wms_cap_parsers.py 22 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635
  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. import xml.etree.ElementTree as etree
  16. import grass.script as grass
  17. class BaseCapabilitiesTree(etree.ElementTree):
  18. def __init__(self, cap_file):
  19. """!Initialize xml.etree.ElementTree
  20. """
  21. try:
  22. etree.ElementTree.__init__(self, file = cap_file)
  23. except etree.ParseError:
  24. raise etree.ParseError(_("Unable to parse XML file"))
  25. except IOError as error:
  26. raise etree.ParseError(_("Unabble to open XML file '%s'.\n%s\n" % (cap_file, error)))
  27. if self.getroot() is None:
  28. raise etree.ParseError(_("Root node was not found."))
  29. class WMSXMLNsHandler:
  30. def __init__(self, caps):
  31. """!Handle XML namespaces according to WMS version of capabilities.
  32. """
  33. self.namespace = "{http://www.opengis.net/wms}"
  34. if caps.getroot().find("Service") is not None:
  35. self.use_ns = False
  36. elif caps.getroot().find(self.namespace + "Service") is not None:
  37. self.use_ns = True
  38. else:
  39. raise etree.ParseError(_("Unable to parse capabilities file.\n\
  40. Tag <%s> was not found.") % "Service")
  41. def Ns(self, tag_name):
  42. """!Add namespace to tag_name according to version
  43. """
  44. if self.use_ns:
  45. tag_name = self.namespace + tag_name
  46. return tag_name
  47. class WMSCapabilitiesTree(BaseCapabilitiesTree):
  48. def __init__(self, cap_file, force_version = None):
  49. """!Parses WMS capabilities file.
  50. If the capabilities file cannot be parsed if it raises xml.etree.ElementTree.ParseError.
  51. The class manges inheritance in 'Layer' elements. Inherited elements
  52. are added to 'Layer' element.
  53. The class also removes elements which are in invalid form and are needed
  54. by wxGUI capabilities dialog.
  55. @param cap_file - capabilities file
  56. @param force_version - force capabilities file version (1.1.1, 1.3.0)
  57. """
  58. BaseCapabilitiesTree.__init__(self, cap_file)
  59. self.xml_ns = WMSXMLNsHandler(self)
  60. grass.debug('Checking WMS capabilities tree.', 4)
  61. if not "version" in self.getroot().attrib:
  62. raise etree.ParseError(_("Missing version attribute root node "
  63. "in Capabilities XML file"))
  64. else:
  65. wms_version = self.getroot().attrib["version"]
  66. if wms_version == "1.3.0":
  67. self.proj_tag = "CRS"
  68. else:
  69. self.proj_tag = "SRS"
  70. if force_version is not None:
  71. if wms_version != force_version:
  72. raise etree.ParseError(_("WMS server does not support '%s' version.") % wms_version)
  73. capability = self._find(self.getroot(), "Capability")
  74. root_layer = self._find(capability, "Layer")
  75. self._checkFormats(capability)
  76. self._checkLayerTree(root_layer)
  77. grass.debug('Check of WMS capabilities tree was finished.', 4)
  78. def _checkFormats(self, capability):
  79. """!Check if format element is defined.
  80. """
  81. request = self._find(capability, "Request")
  82. get_map = self._find(request, "GetMap")
  83. formats = self._findall(get_map, "Format")
  84. def _checkLayerTree(self, parent_layer, first = True):
  85. """!Recursively check layer tree and manage inheritance in the tree
  86. """
  87. if first:
  88. self._initLayer(parent_layer, None)
  89. layers = parent_layer.findall((self.xml_ns.Ns("Layer")))
  90. for l in layers:
  91. self._initLayer(l, parent_layer)
  92. self._checkLayerTree(l, False)
  93. def _initLayer(self, layer, parent_layer):
  94. """Inherit elements from parent layer
  95. @param layer - <Layer> element which inherits
  96. @param parent_layer - <Layer> element which is inherited from
  97. """
  98. if parent_layer is not None:
  99. replaced_elements = [ ["EX_GeographicBoundingBox", "replace"],
  100. ["Attribution", "replace"],
  101. ["MinScaleDenominator", "replace"],
  102. ["MaxScaleDenominator", "replace"],
  103. ["AuthorityURL", "add"]]
  104. for element in replaced_elements:
  105. elems = layer.findall(self.xml_ns.Ns(element[0]))
  106. if len(elems) != 0 or element[1] == "add":
  107. for e in parent_layer.findall(self.xml_ns.Ns(element[0])):
  108. layer.append(e)
  109. inh_arguments = ["queryable", "cascaded", "opaque",
  110. "noSubsets", "fixedWidth", "fixedHeight"]
  111. for attr in parent_layer.attrib:
  112. if attr not in layer.attrib and attr in inh_arguments:
  113. layer.attrib[attr] = parent_layer.attrib[attr]
  114. self._inhNotSame(self.proj_tag, "element_content", layer, parent_layer)
  115. self._inhNotSame("BoundingBox", "attribute", layer, parent_layer, self.proj_tag)
  116. # remove invalid Styles
  117. styles = layer.findall(self.xml_ns.Ns('Style'))
  118. for s in styles:
  119. s_name = s.find(self.xml_ns.Ns('Name'))
  120. if s_name is None or not s_name.text:
  121. grass.debug('Removed invalid <Style> element.', 4)
  122. layer.remove(s)
  123. self._inhNotSame("Style", "child_element_content", layer, parent_layer, "Name")
  124. self._inhNotSame("Dimension", "attribute", layer, parent_layer, "name")
  125. def _inhNotSame(self, element_name, cmp_type, layer, parent_layer, add_arg = None):
  126. """Inherit elements which have unique values.
