From b770ca319e640e2af24221dfe96d9af530c85083 Mon Sep 17 00:00:00 2001
From: "J. Nick Koston" <nick@koston.org>
Date: Sun, 1 May 2022 21:04:05 -0500
Subject: [PATCH] Improve scrape performance by using lxml parser (#71087)

* Improve scape performance by using lxml parser

* load it

* tweak

* tweak

* ensure libxml2 is installed in dev container
---
 Dockerfile.dev                                | 1 +
 homeassistant/components/scrape/manifest.json | 2 +-
 homeassistant/components/scrape/sensor.py     | 2 +-
 requirements_all.txt                          | 3 +++
 requirements_test_all.txt                     | 3 +++
 5 files changed, 9 insertions(+), 2 deletions(-)

diff --git a/Dockerfile.dev b/Dockerfile.dev
index dc04efe56fb..322c63f53dd 100644
--- a/Dockerfile.dev
+++ b/Dockerfile.dev
@@ -18,6 +18,7 @@ RUN \
         libavfilter-dev \
         libpcap-dev \
         libturbojpeg0 \
+        libxml2 \
         git \
         cmake \
     && apt-get clean \
diff --git a/homeassistant/components/scrape/manifest.json b/homeassistant/components/scrape/manifest.json
index bf5865206e4..b1ccbb354a9 100644
--- a/homeassistant/components/scrape/manifest.json
+++ b/homeassistant/components/scrape/manifest.json
@@ -2,7 +2,7 @@
   "domain": "scrape",
   "name": "Scrape",
   "documentation": "https://www.home-assistant.io/integrations/scrape",
-  "requirements": ["beautifulsoup4==4.11.1"],
+  "requirements": ["beautifulsoup4==4.11.1", "lxml==4.8.0"],
   "after_dependencies": ["rest"],
   "codeowners": ["@fabaff"],
   "iot_class": "cloud_polling"
diff --git a/homeassistant/components/scrape/sensor.py b/homeassistant/components/scrape/sensor.py
index 8f2a672ef06..e15f7c5ba97 100644
--- a/homeassistant/components/scrape/sensor.py
+++ b/homeassistant/components/scrape/sensor.py
@@ -154,7 +154,7 @@ class ScrapeSensor(SensorEntity):
 
     def _extract_value(self) -> Any:
         """Parse the html extraction in the executor."""
-        raw_data = BeautifulSoup(self.rest.data, "html.parser")
+        raw_data = BeautifulSoup(self.rest.data, "lxml")
         _LOGGER.debug(raw_data)
 
         try:
diff --git a/requirements_all.txt b/requirements_all.txt
index a331efde6c2..dd808efa0bd 100644
--- a/requirements_all.txt
+++ b/requirements_all.txt
@@ -974,6 +974,9 @@ lupupy==0.0.24
 # homeassistant.components.lw12wifi
 lw12==0.9.2
 
+# homeassistant.components.scrape
+lxml==4.8.0
+
 # homeassistant.components.nmap_tracker
 mac-vendor-lookup==0.1.11
 
diff --git a/requirements_test_all.txt b/requirements_test_all.txt
index 67bba2141d5..26f1c86d1af 100644
--- a/requirements_test_all.txt
+++ b/requirements_test_all.txt
@@ -663,6 +663,9 @@ lru-dict==1.1.7
 # homeassistant.components.luftdaten
 luftdaten==0.7.2
 
+# homeassistant.components.scrape
+lxml==4.8.0
+
 # homeassistant.components.nmap_tracker
 mac-vendor-lookup==0.1.11