| 1 | |
|---|
| 2 | ##################################################################### |
|---|
| 3 | #This software was developed by the University of Tennessee as part of the |
|---|
| 4 | #Distributed Data Analysis of Neutron Scattering Experiments (DANSE) |
|---|
| 5 | #project funded by the US National Science Foundation. |
|---|
| 6 | #See the license text in license.txt |
|---|
| 7 | #copyright 2008, University of Tennessee |
|---|
| 8 | ###################################################################### |
|---|
| 9 | |
|---|
| 10 | """ |
|---|
| 11 | File handler to support different file extensions. |
|---|
| 12 | Uses reflectometry's registry utility. |
|---|
| 13 | |
|---|
| 14 | The default readers are found in the 'readers' sub-module |
|---|
| 15 | and registered by default at initialization time. |
|---|
| 16 | |
|---|
| 17 | To add a new default reader, one must register it in |
|---|
| 18 | the register_readers method found in readers/__init__.py. |
|---|
| 19 | |
|---|
| 20 | A utility method (find_plugins) is available to inspect |
|---|
| 21 | a directory (for instance, a user plug-in directory) and |
|---|
| 22 | look for new readers/writers. |
|---|
| 23 | """ |
|---|
| 24 | |
|---|
| 25 | import os |
|---|
| 26 | import sys |
|---|
| 27 | import logging |
|---|
| 28 | import time |
|---|
| 29 | from zipfile import ZipFile |
|---|
| 30 | from data_util.registry import ExtensionRegistry |
|---|
| 31 | # Default readers are defined in the readers sub-module |
|---|
| 32 | import readers |
|---|
| 33 | from readers import ascii_reader |
|---|
| 34 | from readers import cansas_reader |
|---|
| 35 | |
|---|
| 36 | class Registry(ExtensionRegistry): |
|---|
| 37 | """ |
|---|
| 38 | Registry class for file format extensions. |
|---|
| 39 | Readers and writers are supported. |
|---|
| 40 | """ |
|---|
| 41 | |
|---|
| 42 | def __init__(self): |
|---|
| 43 | super(Registry, self).__init__() |
|---|
| 44 | |
|---|
| 45 | ## Writers |
|---|
| 46 | self.writers = {} |
|---|
| 47 | |
|---|
| 48 | ## List of wildcards |
|---|
| 49 | self.wildcards = ['All (*.*)|*.*'] |
|---|
| 50 | |
|---|
| 51 | ## Creation time, for testing |
|---|
| 52 | self._created = time.time() |
|---|
| 53 | |
|---|
| 54 | # Register default readers |
|---|
| 55 | readers.read_associations(self) |
|---|
| 56 | |
|---|
| 57 | #TODO: remove the following line when ready to switch to |
|---|
| 58 | #the new default readers |
|---|
| 59 | #readers.register_readers(self._identify_plugin) |
|---|
| 60 | |
|---|
| 61 | # Look for plug-in readers |
|---|
| 62 | #self.find_plugins('plugins') |
|---|
| 63 | |
|---|
| 64 | |
|---|
| 65 | def load(self, path, format=None): |
|---|
| 66 | """ |
|---|
| 67 | Call the loader for the file type of path. |
|---|
| 68 | |
|---|
| 69 | :param path: file path |
|---|
| 70 | :param format: explicit extension, to force the use |
|---|
| 71 | of a particular reader |
|---|
| 72 | |
|---|
| 73 | Defaults to the ascii (multi-column) reader |
|---|
| 74 | if no reader was registered for the file's |
|---|
| 75 | extension. |
|---|
| 76 | """ |
|---|
| 77 | try: |
|---|
| 78 | return super(Registry, self).load(path, format=format) |
|---|
| 79 | except: |
|---|
| 80 | try: |
|---|
| 81 | # No reader was found. Default to the ascii reader. |
