next driver

cron.sh

@@ -1,10 +1,10 @@
 # This will pump all the pads with a __PUBLISH__ tag into a folder "publish" as meta.json, txt, html and dhtml
-poetry run etherpump pull --meta --html --text --publish-opt-in --publish __PUBLISH__ --pub publish --css ../stylesheet.css --fix-names
+etherpump pull --meta --html --text --publish-opt-in --publish __PUBLISH__ --pub publish --css ../stylesheet.css --fix-names
 
 echo "Making the Etherpump index now ..."
 
 # This will make an index for the dump
-poetry run etherpump index input \
+etherpump index input \
 	publish/*.meta.json \
 	--templatepath templates \
 	--title "Notes, __MAGICWORDS__, readers & more ..." \

etherpump/commands/pull.py

@@ -171,7 +171,7 @@ def build_argument_parser(args):
 		help="ensure `--publish` is honoured instead of `--nopublish`",
 	)
 	parser.add_argument(
-		"--magic-words",
+		"--magicwords",
 		default=False,
 		action="store_true",
 		help="store all magic words used in a page in the meta.json file",
@@ -375,7 +375,7 @@ async def handle_pad(args, padid, data, info, session):
 
 
 			#  include magic words
-			if args.magic_words:
+			if args.magicwords:
 
 				##########################################
 				## INCLUDE __XXX__ MAGIC WORDS
@@ -383,6 +383,7 @@ async def handle_pad(args, padid, data, info, session):
 				pattern = r'__[a-zA-Z0-9]+?__'
 				magic_words = re.findall(pattern, text)
 				if magic_words:
+					meta["magicwords"] = magic_words
 					print('FOUND MAGIC WORD(s): {} in {}'.format(magic_words, padid))
 
 	links = []
@@ -495,6 +496,33 @@ async def handle_pad(args, padid, data, info, session):
 		# except Exception as exception:
 			# print("PANIC: {}".format(exception))
 
+	if args.all or args.magicwords:
+		url = info["localapiurl"] + "getHTML?" + "padID=" + padid + "&" + 'apikey=' + data["apikey"]
+		# print(url)
+		html = await agetjson(session, url)
+		ver = {"type": "magicwords"}
+		versions.append(ver)
+		# mb: warning, HACK! Added a Try and Except here, as it sometimes bumped into an error, stopping the pull.
+		# try:
+		ver["code"] = html["_code"]
+		if html["_code"] == 200:
+				try:
+						html = html["data"]["html"]
+						ver["path"] = p + ".magicwords.html"
+						ver["url"] = quote(ver["path"])
+						doc = html5lib.parse(
+								html, treebuilder="etree", namespaceHTMLElements=False
+						)
+		# INSERT MAGIC WORDS HIGHLIGHTING STUFF HERE!!!
+						html5tidy(
+								doc, indent=True, title=padid, scripts=args.script, links=links,
+						)
+						async with await trio.open_file(ver["path"], "w") as f:
+								output = ET.tostring(doc, method="html", encoding="unicode")
+								await f.write(output)
+				except TypeError:
+						ver["message"] = html["message"]
+
 	# output meta
 	if args.all or args.meta:
 		ver = {"type": "meta"}