|
| 1 | +package openai |
| 2 | + |
| 3 | +import ( |
| 4 | + "bytes" |
| 5 | + "context" |
| 6 | + "encoding/base64" |
| 7 | + "encoding/binary" |
| 8 | + "encoding/json" |
| 9 | + "fmt" |
| 10 | + "io" |
| 11 | + "math" |
| 12 | + "net/http" |
| 13 | + "net/url" |
| 14 | +) |
| 15 | + |
| 16 | +// Usage tracks API token usage. |
| 17 | +type Usage struct { |
| 18 | + PromptTokens int `json:"prompt_tokens"` |
| 19 | + TotalTokens int `json:"total_tokens"` |
| 20 | +} |
| 21 | + |
| 22 | +// Embedding is openai API vector embedding. |
| 23 | +type Embedding struct { |
| 24 | + Object string `json:"object"` |
| 25 | + Index int `json:"index"` |
| 26 | + Embedding []float64 `json:"embedding"` |
| 27 | +} |
| 28 | + |
| 29 | +// EmbeddingString is base64 encoded embedding. |
| 30 | +type EmbeddingString string |
| 31 | + |
| 32 | +func (s EmbeddingString) Decode() ([]float64, error) { |
| 33 | + decoded, err := base64.StdEncoding.DecodeString(string(s)) |
| 34 | + if err != nil { |
| 35 | + return nil, err |
| 36 | + } |
| 37 | + |
| 38 | + if len(decoded)%8 != 0 { |
| 39 | + return nil, fmt.Errorf("invalid base64 encoded string length") |
| 40 | + } |
| 41 | + |
| 42 | + floats := make([]float64, len(decoded)/8) |
| 43 | + |
| 44 | + for i := 0; i < len(floats); i++ { |
| 45 | + bits := binary.LittleEndian.Uint64(decoded[i*8 : (i+1)*8]) |
| 46 | + floats[i] = math.Float64frombits(bits) |
| 47 | + } |
| 48 | + |
| 49 | + return floats, nil |
| 50 | +} |
| 51 | + |
| 52 | +// EmbeddingRequest is serialized and sent to the API server. |
| 53 | +type EmbeddingRequest struct { |
| 54 | + Input any `json:"input"` |
| 55 | + Model Model `json:"model"` |
| 56 | + User string `json:"user"` |
| 57 | + EncodingFormat EncodingFormat `json:"encoding_format,omitempty"` |
| 58 | +} |
| 59 | + |
| 60 | +// Data is used for deserializing response data. |
| 61 | +type Data[T any] struct { |
| 62 | + Object string `json:"object"` |
| 63 | + Index int `json:"index"` |
| 64 | + Embedding T `json:"embedding"` |
| 65 | +} |
| 66 | + |
| 67 | +// EmbeddingResponse is the API response from a Create embeddings request. |
| 68 | +type EmbeddingResponse[T any] struct { |
| 69 | + Object string `json:"object"` |
| 70 | + Data []Data[T] `json:"data"` |
| 71 | + Model Model `json:"model"` |
| 72 | + Usage Usage `json:"usage"` |
| 73 | +} |
| 74 | + |
| 75 | +func ToEmbeddings[T any](resp io.Reader) ([]*Embedding, error) { |
| 76 | + data := new(T) |
| 77 | + if err := json.NewDecoder(resp).Decode(data); err != nil { |
| 78 | + return nil, err |
| 79 | + } |
| 80 | + |
| 81 | + switch e := any(data).(type) { |
| 82 | + case *EmbeddingResponse[EmbeddingString]: |
| 83 | + embs := make([]*Embedding, 0, len(e.Data)) |
| 84 | + for _, d := range e.Data { |
| 85 | + floats, err := d.Embedding.Decode() |
| 86 | + if err != nil { |
| 87 | + return nil, err |
| 88 | + } |
| 89 | + emb := &Embedding{ |
| 90 | + Object: d.Object, |
| 91 | + Index: d.Index, |
| 92 | + Embedding: floats, |
| 93 | + } |
| 94 | + embs = append(embs, emb) |
| 95 | + } |
| 96 | + return embs, nil |
| 97 | + case *EmbeddingResponse[[]float64]: |
| 98 | + embs := make([]*Embedding, 0, len(e.Data)) |
| 99 | + for _, d := range e.Data { |
| 100 | + emb := &Embedding{ |
| 101 | + Object: d.Object, |
| 102 | + Index: d.Index, |
| 103 | + Embedding: d.Embedding, |
| 104 | + } |
| 105 | + embs = append(embs, emb) |
| 106 | + } |
| 107 | + return embs, nil |
| 108 | + } |
| 109 | + |
| 110 | + return nil, ErrInValidData |
| 111 | +} |
| 112 | + |
| 113 | +// Embeddings returns embeddings for every object in EmbeddingRequest. |
| 114 | +func (c *Client) Embeddings(ctx context.Context, embReq *EmbeddingRequest) ([]*Embedding, error) { |
| 115 | + u, err := url.Parse(c.baseURL + "/embeddings") |
| 116 | + if err != nil { |
| 117 | + return nil, err |
| 118 | + } |
| 119 | + |
| 120 | + var body = &bytes.Buffer{} |
| 121 | + enc := json.NewEncoder(body) |
| 122 | + enc.SetEscapeHTML(false) |
| 123 | + if err := enc.Encode(embReq); err != nil { |
| 124 | + return nil, err |
| 125 | + } |
| 126 | + |
| 127 | + req, err := c.newRequest(ctx, http.MethodPost, u.String(), body) |
| 128 | + if err != nil { |
| 129 | + return nil, err |
| 130 | + } |
| 131 | + |
| 132 | + resp, err := c.doRequest(req) |
| 133 | + if err != nil { |
| 134 | + return nil, err |
| 135 | + } |
| 136 | + defer resp.Body.Close() |
| 137 | + |
| 138 | + switch embReq.EncodingFormat { |
| 139 | + case EncodingBase64: |
| 140 | + return ToEmbeddings[EmbeddingResponse[EmbeddingString]](resp.Body) |
| 141 | + case EncodingFloat: |
| 142 | + return ToEmbeddings[EmbeddingResponse[[]float64]](resp.Body) |
| 143 | + } |
| 144 | + |
| 145 | + return nil, fmt.Errorf("unknown encoding: %v", embReq.EncodingFormat) |
| 146 | +} |
0 commit comments