From 7d828a7c3bf3bc4f09f9b228b86bae2f7a1bc913 Mon Sep 17 00:00:00 2001
From: Marcel Gansfusz <marcel.gansfusz@gmail.com>
Date: Tue, 21 Oct 2025 14:48:06 +0200
Subject: [PATCH] modified the init file to work on local directory; added
 logging to init.py

---
 .gitignore                           |   1 +
 app/__pycache__/main.cpython-313.pyc | Bin 31535 -> 31472 bytes
 app/init.py                          |  81 ++++++++------
 app/init_ssh.py                      | 152 +++++++++++++++++++++++++++
 app/main.py                          |   3 +-
 5 files changed, 200 insertions(+), 37 deletions(-)
 create mode 100644 app/init_ssh.py

diff --git a/.gitignore b/.gitignore
index 20fbbdc..585c4f0 100644
--- a/.gitignore
+++ b/.gitignore
@@ -3,4 +3,5 @@ app/files/
 app/pwfile.json
 app/dest
 app.log
+init.log
 app/__pycache__/
diff --git a/app/__pycache__/main.cpython-313.pyc b/app/__pycache__/main.cpython-313.pyc
index 25c1d4633886ba930a3687795d578ab3d34e705b..95361c3073dd414dafba9257ff590ee7057b2035 100644
GIT binary patch
delta 464
zcmZ4gjq$@*M&8f7yj%=G;2iQTLu@0jk0oQ><S5HCs=<k6sl^H<8L0|snK`KnCHY`>
zdS+Q_o`OqiaY;d9Nrv8LRVz<sMzziPwx1ap#V04*UuBfuZ0V4~$f&V-o}(2bqv7P6
zPPvSIo86rgnOH)_f)poj^pTq6$T?ZuEs7DwVPj%o@O5OE{N7c5^DehbjEq8)dp(v;
zHupTjZV!|y{=&GK$t#F)a*lTvqx$6Q-cG_&49BFn+*z29sYtuoGas{Ob!XYE<0Hn*
zShU&M-=2lBb8~ZW0~6zt%`9PtY>XQ>o5dVqV!W~WcbqH}<NeK&iHjH+IX3S}TFAsG
zyg4rQI5VT>X0^=UOpG?0m2#4q7~MCw<%uvcMr@v4pu{2U%g88ogNyeAGl=nkPw)e)
z@8k;=5?Ubc2Mz{a!3iQ0l4qDs$hs(~c#%ix11DI53n=l08_Z0YEL2$@ZNMmfgOle2
zGmw4&G>sKVe_&$(ve-c^pg95`I6*8fkQg_J#lrx!j2Fb>W8mTczz<>xfLMYcmJkCQ
P=NDlRD{b?WN^eF0a%+tj

delta 497
zcmezHm2v$yM&8f7yj%=G;AHqMLuMndk0oQo<S5HCn|-W2m>E?zPqqEb$S5|s)BY->
z)aH1H6h=n%%|{%q7#R&F|98q|ytX;lIgv>vS`lOf2!x86GD27i49KEE%9A(nNbw;g
zL&bnX0dC5Zzq&>-f>^qYsz4>On`gV7XJizfob9ocwP+L2sL2AJN7(I{7#NCqm^R<`
z3}TeH!@|)a(c%ZDu5(IW<dnS3DSd@SX0on#7Nhp$8QxC9b_~btxZE|Ek8?=7D>ENg
zW_8!t{MK8HnXz~?i@!Y!W7lSv;07kfrJIk28nQ8N+WbBG2ovMY&3ocxnHV2zzLl_u
zk&$zAZPG#}Mv=|(smGZawKhM`_|3#<yZLc;G83c6X7@Z1CdSCkF$GE-!aj_QQa8AG
zKQMzB5BLN>u=-4%SRtVW;(p*@;1!%8G9h_}>4dC{f{GV;ls<5RCAfeRU%0`{xXG6*
z%A<`KrEhTZd|(FB4}hkz0_hKI3_un;hy^r9-~%U!#RU@M2C;Y;fR^!sSbPjT{2%y1
ZECCQp5X2H<VB`EE3}R(%URvqR2mqb~jw}EG

diff --git a/app/init.py b/app/init.py
index 50a598f..3a1feaa 100644
--- a/app/init.py
+++ b/app/init.py
@@ -1,10 +1,12 @@
-import paramiko
+from os.path import isdir
 from stat import S_ISDIR, S_ISREG
 import re
+import pathlib
 
