feat(pinecone): Add rerank task for Pinecone component #773
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -17,9 +17,11 @@ import ( | |
| const ( | ||
| taskQuery = "TASK_QUERY" | ||
| taskUpsert = "TASK_UPSERT" | ||
| taskRerank = "TASK_RERANK" | ||
|
|
||
| upsertPath = "/vectors/upsert" | ||
| queryPath = "/query" | ||
| rerankPath = "/rerank" | ||
| ) | ||
|
|
||
| //go:embed config/definition.json | ||
|
|
@@ -59,7 +61,8 @@ func (c *component) CreateExecution(x base.ComponentExecution) (base.IExecution, | |
| }, nil | ||
| } | ||
|
|
||
| // newIndexClient creates a new httpclient.Client with the index URL provided in setup | ||
| func newIndexClient(setup *structpb.Struct, logger *zap.Logger) *httpclient.Client { | ||
| c := httpclient.New("Pinecone", getURL(setup), | ||
| httpclient.WithLogger(logger), | ||
| httpclient.WithEndUserError(new(errBody)), | ||
|
|
@@ -71,6 +74,23 @@ func newClient(setup *structpb.Struct, logger *zap.Logger) *httpclient.Client { | |
| return c | ||
| } | ||
|
|
||
| // newBaseClient creates a new httpclient.Client with the default Pinecone API URL. | ||
| func newBaseClient(setup *structpb.Struct, logger *zap.Logger) *httpclient.Client { | ||
| c := httpclient.New("Pinecone", "https://api.pinecone.io", | ||
| httpclient.WithLogger(logger), | ||
| httpclient.WithEndUserError(new(errBody)), | ||
| ) | ||
|
|
||
| c.SetHeader("Api-Key", getAPIKey(setup)) | ||
| c.SetHeader("User-Agent", "source_tag=instillai") | ||
|
|
||
| // Currently, by default Pinecone API redirects request to OLDEST stable version i.e. 2024-04 right now and does not support Rerank | ||
| // It is recommended by Pinecone to specify API version to use: https://docs.pinecone.io/reference/api/versioning#specify-an-api-version | ||
| c.SetHeader("X-Pinecone-API-Version", "2024-10") | ||
|
|
||
| return c | ||
| } | ||
|
|
||
| func getAPIKey(setup *structpb.Struct) string { | ||
| return setup.GetFields()["api-key"].GetStringValue() | ||
| } | ||
|
|
@@ -81,8 +101,6 @@ func getURL(setup *structpb.Struct) string { | |
|
|
||
| func (e *execution) Execute(ctx context.Context, jobs []*base.Job) error { | ||
|
|
||
| for _, job := range jobs { | ||
| input, err := job.Input.Read(ctx) | ||
| if err != nil { | ||
|
|
@@ -92,6 +110,8 @@ func (e *execution) Execute(ctx context.Context, jobs []*base.Job) error { | |
| var output *structpb.Struct | ||
| switch e.Task { | ||
| case taskQuery: | ||
| req := newIndexClient(e.Setup, e.GetLogger()).R() | ||
|
|
||
| inputStruct := queryInput{} | ||
| err := base.ConvertFromStructpb(input, &inputStruct) | ||
| if err != nil { | ||
|
|
@@ -122,6 +142,8 @@ func (e *execution) Execute(ctx context.Context, jobs []*base.Job) error { | |
| continue | ||
| } | ||
| case taskUpsert: | ||
| req := newIndexClient(e.Setup, e.GetLogger()).R() | ||
|
|
||
| v := upsertInput{} | ||
| err := base.ConvertFromStructpb(input, &v) | ||
| if err != nil { | ||
|
|
@@ -145,7 +167,34 @@ func (e *execution) Execute(ctx context.Context, jobs []*base.Job) error { | |
| job.Error.Error(ctx, err) | ||
| continue | ||
| } | ||
| case taskRerank: | ||
|
There was a problem hiding this comment. It'd be good to have the test code if possible. |
||
| // rerank task does not need index URL, so using the base client with the default pinecone API URL. | ||
| req := newBaseClient(e.Setup, e.GetLogger()).R() | ||
|
|
||
