FIX: Caps and pytest (#482)

* FIX: Caps and pytest

* FIX: Check in code

* FIX: Check for unique names

* FIX: Minor fixes

* FIX: More checks

.gitignore

@@ -5,3 +5,4 @@ ld
 codespell.egg-info
 *.pyc
 .cache/
+.pytest_cache/

.travis.yml

@@ -39,7 +39,7 @@ before_install:
     - source venv/bin/activate
     - python --version  # just to check
     - pip install -U pip wheel  # upgrade to latest pip find 3.5 wheels; wheel to avoid errors
-    - retry pip install nose flake8 coverage codecov chardet setuptools
+    - retry pip install pytest pytest-cov pytest-sugar flake8 coverage codecov chardet setuptools
     - cd $SRC_DIR
 
 install:
@@ -51,7 +51,7 @@ script:
     # this file has an error
     - "! codespell codespell_lib/tests/test_basic.py"
     - flake8
-    - nosetests
+    - pytest codespell_lib
 
 after_success:
     - codecov

appveyor.yml

@@ -1,8 +1,8 @@
 environment:
   global:
       PYTHON: "C:\\conda"
-      CONDA_DEPENDENCIES: "nose setuptools flake8 coverage chardet"
+      CONDA_DEPENDENCIES: "pytest pytest-cov setuptools flake8 coverage chardet"
-      PIP_DEPENDENCIES: "codecov"
+      PIP_DEPENDENCIES: "codecov pytest-sugar"
   matrix:
       - PYTHON_VERSION: "2.7"
         PYTHON_ARCH: "32"
@@ -21,7 +21,7 @@ build: false  # Not a C# project, build stuff at the test step instead.
 test_script:
   - "codespell --help"
   - "flake8"
-  - "nosetests"
+  - "pytest codespell_lib"
 
 on_success:
   - "codecov"

codespell_lib/_codespell.py

@@ -304,6 +304,10 @@ def build_dict(filename):
     with codecs.open(filename, mode='r', buffering=1, encoding='utf-8') as f:
         for line in f:
             [key, data] = line.split('->')
+            # TODO for now, convert both to lower. Someday we can maybe add
+            # support for fixing caps.
+            key = key.lower()
+            data = data.lower()
             if key in ignore_words:
                 continue
             data = data.strip()
@@ -588,7 +592,7 @@ def main(*args):
 
     dictionaries = options.dictionary or [default_dictionary]
     for dictionary in dictionaries:
-        if dictionary is "-":
+        if dictionary == "-":
             dictionary = default_dictionary
         if not os.path.exists(dictionary):
             print('ERROR: cannot find dictionary file: %s' % dictionary,

