[a916ccc] | 1 | """ |
---|
[daa56d0] | 2 | File handler to support different file extensions. |
---|
| 3 | Uses reflectometry's registry utility. |
---|
[bbc8013] | 4 | |
---|
| 5 | The default readers are found in the 'readers' sub-module |
---|
| 6 | and registered by default at initialization time. |
---|
| 7 | |
---|
| 8 | To add a new default reader, one must register it in |
---|
[f60a8c2] | 9 | the register_readers method found in readers/__init__.py. |
---|
[bbc8013] | 10 | |
---|
[f60a8c2] | 11 | A utility method (find_plugins) is available to inspect |
---|
[bbc8013] | 12 | a directory (for instance, a user plug-in directory) and |
---|
| 13 | look for new readers/writers. |
---|
[a916ccc] | 14 | """ |
---|
[f60a8c2] | 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 | ###################################################################### |
---|
[daa56d0] | 22 | |
---|
[f60a8c2] | 23 | import os |
---|
[daa56d0] | 24 | import sys |
---|
[a916ccc] | 25 | import logging |
---|
[daa56d0] | 26 | import time |
---|
[84545dd] | 27 | from zipfile import ZipFile |
---|
[a7a5886] | 28 | from data_util.registry import ExtensionRegistry |
---|
[bbc8013] | 29 | # Default readers are defined in the readers sub-module |
---|
| 30 | import readers |
---|
[a7a5886] | 31 | from readers import ascii_reader |
---|
| 32 | from readers import cansas_reader |
---|
[bbc8013] | 33 | |
---|
[daa56d0] | 34 | class Registry(ExtensionRegistry): |
---|
| 35 | """ |
---|
[0997158f] | 36 | Registry class for file format extensions. |
---|
| 37 | Readers and writers are supported. |
---|
[daa56d0] | 38 | """ |
---|
| 39 | |
---|
| 40 | def __init__(self): |
---|
| 41 | super(Registry, self).__init__() |
---|
| 42 | |
---|
| 43 | ## Writers |
---|
| 44 | self.writers = {} |
---|
[a916ccc] | 45 | |
---|
[1ed9c57] | 46 | ## List of wildcards |
---|
| 47 | self.wildcards = ['All (*.*)|*.*'] |
---|
| 48 | |
---|
[daa56d0] | 49 | ## Creation time, for testing |
---|
| 50 | self._created = time.time() |
---|
| 51 | |
---|
[bbc8013] | 52 | # Register default readers |
---|
[07f8d9a] | 53 | readers.read_associations(self) |
---|
| 54 | |
---|
[f60a8c2] | 55 | #TODO: remove the following line when ready to switch to |
---|
[a7a5886] | 56 | #the new default readers |
---|
[07f8d9a] | 57 | #readers.register_readers(self._identify_plugin) |
---|
| 58 | |
---|
| 59 | # Look for plug-in readers |
---|
[371cb85] | 60 | #self.find_plugins('plugins') |
---|
[07f8d9a] | 61 | |
---|
[88e289c] | 62 | def load(self, path, format=None): |
---|
| 63 | """ |
---|
[0997158f] | 64 | Call the loader for the file type of path. |
---|
| 65 | |
---|
| 66 | :param path: file path |
---|
[f60a8c2] | 67 | :param format: explicit extension, to force the use |
---|
[a7a5886] | 68 | of a particular reader |
---|
[0997158f] | 69 | |
---|
| 70 | Defaults to the ascii (multi-column) reader |
---|
| 71 | if no reader was registered for the file's |
---|
[f60a8c2] | 72 | extension. |
---|
[88e289c] | 73 | """ |
---|
| 74 | try: |
---|
| 75 | return super(Registry, self).load(path, format=format) |
---|
| 76 | except: |
---|
[f118fe2f] | 77 | try: |
---|
| 78 | # No reader was found. Default to the ascii reader. |
---|
| 79 | ascii_loader = ascii_reader.Reader() |
---|
| 80 | return ascii_loader.read(path) |
---|
| 81 | except: |
---|