  127. @param element_name - name of inherited element
  128. @param cmp_type - 'element_content' - compared value is text of <Layer> element
  129. @param cmp_type - 'child_element_content' - compared value is text of a child of the <Layer> element
  130. @param cmp_type - 'attribute' - compared value is text of the <Layer> element attribute
  131. @param layer - <Layer> element which inherits
  132. @param parent_layer - <Layer> element which is inherited from
  133. @param add_arg - name of child element or attribute
  134. """
  135. elem = layer.findall(self.xml_ns.Ns(element_name))
  136. parent_elems = []
  137. if parent_layer is not None:
  138. parent_elems = parent_layer.findall(self.xml_ns.Ns(element_name))
  139. for par_elem in parent_elems:
  140. parent_cmp_text = None
  141. if cmp_type == "attribute":
  142. if add_arg in par_elem.attrib:
  143. parent_cmp_text = par_elem.attrib[add_arg];
  144. elif cmp_type == "element_content":
  145. parent_cmp_text = par_elem.text
  146. elif cmp_type == "child_element_content":
  147. parent_cmp = par_elem.find(self.xml_ns.Ns(add_arg))
  148. if parent_cmp is not None:
  149. parent_cmp_text = parent_cmp.text
  150. if parent_cmp_text is None:
  151. continue
  152. is_there = False
  153. for elem in elem:
  154. cmp_text = None
  155. if cmp_type == "attribute":
  156. if add_arg in elem.attrib:
  157. cmp_text = elem.attrib[add_arg]
  158. elif cmp_type == "element_content":
  159. cmp_text = elem.text
  160. elif cmp_type == "child_element_content":
  161. cmp = elem.find(self.xml_ns.Ns(add_arg))
  162. if cmp is not None:
  163. cmp_text = cmp.text
  164. if cmpt_text is None or \
  165. cmp_text.lower() == parent_cmp_text.lower():
  166. is_there = True
  167. break
  168. if not is_there:
  169. layer.append(par_elem)
  170. def _find(self, etreeElement, tag):
  171. """!Find child element.
  172. If the element is not found it raises xml.etree.ElementTree.ParseError.
  173. """
  174. res = etreeElement.find(self.xml_ns.Ns(tag))
  175. if res is None:
  176. raise etree.ParseError(_("Unable to parse capabilities file. \n\
  177. Tag <%s> was not found.") % tag)
  178. return res
  179. def _findall(self, etreeElement, tag):
  180. """!Find all children element.
  181. If no element is found it raises xml.etree.ElementTree.ParseError.
  182. """
  183. res = etreeElement.findall(self.xml_ns.Ns(tag))
  184. if not res:
  185. raise etree.ParseError(_("Unable to parse capabilities file. \n\
  186. Tag <%s> was not found.") % tag)
  187. return res
  188. def getprojtag(self):
  189. """!Return projection tag according to version of capabilities (CRS/SRS).
  190. """
  191. return self.proj_tag
  192. def getxmlnshandler(self):
  193. """!Return WMSXMLNsHandler object.
  194. """
  195. return self.xml_ns
  196. class WMTSXMLNsHandler:
  197. """!Handle XML namespaces which are used in WMTS capabilities file.
