python/generators/sql_processing/utils.py

# Copyright (C) 2022 The Android Open Source Project
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#      http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

from enum import Enum
import re
from typing import Dict, List


def is_internal(name: str) -> bool:
  """Check if a name represents an internal artifact (starts with _)."""
  return re.match(r'^_.*', name, re.IGNORECASE) is not None


PKG_COMMON = "common"
PKG_VIZ = "viz"
PKG_CHROME = "chrome"
PKG_ANDROID = "android"

ALLOWED_PREFIXES = {
    'android': ['heap_graph', 'memory'],
    'counters': ['counter'],
    'chrome/util': ['cr'],
    'intervals': ['interval'],
    'graphs': ['graph'],
    'slices': [
        'slice', 'thread_slice', 'process_slice', 'thread_or_process_slice'
    ],
    'linux': ['cpu', 'memory'],
    'stacks': ['cpu_profiling'],
}

# Allows for nonstandard object names.
OBJECT_NAME_ALLOWLIST = {
    'graphs/partition.sql': ['tree_structural_partition_by_group'],
    'slices/stack.sql': [
        'ancestor_slice_by_stack', 'descendant_slice_by_stack'
    ],
}

# Tables that are allowed to use CREATE TABLE instead of CREATE PERFETTO TABLE
CREATE_TABLE_ALLOWLIST = {'_trace_bounds', '_trace_metrics'}

COLUMN_TYPES = [
    # Standard types
    'LONG',
    'DOUBLE',
    'STRING',
    'BOOL',
    'BYTES',
    'ANY',

    # Special types
    'TIMESTAMP',
    'DURATION',
    'ID',
    'JOINID',
    'ARGSETID'
]

MACRO_ARG_TYPES = [
    'TABLEORSUBQUERY',
    'EXPR',
    'COLUMNNAME',
    'COLUMNNAMELIST',
    # Internal macro arg types for advanced use cases
    '_PROJECTIONFRAGMENT',  # Fragment of a SELECT projection list
    '_TABLENAMELIST'  # List of table names
]

NAME = r'[a-zA-Z_\d\{\}]+'
ANY_WORDS = r'[^\s].*'
ANY_NON_QUOTE = r'[^\']*.*'
TYPE = r'[_a-zA-Z\(\)\.]+(?:\.\.\.)?'
SQL = r'[\s\S]*?'
WS = r'\s*'

COMMENT = r' --[^\n]*\n'
COMMENTS = rf'(?:{COMMENT})*'
ARG = rf'{COMMENTS} {NAME} {TYPE}'
ARG_PATTERN = rf'({COMMENTS}) ({NAME}) ({TYPE})'
ARGS = rf'(?:{ARG})?(?: ,{ARG})*'


# Make the pattern more readable by allowing the use of spaces
# and replace then with a wildcard in a separate step.
# NOTE: two whitespaces next to each other are really bad for performance.
# Take special care to avoid them.
def update_pattern(pattern):
  return pattern.replace(' ', WS)


CREATE_TABLE_VIEW_PATTERN = update_pattern(
    # Match create table/view and catch type
    fr'^CREATE (OR REPLACE)? (VIRTUAL|PERFETTO)?'
    fr' (TABLE|VIEW) (?:IF NOT EXISTS)?'
    # Catch the name and optional schema.
    fr' ({NAME}) (?: \( ({ARGS}) \) )? (?:AS|USING)? .*')

CREATE_TABLE_AS_PATTERN = update_pattern(fr'^CREATE TABLE ({NAME}) AS')

CREATE_VIEW_AS_PATTERN = update_pattern(fr'^CREATE VIEW ({NAME}) AS')

DROP_TABLE_VIEW_PATTERN = update_pattern(
    fr'^DROP (VIEW|TABLE|INDEX) (?:IF EXISTS)? ({NAME});$')

INCLUDE_ALL_PATTERN = update_pattern(
    fr'^INCLUDE PERFETTO MODULE [a-zA-Z0-9_\.]*\*;')

