Case insensitive String comparisons #2450

chriscallan · 2016-09-23T18:41:55Z

Added input parameter to all the keywords that commonly compare strings, and could use that feature;
Modified the function logic to trigger off the input parameter and compare strings case-insensitively;

…rings" perform string compares case-insensitively; Modified the Acceptance tests affected by the above 2 keywords; Added an Acceptance test to prove out the case-insensitive string comparison;

…rings" perform string compares case-insensitively; (#1) Modified the Acceptance tests affected by the above 2 keywords; Added an Acceptance test to prove out the case-insensitive string comparison;

…bles, such that returned messages match the currently expected output (e.g. "FAIL: 'test' != 'TEST1'" rather than"FAIL: 'test' != 'test1'" when the user passed in "TEST1"); Added Acceptance tests for all the keywords affected by the case-insensitivity changes;

…d not be changed when running

@krizex

PR #2438 by @krizex already fixed the problem with settings preserving some date between runds he had reported as #2437. This commit slightly cleans up the code and unit tests in the PR. A new unit/integration test to validate the real problem with --rerunfailed that this problem caused is added too.

Added Pygments styles and configured docutils to be compatible with them. Embedding Pygments styles to all Libdoc outputs may be a bit questionable, but they are small enough that I don't think it is a real problem. We may also later add syntax highlighting to Libdoc's default syntax too. Also made background color used by examples etc. in Libdoc outputs and also in logs and reports a little lighter. Syntax highlighting looks better on a little lighter background and this background color is used also in the User Guide.

No need to use external Robot lexer anymore now that Pygments 2.1 suppors all new features. Also made it an error to use invalid syntax highlight language.

…rings" perform string compares case-insensitively; Modified the Acceptance tests affected by the above 2 keywords; Added an Acceptance test to prove out the case-insensitive string comparison;

…bles, such that returned messages match the currently expected output (e.g. "FAIL: 'test' != 'TEST1'" rather than"FAIL: 'test' != 'test1'" when the user passed in "TEST1"); Added Acceptance tests for all the keywords affected by the case-insensitivity changes;

…rk into CaseInsensitive

…rings" perform string compares case-insensitively; Modified the Acceptance tests affected by the above 2 keywords; Added an Acceptance test to prove out the case-insensitive string comparison;

…bles, such that returned messages match the currently expected output (e.g. "FAIL: 'test' != 'TEST1'" rather than"FAIL: 'test' != 'test1'" when the user passed in "TEST1"); Added Acceptance tests for all the keywords affected by the case-insensitivity changes;

…rings" perform string compares case-insensitively; Modified the Acceptance tests affected by the above 2 keywords; Added an Acceptance test to prove out the case-insensitive string comparison;

…bles, such that returned messages match the currently expected output (e.g. "FAIL: 'test' != 'TEST1'" rather than"FAIL: 'test' != 'test1'" when the user passed in "TEST1"); Added Acceptance tests for all the keywords affected by the case-insensitivity changes;

Fixes #2360.

This allows catching these errors using e.g. Run Keyword And Expect Error and also allows teardowns to continue after them.

Fixes #2407.

…d not be changed when running

@krizex

PR #2438 by @krizex already fixed the problem with settings preserving some date between runds he had reported as #2437. This commit slightly cleans up the code and unit tests in the PR. A new unit/integration test to validate the real problem with --rerunfailed that this problem caused is added too.

Added Pygments styles and configured docutils to be compatible with them. Embedding Pygments styles to all Libdoc outputs may be a bit questionable, but they are small enough that I don't think it is a real problem. We may also later add syntax highlighting to Libdoc's default syntax too. Also made background color used by examples etc. in Libdoc outputs and also in logs and reports a little lighter. Syntax highlighting looks better on a little lighter background and this background color is used also in the User Guide.

No need to use external Robot lexer anymore now that Pygments 2.1 suppors all new features. Also made it an error to use invalid syntax highlight language.

…rings" perform string compares case-insensitively; Modified the Acceptance tests affected by the above 2 keywords; Added an Acceptance test to prove out the case-insensitive string comparison;

…bles, such that returned messages match the currently expected output (e.g. "FAIL: 'test' != 'TEST1'" rather than"FAIL: 'test' != 'test1'" when the user passed in "TEST1"); Added Acceptance tests for all the keywords affected by the case-insensitivity changes;

…d not be changed when running

@krizex

PR #2438 by @krizex already fixed the problem with settings preserving some date between runds he had reported as #2437. This commit slightly cleans up the code and unit tests in the PR. A new unit/integration test to validate the real problem with --rerunfailed that this problem caused is added too.

