cognite-neat 0.88.1__py3-none-any.whl → 0.88.3__py3-none-any.whl

cognite/neat/rules/importers/_rdf/_shared.py

@@ -0,0 +1,586 @@
+import datetime
+
+import numpy as np
+import pandas as pd
+from rdflib import OWL, Literal, Namespace
+
+from cognite.neat.rules.models._base import MatchType
+from cognite.neat.rules.models.data_types import _XSD_TYPES
+from cognite.neat.utils.rdf_ import remove_namespace_from_uri
+from cognite.neat.utils.regex_patterns import PATTERNS
+
+
+def parse_raw_classes_dataframe(query_results: list[tuple]) -> pd.DataFrame:
+    df = pd.DataFrame(
+        query_results,
+        columns=[
+            "Class",
+            "Name",
+            "Description",
+            "Parent Class",
+            "Reference",
+            "Match",
+            "Comment",
+        ],
+    )
+
+    if df.empty:
+        return df
+
+    # # remove NaNs
+    df.replace(np.nan, "", regex=True, inplace=True)
+
+    df.Class = df.Class.apply(lambda x: remove_namespace_from_uri(x))
+    df["Match Type"] = len(df) * [MatchType.exact]
+    df["Comment"] = len(df) * [None]
+    df["Parent Class"] = df["Parent Class"].apply(lambda x: remove_namespace_from_uri(x))
+
+    return df
+
+
+def clean_up_classes(df: pd.DataFrame) -> pd.DataFrame:
+    clean_list = [
+        {
+            "Class": class_,
+            "Name": group_df["Name"].unique()[0],
+            "Description": "\n".join(list(group_df.Description.unique())),
+            "Parent Class": ", ".join(list(group_df["Parent Class"].unique())),
+            "Reference": group_df["Reference"].unique()[0],
+            "Match Type": group_df["Match Type"].unique()[0],
+            "Comment": group_df["Comment"].unique()[0],
+        }
+        for class_, group_df in df.groupby("Class")
+    ]
+
+    df = pd.DataFrame(clean_list)
+
+    # bring NaNs back
+    df.replace("", None, inplace=True)
+
+    # split Parent Class column back into list
+    df["Parent Class"] = df["Parent Class"].apply(lambda x: x.split(", ") if isinstance(x, str) else None)
+
+    return df
+
+
+def make_classes_compliant(classes: pd.DataFrame, importer: str = "RDF-based") -> pd.DataFrame:
+    """Make classes compliant.
+
+    Returns:
+        Dataframe containing compliant classes
+
+    !!! note "About the compliant classes"
+        The compliant classes are based on the OWL base ontology, but adapted to NEAT and use in CDF.
+        One thing to note is that this method would not be able to fix issues with class ids which
+        are not compliant with the CDF naming convention. For example, if a class id contains a space,
+        starts with a number, etc. This will cause issues when trying to create the class in CDF.
