Adding extract automation service

ai-services/Makefile

@@ -1,7 +1,7 @@
-REGISTRY ?= icr.io/ai-services-private
-IMAGE = ai-services
-TAG ?= v0.0.449
-CONTAINER_BUILDER ?= podman
+REGISTRY?=icr.io/ai-services-private
+IMAGE=ai-services
+TAG?=v0.0.450
+CONTAINER_BUILDER?=podman
 CREDS_ARG := $(if $(and $(REGISTRY_USER),$(REGISTRY_PASSWORD)),--creds="$(REGISTRY_USER):$(REGISTRY_PASSWORD)")
 
 GO := go

ai-services/assets/catalog/openshift/values.yaml

@@ -9,7 +9,7 @@ ui:
       cpu: "500m"
 
 backend:
-  image: icr.io/ai-services-cicd/ai-services:v0.0.449
+  image: icr.io/ai-services-cicd/ai-services:v0.0.450
   runtime: "openshift"
   adminPasswordHash: ""
   # @generate:password length=32, special=false

ai-services/assets/catalog/podman/values.yaml

@@ -4,7 +4,7 @@ ui:
 
 backend:
   port: ""
-  image: icr.io/ai-services-cicd/ai-services:v0.0.449
+  image: icr.io/ai-services-cicd/ai-services:v0.0.450
   runtime: ""
   adminPasswordHash: ""
   # @generate:password length=32, special=false

ai-services/assets/worker/openshift/values.yaml

@@ -1,5 +1,5 @@
 worker:
-  image: icr.io/ai-services-cicd/ai-services:v0.0.449
+  image: icr.io/ai-services-cicd/ai-services:v0.0.450
   runtime: "openshift"
   token: ""
   gatewayAddr: ""

ai-services/assets/worker/podman/values.yaml

@@ -1,5 +1,5 @@
 worker:
-  image: icr.io/ai-services-cicd/ai-services:v0.0.449
+  image: icr.io/ai-services-cicd/ai-services:v0.0.450
   runtime: "podman"
   token: ""
   gatewayAddr: ""