| 82 | cansas_loader = cansas_reader.Reader() |
---|
| 83 | return cansas_loader.read(path) |
---|
[a7a5886] | 84 | |
---|
[daa56d0] | 85 | def find_plugins(self, dir): |
---|
| 86 | """ |
---|
[0997158f] | 87 | Find readers in a given directory. This method |
---|
| 88 | can be used to inspect user plug-in directories to |
---|
| 89 | find new readers/writers. |
---|
| 90 | |
---|
| 91 | :param dir: directory to search into |
---|
| 92 | |
---|
| 93 | :return: number of readers found |
---|
[daa56d0] | 94 | """ |
---|
| 95 | readers_found = 0 |
---|
[ed61f2a1] | 96 | temp_path = os.path.abspath(dir) |
---|
| 97 | if not os.path.isdir(temp_path): |
---|
| 98 | temp_path = os.path.join(os.getcwd(), dir) |
---|
| 99 | if not os.path.isdir(temp_path): |
---|
| 100 | temp_path = os.path.join(os.path.dirname(__file__), dir) |
---|
| 101 | if not os.path.isdir(temp_path): |
---|
[4e367f6] | 102 | temp_path = os.path.join(os.path.dirname(sys.path[0]), dir) |
---|
[bbc8013] | 103 | |
---|
[ed61f2a1] | 104 | dir = temp_path |
---|
[bbc8013] | 105 | # Check whether the directory exists |
---|
[f60a8c2] | 106 | if not os.path.isdir(dir): |
---|
[ed61f2a1] | 107 | msg = "DataLoader couldn't locate DataLoader plugin folder." |
---|
| 108 | msg += """ "%s" does not exist""" % dir |
---|
| 109 | logging.warning(msg) |
---|
[bbc8013] | 110 | return readers_found |
---|
| 111 | |
---|
[daa56d0] | 112 | for item in os.listdir(dir): |
---|
| 113 | full_path = os.path.join(dir, item) |
---|
[84545dd] | 114 | if os.path.isfile(full_path): |
---|
| 115 | |
---|
| 116 | # Process python files |
---|
| 117 | if item.endswith('.py'): |
---|
| 118 | toks = os.path.splitext(os.path.basename(item)) |
---|
| 119 | try: |
---|
| 120 | sys.path.insert(0, os.path.abspath(dir)) |
---|
| 121 | module = __import__(toks[0], globals(), locals()) |
---|
| 122 | if self._identify_plugin(module): |
---|
[daa56d0] | 123 | readers_found += 1 |
---|
[a7a5886] | 124 | except: |
---|
| 125 | msg = "Loader: Error importing " |
---|
| 126 | msg += "%s\n %s" % (item, sys.exc_value) |
---|
| 127 | logging.error(msg) |
---|
[daa56d0] | 128 | |
---|
[84545dd] | 129 | # Process zip files |
---|
| 130 | elif item.endswith('.zip'): |
---|
| 131 | try: |
---|
| 132 | # Find the modules in the zip file |
---|
| 133 | zfile = ZipFile(item) |
---|
| 134 | nlist = zfile.namelist() |
---|
| 135 | |
---|
| 136 | sys.path.insert(0, item) |
---|
| 137 | for mfile in nlist: |
---|
| 138 | try: |
---|
| 139 | # Change OS path to python path |
---|
| 140 | fullname = mfile.replace('/', '.') |
---|
| 141 | fullname = os.path.splitext(fullname)[0] |
---|
[a7a5886] | 142 | module = __import__(fullname, globals(), |
---|
| 143 | locals(), [""]) |
---|
[84545dd] | 144 | if self._identify_plugin(module): |
---|
| 145 | readers_found += 1 |
---|
| 146 | except: |
---|
[a7a5886] | 147 | msg = "Loader: Error importing" |
---|
| 148 | msg += " %s\n %s" % (mfile, sys.exc_value) |
---|
| 149 | logging.error(msg) |
---|
[84545dd] | 150 | |
---|
| 151 | except: |
---|
[a7a5886] | 152 | msg = "Loader: Error importing " |
---|
| 153 | msg += " %s\n %s" % (item, sys.exc_value) |
---|
| 154 | logging.error(msg) |
---|
[84545dd] | 155 | |
---|
[f60a8c2] | 156 | return readers_found |
---|
[84545dd] | 157 | |
---|
[07f8d9a] | 158 | def associate_file_type(self, ext, module): |
