Experiment with using pdftotext.

Dockerfile

@@ -30,4 +30,5 @@ COPY . /tmp/email2pdf/
 COPY docker/email2pdf/getmail /etc/cron.d/
 WORKDIR /tmp/email2pdf
 RUN make builddeb_real && sh -c 'ls -1 /tmp/email2pdf/*.deb | xargs -L 1 gdebi -n' && cp /tmp/email2pdf/*.deb /tmp
+RUN apt-get install -y xpdf-utils
 RUN apt-get clean && rm -rf /var/lib/apt/lists/* /var/tmp/*

tests/BaseTestClasses.py

@@ -369,18 +369,12 @@ class Email2PDFTestCase(unittest.TestCase):
                 return None
 
     def getPDFText(self, filename):
-        try:
-            with io.StringIO() as retstr:
-                with open(filename, 'rb') as filehandle:
-                    rsrcmgr = PDFResourceManager()
-                    device = TextConverter(rsrcmgr, retstr, laparams=LAParams())
-                    pagenos = set()
-                    process_pdf(rsrcmgr, device, filehandle, pagenos, maxpages=0, password="", caching=True, check_extractable=True)
-                    device.close()
-                    string = retstr.getvalue()
-                    return string
-        except PSException:
-            return None
+        with tempfile.NamedTemporaryFile() as temporaryFile:
+            options = ['pdftotext', filename, temporaryFile.name]
+            p = Popen(options)
+            p.wait()
+            with open(temporaryFile.name, 'rb') as f:
+                return str(f.read(), 'utf-8')
 
     def touch(self, fname):
         open(fname, 'w').close()