+    """
+
+    # Replace empty or non-string values in "Match" column with "exact"
+    classes["Match Type"] = classes["Match Type"].fillna(MatchType.exact)
+    classes["Match Type"] = classes["Match Type"].apply(
+        lambda x: MatchType.exact if not isinstance(x, str) or len(x) == 0 else x
+    )
+
+    # Replace empty or non-string values in "Comment" column with a default value
+    classes["Comment"] = classes["Comment"].fillna(f"Imported using {importer} importer")
+    classes["Comment"] = classes["Comment"].apply(
+        lambda x: (f"Imported using {importer} importer" if not isinstance(x, str) or len(x) == 0 else x)
+    )
+
+    # Add _object_property_class, _data_type_property_class, _thing_class to the dataframe
+    classes = pd.concat(
+        [
+            classes,
+            pd.DataFrame([object_property_class(), data_type_property_class(), thing_class()]),
+        ],
+        ignore_index=True,
+    )
+
+    # Reduce length of elements in the "Description" column to 1024 characters
+    classes["Description"] = classes["Description"].apply(lambda x: x[:1024] if isinstance(x, str) else None)
+
+    # Add missing parent classes to the dataframe
+    classes = pd.concat(
+        [classes, pd.DataFrame(add_parent_class(classes))],
+        ignore_index=True,
+    )
+
+    return classes
+
+
+def object_property_class() -> dict:
+    return {
+        "Class": "ObjectProperty",
+        "Name": None,
+        "Description": "The class of object properties.",
+        "Parent Class": None,
+        "Reference": OWL.ObjectProperty,
+        "Match Type": MatchType.exact,
+        "Comment": "Added by NEAT based on owl:ObjectProperty but adapted to NEAT and use in CDF.",
+    }
+
+
+def data_type_property_class() -> dict:
+    return {
+        "Class": "DatatypeProperty",
+        "Name": None,
+        "Description": "The class of data properties.",
+        "Parent Class": None,
+        "Reference": OWL.DatatypeProperty,
+        "Match Type": MatchType.exact,
+        "Comment": "Added by NEAT based on owl:DatatypeProperty but adapted to NEAT and use in CDF.",
+    }
+
+
+def thing_class() -> dict:
+    return {
+        "Class": "Thing",
+        "Name": None,
+        "Description": "The class of holding class individuals.",
+        "Parent Class": None,
+        "Reference": OWL.Thing,
+        "Match Type": MatchType.exact,
+        "Comment": (
+            "Added by NEAT. "
+            "Imported from OWL base ontology, it is meant for use as a default"
+            " value type for object properties which miss a declared range."
+        ),
+    }
+
+
+def add_parent_class(df: pd.DataFrame) -> list[dict]:
+    parent_set = {
+        item
+        for sublist in df["Parent Class"].tolist()
+        if sublist
+        for item in sublist
+        if item != "" and item is not None
+    }
+    class_set = set(df["Class"].tolist())
+
+    rows = []
+    for missing_parent_class in parent_set.difference(class_set):
+        rows += [
+            {
+                "Class": missing_parent_class,
+                "Name": None,
+                "Description": None,
+                "Parent Class": None,
+                "Reference": None,
+                "Match Type": None,
+                "Comment": (
+                    "Added by NEAT. "
+                    "This is a parent class that is missing in the ontology. "
+                    "It is added by NEAT to make the ontology compliant with CDF."
+                ),
+            }
+        ]
+
+    return rows
+
+
+def parse_raw_properties_dataframe(query_results: list[tuple]) -> pd.DataFrame:
+    df = pd.DataFrame(
+        query_results,
+        columns=[
+            "Class",
+            "Property",
+            "Name",
+            "Description",
+            "Value Type",
+            "Min Count",
+            "Max Count",
+            "Default",
+            "Reference",
+            "Match Type",
+            "Comment",
+            "_property_type",
+        ],
+    )
+    if df.empty:
+        return df
+
+    df.replace(np.nan, "", regex=True, inplace=True)
+
+    df.Class = df.Class.apply(lambda x: remove_namespace_from_uri(x))
+    df.Property = df.Property.apply(lambda x: remove_namespace_from_uri(x))
+    df["Value Type"] = df["Value Type"].apply(lambda x: remove_namespace_from_uri(x))
+    df["Match Type"] = len(df) * [MatchType.exact]
+    df["Comment"] = len(df) * [None]
+    df["_property_type"] = df["_property_type"].apply(lambda x: remove_namespace_from_uri(x))
+
+    return df
+
+
