| #!/usr/bin/env python | 
 | # Copyright (c) 2012 The Chromium Authors. All rights reserved. | 
 | # Use of this source code is governed by a BSD-style license that can be | 
 | # found in the LICENSE file. | 
 |  | 
 | """ Lexer for PPAPI IDL """ | 
 |  | 
 | # | 
 | # IDL Lexer | 
 | # | 
 | # The lexer is uses the PLY lex library to build a tokenizer which understands | 
 | # WebIDL tokens. | 
 | # | 
 | # WebIDL, and WebIDL regular expressions can be found at: | 
 | #   http://dev.w3.org/2006/webapi/WebIDL/ | 
 | # PLY can be found at: | 
 | #   http://www.dabeaz.com/ply/ | 
 |  | 
 | import os.path | 
 | import re | 
 | import sys | 
 |  | 
 | # | 
 | # Try to load the ply module, if not, then assume it is in the third_party | 
 | # directory, relative to ppapi | 
 | # | 
 | try: | 
 |   from ply import lex | 
 | except: | 
 |   module_path, module_name = os.path.split(__file__) | 
 |   third_party = os.path.join(module_path, '..', '..', 'third_party') | 
 |   sys.path.append(third_party) | 
 |   from ply import lex | 
 |  | 
 | from idl_option import GetOption, Option, ParseOptions | 
 |  | 
 |  | 
 | Option('output', 'Generate output.') | 
 |  | 
 | # | 
 | # IDL Lexer | 
 | # | 
 | class IDLLexer(object): | 
 |   # 'tokens' is a value required by lex which specifies the complete list | 
 |   # of valid token types. | 
 |   tokens = [ | 
 |     # Symbol and keywords types | 
 |       'COMMENT', | 
 |       'DESCRIBE', | 
 |       'ENUM', | 
 |       'LABEL', | 
 |       'SYMBOL', | 
 |       'INLINE', | 
 |       'INTERFACE', | 
 |       'STRUCT', | 
 |       'TYPEDEF', | 
 |       'OR', | 
 |  | 
 |     # Extra WebIDL keywords | 
 |       'CALLBACK', | 
 |       'DICTIONARY', | 
 |       'OPTIONAL', | 
 |       'STATIC', | 
 |  | 
 |     # Invented for apps use | 
 |       'NAMESPACE', | 
 |  | 
 |     # Data types | 
 |       'FLOAT', | 
 |       'OCT', | 
 |       'INT', | 
 |       'HEX', | 
 |       'STRING', | 
 |  | 
 |     # Operators | 
 |       'LSHIFT', | 
 |       'RSHIFT' | 
 |   ] | 
 |  | 
 |   # 'keywords' is a map of string to token type.  All SYMBOL tokens are | 
 |   # matched against keywords, to determine if the token is actually a keyword. | 
 |   keywords = { | 
 |     'describe' : 'DESCRIBE', | 
 |     'enum'  : 'ENUM', | 
 |     'label' : 'LABEL', | 
 |     'interface' : 'INTERFACE', | 
 |     'readonly' : 'READONLY', | 
 |     'struct' : 'STRUCT', | 
 |     'typedef' : 'TYPEDEF', | 
 |  | 
 |     'callback' : 'CALLBACK', | 
 |     'dictionary' : 'DICTIONARY', | 
 |     'optional' : 'OPTIONAL', | 
 |     'static' : 'STATIC', | 
 |     'namespace' : 'NAMESPACE', | 
 |  | 
 |     'or' : 'OR', | 
 |   } | 
 |  | 
 |   # 'literals' is a value expected by lex which specifies a list of valid | 
 |   # literal tokens, meaning the token type and token value are identical. | 
 |   literals = '"*.(){}[],;:=+-/~|&^?' | 
 |  | 
 |   # Token definitions | 
 |   # | 
 |   # Lex assumes any value or function in the form of 't_<TYPE>' represents a | 
 |   # regular expression where a match will emit a token of type <TYPE>.  In the | 
 |   # case of a function, the function is called when a match is made. These | 
 |   # definitions come from WebIDL. | 
 |  | 
 |   # 't_ignore' is a special match of items to ignore | 
 |   t_ignore = ' \t' | 
 |  | 
 |   # Constant values | 
 |   t_FLOAT = r'-?(\d+\.\d*|\d*\.\d+)([Ee][+-]?\d+)?|-?\d+[Ee][+-]?\d+' | 
 |   t_INT = r'-?[0-9]+[uU]?' | 
 |   t_OCT = r'-?0[0-7]+' | 
 |   t_HEX = r'-?0[Xx][0-9A-Fa-f]+' | 
 |   t_LSHIFT = r'<<' | 
 |   t_RSHIFT = r'>>' | 
 |  | 
 |   # A line ending '\n', we use this to increment the line number | 
 |   def t_LINE_END(self, t): | 
 |     r'\n+' | 
 |     self.AddLines(len(t.value)) | 
 |  | 
 |   # We do not process escapes in the IDL strings.  Strings are exclusively | 
