[roottest] Precompile regexes in custom_diff.py.

Instead of specifying regexes lazily inside the line filter loop,
precompile them by instantiating a filter class. In this way, adding a
new filter or replacement can be done by adding a single line, and the
regexes run a tiny bit faster.

Author: hageboeck

roottest/scripts/custom_diff.py

@@ -13,55 +13,67 @@
 import sys
 import time
 
+
 #---------------------------------------------------------------------------------------------------------------------------
 #---Filter and substitutions------------------------------------------------------------------------------------------------
+class LineFilter:
+  """A line filter to suppress lines in the diff.
+     self.skip_regexes contains patterns to skip entirely.
+     self.substitutions contains patterns to replace with the given strings.
+     These patterns only need to be compiled once for all the incoming lines.
+  """
 
-def filter(lines, ignoreWhiteSpace = False):
-  outlines = []
-  for line in lines:
-    if sys.platform == 'win32':
-      if 'Creating library ' in line:
-        continue
-      if '_ACLiC_dict' in line:
-        continue
-      if 'Warning in <TInterpreter::ReadRootmapFile>:' in line:
-        continue
-      if 'Warning in <TClassTable::Add>:' in line:
-        continue
-      if 'Error: Removing ' in line:
+  def __init__(self):
+    # Skip these lines
+    self.skip_regexes = [re.compile(pattern) for pattern in [
+      r'^Processing ', # Interpreted macros
+      r'^Info in <\w+::ACLiC>: creating shared library', # Compiled macros
+      r'^In file included from input_line', # Wrapper input line
+      r'^[:space:]*$', # Lines which are empty apart from spaces
+    ]]
+
+    # Replace these patterns in all lines
+    self.substitutions = [
+      (re.compile(r'[ ]@0x[a-fA-F0-9]+'), ''), # Remove pointers from output
+      (re.compile(r'std::__[0-9]+::'), 'std::'), # Canonicalise standard namespaces
+      (re.compile(r'^(\S*/|)([^:/]+)[-:0-9]*(?=: error:)'), r'\2'), # Trim file paths and line numbers from lines with ": error:"
+      (re.compile(r'(input_line_|ROOT_prompt_)[0-9]+'), r'\1--'), # Canonicalise input_line_123, ROOT_prompt_345
+    ]
+
+  def filter(self, lines, ignoreWhiteSpace = False):
+    outlines = []
+    for line in lines:
+      if sys.platform == 'win32':
+        if 'Creating library ' in line:
+          continue
+        if '_ACLiC_dict' in line:
+          continue
+        if 'Warning in <TInterpreter::ReadRootmapFile>:' in line:
+          continue
+        if 'Warning in <TClassTable::Add>:' in line:
+          continue
+        if 'Error: Removing ' in line:
+          continue
+        if ' -nologo -TP -c -nologo -I' in line:
+          continue
+        if 'rootcling -v1 -f ' in line:
+          continue
+        if 'No precompiled header available' in line:
+          continue
+
+      #---Skip all lines matching predefined expressions---------------------------
+      if any(pattern.match(line) is not None for pattern in self.skip_regexes):
         continue
-      if ' -nologo -TP -c -nologo -I' in line:
-        continue
-      if 'rootcling -v1 -f ' in line:
-        continue
-      if 'No precompiled header available' in line:
-        continue
-      #if line in ['\n', '\r\n']:
-      #  continue
-    #---Processing line from interpreter (root.exe)------------------------------
-    if re.match(r'^Processing ', line):
-      continue
-    #---ACLiC info---------------------------------------------------------------
-    if re.match(r'^Info in <\w+::ACLiC>: creating shared library', line):
-      continue
-    #---Compilation error--------------------------------------------------------
-    elif re.search(r': error:', line):
-      nline = re.sub(r'\S+/', '', line)
-      nline = re.sub(r'(:|_)[0-9]+(?=:)', ':--', nline)
-    #---Wrapper input line-------------------------------------------------------
-    elif re.match(r'^In file included from input_line', line):
-      continue
-    else:
-      nline = line
-    #---Remove Addresses in cling/cint-------------------------------------------
-    nline = re.sub(r'[ ]@0x[a-fA-F0-9]+', '', nline)
-    #---Remove versioning in std-------------------------------------------------
-    nline = re.sub(r'std::__[0-9]::', 'std::', nline)
-    #---Remove white spaces------------------------------------------------------
-    if (ignoreWhiteSpace):
-      nline = re.sub(r'[ ]', '', nline)
-    outlines.append(nline)
-  return outlines
+
+      #---Apply predefined replacements--------------------------------------------
+      for pattern, replacement in self.substitutions:
+        line = pattern.sub(replacement, line)
+
+      #---Remove white spaces------------------------------------------------------
+      if (ignoreWhiteSpace):
+        line = re.sub(r'[ ]', '', line)
+      outlines.append(line)
+    return outlines
 
 #-----------------------------------------------------------------------------------------------------------------------------
 def main():
@@ -92,17 +104,18 @@ def main():
     fromlines = open(fromfile, 'r' if sys.version_info >= (3, 4) else 'U').readlines()
     tolines = open(tofile, 'r' if sys.version_info >= (3, 4) else 'U').readlines()
 
-  nows_fromlines = filter(fromlines, True)
-  nows_tolines = filter(tolines, True)
+  lineFilter = LineFilter()
+  nows_fromlines = lineFilter.filter(fromlines, True)
+  nows_tolines = lineFilter.filter(tolines, True)
 
   check = difflib.context_diff(nows_fromlines, nows_tolines)
   try:
     _ = next(check)
   except StopIteration:
     sys.exit(0)
 
-  fromlines = filter(fromlines, False)
-  tolines = filter(tolines, False)
+  fromlines = lineFilter.filter(fromlines, False)
+  tolines = lineFilter.filter(tolines, False)
 
   if options.u:
     diff = difflib.unified_diff(fromlines, tolines, fromfile, tofile, fromdate, todate, n=n)