Behave

-------------------------------------------------------------------------------

Latest stable version.
Same as last development version.

-------------------------------------------------------------------------------

NEWS and CHANGES:

- Install:

* Require now parse>=1.6.2 to enforce log-bugfix 14 (was: parse>=1.6)

- Running:

* load_step_definitions: Are now sorted before loading (due to: Linux, ...).
* NEW: Use lazy-loading for formatters if possible (speed up self-tests by 20%).

- Model:

* location: Now a FileLocation object (was: string), required for ordering.

- Formatters:

* NEW: progress3 formatter, ScenarioStepProgressFormatter (provided by: roignac).
* NEW: sphinx.steps formatter, generate Sphinx-based docs for step definitions (related to 166).
* NEW: steps formatter, shows available step definitions.
* NEW: steps.doc formatter, shows documentation of step definitions (related to: 55).
* NEW: steps.usage formatter, shows where step definitions are used.
* RENAMED: json-pretty, tag_count, tag_location => json.pretty, tags, tags.location
* help: Shows now a better formatted list (improve readability).

IMPROVEMENT:

* issue 166: behave should have a tool (or formatter) that generates Sphinx-based documentation (basics provided).

FIXED:

* issue 172: JUnit report filename sometimes truncated (provided by: roignac).
* issue 171: Importing step from other step file fails with AmbiguousStep Error.
* issue 165: FIX issue 114: do not print a blank line when the feature is skipped (provided by: florentx).
* issue 164: StepRegistry.find_match() extends registered step_type lists.
* issue 122: Failing selftest feature: selftest.features/duplicated_step.feature.
* issue 110: Normalize paths provided at the command line (provided by: jesper).

-------------------------------------------------------------------------------

NEWS and CHANGES:

- Running (and model):

* NEW: Support scenario file locations on command-line, ala: "{filename}:{line}" (related to: 160).
* Formatters are now created only once (was: once for each feature).
* Scenarios can be now be selected by name or regular expression (87).
* Dry-run mode: Detects now undefined steps.
* Dry-run mode: Uses untested counts now (was using: skipped counts).
* Run decision logic: Use ModelElement.mark_skipped() to preselect what not to run.
* Run decision logic: Use ModelElement.should_run() to decide if element should run.

- Parsing (and model):

* Parser: Add support for Scenario/ScenarioOutline descriptions (related to: 79).
* Parser: Refactor to simplify and avoid code duplications (related to: 79).
* Parser: Improve diagnostics when parse errors occur.
* Parser: Check that Backgrounds have no tags.
* NEW: json_parser, parses JSON output and builds model.
* json_parser: Add support for scenario descriptions (related to: 79).

- Formatters:

* INCOMPATIBLE CHANGE:
Formatter Ctor uses now StreamOpener instead of opened Stream.
Formatter output streams are now opened late, under control of the formatter.
This allows the formatter to support also directory mode (if needed).
Needed for RerunFormatter whose file was overwritten before it was read (160).

* NEW: RerunFormatter to simplify to rerun last failing scenarios (related to: 160).
* NEW: TagLocationFormatter, shows where tags are used.
* NEW: TagCountFormatter, shows which tags are used and how often (reborn).
* JSONFormatter: Use JSON array mode now (related to: 161).
* JSONFormatter: Added "type" to Background, Scenario, ScenerioOutline (related to: 161).
* JSONFormatter: Added "error_message" to result (related to: 161).
* JSONFormatter: Use now list<lines> instead of string for multi-line text (related to: 161).
* JSONFormatter: Add support for scenario descriptions (related to: 79).
* JSONFormatter: Generates now valid JSON (well-formed).
* PlainFormatter: Shows now multi-line step parts (text, table), too.
* PrettyFormatter: Enters now monochrome mode if output is piped/redirected.
* ProgressFormatter: Flushes now output to provide better feedback.

- Reporters:

* JUnitReporter: Show complete scenario w/ text/tables. Improve readability.
* SummaryReporter: Summary shows now untested items if one or more exist.

- Testing, development:

* tox: Use tox now in off-line mode per default (use: "tox -e init"...).
* Add utility script to show longest step durations based on JSON data.
* JSON: Add basic JSON schema to support JSON output validation (related to: 161).
* JSON: Add helper script to validate JSON output against its schema (related to: 161).