+def clean_up_properties(df: pd.DataFrame) -> pd.DataFrame:
+    class_grouped_dfs = df.groupby("Class")
+
+    clean_list = []
+
+    for class_, class_grouped_df in class_grouped_dfs:
+        property_grouped_dfs = class_grouped_df.groupby("Property")
+        for property_, property_grouped_df in property_grouped_dfs:
+            clean_list += [
+                {
+                    "Class": class_,
+                    "Property": property_,
+                    "Name": property_grouped_df["Name"].unique()[0],
+                    "Description": "\n".join(list(property_grouped_df["Description"].unique()))[:1024],
+                    "Value Type": property_grouped_df["Value Type"].unique()[0],
+                    "Min Count": property_grouped_df["Min Count"].unique()[0],
+                    "Max Count": property_grouped_df["Max Count"].unique()[0],
+                    "Default": property_grouped_df["Default"].unique()[0],
+                    "Reference": property_grouped_df["Reference"].unique()[0],
+                    "Match Type": property_grouped_df["Match Type"].unique()[0],
+                    "Comment": property_grouped_df["Comment"].unique()[0],
+                    "_property_type": property_grouped_df["_property_type"].unique()[0],
+                }
+            ]
+
+    df = pd.DataFrame(clean_list)
+    df.replace("", None, inplace=True)
+
+    return df
+
+
+def make_properties_compliant(properties: pd.DataFrame, importer: str = "RDF-based") -> pd.DataFrame:
+    # default to 0 if "Min Count" is not specified
+    properties["Min Count"] = properties["Min Count"].apply(lambda x: 0 if not isinstance(x, Literal) or x == "" else x)
+
+    # default to 1 if "Max Count" is not specified
+    properties["Max Count"] = properties["Max Count"].apply(lambda x: 1 if not isinstance(x, Literal) or x == "" else x)
+
+    # Replace empty or non-string values in "Match Type" column with "exact"
+    properties["Match Type"] = properties["Match Type"].fillna("exact")
+    properties["Match Type"] = properties["Match Type"].apply(
+        lambda x: "exact" if not isinstance(x, str) or len(x) == 0 else x
+    )
+
+    # Replace empty or non-string values in "Comment" column with a default value
+    properties["Comment"] = properties["Comment"].fillna(f"Imported using {importer} importer")
+    properties["Comment"] = properties["Comment"].apply(
+        lambda x: (f"Imported using {importer} importer" if not isinstance(x, str) or len(x) == 0 else x)
+    )
+
+    # Reduce length of elements in the "Description" column to 1024 characters
+    properties["Description"] = properties["Description"].apply(lambda x: x[:1024] if isinstance(x, str) else None)
+
+    # fixes and additions
+    properties = fix_dangling_properties(properties)
+    properties = fix_missing_property_value_type(properties)
+
+    return properties
+
+
+def fix_dangling_properties(properties: pd.DataFrame) -> pd.DataFrame:
+    """This method fixes properties which are missing a domain definition in the ontology.
+
+    Args:
+        properties: Dataframe containing properties
+
+    Returns:
+        Dataframe containing properties with fixed domain
+    """
+    domain = {
+        "ObjectProperty": object_property_class()["Class"],
+        "DatatypeProperty": data_type_property_class()["Class"],
+    }
+
+    # apply missing range
+    properties["Class"] = properties.apply(
+        lambda row: (
+            domain[row._property_type]
+            if row._property_type == "ObjectProperty" and pd.isna(row["Class"])
+            else domain["DatatypeProperty"]
+            if pd.isna(row["Class"])
+            else row["Class"]
+        ),
+        axis=1,
+    )
+    return properties
+
+
+def fix_missing_property_value_type(properties: pd.DataFrame) -> pd.DataFrame:
+    """This method fixes properties which are missing a range definition in the ontology.
+
+    Args:
+        properties: Dataframe containing properties
+
+    Returns:
+        Dataframe containing properties with fixed range
+    """
+    # apply missing range
+    properties["Value Type"] = properties.apply(
+        lambda row: (
+            thing_class()["Class"]
+            if row._property_type == "ObjectProperty" and pd.isna(row["Value Type"])
+            else "string"
+            if pd.isna(row["Value Type"])
+            else row["Value Type"]
+        ),
+        axis=1,
+    )
+
+    return properties
+
+
+def make_metadata_compliant(metadata: dict) -> dict:
+    """Attempts to fix errors in metadata, otherwise defaults to values that will pass validation.