  198. """
  199. def NsWmts(self, tag):
  200. """!Add namespace.
  201. """
  202. return "{http://www.opengis.net/wmts/1.0}" + tag
  203. def NsOws(self, tag):
  204. """!Add namespace.
  205. """
  206. return "{http://www.opengis.net/ows/1.1}" + tag
  207. class WMTSCapabilitiesTree(BaseCapabilitiesTree):
  208. def __init__(self, cap_file):
  209. """!Parses WMTS capabilities file.
  210. If the capabilities file cannot be parsed it raises xml.etree.ElementTree.ParseError.
  211. The class also removes elements which are in invalid form and are needed
  212. by wxGUI capabilities dialog or for creation of GetTile request by GRASS WMS library.
  213. @param cap_file - capabilities file
  214. """
  215. BaseCapabilitiesTree.__init__(self, cap_file)
  216. self.xml_ns = WMTSXMLNsHandler()
  217. grass.debug('Checking WMTS capabilities tree.', 4)
  218. contents = self._find(self.getroot(), 'Contents', self.xml_ns.NsWmts)
  219. tile_mat_sets = self._findall(contents, 'TileMatrixSet', self.xml_ns.NsWmts)
  220. for mat_set in tile_mat_sets:
  221. if not self._checkMatSet(mat_set):
  222. grass.debug('Removed invalid <TileMatrixSet> element.', 4)
  223. contents.remove(mat_set)
  224. # are there any <TileMatrixSet> elements after the check
  225. self._findall(contents, 'TileMatrixSet', self.xml_ns.NsWmts)
  226. layers = self._findall(contents, 'Layer', self.xml_ns.NsWmts)
  227. for l in layers:
  228. if not self._checkLayer(l):
  229. grass.debug('Removed invalid <Layer> element.', 4)
  230. contents.remove(l)
  231. # are there any <Layer> elements after the check
  232. self._findall(contents, 'Layer', self.xml_ns.NsWmts)
  233. grass.debug('Check of WMTS capabilities tree was finished.', 4)
  234. def _checkMatSet(self, mat_set):
  235. """!Check <TileMatrixSet>.
  236. """
  237. mat_set_id = mat_set.find(self.xml_ns.NsOws('Identifier'))
  238. if mat_set_id is None or not mat_set_id.text:
  239. return False
  240. mat_set_srs = mat_set.find(self.xml_ns.NsOws('SupportedCRS'))
  241. if mat_set_srs is None or \
  242. not mat_set_srs.text:
  243. return False
  244. tile_mats = mat_set.findall(self.xml_ns.NsWmts('TileMatrix'))
  245. if not tile_mats:
  246. return False
  247. for t_mat in tile_mats:
  248. if not self._checkMat(t_mat):
  249. grass.debug('Removed invalid <TileMatrix> element.', 4)
  250. mat_set.remove(t_mat)
  251. tile_mats = mat_set.findall(self.xml_ns.NsWmts('TileMatrix'))
  252. if not tile_mats:
  253. return False
  254. return True
  255. def _checkMat(self, t_mat):
  256. """!Check <TileMatrix>.
  257. """
  258. def _checkElement(t_mat, e, func):
  259. element = t_mat.find(self.xml_ns.NsWmts(e))
  260. if element is None or not element.text:
  261. return False
  262. try:
  263. e = func(element.text)
  264. except ValueError:
  265. return False
  266. if e < 0:
  267. return False
  268. return True
  269. for e, func in [['ScaleDenominator', float],
  270. ['TileWidth', int],
  271. ['TileHeight', int]]:
  272. if not _checkElement(t_mat, e, func):
  273. return False
  274. tile_mat_id = t_mat.find(self.xml_ns.NsOws('Identifier'))
  275. if tile_mat_id is None or not tile_mat_id.text:
  276. return False
  277. tl_str = t_mat.find(self.xml_ns.NsWmts('TopLeftCorner'))
  278. if tl_str is None or not tl_str.text:
  279. return False
  280. tl = tl_str.text.split(' ')
  281. if len(tl) < 2:
  282. return False
  283. for t in tl:
  284. try:
  285. t = float(t)
  286. except ValueError:
  287. return False
  288. return True
  289. def _checkLayer(self, layer):
  290. """!Check <Layer> element.