codespell_lib/data/dictionary.txt

@@ -192,7 +192,6 @@ achievment->achievement
 achievments->achievements
 achitecture->architecture
 achitectures->architectures
-achive->achieve
 achive->achieve, archive,
 achived->achieved, archived,
 achivement->achievement
@@ -784,7 +783,7 @@ aparment->apartment
 apear->appear
 apeends->appends
 apendix->appendix
-apenines->apennines, Apennines,
+apenines->Apennines
 aplication->application
 aplied->applied
 apllicatin->application
@@ -809,7 +808,7 @@ appedn->append
 appendent->appended
 appendign->appending
 appeneded->appended
-appenines->apennines, Apennines,
+appenines->Apennines
 appens->appends
 apperance->appearance
 apperances->appearances
@@ -1533,7 +1532,7 @@ bootstapped->bootstrapped
 bootstapping->bootstrapping
 bootstaps->bootstraps
 boradcast->broadcast
-bordrelines->borderline
+bordreline->borderline
 bordrelines->borderlines
 borke->broke
 bothe->both
@@ -1903,8 +1902,7 @@ chanel->channel
 changable->changeable
 changuing->changing
 chanined->chained
-chaning->chaining
+chaning->chaining, changing,
-chaning->changing
 channle->channel
 channles->channels
 channnel->channel
@@ -2326,7 +2324,6 @@ competance->competence
 competant->competent
 competative->competitive
 competion->competition, completion,
-competion->completion
 competions->completions
 competitiion->competition
 competive->competitive
@@ -2971,7 +2968,6 @@ covnert->convert
 coyp->copy
 coypright->copyright
 cpation->caption
-cpoy->copy
 cpoy->coy, copy,
 craete->create
 crahses->crashes
@@ -3125,8 +3121,7 @@ decalratiosn->declarations
 deccelerate->decelerate
 decendant->descendant
 decendants->descendants
-decendent->descendant
+decendent->descendent, descendant,
-decendent->descendent
 decendents->descendants
 decideable->decidable
 decidely->decidedly
@@ -5704,7 +5699,6 @@ imigrated->emigrated, immigrated,
 imigration->emigration, immigration,
 imilar->similar
 iminent->eminent, imminent, immanent,
-iminent->imminent
 immeadiately->immediately
 immedate->immediate
 immedately->immediately
@@ -6281,7 +6275,6 @@ intiailises->initialises
 intiailize->initialize
 intiailized->initialized
 intiailizes->initializes
-intiailizing->initialising
 intiailizing->initializing
 intial->initial
 intialisation->initialisation
@@ -6669,7 +6662,6 @@ mamalian->mammalian
 mamory->memory
 managable->manageable, manageably,
 managment->management
-mananged->managed
 manangement->management
 mandetory->mandatory
 maneouvre->manoeuvre
@@ -7030,8 +7022,7 @@ moent->moment
 moeny->money
 mofdified->modified
 mohammedans->muslims
-moil->mohel
+moil->soil, mohel,
-moil->soil
 moint->mount
 moleclues->molecules
 momemtn->moment
@@ -7057,19 +7048,17 @@ mordern->modern
 moreso->more, more so,
 morever->moreover
 morgage->mortgage
-Morisette->Morissette
+morisette->morissette
-morover->moreover
+morrisette->morissette
-Morrisette->Morissette
 morroccan->moroccan
 morrocco->morocco
 morroco->morocco
 mortage->mortgage
-mose->more
+mose->more, mouse,
-mose->mouse
 moslty->mostly
 mosture->moisture
 motiviated->motivated
-mould->mold, mould, module
+mould->mold, mould, module,
 mounth->month
 mouspointer->mousepointer
 movebackwrd->movebackward
@@ -7283,7 +7272,7 @@ noveau->nouveau
 Novermber->November
 nowdays->nowadays
 nowe->now
-nto->not, disable due to \n
+nto->not, disabled due to \n
 nubmer->number
 nubmers->numbers
 nucular->nuclear
@@ -9245,7 +9234,6 @@ sempahore->semaphore
 sempahores->semaphores
 senario->scenario
 senarios->scenarios
-sence->sense
 sence->sense, since,
 sensistive->sensitive
 sensistively->sensitively
@@ -9555,7 +9543,7 @@ specificed->specified
 specificiation->specification
 specificiations->specifications
 specificly->specifically
-specificy->specify, specificity, specifically
+specificy->specify, specificity, specifically,
 specifing->specifying
 specifiy->specify
 specifiying->specifying
@@ -10244,7 +10232,7 @@ tkaing->taking
 tlaking->talking
 to to->to, to do,
 tobbaco->tobacco
-todays->today's, disable because of var names
+todays->today's, disabled because of var names
 todya->today
 togehter->together
 toghether->together
@@ -10890,9 +10878,9 @@ vulnerablility->vulnerability
 vyer->very
 vyre->very
 waht->what
-wan't->want, wasn't
+wan't->want, wasn't,
 wan->want
-wan;t->want, wasn't 
+wan;t->want, wasn't,
 wanna->want to, disabled because one might want to allow informal spelling
 want's->wants
 want;s->wants

codespell_lib/tests/test_basic.py

@@ -10,11 +10,12 @@ import sys
 import tempfile
 import warnings
 
-from nose.tools import with_setup, assert_equal, assert_true
+import pytest
 
 import codespell_lib as cs
 
 
+@pytest.fixture(scope='function')
 def reload_codespell_lib():
     try:
         reload  # Python 2.7
@@ -37,63 +38,63 @@ def test_command():
     """Test running the codespell executable"""
     # With no arguments does "."
     with TemporaryDirectory() as d:
-        assert_equal(run_codespell(cwd=d), 0)
+        assert run_codespell(cwd=d) == 0
         with open(op.join(d, 'bad.txt'), 'w') as f:
             f.write('abandonned\nAbandonned\nABANDONNED\nAbAnDoNnEd')
-        assert_equal(run_codespell(cwd=d), 4)
+        assert run_codespell(cwd=d) == 4
 
