fixed error in compose.yml

Reviewed-on: #1

.gitignore

@@ -5,3 +5,7 @@ app/dest
 app.log
 init.log
 app/__pycache__/
+mariadb/*
+unizeug
+.mypy_cache
+.nvim

.nvim/rsync.toml_alt

@@ -1,2 +0,0 @@
-# remote_path="/srv/http/"
-# remote_path="dev@10.0.0.25:/var/www/html/"

Dockerfile

@@ -0,0 +1,31 @@
+FROM python:3.13-rc-alpine
+WORKDIR /usr/src/
+COPY requirements.txt /usr/src/requirements.txt
+COPY entrypoint.sh /usr/src/entrypoint.sh
+RUN apk add --no-cache \
+    gcc \
+    g++ \
+    musl-dev \
+    python3-dev \
+    libffi-dev \
+    openssl-dev \
+    cargo \
+    make \
+    mariadb-connector-c-dev \
+    jpeg-dev \
+    zlib-dev \
+    freetype-dev \
+    lcms2-dev \
+    openjpeg-dev \
+    tiff-dev \
+    tk-dev \
+    tcl-dev \
+    libwebp-dev \
+    tesseract-ocr \
+    tesseract-ocr-data-deu
+RUN python -m ensurepip --upgrade
+RUN pip install setuptools wheel
+RUN pip install -r requirements.txt
+WORKDIR /python
+CMD /bin/sh /usr/src/entrypoint.sh
+# ENTRYPOINT ["/usr/src/entrypoint.sh"]

app/init.py

@@ -2,11 +2,16 @@ from os.path import isdir
 from stat import S_ISDIR, S_ISREG
 import re
 import pathlib
+import os
 
 # from base64 import decodebytes
 import json
 import mariadb
 import logging
+from pathlib import Path
+import schedule
+import time
+import pytz
 
 CATEGORIES = [
     "Prüfungen",
@@ -18,8 +23,9 @@ CATEGORIES = [
     "Multimedia",
 ]
 SUBCAT_CATEGORIES = ["Klausuren", "Übungen", "Labore"]
-unizeug_path = "/home/wildarch/web/fet_unizeug/unizeug/"
+unizeug_path = os.environ.get("UNIZEUG_PATH", "./unizeug")
-
+APP_ROOT_PATH = Path(os.environ.get("APP_ROOT_PATH", "./app"))
+FILES_IN_PROGRESS = APP_ROOT_PATH / "files/"
 log = logging.getLogger(__name__)
 logging.basicConfig(
     filename="init.log",
@@ -31,7 +37,10 @@ info = log.info
 error = log.error
 
 db = mariadb.connect(
-    host="localhost", user="wildserver", password="DBPassword", database="Unizeug"
+    host=os.environ.get("DB_HOST", "db"),
+    user=os.environ.get("DB_USER", "user"),
+    password=os.environ.get("DB_PASSWORD", "DBPASSWORD"),
+    database=os.environ.get("DB_DATABASE", "unizeug"),
 )
 c = db.cursor()
 try:
@@ -62,9 +71,28 @@ except mariadb.OperationalError:
 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))"
 )
+try:
+    c.execute(
+        "CREATE TABLE FIP(id UUID DEFAULT(UUID()), filename VARCHAR(256), filetype VARCHAR(8),initTimeStamp DATETIME, PRIMARY KEY(id))"
+    )
+except mariadb.OperationalError:
+    pass
 db.commit()
 
 
+def remove_old_FIP_entrys():
+    cur = db.cursor(dictionary=True)
+    cur.execute(
+        "SELECT id,filename FROM FIP WHERE HOUR(TIMEDIFF(NOW(),initTimeStamp)) > 24 "
+    )
+    files = cur.fetchall()
+    info(f"Remove Files: {files}")
+    for file in files:
+        c.execute("DELETE FROM FIP WHERE id=?", (file["id"],))
+        os.remove(FILES_IN_PROGRESS / file["filename"])
+    db.commit()
+
+
 def get_dirstruct():
     # with open("app/pwfile.json", "r") as f:
     # cred = json.load(f)
@@ -76,7 +104,7 @@ def get_dirstruct():
     # 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)
+    folders = pathlib.Path(unizeug_path)
     for entry in folders.iterdir():
         if entry is None:
             continue
@@ -135,6 +163,7 @@ def get_dirstruct():
                         (lid, pid, idx, subcat.name),
                     )
                 db.commit()
+    remove_old_FIP_entrys()
 
