Wrapped watch in exception handling so nothing should terminate it except ctrl-c

Author: virantha

pypdfocr/pypdfocr.py

@@ -15,7 +15,7 @@
 
 import smtplib
 import argparse
-import sys, os
+import sys, os, traceback
 import logging
 import shutil, glob
 import itertools
@@ -326,9 +326,8 @@ def run_conversion(self, pdf_filename):
             :rtype: filename string
         """
         print ("Starting conversion of %s" % pdf_filename)
-        conversion_format = "tiff"
         # Make the images for Tesseract
-        img_dpi, glob_img_filename = self.gs.make_img_from_pdf(pdf_filename, conversion_format)
+        img_dpi, glob_img_filename = self.gs.make_img_from_pdf(pdf_filename)
 
         fns = glob.glob(glob_img_filename)
 
@@ -432,9 +431,17 @@ def go(self, argv):
 
         # Do the actual conversion followed by optional filing and email
         if self.watch:
-            py_watcher = PyPdfWatcher(self.watch_dir, self.config.get('watch'))
-            for pdf_filename in py_watcher.start():
-                self._convert_and_file_email(pdf_filename)
+            while True:  # Make sure the watcher doesn't terminate
+                try:
+                    py_watcher = PyPdfWatcher(self.watch_dir, self.config.get('watch'))
+                    for pdf_filename in py_watcher.start():
+                        self._convert_and_file_email(pdf_filename)
+                except KeyboardInterrupt:
+                    break
+                except Exception as e:
+                    print traceback.print_exc(e)
+                    py_watcher.stop()
+                    
         else:
             self._convert_and_file_email(self.pdf_filename)
 

pypdfocr/pypdfocr_gs.py

@@ -171,7 +171,7 @@ def _run_gs(self, options, output_filename, pdf_filename):
                 error (self.msgs['GS_FAILED'])
 
 
-    def make_img_from_pdf(self, pdf_filename, output_format):
+    def make_img_from_pdf(self, pdf_filename):
         self._get_dpi(pdf_filename) # No need to bother anymore
 
         if not os.path.exists(pdf_filename):

pypdfocr/pypdfocr_watcher.py

@@ -37,23 +37,21 @@ def __init__(self, monitor_dir, config):
         self.scan_interval = config.get('scan_interval', 3) # If no updates in 3 seconds (or user specified option in config file) process file
 
     def start(self):
-        try: 
-            while True:
-                observer = Observer()
-                observer.schedule(self, self.monitor_dir)
-                observer.start()
-                print("Starting to watch for new pdfs in %s" % (self.monitor_dir))
-                while True:
-                    logging.info("Sleeping for %d seconds" % self.scan_interval)
-                    time.sleep(self.scan_interval)
-                    newFile = self.check_queue()
-                    if newFile:
-                        yield newFile
-                observer.join()
-        except KeyboardInterrupt:
-            return
+        self.observer = Observer()
+        self.observer.schedule(self, self.monitor_dir)
+        self.observer.start()
+        print("Starting to watch for new pdfs in %s" % (self.monitor_dir))
+        while True:
+            logging.info("Sleeping for %d seconds" % self.scan_interval)
+            time.sleep(self.scan_interval)
+            newFile = self.check_queue()
+            if newFile:
+                yield newFile
+        self.observer.join()
 
 
+    def stop(self):
+        self.observer.stop()
 
     def rename_file_with_spaces(self, pdf_filename):
         """

test/test_gs.py

@@ -43,7 +43,7 @@ def test_gs_run_nt(self, mock_subprocess, mock_os_name, capsys):
     def test_gs_pdf_missing(self, capsys):
         p = P.PyGs()
         with pytest.raises(SystemExit):
-            p.make_img_from_pdf("missing123.pdf", "")
+            p.make_img_from_pdf("missing123.pdf")
         out,err = capsys.readouterr()
         assert p.msgs['GS_MISSING_PDF'] in out