 
 def test_basic():
     """Test some basic functionality"""
-    assert_equal(cs.main('_does_not_exist_'), 0)
+    assert cs.main('_does_not_exist_') == 0
     with tempfile.NamedTemporaryFile('w') as f:
         pass
     with CaptureStdout() as sio:
-        assert_equal(cs.main('-D', 'foo', f.name), 1)  # missing dictionary
+        assert cs.main('-D', 'foo', f.name) == 1, 'missing dictionary'
     try:
-        assert_true('cannot find dictionary' in sio[1])
+        assert 'cannot find dictionary' in sio[1]
-        assert_equal(cs.main(f.name), 0)  # empty file
+        assert cs.main(f.name) == 0, 'empty file'
         with open(f.name, 'a') as f:
             f.write('this is a test file\n')
-        assert_equal(cs.main(f.name), 0)  # good
+        assert cs.main(f.name) == 0, 'good'
         with open(f.name, 'a') as f:
             f.write('abandonned\n')
-        assert_equal(cs.main(f.name), 1)  # bad
+        assert cs.main(f.name) == 1, 'bad'
         with open(f.name, 'a') as f:
             f.write('abandonned\n')
-        assert_equal(cs.main(f.name), 2)  # worse
+        assert cs.main(f.name) == 2, 'worse'
     finally:
         os.remove(f.name)
     with TemporaryDirectory() as d:
         with open(op.join(d, 'bad.txt'), 'w') as f:
             f.write('abandonned\nAbandonned\nABANDONNED\nAbAnDoNnEd')
-        assert_equal(cs.main(d), 4)
+        assert cs.main(d) == 4
         with CaptureStdout() as sio:
-            assert_equal(cs.main('-w', d), 0)
+            assert cs.main('-w', d) == 0
-        assert_true('FIXED:' in sio[1])
+        assert 'FIXED:' in sio[1]
         with open(op.join(d, 'bad.txt')) as f:
             new_content = f.read()
-        assert_equal(cs.main(d), 0)
+        assert cs.main(d) == 0
-        assert_equal(new_content, 'abandoned\nAbandoned\nABANDONED\nabandoned')
+        assert new_content == 'abandoned\nAbandoned\nABANDONED\nabandoned'
 
         with open(op.join(d, 'bad.txt'), 'w') as f:
             f.write('abandonned abandonned\n')
-        assert_equal(cs.main(d), 2)
+        assert cs.main(d) == 2
         with CaptureStdout() as sio:
-            assert_equal(cs.main('-q', '16', '-w', d), 0)
+            assert cs.main('-q', '16', '-w', d) == 0
-        assert_equal(sio[0], '')
+        assert sio[0] == ''
-        assert_equal(cs.main(d), 0)
+        assert cs.main(d) == 0
 
         # empty directory
         os.mkdir(op.join(d, 'test'))
-        assert_equal(cs.main(d), 0)
+        assert cs.main(d) == 0
 
         # hidden file
         with open(op.join(d, 'test.txt'), 'w') as f:
             f.write('abandonned\n')
-        assert_equal(cs.main(op.join(d, 'test.txt')), 1)
+        assert cs.main(op.join(d, 'test.txt')) == 1
         os.rename(op.join(d, 'test.txt'), op.join(d, '.test.txt'))
-        assert_equal(cs.main(op.join(d, '.test.txt')), 0)
+        assert cs.main(op.join(d, '.test.txt')) == 0
 
