wms_cap_parsers.py 22 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640
  1. """!
  2. @brief Parsers for WMS/WMTS/NASA OnEarth capabilities files.
  3. List of classes:
  4. - wms_cap_parsers::BaseCapabilitiesTree
  5. - wms_cap_parsers::WMSXMLNsHandler
  6. - wms_cap_parsers::WMSCapabilitiesTree
  7. - wms_cap_parsers::WMTSXMLNsHandler
  8. - wms_cap_parsers::WMTSCapabilitiesTree
  9. - wms_cap_parsers::OnEarthCapabilitiesTree
  10. (C) 2012 by the GRASS Development Team
  11. This program is free software under the GNU General Public License
  12. (>=v2). Read the file COPYING that comes with GRASS for details.
  13. @author Stepan Turek <stepan.turek seznam.cz> (Mentor: Martin Landa)
  14. """
  15. try:
  16. from xml.etree.ElementTree import ParseError
  17. except ImportError: # < Python 2.7
  18. from xml.parsers.expat import ExpatError as ParseError
  19. import xml.etree.ElementTree as etree
  20. import grass.script as grass
  21. class BaseCapabilitiesTree(etree.ElementTree):
  22. def __init__(self, cap_file):
  23. """!Initialize xml.etree.ElementTree
  24. """
  25. try:
  26. etree.ElementTree.__init__(self, file = cap_file)
  27. except ParseError:
  28. raise ParseError(_("Unable to parse XML file"))
  29. except IOError as error:
  30. raise ParseError(_("Unable to open XML file '%s'.\n%s\n" % (cap_file, error)))
  31. if self.getroot() is None:
  32. raise ParseError(_("Root node was not found."))
  33. class WMSXMLNsHandler:
  34. def __init__(self, caps):
  35. """!Handle XML namespaces according to WMS version of capabilities.
  36. """
  37. self.namespace = "{http://www.opengis.net/wms}"
  38. if caps.getroot().find("Service") is not None:
  39. self.use_ns = False
  40. elif caps.getroot().find(self.namespace + "Service") is not None:
  41. self.use_ns = True
  42. else:
  43. raise ParseError(_("Unable to parse capabilities file.\n\
  44. Tag <%s> was not found.") % "Service")
  45. def Ns(self, tag_name):
  46. """!Add namespace to tag_name according to version
  47. """
  48. if self.use_ns:
  49. tag_name = self.namespace + tag_name
  50. return tag_name
  51. class WMSCapabilitiesTree(BaseCapabilitiesTree):
  52. def __init__(self, cap_file, force_version = None):
  53. """!Parses WMS capabilities file.
  54. If the capabilities file cannot be parsed if it raises xml.etree.ElementTree.ParseError.
  55. The class manges inheritance in 'Layer' elements. Inherited elements
  56. are added to 'Layer' element.
  57. The class also removes elements which are in invalid form and are needed
  58. by wxGUI capabilities dialog.
  59. @param cap_file - capabilities file
  60. @param force_version - force capabilities file version (1.1.1, 1.3.0)
  61. """
  62. BaseCapabilitiesTree.__init__(self, cap_file)
  63. self.xml_ns = WMSXMLNsHandler(self)
  64. grass.debug('Checking WMS capabilities tree.', 4)
  65. if not "version" in self.getroot().attrib:
  66. raise ParseError(_("Missing version attribute root node "
  67. "in Capabilities XML file"))
  68. else:
  69. wms_version = self.getroot().attrib["version"]
  70. if wms_version == "1.3.0":
  71. self.proj_tag = "CRS"
  72. else:
  73. self.proj_tag = "SRS"
  74. if force_version is not None:
  75. if wms_version != force_version:
  76. raise ParseError(_("WMS server does not support '%s' version.") % wms_version)
  77. capability = self._find(self.getroot(), "Capability")
  78. root_layer = self._find(capability, "Layer")
  79. self._checkFormats(capability)
  80. self._checkLayerTree(root_layer)
  81. grass.debug('Check of WMS capabilities tree was finished.', 4)
  82. def _checkFormats(self, capability):
  83. """!Check if format element is defined.
