Merge pull request #433 from onc-healthit/LANTERN-769-ehealthline-web…

…scraper

LANTERN-769: ehealthline wenscraper

endpointmanager/pkg/chplendpointquerier/chplendpointquerier.go

@@ -192,6 +192,7 @@ var pointclickURL = "https://fhir.pointclickcare.com/"
 var nextgenPracticeURL = "https://www.nextgen.com/api/practice-search"
 var aspmdURL = "https://fhirapi.asp.md:3030/aspmd/fhirserver/fhir_aspmd.asp"
 var axeiumURL = "https://apifhir.axeium.net:8443/reference-server/"
+var ehealthlineURL = "http://ehealthline.com/dev/pdf/FHIR%20API%20Endpoints.htm"
 
 var bundleQuerierArray = [30]string{"https://ac-fhir.harrisambulatory.com/endpoints/r4", "https://dynamicfhirpresentation.dynamicfhirsandbox.com/fhir/r4/endpoints",
 	"https://ct-fhir.harrisambulatory.com/Endpoints/R4", "https://kantime.com/wp-content/uploads/2024/03/fhir-base-urls.json",
@@ -552,6 +553,8 @@ func QueryCHPLEndpointList(chplURL string, fileToWriteTo string) {
 		AxeiumeWebscraper(axeiumURL, fileToWriteTo)
 	} else if contains(bundleQuerierArray, chplURL) {
 		BundleQuerierParser(chplURL, fileToWriteTo)
+	} else if URLsEqual(chplURL, ehealthlineURL) {
+		EhealthlineWebscraper(ehealthlineURL, fileToWriteTo)
 	} else {
 		log.Warnf("Handler is required for url %s", chplURL)
 	}

endpointmanager/pkg/chplendpointquerier/ehealthlinewebscraper.go

@@ -0,0 +1,40 @@
+package chplendpointquerier
+
+import (
+	"regexp"
+	"strings"
+
+	"github.com/PuerkitoBio/goquery"
+	"github.com/onc-healthit/lantern-back-end/endpointmanager/pkg/helpers"
+	log "github.com/sirupsen/logrus"
+)
+
+func EhealthlineWebscraper(CHPLURL string, fileToWriteTo string) {
+
+	var lanternEntryList []LanternEntry
+	var endpointEntryList EndpointList
+
+	ipPattern := `\b\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}(:\d{1,5})?\b`
+	ipRegex := regexp.MustCompile(ipPattern)
+
+	doc, err := helpers.ChromedpQueryEndpointList(CHPLURL, "")
+	if err != nil {
+		log.Fatal(err)
+	}
+
+	doc.Find("td:contains('https')").Each(func(index int, item *goquery.Selection) {
+		if !ipRegex.MatchString(item.Text()) {
+			var entry LanternEntry
+			entry.URL = strings.TrimSpace(item.Find("span").Text())
+			lanternEntryList = append(lanternEntryList, entry)
+		}
+	})
+
+	endpointEntryList.Endpoints = lanternEntryList
+
+	err = WriteCHPLFile(endpointEntryList, fileToWriteTo)
+	if err != nil {
+		log.Fatal(err)
+	}
+
+}