+
+    Args:
+        metadata: Dictionary containing metadata
+
+    Returns:
+        Dictionary containing metadata with fixed errors
+    """
+
+    metadata = fix_namespace(metadata, default=Namespace("http://purl.org/cognite/neat#"))
+    metadata = fix_prefix(metadata)
+    metadata = fix_version(metadata)
+    metadata = fix_date(
+        metadata,
+        date_type="created",
+        default=datetime.datetime.now().replace(microsecond=0),
+    )
+    metadata = fix_date(
+        metadata,
+        date_type="updated",
+        default=datetime.datetime.now().replace(microsecond=0),
+    )
+    metadata = fix_title(metadata)
+    metadata = fix_description(metadata)
+    metadata = fix_author(metadata, "creator")
+    metadata = fix_rights(metadata)
+    metadata = fix_license(metadata)
+
+    return metadata
+
+
+def fix_license(metadata: dict, default: str = "Unknown license") -> dict:
+    if license := metadata.get("license", None):
+        if not isinstance(license, str):
+            metadata["license"] = default
+        elif isinstance(license, str) and len(license) == 0:
+            metadata["license"] = default
+    else:
+        metadata["license"] = default
+    return metadata
+
+
+def fix_rights(metadata: dict, default: str = "Unknown rights") -> dict:
+    if rights := metadata.get("rights", None):
+        if not isinstance(rights, str):
+            metadata["rights"] = default
+        elif isinstance(rights, str) and len(rights) == 0:
+            metadata["rights"] = default
+    else:
+        metadata["rights"] = default
+    return metadata
+
+
+def fix_author(metadata: dict, author_type: str = "creator", default: str = "NEAT") -> dict:
+    if author := metadata.get(author_type, None):
+        if not isinstance(author, str) or isinstance(author, list):
+            metadata[author_type] = default
+        elif isinstance(author, str) and len(author) == 0:
+            metadata[author_type] = default
+    else:
+        metadata[author_type] = default
+    return metadata
+
+
+def fix_description(metadata: dict, default: str = "This model has been inferred from OWL ontology") -> dict:
+    if description := metadata.get("description", None):
+        if not isinstance(description, str) or len(description) == 0:
+            metadata["description"] = default
+        elif isinstance(description, str) and len(description) > 1024:
+            metadata["description"] = metadata["description"][:1024]
+    else:
+        metadata["description"] = default
+    return metadata
+
+
+def fix_prefix(metadata: dict, default: str = "neat") -> dict:
+    if prefix := metadata.get("prefix", None):
+        if not isinstance(prefix, str) or not PATTERNS.prefix_compliance.match(prefix):
+            metadata["prefix"] = default
+    else:
+        metadata["prefix"] = default
+    return metadata
+
+
+def fix_namespace(metadata: dict, default: Namespace) -> dict:
+    if namespace := metadata.get("namespace", None):
+        if not isinstance(namespace, Namespace):
+            try:
+                metadata["namespace"] = Namespace(namespace)
+            except Exception:
+                metadata["namespace"] = default
+    else:
+        metadata["namespace"] = default
+
+    return metadata
+
+
+def fix_date(
+    metadata: dict,
+    date_type: str,
+    default: datetime.datetime,
+) -> dict:
+    if date := metadata.get(date_type, None):
+        try:
+            if isinstance(date, datetime.datetime):
+                return metadata
+            elif isinstance(date, datetime.date):
+                metadata[date_type] = datetime.datetime.combine(metadata[date_type], datetime.datetime.min.time())
+            elif isinstance(date, str):
+                metadata[date_type] = datetime.datetime.strptime(metadata[date_type], "%Y-%m-%dT%H:%M:%SZ")
+            else:
+                metadata[date_type] = default
+        except Exception:
+            metadata[date_type] = default
+    else:
+        metadata[date_type] = default
+
+    return metadata
+
+
+def fix_version(metadata: dict, default: str = "1.0.0") -> dict:
+    if version := metadata.get("version", None):
+        if not PATTERNS.version_compliance.match(version):
+            metadata["version"] = default
+    else:
+        metadata["version"] = default
+
+    return metadata
+
+
+def fix_title(metadata: dict, default: str = "OWL Inferred Data Model") -> dict:
+    if title := metadata.get("title", None):
+        if not isinstance(title, str):
+            metadata["title"] = default
+        elif isinstance(title, str) and len(title) == 0:
+            metadata["title"] = default
+        elif isinstance(title, str) and len(title) > 255:
+            metadata["title"] = metadata["title"][:255]
+        else:
+            pass
+    else:
+        metadata["title"] = default
+
+    return metadata
+
+
+def make_components_compliant(components: dict) -> dict:
+    components = add_missing_classes(components)
+    components = add_missing_value_types(components)
+    components = add_default_property_to_dangling_classes(components)
+
+    return components
+
+
+def add_missing_classes(components: dict[str, list[dict]]) -> dict:
+    """Add missing classes to Classes.