 
 def test_interactivity():
@@ -103,20 +104,20 @@ def test_interactivity():
     with tempfile.NamedTemporaryFile('w') as f:
         pass
     try:
-        assert_equal(cs.main(f.name), 0)  # empty file
+        assert cs.main(f.name) == 0, 'empty file'
         with open(f.name, 'w') as f:
             f.write('abandonned\n')
-        assert_equal(cs.main('-i', '-1', f.name), 1)  # bad
+        assert cs.main('-i', '-1', f.name) == 1, 'bad'
         with FakeStdin('y\n'):
-            assert_equal(cs.main('-i', '3', f.name), 1)
+            assert cs.main('-i', '3', f.name) == 1
         with CaptureStdout() as sio:
             with FakeStdin('n\n'):
-                assert_equal(cs.main('-w', '-i', '3', f.name), 0)
+                assert cs.main('-w', '-i', '3', f.name) == 0
-        assert_true('==>' in sio[0])
+        assert '==>' in sio[0]
         with CaptureStdout():
             with FakeStdin('x\ny\n'):
-                assert_equal(cs.main('-w', '-i', '3', f.name), 0)
+                assert cs.main('-w', '-i', '3', f.name) == 0
-        assert_equal(cs.main(f.name), 0)
+        assert cs.main(f.name) == 0
     finally:
         os.remove(f.name)
 
@@ -126,11 +127,11 @@ def test_interactivity():
     try:
         with open(f.name, 'w') as f:
             f.write('abandonned\n')
-        assert_equal(cs.main(f.name), 1)
+        assert cs.main(f.name) == 1
         with CaptureStdout():
             with FakeStdin(' '):  # blank input -> Y
-                assert_equal(cs.main('-w', '-i', '3', f.name), 0)
+                assert cs.main('-w', '-i', '3', f.name) == 0
-        assert_equal(cs.main(f.name), 0)
+        assert cs.main(f.name) == 0
     finally:
         os.remove(f.name)
 
@@ -141,27 +142,27 @@ def test_interactivity():
         with open(f.name, 'w') as f:
             f.write('ackward\n')
 
-        assert_equal(cs.main(f.name), 1)
+        assert cs.main(f.name) == 1
         with CaptureStdout():
             with FakeStdin(' \n'):  # blank input -> nothing
-                assert_equal(cs.main('-w', '-i', '3', f.name), 0)
+                assert cs.main('-w', '-i', '3', f.name) == 0
-        assert_equal(cs.main(f.name), 1)
+        assert cs.main(f.name) == 1
         with CaptureStdout():
             with FakeStdin('0\n'):  # blank input -> nothing
-                assert_equal(cs.main('-w', '-i', '3', f.name), 0)
+                assert cs.main('-w', '-i', '3', f.name) == 0
-        assert_equal(cs.main(f.name), 0)
+        assert cs.main(f.name) == 0
         with open(f.name, 'r') as f_read:
-            assert_equal(f_read.read(), 'awkward\n')
+            assert f_read.read() == 'awkward\n'
         with open(f.name, 'w') as f:
             f.write('ackward\n')
-        assert_equal(cs.main(f.name), 1)
+        assert cs.main(f.name) == 1
         with CaptureStdout() as sio:
             with FakeStdin('x\n1\n'):  # blank input -> nothing
-                assert_equal(cs.main('-w', '-i', '3', f.name), 0)
+                assert cs.main('-w', '-i', '3', f.name) == 0
-        assert_true('a valid option' in sio[0])
+        assert 'a valid option' in sio[0]
-        assert_equal(cs.main(f.name), 0)
+        assert cs.main(f.name) == 0
         with open(f.name, 'r') as f:
-            assert_equal(f.read(), 'backward\n')
+            assert f.read() == 'backward\n'
     finally:
         os.remove(f.name)
 