 |   # used for attributes, and not used as typical 'C' constants. | 
 |   def t_STRING(self, t): | 
 |     r'"[^"]*"' | 
 |     t.value = t.value[1:-1] | 
 |     self.AddLines(t.value.count('\n')) | 
 |     return t | 
 |  | 
 |   # A C or C++ style comment:  /* xxx */ or // | 
 |   def t_COMMENT(self, t): | 
 |     r'(/\*(.|\n)*?\*/)|(//.*(\n[ \t]*//.*)*)' | 
 |     self.AddLines(t.value.count('\n')) | 
 |     return t | 
 |  | 
 |   # Return a "preprocessor" inline block | 
 |   def t_INLINE(self, t): | 
 |     r'\#inline (.|\n)*?\#endinl.*' | 
 |     self.AddLines(t.value.count('\n')) | 
 |     return t | 
 |  | 
 |   # A symbol or keyword. | 
 |   def t_KEYWORD_SYMBOL(self, t): | 
 |     r'_?[A-Za-z][A-Za-z_0-9]*' | 
 |  | 
 |     # All non-keywords are assumed to be symbols | 
 |     t.type = self.keywords.get(t.value, 'SYMBOL') | 
 |  | 
 |     # We strip leading underscores so that you can specify symbols with the same | 
 |     # value as a keywords (E.g. a dictionary named 'interface'). | 
 |     if t.value[0] == '_': | 
 |       t.value = t.value[1:] | 
 |     return t | 
 |  | 
 |   def t_ANY_error(self, t): | 
 |     msg = "Unrecognized input" | 
 |     line = self.lexobj.lineno | 
 |  | 
 |     # If that line has not been accounted for, then we must have hit | 
 |     # EoF, so compute the beginning of the line that caused the problem. | 
 |     if line >= len(self.index): | 
 |       # Find the offset in the line of the first word causing the issue | 
 |       word = t.value.split()[0] | 
 |       offs = self.lines[line - 1].find(word) | 
 |       # Add the computed line's starting position | 
 |       self.index.append(self.lexobj.lexpos - offs) | 
 |       msg = "Unexpected EoF reached after" | 
 |  | 
 |     pos = self.lexobj.lexpos - self.index[line] | 
 |     file = self.lexobj.filename | 
 |     out = self.ErrorMessage(file, line, pos, msg) | 
 |     sys.stderr.write(out + '\n') | 
 |     self.lex_errors += 1 | 
 |  | 
 |  | 
 |   def AddLines(self, count): | 
 |     # Set the lexer position for the beginning of the next line.  In the case | 
 |     # of multiple lines, tokens can not exist on any of the lines except the | 
 |     # last one, so the recorded value for previous lines are unused.  We still | 
 |     # fill the array however, to make sure the line count is correct. | 
 |     self.lexobj.lineno += count | 
 |     for i in range(count): | 
 |       self.index.append(self.lexobj.lexpos) | 
 |  | 
 |   def FileLineMsg(self, file, line, msg): | 
 |     if file:  return "%s(%d) : %s" % (file, line + 1, msg) | 
 |     return "<BuiltIn> : %s" % msg | 
 |  | 
 |   def SourceLine(self, file, line, pos): | 
 |     caret = '\t^'.expandtabs(pos) | 
 |     # We decrement the line number since the array is 0 based while the | 
 |     # line numbers are 1 based. | 
 |     return "%s\n%s" % (self.lines[line - 1], caret) | 
 |  | 
 |   def ErrorMessage(self, file, line, pos, msg): | 
 |     return "\n%s\n%s" % ( | 
 |         self.FileLineMsg(file, line, msg), | 
 |         self.SourceLine(file, line, pos)) | 
 |  | 
 |   def SetData(self, filename, data): | 
 |     # Start with line 1, not zero | 
 |     self.lexobj.lineno = 1 | 
 |     self.lexobj.filename = filename | 
 |     self.lines = data.split('\n') | 
 |     self.index = [0] | 
 |     self.lexobj.input(data) | 
 |     self.lex_errors = 0 | 
 |  | 
 |   def __init__(self): | 
 |     self.lexobj = lex.lex(object=self, lextab=None, optimize=0) | 
 |  | 
 |  | 
 |  | 
 | # | 
 | # FilesToTokens | 
 | # | 
 | # From a set of source file names, generate a list of tokens. | 
 | # | 
 | def FilesToTokens(filenames, verbose=False): | 
 |   lexer = IDLLexer() | 
 |   outlist = [] | 
 |   for filename in filenames: | 
 |     data = open(filename).read() | 
