Added cli flag to skip spellchecking

checksit/check.py

@@ -20,7 +20,8 @@
 class Checker:
 
     def __init__(self, template="auto", mappings=None, extra_rules=None, specs=None, ignore_attrs=None, 
-                auto_cache=False, verbose=False, log_mode="standard", ignore_warnings=False):
+                auto_cache=False, verbose=False, log_mode="standard", ignore_warnings=False,
+                skip_spellcheck=False):
         self.template = template
         self.mappings = mappings or {}
         self.extra_rules = extra_rules or {}
@@ -30,6 +31,7 @@ def __init__(self, template="auto", mappings=None, extra_rules=None, specs=None,
         self.ignore_warnings = ignore_warnings
         self.verbose = verbose
         self.log_mode = log_mode
+        self.skip_spellcheck = skip_spellcheck
         self._check_context = {}
 
     def _update_check_context(self, file_path, template):
@@ -124,7 +126,8 @@ def _compare_dicts(self, record, template, label, mappings=None, ignore_attrs=No
         return errors
 
     def _check_file(self, file_content, template, mappings=None, extra_rules=None, specs=None,
-                        ignore_attrs=None, log_mode="standard", fmt_errors=None, ignore_warnings=False):
+                        ignore_attrs=None, log_mode="standard", fmt_errors=None,
+                        ignore_warnings=False, skip_spellcheck=False):
 
         if hasattr(file_content, "to_dict"):
             record = file_content.to_dict()
@@ -144,7 +147,7 @@ def _check_file(self, file_content, template, mappings=None, extra_rules=None, s
 
         for spec in specs:
             sr = SpecificationChecker(spec)
-            spec_errors, spec_warnings = sr.run_checks(record)
+            spec_errors, spec_warnings = sr.run_checks(record, skip_spellcheck=skip_spellcheck)
             errors.extend(spec_errors)
             warnings.extend(spec_warnings)
 
@@ -186,7 +189,8 @@ def _check_file(self, file_content, template, mappings=None, extra_rules=None, s
 
 
     def check_file(self, file_path, template="auto", mappings=None, extra_rules=None, specs=None,
-                ignore_attrs=None, auto_cache=False, verbose=False, log_mode="standard", ignore_warnings=False):
+                ignore_attrs=None, auto_cache=False, verbose=False, log_mode="standard",
+                ignore_warnings=False, skip_spellcheck=False):
 
         try:
             fp = FileParser()
@@ -250,7 +254,8 @@ def check_file(self, file_path, template="auto", mappings=None, extra_rules=None
             print(f"\nRunning with:\n\tTemplate: {tmpl_input}\n\tSpec Files: {specs}\n\tDatafile: {file_content.inpt}")
 
         self._check_file(file_content, template=tmpl, mappings=mappings, extra_rules=extra_rules, 
-                        specs=specs, ignore_attrs=ignore_attrs, log_mode=log_mode, ignore_warnings=ignore_warnings)
+                        specs=specs, ignore_attrs=ignore_attrs, log_mode=log_mode,
+                        ignore_warnings=ignore_warnings, skip_spellcheck=skip_spellcheck)
 
 
 class TemplateManager:

checksit/cli.py

@@ -34,9 +34,11 @@ def main():
 @click.option("-v", "--verbose/--no-verbose", default=False)
 @click.option("-t", "--template", default="auto")
 @click.option("-w", "--ignore-warnings", is_flag=True)
+@click.option("-p", "--skip-spellcheck", is_flag=True)
 def check(file_path, mappings=None, rules=None, specs=None, ignore_attrs=None, ignore_all_globals=False,
           ignore_all_dimensions=False, ignore_all_variables=False, ignore_all_variable_attrs=False,
-          auto_cache=False, log_mode="standard", verbose=False, template="auto", ignore_warnings=False):
+          auto_cache=False, log_mode="standard", verbose=False, template="auto", ignore_warnings=False,
+          skip_spellcheck=False):
 
