1 # The Detector determines which Monoglot or Polyglot should be
2 # used to parse a source file.
4 # The Detector primarily uses filename extensions to identify languages.
6 # The hash EXTENSION_MAP maps a filename extension to the name of a parser.
8 # If a filename extension is not enough to determine the correct parser (for
9 # instance, the *.m extension can indicate either a Matlab or Objective-C file),
10 # then the EXTENSION_MAP hash will contain a symbol identifying a Ruby method
11 # which will be invoked. This Ruby method can examine the file
12 # contents and return the name of the correct parser.
14 # Many source files do not have an extension. The method +disambiguate_nil+
15 # is called in these cases. The +file+ command line tool is used to determine
16 # the type of file and select a parser.
18 # The Detector is covered by DetectorTest.
20 class Ohcount::Detector
23 # A performance hack -- once we've checked for the presence of *.m files, the result
24 # is stored here to avoid checking twice.
25 attr_accessor :contains_m
28 # The primary entry point for the detector.
29 # Given a file context containing the file name, content, and an array of
30 # other filenames in the source tree, attempts to detect which
31 # language family (Monoglot or Polyglot) is in use for this file.
33 # Returns nil if the language is not recognized or if the file does not
38 # # List all C/C++ files in the 'src' directory
39 # Dir.entries("src").each do |file|
40 # context = Ohcount::SimpleFileContext.new(file)
41 # polyglot = Ohcount::Detector.detect(context)
42 # puts "#{file}" if polyglot == 'cncpp'
45 def self.detect(file_context)
46 # start with extension
47 polyglot = EXTENSION_MAP[File.extname(file_context.filename).downcase]
51 return polyglot if polyglot.is_a?(String)
53 # extension is ambiguous - requires custom disambiguation
54 self.send(polyglot, file_context)
56 return disambiguate_nil(file_context)
58 raise RuntimeError.new("Unknown file detection type")
62 # Based solely on the filename, makes a judgment whether a file is a binary format.
63 def self.binary_filename?(filename)
91 ignore.include?(File.extname(filename))
94 # If an extension maps to a string, that string must be the name of a glot.
95 # If an extension maps to a Ruby symbol, that symbol must be the name of a
96 # Ruby method which will return the name of a glot.
101 '.asm' => "assembler",
103 '.bas' => "visualbasic",
112 '.el' => "emacslisp",
115 '.cs' => :disambiguate_cs,
118 '.frx' => "visualbasic",
119 '.groovy'=> "groovy",
120 '.h' => :disambiguate_h_header,
128 '.inc' => :disambiguate_inc,
130 '.js' => "javascript",
135 '.m' => :matlab_or_objective_c,
136 '.mm' => "objective_c",
161 '.vb' => "visualbasic",
162 '.vba' => "visualbasic",
163 '.vbs' => "visualbasic",
171 # Returns a count of lines in the buffer matching the given regular expression.
172 def self.lines_matching(buffer, re)
173 buffer.inject(0) { |total, line| line =~ re ? total+1 : total }
176 # For *.m files, differentiates Matlab from Objective-C.
178 # This is done with a weighted heuristic that
179 # scans the *.m file contents for keywords,
180 # and also checks for the presence of matching *.h files.
181 def self.matlab_or_objective_c(file_context)
182 buffer = file_context.contents
184 # if there are .h files in same directory, this probably isn't matlab
186 h_headers = -0.5 if file_context.filenames.select { |a| a =~ /\.h$/ }.any?
188 # if the contents contain 'function (' on a single line - very likely to be matlab
189 # if the contents contain lines starting with '%', its probably matlab comments
190 matlab_signatures = /(^\s*function\s*)|(^\s*%)/
191 matlab_sig_score = 0.1 * lines_matching(buffer, matlab_signatures)
193 # if the contents contains '//' or '/*', likely objective_c
194 objective_c_signatures = /(^\s*\/\/\s*)|(^\s*\/\*)|(^[+-])/
195 obj_c_sig_score = -0.1 * lines_matching(buffer, objective_c_signatures)
197 matlab = h_headers + matlab_sig_score + obj_c_sig_score
199 matlab > 0 ? 'matlab' : 'objective_c'
202 # For *.h files, differentiates C/C++ from Objective-C.
204 # This is done with a weighted heuristic that
205 # scans the *.h file contents for Objective-C keywords,
206 # and also checks for the presence of matching *.m files.
207 def self.disambiguate_h_header(file_context)
208 buffer = file_context.contents
212 # could it be realistically be objective_c ? are there any .m files at all?
213 # Speed hack - remember our findings in case we get the same filenames over and over
214 unless defined?(file_context.filenames.contains_m)
215 file_context.filenames.extend(ContainsM)
216 file_context.filenames.contains_m = file_context.filenames.select { |a| a =~ /\.m$/ }.any?
218 return 'cncpp' unless file_context.filenames.contains_m
220 # if the dir contains a matching *.m file, likely objective_c
221 if file_context.filename =~ /\.h$/
222 m_counterpart = file_context.filename.gsub(/\.h$/, ".m")
223 return 'objective_c' if file_context.filenames.include?(m_counterpart)
226 # ok - it just might be objective_c, let's check contents for objective_c signatures
227 objective_c_signatures = /(^\s*@interface)|(^\s*@end)/
228 objective_c += lines_matching(buffer, objective_c_signatures)
230 return objective_c > 1 ? 'objective_c' : 'cncpp'
233 # Tests whether the provided buffer contains binary or text content.
234 # This is not fool-proof -- we basically just check for zero values
235 # in the early bytes of the buffer. If we find a zero, we know it
236 # is not (ascii) text.
237 def self.binary_buffer?(buffer)
239 return true if buffer[i] == 0
244 # True if the provided buffer includes a '?php' directive
245 def self.php_instruction?(buffer)
249 # For *.inc files, checks for a PHP class.
250 def self.disambiguate_inc(file_context)
251 buffer = file_context.contents
252 return nil if binary_buffer?(buffer)
253 return 'php' if php_instruction?(buffer)
257 # For files with extention *.cs, differentiates C# from Clearsilver.
258 def self.disambiguate_cs(file_context)
259 buffer = file_context.contents
260 return 'clearsilver_template' if lines_matching(file_context.contents, /\<\?cs/) > 0
264 # Attempts to determine the Polyglot for files that do not have a
265 # filename extension.
267 # Relies on the bash +file+ command line tool as its primary method.
269 # There must be a file at <tt>file_context.file_location</tt> for +file+
272 def self.disambiguate_nil(file_context)
273 file_location = file_context.file_location
274 output = `file -b #{ file_location }`
276 when /([\w\/]+) script text/, /script text executable for ([\w\/]+)/
278 if script =~ /\/(\w*)$/
281 known_languages = EXTENSION_MAP.values
282 return script.downcase if known_languages.include?(script.downcase)
283 when /([\w\-]*) shell script text/