ai-services/tests/e2e/README.md

@@ -149,6 +149,54 @@ ginkgo -r -tags "$TAGS" \
   ./tests/e2e -- --app-name=<appname> --template=summarize --runtime=podman
 ```
 
+## Running Entity Extraction Tests
+
+The Entity Extraction Tests validate the `extract-api` service end-to-end, covering both the synchronous extraction API (`POST /v1/extract`) and the asynchronous jobs API (`POST /v1/extract/jobs`).
+
+The suite manages the full application lifecycle: it runs `catalog configure`, `catalog login`, creates a fresh `extract-app-<runID>`, runs all extraction specs, then deletes the app.
+
+### Prerequisites
+
+- Registry credentials must be set so the suite can pull the extract service image.
+- `CATALOG_PASSWORD` must be set — the suite runs `catalog configure` and `catalog login` automatically.
+- The Ginkgo CLI version must match the module version (`v2.28.1`). Install with:
+
+```bash
+go install github.com/onsi/ginkgo/v2/ginkgo@v2.28.1
+export PATH=$PATH:$(go env GOPATH)/bin
+```
+
+### Environment variables
+
+```bash
+# Required
+export CATALOG_PASSWORD="<your-catalog-admin-password>"
+export REGISTRY_URL="icr.io"
+export REGISTRY_USER_NAME="<your-icr-username>"
+export REGISTRY_PASSWORD="<your-icr-api-key>"
+
+# Optional — defaults to true (self-signed certs)
+export CATALOG_INSECURE=true
+
+# Optional — override default extract service port (default: 9500)
+# export EXTRACT_PORT=9500
+```
+
+### Commands to Run Entity Extraction Tests
+
+```bash
+# Using make (recommended — build tags applied automatically)
+make test TEST_ARGS="--label-filter=extraction-tests --timeout=3h" APP_RUNTIME=podman
+
+# Using ginkgo CLI directly (from ai-services/)
+export TAGS="exclude_graphdriver_btrfs containers_image_openpgp remote"
+ginkgo -r --procs=1 -tags "$TAGS" \
+  --label-filter="extraction-tests" \
+  --timeout=3h --v \
+  ./tests/e2e \
+  -- -template=extract -runtime=podman
+```
+
 ## Running Golden Dataset Validation Independently
 
 The RAG Golden Dataset Validation can be executed independently from the full E2E lifecycle. This allows validating an already running RAG application without creating or deleting an application during the test run.

ai-services/tests/e2e/cli/runner.go

@@ -46,6 +46,7 @@ const (
 	svcDigitizeBackend = "digitize-backend"
 	svcSimilarityAPI   = "similarity-api"
 	svcSummarizeAPI    = "summarize-api"
+	svcExtractAPI      = "extract-api"
 )
 
 // ptyWinRows and ptyWinCols define the PTY window size used by runWithPTY.
@@ -497,6 +498,11 @@ func ExtractCatalogSummarizeURL(infoOutput string) string {
 	return extractURLBySubstring(infoOutput, svcSummarizeAPI)
 }
 
+// ExtractExtractURL parses the 'application info' output for the entity extraction service URL.
+func ExtractExtractURL(infoOutput string) string {
+	return extractURLBySubstring(infoOutput, svcExtractAPI)
+}
+
 // getOpenShiftRouteURLs fetches OpenShift routes for the given application namespace
 // via 'oc get routes' and returns the HTTPS URLs found in the output.
 // This is the fallback for OpenShift when 'application info' does not embed URLs.

ai-services/tests/e2e/common/api.go

@@ -0,0 +1,102 @@
+package common
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"net/http"
+	"net/url"
+	"time"
+)
+
+// PaginationInfo holds page metadata for list responses.
+type PaginationInfo struct {
+	Total  int `json:"total"`
+	Limit  int `json:"limit"`
+	Offset int `json:"offset"`
+}
+
+// HTTPConfigForTimeout returns an HTTP client config with the given timeout and no connection pooling.
+func HTTPConfigForTimeout(timeout time.Duration) HTTPClientConfig {
+	return HTTPClientConfig{
+		Timeout:            timeout,
+		InsecureSkipVerify: true,
+		PoolConnections:    false,
+	}
+}
+
+// DeleteURL sends a DELETE request and accepts 200 or 204 as success.
+func DeleteURL(ctx context.Context, rawURL string, config HTTPClientConfig) error {
+	_, statusCode, err := DoDELETE(ctx, rawURL, config)
+	if err != nil {
+		return err
+	}
+
+	if statusCode != http.StatusOK && statusCode != http.StatusNoContent {
+		return fmt.Errorf("unexpected status %d", statusCode)
+	}
+
+	return nil
+}
+
+// AppendQueryParams returns rawURL with the non-empty params added as an encoded query string.
+func AppendQueryParams(rawURL string, params map[string]string) string {
+	query := url.Values{}
+
+	for key, val := range params {
+		if val != "" {
+			query.Set(key, val)
+		}
+	}
+
+	if len(query) == 0 {
+		return rawURL
+	}
+
+	return rawURL + "?" + query.Encode()
+}
+
+// GetJSON sends a GET request, requires HTTP 200 and decodes the JSON body into a T.
+func GetJSON[T any](ctx context.Context, rawURL string, config HTTPClientConfig) (*T, error) {
+	respBody, statusCode, err := DoGET(ctx, rawURL, config)
+	if err != nil {
+		return nil, err
+	}
+
+	if statusCode != http.StatusOK {
+		return nil, fmt.Errorf("unexpected status %d: %s", statusCode, string(respBody))
+	}
+
+	var out T
+	if err := json.Unmarshal(respBody, &out); err != nil {
+		return nil, fmt.Errorf("failed to parse response: %w", err)
+	}
+
+	return &out, nil
+}
+
+// FieldHasValue reports whether data has key set to a non-nil, non-empty-string value.
