Drop Python 2 for scancode-toolkit

Signed-off-by: Abhishek Kumar <abhishek.kasyap09@gmail.com>

Author: Abhishek-Dev09

src/formattedcode/output_html.py

@@ -49,8 +49,6 @@
 from commoncode.fileutils import fsencode
 from commoncode.fileutils import parent_directory
 from commoncode.system import on_linux
-from commoncode.system import py2
-from commoncode.system import py3
 
 from plugincode.output import output_impl
 from plugincode.output import OutputPlugin
@@ -122,10 +120,6 @@ def is_enabled(self, custom_output, custom_template, **kwargs):
     def process_codebase(self, codebase, custom_output, custom_template, **kwargs):
         results = self.get_files(codebase, **kwargs)
         version = codebase.get_or_create_current_header().tool_version
-
-        if on_linux and py2:
-            custom_template = fsencode(custom_template)
-
         template_loc = custom_template
         output_file = custom_output
         write_templated(output_file, results, version, template_loc)
@@ -329,12 +323,8 @@ def create_html_app(output_file, results, version, scanned_path):  # NOQA
 
         # write json data
         # FIXME: this should a regular JSON scan format
-        if py2:
-            mode = 'wb'
-            prefix = b'data='
-        if py3:
-            mode = 'w'
-            prefix = u'data='
+        mode = 'w'
+        prefix = u'data='
         with io.open(join(target_assets_dir, 'data.js'), mode) as f:
             f.write(prefix)
             simplejson.dump(results, f, iterable_as_array=True)

src/formattedcode/output_json.py

@@ -31,8 +31,6 @@
 from six import string_types
 
 from commoncode import compat
-from commoncode.system import py2
-from commoncode.system import py3
 from plugincode.output import output_impl
 from plugincode.output import OutputPlugin
 from scancode import CommandLineOption
@@ -65,19 +63,11 @@ def logger_debug(*args):
                                      and a or repr(a) for a in args))
 
 
-if py2:
-    mode = 'wb'
-    space = b' '
-    comma = b','
-    colon = b':'
-    eol = b'\n'
-
-if py3:
-    mode = 'w'
-    space = u' '
-    comma = u','
-    colon = u':'
-    eol = u'\n'
+mode = 'w'
+space = u' '
+comma = u','
+colon = u':'
+eol = u'\n'
 
 
 @output_impl

src/formattedcode/output_jsonlines.py

@@ -29,8 +29,6 @@
 
 import simplejson
 
-from commoncode.system import py2
-from commoncode.system import py3
 from plugincode.output import output_impl
 from plugincode.output import OutputPlugin
 from scancode import CommandLineOption
@@ -42,21 +40,12 @@
 """
 
 
-if py2:
-    mode = 'wb'
-    space = b' '
-    comma = b','
-    colon = b':'
-    eol = b'\n'
-    file_key = b'files'
-
-if py3:
-    mode = 'w'
-    space = u' '
-    comma = u','
-    colon = u':'
-    eol = u'\n'
-    file_key = u'files'
+mode = 'w'
+space = u' '
+comma = u','
+colon = u':'
+eol = u'\n'
+file_key = u'files'
 
 
 @output_impl

src/licensedcode/cache.py

@@ -38,7 +38,6 @@
 from commoncode.fileutils import resource_iter
 from commoncode.fileutils import create_dir
 from commoncode import ignore
-from commoncode.system import py3
 
 from scancode_config import scancode_cache_dir
 from scancode_config import scancode_src_dir
@@ -285,10 +284,7 @@ def load_index(cache_file, use_loads=False):
                 'Please delete "{cache_file}" and retry.\n'
                 'If the problem persists, copy this error message '
                 'and submit a bug report.\n'.format(**locals()))
-            if py3:
-                raise ex_type(message).with_traceback(ex_traceback)
-            else:
-                six.reraise(ex_type, message, ex_traceback)
+            raise ex_type(message).with_traceback(ex_traceback)
 
 
 _ignored_from_hash = partial(
@@ -314,8 +310,7 @@ def tree_checksum(tree_base_dir=scancode_src_dir, _ignored=_ignored_from_hash):
     resources = resource_iter(tree_base_dir, ignored=_ignored, with_dirs=False)
     hashable = (pth + str(getmtime(pth)) + str(getsize(pth)) for pth in resources)
     hashable = ''.join(sorted(hashable))
-    if py3:
-        hashable=hashable.encode('utf-8')
+    hashable=hashable.encode('utf-8')
     return md5(hashable).hexdigest()
 
 

src/licensedcode/match_hash.py

@@ -29,8 +29,6 @@
 
 from six import string_types
 
-from commoncode.system import py2
-from commoncode.system import py3
 from licensedcode.match import LicenseMatch
 from licensedcode.spans import Span
 
@@ -65,10 +63,7 @@ def tokens_hash(tokens):
     """
     Return a digest binary string computed from a sequence of numeric token ids.
     """
-    if py2:
-        as_bytes = array('h', tokens).tostring()
-    if py3:
-        as_bytes = array('h', tokens).tobytes()
+    as_bytes = array('h', tokens).tobytes()
     return md5(as_bytes).digest()
 