@@ -174,26 +175,25 @@ def test_summary():
         with CaptureStdout() as sio:
             cs.main(f.name)
         for ii in range(2):
-            assert_equal(sio[ii], '')  # no output
+            assert sio[ii] == '', 'no output'
         with CaptureStdout() as sio:
             cs.main(f.name, '--summary')
-        assert_equal(sio[1], '')  # stderr
+        assert sio[1] == '', 'stderr'
-        assert_true('SUMMARY' in sio[0])
+        assert 'SUMMARY' in sio[0]
-        assert_equal(len(sio[0].split('\n')), 5)  # no output
+        assert len(sio[0].split('\n')) == 5, 'no output'
         with open(f.name, 'w') as f:
             f.write('abandonned\nabandonned')
         with CaptureStdout() as sio:
             cs.main(f.name, '--summary')
-        assert_equal(sio[1], '')  # stderr
+        assert sio[1] == '', 'stderr'
-        assert_true('SUMMARY' in sio[0])
+        assert 'SUMMARY' in sio[0]
-        assert_equal(len(sio[0].split('\n')), 7)
+        assert len(sio[0].split('\n')) == 7
-        assert_true('abandonned' in sio[0].split()[-2])
+        assert 'abandonned' in sio[0].split()[-2]
     finally:
         os.remove(f.name)
 
 
-@with_setup(reload_codespell_lib, reload_codespell_lib)
+def test_ignore_dictionary(reload_codespell_lib):
-def test_ignore_dictionary():
     """Test ignore dictionary functionality"""
     with TemporaryDirectory() as d:
         with open(op.join(d, 'bad.txt'), 'w') as f:
@@ -202,17 +202,16 @@ def test_ignore_dictionary():
             pass
         with open(f.name, 'w') as f:
             f.write('abandonned\n')
-        assert_equal(cs.main('-I', f.name, d), 1)
+        assert cs.main('-I', f.name, d) == 1
 
 
-@with_setup(reload_codespell_lib, reload_codespell_lib)
+def test_custom_regex(reload_codespell_lib):
-def test_custom_regex():
     """Test custom word regex"""
     with TemporaryDirectory() as d:
         with open(op.join(d, 'bad.txt'), 'w') as f:
             f.write('abandonned_abondon\n')
-        assert_equal(cs.main(d), 0)
+        assert cs.main(d) == 0
-        assert_equal(cs.main('-r', "[a-z]+", d), 2)
+        assert cs.main('-r', "[a-z]+", d) == 2
 
 
 def test_exclude_file():
@@ -220,13 +219,13 @@ def test_exclude_file():
     with TemporaryDirectory() as d:
         with open(op.join(d, 'bad.txt'), 'wb') as f:
             f.write('abandonned 1\nabandonned 2\n'.encode('utf-8'))
-        assert_equal(cs.main(d), 2)
+        assert cs.main(d) == 2
         with tempfile.NamedTemporaryFile('w') as f:
             pass
         with open(f.name, 'wb') as f:
             f.write('abandonned 1\n'.encode('utf-8'))
-        assert_equal(cs.main(d), 2)
+        assert cs.main(d) == 2
-        assert_equal(cs.main('-x', f.name, d), 1)
+        assert cs.main('-x', f.name, d) == 1
 
 
 def test_encoding():
@@ -235,24 +234,24 @@ def test_encoding():
     with tempfile.NamedTemporaryFile('wb') as f:
         pass
     # with CaptureStdout() as sio:
-    assert_equal(cs.main(f.name), 0)
+    assert cs.main(f.name) == 0
     try:
         with open(f.name, 'wb') as f:
             f.write(u'naïve\n'.encode('utf-8'))
-        assert_equal(cs.main(f.name), 0)
+        assert cs.main(f.name) == 0
-        assert_equal(cs.main('-e', f.name), 0)
+        assert cs.main('-e', f.name) == 0
         with open(f.name, 'ab') as f:
             f.write(u'naieve\n'.encode('utf-8'))
-        assert_equal(cs.main(f.name), 1)
+        assert cs.main(f.name) == 1
         # Binary file warning
         with open(f.name, 'wb') as f:
             f.write(b'\x00\x00naiive\x00\x00')
         with CaptureStdout() as sio:
-            assert_equal(cs.main(f.name), 0)
+            assert cs.main(f.name) == 0
-        assert_true('WARNING: Binary file' in sio[1])
+        assert 'WARNING: Binary file' in sio[1]
         with CaptureStdout() as sio:
-            assert_equal(cs.main('-q', '2', f.name), 0)
+            assert cs.main('-q', '2', f.name) == 0
-        assert_equal(sio[1], '')
+        assert sio[1] == ''
     finally:
         os.remove(f.name)
 
