|
| 1 | +import json |
| 2 | +from io import TextIOWrapper |
| 3 | +from pathlib import Path |
| 4 | +from typing import Any, Union |
| 5 | + |
| 6 | +from package_parser.commands.find_usages import ( |
| 7 | + ClassUsage, |
| 8 | + FunctionUsage, |
| 9 | + UsageStore, |
| 10 | + ValueUsage, |
| 11 | +) |
| 12 | +from package_parser.commands.get_api import API |
| 13 | +from package_parser.utils import ensure_file_exists, parent_qname |
| 14 | + |
| 15 | + |
| 16 | +def suggest_improvements( |
| 17 | + api_file: TextIOWrapper, usages_file: TextIOWrapper |
| 18 | +): |
| 19 | + with api_file: |
| 20 | + api_json = json.load(api_file) |
| 21 | + api = API.from_json(api_json) |
| 22 | + |
| 23 | + with usages_file: |
| 24 | + usages_json = json.load(usages_file) |
| 25 | + usages = UsageStore.from_json(usages_json) |
| 26 | + |
| 27 | + out_dir.mkdir(parents=True, exist_ok=True) |
| 28 | + base_file_name = api_file.name.replace("__api.json", "") |
| 29 | + |
| 30 | + __preprocess_usages(usages, api) |
| 31 | + __print_usage_counts(usages, out_dir, base_file_name) |
| 32 | + __create_usage_distributions(usages, out_dir, base_file_name) |
| 33 | + api_size_after_removal = __remove_rarely_used_api_elements( |
| 34 | + usages, min_usages, out_dir, base_file_name |
| 35 | + ) |
| 36 | + __write_api_size(api, api_size_after_removal, out_dir, base_file_name) |
| 37 | + __optional_vs_required_parameters(usages, api, out_dir, base_file_name) |
| 38 | + |
| 39 | +def __preprocess_usages(usages: UsageStore, api: API) -> None: |
| 40 | + __remove_internal_usages(usages, api) |
| 41 | + __add_unused_api_elements(usages, api) |
| 42 | + __add_implicit_usages_of_default_value(usages, api) |
| 43 | + |
| 44 | + |
| 45 | +def __create_parameter_usage_distribution(usages: UsageStore) -> dict[int, int]: |
| 46 | + """ |
| 47 | + Creates a dictionary X -> N where N indicates the number of parameters that are set at most X times to a value other |
| 48 | + than the most commonly used value (which might differ from the default value). |
| 49 | +
|
| 50 | + :param usages: Usage store. |
| 51 | + :return: The usage distribution. |
| 52 | + """ |
| 53 | + |
| 54 | + result = {} |
| 55 | + |
| 56 | + function_usages = usages.function_usages |
| 57 | + parameter_usages = usages.parameter_usages |
| 58 | + value_usages = usages.value_usages |
| 59 | + |
| 60 | + max_usages = max( |
| 61 | + __n_not_set_to_most_common_value(it, function_usages, value_usages)[1] |
| 62 | + for it in parameter_usages.keys() |
| 63 | + ) |
| 64 | + |
| 65 | + for i in range(max_usages + 1): |
| 66 | + result[i] = len( |
| 67 | + [ |
| 68 | + it |
| 69 | + for it in parameter_usages.keys() |
| 70 | + if usages.n_function_usages(parent_qname(it)) >= i |
| 71 | + and ( |
| 72 | + parent_qname(parent_qname(it)) not in usages.class_usages |
| 73 | + or usages.n_class_usages(parent_qname(parent_qname(it))) >= i |
| 74 | + ) |
| 75 | + and __n_not_set_to_most_common_value(it, function_usages, value_usages)[1] |
| 76 | + >= i |
| 77 | + ] |
| 78 | + ) |
| 79 | + |
| 80 | + return result |
| 81 | + |
| 82 | + |
| 83 | +def __remove_internal_usages(usages: UsageStore, api: API) -> None: |
| 84 | + """ |
| 85 | + Removes usages of internal parts of the API. It might incorrectly remove some calls to methods that are inherited |
| 86 | + from internal classes into a public class but these are just fit/predict/etc., i.e. something we want to keep |