  84. """
  85. request = self._find(capability, "Request")
  86. get_map = self._find(request, "GetMap")
  87. formats = self._findall(get_map, "Format")
  88. def _checkLayerTree(self, parent_layer, first = True):
  89. """!Recursively check layer tree and manage inheritance in the tree
  90. """
  91. if first:
  92. self._initLayer(parent_layer, None)
  93. layers = parent_layer.findall((self.xml_ns.Ns("Layer")))
  94. for l in layers:
  95. self._initLayer(l, parent_layer)
  96. self._checkLayerTree(l, False)
  97. def _initLayer(self, layer, parent_layer):
  98. """Inherit elements from parent layer
  99. @param layer - <Layer> element which inherits
  100. @param parent_layer - <Layer> element which is inherited from
  101. """
  102. if parent_layer is not None:
  103. replaced_elements = [ ["EX_GeographicBoundingBox", "replace"],
  104. ["Attribution", "replace"],
  105. ["MinScaleDenominator", "replace"],
  106. ["MaxScaleDenominator", "replace"],
  107. ["AuthorityURL", "add"]]
  108. for element in replaced_elements:
  109. elems = layer.findall(self.xml_ns.Ns(element[0]))
  110. if len(elems) != 0 or element[1] == "add":
  111. for e in parent_layer.findall(self.xml_ns.Ns(element[0])):
  112. layer.append(e)
  113. inh_arguments = ["queryable", "cascaded", "opaque",
  114. "noSubsets", "fixedWidth", "fixedHeight"]
  115. for attr in parent_layer.attrib:
  116. if attr not in layer.attrib and attr in inh_arguments:
  117. layer.attrib[attr] = parent_layer.attrib[attr]
  118. self._inhNotSame(self.proj_tag, "element_content", layer, parent_layer)
  119. self._inhNotSame("BoundingBox", "attribute", layer, parent_layer, self.proj_tag)
  120. # remove invalid Styles
  121. styles = layer.findall(self.xml_ns.Ns('Style'))
  122. for s in styles:
  123. s_name = s.find(self.xml_ns.Ns('Name'))
  124. if s_name is None or not s_name.text:
  125. grass.debug('Removed invalid <Style> element.', 4)
  126. layer.remove(s)
  127. self._inhNotSame("Style", "child_element_content", layer, parent_layer, "Name")
  128. self._inhNotSame("Dimension", "attribute", layer, parent_layer, "name")
  129. def _inhNotSame(self, element_name, cmp_type, layer, parent_layer, add_arg = None):
  130. """Inherit elements which have unique values.
  131. @param element_name - name of inherited element
  132. @param cmp_type - 'element_content' - compared value is text of <Layer> element
  133. @param cmp_type - 'child_element_content' - compared value is text of a child of the <Layer> element
  134. @param cmp_type - 'attribute' - compared value is text of the <Layer> element attribute
  135. @param layer - <Layer> element which inherits
  136. @param parent_layer - <Layer> element which is inherited from
  137. @param add_arg - name of child element or attribute
  138. """
  139. elem = layer.findall(self.xml_ns.Ns(element_name))
  140. parent_elems = []
  141. if parent_layer is not None:
  142. parent_elems = parent_layer.findall(self.xml_ns.Ns(element_name))
  143. for par_elem in parent_elems:
  144. parent_cmp_text = None
  145. if cmp_type == "attribute":
  146. if add_arg in par_elem.attrib:
  147. parent_cmp_text = par_elem.attrib[add_arg];
  148. elif cmp_type == "element_content":
  149. parent_cmp_text = par_elem.text
  150. elif cmp_type == "child_element_content":
  151. parent_cmp = par_elem.find(self.xml_ns.Ns(add_arg))
  152. if parent_cmp is not None:
  153. parent_cmp_text = parent_cmp.text
  154. if parent_cmp_text is None:
  155. continue
  156. is_there = False
  157. for elem in elem:
  158. cmp_text = None
  159. if cmp_type == "attribute":
  160. if add_arg in elem.attrib:
  161. cmp_text = elem.attrib[add_arg]
  162. elif cmp_type == "element_content":
  163. cmp_text = elem.text
  164. elif cmp_type == "child_element_content":
  165. cmp = elem.find(self.xml_ns.Ns(add_arg))
  166. if cmp is not None:
  167. cmp_text = cmp.text
  168. if cmp_text is None or \
  169. cmp_text.lower() == parent_cmp_text.lower():
  170. is_there = True
  171. break
  172. if not is_there:
  173. layer.append(par_elem)
  174. def _find(self, etreeElement, tag):
  175. """!Find child element.
