wms_cap_parsers.py 22 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657
  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. # i18N
  22. import os
  23. import gettext
  24. gettext.install('grassmods', os.path.join(os.getenv("GISBASE"), 'locale'))
  25. class BaseCapabilitiesTree(etree.ElementTree):
  26. def __init__(self, cap_file):
  27. """!Initialize xml.etree.ElementTree
  28. """
  29. try:
  30. etree.ElementTree.__init__(self, file=cap_file)
  31. except ParseError:
  32. raise ParseError(_("Unable to parse XML file"))
  33. except IOError as error:
  34. raise ParseError(_("Unable to open XML file '%s'.\n%s\n" % (cap_file, error)))
  35. if self.getroot() is None:
  36. raise ParseError(_("Root node was not found."))
  37. class WMSXMLNsHandler:
  38. def __init__(self, caps):
  39. """!Handle XML namespaces according to WMS version of capabilities.
  40. """
  41. self.namespace = "{http://www.opengis.net/wms}"
  42. if caps.getroot().find("Service") is not None:
  43. self.use_ns = False
  44. elif caps.getroot().find(self.namespace + "Service") is not None:
  45. self.use_ns = True
  46. else:
  47. raise ParseError(_("Unable to parse capabilities file.\n\
  48. Tag <%s> was not found.") % "Service")
  49. def Ns(self, tag_name):
  50. """!Add namespace to tag_name according to version
  51. """
  52. if self.use_ns:
  53. tag_name = self.namespace + tag_name
  54. return tag_name
  55. class WMSCapabilitiesTree(BaseCapabilitiesTree):
  56. def __init__(self, cap_file, force_version=None):
  57. """!Parses WMS capabilities file.
  58. If the capabilities file cannot be parsed if it raises xml.etree.ElementTree.ParseError.
  59. The class manges inheritance in 'Layer' elements. Inherited elements
  60. are added to 'Layer' element.
  61. The class also removes elements which are in invalid form and are needed
  62. by wxGUI capabilities dialog.
  63. @param cap_file - capabilities file
  64. @param force_version - force capabilities file version (1.1.1, 1.3.0)
  65. """
  66. BaseCapabilitiesTree.__init__(self, cap_file)
  67. self.xml_ns = WMSXMLNsHandler(self)
  68. grass.debug('Checking WMS capabilities tree.', 4)
  69. if not "version" in self.getroot().attrib:
  70. raise ParseError(_("Missing version attribute root node "
  71. "in Capabilities XML file"))
  72. else:
  73. wms_version = self.getroot().attrib["version"]
  74. if wms_version == "1.3.0":
  75. self.proj_tag = "CRS"
  76. else:
  77. self.proj_tag = "SRS"
  78. if force_version is not None:
  79. if wms_version != force_version:
  80. raise ParseError(_("WMS server does not support '%s' version.") % wms_version)
  81. capability = self._find(self.getroot(), "Capability")
  82. root_layer = self._find(capability, "Layer")
  83. self._checkFormats(capability)
  84. self._checkLayerTree(root_layer)
  85. grass.debug('Check of WMS capabilities tree was finished.', 4)
  86. def _checkFormats(self, capability):
  87. """!Check if format element is defined.
  88. """
  89. request = self._find(capability, "Request")
  90. get_map = self._find(request, "GetMap")
  91. formats = self._findall(get_map, "Format")
  92. def _checkLayerTree(self, parent_layer, first=True):
  93. """!Recursively check layer tree and manage inheritance in the tree
  94. """
  95. if first:
  96. self._initLayer(parent_layer, None)
  97. layers = parent_layer.findall((self.xml_ns.Ns("Layer")))
  98. for l in layers:
  99. self._initLayer(l, parent_layer)
  100. self._checkLayerTree(l, False)
  101. def _initLayer(self, layer, parent_layer):
  102. """Inherit elements from parent layer
  103. @param layer - <Layer> element which inherits
  104. @param parent_layer - <Layer> element which is inherited from
  105. """
  106. if parent_layer is not None:
  107. replaced_elements = [["EX_GeographicBoundingBox", "replace"],
  108. ["Attribution", "replace"],
  109. ["MinScaleDenominator", "replace"],
  110. ["MaxScaleDenominator", "replace"],
  111. ["AuthorityURL", "add"]]
  112. for element in replaced_elements:
  113. elems = layer.findall(self.xml_ns.Ns(element[0]))
  114. if len(elems) != 0 or element[1] == "add":
  115. for e in parent_layer.findall(self.xml_ns.Ns(element[0])):
  116. layer.append(e)
  117. inh_arguments = ["queryable", "cascaded", "opaque",
  118. "noSubsets", "fixedWidth", "fixedHeight"]
  119. for attr in parent_layer.attrib:
  120. if attr not in layer.attrib and attr in inh_arguments:
  121. layer.attrib[attr] = parent_layer.attrib[attr]
  122. self._inhNotSame(self.proj_tag, "element_content", layer, parent_layer)
  123. self._inhNotSame("BoundingBox", "attribute", layer, parent_layer, self.proj_tag)
  124. # remove invalid Styles
  125. styles = layer.findall(self.xml_ns.Ns('Style'))
  126. for s in styles:
  127. s_name = s.find(self.xml_ns.Ns('Name'))
  128. if s_name is None or not s_name.text:
  129. grass.debug('Removed invalid <Style> element.', 4)
  130. layer.remove(s)
  131. self._inhNotSame("Style", "child_element_content", layer, parent_layer, "Name")
  132. self._inhNotSame("Dimension", "attribute", layer, parent_layer, "name")
  133. def _inhNotSame(self, element_name, cmp_type, layer, parent_layer, add_arg=None):
  134. """Inherit elements which have unique values.