---|
| 159 | """ |
---|
[f60a8c2] | 160 | Look into a module to find whether it contains a |
---|
[0997158f] | 161 | Reader class. If so, APPEND it to readers and (potentially) |
---|
| 162 | to the list of writers for the given extension |
---|
| 163 | |
---|
| 164 | :param ext: file extension [string] |
---|
| 165 | :param module: module object |
---|
[07f8d9a] | 166 | """ |
---|
| 167 | reader_found = False |
---|
| 168 | |
---|
| 169 | if hasattr(module, "Reader"): |
---|
| 170 | try: |
---|
| 171 | # Find supported extensions |
---|
| 172 | loader = module.Reader() |
---|
| 173 | if ext not in self.loaders: |
---|
| 174 | self.loaders[ext] = [] |
---|
| 175 | # Append the new reader to the list |
---|
| 176 | self.loaders[ext].append(loader.read) |
---|
| 177 | |
---|
| 178 | reader_found = True |
---|
| 179 | |
---|
| 180 | # Keep track of wildcards |
---|
| 181 | type_name = module.__name__ |
---|
| 182 | if hasattr(loader, 'type_name'): |
---|
| 183 | type_name = loader.type_name |
---|
| 184 | |
---|
[a7a5886] | 185 | wcard = "%s files (*%s)|*%s" % (type_name, ext.lower(), |
---|
| 186 | ext.lower()) |
---|
[07f8d9a] | 187 | if wcard not in self.wildcards: |
---|
| 188 | self.wildcards.append(wcard) |
---|
| 189 | |
---|
| 190 | # Check whether writing is supported |
---|
| 191 | if hasattr(loader, 'write'): |
---|
| 192 | if ext not in self.writers: |
---|
| 193 | self.writers[ext] = [] |
---|
| 194 | # Append the new writer to the list |
---|
| 195 | self.writers[ext].append(loader.write) |
---|
| 196 | |
---|
| 197 | except: |
---|
[a7a5886] | 198 | msg = "Loader: Error accessing" |
---|
| 199 | msg += " Reader in %s\n %s" % (module.__name__, sys.exc_value) |
---|
| 200 | logging.error(msg) |
---|
[07f8d9a] | 201 | return reader_found |
---|
| 202 | |
---|
[fd4b6f8] | 203 | def associate_file_reader(self, ext, loader): |
---|
| 204 | """ |
---|
[0997158f] | 205 | Append a reader object to readers |
---|
| 206 | |
---|
| 207 | :param ext: file extension [string] |
---|
| 208 | :param module: reader object |
---|
[fd4b6f8] | 209 | """ |
---|
| 210 | reader_found = False |
---|
| 211 | |
---|
| 212 | try: |
---|
| 213 | # Find supported extensions |
---|
| 214 | if ext not in self.loaders: |
---|
| 215 | self.loaders[ext] = [] |
---|
| 216 | # Append the new reader to the list |
---|
| 217 | self.loaders[ext].append(loader.read) |
---|
| 218 | |
---|
| 219 | reader_found = True |
---|
| 220 | |
---|
| 221 | # Keep track of wildcards |
---|
| 222 | if hasattr(loader, 'type_name'): |
---|
| 223 | type_name = loader.type_name |
---|
| 224 | |
---|
[f60a8c2] | 225 | wcard = "%s files (*%s)|*%s" % (type_name, ext.lower(), |
---|
[a7a5886] | 226 | ext.lower()) |
---|
[fd4b6f8] | 227 | if wcard not in self.wildcards: |
---|
| 228 | self.wildcards.append(wcard) |
---|
| 229 | |
---|
| 230 | except: |
---|
[a7a5886] | 231 | msg = "Loader: Error accessing Reader " |
---|
| 232 | msg += "in %s\n %s" % (module.__name__, sys.exc_value) |
---|
| 233 | logging.error(msg) |
---|
[fd4b6f8] | 234 | return reader_found |
---|
| 235 | |
---|
[84545dd] | 236 | def _identify_plugin(self, module): |
---|
| 237 | """ |
---|
[0997158f] | 238 | Look into a module to find whether it contains a |
---|
| 239 | Reader class. If so, add it to readers and (potentially) |
---|