+
+    Args:
+        tables: imported tables from owl ontology
+
+    Returns:
+        Updated tables with missing classes added to containers
+    """
+
+    missing_classes = {definition["Class"] for definition in components["Properties"]} - {
+        definition["Class"] for definition in components["Classes"]
+    }
+
+    comment = (
+        "Added by NEAT. "
+        "This is a class that a domain of a property but was not defined in the ontology. "
+        "It is added by NEAT to make the ontology compliant with CDF."
+    )
+
+    for class_ in missing_classes:
+        components["Classes"].append({"Class": class_, "Comment": comment})
+
+    return components
+
+
+def add_missing_value_types(components: dict) -> dict:
+    """Add properties to classes that do not have any properties defined to them
+
+    Args:
+        tables: imported tables from owl ontology
+
+    Returns:
+        Updated tables with missing properties added to containers
+    """
+
+    xsd_types = _XSD_TYPES
+    candidate_value_types = {definition["Value Type"] for definition in components["Properties"]} - {
+        definition["Class"] for definition in components["Classes"]
+    }
+
+    # to avoid issue of case sensitivity for xsd types
+    value_types_lower = {v.lower() for v in candidate_value_types}
+
+    xsd_types_lower = {x.lower() for x in xsd_types}
+
+    # Create a mapping from lowercase strings to original strings
+    value_types_mapping = {v.lower(): v for v in candidate_value_types}
+
+    # Find the difference
+    difference = value_types_lower - xsd_types_lower
+
+    # Convert the difference back to the original case
+    difference_original_case = {value_types_mapping[d] for d in difference}
+
+    for class_ in difference_original_case:
+        components["Classes"].append(
+            {
+                "Class": class_,
+                "Comment": (
+                    "Added by NEAT. "
+                    "This is a class that a domain of a property but was not defined in the ontology. "
+                    "It is added by NEAT to make the ontology compliant with CDF."
+                ),
+            }
+        )
+
+    return components
+
+
+def add_default_property_to_dangling_classes(components: dict[str, list[dict]]) -> dict:
+    """Add missing classes to Classes.
+
+    Args:
+        tables: imported tables from owl ontology
+
+    Returns:
+        Updated tables with missing classes added to containers
+    """
+
+    dangling_classes = {
+        definition["Class"] for definition in components["Classes"] if not definition.get("Parent Class", None)
+    } - {definition["Class"] for definition in components["Properties"]}
+
+    comment = (
+        "Added by NEAT. "
+        "This is property has been added to this class since otherwise it will create "
+        "dangling classes in the ontology."
+    )
+
+    for class_ in dangling_classes:
+        components["Properties"].append(
+            {
+                "Class": class_,
+                "Property": "label",
+                "Value Type": "string",
+                "Comment": comment,
+                "Min Count": 0,
+                "Max Count": 1,
+                "Reference": "http://www.w3.org/2000/01/rdf-schema#label",
+            }
+        )
+
+    return components