|---|
| 82 | ascii_loader = ascii_reader.Reader() |
|---|
| 83 | return ascii_loader.read(path) |
|---|
| 84 | except: |
|---|
| 85 | cansas_loader = cansas_reader.Reader() |
|---|
| 86 | return cansas_loader.read(path) |
|---|
| 87 | |
|---|
| 88 | def find_plugins(self, dir): |
|---|
| 89 | """ |
|---|
| 90 | Find readers in a given directory. This method |
|---|
| 91 | can be used to inspect user plug-in directories to |
|---|
| 92 | find new readers/writers. |
|---|
| 93 | |
|---|
| 94 | :param dir: directory to search into |
|---|
| 95 | |
|---|
| 96 | :return: number of readers found |
|---|
| 97 | """ |
|---|
| 98 | readers_found = 0 |
|---|
| 99 | temp_path = os.path.abspath(dir) |
|---|
| 100 | if not os.path.isdir(temp_path): |
|---|
| 101 | temp_path = os.path.join(os.getcwd(), dir) |
|---|
| 102 | if not os.path.isdir(temp_path): |
|---|
| 103 | temp_path = os.path.join(os.path.dirname(__file__), dir) |
|---|
| 104 | if not os.path.isdir(temp_path): |
|---|
| 105 | temp_path = os.path.join(os.path.dirname(sys.path[0]), dir) |
|---|
| 106 | |
|---|
| 107 | dir = temp_path |
|---|
| 108 | # Check whether the directory exists |
|---|
| 109 | if not os.path.isdir(dir): |
|---|
| 110 | msg = "DataLoader couldn't locate DataLoader plugin folder." |
|---|
| 111 | msg += """ "%s" does not exist""" % dir |
|---|
| 112 | logging.warning(msg) |
|---|
| 113 | return readers_found |
|---|
| 114 | |
|---|
| 115 | for item in os.listdir(dir): |
|---|
| 116 | full_path = os.path.join(dir, item) |
|---|
| 117 | if os.path.isfile(full_path): |
|---|
| 118 | |
|---|
| 119 | # Process python files |
|---|
| 120 | if item.endswith('.py'): |
|---|
| 121 | toks = os.path.splitext(os.path.basename(item)) |
|---|
| 122 | try: |
|---|
| 123 | sys.path.insert(0, os.path.abspath(dir)) |
|---|
| 124 | module = __import__(toks[0], globals(), locals()) |
|---|
| 125 | if self._identify_plugin(module): |
|---|
| 126 | readers_found += 1 |
|---|
| 127 | except: |
|---|
| 128 | msg = "Loader: Error importing " |
|---|
| 129 | msg += "%s\n %s" % (item, sys.exc_value) |
|---|
| 130 | logging.error(msg) |
|---|
| 131 | |
|---|
| 132 | # Process zip files |
|---|
| 133 | elif item.endswith('.zip'): |
|---|
| 134 | try: |
|---|
| 135 | # Find the modules in the zip file |
|---|
| 136 | zfile = ZipFile(item) |
|---|
| 137 | nlist = zfile.namelist() |
|---|
| 138 | |
|---|
| 139 | sys.path.insert(0, item) |
|---|
| 140 | for mfile in nlist: |
|---|
| 141 | try: |
|---|
| 142 | # Change OS path to python path |
|---|
| 143 | fullname = mfile.replace('/', '.') |
|---|
| 144 | fullname = os.path.splitext(fullname)[0] |
|---|
| 145 | module = __import__(fullname, globals(), |
|---|
| 146 | locals(), [""]) |
|---|
| 147 | if self._identify_plugin(module): |
|---|
| 148 | readers_found += 1 |
|---|
| 149 | except: |
|---|
| 150 | msg = "Loader: Error importing" |
|---|
| 151 | msg += " %s\n %s" % (mfile, sys.exc_value) |
|---|
| 152 | logging.error(msg) |
|---|
| 153 | |
|---|
| 154 | except: |
|---|
| 155 | msg = "Loader: Error importing " |
|---|
| 156 | msg += " %s\n %s" % (item, sys.exc_value) |
|---|
| 157 | logging.error(msg) |
|---|
| 158 | |
|---|
| 159 | return readers_found |
|---|
| 160 | |
|---|
| 161 | def associate_file_type(self, ext, module): |
|---|
| 162 | """ |
|---|
| 163 | Look into a module to find whether it contains a |
|---|
| 164 | Reader class. If so, APPEND it to readers and (potentially) |
|---|
| 165 | to the list of writers for the given extension |
|---|
| 166 | |
|---|
| 167 | :param ext: file extension [string] |
|---|
| 168 | :param module: module object |
|---|
| 169 | """ |
|---|
| 170 | reader_found = False |
|---|
| 171 | |
|---|
| 172 | if hasattr(module, "Reader"): |
|---|
| 173 | try: |
|---|
| 174 | # Find supported extensions |
|---|
| 175 | loader = module.Reader() |
|---|
| 176 | if ext not in self.loaders: |
|---|
| 177 | self.loaders[ext] = [] |
|---|
| 178 | # Append the new reader to the list |
|---|
| 179 | self.loaders[ext].append(loader.read) |
|---|
| 180 | |
|---|
| 181 | reader_found = True |
|---|
| 182 | |
|---|
| 183 | # Keep track of wildcards |
|---|
| 184 | type_name = module.__name__ |
|---|
| 185 | if hasattr(loader, 'type_name'): |
|---|
| 186 | type_name = loader.type_name |
|---|
| 187 | |
|---|
| 188 | wcard = "%s files (*%s)|*%s" % (type_name, ext.lower(), |
|---|
| 189 | ext.lower()) |
|---|
| 190 | if wcard not in self.wildcards: |
|---|
| 191 | self.wildcards.append(wcard) |
|---|
| 192 | |
|---|
| 193 | # Check whether writing is supported |
|---|
| 194 | if hasattr(loader, 'write'): |
|---|
| 195 | if ext not in self.writers: |
|---|
| 196 | self.writers[ext] = [] |
|---|
| 197 | # Append the new writer to the list |
|---|
| 198 | self.writers[ext].append(loader.write) |
|---|
| 199 | |
|---|
| 200 | except: |
|---|
| 201 | msg = "Loader: Error accessing" |
|---|
| 202 | msg += " Reader in %s\n %s" % (module.__name__, sys.exc_value) |
|---|
| 203 | logging.error(msg) |
|---|
| 204 | return reader_found |
|---|
| 205 | |
|---|
| 206 | def associate_file_reader(self, ext, loader): |
|---|
| 207 | """ |
|---|
| 208 | Append a reader object to readers |
|---|
| 209 | |
|---|
| 210 | :param ext: file extension [string] |
|---|
| 211 | :param module: reader object |
|---|
| 212 | """ |
|---|
| 213 | reader_found = False |
|---|
| 214 | |
|---|
| 215 | try: |
|---|
| 216 | # Find supported extensions |
|---|
| 217 | if ext not in self.loaders: |
|---|
| 218 | self.loaders[ext] = [] |
|---|
| 219 | # Append the new reader to the list |
|---|
| 220 | self.loaders[ext].append(loader.read) |
|---|
| 221 | |
|---|
| 222 | reader_found = True |
|---|
| 223 | |
|---|
| 224 | # Keep track of wildcards |
|---|
| 225 | if hasattr(loader, 'type_name'): |
|---|
| 226 | type_name = loader.type_name |
|---|
| 227 | |
|---|
| 228 | wcard = "%s files (*%s)|*%s" % (type_name, ext.lower(), |
|---|
| 229 | ext.lower()) |
|---|
| 230 | if wcard not in self.wildcards: |
|---|
| 231 | self.wildcards.append(wcard) |
|---|
| 232 | |
|---|
| 233 | except: |
|---|
| 234 | msg = "Loader: Error accessing Reader " |
|---|
| 235 | msg += "in %s\n %s" % (module.__name__, sys.exc_value) |
|---|
| 236 | logging.error(msg) |
|---|
| 237 | return reader_found |
|---|
| 238 | |
|---|
| 239 | |
|---|
| 240 | def _identify_plugin(self, module): |
|---|
| 241 | """ |
|---|
| 242 | Look into a module to find whether it contains a |
|---|
| 243 | Reader class. If so, add it to readers and (potentially) |
|---|
| 244 | to the list of writers. |
|---|
| 245 | :param module: module object |
|---|
| 246 | |
|---|
| 247 | """ |
|---|
| 248 | reader_found = False |
|---|
| 249 | |
|---|
| 250 | if hasattr(module, "Reader"): |
|---|
| 251 | try: |
|---|
| 252 | # Find supported extensions |
|---|
| 253 | loader = module.Reader() |
|---|
| 254 | for ext in loader.ext: |
|---|
| 255 | if ext not in self.loaders: |
|---|
| 256 | self.loaders[ext] = [] |
|---|
| 257 | # When finding a reader at run time, |
|---|
| 258 | # treat this reader as the new default |
|---|
| 259 | self.loaders[ext].insert(0, loader.read) |
|---|
| 260 | |
|---|
| 261 | reader_found = True |
|---|
| 262 | |
|---|
| 263 | # Keep track of wildcards |
|---|
| 264 | type_name = module.__name__ |
|---|
| 265 | if hasattr(loader, 'type_name'): |
|---|
| 266 | type_name = loader.type_name |
|---|
| 267 | wcard = "%s files (*%s)|*%s" % (type_name, ext.lower(), |
|---|
| 268 | ext.lower()) |
|---|
| 269 | if wcard not in self.wildcards: |
|---|
| 270 | self.wildcards.append(wcard) |
|---|
| 271 | |
|---|
| 272 | # Check whether writing is supported |
|---|
| 273 | if hasattr(loader, 'write'): |
|---|
| 274 | for ext in loader.ext: |
|---|
| 275 | if ext not in self.writers: |
|---|
| 276 | self.writers[ext] = [] |
|---|
| 277 | self.writers[ext].insert(0,loader.write) |
|---|
| 278 | |
|---|
| 279 | except: |
|---|
| 280 | msg = "Loader: Error accessing Reader" |
|---|
| 281 | msg += " in %s\n %s" % (module.__name__, sys.exc_value) |
|---|
| 282 | logging.error(msg) |
|---|
| 283 | return reader_found |
|---|
| 284 | |
|---|
| 285 | def lookup_writers(self, path): |
|---|
| 286 | """ |
|---|
| 287 | :return: the loader associated with the file type of path. |
|---|
| 288 | |
|---|
| 289 | :Raises ValueError: if file type is not known. |
|---|
| 290 | """ |
|---|
| 291 | # Find matching extensions |
|---|
| 292 | extlist = [ext for ext in self.extensions() if path.endswith(ext)] |
|---|
| 293 | # Sort matching extensions by decreasing order of length |
|---|
| 294 | extlist.sort(lambda a, b: len(a) < len(b)) |
|---|
| 295 | # Combine loaders for matching extensions into one big list |
|---|
| 296 | writers = [] |
|---|
| 297 | for L in [self.writers[ext] for ext in extlist]: |
|---|
| 298 | writers.extend(L) |
|---|
| 299 | # Remove duplicates if they exist |
|---|
| 300 | if len(writers) != len(set(writers)): |
|---|
| 301 | result = [] |
|---|
| 302 | for L in writers: |
|---|
| 303 | if L not in result: result.append(L) |
|---|
| 304 | writers = L |
|---|
| 305 | # Raise an error if there are no matching extensions |
|---|
| 306 | if len(writers) == 0: |
|---|
| 307 | raise ValueError, "Unknown file type for " + path |
|---|
| 308 | # All done |
|---|
| 309 | return writers |
|---|
| 310 | |
|---|
| 311 | def save(self, path, data, format=None): |
|---|
| 312 | """ |
|---|
| 313 | Call the writer for the file type of path. |
|---|
| 314 | |
|---|
| 315 | Raises ValueError if no writer is available. |
|---|
| 316 | Raises KeyError if format is not available. |
|---|
| 317 | |
|---|
| 318 | May raise a writer-defined exception if writer fails. |
|---|
| 319 | """ |
|---|
| 320 | if format is None: |
|---|
| 321 | writers = self.lookup_writers(path) |
|---|
| 322 | else: |
|---|
| 323 | writers = self.writers[format] |
|---|
| 324 | for fn in writers: |
|---|
| 325 | try: |
|---|
| 326 | return fn(path, data) |
|---|
| 327 | except: |
|---|
| 328 | pass # give other loaders a chance to succeed |
|---|
| 329 | # If we get here it is because all loaders failed |
|---|
| 330 | raise # reraises last exception |
|---|
| 331 | |
|---|
| 332 | |
|---|
| 333 | class Loader(object): |
|---|
| 334 | """ |
|---|
| 335 | Utility class to use the Registry as a singleton. |
|---|
| 336 | """ |
|---|
| 337 | ## Registry instance |
|---|
| 338 | __registry = Registry() |
|---|
| 339 | |
|---|
| 340 | def associate_file_type(self, ext, module): |
|---|
| 341 | """ |
|---|
| 342 | Look into a module to find whether it contains a |
|---|
| 343 | Reader class. If so, append it to readers and (potentially) |
|---|
| 344 | to the list of writers for the given extension |
|---|
| 345 | |
|---|
| 346 | :param ext: file extension [string] |
|---|
| 347 | :param module: module object |
|---|
| 348 | """ |
|---|
| 349 | return self.__registry.associate_file_type(ext, module) |
|---|
| 350 | |
|---|
| 351 | def associate_file_reader(self, ext, loader): |
|---|
| 352 | """ |
|---|
| 353 | Append a reader object to readers |
|---|
| 354 | |
|---|
| 355 | :param ext: file extension [string] |
|---|
| 356 | :param module: reader object |
|---|
| 357 | """ |
|---|
| 358 | return self.__registry.associate_file_reader(ext, loader) |
|---|
| 359 | |
|---|
| 360 | def load(self, file, format=None): |
|---|
| 361 | """ |
|---|
| 362 | Load a file |
|---|
| 363 | |
|---|
| 364 | :param file: file name (path) |
|---|
| 365 | :param format: specified format to use (optional) |
|---|
| 366 | :return: DataInfo object |
|---|
| 367 | """ |
|---|
| 368 | return self.__registry.load(file, format) |
|---|
| 369 | |
|---|
| 370 | def save(self, file, data, format): |
|---|
| 371 | """ |
|---|
| 372 | Save a DataInfo object to file |
|---|
| 373 | :param file: file name (path) |
|---|
| 374 | :param data: DataInfo object |
|---|
| 375 | :param format: format to write the data in |
|---|
| 376 | """ |
|---|
| 377 | return self.__registry.save(file, data, format) |
|---|
| 378 | |
|---|
| 379 | def _get_registry_creation_time(self): |
|---|
| 380 | """ |
|---|
| 381 | Internal method used to test the uniqueness |
|---|
| 382 | of the registry object |
|---|
| 383 | """ |
|---|
| 384 | return self.__registry._created |
|---|
| 385 | |
|---|
| 386 | def find_plugins(self, dir): |
|---|
| 387 | """ |
|---|
| 388 | Find plugins in a given directory |
|---|
| 389 | |
|---|
| 390 | :param dir: directory to look into to find new readers/writers |
|---|
| 391 | """ |
|---|
| 392 | return self.__registry.find_plugins(dir) |
|---|
| 393 | |
|---|
| 394 | def get_wildcards(self): |
|---|
| 395 | return self.__registry.wildcards |
|---|
| 396 | |
|---|
| 397 | if __name__ == "__main__": |
|---|
| 398 | logging.basicConfig(level=logging.INFO, |
|---|
| 399 | format='%(asctime)s %(levelname)s %(message)s', |
|---|
| 400 | filename='loader.log', |
|---|
| 401 | filemode='w') |
|---|
| 402 | l = Loader() |
|---|
| 403 | test_data = l.load('test/cansas1d.xml') |
|---|
| 404 | l.save('test_file.xml', test_data, '.xml') |
|---|
| 405 | |
|---|
| 406 | print l.get_wildcards() |
|---|
| 407 | |
|---|
| 408 | |
|---|
| 409 | |
|---|
