Decompress RTF data

.travis.yml

@@ -11,7 +11,7 @@ python:
 
 install:
   - pip install -e .
-  - pip install pytest pytest-cov pytest-console-scripts codecov
+  - pip install pytest pytest-cov pytest-console-scripts codecov compressed_rtf
 
 script:
   - pytest --cov

setup.py

@@ -33,7 +33,7 @@
       license='LGPL',
       packages=find_packages(exclude=['ez_setup', 'examples', 'tests']),
       setup_requires=["pytest-runner"],
-      tests_require = ['pytest', 'coverage'],
+      tests_require = ['pytest', 'coverage', 'compressed_rtf'],
       include_package_data=True,
       zip_safe=True,
       entry_points = {

tests/test_cmdline.py

@@ -26,6 +26,6 @@ def test_cmdline_no_htmlbody(script_runner):
 
 def test_cmdline_rtfbody(script_runner):
     ret = script_runner.run('tnefparse', '-rb', 'tests/examples/rtf.tnef')
-    assert len(ret.stdout) == 987
+    assert len(ret.stdout) == 593
     assert ret.stderr == ''
     assert ret.success

tests/test_decoding.py

@@ -68,6 +68,9 @@ def test_decode(tnefspec):
         if body:
             assert getattr(t, body)
 
+        if t.rtfbody:
+            assert t.rtfbody[0:5] == b'{\\rtf'
+
         if objs:
             assert objcodes(t) == objs, "wrong objs: %s" % ["0x%2.2x" % o.name for o in t.objects]
 

tnefparse/cmdline.py

@@ -92,20 +92,20 @@ def tnefparse():
         if args.body:
             body = getattr(t, "body")
             if body:
-                print(body)
+                sys.stdout.write(body.decode('latin-1'))
             else:
                 sys.exit("No body found")
 
         if args.htmlbody:
             body = getattr(t, "htmlbody")
             if body:
-                print(body)
+                sys.stdout.write(body.decode('latin-1'))
             else:
                 sys.exit("No HTML body found")
 
         if args.rtfbody:
             body = getattr(t, "rtfbody")
             if body:
-                print(body)
+                sys.stdout.write(body.decode('latin-1'))
             else:
                 sys.exit("No RTF body found")

tnefparse/tnef.py

@@ -204,7 +204,7 @@ def __init__(self, data, do_checksum=True):
         self.mapiprops = []
         self.body = None
         self.htmlbody = None
-        self.rtfbody = None
+        self._rtfbody = None
         offset = 6
 
         if not do_checksum:
@@ -236,7 +236,7 @@ def __init__(self, data, do_checksum=True):
                     elif p.name == TNEFMAPI_Attribute.MAPI_BODY_HTML:
                         self.htmlbody = p.data
                     elif p.name == TNEFMAPI_Attribute.MAPI_RTF_COMPRESSED:
-                        self.rtfbody = p.data
+                        self._rtfbody = p.data
             elif obj.name == TNEF.ATTBODY:
                 self.body = obj.data
             elif obj.name == TNEF.ATTTNEFVERSION:
@@ -249,7 +249,19 @@ def __init__(self, data, do_checksum=True):
                 logger.debug("Unknown TNEF Object: %s" % obj)
 
     def has_body(self):
-        return True if (self.body or self.htmlbody or self.rtfbody) else False
+        return True if (self.body or self.htmlbody or self._rtfbody) else False
+
+    @property
+    def rtfbody(self):
+        if self._rtfbody:
+            try:
+                from compressed_rtf import decompress
+                return decompress(self._rtfbody + b'\x00')
+            except ImportError:
+                logger.warning("Returning compressed RTF. Install compressed_rtf to decompress")
+                return self._rtfbody
+        else:
+            return None
 
     def __str__(self):
         atts = (", %i attachments" % len(self.attachments)) if self.attachments else ''