From 5ec4699f6bd220c5947736ddb40f3eaaf328ee49 Mon Sep 17 00:00:00 2001
From: sam <wnabee@protonmail.com>
Date: Sat, 13 Aug 2022 15:53:08 +0200
Subject: [PATCH] Fixed dissapearing post bug. Simplified safescrape
 (identified some trivial bugs. Run through Safescrape to circumvent issues).

---
 Seleniumparser.py   | 42 +++++++++++++++++++++++-------------------
 posthistoryhash.txt |  6 ------
 safescrape.sh       |  6 +-----
 3 files changed, 24 insertions(+), 30 deletions(-)

diff --git a/Seleniumparser.py b/Seleniumparser.py
index b53b443..d8b0d46 100644
--- a/Seleniumparser.py
+++ b/Seleniumparser.py
@@ -245,7 +245,7 @@ def postformatter(pagebunch):
                     os.rename(f"html/{ImagePath}", f"html/{CheckedImagePath}")
                     ImagePaths.append(CheckedImagePath)
                     ImageCount+=1
-                    print(CheckedImagePath)
+                print(CheckedImagePath)
  
 
             TitleDate = post[0][1]
@@ -276,25 +276,29 @@ def postformatter(pagebunch):
                     postString = re.sub(Etext, markDownText, postString)
             
             
-            with open("posthistoryhash.txt", "r") as postHistory:
-                if hashedPost in postHistory.read():
-                    postHistory.close()
-                    for image in ImagePaths:
-                        os.remove(f"html/{image}")
-                    print("duplicate")
-                    continue
-                else:
-                    postHistory.close()
-            
-            with open("posthistoryhash.txt", "a") as postHistory:
-                 postHistory.write(f"\n{hashedPost}")
-                 postHistory.close()
-            
-            postString = f"![Profile Picture]({ImagePaths[0]}) ## {TitleDate}  \n {postString}" if ImagePaths else ""
+            postHistory = open("posthistoryhash.txt", "r")
+            if hashedPost in postHistory.read():
+                postHistory.close()
+                for image in ImagePaths:
+                    os.remove(f"html/{image}")
+                print("duplicate")
+            else:
+                postHistory.close()
             
-            with open(filePath, 'w') as postfile:
-                postfile.write(postString)
-            postfile.close()
+                with open("posthistoryhash.txt", "a") as postHistory:
+                     postHistory.write(f"\n{hashedPost}")
+                     postHistory.close()
+                
+                try:
+                    postString = f"![Profile Picture]({ImagePaths[0]}) ## {TitleDate}  \n {postString}"
+                except:
+                    print("ImagePaths empty")
+                    pass
+
+                
+                with open(filePath, 'w') as postfile:
+                    postfile.write(postString)
+                    postfile.close()
 
 def main():
     urllist = configparser()
diff --git a/posthistoryhash.txt b/posthistoryhash.txt
index 0ad3cb4..e69de29 100644
--- a/posthistoryhash.txt
+++ b/posthistoryhash.txt
@@ -1,6 +0,0 @@
-
-7233d60b5dd02ae920b6d1a50b0fd6886d5638f63e17c639ce06cb6160bd0471
-d8c6d6556313a50bf6f95adf6dc301f7674134b50aa40b86cf7b6eacafdc255d
-08f1f621d5bfe1b7a6148e200c8157d138dce68e7a3adcec246b178b36f7ad69
-6f1702a011871bbe48bdd9559f9bc9fe4a96d7810924c63bd26b74e22c6b220d
-e0f5b3d2901e39ca94f7a2e53475a8cde51546660426aaf06ce829f7d9ba0f4b
\ No newline at end of file
diff --git a/safescrape.sh b/safescrape.sh
index b548bd8..b6b01b6 100755
--- a/safescrape.sh
+++ b/safescrape.sh
@@ -7,9 +7,5 @@ do
 	cat pagelist.txt.template > pagelist.txt
 	echo "$Line" >> pagelist.txt
 	python3 Seleniumparser.py
-	~/cshg/cshg -d txt/
-	cp -r html/* ~/.liupeeker/
-	randomwaittime=$(echo $(( $RANDOM % 300 + 300 )))
-	echo "waiting for $randomwaittime"
-	sleep $randomwaittime
+	sleep 2
 done
-- 
GitLab