 # from base64 import decodebytes
 import json
 import mariadb
+import logging
 
 CATEGORIES = [
     "Prüfungen",
@@ -16,7 +18,18 @@ CATEGORIES = [
     "Multimedia",
 ]
 SUBCAT_CATEGORIES = ["Klausuren", "Übungen", "Labore"]
-unizeug_path = "/mnt/save/daten/Unizeug/"
+unizeug_path = "/home/wildarch/web/fet_unizeug/unizeug/"
+
+log = logging.getLogger(__name__)
+logging.basicConfig(
+    filename="init.log",
+    level=logging.INFO,
+    format="[%(asctime)s, %(filename)s:%(lineno)s -> %(funcName)10s() ]%(levelname)s: %(message)s",
+)
+debug = log.debug
+info = log.info
+error = log.error
+
 db = mariadb.connect(
     host="localhost", user="wildserver", password="DBPassword", database="Unizeug"
 )
@@ -53,29 +66,31 @@ db.commit()
 
 
 def get_dirstruct():
-    with open("app/pwfile.json", "r") as f:
-        cred = json.load(f)
-    ssh = paramiko.SSHClient()
-    print(cred["sftpurl"])
-    ssh.set_missing_host_key_policy(paramiko.AutoAddPolicy())
+    # with open("app/pwfile.json", "r") as f:
+        # cred = json.load(f)
+    # ssh = paramiko.SSHClient()
+    # print(cred["sftpurl"])
+    # ssh.set_missing_host_key_policy(paramiko.AutoAddPolicy())
     # key=paramiko.RSAKey(data=decodebytes(bytes(cred["key"],"utf-8")))
     # ssh.get_host_keys().add(cred["sftpurl"], 'ssh-rsa', key)
-    ssh.connect(cred["sftpurl"], username=cred["sftpuser"], password=cred["sftpPW"])
-    sftp = ssh.open_sftp()
-    folders = sftp.listdir_attr(unizeug_path)
-    for entry in folders:
+    # ssh.connect(cred["sftpurl"], username=cred["sftpuser"], password=cred["sftpPW"])
+    # sftp = ssh.open_sftp()
+    # folders = sftp.listdir_attr(unizeug_path)
+    folders=pathlib.Path(unizeug_path)
+    for entry in folders.iterdir():
         if entry is None:
             continue
-        if not S_ISDIR(entry.st_mode):
+        if not entry.is_dir():
             continue
-        fname = str(entry.filename)
+        fname = str(entry.name)
         regex = re.compile(r"Multimedia_only")
         if regex.search(fname):
             continue
         # print(fname)
         lvid = re.search(r"[a-zA-Z0-9]{3}\.[a-zA-Z0-9]{3}", fname)
-        print(lvid)
+        # print(lvid)
         if lvid is None:
+            error(f"Didnt Find LVA ID in Directory {fname}")
             continue
         lvid = lvid.group()[:3] + lvid.group()[4:]
         # name = fname[:-8]
@@ -89,39 +104,35 @@ def get_dirstruct():
         cur.execute("SELECT id FROM LVAs WHERE lvid=?", (lvid,))
         lid = cur.fetchone()[0]
         db.commit()
-        for profsdir in sftp.listdir_attr(unizeug_path + fname + "/"):
-            if profsdir is None or not S_ISDIR(profsdir.st_mode):
+        for profsdir in entry.iterdir():
+            if profsdir is None:
+                continue
+            if not profsdir.is_dir():
                 continue
             # print(profsdir.filename)
             try:
-                lastname, firstname = re.split(r"[_\-\s]", str(profsdir.filename))
+                lastname, firstname = re.split(r"[_\-\s]", str(profsdir.name))
                 pid = link_prof(firstname, lastname, lid)
             except ValueError:
-                print(f"{name} is broken")
+                error(f"Couldnt get Profs from {fname}")
                 continue
-            for cat in sftp.listdir_attr(
-                unizeug_path + fname + "/" + profsdir.filename + "/"
-            ):
-                if cat is None or not S_ISDIR(cat.st_mode):
+            for cat in profsdir.iterdir():
+                if cat is None:
                     continue
-                if cat.filename not in SUBCAT_CATEGORIES:
+                if not cat.is_dir():
                     continue
-                idx = CATEGORIES.index(cat.filename)
-                for subcat in sftp.listdir_attr(
-                    unizeug_path
-                    + fname
-                    + "/"
-                    + profsdir.filename
-                    + "/"
-                    + cat.filename
-                    + "/"
-                ):
-                    if subcat is None or not S_ISDIR(subcat.st_mode):
+                if cat.name not in SUBCAT_CATEGORIES:
+                    continue
+                idx = CATEGORIES.index(cat.name)
+                for subcat in cat.iterdir():
+                    if subcat is None:
+                        continue
+                    if not subcat.is_dir():
                         continue
                     cur = db.cursor()
                     cur.execute(
                         "INSERT INTO SubCats (LId,PId,cat,name) VALUES(?,?,?,?)",
-                        (lid, pid, idx, subcat.filename),
+                        (lid, pid, idx, subcat.name),
                     )
                 db.commit()
 