  176. If the element is not found it raises xml.etree.ElementTree.ParseError.
  177. """
  178. res = etreeElement.find(self.xml_ns.Ns(tag))
  179. if res is None:
  180. raise ParseError(_("Unable to parse capabilities file. \n\
  181. Tag <%s> was not found.") % tag)
  182. return res
  183. def _findall(self, etreeElement, tag):
  184. """!Find all children element.
  185. If no element is found it raises xml.etree.ElementTree.ParseError.
  186. """
  187. res = etreeElement.findall(self.xml_ns.Ns(tag))
  188. if not res:
  189. raise ParseError(_("Unable to parse capabilities file. \n\
  190. Tag <%s> was not found.") % tag)
  191. return res
  192. def getprojtag(self):
  193. """!Return projection tag according to version of capabilities (CRS/SRS).
  194. """
  195. return self.proj_tag
  196. def getxmlnshandler(self):
  197. """!Return WMSXMLNsHandler object.
  198. """
  199. return self.xml_ns
  200. class WMTSXMLNsHandler:
  201. """!Handle XML namespaces which are used in WMTS capabilities file.
  202. """
  203. def NsWmts(self, tag):
  204. """!Add namespace.
  205. """
  206. return "{http://www.opengis.net/wmts/1.0}" + tag
  207. def NsOws(self, tag):
  208. """!Add namespace.
  209. """
  210. return "{http://www.opengis.net/ows/1.1}" + tag
  211. class WMTSCapabilitiesTree(BaseCapabilitiesTree):
  212. def __init__(self, cap_file):
  213. """!Parses WMTS capabilities file.
  214. If the capabilities file cannot be parsed it raises xml.etree.ElementTree.ParseError.
  215. The class also removes elements which are in invalid form and are needed
  216. by wxGUI capabilities dialog or for creation of GetTile request by GRASS WMS library.
  217. @param cap_file - capabilities file
  218. """
  219. BaseCapabilitiesTree.__init__(self, cap_file)
  220. self.xml_ns = WMTSXMLNsHandler()
  221. grass.debug('Checking WMTS capabilities tree.', 4)
  222. contents = self._find(self.getroot(), 'Contents', self.xml_ns.NsWmts)
  223. tile_mat_sets = self._findall(contents, 'TileMatrixSet', self.xml_ns.NsWmts)
  224. for mat_set in tile_mat_sets:
  225. if not self._checkMatSet(mat_set):
  226. grass.debug('Removed invalid <TileMatrixSet> element.', 4)
  227. contents.remove(mat_set)
  228. # are there any <TileMatrixSet> elements after the check
  229. self._findall(contents, 'TileMatrixSet', self.xml_ns.NsWmts)
  230. layers = self._findall(contents, 'Layer', self.xml_ns.NsWmts)
  231. for l in layers:
  232. if not self._checkLayer(l):
  233. grass.debug('Removed invalid <Layer> element.', 4)
  234. contents.remove(l)
  235. # are there any <Layer> elements after the check
  236. self._findall(contents, 'Layer', self.xml_ns.NsWmts)
  237. grass.debug('Check of WMTS capabilities tree was finished.', 4)
  238. def _checkMatSet(self, mat_set):
  239. """!Check <TileMatrixSet>.