 

src/licensedcode/tokenize.py

@@ -40,8 +40,6 @@
 from binascii import crc32
 import re
 
-from commoncode.system import py2
-from commoncode.system import py3
 from licensedcode.stopwords import STOPWORDS
 from textcode.analysis import numbered_text_lines
 
@@ -229,8 +227,7 @@ def select_ngrams(ngrams, with_pos=False):
         # FIXME: use a proper hash
         nghs = []
         for ng in ngram:
-            if ((py2 and isinstance(ng, basestring))
-                    or (py3 and isinstance(ng, str))):
+            if isinstance(ng, str):
                 ng = bytearray(ng, encoding='utf-8')
             else:
                 ng = bytearray(str(ng).encode('utf-8'))

src/packagedcode/jar_manifest.py

@@ -32,7 +32,6 @@
 import attr
 
 from commoncode.fileutils import as_posixpath
-from commoncode.fileutils import py2
 from packagedcode.utils import normalize_vcs_url
 from packagedcode.maven import parse_scm_connection
 from packagedcode.models import Package
@@ -82,10 +81,7 @@ def parse_manifest(location):
     """
     Return a Manifest parsed from the file at `location` or None if this
     cannot be parsed.         """
-    if py2:
-        mode = 'rb'
-    else:
-        mode = 'r'
+    mode = 'r'
     with open(location, mode) as manifest:
         return parse_manifest_data(manifest.read())
 

src/packagedcode/pypi.py

@@ -49,7 +49,6 @@
 
 from commoncode import filetype
 from commoncode import fileutils
-from commoncode.system import py2
 from packagedcode import models
 from packagedcode.utils import build_description
 from packagedcode.utils import combine_expressions
@@ -243,10 +242,7 @@ def parse_with_dparse(location):
                          filetypes.pipfile,
                          filetypes.pipfile_lock):
         return
-    if py2:
-        mode = 'rb'
-    else:
-        mode = 'r'
+    mode = 'r'
     with open(location, mode) as f:
         content = f.read()
 
@@ -329,10 +325,7 @@ def parse_setup_py(location):
         return
 
     # FIXME: what if this is unicode text?
-    if py2:
-        mode = 'rb'
-    else:
-        mode = 'r'
+    mode = 'r'
     with open(location, mode) as inp:
         setup_text = inp.read()
 

src/packagedcode/recognize.py

@@ -37,7 +37,6 @@
 from commoncode.fileutils import fsencode
 from commoncode.fileutils import splitext_name
 from commoncode.system import on_linux
-from commoncode.system import py2
 from packagedcode import PACKAGE_TYPES
 from typecode import contenttype
 
@@ -90,9 +89,6 @@ def recognize_packages(location):
     for package_type in PACKAGE_TYPES:
         # Note: default to True if there is nothing to match against
         metafiles = package_type.metafiles
-        if on_linux and py2:
-            metafiles = (fsencode(m) for m in metafiles)
-
         if any(fnmatch.fnmatchcase(filename, metaf) for metaf in metafiles):
             for recognized in package_type.recognize(location):
                 if TRACE:logger_debug('recognize_packages: metafile matching: recognized:', recognized)
@@ -114,9 +110,6 @@ def recognize_packages(location):
         extension_matched = False
         extensions = package_type.extensions
         if extensions:
-            if on_linux and py2:
-                extensions = (fsencode(e) for e in extensions)
-
             extensions = (e.lower() for e in extensions)
             extension_matched = any(fnmatch.fnmatchcase(extension, ext_pat)
                                     for ext_pat in extensions)

src/scancode/cli.py

@@ -68,7 +68,6 @@ class WindowsError(Exception):
 from commoncode.fileutils import PATH_TYPE
 from commoncode.fileutils import POSIX_PATH_SEP
 from commoncode.timeutils import time2tstamp
-from commoncode.system import py2
 from commoncode.system import on_windows
 from commoncode.system import on_linux
 
@@ -590,10 +589,7 @@ def echo_func(*_args, **_kwargs):
 
     if not isinstance(input, (list, tuple)):
         # nothing else todo
-        if on_linux and py2:
-            assert isinstance(input, bytes)
-        else:
-            assert isinstance(input, compat.unicode)
+        assert isinstance(input, compat.unicode)
 
     elif len(input) == 1:
         # we received a single input path, so we treat this as a single path
@@ -1206,10 +1202,7 @@ def scan_codebase(codebase, scanners, processes=1, timeout=DEFAULT_TIMEOUT,
 
         while True:
             try:
-                if py2:
-                    location, rid, scan_errors, scan_time, scan_result, scan_timings = scans.next()
-                else:
-                    location, rid, scan_errors, scan_time, scan_result, scan_timings = next(scans)
+                location, rid, scan_errors, scan_time, scan_result, scan_timings = next(scans)
 
                 if TRACE_DEEP:
                     logger_debug(