 
 def link_prof(firstname, lastname, lid):
@@ -161,3 +190,8 @@ def link_prof(firstname, lastname, lid):
 
 if __name__ == "__main__":
     get_dirstruct()
+    info("Database updated")
+    schedule.every().day.at("04:00", "Europe/Vienna").do(get_dirstruct)
+    while True:
+        schedule.run_pending()
+        time.sleep(1)

app/main.py

@@ -21,30 +21,42 @@ import re
 
 
 import os
+import signal
 import mariadb
+import sys
 
 import filetype
 
-import datetime
 
 import logging
-import inspect
 import pathlib
+from pathlib import Path
 
 from starlette.types import HTTPExceptionHandler
 
 log = logging.getLogger(__name__)
 logging.basicConfig(
-    filename="app.log",
+    filename=os.environ.get("APP_LOG_PATH"),
     level=logging.INFO,
-    format="[%(asctime)s, %(filename)s:%(lineno)s -> %(funcName)10s() ]%(levelname)s: %(message)s",
+    format="[%(asctime)s, %(filename)s:%(lineno)s -> %(funcName)10s()] %(levelname)s: %(message)s",
 )
 debug = log.debug
 info = log.info
 error = log.error
+critical = log.critical
+
+
+def exception_handler(etype, value, tb):
+    log.exception(f"Uncought Exception: {value}")
+
+
+sys.excepthook = exception_handler
 
 db = mariadb.connect(
-    host="localhost", user="wildserver", password="DBPassword", database="Unizeug"
+    host=os.environ.get("DB_HOST", "db"),
+    user=os.environ.get("DB_USER", "user"),
+    password=os.environ.get("DB_PASSWORD", "DBPASSWORD"),
+    database=os.environ.get("DB_DATABASE", "unizeug"),
 )
 
 info("App Started")
@@ -53,8 +65,6 @@ info("App Started")
 
 # startup()
 app = FastAPI()
-app.mount("/favicon", StaticFiles(directory="./favicon"), name="favicon")
-app.mount("/static", StaticFiles(directory="./static"), name="static")
 
 
 CATEGORIES = [
@@ -66,15 +76,18 @@ CATEGORIES = [
     "Zusammenfassungen",
     "Multimedia",
 ]
+APP_ROOT_PATH = Path(os.environ.get("APP_ROOT_PATH", "./app"))
 SUBCAT_CATEGORIES = ["Klausuren", "Übungen", "Labore"]
 SUBCAT_CATEGORIES_I = [1, 2, 3]
 EX_DATE_CATEGORIES = ["Prüfungen", "Klausuren"]
 EX_DATE_CATEGORIES_I = [0, 1]
-UNIZEUG_PATH = "./app/dest/"
+UNIZEUG_PATH = Path(os.environ.get("UNIZEUG_PATH", "./app/dest"))
-FILES_IN_PROGRESS = "./app/files/"
+FILES_IN_PROGRESS = APP_ROOT_PATH / "files/"
-EMPTYFILE = "./app/graphics/empty.pdf"
+EMPTYFILE = APP_ROOT_PATH / "graphics/empty.pdf"
-UNSUPPORTEDFILE = "./app/graphics/unsupported.pdf"
+UNSUPPORTEDFILE = APP_ROOT_PATH / "graphics/unsupported.pdf"
-GREETINGFILE = "./app/graphics/greeting.pdf"
+GREETINGFILE = APP_ROOT_PATH / "graphics/greeting.pdf"
+FAVICON = APP_ROOT_PATH / "favicon"
+STATIC_FILES = APP_ROOT_PATH / "static"
 
