Add support of multiple personal word list files (multiple options -P/--pwl) (closes #5)

ChangeLog.adoc

@@ -4,6 +4,10 @@
 :lang: en
+== Version 3.0 (under dev)
+
+* add support of multiple personal word list files (multiple options -P/--pwl) (issue #5)
+
 == Version 2.9 (2018-01-15)
 * add option "-n" (or "--skip-noqa") to not check "noqa"-commented lines (issue #2, issue #7)

msgcheck/msgcheck.py

@@ -76,9 +76,10 @@ The script returns:
     parser.add_argument('-d', '--dicts',
                         help='comma-separated list of extra dictionaries '
                         'to use (in addition to file language)')
-    parser.add_argument('-P', '--pwl',
-                        help='file with personal list of words used when '
-                        'checking spelling')
+    parser.add_argument('-P', '--pwl', action='append',
+                        help='file(s) with personal list of words used when '
+                        'checking spelling (this option can be given multiple '
+                        'times)')
     parser.add_argument('-m', '--only-misspelled', action='store_true',
                         help='display only misspelled words (no error, '
                         'line number and translation)')

msgcheck/po.py

@@ -29,6 +29,7 @@ import os
 import re
 import subprocess
 import sys
+import tempfile
 # enchant module is optional, spelling is checked on demand
 # (argument -s/--spell)
@@ -42,6 +43,19 @@ except ImportError:
 from . utils import count_lines, replace_formatters
+def build_temp_file_concat_files(filenames):
+    """Build a temporary file with concatenation of multiple files."""
+    if not filenames:
+        return None
+    tmp_file = tempfile.NamedTemporaryFile()
+    for filename in filenames:
+        if not os.path.isfile(filename):
+            raise IOError('file "{0}" not found'.format(filename))
+        tmp_file.write(open(filename, 'rb').read())
+    tmp_file.flush()
+    return tmp_file
+
+
 class PoReport(object):  # pylint: disable=too-few-public-methods
     """A message in report (commonly an error in detected in gettext file)."""
@@ -420,15 +434,11 @@ class PoCheck(object):
         if check in self.checks:
             self.checks[check] = bool(state)
-    def set_spelling_options(self, spelling, dicts, pwl):
+    def set_spelling_options(self, spelling, dicts, pwl_files):
         """Set spelling options."""
         self.spelling = spelling
         self.dicts = dicts
-        self.pwl = pwl
-
-        # check if pwl file exists
-        if pwl and not os.path.isfile(pwl):
-            raise IOError('pwl file "{0}" not found'.format(pwl))
+        self.pwl = build_temp_file_concat_files(pwl_files)
         # build extra checkers with dicts
         self.extra_checkers = []
@@ -454,7 +464,7 @@ class PoCheck(object):
             lang = po_file.props['language'] \
                 if self.spelling == 'str' else 'en'
             try:
-                _dict = DictWithPWL(lang, self.pwl)
+                _dict = DictWithPWL(lang, self.pwl.name if self.pwl else None)
                 checker.append(SpellChecker(_dict))
             except DictNotFoundError:
                 reports.append(PoReport(

tests/fr_spelling_id.po

@@ -36,8 +36,12 @@ msgstr ""
 "Plural-Forms: nplurals=2; plural=(n > 1);\n"
 #, c-format
-msgid "%.3fThsi is a test.\n"
-msgstr "%.3fCeci est un test.\n"
+msgid "%.3fThsi is a testone.\n"
+msgstr "%.3fCeci est un testone.\n"
+
+#, c-format
+msgid "%.3fThis is a testtwo.\n"
+msgstr "%.3fCeci est un testtwo.\n"
 msgid "Test 1"
 msgstr "Test 1"

tests/fr_spelling_str.po

@@ -36,8 +36,12 @@ msgstr ""
 "Plural-Forms: nplurals=2; plural=(n > 1);\n"
 #, c-format
-msgid "%.3fThis is a test.\n"
-msgstr "%.3fCeci est un test.\n"
+msgid "%.3fThis is a testone.\n"
+msgstr "%.3fCeci est un testone.\n"
+
+#, c-format
+msgid "%.3fThis is a testtwo.\n"
+msgstr "%.3fCeci est un testtwo.\n"
 #, c-format
 msgid "%.3fThis is another test.\n"

