fixed error in compose.yml

Dockerfile

@@ -21,7 +21,8 @@ RUN apk add --no-cache \
     tk-dev \
     tcl-dev \
     libwebp-dev \
-    tesseract-ocr
+    tesseract-ocr \
+    tesseract-ocr-data-deu
 RUN python -m ensurepip --upgrade
 RUN pip install setuptools wheel
 RUN pip install -r requirements.txt

app/init.py

@@ -8,7 +8,7 @@ import os
 import json
 import mariadb
 import logging
-
+from pathlib import Path
 import schedule
 import time
 import pytz
@@ -24,7 +24,8 @@ CATEGORIES = [
 ]
 SUBCAT_CATEGORIES = ["Klausuren", "Übungen", "Labore"]
 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",
@@ -79,6 +80,19 @@ except mariadb.OperationalError:
 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)
@@ -149,6 +163,7 @@ def get_dirstruct():
                         (lid, pid, idx, subcat.name),
                     )
                 db.commit()
+    remove_old_FIP_entrys()
 
 
 def link_prof(firstname, lastname, lid):

app/main.py

@@ -671,12 +671,31 @@ def censor_pdf_ocr(
         censor_status_datas[file_id]["done"] = False
         censor_status_update_events[file_id].set()
         # THis Costs us dearly
-        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))
+        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)

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(
-      window.location + "get_censor_status/" + doc.fID,
-    );
+    updateEventSource = new EventSource("/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;
     });
+  } catch {
+    console.error(
+      "Error geting eventsource for updating censoring page count: " + error,
+    );
+  }
+  try {
     const response = await fetch("/submit/", {
       method: "POST",
       body: formData,
     });
-    updateEventSource.close();
+    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) {

compose.yml

@@ -57,6 +57,7 @@ services:
     environment:
       ENTRY_COMMAND: python /python/init.py
       UNIZEUG_PATH: /unizeug
+      APP_ROOT_PATH: /python
       DB_HOST: db
       DB_USER: app
       DB_PASSWORD: DBPassword