From a12c38924fab0e5bab568fdde3c92199be1add33 Mon Sep 17 00:00:00 2001
From: "octo@swiss" <octospacc@gitlab.com>
Date: Mon, 22 Aug 2022 23:19:33 +0200
Subject: [PATCH] MastodonFeedHTML: Ordered recurse into previus pages

---
 MastodonFeedHTML/Example.Config.py   |  9 +++++--
 MastodonFeedHTML/MastodonFeedHTML.py | 35 +++++++++++++++++++++-------
 2 files changed, 33 insertions(+), 11 deletions(-)

diff --git a/MastodonFeedHTML/Example.Config.py b/MastodonFeedHTML/Example.Config.py
index 6dfab9d..722d9ab 100644
--- a/MastodonFeedHTML/Example.Config.py
+++ b/MastodonFeedHTML/Example.Config.py
@@ -19,8 +19,13 @@ MailPort = 465
 # How often to refresh the feeds (in seconds). Set to 0 for a single run, instead of having the program sleep.
 LoopTime = 300
 
-# Additional time (in seconds) to sleep between every mail sent, to prevent spam.
-MailSleep = 10
+# Additional time (in seconds) to sleep in code sections, to prevent ratelimiting.
+PageSleep = 2 # Between every scraped page
+ItemSleep = 1 # Between every scaped item
+MailSleep = 9 # Between every sent mail
+
+# Stop recursive navigation across posts pages if limit is reached. Set 0 for no limit (use with caution on new profiles with many posts).
+MaxPagesRecursion = 10
 
 # Whether or not to allow spaces in file names.
 SpacesInFiles = True
diff --git a/MastodonFeedHTML/MastodonFeedHTML.py b/MastodonFeedHTML/MastodonFeedHTML.py
index 44ec853..afe8c2e 100755
--- a/MastodonFeedHTML/MastodonFeedHTML.py
+++ b/MastodonFeedHTML/MastodonFeedHTML.py
@@ -21,6 +21,10 @@ def MakePathStr(Str):
 		Str = Str.replace(' ', '_')
 	return Str
 
+def SleepPrint(s):
+	print(f"[I] Sleeping for {s}s...")
+	time.sleep(s)
+
 def HandleFeedsList(List):
 	for Feed in List:
 		print(f"[I] Handling Feed ->\n: {Feed}")
@@ -38,14 +42,23 @@ def HandleFeed(URLs, IncludeRetoots, IncludeReplies, LocalSave, SendMail, MailTo
 			print("[I] Partial dry-run for this URL (LocalSave and SendMail are disabled).")
 		URL = URL.removesuffix('/').removesuffix('/with_replies') + '/with_replies'
 		Usertag = f"{URL.split('/')[-2]}@{URL.split('/')[-3]}"
-		LastEntryIsNew, PageOlder = HandleURL(URL, Usertag, IncludeRetoots, IncludeReplies, LocalSave, SendMail, MailTo)
-		while LastEntryIsNew and PageOlder: # TODO: Fix this, make a single run for all items / start from oldest, otherwise order is messed up
-			LastEntryIsNew, PageOlder = HandleURL(PageOlder, Usertag, IncludeRetoots, IncludeReplies, LocalSave, SendMail, MailTo)
+		Pages = []
+		LastEntryIsNew, PageOlder = HandleURL(True, URL, Usertag, IncludeRetoots, IncludeReplies, LocalSave, SendMail, MailTo)
+		if LastEntryIsNew and PageOlder:
+			Pages += [PageOlder]
+		while LastEntryIsNew and PageOlder and len(Pages) < MaxPagesRecursion:
+			LastEntryIsNew, PageOlder = HandleURL(True, PageOlder, Usertag, IncludeRetoots, IncludeReplies, LocalSave, SendMail, MailTo)
+			if LastEntryIsNew and PageOlder and MaxPagesRecursion:
+				Pages += [PageOlder]
+		Pages.reverse()
+		for Page in Pages:
+			HandleURL(False, Page, Usertag, IncludeRetoots, IncludeReplies, LocalSave, SendMail, MailTo)
 
-def HandleURL(URL, Usertag, IncludeRetoots, IncludeReplies, LocalSave, SendMail, MailTo):
+def HandleURL(IsFirstRun, URL, Usertag, IncludeRetoots, IncludeReplies, LocalSave, SendMail, MailTo):
 	LastEntryIsNew = False
 	PageOlder = ''
 	try:
+		print(f"-> Page: {URL}")
 		Response = urlopen(Request(URL, headers={'User-Agent':UserAgent}))
 		Data = Response.read()
 		Soup = BeautifulSoup(Data, 'html.parser')
@@ -73,6 +86,9 @@ def HandleURL(URL, Usertag, IncludeRetoots, IncludeReplies, LocalSave, SendMail,
 
 			if Index == 1:
 				LastEntryIsNew = True
+			if IsFirstRun:
+				break
+
 			LocalId = GlobalId.split('/')[-1]
 			Username = Entry.find('a', class_='status__display-name').get_text().strip()
 			Content = Entry.find('div', class_='e-content')
@@ -90,7 +106,7 @@ def HandleURL(URL, Usertag, IncludeRetoots, IncludeReplies, LocalSave, SendMail,
 			Entry.find('img', class_='u-photo account__avatar')['style'] = 'max-height:4em;' # Profile pics
 			Entry.find('div', class_='status__action-bar').replace_with('')
 
-			print(f"-> {LocalId} - {Title}")
+			print(f"-> Item: {LocalId} - {Title}")
 			HTML = f"""\
 <h1>{Title}</h1>
 
@@ -143,8 +159,7 @@ def HandleURL(URL, Usertag, IncludeRetoots, IncludeReplies, LocalSave, SendMail,
 				with smtplib.SMTP_SSL(MailServer, MailPort, context=ssl.create_default_context()) as Client:
 					Client.login(MailUsername, MailPassword)
 					Client.sendmail(MailUsername, MailTo, Message.as_string())
-				time.sleep(MailSleep)
-
+				SleepPrint(MailSleep)
 			if LocalSave:
 				LocalBackupDir = MakePathStr(Usertag)
 				if not os.path.isdir(LocalBackupDir):
@@ -156,6 +171,9 @@ def HandleURL(URL, Usertag, IncludeRetoots, IncludeReplies, LocalSave, SendMail,
 			with open(f'{AppName}.db', 'a') as Db:
 				Db.write(f'{Usertag} {GlobalId}' + '\n')
 
+			SleepPrint(ItemSleep)
+		SleepPrint(PageSleep)
+
 		return LastEntryIsNew, PageOlder
 
 	except Exception:
@@ -167,5 +185,4 @@ if __name__ == '__main__':
 		HandleFeedsList(Feeds)
 		if LoopTime <= 0:
 			exit()
-		print(f"[I] Sleeping for {LoopTime}s...")
-		time.sleep(LoopTime)
+		SleepPrint(LoopTime)