     if ignore_all_globals or ignore_all_dimensions or ignore_all_variables or ignore_all_variable_attrs:
         raise Exception("Options not implemented yet!!!!!")
@@ -56,7 +58,7 @@ def check(file_path, mappings=None, rules=None, specs=None, ignore_attrs=None, i
     return check_file(file_path, template=template, mappings=mappings, extra_rules=rules, 
                 specs=specs, ignore_attrs=ignore_attrs, 
                 auto_cache=auto_cache, verbose=verbose, 
-                log_mode=log_mode, ignore_warnings=ignore_warnings)
+                log_mode=log_mode, ignore_warnings=ignore_warnings, skip_spellcheck=skip_spellcheck)
 
 
 @main.command()

checksit/generic.py

@@ -40,7 +40,7 @@ def search_close_match(search_for, search_in):
     return ""
 
 
-def check_var_attrs(dct, defined_attrs, ignore_bounds=True):
+def check_var_attrs(dct, defined_attrs, ignore_bounds=True, skip_spellcheck=False):
     """
     Check that variable attributes are defined.
 
@@ -61,7 +61,7 @@ def check_var_attrs(dct, defined_attrs, ignore_bounds=True):
     return errors, warnings
 
 
-def check_global_attrs(dct, defined_attrs=None, vocab_attrs=None, regex_attrs=None, rules_attrs=None):
+def check_global_attrs(dct, defined_attrs=None, vocab_attrs=None, regex_attrs=None, rules_attrs=None, skip_spellcheck=False):
     """
     Check that required global attributes are correct.
 
@@ -78,21 +78,21 @@ def check_global_attrs(dct, defined_attrs=None, vocab_attrs=None, regex_attrs=No
 
     for attr in defined_attrs:
         if attr not in dct['global_attributes']:
-            errors.append(f"[global-attributes:**************:{attr}]: Attribute '{attr}' does not exist. {search_close_match(attr, dct['global_attributes'].keys())}")
+            errors.append(f"[global-attributes:**************:{attr}]: Attribute '{attr}' does not exist. {search_close_match(attr, dct['global_attributes'].keys()) if not skip_spellcheck else ''}")
         elif is_undefined(dct['global_attributes'].get(attr)):
             errors.append(f"[global-attributes:**************:{attr}]: No value defined for attribute '{attr}'.")
 
     for attr in vocab_attrs:
         if attr not in dct['global_attributes']:
-            errors.append(f"[global-attributes:**************:{attr}]: Attribute '{attr}' does not exist. {search_close_match(attr, dct['global_attributes'].keys())}")
+            errors.append(f"[global-attributes:**************:{attr}]: Attribute '{attr}' does not exist. {search_close_match(attr, dct['global_attributes'].keys()) if not skip_spellcheck else ''}")
         elif is_undefined(dct['global_attributes'].get(attr)):
             errors.append(f"[global-attributes:**************:{attr}]: No value defined for attribute '{attr}'.")
         else:
             errors.extend(vocabs.check(vocab_attrs[attr], dct["global_attributes"].get(attr), label=f"[global-attributes:******:{attr}]***"))
 
     for attr in regex_attrs:
         if attr not in dct['global_attributes']:
-            errors.append(f"[global-attributes:**************:{attr}]: Attribute '{attr}' does not exist. {search_close_match(attr, dct['global_attributes'].keys())}")
+            errors.append(f"[global-attributes:**************:{attr}]: Attribute '{attr}' does not exist. {search_close_match(attr, dct['global_attributes'].keys()) if not skip_spellcheck else ''}")
         elif is_undefined(dct['global_attributes'].get(attr)):
             errors.append(f"[global-attributes:**************:{attr}]: No value defined for attribute '{attr}'.")
         elif not re.match(regex_attrs[attr], dct['global_attributes'].get(attr)):
@@ -101,7 +101,7 @@ def check_global_attrs(dct, defined_attrs=None, vocab_attrs=None, regex_attrs=No
     for attr in rules_attrs:
         #errors.extend(rules.check(rules_attrs[attr], dct['global_attributes'].get(attr, UNDEFINED), label=f"[global-attributes:******:{attr}]***"))
         if attr not in dct['global_attributes']:
-            errors.append(f"[global-attributes:**************:{attr}]: Attribute '{attr}' does not exist. {search_close_match(attr, dct['global_attributes'].keys())}")
+            errors.append(f"[global-attributes:**************:{attr}]: Attribute '{attr}' does not exist. {search_close_match(attr, dct['global_attributes'].keys()) if not skip_spellcheck else ''}")
         elif is_undefined(dct['global_attributes'].get(attr)):
             errors.append(f"[global-attributes:**************:{attr}]: No value defined for attribute '{attr}'.")
         else:
@@ -111,7 +111,7 @@ def check_global_attrs(dct, defined_attrs=None, vocab_attrs=None, regex_attrs=No
     return errors, warnings
 
 
-def check_var_exists(dct, variables):
+def check_var_exists(dct, variables, skip_spellcheck=False):
     """
     Check that variables exist
 
@@ -124,15 +124,15 @@ def check_var_exists(dct, variables):
         if ':__OPTIONAL__' in var:
             var = var.split(':')[0]
             if var not in dct["variables"].keys():
-                warnings.append(f"[variable**************:{var}]: Optional variable does not exist in file. {search_close_match(var, dct['variables'].keys())}")
+                warnings.append(f"[variable**************:{var}]: Optional variable does not exist in file. {search_close_match(var, dct['variables'].keys()) if not skip_spellcheck else ''}")
         else:
             if var not in dct["variables"].keys():
-                errors.append(f"[variable**************:{var}]: Does not exist in file. {search_close_match(var, dct['variables'].keys())}")
+                errors.append(f"[variable**************:{var}]: Does not exist in file. {search_close_match(var, dct['variables'].keys()) if not skip_spellcheck else ''}")
 
     return errors, warnings
 
 
-def check_dim_exists(dct, dimensions):
+def check_dim_exists(dct, dimensions, skip_spellcheck=False):
     """
     Check that variables exist
 
@@ -145,15 +145,15 @@ def check_dim_exists(dct, dimensions):
         if ':__OPTIONAL__' in dim:
             dim = dim.split(':')[0]
             if dim not in dct["dimensions"].keys():
-                warnings.append(f"[dimension**************:{dim}]: Optional dimension does not exist in file. {search_close_match(dim, dct['dimensions'].keys())}")
+                warnings.append(f"[dimension**************:{dim}]: Optional dimension does not exist in file. {search_close_match(dim, dct['dimensions'].keys()) if not skip_spellcheck else ''}")
         else:
             if dim not in dct["dimensions"].keys():
-                errors.append(f"[dimension**************:{dim}]: Does not exist in file. {search_close_match(dim, dct['dimensions'].keys())}")
+                errors.append(f"[dimension**************:{dim}]: Does not exist in file. {search_close_match(dim, dct['dimensions'].keys()) if not skip_spellcheck else ''}")
 
