add get_valid_identifier

Author: Xinyue-Yang

tools/generated_classes.py

@@ -0,0 +1,11 @@
+from typing import Any, Union
+
+class AggregateExpression:
+    def __init__(self, agg: str, label: str = None):
+        self.agg = agg
+        self.label = label
+
+
+class AggregateTransform:
+    def __init__(self, value: Union["Argmax", "Argmin", "Avg", "Count", "Max", "Min", "First", "Last", "Median", "Mode", "Product", "Quantile", "Stddev", "StddevPop", "Sum", "Variance", "VarPop"]):
+        self.value = value

tools/schema_wrapper/utils.py

@@ -37,4 +37,67 @@
     "boolean": "bool",
     "array": "list",
     "null": "None",
-}
+}
+
+
+def get_valid_identifier(
+    prop: str,
+    replacement_character: str = "",
+    allow_unicode: bool = False,
+    url_decode: bool = True,
+) -> str:
+    """
+    Given a string property, generate a valid Python identifier.
+
+    Parameters
+    ----------
+    prop: string
+        Name of property to decode.
+    replacement_character: string, default ''
+        The character to replace invalid characters with.
+    allow_unicode: boolean, default False
+        If True, then allow Python 3-style unicode identifiers.
+    url_decode: boolean, default True
+        If True, decode URL characters in identifier names.
+
+    Examples
+    --------
+    >>> get_valid_identifier("my-var")
+    'myvar'
+
+    >>> get_valid_identifier("if")
+    'if_'
+
+    >>> get_valid_identifier("$schema", "_")
+    '_schema'
+
+    >>> get_valid_identifier("$*#$")
+    '_'
+
+    >>> get_valid_identifier("Name%3Cstring%3E")
+    'Namestring'
+    """
+    # Decode URL characters.
+    if url_decode:
+        prop = urllib.parse.unquote(prop)
+
+    # Deal with []
+    prop = prop.replace("[]", "Array")
+
+    # First substitute-out all non-valid characters.
+    flags = re.UNICODE if allow_unicode else re.ASCII
+    valid = re.sub(r"\W", replacement_character, prop, flags=flags)
+
+    # If nothing is left, use just an underscore
+    if not valid:
+        valid = "_"
+
+    # first character must be a non-digit. Prefix with an underscore
+    # if needed
+    if re.match(r"^[\d\W]", valid):
+        valid = "_" + valid
+
+    # if the result is a reserved keyword, then add an underscore at the end
+    if keyword.iskeyword(valid):
+        valid += "_"
+    return valid