feat: remove image from the description if configured

Author: ogajduse

custom_components/feedparser/sensor.py

@@ -35,12 +35,14 @@
 CONF_INCLUSIONS = "inclusions"
 CONF_EXCLUSIONS = "exclusions"
 CONF_SHOW_TOPN = "show_topn"
+CONF_REMOVE_SUMMARY_IMG = "remove_summary_image"
 
 DEFAULT_DATE_FORMAT = "%a, %b %d %I:%M %p"
 DEFAULT_SCAN_INTERVAL = timedelta(hours=1)
 DEFAULT_THUMBNAIL = "https://www.home-assistant.io/images/favicon-192x192-full.png"
 DEFAULT_TOPN = 9999
 USER_AGENT = f"Home Assistant Feed-parser Integration {__version__}"
+IMAGE_REGEX = r"<img.+?src=\"(.+?)\".+?>"
 
 PLATFORM_SCHEMA = PLATFORM_SCHEMA.extend(
     {
@@ -49,6 +51,7 @@
         vol.Required(CONF_DATE_FORMAT, default=DEFAULT_DATE_FORMAT): cv.string,
         vol.Optional(CONF_LOCAL_TIME, default=False): cv.boolean,
         vol.Optional(CONF_SHOW_TOPN, default=DEFAULT_TOPN): cv.positive_int,
+        vol.Optional(CONF_REMOVE_SUMMARY_IMG, default=False): cv.boolean,
         vol.Optional(CONF_INCLUSIONS, default=[]): vol.All(cv.ensure_list, [cv.string]),
         vol.Optional(CONF_EXCLUSIONS, default=[]): vol.All(cv.ensure_list, [cv.string]),
         vol.Optional(CONF_SCAN_INTERVAL, default=DEFAULT_SCAN_INTERVAL): cv.time_period,
@@ -72,6 +75,7 @@ async def async_setup_platform(
                 name=config[CONF_NAME],
                 date_format=config[CONF_DATE_FORMAT],
                 show_topn=config[CONF_SHOW_TOPN],
+                remove_summary_image=config[CONF_REMOVE_SUMMARY_IMG],
                 inclusions=config[CONF_INCLUSIONS],
                 exclusions=config[CONF_EXCLUSIONS],
                 scan_interval=config[CONF_SCAN_INTERVAL],
@@ -95,6 +99,7 @@ def __init__(
         name: str,
         date_format: str,
         show_topn: int,
+        remove_summary_image: bool,
         exclusions: list[str | None],
         inclusions: list[str | None],
         scan_interval: timedelta,
@@ -106,6 +111,7 @@ def __init__(
         self._attr_icon = "mdi:rss"
         self._date_format = date_format
         self._show_topn: int = show_topn
+        self._remove_summary_image = remove_summary_image
         self._inclusions = inclusions
         self._exclusions = exclusions
         self._scan_interval = scan_interval
@@ -119,7 +125,9 @@ def __repr__(self: FeedParserSensor) -> str:
         """Return the representation."""
         return (
             f'FeedParserSensor(name="{self.name}", feed="{self._feed}", '
-            f"show_topn={self._show_topn}, inclusions={self._inclusions}, "
+            f"show_topn={self._show_topn}, "
+            f"remove_summary_image={self._remove_summary_image}, "
+            f"inclusions={self._inclusions}, "
             f"exclusions={self._exclusions}, scan_interval={self._scan_interval}, "
             f'local_time={self._local_time}, date_format="{self._date_format}")'
         )
@@ -199,6 +207,12 @@ def _generate_sensor_entry(
             and (processed_link := self._process_link(feed_entry))
         ):
             sensor_entry["link"] = processed_link
+        if self._remove_summary_image and "summary" in sensor_entry:
+            sensor_entry["summary"] = re.sub(
+                IMAGE_REGEX,
+                "",
+                sensor_entry["summary"],
+            )
         _LOGGER.debug("Feed %s: Generated sensor entry: %s", self.name, sensor_entry)
         return sensor_entry
 
@@ -248,7 +262,7 @@ def _process_image(self: FeedParserSensor, feed_entry: FeedParserDict) -> str:
                 return images[0]["href"]
         elif "summary" in feed_entry:
             images = re.findall(
-                r"<img.+?src=\"(.+?)\".+?>",
+                IMAGE_REGEX,
                 feed_entry["summary"],
             )
             if images:

pyproject.toml

@@ -150,7 +150,7 @@ target-version = "py311"
 "tests/**" = ["S101"]
 
 [tool.ruff.pylint]
-max-args = 9
+max-args = 10
 
 [[tool.mypy.overrides]]
 module = "feedparser.*"

tests/feedsource.py

@@ -111,6 +111,7 @@ def _common_config(self: "FeedSource") -> dict[str, str | int | bool | list[str]
             "name": self.name,
             "date_format": self.sensor_config.date_format,
             "show_topn": self.sensor_config.show_topn,
+            "remove_summary_image": self.sensor_config.remove_summary_image,
             "inclusions": self.sensor_config.inclusions,
             "exclusions": self.sensor_config.exclusions,
             "local_time": self.sensor_config.local_time,
@@ -191,6 +192,11 @@ def show_topn(self: "FeedConfig") -> int:
         """Return show_topn."""
         return self.raw.get("show_topn", 9999)
 
+    @property
+    def remove_summary_image(self: "FeedConfig") -> bool:
+        """Return remove_summary_image."""
+        return self.raw.get("remove_summary_image", False)
+
     @property
     def scan_interval(self: "FeedConfig") -> int:
         """Return scan_interval in seconds."""

tests/test_sensors.py

@@ -33,6 +33,7 @@ def test_update_sensor(feed: FeedSource) -> None:
         date_format=DATE_FORMAT,
         local_time=False,
         show_topn=9999,
+        remove_summary_image=feed.sensor_config.remove_summary_image,
         inclusions=["image", "title", "link", "published"],
         exclusions=[],
         scan_interval=DEFAULT_SCAN_INTERVAL,
@@ -96,6 +97,7 @@ def test_update_sensor_with_topn(feed: FeedSource) -> None:
         date_format=DATE_FORMAT,
         local_time=False,
         show_topn=show_topn,
+        remove_summary_image=False,
         inclusions=["image", "title", "link", "published"],
         exclusions=[],
         scan_interval=DEFAULT_SCAN_INTERVAL,
@@ -172,6 +174,7 @@ def test_fetch_data_headers_required(online_feed: dict) -> None:
         date_format=DATE_FORMAT,
         local_time=False,
         show_topn=9999,
+        remove_summary_image=False,
         inclusions=["image", "title", "link", "published"],
         exclusions=[],
         scan_interval=DEFAULT_SCAN_INTERVAL,