 
 # cur = db.cursor()
@@ -114,12 +127,40 @@ def _sql_quarry(
         )
 
 
+def sql_connector_is_active(connector: mariadb.Connection) -> bool:
+    try:
+        connector.ping()
+    except mariadb.Error as e:
+        return False
+    return True
+
+
+def sql_connect(connector: mariadb.Connection) -> mariadb.Connection:
+    try:
+        connector = mariadb.connect(
+            host=os.environ.get("DB_HOST", "db"),
+            user=os.environ.get("DB_USER", "user"),
+            password=os.environ.get("DB_PASSWORD", "DBPASSWORD"),
+            database=os.environ.get("DB_DATABASE", "Unizeug"),
+        )
+    except mariadb.Error as e:
+        critical(
+            f"Cannot reconnect to Database {os.environ.get('DB_DATABASE', 'Unizeug')} on {os.environ.get('DB_HOST', 'db')}. Got Mariadb Error: {e}"
+        )
+        os.kill(os.getpid(), signal.SIGTERM)
+        raise HTTPException(500, detail="Database failed")
+    return connector
+
+
 def sql(
     querry: str,
     data: Tuple[str | int, ...] | str | int = (),
     return_result: bool = True,
     commit: bool = False,
 ) -> List[Tuple]:
+    global db
+    if not sql_connector_is_active(db):
+        db = sql_connect(db)
     cur = db.cursor(dictionary=False)
     return _sql_quarry(cur, querry, data, return_result, commit)
 
@@ -130,6 +171,10 @@ def sqlT(
     return_result: bool = True,
     commit: bool = False,
 ) -> List[Dict]:
+    global db
+    if not sql_connector_is_active(db):
+        db = sql_connect(db)
+
     cur = db.cursor(dictionary=True)
     return _sql_quarry(cur, querry, data, return_result, commit)
 
@@ -148,10 +193,22 @@ def sqlT(
     #     )
 
 
+app.mount(
+    "/favicon",
+    StaticFiles(directory=os.environ.get("FAVICON_PATH", FAVICON)),
+    name="favicon",
+)
+app.mount(
+    "/static",
+    StaticFiles(directory=os.environ.get("STATIC_PATH", STATIC_FILES)),
+    name="static",
+)
+
+
 @app.get("/")
 async def get_index():
     """gives the Index.html file"""
-    return FileResponse("./index.html")
+    return FileResponse(APP_ROOT_PATH / "index.html")
 
 
 @app.get("/files/{file_id}")
@@ -178,7 +235,7 @@ async def get_file(file_id: str):
     # status_code=500, detail="Somethings wrong with the database"
     # )
     # filename = cur.fetchone()[0]
-    return FileResponse(FILES_IN_PROGRESS + filename)
+    return FileResponse(FILES_IN_PROGRESS / filename)
 
 
 @app.get("/search/lva")
@@ -224,6 +281,9 @@ async def search_lva(
             )
             # res += cur.fetchall()
         res = remove_duplicates(res + zw)
+    info(
+        f"LVA Search: {searchterm}; Result: {res[: (searchlim if searchlim != 0 else -1)]}"
+    )
     if searchlim == 0:
         return res
     else:
@@ -258,6 +318,9 @@ async def search_profs(
         )
         # res += cur.fetchall()
     res = remove_duplicates(res + zw)
+    info(
+        f"Prof Search: {searchterm}; Result: {res[: (searchlim if searchlim != 0 else -1)]}"
+    )
     if searchlim == 0:
         return res
     else:
@@ -298,6 +361,9 @@ async def search_subcats(
         )
         # res += cur.fetchall()
     res = remove_duplicates(res + rest)
+    info(
+        f"Subcatrgory Search: {searchterm}; Result: {res[: (searchlim if searchlim != 0 else -1)]}"
+    )
     if searchlim == 0:
         return res
     else:
@@ -354,7 +420,7 @@ async def create_upload_file(files: List[UploadFile], c2pdf: bool = True):
             content = doc.tobytes()
     if ft != "dir":
         filename = make_filename_unique(filename)
-    locpath = FILES_IN_PROGRESS + filename
+    locpath = FILES_IN_PROGRESS / filename
     # locpaths.append(locpath)
     # cur = db.cursor()
     # try:
@@ -421,7 +487,7 @@ async def get_submission(
         f"lva: {lva}, prof: {prof}, fname {fname}, stype: {stype}, subcat: {subcat}, sem: {sem}, ex_date: {ex_date}, rects: {rects}, pagescales: {pagescales}, ocr: {ocr}"
     )
     info(
-        f"lva: {lva}, prof: {prof}, fname {fname}, stype: {stype}, subcat: {subcat}, sem: {sem}, ex_date: {ex_date}, rects: {rects}, pagescales: {pagescales}, ocr: {ocr}"
+        f"Got Submission: lva: {lva}, prof: {prof}, fname {fname}, stype: {stype}, subcat: {subcat}, sem: {sem}, ex_date: {ex_date}, rects: {rects}, pagescales: {pagescales}, ocr: {ocr}"
     )
     rects_p = json.loads(rects)
     scales_p = json.loads(pagescales)
@@ -438,7 +504,7 @@ async def get_submission(
             error(f"User tried to upload a file without specifying the {th[1]}")
             raise HTTPException(400, f"You need to specify a {th[1]}")
 
