From 9b68e3c3c14579f3e4256d80ef5f112d388ce387 Mon Sep 17 00:00:00 2001
From: archita-ekkirala <archita.ekkirala@mettles.com>
Date: Fri, 8 Nov 2024 09:44:49 -0600
Subject: [PATCH] LANTERN-782-Add-curemd-webscraper

---
 .../chplendpointquerier.go                    |  3 ++
 .../chplendpointquerier/curemdwebscraper.go   | 45 +++++++++++++++++++
 2 files changed, 48 insertions(+)
 create mode 100644 endpointmanager/pkg/chplendpointquerier/curemdwebscraper.go

diff --git a/endpointmanager/pkg/chplendpointquerier/chplendpointquerier.go b/endpointmanager/pkg/chplendpointquerier/chplendpointquerier.go
index 4d7c2013d..ebe9e75f9 100644
--- a/endpointmanager/pkg/chplendpointquerier/chplendpointquerier.go
+++ b/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 curemdURL = "https://www.curemd.com/developer/base-fhir-urls/"
 
 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, curemdURL) {
+		CuremdWebscraper(curemdURL, fileToWriteTo)
 	} else {
 		log.Warnf("Handler is required for url %s", chplURL)
 	}
diff --git a/endpointmanager/pkg/chplendpointquerier/curemdwebscraper.go b/endpointmanager/pkg/chplendpointquerier/curemdwebscraper.go
new file mode 100644
index 000000000..40a561370
--- /dev/null
+++ b/endpointmanager/pkg/chplendpointquerier/curemdwebscraper.go
@@ -0,0 +1,45 @@
+package chplendpointquerier
+
+import (
+	"strings"
+
+	"github.com/PuerkitoBio/goquery"
+	"github.com/onc-healthit/lantern-back-end/endpointmanager/pkg/helpers"
+	log "github.com/sirupsen/logrus"
+)
+
+func CuremdWebscraper(CHPLURL string, fileToWriteTo string) {
+
+	var entry LanternEntry
+	var lanternEntryList []LanternEntry
+	var endpointEntryList EndpointList
+
+	doc, err := helpers.ChromedpQueryEndpointList(CHPLURL, "")
+	if err != nil {
+		log.Fatal(err)
+	}
+
+	doc.Find("label").Each(func(index int, item *goquery.Selection) {
+		labelText := strings.TrimSpace(item.Text())
+		if labelText == "Capability Statement:" {
+			pTag := item.NextFiltered("p")
+			if pTag.Length() > 0 {
+				pTag.Find("a").Each(func(i int, link *goquery.Selection) {
+					url, exists := link.Attr("href")
+					if exists {
+						entry.URL = url
+						lanternEntryList = append(lanternEntryList, entry)
+					}
+				})
+			}
+		}
+	})
+
+	endpointEntryList.Endpoints = lanternEntryList
+
+	err = WriteCHPLFile(endpointEntryList, fileToWriteTo)
+	if err != nil {
+		log.Fatal(err)
+	}
+
+}