From df3c27d79dec231809deb4e617070a16858c306d Mon Sep 17 00:00:00 2001
From: jvoisin
Date: Sun, 18 Mar 2018 21:42:12 +0100
Subject: Improve the testsuite

---
 src/parsers/abstract.py |  1 +
 src/parsers/pdf.py      | 25 ++++++-------------------
 2 files changed, 7 insertions(+), 19 deletions(-)

(limited to 'src/parsers')

diff --git a/src/parsers/abstract.py b/src/parsers/abstract.py
index a9129cc..d0e7108 100644
--- a/src/parsers/abstract.py
+++ b/src/parsers/abstract.py
@@ -1,6 +1,7 @@
 class AbstractParser(object):
     def __init__(self, filename: str):
         self.filename = filename
+        self.output_filename = filename + '.cleaned'
         self.meta_list = set()
 
     def get_meta(self):
diff --git a/src/parsers/pdf.py b/src/parsers/pdf.py
index c25b324..a77eabd 100644
--- a/src/parsers/pdf.py
+++ b/src/parsers/pdf.py
@@ -31,20 +31,6 @@ class PDFParser(abstract.AbstractParser):
         self.uri = 'file://' + os.path.abspath(self.filename)
         self.password = None
 
-    def __optimize_image_size(self, img: io.BytesIO) -> io.BytesIO:
-        """ This is useless as fuck. """
-        if Image is None:
-            return img
-        ret = io.BytesIO()
-        im = Image.open(img)
-        w, h = im.size
-        resized = im.resize((w, h), Image.ANTIALIAS)
-        resized.save(ret, optimize=True, format="PNG")
-        ret.seek(0)
-
-        return ret
-
-
     def remove_all(self):
         """
             Load the document into Poppler, render pages on PNG,
@@ -57,7 +43,7 @@ class PDFParser(abstract.AbstractParser):
         """
         document = Poppler.Document.new_from_file(self.uri, self.password)
 
-        pdf_surface = cairo.PDFSurface("OUT.pdf", 128, 128)
+        pdf_surface = cairo.PDFSurface(self.output_filename, 128, 128)
         pdf_context = cairo.Context(pdf_surface)
 
         for pagenum in range(document.get_n_pages()):
@@ -87,10 +73,11 @@ class PDFParser(abstract.AbstractParser):
 
         pdf_surface.finish()
 
-        document = Poppler.Document.new_from_file('file://' + os.path.abspath('OUT.pdf'), self.password)
-        document.set_producer('totally not MAT2 ;)')
-        document.set_creator('')
-        document.save('file://' + os.path.abspath("OUT_clean.pdf"))
+        # This is removing metadata
+        #document = Poppler.Document.new_from_file('file://' + os.path.abspath('OUT.pdf'), self.password)
+        #document.set_producer('totally not MAT2 ;)')
+        #document.set_creator('')
+        #document.save('file://' + os.path.abspath("OUT_clean.pdf"))
 
         return True
 
-- 
cgit v1.3