| 1 |
import keyword |
| 2 |
import re |
| 3 |
|
| 4 |
from django.core.management.base import BaseCommand, CommandError |
| 5 |
from django.db import DEFAULT_DB_ALIAS, connections |
| 6 |
from django.db.models.constants import LOOKUP_SEP |
| 7 |
|
| 8 |
|
| 9 |
class Command(BaseCommand): |
| 10 |
help = ( |
| 11 |
"Introspects the database tables in the given database and outputs a Django " |
| 12 |
"model module." |
| 13 |
) |
| 14 |
requires_system_checks = [] |
| 15 |
stealth_options = ("table_name_filter",) |
| 16 |
db_module = "django.db" |
| 17 |
|
| 18 |
def add_arguments(self, parser): |
| 19 |
parser.add_argument( |
| 20 |
"table", |
| 21 |
nargs="*", |
| 22 |
type=str, |
| 23 |
help="Selects what tables or views should be introspected.", |
| 24 |
) |
| 25 |
parser.add_argument( |
| 26 |
"--database", |
| 27 |
default=DEFAULT_DB_ALIAS, |
| 28 |
choices=tuple(connections), |
| 29 |
help=( |
| 30 |
'Nominates a database to introspect. Defaults to using the "default" ' |
| 31 |
"database." |
| 32 |
), |
| 33 |
) |
| 34 |
parser.add_argument( |
| 35 |
"--include-partitions", |
| 36 |
action="store_true", |
| 37 |
help="Also output models for partition tables.", |
| 38 |
) |
| 39 |
parser.add_argument( |
| 40 |
"--include-views", |
| 41 |
action="store_true", |
| 42 |
help="Also output models for database views.", |
| 43 |
) |
| 44 |
|
| 45 |
def handle(self, **options): |
| 46 |
try: |
| 47 |
for line in self.handle_inspection(options): |
| 48 |
self.stdout.write(line) |
| 49 |
except NotImplementedError: |
| 50 |
raise CommandError( |
| 51 |
"Database inspection isn't supported for the currently selected " |
| 52 |
"database backend." |
| 53 |
) |
| 54 |
|
| 55 |
def handle_inspection(self, options): |
| 56 |
connection = connections[options["database"]] |
| 57 |
# 'table_name_filter' is a stealth option |
| 58 |
table_name_filter = options.get("table_name_filter") |
| 59 |
|
| 60 |
with connection.cursor() as cursor: |
| 61 |
yield "# This is an auto-generated Django model module." |
| 62 |
yield "# You'll have to do the following manually to clean this up:" |
| 63 |
yield "# * Rearrange models' order" |
| 64 |
yield "# * Make sure each model has one field with primary_key=True" |
| 65 |
yield ( |
| 66 |
"# * Make sure each ForeignKey and OneToOneField has `on_delete` set " |
| 67 |
"to the desired behavior" |
| 68 |
) |
| 69 |
yield ( |
| 70 |
"# * Remove `managed = False` lines if you wish to allow " |
| 71 |
"Django to create, modify, and delete the table" |
| 72 |
) |
| 73 |
yield ( |
| 74 |
"# Feel free to rename the models, but don't rename db_table values or " |
| 75 |
"field names." |
| 76 |
) |
| 77 |
yield "from %s import models" % self.db_module |
| 78 |
known_models = [] |
| 79 |
# Determine types of tables and/or views to be introspected. |
| 80 |
types = {"t"} |
| 81 |
if options["include_partitions"]: |
| 82 |
types.add("p") |
| 83 |
if options["include_views"]: |
| 84 |
types.add("v") |
| 85 |
table_info = connection.introspection.get_table_list(cursor) |
| 86 |
table_info = {info.name: info for info in table_info if info.type in types} |
| 87 |
|
| 88 |
for table_name in options["table"] or sorted(name for name in table_info): |
| 89 |
if table_name_filter is not None and callable(table_name_filter): |
| 90 |
if not table_name_filter(table_name): |
| 91 |
continue |
| 92 |
try: |
| 93 |
try: |
| 94 |
relations = connection.introspection.get_relations( |
| 95 |
cursor, table_name |
| 96 |
) |
| 97 |
except NotImplementedError: |
| 98 |
relations = {} |
| 99 |
try: |
| 100 |
constraints = connection.introspection.get_constraints( |
| 101 |
cursor, table_name |
| 102 |
) |
| 103 |
except NotImplementedError: |
| 104 |
constraints = {} |
| 105 |
primary_key_columns = ( |
| 106 |
connection.introspection.get_primary_key_columns( |
| 107 |