 |     lexer.SetData(filename, data) | 
 |     if verbose: sys.stdout.write('  Loaded %s...\n' % filename) | 
 |     while 1: | 
 |       t = lexer.lexobj.token() | 
 |       if t is None: break | 
 |       outlist.append(t) | 
 |   return outlist | 
 |  | 
 |  | 
 | def TokensFromText(text): | 
 |   lexer = IDLLexer() | 
 |   lexer.SetData('unknown', text) | 
 |   outlist = [] | 
 |   while 1: | 
 |     t = lexer.lexobj.token() | 
 |     if t is None: break | 
 |     outlist.append(t.value) | 
 |   return outlist | 
 |  | 
 | # | 
 | # TextToTokens | 
 | # | 
 | # From a block of text, generate a list of tokens | 
 | # | 
 | def TextToTokens(source): | 
 |   lexer = IDLLexer() | 
 |   outlist = [] | 
 |   lexer.SetData('AUTO', source) | 
 |   while 1: | 
 |     t = lexer.lexobj.token() | 
 |     if t is None: break | 
 |     outlist.append(t.value) | 
 |   return outlist | 
 |  | 
 |  | 
 | # | 
 | # TestSame | 
 | # | 
 | # From a set of token values, generate a new source text by joining with a | 
 | # single space.  The new source is then tokenized and compared against the | 
 | # old set. | 
 | # | 
 | def TestSame(values1): | 
 |   # Recreate the source from the tokens.  We use newline instead of whitespace | 
 |   # since the '//' and #inline regex are line sensitive. | 
 |   text = '\n'.join(values1) | 
 |   values2 = TextToTokens(text) | 
 |  | 
 |   count1 = len(values1) | 
 |   count2 = len(values2) | 
 |   if count1 != count2: | 
 |     print "Size mismatch original %d vs %d\n" % (count1, count2) | 
 |     if count1 > count2: count1 = count2 | 
 |  | 
 |   for i in range(count1): | 
 |     if values1[i] != values2[i]: | 
 |       print "%d >>%s<< >>%s<<" % (i, values1[i], values2[i]) | 
 |  | 
 |   if GetOption('output'): | 
 |     sys.stdout.write('Generating original.txt and tokenized.txt\n') | 
 |     open('original.txt', 'w').write(src1) | 
 |     open('tokenized.txt', 'w').write(src2) | 
 |  | 
 |   if values1 == values2: | 
 |     sys.stdout.write('Same: Pass\n') | 
 |     return 0 | 
 |  | 
 |   print "****************\n%s\n%s***************\n" % (src1, src2) | 
 |   sys.stdout.write('Same: Failed\n') | 
 |   return -1 | 
 |  | 
 |  | 
 | # | 
 | # TestExpect | 
 | # | 
 | # From a set of tokens pairs, verify the type field of the second matches | 
 | # the value of the first, so that: | 
 | # INT 123 FLOAT 1.1 | 
 | # will generate a passing test, where the first token is the SYMBOL INT, | 
 | # and the second token is the INT 123, third token is the SYMBOL FLOAT and | 
 | # the fourth is the FLOAT 1.1, etc... | 
 | def TestExpect(tokens): | 
 |   count = len(tokens) | 
 |   index = 0 | 
 |   errors = 0 | 
 |   while index < count: | 
 |     type = tokens[index].value | 
 |     token = tokens[index + 1] | 
 |     index += 2 | 
 |  | 
 |     if type != token.type: | 
 |       sys.stderr.write('Mismatch:  Expected %s, but got %s = %s.\n' % | 
 |                        (type, token.type, token.value)) | 
 |       errors += 1 | 
 |  | 
 |   if not errors: | 
 |     sys.stdout.write('Expect: Pass\n') | 
 |     return 0 | 
 |  | 
 |   sys.stdout.write('Expect: Failed\n') | 
 |   return -1 | 
 |  | 
 |  | 
 | def Main(args): | 
 |   filenames = ParseOptions(args) | 
 |  | 
 |   try: | 
 |     tokens = FilesToTokens(filenames, GetOption('verbose')) | 
 |     values = [tok.value for tok in tokens] | 
 |     if GetOption('output'): sys.stdout.write(' <> '.join(values) + '\n') | 
 |     if GetOption('test'): | 
 |       if TestSame(values): | 
 |         return -1 | 
 |       if TestExpect(tokens): | 
 |         return -1 | 
 |     return 0 | 
 |  | 
 |   except lex.LexError as le: | 
 |     sys.stderr.write('%s\n' % str(le)) | 
 |   return -1 | 
 |  | 
 |  | 
 | if __name__ == '__main__': | 
 |   sys.exit(Main(sys.argv[1:])) |