| 240 | to the list of writers. |
---|
| 241 | :param module: module object |
---|
| 242 | |
---|
[84545dd] | 243 | """ |
---|
| 244 | reader_found = False |
---|
| 245 | |
---|
| 246 | if hasattr(module, "Reader"): |
---|
| 247 | try: |
---|
| 248 | # Find supported extensions |
---|
| 249 | loader = module.Reader() |
---|
| 250 | for ext in loader.ext: |
---|
| 251 | if ext not in self.loaders: |
---|
| 252 | self.loaders[ext] = [] |
---|
[a7a5886] | 253 | # When finding a reader at run time, |
---|
| 254 | # treat this reader as the new default |
---|
| 255 | self.loaders[ext].insert(0, loader.read) |
---|
[954c045] | 256 | |
---|
[84545dd] | 257 | reader_found = True |
---|
[954c045] | 258 | |
---|
[1ed9c57] | 259 | # Keep track of wildcards |
---|
[07f8d9a] | 260 | type_name = module.__name__ |
---|
| 261 | if hasattr(loader, 'type_name'): |
---|
| 262 | type_name = loader.type_name |
---|
[a7a5886] | 263 | wcard = "%s files (*%s)|*%s" % (type_name, ext.lower(), |
---|
| 264 | ext.lower()) |
---|
[07f8d9a] | 265 | if wcard not in self.wildcards: |
---|
[f60a8c2] | 266 | self.wildcards.append(wcard) |
---|
[1ed9c57] | 267 | |
---|
[84545dd] | 268 | # Check whether writing is supported |
---|
| 269 | if hasattr(loader, 'write'): |
---|
| 270 | for ext in loader.ext: |
---|
| 271 | if ext not in self.writers: |
---|
| 272 | self.writers[ext] = [] |
---|
[f60a8c2] | 273 | self.writers[ext].insert(0, loader.write) |
---|
[954c045] | 274 | |
---|
[84545dd] | 275 | except: |
---|
[a7a5886] | 276 | msg = "Loader: Error accessing Reader" |
---|
| 277 | msg += " in %s\n %s" % (module.__name__, sys.exc_value) |
---|
| 278 | logging.error(msg) |
---|
[84545dd] | 279 | return reader_found |
---|
[d22da51] | 280 | |
---|
[daa56d0] | 281 | def lookup_writers(self, path): |
---|
| 282 | """ |
---|
[0997158f] | 283 | :return: the loader associated with the file type of path. |
---|
[daa56d0] | 284 | |
---|
[0997158f] | 285 | :Raises ValueError: if file type is not known. |
---|
[f60a8c2] | 286 | """ |
---|
[daa56d0] | 287 | # Find matching extensions |
---|
| 288 | extlist = [ext for ext in self.extensions() if path.endswith(ext)] |
---|
| 289 | # Sort matching extensions by decreasing order of length |
---|
[a7a5886] | 290 | extlist.sort(lambda a, b: len(a) < len(b)) |
---|
[daa56d0] | 291 | # Combine loaders for matching extensions into one big list |
---|
| 292 | writers = [] |
---|
| 293 | for L in [self.writers[ext] for ext in extlist]: |
---|
| 294 | writers.extend(L) |
---|
| 295 | # Remove duplicates if they exist |
---|
| 296 | if len(writers) != len(set(writers)): |
---|
| 297 | result = [] |
---|
| 298 | for L in writers: |
---|
[f60a8c2] | 299 | if L not in result: |
---|
| 300 | result.append(L) |
---|
[daa56d0] | 301 | writers = L |
---|
| 302 | # Raise an error if there are no matching extensions |
---|
| 303 | if len(writers) == 0: |
---|
[a7a5886] | 304 | raise ValueError, "Unknown file type for " + path |
---|
[daa56d0] | 305 | # All done |
---|
| 306 | return writers |
---|
| 307 | |
---|
| 308 | def save(self, path, data, format=None): |
---|
| 309 | """ |
---|
| 310 | Call the writer for the file type of path. |
---|
[d22da51] | 311 | |
---|
[daa56d0] | 312 | Raises ValueError if no writer is available. |
---|
| 313 | Raises KeyError if format is not available. |
---|