  240. """
  241. mat_set_id = mat_set.find(self.xml_ns.NsOws('Identifier'))
  242. if mat_set_id is None or not mat_set_id.text:
  243. return False
  244. mat_set_srs = mat_set.find(self.xml_ns.NsOws('SupportedCRS'))
  245. if mat_set_srs is None or \
  246. not mat_set_srs.text:
  247. return False
  248. tile_mats = mat_set.findall(self.xml_ns.NsWmts('TileMatrix'))
  249. if not tile_mats:
  250. return False
  251. for t_mat in tile_mats:
  252. if not self._checkMat(t_mat):
  253. grass.debug('Removed invalid <TileMatrix> element.', 4)
  254. mat_set.remove(t_mat)
  255. tile_mats = mat_set.findall(self.xml_ns.NsWmts('TileMatrix'))
  256. if not tile_mats:
  257. return False
  258. return True
  259. def _checkMat(self, t_mat):
  260. """!Check <TileMatrix>.
  261. """
  262. def _checkElement(t_mat, e, func):
  263. element = t_mat.find(self.xml_ns.NsWmts(e))
  264. if element is None or not element.text:
  265. return False
  266. try:
  267. e = func(element.text)
  268. except ValueError:
  269. return False
  270. if e < 0:
  271. return False
  272. return True
  273. for e, func in [['ScaleDenominator', float],
  274. ['TileWidth', int],
  275. ['TileHeight', int]]:
  276. if not _checkElement(t_mat, e, func):
  277. return False
  278. tile_mat_id = t_mat.find(self.xml_ns.NsOws('Identifier'))
  279. if tile_mat_id is None or not tile_mat_id.text:
  280. return False
  281. tl_str = t_mat.find(self.xml_ns.NsWmts('TopLeftCorner'))
  282. if tl_str is None or not tl_str.text:
  283. return False
  284. tl = tl_str.text.split(' ')
  285. if len(tl) < 2:
  286. return False
  287. for t in tl:
  288. try:
  289. t = float(t)
  290. except ValueError:
  291. return False
  292. return True
  293. def _checkLayer(self, layer):
  294. """!Check <Layer> element.
  295. """
  296. layer_id = layer.find(self.xml_ns.NsOws('Identifier'))
  297. if layer_id is None or not layer_id.text:
  298. return False
  299. mat_set_links = layer.findall(self.xml_ns.NsWmts('TileMatrixSetLink'))
  300. if not mat_set_links:
  301. return False
  302. styles = layer.findall(self.xml_ns.NsWmts('Style'))
  303. if not styles:
  304. return False
  305. for s in styles:
  306. s_name = s.find(self.xml_ns.NsOws('Identifier'))
  307. if s_name is None or not s_name.text:
  308. grass.debug('Removed invalid <Style> element.', 4)
  309. layer.remove(s_name)
  310. contents = self.getroot().find(self.xml_ns.NsWmts('Contents'))
  311. mat_sets = contents.findall(self.xml_ns.NsWmts('TileMatrixSet'))
  312. for link in mat_set_links:
  313. # <TileMatrixSetLink> does not point to existing <TileMatrixSet>
  314. if not self._checkMatSetLink(link, mat_sets):
  315. grass.debug('Removed invalid <TileMatrixSetLink> element.', 4)
  316. layer.remove(link)
  317. return True
  318. def _checkMatSetLink(self, link, mat_sets):
  319. """!Check <TileMatrixSetLink> element.
