[#73912] implement more robust SAIF parser

nodist/verilator_saif_diff

@@ -0,0 +1,292 @@
+#!/usr/bin/env python3
+# pylint: disable=C0103,C0114,C0115,C0116,C0209,C0302,R0902,R0903,R0904,R0912,R0913,R0914,R0915,R0916,W0212,W0511,W0603,W1201
+######################################################################
+
+import argparse
+import re
+
+SUCCESS_CODE = 0
+FAILURE_CODE = 1
+
+INSTANCE_TYPE = "INSTANCE"
+NET_LIST_TYPE = "NET"
+SIGNAL_TYPE = "SIGNAL"
+
+EOF_ERROR = "Unexpected EOF"
+
+def saif_assert(expression, message):
+    if not expression:
+        raise Exception(message)
+    
+def saif_error(message):
+    raise Exception(message)
+
+class SAIFSignalBit:
+    name: str
+    high_time: int
+    low_time: int
+    transitions: int
+
+    def __init__(self, name):
+        self.name = name
+        self.high_time = 0
+        self.low_time = 0
+        self.transitions = 0
+
+
+class SAIFInstance:
+
+    def __init__(self, scope_name):
+        self.scope_name = scope_name
+        self.parent_instance = None
+        self.nets = {}
+        self.child_instances = {}
+
+
+class SAIFToken:
+
+    def __init__(self, token):
+        self.token = token
+        self.type = ''
+        self.value = ''
+
+
+class SAIFParser:
+
+    def __init__(self):
+        self.token_stack = []
+
+        # for parsing simplicity
+        self.token_stack.append(SAIFToken('saif_root'))
+
+        self.current_instance = None
+        
+        self.has_saifile_header = False
+        self.direction = ''
+        self.saif_version = ''
+        self.top_instances = {}
+        self.duration = 0
+        self.divider = ''
+        self.timescale = ''
+
+    def parse(self, saif_filename):
+        file_contents = ''
+
+        with open(saif_filename, 'r', encoding="utf8") as saif_file:
+            file_contents = saif_file.read()
+
+        tokens = file_contents.replace('(', ' ( ').replace(')', ' ) ').split()
+        num_of_tokens = len(tokens)
+
+        index = 0
+        while index < num_of_tokens:
+            token = tokens[index]
+            index += 1
+
+            if token == '(':
+                self.token_stack.append(SAIFToken(token))
+                self.token_stack[-1].type = self.token_stack[-2].type
+                self.token_stack[-1].value = self.token_stack[-2].value
+
+                continue
+
+            if token == ')':
+                if self.token_stack[-1].type == INSTANCE_TYPE:
+                    self.current_instance = self.current_instance.parent_instance
+
+                self.token_stack.pop()
+                continue
+
+            if re.match(r'SAIFILE', token):
+                self.has_saifile_header = True
+
+                continue
+
+            if re.match(r'DIRECTION', token):
+                saif_assert(index < num_of_tokens, EOF_ERROR)
+                self.direction = tokens[index].replace('\"', '')
+                index += 1
+
+                continue
+
+            if re.match(r'SAIFVERSION', token):
+                saif_assert(index < num_of_tokens, EOF_ERROR)
+                self.saif_version = tokens[index].replace('\"', '')
+                index += 1
+
+                continue
+
+            if re.match(r'DESIGN|DATE|VENDOR|PROGRAM_NAME|VERSION', token):
+                # noop, only skip value
+                saif_assert(index < num_of_tokens, EOF_ERROR)
+                index += 1
+
+                continue
+
+            if re.match(r'DIVIDER', token):
+                saif_assert(index < num_of_tokens, EOF_ERROR)
+                self.divider = tokens[index]
+                index += 1
+
+                continue
+
+            if re.match(r'TIMESCALE', token):
+                saif_assert(index < num_of_tokens, EOF_ERROR)
+                self.timescale = tokens[index]
+                index += 1
+
+                continue
+
+            if re.match(r'DURATION', token):
+                saif_assert(index < num_of_tokens, EOF_ERROR)
+                self.duration = tokens[index]
+                index += 1
+
+                continue
+
+            if re.match(r'INSTANCE', token):
+                saif_assert(index < num_of_tokens, EOF_ERROR)
+                instance_name = tokens[index]
+                index += 1
+
+                self.token_stack[-1].type = INSTANCE_TYPE
+                self.token_stack[-1].value = instance_name
+
+                instance = SAIFInstance(instance_name)
+
+                if self.current_instance is None:
+                    self.top_instances[instance_name] = instance
+                else:
+                    self.current_instance.child_instances[instance_name] = instance
+
+                instance.parent_instance = self.current_instance
+                self.current_instance = instance
+
+                continue
+
+            if re.match(r'NET', token):
+                self.token_stack[-1].type = NET_LIST_TYPE
+
+                continue
+
+            if re.match(r'T1', token):