diff --git a/app/init_ssh.py b/app/init_ssh.py
new file mode 100644
index 0000000..50a598f
--- /dev/null
+++ b/app/init_ssh.py
@@ -0,0 +1,152 @@
+import paramiko
+from stat import S_ISDIR, S_ISREG
+import re
+
+# from base64 import decodebytes
+import json
+import mariadb
+
+CATEGORIES = [
+    "Prüfungen",
+    "Klausuren",
+    "Übungen",
+    "Labore",
+    "Unterlagen",
+    "Zusammenfassungen",
+    "Multimedia",
+]
+SUBCAT_CATEGORIES = ["Klausuren", "Übungen", "Labore"]
+unizeug_path = "/mnt/save/daten/Unizeug/"
+db = mariadb.connect(
+    host="localhost", user="wildserver", password="DBPassword", database="Unizeug"
+)
+c = db.cursor()
+try:
+    c.execute("DROP TABLE LVAs")
+except mariadb.OperationalError:
+    pass
+c.execute(
+    "CREATE TABLE LVAs(id BIGINT(20) unsigned NOT NULL AUTO_INCREMENT,lvid VARCHAR(6), lvname VARCHAR(256), lvpath VARCHAR(256),PRIMARY KEY(id))"
+)
+try:
+    c.execute("DROP TABLE Profs")
+except mariadb.OperationalError:
+    pass
+c.execute(
+    "CREATE TABLE Profs(id BIGINT(20) unsigned NOT NULL AUTO_INCREMENT,name VARCHAR(256),PRIMARY KEY(id))"
+)
+try:
+    c.execute("DROP TABLE LPLink")
+except mariadb.OperationalError:
+    pass
+c.execute(
+    "CREATE TABLE LPLink(id BIGINT(20) unsigned NOT NULL AUTO_INCREMENT,LId bigint(20),PId bigint(20),PRIMARY KEY(id))"
+)
+try:
+    c.execute("DROP TABLE SubCats")
+except mariadb.OperationalError:
+    pass
+c.execute(
+    "CREATE TABLE SubCats(id BIGINT(20) UNSIGNED NOT NULL AUTO_INCREMENT,LId BIGINT(20),PId BIGINT(20),cat TINYINT UNSIGNED,name VARCHAR(256), PRIMARY KEY(id))"
+)
+db.commit()
+
+
+def get_dirstruct():
+    with open("app/pwfile.json", "r") as f:
+        cred = json.load(f)
+    ssh = paramiko.SSHClient()
+    print(cred["sftpurl"])
+    ssh.set_missing_host_key_policy(paramiko.AutoAddPolicy())
+    # key=paramiko.RSAKey(data=decodebytes(bytes(cred["key"],"utf-8")))
+    # ssh.get_host_keys().add(cred["sftpurl"], 'ssh-rsa', key)
+    ssh.connect(cred["sftpurl"], username=cred["sftpuser"], password=cred["sftpPW"])
+    sftp = ssh.open_sftp()
+    folders = sftp.listdir_attr(unizeug_path)
+    for entry in folders:
+        if entry is None:
+            continue
+        if not S_ISDIR(entry.st_mode):
+            continue
+        fname = str(entry.filename)
+        regex = re.compile(r"Multimedia_only")
+        if regex.search(fname):
+            continue
+        # print(fname)
+        lvid = re.search(r"[a-zA-Z0-9]{3}\.[a-zA-Z0-9]{3}", fname)
+        print(lvid)
+        if lvid is None:
+            continue
+        lvid = lvid.group()[:3] + lvid.group()[4:]
+        # name = fname[:-8]
+        name = re.sub(r"[a-zA-Z0-9]{3}\.[a-zA-Z0-9]{3}", "", fname)
+        # print(name)
+        # print(lvid)
+        cur = db.cursor()
+        cur.execute(
+            "INSERT INTO LVAs (lvid, lvname, lvpath) VALUES(?,?,?)", (lvid, name, fname)
+        )
+        cur.execute("SELECT id FROM LVAs WHERE lvid=?", (lvid,))
+        lid = cur.fetchone()[0]
+        db.commit()
+        for profsdir in sftp.listdir_attr(unizeug_path + fname + "/"):