@@ -262,20 +261,20 @@ def test_ignore():
     with TemporaryDirectory() as d:
         with open(op.join(d, 'good.txt'), 'w') as f:
             f.write('this file is okay')
-        assert_equal(cs.main(d), 0)
+        assert cs.main(d) == 0
         with open(op.join(d, 'bad.txt'), 'w') as f:
             f.write('abandonned')
-        assert_equal(cs.main(d), 1)
+        assert cs.main(d) == 1
-        assert_equal(cs.main('--skip=bad*', d), 0)
+        assert cs.main('--skip=bad*', d) == 0
-        assert_equal(cs.main('--skip=bad.txt', d), 0)
+        assert cs.main('--skip=bad.txt', d) == 0
         subdir = op.join(d, 'ignoredir')
         os.mkdir(subdir)
         with open(op.join(subdir, 'bad.txt'), 'w') as f:
             f.write('abandonned')
-        assert_equal(cs.main(d), 2)
+        assert cs.main(d) == 2
-        assert_equal(cs.main('--skip=bad*', d), 0)
+        assert cs.main('--skip=bad*', d) == 0
-        assert_equal(cs.main('--skip=*ignoredir*', d), 1)
+        assert cs.main('--skip=*ignoredir*', d) == 1
-        assert_equal(cs.main('--skip=ignoredir', d), 1)
+        assert cs.main('--skip=ignoredir', d) == 1
 
 
 def test_check_filename():
@@ -283,7 +282,7 @@ def test_check_filename():
     with TemporaryDirectory() as d:
         with open(op.join(d, 'abandonned.txt'), 'w') as f:
             f.write('.')
-        assert_equal(cs.main('-f', d), 1)
+        assert cs.main('-f', d) == 1
 
 
 class TemporaryDirectory(object):
@@ -385,3 +384,52 @@ def FakeStdin(text):
         yield
     finally:
         sys.stdin = oldin
+
+
+def test_dictionary_formatting():
+    """Test that all dictionary entries are in lower case and non-empty."""
+    err_dict = dict()
+    with open(op.join(op.dirname(__file__), '..', 'data',
+                      'dictionary.txt'), 'rb') as fid:
+        for line in fid:
+            err, rep = line.decode('utf-8').split('->')
+            err = err.lower()
+            assert err not in err_dict, 'entry already exists'
+            rep = rep.rstrip('\n')
+            assert len(rep) > 0, 'corrections must be non-empty'
+            if rep.count(','):
+                if not rep.endswith(','):
+                    assert 'disabled' in rep.split(',')[-1], \
+                        ('currently corrections must end with trailing "," (if'
+                         ' multiple corrections are available) or '
+                         'have "disabled" in the comment')
+            err_dict[err] = rep
+            reps = [r.strip() for r in rep.lower().split(',')]
+            reps = [r for r in reps if len(r)]
+            unique = list()
+            for r in reps:
+                if r not in unique:
+                    unique.append(r)
+            assert reps == unique, 'entries are not (lower-case) unique'
+
+
+def test_case_handling(reload_codespell_lib):
+    """Test that capitalized entries get detected properly."""
+    # Some simple Unicode things
+    with tempfile.NamedTemporaryFile('wb') as f:
+        pass
+    # with CaptureStdout() as sio:
+    assert cs.main(f.name) == 0
+    try:
+        with open(f.name, 'wb') as f:
+            f.write('this has an ACII error'.encode('utf-8'))
+        with CaptureStdout() as sio:
+            assert cs.main(f.name) == 1
+        assert 'ASCII' in sio[0]
+        with CaptureStdout() as sio:
+            assert cs.main('-w', f.name) == 0
+        assert 'FIXED' in sio[1]
+        with open(f.name, 'rb') as f:
+            assert f.read().decode('utf-8') == 'this has an ASCII error'
+    finally:
+        os.remove(f.name)

setup.cfg

@@ -1,8 +1,5 @@
-[nosetests]
+[tool:pytest]
-with-coverage = 1
+addopts = --cov=codespell_lib --showlocals -rs --cov-report=
-cover-package = codespell_lib
-detailed-errors = 1
-verbosity = 2
 
 [flake8]
 exclude = build, ci-helpers