| 87 | + unchanged anyway. |
| 88 | +
|
| 89 | + :param usages: Usage store |
| 90 | + :param api: Description of the API |
| 91 | + """ |
| 92 | + |
| 93 | + # Internal classes |
| 94 | + for class_qname in list(usages.class_usages.keys()): |
| 95 | + if not api.is_public_class(class_qname): |
| 96 | + print(f"Removing usages of internal class {class_qname}") |
| 97 | + usages.remove_class(class_qname) |
| 98 | + |
| 99 | + # Internal functions |
| 100 | + for function_qname in list(usages.function_usages.keys()): |
| 101 | + if not api.is_public_function(function_qname): |
| 102 | + print(f"Removing usages of internal function {function_qname}") |
| 103 | + usages.remove_function(function_qname) |
| 104 | + |
| 105 | + # Internal parameters |
| 106 | + parameter_qnames = set(api.parameters().keys()) |
| 107 | + |
| 108 | + for parameter_qname in list(usages.parameter_usages.keys()): |
| 109 | + function_qname = parent_qname(parameter_qname) |
| 110 | + if parameter_qname not in parameter_qnames or not api.is_public_function( |
| 111 | + function_qname |
| 112 | + ): |
| 113 | + print(f"Removing usages of internal parameter {parameter_qname}") |
| 114 | + usages.remove_parameter(parameter_qname) |
| 115 | + |
| 116 | + |
| 117 | +def __add_unused_api_elements(usages: UsageStore, api: API) -> None: |
| 118 | + # Public classes |
| 119 | + for class_qname in api.classes: |
| 120 | + if api.is_public_class(class_qname): |
| 121 | + usages.init_class(class_qname) |
| 122 | + |
| 123 | + # Public functions |
| 124 | + for function in api.functions.values(): |
| 125 | + if api.is_public_function(function.qname): |
| 126 | + usages.init_function(function.qname) |
| 127 | + |
| 128 | + # "Public" parameters |
| 129 | + for parameter in function.parameters: |
| 130 | + parameter_qname = f"{function.qname}.{parameter.name}" |
| 131 | + usages.init_parameter(parameter_qname) |
| 132 | + usages.init_value(parameter_qname) |
| 133 | + |
| 134 | + |
| 135 | +def __add_implicit_usages_of_default_value(usages: UsageStore, api: API) -> None: |
| 136 | + for parameter_qname, parameter_usage_list in list(usages.parameter_usages.items()): |
| 137 | + default_value = api.get_default_value(parameter_qname) |
| 138 | + if default_value is None: |
| 139 | + continue |
| 140 | + |
| 141 | + function_qname = parent_qname(parameter_qname) |
| 142 | + function_usage_list = usages.function_usages[function_qname] |
| 143 | + |
| 144 | + locations_of_implicit_usages_of_default_value = set( |
| 145 | + [it.location for it in function_usage_list] |
| 146 | + ) - set([it.location for it in parameter_usage_list]) |
| 147 | + |
| 148 | + for location in locations_of_implicit_usages_of_default_value: |
| 149 | + usages.add_value_usage(parameter_qname, default_value, location) |
| 150 | + |
| 151 | + |
| 152 | +def __n_not_set_to_most_common_value( |
| 153 | + parameter_qname: str, |
| 154 | + function_usages: dict[str, list[FunctionUsage]], |
| 155 | + value_usages: dict[str, dict[str, list[ValueUsage]]], |
| 156 | +) -> tuple: |
| 157 | + """Counts how often a parameter is set to a value other than the most commonly used value.""" |
| 158 | + |
| 159 | + n_total_usage = len(function_usages[parent_qname(parameter_qname)]) |
| 160 | + |
| 161 | + # Parameter is unused |
| 162 | + # Checking both conditions even though one implies the other to ensure correctness of the program |
| 163 | + if n_total_usage == 0 and len(value_usages[parameter_qname].values()) == 0: |