tests/pwl.txt

-test

tests/pwl1.txt

+testone

tests/pwl2.txt

+testtwo

tests/test_msgcheck.py

@@ -160,7 +160,33 @@ class TestMsgCheck(unittest.TestCase):  # pylint: disable=too-many-public-method
     def test_spelling_id(self):
         """Test spelling on source messages (English) of gettext files."""
         po_check = PoCheck()
-        po_check.set_spelling_options('id', None, local_path('pwl.txt'))
+        pwl_files = [local_path('pwl1.txt')]
+        po_check.set_spelling_options('id', None, pwl_files)
+        result = po_check.check_files([local_path('fr_spelling_id.po')])
+
+        # be sure we have 1 file in result
+        self.assertEqual(len(result), 1)
+
+        # the file has 2 spelling errors: "Thsi" and "errro"
+        errors = result[0][1]
+        self.assertEqual(len(errors), 3)
+        for i, word in enumerate(('Thsi', 'testtwo', 'errro')):
+            self.assertEqual(errors[i].idmsg, 'spelling-id')
+            self.assertTrue(isinstance(errors[i].message, list))
+            self.assertEqual(len(errors[i].message), 1)
+            self.assertEqual(errors[i].message[0], word)
+
+    def test_spelling_id_multilpe_pwl(self):
+        """
+        Test spelling on source messages (English) of gettext files
+        using multiple personal word lists.
+        """
+        po_check = PoCheck()
+        pwl_files = [
+            local_path('pwl1.txt'),
+            local_path('pwl2.txt'),
+        ]
+        po_check.set_spelling_options('id', None, pwl_files)
         result = po_check.check_files([local_path('fr_spelling_id.po')])
         # be sure we have 1 file in result
@@ -178,7 +204,39 @@ class TestMsgCheck(unittest.TestCase):  # pylint: disable=too-many-public-method
     def test_spelling_str(self):
         """Test spelling on translated messages of gettext files."""
         po_check = PoCheck()
-        po_check.set_spelling_options('str', None, local_path('pwl.txt'))
+        pwl_files = [local_path('pwl1.txt')]
+        po_check.set_spelling_options('str', None, pwl_files)
+        result = po_check.check_files([local_path('fr_spelling_str.po'),
+                                       local_path('fr_language.po')])
+
+        # be sure we have 2 files in result
+        self.assertEqual(len(result), 2)
+
+        # first file has 3 spelling errors: "CecX", "aabbcc" and "xxyyzz"
+        errors = result[0][1]
+        self.assertEqual(len(errors), 4)
+        for i, word in enumerate(('testtwo', 'CecX', 'aabbcc', 'xxyyzz')):
+            self.assertEqual(errors[i].idmsg, 'spelling-str')
+            self.assertTrue(isinstance(errors[i].message, list))
+            self.assertEqual(len(errors[i].message), 1)
+            self.assertEqual(errors[i].message[0], word)
+
+        # second file has 1 error: dict/language "xyz" not found
+        errors = result[1][1]
+        self.assertEqual(len(errors), 1)
+        self.assertEqual(errors[0].idmsg, 'dict')
+
+    def test_spelling_str_multiple_pwl(self):
+        """
+        Test spelling on translated messages of gettext files
+        using multiple personal word lists.
+        """
+        po_check = PoCheck()
+        pwl_files = [
+            local_path('pwl1.txt'),
+            local_path('pwl2.txt'),
+        ]
+        po_check.set_spelling_options('str', None, pwl_files)
         result = po_check.check_files([local_path('fr_spelling_str.po'),
                                        local_path('fr_language.po')])
@@ -209,8 +267,8 @@ class TestMsgCheck(unittest.TestCase):  # pylint: disable=too-many-public-method
         """Test spelling with a bad pwl option."""
         po_check = PoCheck()
         try:
-            po_check.set_spelling_options('str', None,
-                                          local_path('pwl_does_not_exist.txt'))
+            pwl_files = [local_path('pwl_does_not_exist.txt')]
+            po_check.set_spelling_options('str', None, pwl_files)
         except IOError:
             pass  # this exception is expected
         else: