Merge pull request #771 from IanCa/develop

Add a string based search first pass

hed/errors/error_types.py

@@ -107,7 +107,7 @@ class SidecarErrors:
 
 class SchemaErrors:
     SCHEMA_DUPLICATE_NODE = 'SCHEMA_DUPLICATE_NODE'
-    SCHEMA_ATTRIBUTE_INVALID = 'SCHEMA_ATTRIBUTE_INVALID'
+
     SCHEMA_DUPLICATE_FROM_LIBRARY = "SCHEMA_LIBRARY_INVALID"
 
 
@@ -119,19 +119,22 @@ class SchemaWarnings:
     SCHEMA_CHARACTER_INVALID = "SCHEMA_CHARACTER_INVALID"
     SCHEMA_INVALID_CAPITALIZATION = 'invalidCaps'
     SCHEMA_NON_PLACEHOLDER_HAS_CLASS = 'SCHEMA_NON_PLACEHOLDER_HAS_CLASS'
-    SCHEMA_INVALID_ATTRIBUTE = "SCHEMA_INVALID_ATTRIBUTE"
 
 
 class SchemaAttributeErrors:
+    SCHEMA_ATTRIBUTE_INVALID = 'SCHEMA_ATTRIBUTE_INVALID'
+    SCHEMA_ATTRIBUTE_VALUE_INVALID = 'SCHEMA_ATTRIBUTE_VALUE_INVALID'
     SCHEMA_DEPRECATED_INVALID = "SCHEMA_DEPRECATED_INVALID"
     SCHEMA_SUGGESTED_TAG_INVALID = "SCHEMA_SUGGESTED_TAG_INVALID"
-    SCHEMA_RELATED_TAG_INVALID = "SCHEMA_RELATED_TAG_INVALID"
 
     SCHEMA_UNIT_CLASS_INVALID = "SCHEMA_UNIT_CLASS_INVALID"
     SCHEMA_VALUE_CLASS_INVALID = "SCHEMA_VALUE_CLASS_INVALID"
+    SCHEMA_ALLOWED_CHARACTERS_INVALID = "SCHEMA_ALLOWED_CHARACTERS_INVALID"
+    SCHEMA_IN_LIBRARY_INVALID = "SCHEMA_IN_LIBRARY_INVALID"
 
     SCHEMA_DEFAULT_UNITS_INVALID = "SCHEMA_DEFAULT_UNITS_INVALID"
-    SCHEMA_CHILD_OF_DEPRECATED = "SCHEMA_CHILD_OF_DEPRECATED" # Reported as SCHEMA_DEPRECATED_INVALID
+    SCHEMA_CHILD_OF_DEPRECATED = "SCHEMA_CHILD_OF_DEPRECATED"
+    SCHEMA_CONVERSION_FACTOR_NOT_POSITIVE = "SCHEMA_CONVERSION_FACTOR_NOT_POSITIVE"
 
 
 class DefinitionErrors:

hed/errors/exceptions.py

@@ -14,8 +14,9 @@ class HedExceptions:
     INVALID_DATAFRAME = 'INVALID_DATAFRAME'
     INVALID_FILE_FORMAT = 'INVALID_FILE_FORMAT'
     # These are actual schema issues, not that the file cannot be found or parsed
-    SCHEMA_HEADER_MISSING = 'HED_SCHEMA_HEADER_INVALID'
-    HED_SCHEMA_HEADER_INVALID = 'HED_SCHEMA_HEADER_INVALID'
+    SCHEMA_HEADER_MISSING = 'SCHEMA_HEADER_INVALID'
+    SCHEMA_HEADER_INVALID = 'SCHEMA_HEADER_INVALID'
+    SCHEMA_UNKNOWN_HEADER_ATTRIBUTE = "SCHEMA_HEADER_INVALID"
 
     SCHEMA_LIBRARY_INVALID = "SCHEMA_LIBRARY_INVALID"
     BAD_HED_LIBRARY_NAME = 'SCHEMA_LIBRARY_INVALID'
@@ -26,14 +27,14 @@ class HedExceptions:
     ROOTED_TAG_DOES_NOT_EXIST = "SCHEMA_LIBRARY_INVALID"
     IN_LIBRARY_IN_UNMERGED = "SCHEMA_LIBRARY_INVALID"
 