cursor, table_name |
| 108 |
) |
| 109 |
or [] |
| 110 |
) |
| 111 |
primary_key_column = ( |
| 112 |
primary_key_columns[0] |
| 113 |
if len(primary_key_columns) == 1 |
| 114 |
else None |
| 115 |
) |
| 116 |
unique_columns = [ |
| 117 |
c["columns"][0] |
| 118 |
for c in constraints.values() |
| 119 |
if c["unique"] and len(c["columns"]) == 1 |
| 120 |
] |
| 121 |
table_description = connection.introspection.get_table_description( |
| 122 |
cursor, table_name |
| 123 |
) |
| 124 |
except Exception as e: |
| 125 |
yield "# Unable to inspect table '%s'" % table_name |
| 126 |
yield "# The error was: %s" % e |
| 127 |
continue |
| 128 |
|
| 129 |
model_name = self.normalize_table_name(table_name) |
| 130 |
yield "" |
| 131 |
yield "" |
| 132 |
yield "class %s(models.Model):" % model_name |
| 133 |
known_models.append(model_name) |
| 134 |
|
| 135 |
if len(primary_key_columns) > 1: |
| 136 |
fields = ", ".join([f"'{col}'" for col in primary_key_columns]) |
| 137 |
yield f" pk = models.CompositePrimaryKey({fields})" |
| 138 |
|
| 139 |
used_column_names = [] # Holds column names used in the table so far |
| 140 |
column_to_field_name = {} # Maps column names to names of model fields |
| 141 |
used_relations = set() # Holds foreign relations used in the table. |
| 142 |
for row in table_description: |
| 143 |
comment_notes = ( |
| 144 |
[] |
| 145 |
) # Holds Field notes, to be displayed in a Python comment. |
| 146 |
extra_params = {} # Holds Field parameters such as 'db_column'. |
| 147 |
column_name = row.name |
| 148 |
is_relation = column_name in relations |
| 149 |
|
| 150 |
att_name, params, notes = self.normalize_col_name( |
| 151 |
column_name, used_column_names, is_relation |
| 152 |
) |
| 153 |
extra_params.update(params) |
| 154 |
comment_notes.extend(notes) |
| 155 |
|
| 156 |
used_column_names.append(att_name) |
| 157 |
column_to_field_name[column_name] = att_name |
| 158 |
|
| 159 |
# Add primary_key and unique, if necessary. |
| 160 |
if column_name == primary_key_column: |
| 161 |
extra_params["primary_key"] = True |
| 162 |
elif column_name in unique_columns: |
| 163 |
extra_params["unique"] = True |
| 164 |
|
| 165 |
if is_relation: |
| 166 |
ref_db_column, ref_db_table = relations[column_name] |
| 167 |
if extra_params.pop("unique", False) or extra_params.get( |
| 168 |
"primary_key" |
| 169 |
): |
| 170 |
rel_type = "OneToOneField" |
| 171 |
else: |
| 172 |
rel_type = "ForeignKey" |
| 173 |
ref_pk_column = ( |
| 174 |
connection.introspection.get_primary_key_column( |
| 175 |
cursor, ref_db_table |
| 176 |
) |
| 177 |
) |
| 178 |
if ref_pk_column and ref_pk_column != ref_db_column: |
| 179 |
extra_params["to_field"] = ref_db_column |
| 180 |
rel_to = ( |
| 181 |
"self" |
| 182 |
if ref_db_table == table_name |
| 183 |
else self.normalize_table_name(ref_db_table) |
| 184 |
) |
| 185 |
if rel_to in known_models: |
| 186 |
field_type = "%s(%s" % (rel_type, rel_to) |
| 187 |
else: |
| 188 |
field_type = "%s('%s'" % (rel_type, rel_to) |
| 189 |
if rel_to in used_relations: |
| 190 |
extra_params["related_name"] = "%s_%s_set" % ( |
| 191 |
model_name.lower(), |
| 192 |
att_name, |
| 193 |
) |
| 194 |
used_relations.add(rel_to) |
| 195 |
else: |
| 196 |
# Calling `get_field_type` to get the field type string and any |
| 197 |
# additional parameters and notes. |
| 198 |
field_type, field_params, field_notes = self.get_field_type( |
| 199 |
connection, table_name, row |
| 200 |
) |
| 201 |
extra_params.update(field_params) |
| 202 |
comment_notes.extend(field_notes) |
| 203 |
|
| 204 |
field_type += "(" |
| 205 |
|
| 206 |
# Don't output 'id = meta.AutoField(primary_key=True)', because |
| 207 |
# that's assumed if it doesn't exist. |
| 208 |
if att_name == "id" and extra_params == {"primary_key": True}: |
| 209 |
if field_type == "AutoField(": |
| 210 |
