1#!/usr/bin/env python 2"""A script to generate FileCheck statements for mlir unit tests. 3 4This script is a utility to add FileCheck patterns to an mlir file. 5 6NOTE: The input .mlir is expected to be the output from the parser, not a 7stripped down variant. 8 9Example usage: 10$ generate-test-checks.py foo.mlir 11$ mlir-opt foo.mlir -transformation | generate-test-checks.py 12 13The script will heuristically insert CHECK/CHECK-LABEL commands for each line 14within the file. By default this script will also try to insert string 15substitution blocks for all SSA value names. The script is designed to make 16adding checks to a test case fast, it is *not* designed to be authoritative 17about what constitutes a good test! 18""" 19 20# Copyright 2019 The MLIR Authors. 21# 22# Licensed under the Apache License, Version 2.0 (the "License"); 23# you may not use this file except in compliance with the License. 24# You may obtain a copy of the License at 25# 26# http://www.apache.org/licenses/LICENSE-2.0 27# 28# Unless required by applicable law or agreed to in writing, software 29# distributed under the License is distributed on an "AS IS" BASIS, 30# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 31# See the License for the specific language governing permissions and 32# limitations under the License. 33 34import argparse 35import os # Used to advertise this file's name ("autogenerated_note"). 36import re 37import sys 38 39ADVERT = '// NOTE: Assertions have been autogenerated by ' 40 41# Regex command to match an SSA identifier. 42SSA_RE_STR = '[0-9]+|[a-zA-Z$._-][a-zA-Z0-9$._-]*' 43SSA_RE = re.compile(SSA_RE_STR) 44 45 46# Class used to generate and manage string substitution blocks for SSA value 47# names. 48class SSAVariableNamer: 49 50 def __init__(self): 51 self.scopes = [] 52 self.name_counter = 0 53 54 # Generate a substitution name for the given ssa value name. 55 def generate_name(self, ssa_name): 56 variable = 'VAL_' + str(self.name_counter) 57 self.name_counter += 1 58 self.scopes[-1][ssa_name] = variable 59 return variable 60 61 # Push a new variable name scope. 62 def push_name_scope(self): 63 self.scopes.append({}) 64 65 # Pop the last variable name scope. 66 def pop_name_scope(self): 67 self.scopes.pop() 68 69 70# Process a line of input that has been split at each SSA identifier '%'. 71def process_line(line_chunks, variable_namer): 72 output_line = '' 73 74 # Process the rest that contained an SSA value name. 75 for chunk in line_chunks: 76 m = SSA_RE.match(chunk) 77 ssa_name = m.group(0) 78 79 # Check if an existing variable exists for this name. 80 variable = None 81 for scope in variable_namer.scopes: 82 variable = scope.get(ssa_name) 83 if variable is not None: 84 break 85 86 # If one exists, then output the existing name. 87 if variable is not None: 88 output_line += '[[' + variable + ']]' 89 else: 90 # Otherwise, generate a new variable. 91 variable = variable_namer.generate_name(ssa_name) 92 output_line += '[[' + variable + ':%.*]]' 93 94 # Append the non named group. 95 output_line += chunk[len(ssa_name):] 96 97 return output_line + '\n' 98 99 100# Pre-process a line of input to remove any character sequences that will be 101# problematic with FileCheck. 102def preprocess_line(line): 103 # Replace any double brackets, '[[' with escaped replacements. '[[' 104 # corresponds to variable names in FileCheck. 105 output_line = line.replace('[[', '{{\\[\\[}}') 106 107 # Replace any single brackets that are followed by an SSA identifier, the 108 # identifier will be replace by a variable; Creating the same situation as 109 # above. 110 output_line = output_line.replace('[%', '{{\\[}}%') 111 112 return output_line 113 114 115def main(): 116 parser = argparse.ArgumentParser( 117 description=__doc__, formatter_class=argparse.RawTextHelpFormatter) 118 parser.add_argument( 119 '--check-prefix', default='CHECK', help='Prefix to use from check file.') 120 parser.add_argument( 121 '-o', 122 '--output', 123 nargs='?', 124 type=argparse.FileType('w'), 125 default=sys.stdout) 126 parser.add_argument( 127 'input', 128 nargs='?', 129 type=argparse.FileType('r'), 130 default=sys.stdin) 131 args = parser.parse_args() 132 133 # Open the given input file. 134 input_lines = [l.rstrip() for l in args.input] 135 args.input.close() 136 137 output_lines = [] 138 139 # Generate a note used for the generated check file. 140 script_name = os.path.basename(__file__) 141 autogenerated_note = (ADVERT + 'utils/' + script_name) 142 output_lines.append(autogenerated_note + '\n') 143 144 # A map containing data used for naming SSA value names. 145 variable_namer = SSAVariableNamer() 146 for input_line in input_lines: 147 if not input_line: 148 continue 149 lstripped_input_line = input_line.lstrip() 150 151 # Lines with blocks begin with a ^. These lines have a trailing comment 152 # that needs to be stripped. 153 is_block = lstripped_input_line[0] == '^' 154 if is_block: 155 input_line = input_line.rsplit('//', 1)[0].rstrip() 156 157 # Top-level operations are heuristically the operations at nesting level 1. 158 is_toplevel_op = (not is_block and input_line.startswith(' ') and 159 input_line[2] != ' ' and input_line[2] != '}') 160 161 # If the line starts with a '}', pop the last name scope. 162 if lstripped_input_line[0] == '}': 163 variable_namer.pop_name_scope() 164 165 # If the line ends with a '{', push a new name scope. 166 if input_line[-1] == '{': 167 variable_namer.push_name_scope() 168 169 # Preprocess the input to remove any sequences that may be problematic with 170 # FileCheck. 171 input_line = preprocess_line(input_line) 172 173 # Split the line at the each SSA value name. 174 ssa_split = input_line.split('%') 175 176 # If this is a top-level operation use 'CHECK-LABEL', otherwise 'CHECK:'. 177 if not is_toplevel_op or not ssa_split[0]: 178 output_line = '// ' + args.check_prefix + ': ' 179 # Pad to align with the 'LABEL' statements. 180 output_line += (' ' * len('-LABEL')) 181 182 # Output the first line chunk that does not contain an SSA name. 183 output_line += ssa_split[0] 184 185 # Process the rest of the input line. 186 output_line += process_line(ssa_split[1:], variable_namer) 187 188 else: 189 # Append a newline to the output to separate the logical blocks. 190 output_lines.append('\n') 191 output_line = '// ' + args.check_prefix + '-LABEL: ' 192 193 # Output the first line chunk that does not contain an SSA name for the 194 # label. 195 output_line += ssa_split[0] + '\n' 196 197 # Process the rest of the input line on a separate check line. 198 if len(ssa_split) > 1: 199 output_line += '// ' + args.check_prefix + '-SAME: ' 200 201 # Pad to align with the original position in the line. 202 output_line += ' ' * len(ssa_split[0]) 203 204 # Process the rest of the line. 205 output_line += process_line(ssa_split[1:], variable_namer) 206 207 # Append the output line. 208 output_lines.append(output_line) 209 210 # Write the output. 211 for output_line in output_lines: 212 args.output.write(output_line) 213 args.output.write('\n') 214 args.output.close() 215 216 217if __name__ == '__main__': 218 main() 219