| 164 | + return 0 |
| 165 | + |
| 166 | + max_value = 0 |
| 167 | + max_key = None |
| 168 | + for key, value in value_usages[parameter_qname].entries(): |
| 169 | + if max_value < len(value): |
| 170 | + max_key = key |
| 171 | + max_value = len(value) |
| 172 | + |
| 173 | + return max_key, (n_total_usage - max_value) |
| 174 | + |
| 175 | + |
| 176 | +def __remove_rarely_used_api_elements( |
| 177 | + usages: UsageStore, min_usages: int, out_dir: Path, base_file_name: str |
| 178 | +) -> dict[str, Any]: |
| 179 | + """ |
| 180 | + Removes API elements that are used fewer than min_usages times. |
| 181 | +
|
| 182 | + :return: The API size after the individual steps. |
| 183 | + """ |
| 184 | + |
| 185 | + rarely_used_classes = __remove_rarely_used_classes(usages, min_usages) |
| 186 | + api_size_after_unused_class_removal = __api_size_to_json( |
| 187 | + len(usages.class_usages), |
| 188 | + len(usages.function_usages), |
| 189 | + len(usages.parameter_usages), |
| 190 | + ) |
| 191 | + with out_dir.joinpath( |
| 192 | + f"{base_file_name}__classes_used_fewer_than_{min_usages}_times.json" |
| 193 | + ).open("w") as f: |
| 194 | + json.dump(rarely_used_classes, f, indent=2) |
| 195 | + |
| 196 | + rarely_used_functions = __remove_rarely_used_functions(usages, min_usages) |
| 197 | + api_size_after_unused_function_removal = __api_size_to_json( |
| 198 | + len(usages.class_usages), |
| 199 | + len(usages.function_usages), |
| 200 | + len(usages.parameter_usages), |
| 201 | + ) |
| 202 | + with out_dir.joinpath( |
| 203 | + f"{base_file_name}__functions_used_fewer_than_{min_usages}_times.json" |
| 204 | + ).open("w") as f: |
| 205 | + json.dump(rarely_used_functions, f, indent=2) |
| 206 | + |
| 207 | + rarely_used_parameters = __remove_rarely_used_parameters(usages, min_usages) |
| 208 | + api_size_after_unused_parameter_removal = __api_size_to_json( |
| 209 | + len(usages.class_usages), |
| 210 | + len(usages.function_usages), |
| 211 | + len(usages.parameter_usages), |
| 212 | + ) |
| 213 | + with out_dir.joinpath( |
| 214 | + f"{base_file_name}__parameters_used_fewer_than_{min_usages}_times.json" |
| 215 | + ).open("w") as f: |
| 216 | + json.dump(rarely_used_parameters, f, indent=2) |
| 217 | + |
| 218 | + mostly_useless_parameters = __remove_mostly_useless_parameters(usages, min_usages) |
| 219 | + api_size_after_useless_parameter_removal = __api_size_to_json( |
| 220 | + len(usages.class_usages), |
| 221 | + len(usages.function_usages), |
| 222 | + len(usages.parameter_usages), |
| 223 | + ) |
| 224 | + with out_dir.joinpath( |
| 225 | + f"{base_file_name}__parameters_set_fewer_than_{min_usages}_times_to_value_other_than_most_common.json" |
| 226 | + ).open("w") as f: |
| 227 | + json.dump(mostly_useless_parameters, f, indent=2) |
| 228 | + |
| 229 | + return { |
| 230 | + "after_unused_class_removal": api_size_after_unused_class_removal, |
| 231 | + "after_unused_function_removal": api_size_after_unused_function_removal, |
| 232 | + "after_unused_parameter_removal": api_size_after_unused_parameter_removal, |
| 233 | + "after_useless_parameter_removal": api_size_after_useless_parameter_removal, |
| 234 | + } |
| 235 | + |
| 236 | + |
| 237 | +def __remove_rarely_used_classes(usages: UsageStore, min_usages: int) -> list[str]: |
| 238 | + result = [] |
| 239 | + |
| 240 | + for class_qname in list(usages.class_usages.keys()): |
| 241 | + if usages.n_class_usages(class_qname) < min_usages: |