continue |
| 211 |
elif ( |
| 212 |
field_type |
| 213 |
== connection.features.introspected_field_types["AutoField"] |
| 214 |
+ "(" |
| 215 |
): |
| 216 |
comment_notes.append("AutoField?") |
| 217 |
|
| 218 |
# Add 'null' and 'blank', if the 'null_ok' flag was present in the |
| 219 |
# table description. |
| 220 |
if row.null_ok: # If it's NULL... |
| 221 |
extra_params["blank"] = True |
| 222 |
extra_params["null"] = True |
| 223 |
|
| 224 |
field_desc = "%s = %s%s" % ( |
| 225 |
att_name, |
| 226 |
# Custom fields will have a dotted path |
| 227 |
"" if "." in field_type else "models.", |
| 228 |
field_type, |
| 229 |
) |
| 230 |
if field_type.startswith(("ForeignKey(", "OneToOneField(")): |
| 231 |
field_desc += ", models.DO_NOTHING" |
| 232 |
|
| 233 |
# Add comment. |
| 234 |
if connection.features.supports_comments and row.comment: |
| 235 |
extra_params["db_comment"] = row.comment |
| 236 |
|
| 237 |
if extra_params: |
| 238 |
if not field_desc.endswith("("): |
| 239 |
field_desc += ", " |
| 240 |
field_desc += ", ".join( |
| 241 |
"%s=%r" % (k, v) for k, v in extra_params.items() |
| 242 |
) |
| 243 |
field_desc += ")" |
| 244 |
if comment_notes: |
| 245 |
field_desc += " # " + " ".join(comment_notes) |
| 246 |
yield " %s" % field_desc |
| 247 |
comment = None |
| 248 |
if info := table_info.get(table_name): |
| 249 |
is_view = info.type == "v" |
| 250 |
is_partition = info.type == "p" |
| 251 |
if connection.features.supports_comments: |
| 252 |
comment = info.comment |
| 253 |
else: |
| 254 |
is_view = False |
| 255 |
is_partition = False |
| 256 |
yield from self.get_meta( |
| 257 |
table_name, |
| 258 |
constraints, |
| 259 |
column_to_field_name, |
| 260 |
is_view, |
| 261 |
is_partition, |
| 262 |
comment, |
| 263 |
) |
| 264 |
|
| 265 |
def normalize_col_name(self, col_name, used_column_names, is_relation): |
| 266 |
""" |
| 267 |
Modify the column name to make it Python-compatible as a field name |
| 268 |
""" |
| 269 |
field_params = {} |
| 270 |
field_notes = [] |
| 271 |
|
| 272 |
new_name = col_name.lower() |
| 273 |
if new_name != col_name: |
| 274 |
field_notes.append("Field name made lowercase.") |
| 275 |
|
| 276 |
if is_relation: |
| 277 |
if new_name.endswith("_id"): |
| 278 |
new_name = new_name.removesuffix("_id") |
| 279 |
else: |
| 280 |
field_params["db_column"] = col_name |
| 281 |
|
| 282 |
new_name, num_repl = re.subn(r"\W", "_", new_name) |
| 283 |
if num_repl > 0: |
| 284 |
field_notes.append("Field renamed to remove unsuitable characters.") |
| 285 |
|
| 286 |
if new_name.find(LOOKUP_SEP) >= 0: |
| 287 |
while new_name.find(LOOKUP_SEP) >= 0: |
| 288 |
new_name = new_name.replace(LOOKUP_SEP, "_") |
| 289 |
if col_name.lower().find(LOOKUP_SEP) >= 0: |
| 290 |
# Only add the comment if the double underscore was in the original name |
| 291 |
field_notes.append( |
| 292 |
"Field renamed because it contained more than one '_' in a row." |
| 293 |
) |
| 294 |
|
| 295 |
if new_name.startswith("_"): |
| 296 |
new_name = "field%s" % new_name |
| 297 |
field_notes.append("Field renamed because it started with '_'.") |
| 298 |
|
| 299 |
if new_name.endswith("_"): |
| 300 |
new_name = "%sfield" % new_name |
| 301 |
field_notes.append("Field renamed because it ended with '_'.") |
| 302 |
|
| 303 |
if keyword.iskeyword(new_name): |
| 304 |
new_name += "_field" |
| 305 |
field_notes.append("Field renamed because it was a Python reserved word.") |
| 306 |
|
| 307 |
if new_name[0].isdigit(): |
| 308 |
new_name = "number_%s" % new_name |
| 309 |
field_notes.append( |
| 310 |
"Field renamed because it wasn't a valid Python identifier." |
| 311 |
) |
| 312 |
|
| 313 |
if new_name in used_column_names: |
| 314 |
num = 0 |
| 315 |
while "%s_%d" % (new_name, num) in used_column_names: |
| 316 |
num += 1 |
| 317 |
new_name = "%s_%d" % (new_name, num) |