…rings" perform string compares case-insensitively; Modified the Acceptance tests affected by the above 2 keywords; Added an Acceptance test to prove out the case-insensitive string comparison;

…bles, such that returned messages match the currently expected output (e.g. "FAIL: 'test' != 'TEST1'" rather than"FAIL: 'test' != 'test1'" when the user passed in "TEST1"); Added Acceptance tests for all the keywords affected by the case-insensitivity changes;

…rings" perform string compares case-insensitively; Modified the Acceptance tests affected by the above 2 keywords; Added an Acceptance test to prove out the case-insensitive string comparison;

…bles, such that returned messages match the currently expected output (e.g. "FAIL: 'test' != 'TEST1'" rather than"FAIL: 'test' != 'test1'" when the user passed in "TEST1"); Added Acceptance tests for all the keywords affected by the case-insensitivity changes;

…rings" perform string compares case-insensitively; Modified the Acceptance tests affected by the above 2 keywords; Added an Acceptance test to prove out the case-insensitive string comparison;

…bles, such that returned messages match the currently expected output (e.g. "FAIL: 'test' != 'TEST1'" rather than"FAIL: 'test' != 'test1'" when the user passed in "TEST1"); Added Acceptance tests for all the keywords affected by the case-insensitivity changes;

pekkaklarck

I didn't yet look at tests, but code looks very good. Most important things to be fixed:

Keywords that accept a container should be enhanced to convert items in the container to lower case.
Docs and implementation with keywords that always expect arguments to be passed as strings can be simplified.

pekkaklarck · 2016-10-05T11:25:21Z

src/robot/libraries/BuiltIn.py

+        ``ignore_case`` is False by default.  If True, it indicates that
+        ``first`` and ``second`` should be compared case-insensitively,
+        provided that ``first`` and ``second`` are string types.  See `Boolean
+        arguments` section for more details.  (New in RF 3.0.1)


Very good docs and formatting. Two minor issues:

(New in RF 3.0.1) is a inconsistent with how this kind of information is displayed elsewhere. Could you change it to, for example, This option is new in Robot Framework 3.0.1.?

ignore_case is False by default. If True, it indicates ... is a bit redundant considering that the argument is ignore_case=False. The current wording is fine, but perhaps something like If ignore_case is given a true value, it indicates ... would be a little better.

pekkaklarck · 2016-10-05T11:30:44Z

src/robot/libraries/BuiltIn.py

        If both arguments are multiline strings, the comparison is done using
        `multiline string comparisons`.
        """
+        if is_truthy(ignore_case) and is_string(first) and is_string(second):
+            first = first.lower()
+            second = second.lower()


This code is great, but it is unfortunately subject to this annoying IronPython bug:
http://ironpython.codeplex.com/workitem/33133

We have utils.lower() utility that could be used instead of string.lower() or alternatively we could use string.upper() here. Or then we could argue that we don't need to care about this because the bug is fixed in latest IronPython 2.7.x releases. What do you think?

I'd prefer to not add workarounds for Python engines that have bugs which need to be resolved.

Adding workarounds like that is annoying, but the framework working incorrectly on some platform is worse. It's especially bad if tests fail on that platform when run on CI. In this particular case we can argue that people ought to use newish IronPython versions where this problem is fixed. Like we do on our CI.

So, are you advocating for the stance that users of IronPython 2.7.x and below should upgrade to fix the bug with "string.lower()" in their interpreter or modify all the "string.lower()" calls to "utils.lower(string)" to workaround that bug?
My opinion says to leave the string conversions and have the users upgrade since bugs with "string.lower()" are likely giving problems elsewhere.

pekkaklarck · 2016-10-05T11:38:10Z

src/robot/libraries/BuiltIn.py

+        ``ignore_case`` is False by default.  If True, it indicates that
+        ``first`` and ``second`` should be compared case-insensitively,
+        provided that ``first`` and ``second`` are string types.  See `Boolean
+        arguments` section for more details.  (New in RF 3.0.1)


With this particular keyword, provided that first and second are string types is misleading. This keyword first converts inputs to strings and thus values are always compared case-insensitively if when ignore_case is enabled. I guess the whole provided that ... piece could just be removed.

