warning struct {
+	Parameter string `json:"parameter"`
+	Type      string `json:"type"`
+	Message   string `json:"message"`
+}
+
+type Hit struct {
+	Confidence float64 `json:"confidence"`
+	Start      float64 `json:"start"`
+	End        float64 `json:"end"`
+	Snippet    string  `json:"snippet"`
+}
+
+type Search struct {
+	Query string `json:"query"`
+	Hits  []Hit  `json:"hits"`
+}
+
+type WordBase struct {
+	Word              string  `json:"word"`
+	Start             float64 `json:"start"`
+	End               float64 `json:"end"`
+	Confidence        float64 `json:"confidence"`
+	Speaker           *int    `json:"speaker,omitempty"`
+	SpeakerConfidence float64 `json:"speaker_confidence,omitempty"`
+	Punctuated_Word   string  `json:"punctuated_word,omitempty"`
+	Sentiment         string  `json:"sentiment,omitempty"`
+}
+
+type Alternative struct {
+	Transcript string          `json:"transcript"`
+	Confidence float64         `json:"confidence"`
+	Words      []WordBase      `json:"words"`
+	Summaries  []*SummaryV1    `json:"summaries,omitempty"`
+	Paragraphs *ParagraphGroup `json:"paragraphs,omitempty"`
+	Topics     []*TopicBase    `json:"topics,omitempty"`
+	Entities   []*EntityBase   `json:"entities,omitempty"`
+}
+
+type ParagraphGroup struct {
+	Transcript string          `json:"transcript"`
+	Paragraphs []ParagraphBase `json:"paragraphs"`
+}
+
+type ParagraphBase struct {
+	Sentences []SentenceBase `json:"sentences"`
+	NumWords  int            `json:"num_words"`
+	Start     float64        `json:"start"`
+	End       float64        `json:"end"`
+}
+
+type SentenceBase struct {
+	Text  string  `json:"text"`
+	Start float64 `json:"start"`
+	End   float64 `json:"end"`
+}
+
+type EntityBase struct {
+	Label      string  `json:"label"`
+	Value      string  `json:"value"`
+	Confidence float64 `json:"confidence"`
+	StartWord  int     `json:"start_word"`
+	EndWord    int     `json:"end_word"`
+}
+
+type TopicBase struct {
+	Text      string  `json:"text"`
+	StartWord int     `json:"start_word"`
+	EndWord   int     `json:"end_word"`
+	Topics    []Topic `json:"topics"`
+}
+
+type Topic struct {
+	Topic      string  `json:"topic"`
+	Confidence float64 `json:"confidence"`
+}
+
+type Channel struct {
+	Search           []*Search     `json:"search,omitempty"`
+	Alternatives     []Alternative `json:"alternatives"`
+	DetectedLanguage string        `json:"detected_language,omitempty"`
+}
+
+type Utterance struct {
+	Start      float64    `json:"start"`
+	End        float64    `json:"end"`
+	Confidence float64    `json:"confidence"`
+	Channel    int        `json:"channel"`
+	Transcript string     `json:"transcript"`
+	Words      []WordBase `json:"words"`
+	Speaker    *int       `json:"speaker,omitempty"`
+	Id         string     `json:"id"`
+}
+
+type Results struct {
+	Utterances []*Utterance `json:"utterances,omitempty"`
+	Channels   []Channel    `json:"channels"`
+	Summary    *SummaryV2   `json:"summary,omitempty"`
+}
+
+type SummaryV1 struct {
+	Summary   string `json:"summary"`
+	StartWord int    `json:"start_word"`
+	EndWord   int    `json:"end_word"`
+}
+
+type SummaryV2 struct {
+	Short  string `json:"short"`
+	Result string `json:"result"`
+}
+
+// Response
+type PreRecordedResponse struct {
+	Request_id string   `json:"request_id,omitempty"`
+	Metadata   Metadata `json:"metadata"`
+	Results    Results  `json:"results"`
+}