Merge lp:~thelinuxguy/openlp/fix-newline-bug into lp:openlp

[Simon Hanna]

Fix the fix for 1727517

[Tim Bentley (trb143)]

One minor fix as I do not like the rename of the field

review: Needs Fixing

[Phill (phill-ridout)]

See in line. One question and a few (more) minor fixes please.

review: Needs Fixing

[Simon Hanna (thelinuxguy)]

will update shortly

[Phill (phill-ridout)]

I've done some research in to this. Officially the only code points allowed are:

"
[2] Char ::= #x9 | #xA | #xD | [#x20-#xD7FF] | [#xE000-#xFFFD] | [#x10000-#x10FFFF] /* any Unicode character, excluding the surrogate blocks, FFFE, and FFFF. */
" ( https://www.w3.org/TR/REC-xml/#charsets)

So going on this our regex should be:

re.compile(r'[\x00-\x08\x0B\x0C\x0E-\x1F\x7F-\x9F]')

we probably should filter out the other code points too.

Its worth noting that REPLACMENT_CHARS_MAP replaces the vertical tab and form feed chars with "\n\n" before the CONTROL_CHARS regex filters them out!

So in summary please replace the regex with the one in the inline comment. (Note: its not tested)

review: Needs Fixing

[Simon Hanna (thelinuxguy)]

I wonder what the reason behind replacing vertial tabs with new lines is...

Looking at where the regex is being used I'm inclined to say the whole thing needs to be reworked.

It's also used in the filename cleaning where linefeeds are strange, but then again I don't know what the input always is.

I'm changing it to what you request, but you should really take a serious look at it. From what I see, you are the last one that touched the invalid file regex, I highly doubt this is what you actually want to happen. If it is, it is a mystery to me...

[Phill (phill-ridout)]

One small typo, and a couple comments that need changing.

review: Needs Fixing

[Phill (phill-ridout)]

> One small typo, and a couple comments that need changing.

Sorry!

[lp:~thelinuxguy/openlp/fix-newline-bug]

2821. By Simon Hanna on 2018-04-16

Fix typo

2822. By Simon Hanna on 2018-04-17

Fix comment

2823. By Simon Hanna on 2018-04-17

Change parameter name

Unmerged revisions

=== modified file 'openlp/core/common/__init__.py'
--- openlp/core/common/__init__.py	2018-02-24 16:10:02 +0000
+++ openlp/core/common/__init__.py	2018-04-17 10:27:58 +0000
@@ -44,7 +44,7 @@
 
 FIRST_CAMEL_REGEX = re.compile('(.)([A-Z][a-z]+)')
 SECOND_CAMEL_REGEX = re.compile('([a-z0-9])([A-Z])')
-CONTROL_CHARS = re.compile(r'[\x00-\x1F\x7F-\x9F]')
+CONTROL_CHARS = re.compile(r'[\x00-\x08\x0B\x0C\x0E-\x1F\x7F-\x9F]')
 INVALID_FILE_CHARS = re.compile(r'[\\/:\*\?"<>\|\+\[\]%]')
 IMAGES_FILTER = None
 REPLACMENT_CHARS_MAP = str.maketrans({'\u2018': '\'', '\u2019': '\'', '\u201c': '"', '\u201d': '"', '\u2026': '...',
@@ -471,15 +471,15 @@
         log.exception('Error detecting file encoding')
 
 
-def normalize_str(irreg_str):
+def normalize_str(string):
     """
     Normalize the supplied string. Remove unicode control chars and tidy up white space.
 
-    :param str irreg_str: The string to normalize.
+    :param str string: The string to normalize.
     :return: The normalized string
     :rtype: str
     """
-    irreg_str = irreg_str.translate(REPLACMENT_CHARS_MAP)
-    irreg_str = CONTROL_CHARS.sub('', irreg_str)
-    irreg_str = NEW_LINE_REGEX.sub('\n', irreg_str)
-    return WHITESPACE_REGEX.sub(' ', irreg_str)
+    string = string.translate(REPLACMENT_CHARS_MAP)
+    string = CONTROL_CHARS.sub('', string)
+    string = NEW_LINE_REGEX.sub('\n', string)
+    return WHITESPACE_REGEX.sub(' ', string)
 
=== modified file 'tests/functional/openlp_core/common/test_common.py'
--- tests/functional/openlp_core/common/test_common.py	2017-12-29 09:15:48 +0000
+++ tests/functional/openlp_core/common/test_common.py	2018-04-17 10:27:58 +0000
@@ -25,8 +25,8 @@
 from unittest import TestCase
 from unittest.mock import MagicMock, call, patch
 
-from openlp.core.common import clean_button_text, de_hump, extension_loader, is_macosx, is_linux, is_win, \
-    path_to_module, trace_error_handler
+from openlp.core.common import clean_button_text, de_hump, extension_loader, is_macosx, is_linux, \
+    is_win, normalize_str, path_to_module, trace_error_handler
 from openlp.core.common.path import Path
 
 
@@ -211,6 +211,30 @@
             assert is_win() is False, 'is_win() should return False'
             assert is_macosx() is False, 'is_macosx() should return False'
 
+    def test_normalize_str_leaves_newlines(self):
+        # GIVEN: a string containing newlines
+        string = 'something\nelse'
+        # WHEN: normalize is called
+        normalized_string = normalize_str(string)
+        # THEN: string is unchanged
+        assert normalized_string == string
+
+    def test_normalize_str_removes_null_byte(self):
+        # GIVEN: a string containing a null byte
+        string = 'somet\x00hing'
+        # WHEN: normalize is called
+        normalized_string = normalize_str(string)
+        # THEN: nullbyte is removed
+        assert normalized_string == 'something'
+
+    def test_normalize_str_replaces_crlf_with_lf(self):
+        # GIVEN: a string containing crlf
+        string = 'something\r\nelse'
+        # WHEN: normalize is called
+        normalized_string = normalize_str(string)
+        # THEN: crlf is replaced with lf
+        assert normalized_string == 'something\nelse'
+
     def test_clean_button_text(self):
         """
         Test the clean_button_text() function.