CREATE_FUNCTION_PATTERN = update_pattern(
    # Function name.
    fr"CREATE (OR REPLACE)? PERFETTO FUNCTION ({NAME}) "
    # Args: anything in the brackets.
    fr" \( ({ARGS}) \)"
    # Type: word after RETURNS.
    fr"({COMMENTS})"
    fr" RETURNS ({TYPE}) AS ")

CREATE_TABLE_FUNCTION_PATTERN = update_pattern(
    fr"CREATE (OR REPLACE)? PERFETTO FUNCTION ({NAME}) "
    # Args: anything in the brackets.
    fr" \( ({ARGS}) \) "
    # Type: table definition after RETURNS.
    fr"({COMMENTS})"
    fr" RETURNS TABLE \( ({ARGS}) \) AS ")

CREATE_MACRO_PATTERN = update_pattern(
    fr"CREATE (OR REPLACE)? PERFETTO MACRO ({NAME}) "
    # Args: anything in the brackets.
    fr" \( ({ARGS}) \) "
    # Type: word after RETURNS.
    fr"({COMMENTS})"
    fr" RETURNS ({TYPE})")

INCLUDE_PATTERN = update_pattern(
    fr'^INCLUDE PERFETTO MODULE ([A-Za-z0-9_.*]*);$')

NAME_AND_TYPE_PATTERN = update_pattern(fr' ({NAME})\s+({TYPE}) ')

ARG_DEFINITION_PATTERN = update_pattern(ARG_PATTERN)

FUNCTION_RETURN_PATTERN = update_pattern(fr'^ ({TYPE})\s+({ANY_WORDS})')

ANY_PATTERN = r'(?:\s|.)*'


class ObjKind(str, Enum):
  table_view = 'table_view'
  function = 'function'
  table_function = 'table_function'
  macro = 'macro'
  include = 'include'


PATTERN_BY_KIND = {
    ObjKind.table_view: CREATE_TABLE_VIEW_PATTERN,
    ObjKind.function: CREATE_FUNCTION_PATTERN,
    ObjKind.table_function: CREATE_TABLE_FUNCTION_PATTERN,
    ObjKind.macro: CREATE_MACRO_PATTERN,
    ObjKind.include: INCLUDE_PATTERN
}


# Given a regex pattern and a string to match against, returns all the
# matching positions. Specifically, it returns a dictionary from the line
# number of the match to the regex match object.
# Note: this resuts a dict[int, re.Match], but re.Match exists only in later
# versions of python3, prior to that it was _sre.SRE_Match.
def match_pattern(pattern: str, file_str: str) -> Dict[int, object]:
  line_number_to_matches = {}
  for match in re.finditer(pattern, file_str, re.MULTILINE | re.IGNORECASE):
    line_id = file_str[:match.start()].count('\n')
    line_number_to_matches[line_id] = match.groups()
  return line_number_to_matches


# Given a list of lines in a text and the line number, scans backwards to find
# all the comments.
def extract_comment(lines: List[str], line_number: int) -> List[str]:
  comments = []
  for line in lines[line_number - 1::-1]:
    # Break on empty line, as that suggests it is no longer a part of
    # this comment.
    if not line or not line.startswith('--'):
      break
    comments.append(line)

  # Reverse as the above was reversed
  comments.reverse()
  return comments


# Given SQL string check whether any of the words is used, and create error
# string if needed.
def check_banned_words(sql: str) -> List[str]:
  lines = [l.strip() for l in sql.split('\n')]
  errors = []

  # Ban the use of LIKE in non-comment lines.
  for line in lines:
    if line.startswith('--'):
      continue

    if 'like' in line.casefold():
      errors.append('LIKE is banned in trace processor metrics and stdlib. '
                    'Prefer GLOB instead.\n')
      continue

    if 'create_function' in line.casefold():
      errors.append('CREATE_FUNCTION is deprecated in trace processor. '
                    'Use CREATE PERFETTO FUNCTION instead.')