+                net_name = self.token_stack[-1].value
+                saif_assert(index < num_of_tokens, EOF_ERROR)
+                self.current_instance.nets[net_name].high_time = tokens[index]
+                index += 1
+
+                continue
+
+            if re.match(r'T0', token):
+                net_name = self.token_stack[-1].value
+                saif_assert(index < num_of_tokens, EOF_ERROR)
+                self.current_instance.nets[net_name].low_time = tokens[index]
+                index += 1
+
+                continue
+
+            if re.match(r'TC', token):
+                net_name = self.token_stack[-1].value
+                saif_assert(index < num_of_tokens, EOF_ERROR)
+                self.current_instance.nets[net_name].transitions = tokens[index]
+                index += 1
+
+                continue
+
+            if re.match(r'TZ|TX|TB|TG|IG|IK', token):
+                # noop, only skip value
+                index += 1
+                continue
+            
+            if self.token_stack[-2].type == NET_LIST_TYPE:
+                self.token_stack[-1].type = SIGNAL_TYPE
+                self.token_stack[-1].value = token
+                self.current_instance.nets[token] = (SAIFSignalBit(token))
+
+        saif_assert(self.has_saifile_header, "SAIF file doesn't contain a SAIFILE keyword")
+        saif_assert(self.direction == "backward", f"SAIF file doesn't have a valid/compatible direction: {self.direction}")
+        saif_assert(self.saif_version == "2.0", f"SAIF file doesn't have a valid/compatible version: {self.saif_version}")
+
+        # only 'saif_root' token should be left
+        saif_assert(len(self.token_stack) == 1, "Incorrect nesting of scopes")
+
+                        
+def compare_saif_instances(first: SAIFInstance, second: SAIFInstance):
+    if len(first.nets) != len(second.nets):
+        saif_error(f"Number of nets doesn't match in {first.scope_name}: "
+                    f"{len(first.nets)} != {len(second.nets)}")
+
+    for signal_name, saif_signal in first.nets.items():
+        if signal_name not in second.nets:
+            saif_error(f"Signal {signal_name} doesn't exist in the second object\n")
+
+        other_signal = second.nets[signal_name]
+
+        if (saif_signal.high_time != other_signal.high_time
+                or saif_signal.low_time != other_signal.low_time
+                or saif_signal.transitions != other_signal.transitions):
+            saif_error("Incompatible signal bit parameters in "
+                        f"{signal_name}\n")
+
+    if len(first.child_instances) != len(second.child_instances):
+        saif_error(f"Number of child instances doesn't match in {first.scope_name}: "
+                    f"{len(first.child_instances)} != {len(second.child_instances)}")
+
+    for instance_name, instance in first.child_instances.items():
+        if instance_name not in second.child_instances:
+            saif_error(f"Instance {instance_name} doesn't exist in the second object\n")
+
+        compare_saif_instances(instance, second.child_instances[instance_name])
+
+
+def compare_saif_contents(first_file: str, second_file: str):
+    """Test if second SAIF file has the same values as the first"""
+
+    first_saif = SAIFParser()
+    first_saif.parse(first_file)
+
+    second_saif = SAIFParser()
+    second_saif.parse(second_file)
+
+    if first_saif.duration != second_saif.duration:
+        saif_error(f"Duration of trace doesn't match: {first_saif.duration} != {second_saif.duration}")
+
+    if first_saif.divider != second_saif.divider:
+        saif_error(f"Dividers don't match: {first_saif.divider} != {second_saif.divider}")
+
+    if first_saif.timescale != second_saif.timescale:
+        saif_error(f"Timescale doesn't match: {first_saif.timescale} != {second_saif.timescale}")
+
+    if len(first_saif.top_instances) != len(second_saif.top_instances):
+        saif_error("Number of top instances doesn't match: "
+                    f"{len(first_saif.top_instances)} != {len(second_saif.top_instances)}")
+
+    for top_instance_name, top_instance in first_saif.top_instances.items():
+        if top_instance_name not in second_saif.top_instances:
+            saif_error(f"Top instance {top_instance_name} missing in other SAIF")
+
+        compare_saif_instances(top_instance, second_saif.top_instances[top_instance_name])
+
+    return SUCCESS_CODE
+
+
+parser = argparse.ArgumentParser(
+    allow_abbrev=False,
+    formatter_class=argparse.RawDescriptionHelpFormatter,
+    description="""""")
+
+parser.add_argument('--first',
+                    action='store',
+                    help='First SAIF file')
+
+parser.add_argument('--second',
+                    action='store',
+                    help='Second SAIF file')
+
+parser.set_defaults(stop=True)
+args = parser.parse_args()
+
+try:
+    compare_saif_contents(args.first, args.second)
+except Exception as error:
+    print(error)