  291. """
  292. layer_id = layer.find(self.xml_ns.NsOws('Identifier'))
  293. if layer_id is None or not layer_id.text:
  294. return False
  295. mat_set_links = layer.findall(self.xml_ns.NsWmts('TileMatrixSetLink'))
  296. if not mat_set_links:
  297. return False
  298. styles = layer.findall(self.xml_ns.NsWmts('Style'))
  299. if not styles:
  300. return False
  301. for s in styles:
  302. s_name = s.find(self.xml_ns.NsOws('Identifier'))
  303. if s_name is None or not s_name.text:
  304. grass.debug('Removed invalid <Style> element.', 4)
  305. layer.remove(s_name)
  306. contents = self.getroot().find(self.xml_ns.NsWmts('Contents'))
  307. mat_sets = contents.findall(self.xml_ns.NsWmts('TileMatrixSet'))
  308. for link in mat_set_links:
  309. # <TileMatrixSetLink> does not point to existing <TileMatrixSet>
  310. if not self._checkMatSetLink(link, mat_sets):
  311. grass.debug('Removed invalid <TileMatrixSetLink> element.', 4)
  312. layer.remove(link)
  313. return True
  314. def _checkMatSetLink(self, link, mat_sets):
  315. """!Check <TileMatrixSetLink> element.
  316. """
  317. mat_set_link_id = link.find(self.xml_ns.NsWmts('TileMatrixSet')).text
  318. found = False
  319. for mat_set in mat_sets:
  320. mat_set_id = mat_set.find(self.xml_ns.NsOws('Identifier')).text
  321. if mat_set_id != mat_set_link_id:
  322. continue
  323. # the link points to existing <TileMatrixSet>
  324. found = True
  325. tile_mat_set_limits = link.find(self.xml_ns.NsWmts('TileMatrixSetLimits'))
  326. if tile_mat_set_limits is None:
  327. continue
  328. tile_mat_limits = tile_mat_set_limits.findall(self.xml_ns.NsWmts('TileMatrixLimits'))
  329. for limit in tile_mat_limits:
  330. if not self._checkMatSetLimit(limit):
  331. grass.debug('Removed invalid <TileMatrixLimits> element.', 4)
  332. tile_mat_limits.remove(limit)
  333. # are there any <TileMatrixLimits> elements after the check
  334. tile_mat_limits = tile_mat_set_limits.findall(self.xml_ns.NsWmts('TileMatrixLimits'))
  335. if not tile_mat_limits:
  336. grass.debug('Removed invalid <TileMatrixSetLimits> element.', 4)
  337. link.remove(tile_mat_set_limits)
  338. if not found:
  339. return False
  340. return True
  341. def _checkMatSetLimit(self, limit):
  342. """!Check <TileMatrixLimits> element.
  343. """
  344. limit_tile_mat = limit.find(self.xml_ns.NsWmts('TileMatrix'))
  345. if limit_tile_mat is None or not limit_tile_mat.text:
  346. return False
  347. for i in ['MinTileRow', 'MaxTileRow', 'MinTileCol', 'MaxTileCol']:
  348. i_tag = limit.find(self.xml_ns.NsWmts(i))
  349. if i_tag is None:
  350. return False
  351. try:
  352. int(i_tag.text)
  353. except ValueError:
  354. return False
  355. return True
  356. def _find(self, etreeElement, tag, ns = None):
  357. """!Find child element.
  358. If the element is not found it raises xml.etree.ElementTree.ParseError.
  359. """
  360. if not ns:
  361. res = etreeElement.find(tag)
  362. else:
  363. res = etreeElement.find(ns(tag))
  364. if res is None:
  365. raise etree.ParseError(_("Unable to parse capabilities file. \n\
  366. Tag '%s' was not found.") % tag)
  367. return res
  368. def _findall(self, etreeElement, tag, ns = None):
  369. """!Find all children element.
  370. If no element is found it raises xml.etree.ElementTree.ParseError.
  371. """
  372. if not ns:
  373. res = etreeElement.findall(tag)
  374. else:
  375. res = etreeElement.findall(ns(tag))
  376. if not res:
  377. raise etree.ParseError(_("Unable to parse capabilities file. \n\
  378. Tag '%s' was not found.") % tag)
  379. return res
  380. def getxmlnshandler(self):
  381. """!Return WMTSXMLNsHandler object.
  382. """
  383. return self.xml_ns
  384. class OnEarthCapabilitiesTree(BaseCapabilitiesTree):
  385. def __init__(self, cap_file):
  386. """!Parse NASA OnEarth tile service file.
  387. If the file cannot be parsed it raises xml.etree.ElementTree.ParseError.
  388. The class also removes elements which are in invalid form and are needed
  389. by wxGUI capabilities dialog or for creation of GetMap request by GRASS WMS library.