Fair enough.

pekkaklarck · 2016-10-05T11:39:23Z

src/robot/libraries/BuiltIn.py

        See `Should Be Equal` for an explanation on how to override the default
        error message with ``msg`` and ``values``.
        """
        self._log_types_at_info_if_different(first, second)
        first, second = [self._convert_to_string(i) for i in (first, second)]
+        if is_truthy(ignore_case) and is_string(first) and is_string(second):


No need to use is_string here.

Good point.

pekkaklarck · 2016-10-05T11:40:06Z

src/robot/libraries/BuiltIn.py

        """Fails if objects are unequal after converting them to strings.

        See `Should Be Equal` for an explanation on how to override the default
        error message with ``msg`` and ``values``.

+        ``ignore_case`` is False by default.  If True, it indicates that
+        ``first`` and ``second`` should be compared case-insensitively,
+        provided that ``first`` and ``second`` are string types.  See `Boolean


provided that ... are string types is wrong also here.

pekkaklarck · 2016-10-05T12:38:00Z

src/robot/libraries/BuiltIn.py

        """Fails if ``item1`` does not contain ``item2`` ``count`` times.

        Works with strings, lists and all objects that `Get Count` works
        with. The default error message can be overridden with ``msg`` and
        the actual count is always logged.

+        ``ignore_case`` is False by default.  If True, it indicates that
+        ``item1`` and ``item2`` should be compared case-insensitively,
+        provided that ``item1`` and ``item2`` are string types.  See `Boolean


Same discussion about docs and implementation as with should_not_contain applies also to this keyword.

pekkaklarck · 2016-10-05T12:39:11Z

src/robot/libraries/BuiltIn.py

        """Fails if the given ``string`` matches the given ``pattern``.

        Pattern matching is similar as matching files in a shell, and it is
        always case-sensitive. In the pattern ``*`` matches to anything and
        ``?`` matches to any single character.

+        ``ignore_case`` is False by default.  If True, it indicates that
+        ``first`` and ``second`` should be compared case-insensitively,
+        provided that ``first`` and ``second`` are string types.  See `Boolean


Similarly as with other keywords that should always be used with strings, provided that ... are string types is unnecessary.

pekkaklarck · 2016-10-05T12:40:18Z

src/robot/libraries/BuiltIn.py

+        if is_truthy(ignore_case) and is_string(string) and is_string(pattern):
+            prv_string = string.lower()
+            prv_pattern = pattern.lower()
+        if self._matches(prv_string, prv_pattern):


No need to explicitly check are string and pattern strings or not. Simply using

if is_truthy(ignore_case): string = string.lower() pattern = pattern.lower()

ought to be enough.

pekkaklarck · 2016-10-05T12:41:01Z

src/robot/libraries/BuiltIn.py

        """Fails unless the given ``string`` matches the given ``pattern``.

        Pattern matching is similar as matching files in a shell, and it is
        always case-sensitive. In the pattern, ``*`` matches to anything and
        ``?`` matches to any single character.

+        ``ignore_case`` is False by default.  If True, it indicates that
+        ``string`` and ``pattern`` should be compared case-insensitively,
+        provided that ``string`` and ``pattern`` are string types.  See `Boolean


provided that ... are string types is unnecessary here too.

pekkaklarck · 2016-10-05T12:41:20Z

src/robot/libraries/BuiltIn.py

+        if is_truthy(ignore_case) and is_string(string) and is_string(pattern):
+            prv_string = string.lower()
+            prv_pattern = pattern.lower()
+        if not self._matches(prv_string, prv_pattern):


No need to check are string and pattern strings here either.

pekkaklarck · 2016-10-05T12:48:15Z

This is quite a big PR. It would be easier to review changes if each keyword and its "not" variant would be in a separate pull request. Simple cases like Should (Not) Be Equal would be ready to be merged with minor changes, but there were also more complicated ones like Should Not Contain. I guess we can manage also with just this one PR, but take that into account if you end up recreating the PR for other reasons.

pekkaklarck · 2016-10-07T09:57:53Z

As already discussed in code comments, some keywords currently show the lower cased strings in the error message and others show the original. Probably showing the original is better, but that needs to be done consistently. If we agree it is a good idea, we should enhance _get_string_msg to add, for example, (case-insensitively) at the end of the default error message it creates when case_insensitive=True. Otherwise we will get confusing errors like 'foo' == 'FOO' or 'Foo' starts with 'f'. Something like 'foo' == 'FOO' (case-insensitively) and 'Foo' starts with 'f' (case-insensitively) is better.