| 242 | + result.append(class_qname) |
| 243 | + usages.remove_class(class_qname) |
| 244 | + |
| 245 | + return sorted(result) |
| 246 | + |
| 247 | + |
| 248 | +def __remove_rarely_used_functions(usages: UsageStore, min_usages: int) -> list[str]: |
| 249 | + result = [] |
| 250 | + |
| 251 | + for function_qname in list(usages.function_usages.keys()): |
| 252 | + if usages.n_function_usages(function_qname) < min_usages: |
| 253 | + result.append(function_qname) |
| 254 | + usages.remove_function(function_qname) |
| 255 | + |
| 256 | + return sorted(result) |
| 257 | + |
| 258 | + |
| 259 | +def __remove_rarely_used_parameters(usages: UsageStore, min_usages: int) -> list[str]: |
| 260 | + result = [] |
| 261 | + |
| 262 | + for parameter_qname in list(usages.parameter_usages.keys()): |
| 263 | + if usages.n_parameter_usages(parameter_qname) < min_usages: |
| 264 | + result.append(parameter_qname) |
| 265 | + usages.remove_parameter(parameter_qname) |
| 266 | + |
| 267 | + return sorted(result) |
| 268 | + |
| 269 | + |
| 270 | +def __remove_mostly_useless_parameters(usages: UsageStore) -> dict[str, int]: |
| 271 | + result = {} |
| 272 | + |
| 273 | + for parameter_qname in list(usages.parameter_usages.keys()): |
| 274 | + usage_key, usage_count = __n_not_set_to_most_common_value( |
| 275 | + parameter_qname, usages.function_usages, usages.value_usages |
| 276 | + ) |
| 277 | + |
| 278 | + if usage_count < 1: |
| 279 | + result[parameter_qname] = usage_key |
| 280 | + usages.remove_parameter(parameter_qname) |
| 281 | + |
| 282 | + return result |
| 283 | + |
| 284 | + |
| 285 | +def __write_api_size( |
| 286 | + api: API, api_size_after_removal: dict[str, Any], out_dir: Path, base_file_name: str |
| 287 | +) -> None: |
| 288 | + with out_dir.joinpath(f"{base_file_name}__api_size.json").open("w") as f: |
| 289 | + json.dump( |
| 290 | + { |
| 291 | + "full": __api_size_to_json( |
| 292 | + api.class_count(), api.function_count(), api.parameter_count() |
| 293 | + ), |
| 294 | + "public": __api_size_to_json( |
| 295 | + api.public_class_count(), |
| 296 | + api.public_function_count(), |
| 297 | + api.public_parameter_count(), |
| 298 | + ), |
| 299 | + "after_unused_class_removal": api_size_after_removal[ |
| 300 | + "after_unused_class_removal" |
| 301 | + ], |
| 302 | + "after_unused_function_removal": api_size_after_removal[ |
| 303 | + "after_unused_function_removal" |
| 304 | + ], |
| 305 | + "after_unused_parameter_removal": api_size_after_removal[ |
| 306 | + "after_unused_parameter_removal" |
| 307 | + ], |
| 308 | + "after_useless_parameter_removal": api_size_after_removal[ |
| 309 | + "after_useless_parameter_removal" |
| 310 | + ], |
| 311 | + }, |
| 312 | + f, |
| 313 | + indent=2, |
| 314 | + ) |
| 315 | + |
| 316 | + |
| 317 | +def __api_size_to_json(n_classes: int, n_functions: int, n_parameters: int) -> Any: |
| 318 | + return { |
| 319 | + "n_classes": n_classes, |
| 320 | + "n_functions": n_functions, |
| 321 | + "n_parameters": n_parameters, |
| 322 | + } |
| 323 | + |
| 324 | + |
| 325 | +def __optional_vs_required_parameters( |
| 326 | + usages: UsageStore, public_api: API, out_dir: Path, base_file_name: str |
| 327 | +) -> None: |
| 328 | + # TODO: Determine whether parameter should be constant (already removed)/required/optional based on entropy |
| 329 | + # TODO: Use must commonly set value as default |
| 330 | + |
| 331 | + pass |
0 commit comments