Skip to content

Latest commit

 

History

History
206 lines (171 loc) · 8 KB

README.md

File metadata and controls

206 lines (171 loc) · 8 KB
Raw

I would like to personally express my gratitute to Professor Dimitris Kolovos and to Doctoral student Sina Madani for all their help in this project!

Training & Linting through bash (CLI) 
sh cli.sh -t <training_directory> -l <lint_page> -r <relations_dump_file> -c <depth_cap> -d <html_file_to_save_documentation>
Loading dumped rules & Linting through bash (CLI) 
sh cli.sh -l <lint_page> -r <relations_dump_file>

Example

sh cli.sh -l 'scenarios/scenario_12/coreui-free-bootstrap-admin-template-3-next/src/index.html' -r 'exported_rules_files/rules.json'
Create an HTML documentation page with the rules constructed from the training pages

Click here for a demo

sh cli.sh -r <relations_dump_file> -d <html_file_to_save_documentation>
OR
sh cli.sh -t <training_directory>> -d <html_file_to_save_documentation>

Example

sh cli.sh -r 'exported_rules_files/rules.json' -d 'documentation.html'

If you are not using the shell script then load the Virtual Environment to have all the dependencies installed

source venv/bin/activate
For running a specific test with pytest and not capturing any print statements (i.e. be verbose) 
pytest -s test_all_scenarios.py::TestScenario22::test_shuffled_stuff
How to train the linter 
from main import RuleComposer
rule_composer_class = RuleComposer(threshold=1, 
                                   train_set='path/to/training/directory')
If you want to train the linter and have * (any depth) relation when an x amount of distinct depths is encountered then use the parameter "star_depth_threshold" (default: None) 
from main import RuleComposer
rule_composer_class = RuleComposer(threshold=1, 
                                   train_set='path/to/training/directory',
                                   star_depth_threshold=None)
If you want the linter to be verbose while training then use the parameter "debug" (default: False) 
from main import RuleComposer
rule_composer_class = RuleComposer(threshold=1, 
                                   train_set='path/to/training/directory',
                                   star_depth_threshold=None,
                                   debug=True)
If you already have a json file with rules from a different training then import it with the parameter "json_rules_filename" (default: None) 
from main import RuleComposer
rule_composer_class = RuleComposer(threshold=1, 
                                   train_set='path/to/training/directory',
                                   star_depth_threshold=None,
                                   debug=True,
                                   json_rules_filename=None)
How to lint a page 
from main import RuleComposer
rule_composer_class = RuleComposer(threshold=1, 
                                   train_set='path/to/training/directory',
                                   star_depth_threshold=None,
                                   debug=True,
                                   json_rules_filename=None)
rule_composer_class.compare_test_page(test_page='path/to/test/page')
rule_composer_class.print_parent_level_errors()
If you want to lint a page and only check for collocated class relations then use the parameter "allow_fine_grain_relations" (default: True) 
from main import RuleComposer
rule_composer_class = RuleComposer(threshold=1, 
                                   train_set='path/to/training/directory',
                                   star_depth_threshold=None,
                                   debug=True,
                                   json_rules_filename=None)

rule_composer_class.compare_test_page(test_page='path/to/test/page',
                                      allow_fine_grain_relations=False)
rule_composer_class.print_parent_level_errors()
If you want to considered untrain/unseen CSS classes then use the parameter "ignore_unseen_classes" (default: True) 
from main import RuleComposer
rule_composer_class = RuleComposer(threshold=1, 
                                   train_set='path/to/training/directory',
                                   star_depth_threshold=None,
                                   debug=True,
                                   json_rules_filename=None)

rule_composer_class.compare_test_page(test_page='path/to/test/page',
                                      allow_fine_grain_relations=False,
                                      ignore_unseen_classes=False)
rule_composer_class.print_parent_level_errors()
If you want to report errors regarding unseen classes as warnings, then use the parameter "include_warnings" (default: False) 
from main import RuleComposer
rule_composer_class = RuleComposer(threshold=1, 
                                   train_set='path/to/training/directory',
                                   star_depth_threshold=None,
                                   debug=True,
                                   json_rules_filename=None)

rule_composer_class.compare_test_page(test_page='path/to/test/page',
                                      allow_fine_grain_relations=False,
                                      ignore_unseen_classes=True,
                                      include_warnings=True)
rule_composer_class.print_parent_level_errors()
If you want to restrict the depth of the linter lookup during relations (i.e. how deep to look for violations, let us say 2) then you can use the parameter "depth_cap" (default: None) 
from main import RuleComposer
rule_composer_class = RuleComposer(threshold=1, 
                                   train_set='path/to/training/directory',
                                   star_depth_threshold=None,
                                   debug=True,
                                   json_rules_filename=None)

rule_composer_class.compare_test_page(test_page='path/to/test/page',
                                      allow_fine_grain_relations=False,
                                      ignore_unseen_classes=True,
                                      include_warnings=True,
                                      depth_cap=2)
rule_composer_class.print_parent_level_errors()
If you want to convert a CSS class (that has i.e. 3 relations with other classes) to a wildcard-class (can appear anywhere in the code) then use the parameter "relation_cap" (default: None) 
from main import RuleComposer
rule_composer_class = RuleComposer(threshold=1, 
                                   train_set='path/to/training/directory',
                                   star_depth_threshold=None,
                                   debug=True,
                                   json_rules_filename=None)

rule_composer_class.compare_test_page(test_page='path/to/test/page',
                                      allow_fine_grain_relations=False,
                                      ignore_unseen_classes=True,
                                      include_warnings=True,
                                      depth_cap=2,
                                      relation_cap=3)
rule_composer_class.print_parent_level_errors()