+func FieldHasValue(data map[string]interface{}, key string) bool {
+	val, ok := data[key]
+	if !ok || val == nil {
+		return false
+	}
+
+	if s, isStr := val.(string); isStr {
+		return s != ""
+	}
+
+	return true
+}
+
+// AllFieldsPresent reports whether every key in expected exists in data.
+// When one is missing it returns that key and false.
+func AllFieldsPresent(data map[string]interface{}, expected []string) (string, bool) {
+	for _, key := range expected {
+		if _, ok := data[key]; !ok {
+			return key, false
+		}
+	}
+
+	return "", true
+}

ai-services/tests/e2e/common/files.go

@@ -82,3 +82,22 @@ func RemoveDirContents(dir string) error {
 
 	return nil
 }
+
+// CreateTempFile writes content to a new temporary file whose name matches pattern
+// (e.g. "extract-text-*.txt") and returns its path. The caller must remove the file.
+func CreateTempFile(pattern, content string) (string, error) {
+	f, err := os.CreateTemp("", pattern)
+	if err != nil {
+		return "", fmt.Errorf("failed to create temp file: %w", err)
+	}
+
+	defer func() { _ = f.Close() }()
+
+	if _, err := f.WriteString(content); err != nil {
+		_ = os.Remove(f.Name())
+
+		return "", fmt.Errorf("failed to write temp file: %w", err)
+	}
+
+	return f.Name(), nil
+}

ai-services/tests/e2e/common/jobs.go

@@ -0,0 +1,102 @@
+package common
+
+import (
+	"context"
+	"fmt"
+	"time"
+
+	"github.com/project-ai-services/ai-services/internal/pkg/logger"
+)
+
+// JobStatus is the lifecycle state of an async job.
+type JobStatus string
+
+const (
+	JobStatusAccepted            JobStatus = "accepted"
+	JobStatusInProgress          JobStatus = "in_progress"
+	JobStatusCompleted           JobStatus = "completed"
+	JobStatusCompletedWithErrors JobStatus = "completed_with_errors"
+	JobStatusFailed              JobStatus = "failed"
+)
+
+// JobDocStatus is the lifecycle state of a single document inside a job.
+type JobDocStatus string
+
+const (
+	JobDocStatusPending    JobDocStatus = "pending"
+	JobDocStatusInProgress JobDocStatus = "in_progress"
+	JobDocStatusCompleted  JobDocStatus = "completed"
+	JobDocStatusFailed     JobDocStatus = "failed"
+)
+
+// JobPollConfig controls how WaitForJob polls a job.
+type JobPollConfig struct {
+	Timeout   time.Duration
+	Interval  time.Duration
+	LogPrefix string
+}
+
+// JobState is the part of a job response that WaitForJob needs to decide whether to keep polling.
+type JobState struct {
+	Status JobStatus
+	Error  *string
+}
+
+// classifyJobState reports whether polling is finished and the error, if any, for the given state.
+func classifyJobState(state JobState) (bool, error) {
+	switch state.Status {
+	case JobStatusCompleted, JobStatusCompletedWithErrors:
+		return true, nil
+	case JobStatusFailed:
+		errMsg := "unknown error"
+		if state.Error != nil {
+			errMsg = *state.Error
+		}
+
+		return true, fmt.Errorf("job failed: %s", errMsg)
+	case JobStatusAccepted, JobStatusInProgress:
+		return false, nil
+	default:
+		return true, fmt.Errorf("unexpected job status: %s", state.Status)
+	}
+}
+
+// WaitForJob polls fetch until the job reaches a terminal state, ctx is cancelled or the timeout elapses.
+// stateOf extracts the status and error message from the fetched response.
+func WaitForJob[T any](
+	ctx context.Context,
+	jobID string,
+	cfg JobPollConfig,
+	fetch func(context.Context) (*T, error),
+	stateOf func(*T) JobState,
+) (*T, error) {
+	deadline := time.Now().Add(cfg.Timeout)
+	ticker := time.NewTicker(cfg.Interval)
+
+	defer ticker.Stop()
+
+	for {
+		select {
+		case <-ctx.Done():
+			return nil, ctx.Err()
+		case <-ticker.C:
+			if time.Now().After(deadline) {
+				return nil, fmt.Errorf("timed out waiting for job %s", jobID)
+			}
+
+			job, err := fetch(ctx)
+			if err != nil {
+				logger.Warningf("[%s] get job detail failed: %v", cfg.LogPrefix, err)
+
+				continue
+			}
+
+			state := stateOf(job)
+			logger.Infof("[%s] job %s status: %s", cfg.LogPrefix, jobID, state.Status)
+
+			if done, jobErr := classifyJobState(state); done {
+				return job, jobErr
+			}
+		}
+	}
+}

ai-services/tests/e2e/digitization/digitize.go

@@ -99,15 +99,8 @@ type JobStatusResponse struct {
 
 // JobsListResponse represents the response when listing jobs.
 type JobsListResponse struct {
-	Data       []JobStatusResponse `json:"data"`
-	Pagination PaginationInfo      `json:"pagination"`
-}
-
-// PaginationInfo represents pagination metadata.
-type PaginationInfo struct {
-	Total  int `json:"total"`
-	Limit  int `json:"limit"`
-	Offset int `json:"offset"`
+	Data       []JobStatusResponse   `json:"data"`
+	Pagination common.PaginationInfo `json:"pagination"`
 }
 
 // DocumentDetailResponse represents a document returned by both the list and detail endpoints (same JSON shape).
@@ -129,8 +122,8 @@ type DocumentListItem = DocumentDetailResponse
 
 // DocumentsListResponse represents the response when listing documents.
 type DocumentsListResponse struct {
-	Data       []DocumentListItem `json:"data"`
-	Pagination PaginationInfo     `json:"pagination"`
+	Data       []DocumentListItem    `json:"data"`
+	Pagination common.PaginationInfo `json:"pagination"`
 }
 
 // DocumentContentResponse represents the document content.