| 318 |
field_notes.append("Field renamed because of name conflict.") |
| 319 |
|
| 320 |
if col_name != new_name and field_notes: |
| 321 |
field_params["db_column"] = col_name |
| 322 |
|
| 323 |
return new_name, field_params, field_notes |
| 324 |
|
| 325 |
def normalize_table_name(self, table_name): |
| 326 |
"""Translate the table name to a Python-compatible model name.""" |
| 327 |
return re.sub(r"[^a-zA-Z0-9]", "", table_name.title()) |
| 328 |
|
| 329 |
def get_field_type(self, connection, table_name, row): |
| 330 |
""" |
| 331 |
Given the database connection, the table name, and the cursor row |
| 332 |
description, this routine will return the given field type name, as |
| 333 |
well as any additional keyword parameters and notes for the field. |
| 334 |
""" |
| 335 |
field_params = {} |
| 336 |
field_notes = [] |
| 337 |
|
| 338 |
try: |
| 339 |
field_type = connection.introspection.get_field_type(row.type_code, row) |
| 340 |
except KeyError: |
| 341 |
field_type = "TextField" |
| 342 |
field_notes.append("This field type is a guess.") |
| 343 |
|
| 344 |
# Add max_length for all CharFields. |
| 345 |
if field_type == "CharField" and row.display_size: |
| 346 |
if (size := int(row.display_size)) and size > 0: |
| 347 |
field_params["max_length"] = size |
| 348 |
|
| 349 |
if field_type in {"CharField", "TextField"} and row.collation: |
| 350 |
field_params["db_collation"] = row.collation |
| 351 |
|
| 352 |
if field_type == "DecimalField": |
| 353 |
if row.precision is None or row.scale is None: |
| 354 |
field_notes.append( |
| 355 |
"max_digits and decimal_places have been guessed, as this " |
| 356 |
"database handles decimal fields as float" |
| 357 |
) |
| 358 |
field_params["max_digits"] = ( |
| 359 |
row.precision if row.precision is not None else 10 |
| 360 |
) |
| 361 |
field_params["decimal_places"] = ( |
| 362 |
row.scale if row.scale is not None else 5 |
| 363 |
) |
| 364 |
else: |
| 365 |
field_params["max_digits"] = row.precision |
| 366 |
field_params["decimal_places"] = row.scale |
| 367 |
|
| 368 |
return field_type, field_params, field_notes |
| 369 |
|
| 370 |
def get_meta( |
| 371 |
self, |
| 372 |
table_name, |
| 373 |
constraints, |
| 374 |
column_to_field_name, |
| 375 |
is_view, |
| 376 |
is_partition, |
| 377 |
comment, |
| 378 |
): |
| 379 |
""" |
| 380 |
Return a sequence comprising the lines of code necessary |
| 381 |
to construct the inner Meta class for the model corresponding |
| 382 |
to the given database table name. |
| 383 |
""" |
| 384 |
unique_together = [] |
| 385 |
has_unsupported_constraint = False |
| 386 |
for params in constraints.values(): |
| 387 |
if params["unique"]: |
| 388 |
columns = params["columns"] |
| 389 |
if None in columns: |
| 390 |
has_unsupported_constraint = True |
| 391 |
columns = [ |
| 392 |
x for x in columns if x is not None and x in column_to_field_name |
| 393 |
] |
| 394 |
if len(columns) > 1 and not params["primary_key"]: |
| 395 |
unique_together.append( |
| 396 |
str(tuple(column_to_field_name[c] for c in columns)) |
| 397 |
) |
| 398 |
if is_view: |
| 399 |
managed_comment = " # Created from a view. Don't remove." |
| 400 |
elif is_partition: |
| 401 |
managed_comment = " # Created from a partition. Don't remove." |
| 402 |
else: |
| 403 |
managed_comment = "" |
| 404 |
meta = [""] |
| 405 |
if has_unsupported_constraint: |
| 406 |
meta.append(" # A unique constraint could not be introspected.") |
| 407 |
meta += [ |
| 408 |
" class Meta:", |
| 409 |
" managed = False%s" % managed_comment, |
| 410 |
" db_table = %r" % table_name, |
| 411 |
] |
| 412 |
if unique_together: |
| 413 |
tup = "(" + ", ".join(unique_together) + ",)" |
| 414 |
meta += [" unique_together = %s" % tup] |
| 415 |
if comment: |
| 416 |
meta += [f" db_table_comment = {comment!r}"] |
| 417 |
return meta |