-    filepath = "./app/files/" + res[0][0]
+    filepath = FILES_IN_PROGRESS / res[0][0]
     # except mariadb.Error as e:
     # print(f"Mariadb Error: {e}")
     # raise HTTPException(
@@ -510,8 +576,8 @@ async def yield_censor_status(file_id: str):
 
 
 def censor_pdf(
-    path: str,
+    path: os.PathLike,
-    destpath: str,
+    destpath: os.PathLike,
     rects: List[List[List[float]]],
     scales: List[Dict[str, float]],
     file_id: str,
@@ -526,6 +592,7 @@ def censor_pdf(
     Returns:
         None
     """
+    info(f"started Censoring for file {path} to be saved to {destpath}")
     doc = pymupdf.open(path)
     page = doc[0]
     npage = doc.page_count
@@ -558,8 +625,8 @@ def censor_pdf(
 
 
 def censor_pdf_ocr(
-    path: str,
+    path: os.PathLike,
-    destpath: str,
+    destpath: os.PathLike,
     rects: List[List[List[float]]],
     scales: List[Dict[str, float]],
     file_id: str,
@@ -576,6 +643,7 @@ def censor_pdf_ocr(
     Returns:
         None
     """
+    info(f"started Censoring in OCR Mode for file {path} to be saved to {destpath}")
     doc = pymupdf.open(path)
     output = pymupdf.open()
     page = doc[0]
@@ -603,12 +671,31 @@ def censor_pdf_ocr(
         censor_status_datas[file_id]["done"] = False
         censor_status_update_events[file_id].set()
         # THis Costs us dearly
+        try:
             bitmap = page.get_pixmap(dpi=400)
             pdf_bytes = bitmap.pdfocr_tobytes(
                 language="deu",
                 tessdata="/usr/share/tessdata/",  # tesseract needs to be installed; this is the path to thetesseract files
             )
             output.insert_pdf(pymupdf.Document(stream=pdf_bytes))
+        except RuntimeError as e:
+            error(
+                f"Error in OCR for document: {destpath}. Error: {e}. Falling back to standard mode."
+            )
+            if i < len(rects) and rects[i] != []:
+                for rect in rects[i]:
+                    prect = pymupdf.Rect(
+                        rect[0] * wfac,
+                        rect[1] * hfac,
+                        (rect[0] + rect[2]) * wfac,
+                        (rect[1] + rect[3]) * hfac,
+                    )
+                    page.add_redact_annot(
+                        prect,
+                        fill=(0, 0, 0),
+                    )
+                page.apply_redactions()
+            output.insert_pdf(page.parent, from_page=page.number, to_page=page.number)
         # End of the costly part
         print(f"Page {i + 1}/{npage}: CENSORING DONE")
     output.save(destpath)
@@ -665,21 +752,22 @@ def make_savepath(
     ex_date: str,
     fname: str,
     ftype: str,
-) -> str:
+) -> os.PathLike:
     """Generates the path, the file is saved to after the upload process is finished. It creates all nessecery directories."""
+    info(f"Started to make Savepath for '{fname}' in '{lva}' with prof '{prof}'.")
     lv = get_lvpath(lva)
-    lvpath = lv[1] + "/"
+    lvpath = Path(lv[1])
     pf = get_profpath(prof, lv[0])
-    pfpath = pf[1] + "/"
+    pfpath = Path(pf[1])
-    catpath = CATEGORIES[int(cat)] + "/"
+    catpath = Path(CATEGORIES[int(cat)])
-    scpath = ""
+    scpath: str | os.PathLike = ""
     if int(cat) in SUBCAT_CATEGORIES_I and subcat != "":
         sc = get_subcatpath(subcat, int(cat), pf[0], lv[0])
-        scpath = sc[1] + "/"
+        scpath = Path(sc[1])
     if int(cat) == 6:
-        savepath = UNIZEUG_PATH + lv[1] + "_Multimedia_only/" + pfpath
+        savepath = UNIZEUG_PATH / (lv[1] + "_Multimedia_only/") / pfpath
     else:
-        savepath = UNIZEUG_PATH + lvpath + pfpath + catpath + scpath
+        savepath = UNIZEUG_PATH / lvpath / pfpath / catpath / scpath
     os.makedirs(savepath, exist_ok=True)
     filename = sem + "_"
     if int(cat) in EX_DATE_CATEGORIES_I:
@@ -697,14 +785,16 @@ def make_savepath(
     filename += fname
     file = filename + "." + ftype
 
