Gemlog format fix; Cringe XML sitemap generation

README.md

@@ -36,7 +36,7 @@ Feel free to experiment with all of this stuff!
 - [ ] Pug support for base templates and page side parts
 - [ ] Differential recompile (to optimize resource waste on non-ephemeral servers)
 - [ ] Hot-recompile (for website development)
-- [ ] XML sitemap generation
+- [x] XML sitemap generation
 - [x] Atom + RSS feed generation for posts
 - [x] Generation of website page tree in left sidebar
 - [x] Generation of titles in right sidebar with clickable links

Source/Build.py

@@ -472,7 +472,7 @@ def GetConfMenu(Conf):
 	print(Menu)
 	return Menu
 
-def Main(Args, FeedEntries):
+def Main(Args, FeedEntries, SitemapOut):
 	HavePages, HavePosts = False, False
 	SiteConf = LoadConf('Site.ini')
 	#SiteMenu = GetConfMenu(SiteConf)
@@ -541,6 +541,19 @@ def Main(Args, FeedEntries):
 			SiteDomain=SiteDomain,
 			MaxEntries=FeedEntries,
 			Lang=SiteLang,
+			FullMap=False,
+			Minify=True if Args.Minify and Args.Minify not in ('False', 'None') else False)
+
+	if SitemapOut:
+		print("[I] Generating Sitemap")
+		MakeFeed(
+			Pages=Pages,
+			SiteName=SiteName,
+			SiteTagline=SiteTagline,
+			SiteDomain=SiteDomain,
+			MaxEntries=FeedEntries,
+			Lang=SiteLang,
+			FullMap=True,
 			Minify=True if Args.Minify and Args.Minify not in ('False', 'None') else False)
 
 	if ActivityPub and MastodonURL and MastodonToken and SiteDomain:
@@ -593,6 +606,7 @@ if __name__ == '__main__':
 	Parser.add_argument('--GemtextOut', type=bool)
 	Parser.add_argument('--GemtextHeader', type=str)
 	Parser.add_argument('--SiteTagline', type=str)
+	Parser.add_argument('--SitemapOut', type=bool)
 	Parser.add_argument('--FeedEntries', type=int)
 	Parser.add_argument('--FolderRoots', type=str)
 	Parser.add_argument('--ContextParts', type=str)
@@ -606,10 +620,13 @@ if __name__ == '__main__':
 		import lxml
 		from Modules.Feed import *
 		FeedEntries = Args.FeedEntries if Args.FeedEntries or Args.FeedEntries == 0 else 10
+		SitemapOut = True if Args.SitemapOut else False
 	except:
-		print("[E] Can't load the Atom/RSS feed libraries. Their generation is disabled. Make sure the 'lxml' library is installed.")
+		print("[E] Can't load the XML libraries. XML Feeds and Sitemaps generation is disabled. Make sure the 'lxml' library is installed.")
 		FeedEntries = 0
+		SitemapOut = False
 
 	Main(
 		Args=Args,
-		FeedEntries=FeedEntries)
+		FeedEntries=FeedEntries,
+		SitemapOut=SitemapOut)

Source/Modules/Feed.py

@@ -12,7 +12,7 @@
 from Libs.feedgen.feed import FeedGenerator
 from Modules.Utils import *
 
-def MakeFeed(Pages, SiteName, SiteTagline, SiteDomain, MaxEntries, Lang, Minify=False):
+def MakeFeed(Pages, SiteName, SiteTagline, SiteDomain, MaxEntries, Lang, FullMap=False, Minify=False):
 	Feed = FeedGenerator()
 	Link = SiteDomain if SiteDomain else ' '
 	Feed.id(Link)
@@ -24,21 +24,22 @@ def MakeFeed(Pages, SiteName, SiteTagline, SiteDomain, MaxEntries, Lang, Minify=
 	Feed.language(Lang)
 
 	DoPages = []
+	if FullMap:
+		MaxEntries = 50000 # Sitemap standard limit
 	for e in Pages:
-		if MaxEntries != 0 and e[3]['Type'] == 'Post':
+		if MaxEntries != 0 and (FullMap or (not FullMap and e[3]['Type'] == 'Post')):
 			DoPages += [e]
 			MaxEntries -= 1
 	DoPages.reverse()
 
 	for File, Content, Titles, Meta, ContentHTML, SlimHTML, Description, Image in DoPages:
-		if Meta['Type'] == 'Post':
+		if FullMap or (not FullMap and Meta['Type'] == 'Post'):
 			Entry = Feed.add_entry()
 			File = '{}.html'.format(StripExt(File))
 			Content = ReadFile('public/'+File)
 			Link = SiteDomain + '/' + File if SiteDomain else ' '
 			CreatedOn = GetFullDate(Meta['CreatedOn'])
 			EditedOn = GetFullDate(Meta['EditedOn'])
-
 			Entry.id(Link)
 			Entry.title(Meta['Title'] if Meta['Title'] else ' ')
 			Entry.description(Description)
@@ -49,6 +50,9 @@ def MakeFeed(Pages, SiteName, SiteTagline, SiteDomain, MaxEntries, Lang, Minify=
 			EditedOn = EditedOn if EditedOn else CreatedOn if CreatedOn and not EditedOn else '1970-01-01T00:00+00:00'
 			Entry.updated(EditedOn)
 
+	if FullMap:
+		Feed.atom_file('public/sitemap.xml', pretty=(not Minify))
+	else:
 		os.mkdir('public/feed')
 		Feed.atom_file('public/feed/atom.xml', pretty=(not Minify))
 		Feed.rss_file('public/feed/rss.xml', pretty=(not Minify))

Source/Modules/Gemini.py

@@ -31,6 +31,15 @@ def StripAttrs(HTML):
 			t.attrs = {}
 	return str(Soup)
 
+def FixGemlogDateLine(Line):
+	if len(Line) >= 2 and Line[0] == '[' and Line[1].isdigit():
+		Line = Line[1:]
+	else:
+		Words = Line.split(' ')
+		if len(Words) >= 2 and len(Words[1]) >= 2 and Words[1][0] == '[' and Words[1][1].isdigit():
+			Line = Words[0] + '\n' + Words[1][1:] + ' ' + ' '.join(Words[2:])
+	return Line
+
 def GemtextCompileList(Pages, Header=''):
 	Cmd = ''
 	for File, Content, Titles, Meta, ContentHTML, SlimHTML, Description, Image in Pages:
@@ -45,7 +54,11 @@ def GemtextCompileList(Pages, Header=''):
 	os.system(Cmd)
 	for File, Content, Titles, Meta, ContentHTML, SlimHTML, Description, Image in Pages:
 		Dst = 'public.gmi/{}.gmi'.format(StripExt(File))
-		WriteFile(Dst, Header + ReadFile(Dst))
+		Gemtext = ''
+		for Line in ReadFile(Dst).splitlines():
+			Line = FixGemlogDateLine(Line)
+			Gemtext += Line + '\n'
+		WriteFile(Dst, Header + Gemtext)
 
 def FindEarliest(Str, Items):
 	Pos, Item = 0, ''