parser.py

try:
    from .utils import is_json_schema
except ImportError:
    from utils import is_json_schema


def parse_yaml_to_columns(content: dict) -> list:
    if isinstance(content, dict):
        if is_json_schema(content):
            schema_def = list(content.values())[0]
            if isinstance(schema_def, dict) and (
                "properties" in schema_def or "allOf" in schema_def
            ):
                columns = ["    id SERIAL PRIMARY KEY"]
                properties = schema_def.get("properties", {})
                for prop_name, prop_def in properties.items():
                    if prop_name == "id":
                        continue
                    sql_type = "TEXT"
                    if isinstance(prop_def, dict) and "type" in prop_def:
                        t = prop_def["type"]
                        if t == "string":
                            if prop_def.get("format") == "date-time":
                                sql_type = "TIMESTAMP"
                            elif "url" in prop_name.lower():
                                sql_type = "VARCHAR(500)"
                            elif prop_name == "hash":
                                sql_type = "VARCHAR(64)"
                            elif prop_name == "inLanguage":
                                sql_type = "VARCHAR(10)"
                            elif prop_name in ["author", "publisher"]:
                                sql_type = "VARCHAR(255)"
                        elif t == "integer":
                            sql_type = "INTEGER"
                        elif t == "number":
                            sql_type = "DECIMAL"
                        elif t == "boolean":
                            sql_type = "BOOLEAN"
                    columns.append(f"    {prop_name} {sql_type}")
                return columns
        return [f"    {key} TEXT" for key in content.keys()]
    elif isinstance(content, list) and content and isinstance(content[0], dict):
        return [f"    {key} TEXT" for key in content[0].keys()]
    return ["    id SERIAL PRIMARY KEY"]