  320. """
  321. mat_set_link_id = link.find(self.xml_ns.NsWmts('TileMatrixSet')).text
  322. found = False
  323. for mat_set in mat_sets:
  324. mat_set_id = mat_set.find(self.xml_ns.NsOws('Identifier')).text
  325. if mat_set_id != mat_set_link_id:
  326. continue
  327. # the link points to existing <TileMatrixSet>
  328. found = True
  329. tile_mat_set_limits = link.find(self.xml_ns.NsWmts('TileMatrixSetLimits'))
  330. if tile_mat_set_limits is None:
  331. continue
  332. tile_mat_limits = tile_mat_set_limits.findall(self.xml_ns.NsWmts('TileMatrixLimits'))
  333. for limit in tile_mat_limits:
  334. if not self._checkMatSetLimit(limit):
  335. grass.debug('Removed invalid <TileMatrixLimits> element.', 4)
  336. tile_mat_limits.remove(limit)
  337. # are there any <TileMatrixLimits> elements after the check
  338. tile_mat_limits = tile_mat_set_limits.findall(self.xml_ns.NsWmts('TileMatrixLimits'))
  339. if not tile_mat_limits:
  340. grass.debug('Removed invalid <TileMatrixSetLimits> element.', 4)
  341. link.remove(tile_mat_set_limits)
  342. if not found:
  343. return False
  344. return True
  345. def _checkMatSetLimit(self, limit):
  346. """!Check <TileMatrixLimits> element.
  347. """
  348. limit_tile_mat = limit.find(self.xml_ns.NsWmts('TileMatrix'))
  349. if limit_tile_mat is None or not limit_tile_mat.text:
  350. return False
  351. for i in ['MinTileRow', 'MaxTileRow', 'MinTileCol', 'MaxTileCol']:
  352. i_tag = limit.find(self.xml_ns.NsWmts(i))
  353. if i_tag is None:
  354. return False
  355. try:
  356. int(i_tag.text)
  357. except ValueError:
  358. return False
  359. return True
  360. def _find(self, etreeElement, tag, ns = None):
  361. """!Find child element.
  362. If the element is not found it raises xml.etree.ElementTree.ParseError.
  363. """
  364. if not ns:
  365. res = etreeElement.find(tag)
  366. else:
  367. res = etreeElement.find(ns(tag))
  368. if res is None:
  369. raise ParseError(_("Unable to parse capabilities file. \n\
  370. Tag '%s' was not found.") % tag)
  371. return res
  372. def _findall(self, etreeElement, tag, ns = None):
  373. """!Find all children element.
  374. If no element is found it raises xml.etree.ElementTree.ParseError.
  375. """
  376. if not ns:
  377. res = etreeElement.findall(tag)
  378. else:
  379. res = etreeElement.findall(ns(tag))
  380. if not res:
  381. raise ParseError(_("Unable to parse capabilities file. \n\
  382. Tag '%s' was not found.") % tag)
  383. return res
  384. def getxmlnshandler(self):
  385. """!Return WMTSXMLNsHandler object.
  386. """
  387. return self.xml_ns
  388. class OnEarthCapabilitiesTree(BaseCapabilitiesTree):
  389. def __init__(self, cap_file):
  390. """!Parse NASA OnEarth tile service file.
  391. If the file cannot be parsed it raises xml.etree.ElementTree.ParseError.
  392. The class also removes elements which are in invalid form and are needed
  393. by wxGUI capabilities dialog or for creation of GetMap request by GRASS WMS library.
  394. @param cap_file - capabilities file
  395. """
  396. BaseCapabilitiesTree.__init__(self, cap_file)
  397. grass.debug('Checking OnEarth capabilities tree.', 4)
  398. self._checkLayerTree(self.getroot())
  399. grass.debug('Check if OnEarth capabilities tree was finished.', 4)
  400. def _checkLayerTree(self, parent_layer, first = True):
  401. """!Recursively check layer tree.