+            if profsdir is None or not S_ISDIR(profsdir.st_mode):
+                continue
+            # print(profsdir.filename)
+            try:
+                lastname, firstname = re.split(r"[_\-\s]", str(profsdir.filename))
+                pid = link_prof(firstname, lastname, lid)
+            except ValueError:
+                print(f"{name} is broken")
+                continue
+            for cat in sftp.listdir_attr(
+                unizeug_path + fname + "/" + profsdir.filename + "/"
+            ):
+                if cat is None or not S_ISDIR(cat.st_mode):
+                    continue
+                if cat.filename not in SUBCAT_CATEGORIES:
+                    continue
+                idx = CATEGORIES.index(cat.filename)
+                for subcat in sftp.listdir_attr(
+                    unizeug_path
+                    + fname
+                    + "/"
+                    + profsdir.filename
+                    + "/"
+                    + cat.filename
+                    + "/"
+                ):
+                    if subcat is None or not S_ISDIR(subcat.st_mode):
+                        continue
+                    cur = db.cursor()
+                    cur.execute(
+                        "INSERT INTO SubCats (LId,PId,cat,name) VALUES(?,?,?,?)",
+                        (lid, pid, idx, subcat.filename),
+                    )
+                db.commit()
+
+
+def link_prof(firstname, lastname, lid):
+    cur = db.cursor()
+    cur.execute("SELECT id from Profs WHERE name=?", (lastname + " " + firstname,))
+    res = cur.fetchone()
+    if res is not None:
+        cur.execute("INSERT INTO LPLink (LId,PId) VALUES(?,?)", (lid, res[0]))
+        db.commit()
+        return res[0]
+    cur.execute("SELECT id from Profs WHERE name=?", (firstname + " " + lastname,))
+    res = cur.fetchone()
+    if res is not None:
+        cur.execute("INSERT INTO LPLink (LId,PId) VALUES(?,?)", (lid, res[0]))
+        db.commit()
+        return res[0]
+    cur.execute("INSERT INTO Profs (name) VALUES(?)", (lastname + " " + firstname,))
+    cur.execute("SELECT id FROM Profs WHERE name=?", (lastname + " " + firstname,))
+    res = cur.fetchone()
+    cur.execute("INSERT INTO LPLink (LId,PId) VALUES(?,?)", (lid, res[0]))
+    db.commit()
+    return res[0]
+
+
+if __name__ == "__main__":
+    get_dirstruct()
diff --git a/app/main.py b/app/main.py
index bc4ef07..78118a8 100644
--- a/app/main.py
+++ b/app/main.py
@@ -461,6 +461,7 @@ async def censor_pdf(
 ):
     """Censors pdf and runs OCR
     If Secure is True the file is converted to Pixels and then recreated; else the censored sections are just covering the text below and can be easiliy removed with e.g. Inkscape
+    Saves the file to the given Destpath.
     Args:
         path: path to the pdf document
         destpath: Path where the result is supposed to be saved to
@@ -614,8 +615,6 @@ def make_savepath(
 
     destpath = pathlib.Path(savepath + file)
     i = 0
-    info(destpath)
-    info(destpath.is_file())
     while destpath.is_file():
         file = filename + f"_{i}." + ftype
         i += 1