[0997158f] | 314 | |
---|
[f60a8c2] | 315 | May raise a writer-defined exception if writer fails. |
---|
[daa56d0] | 316 | """ |
---|
| 317 | if format is None: |
---|
| 318 | writers = self.lookup_writers(path) |
---|
| 319 | else: |
---|
| 320 | writers = self.writers[format] |
---|
| 321 | for fn in writers: |
---|
| 322 | try: |
---|
| 323 | return fn(path, data) |
---|
| 324 | except: |
---|
[f60a8c2] | 325 | pass # give other loaders a chance to succeed |
---|
[daa56d0] | 326 | # If we get here it is because all loaders failed |
---|
[f60a8c2] | 327 | raise # reraises last exception |
---|
[daa56d0] | 328 | |
---|
| 329 | |
---|
[1b0b3ca] | 330 | class Loader(object): |
---|
[d22da51] | 331 | """ |
---|
[0997158f] | 332 | Utility class to use the Registry as a singleton. |
---|
[d22da51] | 333 | """ |
---|
[daa56d0] | 334 | ## Registry instance |
---|
| 335 | __registry = Registry() |
---|
| 336 | |
---|
[07f8d9a] | 337 | def associate_file_type(self, ext, module): |
---|
| 338 | """ |
---|
[f60a8c2] | 339 | Look into a module to find whether it contains a |
---|
[0997158f] | 340 | Reader class. If so, append it to readers and (potentially) |
---|
| 341 | to the list of writers for the given extension |
---|
| 342 | |
---|
| 343 | :param ext: file extension [string] |
---|
| 344 | :param module: module object |
---|
[07f8d9a] | 345 | """ |
---|
| 346 | return self.__registry.associate_file_type(ext, module) |
---|
| 347 | |
---|
[fd4b6f8] | 348 | def associate_file_reader(self, ext, loader): |
---|
| 349 | """ |
---|
[0997158f] | 350 | Append a reader object to readers |
---|
| 351 | |
---|
| 352 | :param ext: file extension [string] |
---|
| 353 | :param module: reader object |
---|
[fd4b6f8] | 354 | """ |
---|
| 355 | return self.__registry.associate_file_reader(ext, loader) |
---|
| 356 | |
---|
[daa56d0] | 357 | def load(self, file, format=None): |
---|
| 358 | """ |
---|
[0997158f] | 359 | Load a file |
---|
| 360 | |
---|
| 361 | :param file: file name (path) |
---|
| 362 | :param format: specified format to use (optional) |
---|
| 363 | :return: DataInfo object |
---|
[daa56d0] | 364 | """ |
---|
| 365 | return self.__registry.load(file, format) |
---|
[d22da51] | 366 | |
---|
[daa56d0] | 367 | def save(self, file, data, format): |
---|
| 368 | """ |
---|
[0997158f] | 369 | Save a DataInfo object to file |
---|
| 370 | :param file: file name (path) |
---|
| 371 | :param data: DataInfo object |
---|
| 372 | :param format: format to write the data in |
---|
[daa56d0] | 373 | """ |
---|
| 374 | return self.__registry.save(file, data, format) |
---|
| 375 | |
---|
| 376 | def _get_registry_creation_time(self): |
---|
| 377 | """ |
---|
[0997158f] | 378 | Internal method used to test the uniqueness |
---|
| 379 | of the registry object |
---|
[daa56d0] | 380 | """ |
---|
| 381 | return self.__registry._created |
---|
[d22da51] | 382 | |
---|
[daa56d0] | 383 | def find_plugins(self, dir): |
---|
| 384 | """ |
---|
[0997158f] | 385 | Find plugins in a given directory |
---|
| 386 | |
---|
| 387 | :param dir: directory to look into to find new readers/writers |
---|
[daa56d0] | 388 | """ |
---|
| 389 | return self.__registry.find_plugins(dir) |
---|
[a916ccc] | 390 | |
---|
[1ed9c57] | 391 | def get_wildcards(self): |
---|
| 392 | return self.__registry.wildcards |
---|