-    HED_SCHEMA_VERSION_INVALID = 'HED_SCHEMA_VERSION_INVALID'
-    SCHEMA_START_MISSING = 'HED_WIKI_SEPARATOR_INVALID'
-    SCHEMA_END_INVALID = 'HED_WIKI_SEPARATOR_INVALID'
-    HED_END_INVALID = 'HED_WIKI_SEPARATOR_INVALID'
-    INVALID_SECTION_SEPARATOR = 'invalidSectionSeparator'
+    SCHEMA_VERSION_INVALID = 'SCHEMA_VERSION_INVALID'
+    SCHEMA_SECTION_MISSING = 'SCHEMA_SECTION_MISSING'
+
+    WIKI_SEPARATOR_INVALID = 'invalidSectionSeparator'
 
     # This issue will contain a list of lines with issues.
-    HED_WIKI_DELIMITERS_INVALID = 'HED_WIKI_DELIMITERS_INVALID'
+    WIKI_DELIMITERS_INVALID = 'WIKI_DELIMITERS_INVALID'
+    WIKI_LINE_START_INVALID = 'WIKI_LINE_START_INVALID'
     HED_SCHEMA_NODE_NAME_INVALID = 'HED_SCHEMA_NODE_NAME_INVALID'
 
     SCHEMA_DUPLICATE_PREFIX = 'schemaDuplicatePrefix'

hed/errors/known_error_codes.py

@@ -31,6 +31,7 @@
     ],
     "schema_validation_errors": [
         "SCHEMA_ATTRIBUTE_INVALID",
+        "SCHEMA_ATTRIBUTE_VALUE_INVALID",
         "SCHEMA_CHARACTER_INVALID",
         "SCHEMA_DUPLICATE_NODE",
         "SCHEMA_HEADER_INVALID",

hed/errors/schema_error_messages.py

@@ -16,7 +16,7 @@ def schema_error_hed_duplicate_from_library(tag, duplicate_tag_list, section):
            f"{tag_join_delimiter}{tag_join_delimiter.join(duplicate_tag_list)}"
 
 
-@hed_error(SchemaErrors.SCHEMA_ATTRIBUTE_INVALID)
+@hed_error(SchemaAttributeErrors.SCHEMA_ATTRIBUTE_INVALID)
 def schema_error_unknown_attribute(attribute_name, source_tag):
     return f"Attribute '{attribute_name}' used by '{source_tag}' was not defined in the schema, " \
            f"or was used outside of it's defined class."
@@ -40,45 +40,58 @@ def schema_warning_SCHEMA_INVALID_CAPITALIZATION(tag_name, problem_char, char_in
            f"Found character '{problem_char}' in tag '{tag_name}' at position {char_index}."
 
 
-@hed_error(SchemaWarnings.SCHEMA_NON_PLACEHOLDER_HAS_CLASS, default_severity=ErrorSeverity.WARNING)
+@hed_error(SchemaWarnings.SCHEMA_NON_PLACEHOLDER_HAS_CLASS, default_severity=ErrorSeverity.WARNING,
+           actual_code=SchemaAttributeErrors.SCHEMA_ATTRIBUTE_VALUE_INVALID)
 def schema_warning_non_placeholder_class(tag_name, invalid_attribute_name):
     return "Only placeholder nodes('#') can have a unit class, value class, or takes value." + \
            f"Found {invalid_attribute_name} on {tag_name}"
 
 
-@hed_error(SchemaWarnings.SCHEMA_INVALID_ATTRIBUTE, default_severity=ErrorSeverity.ERROR)
-def schema_error_SCHEMA_INVALID_ATTRIBUTE(tag_name, invalid_attribute_name):
-    return f"'{invalid_attribute_name}' should not be present in a loaded schema, found on '{tag_name}'." \
-           f"Something went very wrong."
 
-
-@hed_error(SchemaAttributeErrors.SCHEMA_DEPRECATED_INVALID)
+@hed_error(SchemaAttributeErrors.SCHEMA_DEPRECATED_INVALID, actual_code=SchemaAttributeErrors.SCHEMA_ATTRIBUTE_VALUE_INVALID)
 def schema_error_SCHEMA_DEPRECATED_INVALID(tag_name, invalid_deprecated_version):
     return f"'{tag_name}' has invalid or unknown value in attribute deprecatedFrom: '{invalid_deprecated_version}'."
 
 
 @hed_error(SchemaAttributeErrors.SCHEMA_CHILD_OF_DEPRECATED,
-           actual_code=SchemaAttributeErrors.SCHEMA_DEPRECATED_INVALID)
+           actual_code=SchemaAttributeErrors.SCHEMA_ATTRIBUTE_VALUE_INVALID)
 def schema_error_SCHEMA_CHILD_OF_DEPRECATED(deprecated_tag, non_deprecated_child):
     return f"Deprecated tag '{deprecated_tag}' has a child that is not deprecated: '{non_deprecated_child}'."
 