On the code level we have two general possibilities to preserve the original values. Either we can just keep a reference to the original values and use them in the error message like this:

orig_first = first
orig_second = second
if is_truthy(case_insensitive):
    first = first.lower()
    second = second.lower()
if not first.startswith(second):
    # raise exception using `orig_first` and `orig_second`

Alternatively we could do the actual validation differently depending on case-sensitivity:

if not is_truthy(case_insensitive):
    passed = first.startswith(second)
else:
    passed = first.lower().startswith(second.lower())
if not passed:
    # raise exception using `first` and `second`

Probably the former is more explicit and easier to understand. In this case the latter is shorted, but if the validation part is more complex, duplicating it isn't a good idea.

…ileLinksInOutput # Conflicts: # src/robot/libraries/BuiltIn.py

chriscallan · 2016-10-10T19:52:30Z

I just revisited all of the above, and I'm more convinced that your suggestion to handle this update as multiple bite-size PR's is much wiser idea. I'll go ahead and close this PR and then reintroduce them as a single PR focused on one logical grouping of keywords (i.e. "Should Be Equal"/"Should Not Be Equal"), keeping all your above feedback in mind of course :-)
Sound good?

pekkaklarck · 2016-10-12T14:45:19Z

Sounds great!

chriscallan and others added 19 commits September 22, 2016 08:19

Added the input parameters and logic to have "Should Be Equal"/"As St…

3edb9ea

…rings" perform string compares case-insensitively; Modified the Acceptance tests affected by the above 2 keywords; Added an Acceptance test to prove out the case-insensitive string comparison;

Added the input parameters and logic to have "Should Be Equal"/"As St…

f662c68

…rings" perform string compares case-insensitively; (#1) Modified the Acceptance tests affected by the above 2 keywords; Added an Acceptance test to prove out the case-insensitive string comparison;

#2437 Options' default value of RobotSettings and RebotSettings shoul…

87297c2

…d not be changed when running

Fix updated note in UG

c5a1f2f

Cleaned up unit tests for handling state between runs

ba8a32a

ug2html.py: Require new enough Pygments version.

4e63a85

No need to use external Robot lexer anymore now that Pygments 2.1 suppors all new features. Also made it an error to use invalid syntax highlight language.

Added the input parameters and logic to have "Should Be Equal"/"As St…

7677df1

…rings" perform string compares case-insensitively; Modified the Acceptance tests affected by the above 2 keywords; Added an Acceptance test to prove out the case-insensitive string comparison;

Merge branch 'CaseInsensitive' of github.com:chriscallan/robotframewo…

0610eb7

…rk into CaseInsensitive

Added the input parameters and logic to have "Should Be Equal"/"As St…

d9eedaa

…rings" perform string compares case-insensitively; Modified the Acceptance tests affected by the above 2 keywords; Added an Acceptance test to prove out the case-insensitive string comparison;

Added the input parameters and logic to have "Should Be Equal"/"As St…

988fdee

…rings" perform string compares case-insensitively; Modified the Acceptance tests affected by the above 2 keywords; Added an Acceptance test to prove out the case-insensitive string comparison;

Merge branch 'master' into CaseInsensitive

966d672

Merge branch 'CaseInsensitive'

338259c

Added test case for "Should Be Equal" that exercises case insensitivity;

7d00973

chriscallan mentioned this pull request Sep 30, 2016

Builtin: Add case-insensitivity support to comparison keywords #2439

Closed

8 tasks

chriscallan and others added 10 commits October 3, 2016 09:23

Added my name to the authors.txt file

c35c022

Minor grammar adjustment to make OS doc more clear.

5344c51

Updated copyrights of Java files. #2309

b9bde24

Updated copyrights of Python files. #2309

ffa0745

Update copyrights in test READMEs. #2309

b54a167

Added Robot Framework Foundation to the main copyright file #2309

4c5fc59

Updated UG and API doc copyrights. #2309

ae0174e

Minor --help update. Including removing NSN reference (#2309)

1eefb8b

Updated copyright/sponsor/license docs in README and UG. #2309

45ef9d7

Added ItemList.pop(). Fixes #2303.

dac751f

pekkaklarck and others added 22 commits October 3, 2016 09:44

Fix HTML failures in setups/teardowns and with continuable failures.

db87277

Fixes #2360.

Variable assign errors are not syntax errors anymore.

206a9bb

This allows catching these errors using e.g. Run Keyword And Expect Error and also allows teardowns to continue after them.

Added -X as a short opt for --exitonfailure.

f7ed184

Fixes #2407.

fix-python-code-in-doc

4d18f90

fix For-in-range loop Documentation in AdvancedFeatures.rst

9ccb016

Fix updated note in UG

180cf85

#2437 Options' default value of RobotSettings and RebotSettings shoul…

1f58853

…d not be changed when running

Cleaned up unit tests for handling state between runs

7c95c18

ug2html.py: Require new enough Pygments version.

bd30600

No need to use external Robot lexer anymore now that Pygments 2.1 suppors all new features. Also made it an error to use invalid syntax highlight language.

Added the input parameters and logic to have "Should Be Equal"/"As St…

5520cfc

…rings" perform string compares case-insensitively; Modified the Acceptance tests affected by the above 2 keywords; Added an Acceptance test to prove out the case-insensitive string comparison;

#2437 Options' default value of RobotSettings and RebotSettings shoul…

154edd9

…d not be changed when running

Added the input parameters and logic to have "Should Be Equal"/"As St…

a5e87ac

…rings" perform string compares case-insensitively; Modified the Acceptance tests affected by the above 2 keywords; Added an Acceptance test to prove out the case-insensitive string comparison;

Added the input parameters and logic to have "Should Be Equal"/"As St…

ddde7a5

…rings" perform string compares case-insensitively; Modified the Acceptance tests affected by the above 2 keywords; Added an Acceptance test to prove out the case-insensitive string comparison;

Added the input parameters and logic to have "Should Be Equal"/"As St…

24389ee

…rings" perform string compares case-insensitively; Modified the Acceptance tests affected by the above 2 keywords; Added an Acceptance test to prove out the case-insensitive string comparison;

Added my name to the authors.txt file

682a584

pekkaklarck requested changes Oct 5, 2016

View reviewed changes

chriscallan added 2 commits October 10, 2016 12:00

Set of changes based on the feedback from pekkaklarck

4f55698

Merge branch 'master' of github.com:chriscallan/robotframework into F…

35002cd

…ileLinksInOutput # Conflicts: # src/robot/libraries/BuiltIn.py

chriscallan closed this Oct 12, 2016

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Case insensitive String comparisons #2450

Case insensitive String comparisons #2450

chriscallan commented Sep 23, 2016

pekkaklarck left a comment

pekkaklarck Oct 5, 2016 •

edited

Loading

pekkaklarck Oct 5, 2016

chriscallan Oct 6, 2016

pekkaklarck Oct 7, 2016 •

edited

Loading

chriscallan Oct 10, 2016 •

edited

Loading

pekkaklarck Oct 5, 2016

chriscallan Oct 6, 2016

pekkaklarck Oct 5, 2016

chriscallan Oct 6, 2016

pekkaklarck Oct 5, 2016

pekkaklarck Oct 5, 2016

pekkaklarck Oct 5, 2016

pekkaklarck Oct 5, 2016

pekkaklarck Oct 5, 2016

pekkaklarck Oct 5, 2016

pekkaklarck commented Oct 5, 2016 •

edited

Loading

pekkaklarck commented Oct 7, 2016 •

edited

Loading

chriscallan commented Oct 10, 2016

pekkaklarck commented Oct 12, 2016

Case insensitive String comparisons #2450

Case insensitive String comparisons #2450

Conversation

chriscallan commented Sep 23, 2016

pekkaklarck left a comment

Choose a reason for hiding this comment

pekkaklarck Oct 5, 2016 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

pekkaklarck Oct 7, 2016 • edited Loading

Choose a reason for hiding this comment

chriscallan Oct 10, 2016 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

pekkaklarck commented Oct 5, 2016 • edited Loading

pekkaklarck commented Oct 7, 2016 • edited Loading

chriscallan commented Oct 10, 2016

pekkaklarck commented Oct 12, 2016

pekkaklarck Oct 5, 2016 •

edited

Loading

pekkaklarck Oct 7, 2016 •

edited

Loading

chriscallan Oct 10, 2016 •

edited

Loading

pekkaklarck commented Oct 5, 2016 •

edited

Loading

pekkaklarck commented Oct 7, 2016 •

edited

Loading