wiktextract: disable categories by default with option to enable it

ilius · Dec 18, 2024 · 3666cdd · 3666cdd
1 parent 26f13ed
commit 3666cdd
Show file tree

Hide file tree

Showing 3 changed files with 21 additions and 8 deletions.
diff --git a/doc/p/wiktextract.md b/doc/p/wiktextract.md
@@ -27,6 +27,7 @@
 | example_padding | `10px 20px`      | str  | Padding for examples (css value)               |
 | audio           | `True`           | bool | Enable audio                                   |
 | audio_formats   | `['ogg', 'mp3']` | list | List of audio formats to use                   |
+| categories      | `False`          | bool | Enable categories                              |
 
 ### Dependencies for reading
 

diff --git a/plugins-meta/index.json b/plugins-meta/index.json
@@ -1820,6 +1820,11 @@
 				"class": "ListOption",
 				"type": "list",
 				"comment": "List of audio formats to use"
+			},
+			"categories": {
+				"class": "BoolOption",
+				"type": "bool",
+				"comment": "Enable categories"
 			}
 		},
 		"canRead": true,
@@ -1833,7 +1838,8 @@
 			"audio_formats": [
 				"ogg",
 				"mp3"
-			]
+			],
+			"categories": false
 		},
 		"readDepends": {
 			"lxml": "lxml"

diff --git a/pyglossary/plugins/wiktextract.py b/pyglossary/plugins/wiktextract.py
@@ -78,6 +78,9 @@
 	"audio_formats": ListOption(
 		comment="List of audio formats to use",
 	),
+	"categories": BoolOption(
+		comment="Enable categories",
+	),
 }
 
 
@@ -98,6 +101,8 @@ class Reader:
 
 	_audio_formats: list[str] = ["ogg", "mp3"]
 
+	_categories: bool = False
+
 	topicStyle = (
 		"color:white;"
 		"background:green;"
@@ -167,7 +172,7 @@ def __iter__(self) -> Iterator[EntryType]:
 	def warning(self, msg: str) -> None:
 		self._warnings[msg] += 1
 
-	def makeEntry(self, data: dict[str, Any]) -> EntryType:
+	def makeEntry(self, data: dict[str, Any]) -> EntryType:  # noqa: PLR0912
 		from lxml import etree as ET
 
 		glos = self._glos
@@ -234,12 +239,13 @@ def br() -> Element:
 					with hf.element("div"):
 						hf.write(f"Etymology: {etymology}")
 
-				categories = []
-				for sense in senses:
-					senseCats = sense.get("categories")
-					if senseCats:
-						categories += senseCats
-				self.writeSenseCategories(hf_, categories)
+				if self._categories:
+					categories = []
+					for sense in senses:
+						senseCats = sense.get("categories")
+						if senseCats:
+							categories += senseCats
+					self.writeSenseCategories(hf_, categories)
 
 		defi = f.getvalue().decode("utf-8")
 		# defi = defi.replace("\xa0", "&nbsp;")  # do we need to do this?