IMPROVEMENT:

* issue 161: JSONFormatter: Should use a slightly different output schema (provided by: jenisys)
* issue 160: Support rerun file with failed features/scenarios during the last test run (provided by: jenisys)
* issue 154: Support multiple formatters (provided by: roignac, jenisys)
* issue 103: sort feature file by name in a given directory (provided by: gurneyalex).
* issue 102: Add configuration file setting for specifying default feature paths (provided by: lrowe).
* issue 87: Add --name option support (provided by: johbo, jenisys).
* issue 79: Provide Support for Scenario Descriptions (provided by: caphrim007, jenisys).
* issue 42: Show all undefined steps taking tags into account (provided by: roignac, jenisys)

FIXED:

* issue 162 Unnecessary ContextMaskWarnings when assert fails or exception is raised (provided by: jenisys).
* issue 159: output stream is wrapped twice in the codecs.StreamWriter (provided by: florentx).
* issue 153: The runtime should not by-pass the formatter to print line breaks minor.
* issue 152: Fix encoding issues (provided by: devainandor)
* issue 145: before_feature/after_feature should not be skipped (provided by: florentx).
* issue 141: Don't check for full package in issue 112 (provided by: roignac).
* issue 125: Duplicate "Captured stdout" if substep has failed (provided by: roignac).
* issue 60: JSONFormatter has several problems (last problem fixed).
* issue 48: Docs aren't clear on how Background is to be used.
* issue 47: Formatter processing chain is broken (solved by: 154).
* issue 33: behave 1.1.0: Install fails under Windows (verified, solved already some time ago).
* issue 28: Install fails on Windows (verified, solved already some time ago).

-------------------------------------------------------------------------------

NEWS and CHANGES:

* NullFormatter provided
* model.Row: Changed Ctor parameter ordering, move seldom used to the end.
* model.Row: Add methods .get(), .as_dict() and len operator (related to: 27).
* Introduce ``behave.compat`` as compatibility layer for Python versions.

IMPROVEMENT:

* issue 117: context.execute_steps() should also support steps with multi-line text or table
* issue 116: SummaryReporter shows list of failing scenarios (provided by: roignac).
* issue 112: Improvement to AmbiguousStep error diagnostics
* issue 74: django-behave module now available at pypi (done: 2012-10-04).
* issue 27: Row should support .get() to be more dict-like

FIXED:

* issue 135: Avoid leaking globals between step modules.
* issue 114: No blank lines when option --no-skipped is used (provided by: florentx).
* issue 111: Comment following wip tag results in scenario being ignored
* issue 83: behave.__main__:main() Various sys.exit issues
* issue 80: source file names not properly printed with python 3.3.0
* issue 62: --format=json: Background steps are missing (fixed: some time ago).

RESOLVED:

* issue 98: Summary should include the names of the first X tests that failed (solved by: 116).

-------------------------------------------------------------------------------

NEW:

* "progress" formatter added (from jenisy-repo).
* Add "issue.features/" to simplify verification/validation of issues (from jenisy-repo).

FIXED:

* issue 107: test/ directory gets installed into site-packages
* issue 99: Layout variation "a directory containing your feature files" is broken for running single features
* issue 96: Sub-steps failed without any error info to help debug issue
* issue 85: AssertionError with nested regex and pretty formatter
* issue 84: behave.runner behave does not reliably detected failed test runs
* issue 75: behave list_of_features.txt is broken.
* issue 73: current_matcher is not predictable.
* issue 72: Using GHERKIN_COLORS caused an TypeError.
* issue 70: JUnitReporter: Generates invalid UTF-8 in CDATA sections (stdout/stderr output) when ANSI escapes are used.
* issue 69: JUnitReporter: Fault when processing ScenarioOutlines with failing steps
* issue 67: JSON formatter cannot serialize tables.
* issue 66: context.table and context.text are not cleared.
* issue 65: unrecognized --tag-help argument.
* issue 64: Exit status not set to 1 even there are failures in certain cases (related to: 52)
* issue 63: 'ScenarioOutline' object has no attribute 'stdout'.
* issue 35: "behave --format=plain --tags one" seems to execute right scenario w/ wrong steps
* issue 32: "behave ... --junit-directory=xxx" fails for more than 1 level

RESOLVED:

* issue 81: Allow defining steps in a separate library.
* issue 78: Added references to django-behave (pull-request).
* issue 77: Does not capture stdout from sub-processes

REJECTED:

* issue 109: Insists that implemented tests are not implemented (not reproducable)
* issue 100: Forked package installs but won't run on RHEL.
* issue 88: Python 3 compatibility changes (=> use 2to3 tool instead).

DUPLICATED:

* issue 106: When path is to a feature file only one folder level usable (same as 99).
* issue 105: behave's exit code only depends on the last scenario of the last feature (same as 95).
* issue 95: Failed test run still returns exit code 0 (same as 84, 64).
* issue 94: JUnit format does not handle ScenarioOutlines (same as 69).
* issue 92: Output from --format=plain shows skipped steps in next scenario (same as 35).
* issue 34: "behave --version" runs features, but shows no version (same as 30)

-------------------------------------------------------------------------------

* Fix for an error when an assertion message contains Unicode characters.
* Don't repr() the step text in snippets to avoid turning Unicode text into
backslash hell.