-    destpath = pathlib.Path(savepath + file)
+    destpath = savepath / file
     i = 0
     while destpath.is_file():
+        info(f"{destpath} already exists.")
         file = filename + f"_{i}." + ftype
         i += 1
-        destpath = pathlib.Path(savepath + file)
+        destpath = savepath / file
     destpath.touch()
-    return savepath + file
+    info(f"Path for file to be saved generated as: {savepath / file}")
+    return savepath / file
 
 
 def get_lvpath(lva: str) -> Tuple[int, str]:
@@ -897,10 +987,10 @@ async def save_files_to_folder(files: List[UploadFile]) -> str:
     if filename == "":
         filename = "None"
     filename = make_filename_unique(filename)
-    os.mkdir(FILES_IN_PROGRESS + filename)
+    os.mkdir(FILES_IN_PROGRESS / filename)
     for idx, file in enumerate(files):
         fn = file.filename if file.filename is not None else "None" + str(idx)
-        with open(FILES_IN_PROGRESS + filename + "/" + fn, "wb") as f:
+        with open(FILES_IN_PROGRESS / filename / fn, "wb") as f:
             f.write(await file.read())
     return filename
 
@@ -928,13 +1018,13 @@ async def remove_old_FIP_entrys():
     info(f"Remove Files: {files}")
     for file in files:
         sql("DELETE FROM FIP WHERE id=?", (file["id"]), return_result=False)
-        os.remove(FILES_IN_PROGRESS + file["filename"])
+        os.remove(FILES_IN_PROGRESS / file["filename"])
     # sql(
     #     "DELETE FROM FIP WHERE HOUR(TIMEDIFF(NOW(),initTimeStamp)) > 24",
     #     return_result=False,
     # )
     db.commit()
-    return FileResponse("./index.html")
+    return FileResponse(APP_ROOT_PATH / "index.html")
 
 
 def delete_from_FIP(uuid: str):
@@ -942,4 +1032,4 @@ def delete_from_FIP(uuid: str):
     if len(res) < 1:
         raise HTTPException(500, "I am trying to delete a file that dose not exist")
     sql("DELETE FROM FIP WHERE id=?", (uuid,), return_result=False, commit=True)
-    os.remove(FILES_IN_PROGRESS + res[0]["filename"])
+    os.remove(FILES_IN_PROGRESS / res[0]["filename"])