| // parse input struct | ||
| inputStruct := rerankInput{} | ||
| err := base.ConvertFromStructpb(input, &inputStruct) | ||
| if err != nil { | ||
| job.Error.Error(ctx, err) | ||
| continue | ||
| } | ||
|
|
||
| // make API request to rerank task | ||
| resp := rerankResp{} | ||
| req.SetResult(&resp).SetBody(inputStruct.asRequest()) | ||
| if _, err := req.Post(rerankPath); err != nil { | ||
| job.Error.Error(ctx, httpclient.WrapURLError(err)) | ||
|
There was a problem hiding this comment. Recently, I found the clients don't reply the meaningful error message in Could you add the errMsg like this? There was a problem hiding this comment. WrapURLError is a helper to add an end-user message to http transport errors. |
||
| continue | ||
| } | ||
|
|
||
| // convert response to output struct | ||
| output, err = base.ConvertToStructpb(resp.toOutput()) | ||
| if err != nil { | ||
| job.Error.Error(ctx, err) | ||
| continue | ||
| } | ||
| } | ||
|
|
||
| err = job.Output.Write(ctx, output) | ||
| if err != nil { | ||
| job.Error.Error(ctx, err) | ||
|
|
||
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -83,6 +83,65 @@ type upsertOutput struct { | |
| RecordsUpserted int64 `json:"upserted-count"` | ||
| } | ||
|
|
||
| type Document struct { | ||
| Text string `json:"text"` | ||
| } | ||
|
|
||
| type rerankInput struct { | ||
| // not taking model as input for now as only one model is supported for rerank task: https://docs.pinecone.io/guides/inference/understanding-inference#models | ||
| //ModelName string `json:"model-name"` | ||
| Query string `json:"query"` | ||
| Documents []string `json:"documents"` | ||
| TopN int `json:"top-n"` | ||
| } | ||
|
|
||
| func (r *rerankInput) asRequest() *rerankReq { | ||
| reqDocuments := make([]Document, 0, len(r.Documents)) | ||
| for _, doc := range r.Documents { | ||
| reqDocuments = append(reqDocuments, Document{Text: doc}) | ||
| } | ||
|
|
||
| return &rerankReq{ | ||
| Model: "bge-reranker-v2-m3", | ||
|
There was a problem hiding this comment. Though it only provides a model now, we can make it a field with a default value. You can refer to this. |
||
| Query: r.Query, | ||
| TopN: r.TopN, | ||
| Documents: reqDocuments, | ||
| } | ||
| } | ||
|
|
||
| type rerankReq struct { | ||
| Model string `json:"model"` | ||
| Query string `json:"query"` | ||
| TopN int `json:"top_n,omitempty"` | ||
| Documents []Document `json:"documents"` | ||
| } | ||
|
|
||
| type rerankResp struct { | ||
| Data []struct { | ||
| Index int `json:"index"` | ||
| Document Document `json:"document"` | ||
| Score float64 `json:"score"` | ||
| } `json:"data"` | ||
| } | ||
|
|
||
| func (r *rerankResp) toOutput() rerankOutput { | ||
| documents := make([]string, 0, len(r.Data)) | ||
| scores := make([]float64, 0, len(r.Data)) | ||
| for _, d := range r.Data { | ||
| documents = append(documents, d.Document.Text) | ||
| scores = append(scores, d.Score) | ||
| } | ||
| return rerankOutput{ | ||
| Documents: documents, | ||
| Scores: scores, | ||
| } | ||
| } | ||
|
|
||
| type rerankOutput struct { | ||
| Documents []string `json:"documents"` | ||
| Scores []float64 `json:"scores"` | ||
| } | ||
|
|
||
| type errBody struct { | ||
| Msg string `json:"message"` | ||
| } | ||
|
|
||
Oops, something went wrong.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Now, we always add period in the end for description fields.