    if 'create_view_function' in line.casefold():
      errors.append('CREATE_VIEW_FUNCTION is deprecated in trace processor. '
                    'Use CREATE PERFETTO FUNCTION $name RETURNS TABLE instead.')

    if 'import(' in line.casefold():
      errors.append('SELECT IMPORT is deprecated in trace processor. '
                    'Use INCLUDE PERFETTO MODULE instead.')

  return errors


# Given SQL string check whether there is (not allowlisted) usage of
# CREATE TABLE {name} AS.
def check_banned_create_table_as(sql: str) -> List[str]:
  errors = []
  for _, matches in match_pattern(CREATE_TABLE_AS_PATTERN, sql).items():
    name = matches[0]
    if name not in CREATE_TABLE_ALLOWLIST:
      errors.append(
          f"Table '{name}' uses CREATE TABLE which is deprecated "
          "and this table is not allowlisted. Use CREATE PERFETTO TABLE.")
  return errors


# Given SQL string check whether there is usage of CREATE VIEW {name} AS.
def check_banned_create_view_as(sql: str) -> List[str]:
  errors = []
  for _, matches in match_pattern(CREATE_VIEW_AS_PATTERN, sql).items():
    name = matches[0]
    errors.append(f"CREATE VIEW '{name}' is deprecated. "
                  "Use CREATE PERFETTO VIEW instead.")
  return errors


# Given SQL string check whether there is usage of DROP TABLE/VIEW/MACRO/INDEX.
def check_banned_drop(sql: str) -> List[str]:
  errors = []
  for _, matches in match_pattern(DROP_TABLE_VIEW_PATTERN, sql).items():
    sql_type = matches[0]
    name = matches[1]
    errors.append(f"Dropping object {sql_type} '{name}' is banned.")
  return errors


# Given SQL string check whether there is usage of CREATE VIEW {name} AS.
def check_banned_include_all(sql: str) -> List[str]:
  errors = []
  for _ in match_pattern(INCLUDE_ALL_PATTERN, sql).items():
    errors.append(
        "INCLUDE PERFETTO MODULE with wildcards is not allowed in stdlib. "
        "Import specific modules instead.")
  return errors


# Validates a parsed SQL module for banned words, valid includes and deprecated patterns.
def check_banned_patterns(parsed, sql: str) -> List[str]:
  errors = []
  # Check for banned statements
  lines = [l.strip() for l in sql.split('\n')]
  for line in lines:
    if line.startswith('--'):
      continue
    if 'run_metric' in line.casefold():
      errors.append("RUN_METRIC is banned in standard library.")
    if 'insert into' in line.casefold():
      errors.append("INSERT INTO table is not allowed in standard library.")

  # Validate includes
  package = parsed.package_name.lower() if parsed.package_name else ''
  for include in parsed.includes:
    include_package = include.package.lower() if include.package else ''

    if include_package == PKG_COMMON:
      errors.append(
          "Common module has been deprecated in the standard library. "
          "Please check `slices.with_context` for a replacement for "
          "`common.slices` and `time.conversion` for replacement for "
          "`common.timestamps`")

    if package != PKG_VIZ and include_package == PKG_VIZ:
      errors.append(
          f"No modules can depend on '{PKG_VIZ}' outside '{PKG_VIZ}' package.")

    if package == PKG_CHROME and include_package == PKG_ANDROID:
      errors.append(
          f"Modules from package '{PKG_CHROME}' can't include '{include.module}' "
          f"from package '{PKG_ANDROID}'")

    if package == PKG_ANDROID and include_package == PKG_CHROME:
      errors.append(
          f"Modules from package '{PKG_ANDROID}' can't include '{include.module}' "
          f"from package '{PKG_CHROME}'")
  # Add parsing errors and validation errors
  errors += [
      *parsed.errors, *check_banned_words(sql),
      *check_banned_create_table_as(sql), *check_banned_create_view_as(sql),
      *check_banned_include_all(sql), *check_banned_drop(sql)
  ]
  return errors