app/static/app.js

@@ -283,10 +283,10 @@ function submitPdf(eve) {
   submitForm(formdata);
 }
 async function submitForm(formData) {
+  var updateEventSource = null;
   try {
-    const updateEventSource = new EventSource(
+    updateEventSource = new EventSource("/get_censor_status/" + doc.fID);
-      "http://127.0.0.1:8000/get_censor_status/" + doc.fID,
+
-    );
     modal.style.display = "flex";
     // console.log("http://127.0.0.1:8000/get_censor_status/" + doc.fID);
     updateEventSource.addEventListener("censorUpdate", function(eve) {
@@ -295,11 +295,19 @@ async function submitForm(formData) {
       upload_status.innerText =
         "Censoring Page " + data.page + "/" + data.pages;
     });
-    const response = await fetch("http://127.0.0.1:8000/submit", {
+  } catch {
+    console.error(
+      "Error geting eventsource for updating censoring page count: " + error,
+    );
+  }
+  try {
+    const response = await fetch("/submit/", {
       method: "POST",
       body: formData,
     });
+    if (updateEventSource !== null) {
       updateEventSource.close();
+    }
     modal.style.display = "none";
     //let responseJSON=await response.json();
     if (response.ok) {
@@ -320,7 +328,7 @@ async function submitForm(formData) {
       window.alert("Error: " + (await response.json())["detail"]);
     }
   } catch (error) {
-    console.error("Error" + error);
+    console.error("Error submitting: " + error);
   }
 }
 function uploadPdf(eve) {
@@ -338,7 +346,7 @@ function uploadPdf(eve) {
 }
 async function uploadFile(formData) {
   try {
-    const response = await fetch("http://127.0.0.1:8000/uploadfile", {
+    const response = await fetch("/uploadfile/", {
       method: "POST",
       body: formData,
     });

app/static/autocomplete.js

@@ -1,4 +1,4 @@
-var url = "http://127.0.0.1:8000/search/";
+var url = "/search/";
 var lid = null;
 var pid = null;
 var activeAutocompletion = null;

compose.yml

@@ -0,0 +1,67 @@
+services:
+  app:
+    container_name: python-app
+    # command: python -m uvicorn app.main:app --host 0.0.0.0 --port 80
+    build:
+      context: .
+      dockerfile: Dockerfile
+    volumes:
+      - ./app:/python
+      - ./unizeug:/unizeug:source
+    ports:
+      - 80:80
+    restart: unless-stopped
+    environment:
+      ENTRY_COMMAND: python -m uvicorn main:app --host 0.0.0.0 --port 80
+      APP_LOG_PATH: /python/app.log
+      APP_ROOT_PATH: /python
+      UNIZEUG_PATH: /unizeug
+      DB_HOST: db
+      DB_USER: app
+      DB_PASSWORD: DBPassword
+      DB_DATABASE: Unizeug
+      TZ: "Europe/Vienna"
+
+    depends_on:
+      - db
+      - scaner
+  db:
+    container_name: db
+    image: mariadb
+    restart: unless-stopped
+    environment:
+      MARIADB_ROOT_PASSWORD: DBPassword
+      MARIADB_USER: app
+      UNIZEUG_PATH: /unizeug
+      MARIADB_PASSWORD: DBPassword
+      MARIADB_DATABASE: Unizeug
+      TZ: "Europe/Vienna"
+    healthcheck:
+      test: ["CMD", "healthcheck.sh", "--connect", "--innodb_initialized"]
+      start_period: 10s
+      interval: 10s
+      timeout: 5s
+      retries: 3
+    volumes:
+      - ./mariadb:/var/lib/mysql
+  scaner:
+    container_name: python-scaner
+    # command: python /python/init.py
+    build:
+      context: .
+      dockerfile: Dockerfile
+    volumes:
+      - ./app:/python
+      - ./unizeug:/unizeug:source
+    restart: unless-stopped
+    environment:
+      ENTRY_COMMAND: python /python/init.py
+      UNIZEUG_PATH: /unizeug
+      APP_ROOT_PATH: /python
+      DB_HOST: db
+      DB_USER: app
+      DB_PASSWORD: DBPassword
+      DB_DATABASE: Unizeug
+      TZ: "Europe/Vienna"
+    depends_on:
+      - db

entrypoint.sh

@@ -0,0 +1,4 @@
+#!/bin/sh 
+if [[ -n "$ENTRY_COMMAND" ]]; then
+  /bin/sh -c "$ENTRY_COMMAND"
+fi

requirements.txt

@@ -44,10 +44,12 @@ pypdf==5.2.0
 pytesseract==0.3.13
 python-dotenv==1.0.1
 python-multipart==0.0.20
+pytz==2025.2
 PyYAML==6.0.2
 requests==2.32.3
 rich==13.9.4
 rich-toolkit==0.13.2
+schedule==1.2.2
 shellingham==1.5.4
 sniffio==1.3.1
 starlette==0.45.3