     return errors, warnings 
 
 
-def check_var(dct, variable, defined_attrs):
+def check_var(dct, variable, defined_attrs, skip_spellcheck=False):
     """
     Check variable exists and has attributes defined.
     """
@@ -165,7 +165,7 @@ def check_var(dct, variable, defined_attrs):
     if ':__OPTIONAL__' in variable:
         variable = variable.split(':')[0]
         if variable not in dct["variables"].keys():
-            warnings.append(f"[variable**************:{variable}]: Optional variable does not exist in file. {search_close_match(variable, dct['variables'].keys())}")
+            warnings.append(f"[variable**************:{variable}]: Optional variable does not exist in file. {search_close_match(variable, dct['variables'].keys()) if not skip_spellcheck else ''}")
         else:
             for attr in defined_attrs:
                 if isinstance(attr, dict) and len(attr.keys()) == 1:
@@ -174,7 +174,7 @@ def check_var(dct, variable, defined_attrs):
                 attr_key = attr.split(':')[0]
                 attr_value = ':'.join(attr.split(':')[1:])
                 if attr_key not in dct["variables"][variable]:
-                    errors.append(f"[variable**************:{variable}]: Attribute '{attr_key}' does not exist. {search_close_match(attr_key, dct['variables'][variable])}")
+                    errors.append(f"[variable**************:{variable}]: Attribute '{attr_key}' does not exist. {search_close_match(attr_key, dct['variables'][variable]) if not skip_spellcheck else ''}")
                 elif '<derived from file>' in attr_value:
                     # work this out
                     pass
@@ -183,21 +183,21 @@ def check_var(dct, variable, defined_attrs):
                     attr_value = [ int(i.strip('b')) for i in attr_value.split(',') ]
                     attr_value = np.array(attr_value, dtype=np.int8)
                     if not np.all(dct["variables"][variable].get(attr_key) == attr_value):
-                        errors.append(f"[variable**************:{variable}]: Attribute '{attr_key}' must have definition {attr_value}, not {dct['variables'][variable].get(attr_key)}.")
+                        errors.append(f"[variable**************:{variable}]: Attribute '{attr_key}' must have definition {attr_value}, not {dct['variables'][variable].get(attr_key) if skip_spellcheck else ''}.")
                 #elif attr_key == 'flag_meanings':
                 #    print(attr_value)
                 #    print(dct["variables"][variable].get(attr_key))
                 elif not str(dct["variables"][variable].get(attr_key)) == attr_value:
                     errors.append(f"[variable**************:{variable}]: Attribute '{attr_key}' must have definition {attr_value}, not {dct['variables'][variable].get(attr_key).encode('unicode_escape').decode('utf-8')}.")
     else:
         if variable not in dct["variables"].keys():
-            errors.append(f"[variable**************:{variable}]: Optional variable does not exist in file. {search_close_match(variable, dct['variables'].keys())}")
+            errors.append(f"[variable**************:{variable}]: Optional variable does not exist in file. {search_close_match(variable, dct['variables'].keys()) if not skip_spellcheck else ''}")
         else:
             for attr in defined_attrs:
                 attr_key = attr.split(':')[0]
                 attr_value = ':'.join(attr.split(':')[1:]) 
                 if attr_key not in dct["variables"][variable]:
-                    errors.append(f"[variable**************:{variable}]: Attribute '{attr_key}' does not exist. {search_close_match(attr_key, dct['variables'][variable])}")
+                    errors.append(f"[variable**************:{variable}]: Attribute '{attr_key}' does not exist. {search_close_match(attr_key, dct['variables'][variable]) if not skip_spellcheck else ''}")
                 elif '<' in attr_value:
                     # work this out
                     pass

checksit/specs.py

@@ -48,22 +48,25 @@ def _setup(self, spec_id):
         self.spec_id = spec_id
         self.spec = load_specs([spec_id])[spec_id]
 
-    def _run_check(self, record, check_dict):
+    def _run_check(self, record, check_dict, skip_spellcheck=False):
         d = check_dict
         parts = d["func"].split(".")
 
         mod_path, func = ".".join(parts[:-1]), parts[-1]
         func = getattr(importlib.import_module(mod_path), func)
 
         params = d["params"]
+        params["skip_spellcheck"] = skip_spellcheck
         return func(record, **params)
 
-    def run_checks(self, record):
+    def run_checks(self, record, skip_spellcheck=False):
         errors = []
         warnings = []
 
         for check_id, check_dict in self.spec.items():
-            check_errors, check_warnings = self._run_check(record, check_dict)
+            check_errors, check_warnings = self._run_check(
+                                               record, check_dict, skip_spellcheck=skip_spellcheck
+                                           )
             errors.extend(check_errors)
             warnings.extend(check_warnings)