 
-@hed_error(SchemaAttributeErrors.SCHEMA_SUGGESTED_TAG_INVALID)
+@hed_error(SchemaAttributeErrors.SCHEMA_SUGGESTED_TAG_INVALID, actual_code=SchemaAttributeErrors.SCHEMA_ATTRIBUTE_VALUE_INVALID)
 def schema_error_SCHEMA_SUGGESTED_TAG_INVALID(suggestedTag, invalidSuggestedTag, attribute_name):
     return f"Tag '{suggestedTag}' has an invalid {attribute_name}: '{invalidSuggestedTag}'."
 
 
-@hed_error(SchemaAttributeErrors.SCHEMA_UNIT_CLASS_INVALID)
+@hed_error(SchemaAttributeErrors.SCHEMA_UNIT_CLASS_INVALID, actual_code=SchemaAttributeErrors.SCHEMA_ATTRIBUTE_VALUE_INVALID)
 def schema_error_SCHEMA_UNIT_CLASS_INVALID(tag, unit_class, attribute_name):
     return f"Tag '{tag}' has an invalid {attribute_name}: '{unit_class}'."
 
 
-@hed_error(SchemaAttributeErrors.SCHEMA_VALUE_CLASS_INVALID)
+@hed_error(SchemaAttributeErrors.SCHEMA_VALUE_CLASS_INVALID, actual_code=SchemaAttributeErrors.SCHEMA_ATTRIBUTE_VALUE_INVALID)
 def schema_error_SCHEMA_VALUE_CLASS_INVALID(tag, unit_class, attribute_name):
     return f"Tag '{tag}' has an invalid {attribute_name}: '{unit_class}'."
 
 
-@hed_error(SchemaAttributeErrors.SCHEMA_DEFAULT_UNITS_INVALID)
+@hed_error(SchemaAttributeErrors.SCHEMA_DEFAULT_UNITS_INVALID, actual_code=SchemaAttributeErrors.SCHEMA_ATTRIBUTE_VALUE_INVALID)
 def schema_error_SCHEMA_DEFAULT_UNITS_INVALID(tag, bad_unit, valid_units):
     valid_units = ",".join(valid_units)
     return f"Tag '{tag}' has an invalid defaultUnit '{bad_unit}'.  Valid units are: '{valid_units}'."
+
+
+@hed_error(SchemaAttributeErrors.SCHEMA_CONVERSION_FACTOR_NOT_POSITIVE, actual_code=SchemaAttributeErrors.SCHEMA_ATTRIBUTE_VALUE_INVALID)
+def schema_error_SCHEMA_CONVERSION_FACTOR_NOT_POSITIVE(tag, conversion_factor):
+    return f"Tag '{tag}' has an invalid conversionFactor '{conversion_factor}'.  Conversion factor must be positive."
+
+
+@hed_error(SchemaAttributeErrors.SCHEMA_ALLOWED_CHARACTERS_INVALID, actual_code=SchemaAttributeErrors.SCHEMA_ATTRIBUTE_VALUE_INVALID)
+def schema_error_SCHEMA_ALLOWED_CHARACTERS_INVALID(tag, invalid_character):
+    return (f"Tag '{tag}' has an invalid allowedCharacter: '{invalid_character}'.  "
+            f"Allowed characters are: a single character, "
+            f"or one of the following - letters, blank, digits, alphanumeric.")
+
+
+@hed_error(SchemaAttributeErrors.SCHEMA_IN_LIBRARY_INVALID, actual_code=SchemaAttributeErrors.SCHEMA_ATTRIBUTE_VALUE_INVALID)
+def schema_error_SCHEMA_IN_LIBRARY_INVALID(tag, bad_library):
+    return (f"Tag '{tag}' has an invalid inLibrary: '{bad_library}'.  ")

hed/models/base_input.py

@@ -137,7 +137,7 @@ def _indexed_dict_from_onsets(onsets):
 
     @staticmethod
     def _filter_by_index_list(original_series, indexed_dict):
-        new_series = ["n/a"] * len(original_series)  # Initialize new_series with "n/a"
+        new_series = pd.Series(["n/a"] * len(original_series))
 
         for onset, indices in indexed_dict.items():
             if indices: