[4749514] | 1 | """ |
---|
| 2 | File handler to support different file extensions. |
---|
[270c882b] | 3 | Uses reflectometer registry utility. |
---|
[f06d7fc] | 4 | |
---|
[4749514] | 5 | The default readers are found in the 'readers' sub-module |
---|
| 6 | and registered by default at initialization time. |
---|
[f06d7fc] | 7 | |
---|
[4749514] | 8 | To add a new default reader, one must register it in |
---|
| 9 | the register_readers method found in readers/__init__.py. |
---|
[f06d7fc] | 10 | |
---|
[4749514] | 11 | A utility method (find_plugins) is available to inspect |
---|
| 12 | a directory (for instance, a user plug-in directory) and |
---|
| 13 | look for new readers/writers. |
---|
| 14 | """ |
---|
| 15 | ##################################################################### |
---|
| 16 | #This software was developed by the University of Tennessee as part of the |
---|
| 17 | #Distributed Data Analysis of Neutron Scattering Experiments (DANSE) |
---|
| 18 | #project funded by the US National Science Foundation. |
---|
| 19 | #See the license text in license.txt |
---|
| 20 | #copyright 2008, University of Tennessee |
---|
| 21 | ###################################################################### |
---|
| 22 | |
---|
| 23 | import os |
---|
| 24 | import sys |
---|
| 25 | import logging |
---|
| 26 | import time |
---|
| 27 | from zipfile import ZipFile |
---|
[b699768] | 28 | from sas.sascalc.data_util.registry import ExtensionRegistry |
---|
[4749514] | 29 | # Default readers are defined in the readers sub-module |
---|
| 30 | import readers |
---|
[270c882b] | 31 | from loader_exceptions import NoKnownLoaderException, FileContentsException |
---|
[4749514] | 32 | from readers import ascii_reader |
---|
| 33 | from readers import cansas_reader |
---|
[d0764bf] | 34 | from readers import cansas_reader_HDF5 |
---|
[4749514] | 35 | |
---|
[270c882b] | 36 | |
---|
[463e7ffc] | 37 | logger = logging.getLogger(__name__) |
---|
[c155a16] | 38 | |
---|
[4749514] | 39 | class Registry(ExtensionRegistry): |
---|
| 40 | """ |
---|
| 41 | Registry class for file format extensions. |
---|
| 42 | Readers and writers are supported. |
---|
| 43 | """ |
---|
| 44 | def __init__(self): |
---|
| 45 | super(Registry, self).__init__() |
---|
[f06d7fc] | 46 | |
---|
[270c882b] | 47 | # Writers |
---|
[4749514] | 48 | self.writers = {} |
---|
[f06d7fc] | 49 | |
---|
[270c882b] | 50 | # List of wildcards |
---|
[4749514] | 51 | self.wildcards = ['All (*.*)|*.*'] |
---|
[f06d7fc] | 52 | |
---|
[270c882b] | 53 | # Creation time, for testing |
---|
[4749514] | 54 | self._created = time.time() |
---|
[f06d7fc] | 55 | |
---|
[4749514] | 56 | # Register default readers |
---|
| 57 | readers.read_associations(self) |
---|
| 58 | |
---|
| 59 | def load(self, path, format=None): |
---|
| 60 | """ |
---|
| 61 | Call the loader for the file type of path. |
---|
| 62 | |
---|
| 63 | :param path: file path |
---|
| 64 | :param format: explicit extension, to force the use |
---|
| 65 | of a particular reader |
---|
| 66 | |
---|
[b9b612a] | 67 | Defaults to the ascii (multi-column), cansas XML, and cansas NeXuS |
---|
| 68 | readers if no reader was registered for the file's extension. |
---|
[4749514] | 69 | """ |
---|
| 70 | try: |
---|
| 71 | return super(Registry, self).load(path, format=format) |
---|
[270c882b] | 72 | except NoKnownLoaderException as e: |
---|
| 73 | pass # try the ASCII reader |
---|
[b9b612a] | 74 | try: |
---|
| 75 | ascii_loader = ascii_reader.Reader() |
---|
| 76 | return ascii_loader.read(path) |
---|
[270c882b] | 77 | except FileContentsException: |
---|
| 78 | pass # try the cansas XML reader |
---|
[b9b612a] | 79 | try: |
---|
| 80 | cansas_loader = cansas_reader.Reader() |
---|
| 81 | return cansas_loader.read(path) |
---|
[270c882b] | 82 | except FileContentsException: |
---|
| 83 | pass # try the cansas NeXuS reader |
---|
[b9b612a] | 84 | try: |
---|
| 85 | cansas_nexus_loader = cansas_reader_HDF5.Reader() |
---|
| 86 | return cansas_nexus_loader.read(path) |
---|
[270c882b] | 87 | except FileContentsException: |
---|
[7f75a3f] | 88 | logging.errors("No default loader can load the data") |
---|
[b9b612a] | 89 | # No known reader available. Give up and throw an error |
---|
| 90 | msg = "\n\tUnknown data format: %s.\n\tThe file is not a " % path |
---|
[7f75a3f] | 91 | msg += "known format that can be loaded by SasView.\n" |
---|
| 92 | msg += "Traceback:\n%s" % e.message |
---|
| 93 | raise NoKnownLoaderException, msg |
---|
[f06d7fc] | 94 | |
---|
[4749514] | 95 | def find_plugins(self, dir): |
---|
| 96 | """ |
---|
| 97 | Find readers in a given directory. This method |
---|
| 98 | can be used to inspect user plug-in directories to |
---|
| 99 | find new readers/writers. |
---|
[f06d7fc] | 100 | |
---|
[4749514] | 101 | :param dir: directory to search into |
---|
| 102 | :return: number of readers found |
---|
| 103 | """ |
---|
| 104 | readers_found = 0 |
---|
| 105 | temp_path = os.path.abspath(dir) |
---|
| 106 | if not os.path.isdir(temp_path): |
---|
| 107 | temp_path = os.path.join(os.getcwd(), dir) |
---|
| 108 | if not os.path.isdir(temp_path): |
---|
| 109 | temp_path = os.path.join(os.path.dirname(__file__), dir) |
---|
| 110 | if not os.path.isdir(temp_path): |
---|
| 111 | temp_path = os.path.join(os.path.dirname(sys.path[0]), dir) |
---|
[f06d7fc] | 112 | |
---|
[4749514] | 113 | dir = temp_path |
---|
| 114 | # Check whether the directory exists |
---|
| 115 | if not os.path.isdir(dir): |
---|
| 116 | msg = "DataLoader couldn't locate DataLoader plugin folder." |
---|
| 117 | msg += """ "%s" does not exist""" % dir |
---|
[c155a16] | 118 | logger.warning(msg) |
---|
[4749514] | 119 | return readers_found |
---|
[f06d7fc] | 120 | |
---|
[4749514] | 121 | for item in os.listdir(dir): |
---|
| 122 | full_path = os.path.join(dir, item) |
---|
| 123 | if os.path.isfile(full_path): |
---|
[f06d7fc] | 124 | |
---|
[4749514] | 125 | # Process python files |
---|
| 126 | if item.endswith('.py'): |
---|
| 127 | toks = os.path.splitext(os.path.basename(item)) |
---|
| 128 | try: |
---|
| 129 | sys.path.insert(0, os.path.abspath(dir)) |
---|
| 130 | module = __import__(toks[0], globals(), locals()) |
---|
| 131 | if self._identify_plugin(module): |
---|
| 132 | readers_found += 1 |
---|
| 133 | except: |
---|
| 134 | msg = "Loader: Error importing " |
---|
| 135 | msg += "%s\n %s" % (item, sys.exc_value) |
---|
[c155a16] | 136 | logger.error(msg) |
---|
[f06d7fc] | 137 | |
---|
[4749514] | 138 | # Process zip files |
---|
| 139 | elif item.endswith('.zip'): |
---|
| 140 | try: |
---|
| 141 | # Find the modules in the zip file |
---|
| 142 | zfile = ZipFile(item) |
---|
| 143 | nlist = zfile.namelist() |
---|
[f06d7fc] | 144 | |
---|
[4749514] | 145 | sys.path.insert(0, item) |
---|
| 146 | for mfile in nlist: |
---|
| 147 | try: |
---|
| 148 | # Change OS path to python path |
---|
| 149 | fullname = mfile.replace('/', '.') |
---|
| 150 | fullname = os.path.splitext(fullname)[0] |
---|
| 151 | module = __import__(fullname, globals(), |
---|
[f06d7fc] | 152 | locals(), [""]) |
---|
[4749514] | 153 | if self._identify_plugin(module): |
---|
| 154 | readers_found += 1 |
---|
| 155 | except: |
---|
| 156 | msg = "Loader: Error importing" |
---|
| 157 | msg += " %s\n %s" % (mfile, sys.exc_value) |
---|
[c155a16] | 158 | logger.error(msg) |
---|
[f06d7fc] | 159 | |
---|
[4749514] | 160 | except: |
---|
| 161 | msg = "Loader: Error importing " |
---|
| 162 | msg += " %s\n %s" % (item, sys.exc_value) |
---|
[c155a16] | 163 | logger.error(msg) |
---|
[f06d7fc] | 164 | |
---|
[4749514] | 165 | return readers_found |
---|
[f06d7fc] | 166 | |
---|
[4749514] | 167 | def associate_file_type(self, ext, module): |
---|
| 168 | """ |
---|
| 169 | Look into a module to find whether it contains a |
---|
| 170 | Reader class. If so, APPEND it to readers and (potentially) |
---|
| 171 | to the list of writers for the given extension |
---|
[f06d7fc] | 172 | |
---|
[4749514] | 173 | :param ext: file extension [string] |
---|
| 174 | :param module: module object |
---|
| 175 | """ |
---|
| 176 | reader_found = False |
---|
[f06d7fc] | 177 | |
---|
[4749514] | 178 | if hasattr(module, "Reader"): |
---|
| 179 | try: |
---|
| 180 | # Find supported extensions |
---|
| 181 | loader = module.Reader() |
---|
| 182 | if ext not in self.loaders: |
---|
| 183 | self.loaders[ext] = [] |
---|
| 184 | # Append the new reader to the list |
---|
| 185 | self.loaders[ext].append(loader.read) |
---|
| 186 | |
---|
| 187 | reader_found = True |
---|
[f06d7fc] | 188 | |
---|
[4749514] | 189 | # Keep track of wildcards |
---|
| 190 | type_name = module.__name__ |
---|
| 191 | if hasattr(loader, 'type_name'): |
---|
| 192 | type_name = loader.type_name |
---|
[f06d7fc] | 193 | |
---|
[4749514] | 194 | wcard = "%s files (*%s)|*%s" % (type_name, ext.lower(), |
---|
[f06d7fc] | 195 | ext.lower()) |
---|
[4749514] | 196 | if wcard not in self.wildcards: |
---|
| 197 | self.wildcards.append(wcard) |
---|
[f06d7fc] | 198 | |
---|
[4749514] | 199 | # Check whether writing is supported |
---|
| 200 | if hasattr(loader, 'write'): |
---|
| 201 | if ext not in self.writers: |
---|
| 202 | self.writers[ext] = [] |
---|
| 203 | # Append the new writer to the list |
---|
| 204 | self.writers[ext].append(loader.write) |
---|
[f06d7fc] | 205 | |
---|
[4749514] | 206 | except: |
---|
| 207 | msg = "Loader: Error accessing" |
---|
| 208 | msg += " Reader in %s\n %s" % (module.__name__, sys.exc_value) |
---|
[c155a16] | 209 | logger.error(msg) |
---|
[4749514] | 210 | return reader_found |
---|
| 211 | |
---|
| 212 | def associate_file_reader(self, ext, loader): |
---|
| 213 | """ |
---|
| 214 | Append a reader object to readers |
---|
[f06d7fc] | 215 | |
---|
[4749514] | 216 | :param ext: file extension [string] |
---|
| 217 | :param module: reader object |
---|
| 218 | """ |
---|
| 219 | reader_found = False |
---|
[f06d7fc] | 220 | |
---|
[4749514] | 221 | try: |
---|
| 222 | # Find supported extensions |
---|
| 223 | if ext not in self.loaders: |
---|
| 224 | self.loaders[ext] = [] |
---|
| 225 | # Append the new reader to the list |
---|
| 226 | self.loaders[ext].append(loader.read) |
---|
| 227 | |
---|
| 228 | reader_found = True |
---|
[f06d7fc] | 229 | |
---|
[4749514] | 230 | # Keep track of wildcards |
---|
| 231 | if hasattr(loader, 'type_name'): |
---|
| 232 | type_name = loader.type_name |
---|
[f06d7fc] | 233 | |
---|
[4749514] | 234 | wcard = "%s files (*%s)|*%s" % (type_name, ext.lower(), |
---|
| 235 | ext.lower()) |
---|
| 236 | if wcard not in self.wildcards: |
---|
| 237 | self.wildcards.append(wcard) |
---|
[f06d7fc] | 238 | |
---|
[4749514] | 239 | except: |
---|
| 240 | msg = "Loader: Error accessing Reader " |
---|
| 241 | msg += "in %s\n %s" % (loader.__name__, sys.exc_value) |
---|
[c155a16] | 242 | logger.error(msg) |
---|
[4749514] | 243 | return reader_found |
---|
| 244 | |
---|
| 245 | def _identify_plugin(self, module): |
---|
| 246 | """ |
---|
[f06d7fc] | 247 | Look into a module to find whether it contains a |
---|
[4749514] | 248 | Reader class. If so, add it to readers and (potentially) |
---|
| 249 | to the list of writers. |
---|
| 250 | :param module: module object |
---|
[f06d7fc] | 251 | |
---|
[4749514] | 252 | """ |
---|
| 253 | reader_found = False |
---|
[f06d7fc] | 254 | |
---|
[4749514] | 255 | if hasattr(module, "Reader"): |
---|
| 256 | try: |
---|
| 257 | # Find supported extensions |
---|
| 258 | loader = module.Reader() |
---|
| 259 | for ext in loader.ext: |
---|
| 260 | if ext not in self.loaders: |
---|
| 261 | self.loaders[ext] = [] |
---|
| 262 | # When finding a reader at run time, |
---|
| 263 | # treat this reader as the new default |
---|
| 264 | self.loaders[ext].insert(0, loader.read) |
---|
| 265 | |
---|
| 266 | reader_found = True |
---|
[f06d7fc] | 267 | |
---|
[4749514] | 268 | # Keep track of wildcards |
---|
| 269 | type_name = module.__name__ |
---|
| 270 | if hasattr(loader, 'type_name'): |
---|
| 271 | type_name = loader.type_name |
---|
| 272 | wcard = "%s files (*%s)|*%s" % (type_name, ext.lower(), |
---|
[f06d7fc] | 273 | ext.lower()) |
---|
[4749514] | 274 | if wcard not in self.wildcards: |
---|
| 275 | self.wildcards.append(wcard) |
---|
[f06d7fc] | 276 | |
---|
[4749514] | 277 | # Check whether writing is supported |
---|
| 278 | if hasattr(loader, 'write'): |
---|
| 279 | for ext in loader.ext: |
---|
| 280 | if ext not in self.writers: |
---|
| 281 | self.writers[ext] = [] |
---|
| 282 | self.writers[ext].insert(0, loader.write) |
---|
[f06d7fc] | 283 | |
---|
[4749514] | 284 | except: |
---|
| 285 | msg = "Loader: Error accessing Reader" |
---|
| 286 | msg += " in %s\n %s" % (module.__name__, sys.exc_value) |
---|
[c155a16] | 287 | logger.error(msg) |
---|
[4749514] | 288 | return reader_found |
---|
| 289 | |
---|
| 290 | def lookup_writers(self, path): |
---|
| 291 | """ |
---|
| 292 | :return: the loader associated with the file type of path. |
---|
| 293 | :Raises ValueError: if file type is not known. |
---|
| 294 | """ |
---|
| 295 | # Find matching extensions |
---|
| 296 | extlist = [ext for ext in self.extensions() if path.endswith(ext)] |
---|
| 297 | # Sort matching extensions by decreasing order of length |
---|
| 298 | extlist.sort(lambda a, b: len(a) < len(b)) |
---|
| 299 | # Combine loaders for matching extensions into one big list |
---|
| 300 | writers = [] |
---|
| 301 | for L in [self.writers[ext] for ext in extlist]: |
---|
| 302 | writers.extend(L) |
---|
| 303 | # Remove duplicates if they exist |
---|
| 304 | if len(writers) != len(set(writers)): |
---|
| 305 | result = [] |
---|
| 306 | for L in writers: |
---|
| 307 | if L not in result: |
---|
| 308 | result.append(L) |
---|
| 309 | writers = L |
---|
| 310 | # Raise an error if there are no matching extensions |
---|
| 311 | if len(writers) == 0: |
---|
| 312 | raise ValueError, "Unknown file type for " + path |
---|
| 313 | # All done |
---|
| 314 | return writers |
---|
| 315 | |
---|
| 316 | def save(self, path, data, format=None): |
---|
| 317 | """ |
---|
| 318 | Call the writer for the file type of path. |
---|
| 319 | |
---|
| 320 | Raises ValueError if no writer is available. |
---|
| 321 | Raises KeyError if format is not available. |
---|
| 322 | May raise a writer-defined exception if writer fails. |
---|
| 323 | """ |
---|
| 324 | if format is None: |
---|
| 325 | writers = self.lookup_writers(path) |
---|
| 326 | else: |
---|
| 327 | writers = self.writers[format] |
---|
| 328 | for fn in writers: |
---|
| 329 | try: |
---|
| 330 | return fn(path, data) |
---|
| 331 | except: |
---|
| 332 | pass # give other loaders a chance to succeed |
---|
| 333 | # If we get here it is because all loaders failed |
---|
| 334 | raise # reraises last exception |
---|
| 335 | |
---|
[f06d7fc] | 336 | |
---|
[4749514] | 337 | class Loader(object): |
---|
| 338 | """ |
---|
| 339 | Utility class to use the Registry as a singleton. |
---|
| 340 | """ |
---|
| 341 | ## Registry instance |
---|
| 342 | __registry = Registry() |
---|
[f06d7fc] | 343 | |
---|
[4749514] | 344 | def associate_file_type(self, ext, module): |
---|
| 345 | """ |
---|
| 346 | Look into a module to find whether it contains a |
---|
| 347 | Reader class. If so, append it to readers and (potentially) |
---|
| 348 | to the list of writers for the given extension |
---|
[f06d7fc] | 349 | |
---|
[4749514] | 350 | :param ext: file extension [string] |
---|
| 351 | :param module: module object |
---|
| 352 | """ |
---|
| 353 | return self.__registry.associate_file_type(ext, module) |
---|
| 354 | |
---|
| 355 | def associate_file_reader(self, ext, loader): |
---|
| 356 | """ |
---|
| 357 | Append a reader object to readers |
---|
[f06d7fc] | 358 | |
---|
[4749514] | 359 | :param ext: file extension [string] |
---|
| 360 | :param module: reader object |
---|
| 361 | """ |
---|
| 362 | return self.__registry.associate_file_reader(ext, loader) |
---|
| 363 | |
---|
| 364 | def load(self, file, format=None): |
---|
| 365 | """ |
---|
| 366 | Load a file |
---|
[f06d7fc] | 367 | |
---|
[4749514] | 368 | :param file: file name (path) |
---|
| 369 | :param format: specified format to use (optional) |
---|
| 370 | :return: DataInfo object |
---|
| 371 | """ |
---|
| 372 | return self.__registry.load(file, format) |
---|
[f06d7fc] | 373 | |
---|
[4749514] | 374 | def save(self, file, data, format): |
---|
| 375 | """ |
---|
| 376 | Save a DataInfo object to file |
---|
| 377 | :param file: file name (path) |
---|
| 378 | :param data: DataInfo object |
---|
[f06d7fc] | 379 | :param format: format to write the data in |
---|
[4749514] | 380 | """ |
---|
| 381 | return self.__registry.save(file, data, format) |
---|
[f06d7fc] | 382 | |
---|
[4749514] | 383 | def _get_registry_creation_time(self): |
---|
| 384 | """ |
---|
| 385 | Internal method used to test the uniqueness |
---|
| 386 | of the registry object |
---|
| 387 | """ |
---|
| 388 | return self.__registry._created |
---|
[f06d7fc] | 389 | |
---|
| 390 | def find_plugins(self, directory): |
---|
[4749514] | 391 | """ |
---|
| 392 | Find plugins in a given directory |
---|
[f06d7fc] | 393 | |
---|
[4749514] | 394 | :param dir: directory to look into to find new readers/writers |
---|
| 395 | """ |
---|
[f06d7fc] | 396 | return self.__registry.find_plugins(directory) |
---|
| 397 | |
---|
[4749514] | 398 | def get_wildcards(self): |
---|
[f06d7fc] | 399 | """ |
---|
| 400 | Return the list of wildcards |
---|
| 401 | """ |
---|
[4749514] | 402 | return self.__registry.wildcards |
---|