Avoid using close methods on pdfParser objects

Signed-off-by: Ayan Sinha Mahapatra <[email protected]>

Author: AyanSinhaMahapatra

src/textcode/pdf.py

@@ -28,20 +28,20 @@ def get_text_lines(location, max_pages=5):
     extracted_text = BytesIO()
     laparams = LAParams()
     with open(location, 'rb') as pdf_file:
-        with contextlib.closing(PDFParser(pdf_file)) as parser:
-            document = PDFDocument(parser)
-            if not document.is_extractable:
-                raise PDFTextExtractionNotAllowed(
-                    'Encrypted PDF document: text extraction is not allowed')
+        parser = PDFParser(pdf_file)
+        document = PDFDocument(parser)
+        if not document.is_extractable:
+            raise PDFTextExtractionNotAllowed(
+                'Encrypted PDF document: text extraction is not allowed')
 
-            manager = PDFResourceManager()
-            with contextlib.closing(
-                TextConverter(manager, extracted_text, laparams=laparams)) as extractor:
-                interpreter = PDFPageInterpreter(manager, extractor)
-                pages = PDFPage.create_pages(document)
-                for page_num, page in enumerate(pages, 1):
-                    interpreter.process_page(page)
-                    if max_pages and page_num == max_pages:
-                        break
+        manager = PDFResourceManager()
+        with contextlib.closing(
+            TextConverter(manager, extracted_text, laparams=laparams)) as extractor:
+            interpreter = PDFPageInterpreter(manager, extractor)
+            pages = PDFPage.create_pages(document)
+            for page_num, page in enumerate(pages, 1):
+                interpreter.process_page(page)
+                if max_pages and page_num == max_pages:
+                    break
                 extracted_text.seek(0)
                 return extracted_text.readlines()

tests/licensedcode/data/datadriven/lic4/should_detect_something_5.pdf.yml

@@ -1,2 +1,4 @@
 license_expressions:
   - sun-sissl-1.1
+  - proprietary-license
+  - cpal-1.0

tests/textcode/test_pdf.py

@@ -98,13 +98,7 @@ def test_numbered_text_lines_does_not_fail_on_autocad_test_pdf(self):
     b'It  is  a  print  formatter  driv-\n',
     b'en  by  XSL  formatting  ob-\n',
     b'jects (XSL-FO) and an out-\n',
-    b'(XSL-FO) and an output in-\n',
-    b'dependent formatter. It is a\n',
-    b'Java application that reads\n',
-    b'Line 2 of item 1\n',
-    b'Apache  FOP  (Formatting\n',
-    b'Objects  Processor)  est\n',
-    b'une application de mise en\n',
-    b'The  end  of  the  document\n',
-    b'has now been reached.\n',
-]
+    b'ter1. FOP has a nice logo:\n',
+    b'Header 1.1 Header 1.2\n',
+    b'See the FOP website for more information\n'
+]