Additional fixes to port map file.

Author: vladd-bit

config.py

@@ -3,15 +3,15 @@
 
 from sys import platform
 
-OCR_SERVICE_VERSION = "0.2.0"
+OCR_SERVICE_VERSION = "0.2.3"
 # 50 - CRITICAL, 40 - ERROR, 30 - WARNING, 20 - INFO, 10 - DEBUG, 0 - NOTSET
 LOG_LEVEL = int(os.environ.get("OCR_SERVICE_LOG_LEVEL", 40))
 
 ROOT_DIR = os.path.abspath(os.curdir)
 TMP_FILE_DIR = os.path.join(ROOT_DIR, "tmp")
 
 # Should we actually ocr or just extract text from PDFs ? NOTE: OCR IS STILL APPLIED TO IMAGES if detected | possible vals : "OCR", "NO_OCR"
-OPERATION_MODE= os.environ.get("OCR_SERVICE_OPERATION_MODE", "OCR")
+OPERATION_MODE = os.environ.get("OCR_SERVICE_OPERATION_MODE", "OCR")
 
 # basic app settings
 OCR_SERVICE_PORT = os.environ.get("OCR_SERVICE_PORT", 8090)

gunicorn.conf.py

@@ -11,7 +11,7 @@
 def pre_fork(server, worker):
     global counter
 
-    time.sleep(2 + counter * 1.5)
+    time.sleep(1 + counter)
     counter += 1
 
 def post_fork(server, worker):

ocr_service/app/app.py

@@ -30,15 +30,14 @@ def start_office_converter_servers():
     port_count = 0
 
     for port_num in LIBRE_OFFICE_LISTENER_PORT_RANGE:
-        if (port_count < OCR_WEB_SERVICE_WORKERS or port_count < OCR_WEB_SERVICE_THREADS):
-            port_count += 1
+        if port_num == get_assigned_port(os.getpid()) and OCR_WEB_SERVICE_THREADS == 1:
             print("STARTED WORKER ON PORT: " + str(port_num))
             process = start_office_server(port_num)
             loffice_processes[port_num] = process
-            if port_num == get_assigned_port(os.getpid()) and OCR_WEB_SERVICE_THREADS == 1:
-                break
-            elif OCR_WEB_SERVICE_WORKERS == 1:
-                continue
+            break
+        elif OCR_WEB_SERVICE_WORKERS == 1 and OCR_WEB_SERVICE_THREADS > 1:
+            process = start_office_server(port_num)
+            loffice_processes[port_num] = process
 
     return loffice_processes
 

ocr_service/utils/utils.py

@@ -99,7 +99,14 @@ def get_process_id_by_process_name(process_name: str = "") -> int:
     return pid
 
 def sync_port_mapping(worker_id = None, worker_pid = None):
-      with open(os.path.join(TMP_FILE_DIR, './worker_process_data.txt'), encoding="utf-8", mode='a+') as f:
+      
+
+      f_path = os.path.join(TMP_FILE_DIR, './worker_process_data.txt')
+      open_mode = "r+"
+      if not os.path.exists(f_path):
+          open_mode = "w+"
+      
+      with open(f_path, encoding="utf-8", mode=open_mode) as f:
             fcntl.lockf(f, fcntl.LOCK_EX)
 
             port_mapping = {}