  402. """
  403. if first:
  404. tiled_patterns = self._find(parent_layer, 'TiledPatterns')
  405. layers = tiled_patterns.findall('TiledGroup')
  406. layers += tiled_patterns.findall('TiledGroups')
  407. parent_layer = tiled_patterns
  408. else:
  409. layers = parent_layer.findall('TiledGroup')
  410. layers += parent_layer.findall('TiledGroups')
  411. for l in layers:
  412. if not self._checkLayer(l):
  413. grass.debug(('Removed invalid <%s> element.' % l.tag), 4)
  414. parent_layer.remove(l)
  415. if l.tag == 'TiledGroups':
  416. self._checkLayerTree(l, False)
  417. def _find(self, etreeElement, tag):
  418. """!Find child element.
  419. If the element is not found it raises xml.etree.ElementTree.ParseError.
  420. """
  421. res = etreeElement.find(tag)
  422. if res is None:
  423. raise ParseError(_("Unable to parse tile service file. \n\
  424. Tag <%s> was not found.") % tag)
  425. return res
  426. def _checkLayer(self, layer):
  427. """!Check <TiledGroup>/<TiledGroups> elements.
  428. """
  429. if layer.tag == 'TiledGroups':
  430. return True
  431. name = layer.find('Name')
  432. if name is None or not name.text:
  433. return False
  434. t_patts = layer.findall('TilePattern')
  435. for patt in t_patts:
  436. urls = self._getUrls(patt)
  437. for url in urls:
  438. if not self.gettilepatternurldata(url):
  439. urls.remove(url)
  440. # check if there are any vaild urls
  441. if not urls:
  442. grass.debug('<TilePattern> was removed. It has no valid url.', 4)
  443. layer.remove(patt)
  444. patt.text = '\n'.join(urls)
  445. t_patts = layer.findall('TilePattern')
  446. if not t_patts:
  447. return False
  448. return True
  449. def _getUrls(self, tile_pattern):
  450. """!Get all urls from tile pattern.
  451. """
  452. urls = []
  453. if tile_pattern.text is not None:
  454. tile_patt_lines = tile_pattern.text.split('\n')
  455. for line in tile_patt_lines:
  456. if 'request=GetMap' in line:
  457. urls.append(line.strip())
  458. return urls
  459. def gettilepatternurldata(self, url):
  460. """!Parse url string in Tile Pattern.
  461. """
  462. par_url = bbox = width = height = None
  463. bbox_idxs = self.geturlparamidxs(url, "bbox=")
  464. if bbox_idxs is None:
  465. return None
  466. par_url = [url[:bbox_idxs[0] - 1], url[bbox_idxs[1]:]]
  467. bbox = url[bbox_idxs[0] + len('bbox=') : bbox_idxs[1]]
  468. bbox_list = bbox.split(',')
  469. if len(bbox_list) < 4:
  470. return None
  471. try:
  472. bbox = map(float, bbox.split(','))
  473. except ValueError:
  474. return None
  475. width_idxs = self.geturlparamidxs(url, "width=")
  476. if width_idxs is None:
  477. return None
  478. try:
  479. width = int(url[width_idxs[0] + len('width=') : width_idxs[1]])
  480. except ValueError:
  481. return None
  482. height_idxs = self.geturlparamidxs(url, "height=")
  483. if height_idxs is None:
  484. return None
  485. try:
  486. height = int(url[height_idxs[0] + len('height=') : height_idxs[1]])
  487. except ValueError:
  488. return None
  489. if height < 0 or width < 0:
  490. return None
  491. return par_url, bbox, width, height
  492. def geturlparamidxs(self, params_str, param_key):
  493. """!Find start and end index of parameter and it's value in url string
  494. """
  495. start_i = params_str.lower().find(param_key)
  496. if start_i < 0:
  497. return None
  498. end_i = params_str.find("&", start_i)
  499. if end_i < 0:
  500. end_i = len(params_str)
  501. return (start_i, end_i)