test_regress/driver.py

@@ -52,6 +52,7 @@ Quitting = False
 Vltmt_Threads = 3
 forker = None
 Start = None
+nodist_directory = "../nodist"
 
 # So an 'import vltest_bootstrap' inside test files will do nothing
 sys.modules['vltest_bootstrap'] = {}
@@ -67,145 +68,6 @@ class staticproperty(property):
         return self.fget()
 
 
-#######################################################################
-#######################################################################
-# SAIF parser utilities
-
-
-class SAIFSignalBit:
-    last_val: int
-    high_time: int
-    low_time: int
-    transitions: int
-
-    def __init__(self):
-        self.last_val = 0
-        self.high_time = 0
-        self.low_time = 0
-        self.transitions = 0
-
-    def aggregate(self, dt: int, new_val: int):
-        if new_val != self.last_val:
-            self.transitions += 1
-
-        if self.last_val == 1:
-            self.high_time += dt
-
-        self.last_val = new_val
-
-
-class SAIFSignal:
-
-    def __init__(self, signal_name, signal_width=0):
-        self.name = signal_name
-        self.width = signal_width
-        self.last_time = 0
-
-        self.bits = []
-        for _ in range(self.width):
-            self.bits.append(SAIFSignalBit())
-
-
-class SAIFInstance:
-
-    def __init__(self, scope_name):
-        self.scope_name = scope_name
-        self.parent_instance = None
-        self.nets = {}
-        self.child_instances = {}
-
-
-class SAIFParser:
-
-    def __init__(self):
-        self.top_instances = {}
-        self.current_instance = None
-        self.traversing_nets = False
-        self.duration = 0
-        self.divider = ''
-        self.timescale = ''
-
-    def parse(self, saif_filename):
-        with open(saif_filename, 'r', encoding="utf8") as saif_file:
-            for line in saif_file:
-                line = line.strip()
-                if not line:
-                    continue
-
-                match = re.search(r'\(DIVIDER\s+(.)', line)
-                if match:
-                    self.divider = match.groups()[0]
-
-                match = re.search(r'\(TIMESCALE\s+(\d+\s*\w+)', line)
-                if match:
-                    self.timescale = match.groups()[0]
-
-                match = re.search(r'\s*\(DURATION\s+(\d+)', line)
-                if match:
-                    self.duration = int(match.groups()[0])
-
-                match = re.search(r'INSTANCE\s+([\w\.\$]+)', line)
-                if match:
-                    instance_name = match.groups()[0]
-
-                    instance = SAIFInstance(instance_name)
-
-                    if self.current_instance is None:
-                        self.top_instances[instance_name] = instance
-                    else:
-                        self.current_instance.child_instances[instance_name] = instance
-
-                    instance.parent_instance = self.current_instance
-                    self.current_instance = instance
-
-                match = re.search(r'NET', line)
-                if match:
-                    self.traversing_nets = True
-
-                match = re.search(r'((?:[\w\[\]])+)(?:\\\[(\d+)\\\])*\s+(\(T.+\))+', line)
-                if match:
-                    signal_name, bit_index, bit_values = match.groups()
-
-                    if bit_index is None:
-                        bit_index = 0
-
-                    if signal_name not in self.current_instance.nets:
-                        saif_signal = SAIFSignal(signal_name)
-                        self.current_instance.nets[signal_name] = saif_signal
-
-                    current_signal = self.current_instance.nets[signal_name]
-
-                    for _ in range(0, int(bit_index) - current_signal.width + 1):
-                        current_signal.bits.append(SAIFSignalBit())
-
-                    current_signal.width = int(bit_index) + 1
-
-                    match = re.search(r'T0 (\d+)', bit_values)
-                    if match:
-                        low_time = match.groups()[0]
-
-                        current_signal.bits[int(bit_index)].low_time = int(low_time)
-
-                    match = re.search(r'T1 (\d+)', bit_values)
-                    if match:
-                        high_time = match.groups()[0]
-
-                        current_signal.bits[int(bit_index)].high_time = int(high_time)
-
-                    match = re.search(r'TC (\d+)', bit_values)
-                    if match:
-                        toggle_count = match.groups()[0]
-
-                        current_signal.bits[int(bit_index)].transitions = int(toggle_count)
-
-                match = re.match(r'\s+\)\s+', line)
-                if match:
-                    if self.traversing_nets:
-                        self.traversing_nets = False
-                    else:
-                        self.current_instance = self.current_instance.parent_instance
-
-
 #######################################################################
 #######################################################################
 # VtOs -- OS extensions
@@ -2507,71 +2369,15 @@ class VlTest:
         self.fst2vcd(fn1, tmp)
         self.vcd_identical(tmp, fn2)
 
