Merge pull request #391 from onc-healthit/LANTERN-630-KaiserURLWebscr…

…aper

LANTERN-630: Modified the KaiserURLWebscraper

endpointmanager/pkg/chplendpointquerier/chplendpointquerier.go

@@ -166,7 +166,8 @@ var anthemURL = "https://patient360.anthem.com/P360Member/fhir"
 // var guidewellURL = "https://developer.bcbsfl.com/interop/interop-developer-portal/product/469/api/466#/PatientAccessAPI_105/overview"
 // var hcscURL = "https://interoperability.hcsc.com/s/patient-access-api"
 // var humanaURL = "https://developers.humana.com/patient-api/doc"
-// var kaiserURL = "https://developer.kp.org/#/apis/639c015049655aa96ab5b2f1"
+var kaiserURL = "https://developer.kp.org/#/apis/639c015049655aa96ab5b2f1"
+
 // var molinaURL = "https://developer.interop.molinahealthcare.com/api-details#api=patient-access&operation=5f72ab665269f310ef58b361"
 var unitedHealthURL = "https://www.uhc.com/legal/interoperability-apis"
 var meldrxURL = "https://app.meldrx.com/api/Directories/fhir/endpoints"
@@ -512,8 +513,8 @@ func QueryCHPLEndpointList(chplURL string, fileToWriteTo string) {
 		// 	HcscURLWebscraper(chplURL, fileToWriteTo)
 		// } else if URLsEqual(chplURL, humanaURL) {
 		// 	HumanaURLWebscraper(chplURL, fileToWriteTo)
-		//} else if URLsEqual(chplURL, kaiserURL) {
-		//KaiserURLWebscraper(chplURL, fileToWriteTo)
+	} else if URLsEqual(chplURL, kaiserURL) {
+		KaiserURLWebscraper(chplURL, fileToWriteTo)
 		// } else if URLsEqual(chplURL, molinaURL) {
 		// 	MolinaURLWebscraper(chplURL, fileToWriteTo)
 	} else if URLsEqual(chplURL, unitedHealthURL) {

endpointmanager/pkg/chplendpointquerier/kaiserwebscraper.go

@@ -1,10 +1,12 @@
 package chplendpointquerier
 
 import (
+	"context"
 	"strings"
+	"time"
 
 	"github.com/PuerkitoBio/goquery"
-	"github.com/onc-healthit/lantern-back-end/endpointmanager/pkg/helpers"
+	"github.com/chromedp/chromedp"
 	log "github.com/sirupsen/logrus"
 )
 
@@ -13,7 +15,7 @@ func KaiserURLWebscraper(CHPLURL string, fileToWriteTo string) {
 	var lanternEntryList []LanternEntry
 	var endpointEntryList EndpointList
 
-	doc, err := helpers.ChromedpQueryEndpointList(CHPLURL, ".language-json")
+	doc, err := KaiserChromedpQueryEndpointList(CHPLURL, ".opblock-tag-section")
 	if err != nil {
 		log.Fatal(err)
 	}
@@ -25,9 +27,10 @@ func KaiserURLWebscraper(CHPLURL string, fileToWriteTo string) {
 				found = true
 			}
 			if found {
-				if strings.Contains(spanhtml.Text(), "/FHIR/api") {
+				if strings.HasSuffix(spanhtml.Text(), "/FHIR/api\"") {
 					var entry LanternEntry
 					URL := strings.TrimSpace(spanhtml.Text())
+					URL = strings.ReplaceAll(URL, "\"", "")
 					entry.URL = URL
 					processed = true
 					lanternEntryList = append(lanternEntryList, entry)
@@ -49,3 +52,53 @@ func KaiserURLWebscraper(CHPLURL string, fileToWriteTo string) {
 	}
 
 }
+
+// KaiserChromedpQueryEndpointList queries the given endpoint list and clicks buttons using chromedp and returns the html document
+func KaiserChromedpQueryEndpointList(endpointListURL string, waitVisibleElement string) (*goquery.Document, error) {
+
+	ctx, cancel := chromedp.NewContext(context.Background())
+	defer cancel()
+
+	timeoutContext, cancel := context.WithTimeout(ctx, 90*time.Second)
+	defer cancel()
+
+	var htmlContent string
+	var err error
+
+	if len(waitVisibleElement) > 0 {
+		// Chromedp will wait a max of 30 seconds for webpage to run javascript code to generate api search results before grapping HTML
+		err = chromedp.Run(timeoutContext,
+			chromedp.Navigate(endpointListURL),
+			chromedp.WaitVisible(waitVisibleElement, chromedp.ByQuery),
+
+			// Expand the Metadata section
+			chromedp.WaitVisible(`.expand-operation`),
+			chromedp.Click(`.expand-operation`, chromedp.ByQuery),
+
+			// Expand the Metadata endpoint section
+			chromedp.WaitVisible(`.opblock-summary-control`),
+			chromedp.Click(`.opblock-summary-control`, chromedp.ByQuery),
+
+			// Wait till the code snippet is rendered
+			chromedp.WaitVisible(`.language-json`),
+
+			chromedp.OuterHTML("html", &htmlContent, chromedp.ByQuery),
+		)
+	} else {
+		err = chromedp.Run(timeoutContext,
+			chromedp.Navigate(endpointListURL),
+			chromedp.OuterHTML("html", &htmlContent, chromedp.ByQuery),
+		)
+	}
+
+	if err != nil {
+		return nil, err
+	}
+
+	doc, err := goquery.NewDocumentFromReader(strings.NewReader(htmlContent))
+	if err != nil {
+		return nil, err
+	}
+
+	return doc, nil
+}