  390. @param cap_file - capabilities file
  391. """
  392. BaseCapabilitiesTree.__init__(self, cap_file)
  393. grass.debug('Checking OnEarth capabilities tree.', 4)
  394. self._checkLayerTree(self.getroot())
  395. grass.debug('Check if OnEarth capabilities tree was finished.', 4)
  396. def _checkLayerTree(self, parent_layer, first = True):
  397. """!Recursively check layer tree.
  398. """
  399. if first:
  400. tiled_patterns = self._find(parent_layer, 'TiledPatterns')
  401. layers = tiled_patterns.findall('TiledGroup')
  402. layers += tiled_patterns.findall('TiledGroups')
  403. parent_layer = tiled_patterns
  404. else:
  405. layers = parent_layer.findall('TiledGroup')
  406. layers += parent_layer.findall('TiledGroups')
  407. for l in layers:
  408. if not self._checkLayer(l):
  409. grass.debug(('Removed invalid <%s> element.' % l.tag), 4)
  410. parent_layer.remove(l)
  411. if l.tag == 'TiledGroups':
  412. self._checkLayerTree(l, False)
  413. def _find(self, etreeElement, tag):
  414. """!Find child element.
  415. If the element is not found it raises xml.etree.ElementTree.ParseError.
  416. """
  417. res = etreeElement.find(tag)
  418. if res is None:
  419. raise etree.ParseError(_("Unable to parse tile service file. \n\
  420. Tag <%s> was not found.") % tag)
  421. return res
  422. def _checkLayer(self, layer):
  423. """!Check <TiledGroup>/<TiledGroups> elements.
  424. """
  425. if layer.tag == 'TiledGroups':
  426. return True
  427. name = layer.find('Name')
  428. if name is None or not name.text:
  429. return False
  430. t_patts = layer.findall('TilePattern')
  431. for patt in t_patts:
  432. urls = self._getUrls(patt)
  433. for url in urls:
  434. if not self.gettilepatternurldata(url):
  435. urls.remove(url)
  436. # check if there are any vaild urls
  437. if not urls:
  438. grass.debug('<TilePattern> was removed. It has no valid url.', 4)
  439. layer.remove(patt)
  440. patt.text = '\n'.join(urls)
  441. t_patts = layer.findall('TilePattern')
  442. if not t_patts:
  443. return False
  444. return True
  445. def _getUrls(self, tile_pattern):
  446. """!Get all urls from tile pattern.
  447. """
  448. urls = []
  449. if tile_pattern.text is not None:
  450. tile_patt_lines = tile_pattern.text.split('\n')
  451. for line in tile_patt_lines:
  452. if 'request=GetMap' in line:
  453. urls.append(line.strip())
  454. return urls
  455. def gettilepatternurldata(self, url):
  456. """!Parse url string in Tile Pattern.
  457. """
  458. par_url = bbox = width = height = None
  459. bbox_idxs = self.geturlparamidxs(url, "bbox=")
  460. if bbox_idxs is None:
  461. return None
  462. par_url = [url[:bbox_idxs[0] - 1], url[bbox_idxs[1]:]]
  463. bbox = url[bbox_idxs[0] + len('bbox=') : bbox_idxs[1]]
  464. bbox_list = bbox.split(',')
  465. if len(bbox_list) < 4:
  466. return None
  467. try:
  468. bbox = map(float, bbox.split(','))
  469. except ValueError:
  470. return None
  471. width_idxs = self.geturlparamidxs(url, "width=")
  472. if width_idxs is None:
  473. return None
  474. try:
  475. width = int(url[width_idxs[0] + len('width=') : width_idxs[1]])
  476. except ValueError:
  477. return None
  478. height_idxs = self.geturlparamidxs(url, "height=")
  479. if height_idxs is None:
  480. return None
  481. try:
  482. height = int(url[height_idxs[0] + len('height=') : height_idxs[1]])
  483. except ValueError:
  484. return None
  485. if height < 0 or width < 0:
  486. return None
  487. return par_url, bbox, width, height
  488. def geturlparamidxs(self, params_str, param_key):
  489. """!Find start and end index of parameter and it's value in url string
  490. """
  491. start_i = params_str.lower().find(param_key)
  492. if start_i < 0:
  493. return None
  494. end_i = params_str.find("&", start_i)
  495. if end_i < 0:
  496. end_i = len(params_str)
  497. return (start_i, end_i)