-    def compare_saif_instances(self, first: SAIFInstance, second: SAIFInstance):
-        if len(first.nets) != len(second.nets):
-            self.error(f"Number of nets doesn't match in {first.scope_name}: "
-                       f"{len(first.nets)} != {len(second.nets)}")
-
-        for signal_name, saif_signal in first.nets.items():
-            if signal_name not in second.nets:
-                self.error(f"Signal {signal_name} doesn't exist in the second object\n")
-
-            other_signal = second.nets[signal_name]
-            if other_signal.width != saif_signal.width:
-                self.error("Incompatible signal width in "
-                           f"{signal_name} {saif_signal.width} != {other_signal.width}\n")
-
-            for bit_index in range(saif_signal.width):
-                signal_bit = saif_signal.bits[bit_index]
-                other_signal_bit = other_signal.bits[bit_index]
-
-                if (signal_bit.high_time != other_signal_bit.high_time
-                        or signal_bit.low_time != other_signal_bit.low_time
-                        or signal_bit.transitions != other_signal_bit.transitions):
-                    self.error("Incompatible signal bit parameters in "
-                               f"{signal_name}[{bit_index}]\n")
-
-        if len(first.child_instances) != len(second.child_instances):
-            self.error(f"Number of child instances doesn't match in {first.scope_name}: "
-                       f"{len(first.child_instances)} != {len(second.child_instances)}")
-
-        for instance_name, instance in first.child_instances.items():
-            if instance_name not in second.child_instances:
-                self.error(f"Instance {instance_name} doesn't exist in the second object\n")
-
-            self.compare_saif_instances(instance, second.child_instances[instance_name])
-
-    def compare_saif_contents(self, first: SAIFParser, second: SAIFParser):
-        """Test if second SAIF file has the same values as the first"""
-
-        if first.duration != second.duration:
-            self.error(f"Duration of trace doesn't match: {first.duration} != {second.duration}")
-
-        if first.divider != second.divider:
-            self.error(f"Dividers don't match: {first.divider} != {second.divider}")
-
-        if first.timescale != second.timescale:
-            self.error(f"Timescale doesn't match: {first.timescale} != {second.timescale}")
-
-        if len(first.top_instances) != len(second.top_instances):
-            self.error("Number of top instances doesn't match: "
-                       f"{len(first.top_instances)} != {len(second.top_instances)}")
-
-        for top_instance_name, top_instance in first.top_instances.items():
-            if top_instance_name not in second.top_instances:
-                self.error(f"Top instance {top_instance_name} missing in other SAIF")
-
-            self.compare_saif_instances(top_instance, second.top_instances[top_instance_name])
-
     def saif_identical(self, fn1: str, fn2: str) -> None:
         """Test if two SAIF files have logically-identical contents"""
-        test_result_saif = SAIFParser()
-        test_result_saif.parse(fn1)
 
-        golden_saif = SAIFParser()
+        cmd = nodist_directory + '/verilator_saif_diff --first "' + fn1 + '" --second "' + fn2 + '"'
-        golden_saif.parse(fn2)
+        print("\t " + cmd + "\n")
-
+        out = test.run_capture(cmd, check=True)
-        self.compare_saif_contents(golden_saif, test_result_saif)
+        if out != '':
+            print(out)
+            self.error("SAIF files don't match!")
 
     def _vcd_read(self, filename: str) -> str:
         data = {}