  135. @param element_name - name of inherited element
  136. @param cmp_type - 'element_content' - compared value is text of <Layer> element
  137. @param cmp_type - 'child_element_content' - compared value is text of a child of the <Layer> element
  138. @param cmp_type - 'attribute' - compared value is text of the <Layer> element attribute
  139. @param layer - <Layer> element which inherits
  140. @param parent_layer - <Layer> element which is inherited from
  141. @param add_arg - name of child element or attribute
  142. """
  143. elem = layer.findall(self.xml_ns.Ns(element_name))
  144. parent_elems = []
  145. if parent_layer is not None:
  146. parent_elems = parent_layer.findall(self.xml_ns.Ns(element_name))
  147. for par_elem in parent_elems:
  148. parent_cmp_text = None
  149. if cmp_type == "attribute":
  150. if add_arg in par_elem.attrib:
  151. parent_cmp_text = par_elem.attrib[add_arg]
  152. elif cmp_type == "element_content":
  153. parent_cmp_text = par_elem.text
  154. elif cmp_type == "child_element_content":
  155. parent_cmp = par_elem.find(self.xml_ns.Ns(add_arg))
  156. if parent_cmp is not None:
  157. parent_cmp_text = parent_cmp.text
  158. if parent_cmp_text is None:
  159. continue
  160. is_there = False
  161. for elem in elem:
  162. cmp_text = None
  163. if cmp_type == "attribute":
  164. if add_arg in elem.attrib:
  165. cmp_text = elem.attrib[add_arg]
  166. elif cmp_type == "element_content":
  167. cmp_text = elem.text
  168. elif cmp_type == "child_element_content":
  169. cmp = elem.find(self.xml_ns.Ns(add_arg))
  170. if cmp is not None:
  171. cmp_text = cmp.text
  172. if cmp_text is None or \
  173. cmp_text.lower() == parent_cmp_text.lower():
  174. is_there = True
  175. break
  176. if not is_there:
  177. layer.append(par_elem)
  178. def _find(self, etreeElement, tag):
  179. """!Find child element.
  180. If the element is not found it raises xml.etree.ElementTree.ParseError.
  181. """
  182. res = etreeElement.find(self.xml_ns.Ns(tag))
  183. if res is None:
  184. raise ParseError(_("Unable to parse capabilities file. \n\
  185. Tag <%s> was not found.") % tag)
  186. return res
  187. def _findall(self, etreeElement, tag):
  188. """!Find all children element.
  189. If no element is found it raises xml.etree.ElementTree.ParseError.
  190. """
  191. res = etreeElement.findall(self.xml_ns.Ns(tag))
  192. if not res:
  193. raise ParseError(_("Unable to parse capabilities file. \n\
  194. Tag <%s> was not found.") % tag)
  195. return res
  196. def getprojtag(self):
  197. """!Return projection tag according to version of capabilities (CRS/SRS).
  198. """
  199. return self.proj_tag
  200. def getxmlnshandler(self):
  201. """!Return WMSXMLNsHandler object.
  202. """
  203. return self.xml_ns
  204. class WMTSXMLNsHandler:
  205. """!Handle XML namespaces which are used in WMTS capabilities file.
  206. """
  207. def NsWmts(self, tag):
  208. """!Add namespace.
  209. """
  210. return "{http://www.opengis.net/wmts/1.0}" + tag
  211. def NsOws(self, tag):
  212. """!Add namespace.
  213. """
  214. return "{http://www.opengis.net/ows/1.1}" + tag
  215. class WMTSCapabilitiesTree(BaseCapabilitiesTree):
  216. def __init__(self, cap_file):
  217. """!Parses WMTS capabilities file.
  218. If the capabilities file cannot be parsed it raises xml.etree.ElementTree.ParseError.
  219. The class also removes elements which are in invalid form and are needed
  220. by wxGUI capabilities dialog or for creation of GetTile request by GRASS WMS library.
  221. @param cap_file - capabilities file
  222. """
  223. BaseCapabilitiesTree.__init__(self, cap_file)
  224. self.xml_ns = WMTSXMLNsHandler()
  225. grass.debug('Checking WMTS capabilities tree.', 4)
  226. contents = self._find(self.getroot(), 'Contents', self.xml_ns.NsWmts)
  227. tile_mat_sets = self._findall(contents, 'TileMatrixSet', self.xml_ns.NsWmts)
  228. for mat_set in tile_mat_sets:
  229. if not self._checkMatSet(mat_set):
  230. grass.debug('Removed invalid <TileMatrixSet> element.', 4)
  231. contents.remove(mat_set)
  232. # are there any <TileMatrixSet> elements after the check
  233. self._findall(contents, 'TileMatrixSet', self.xml_ns.NsWmts)
  234. layers = self._findall(contents, 'Layer', self.xml_ns.NsWmts)
  235. for l in layers:
  236. if not self._checkLayer(l):
  237. grass.debug('Removed invalid <Layer> element.', 4)
  238. contents.remove(l)
  239. # are there any <Layer> elements after the check
  240. self._findall(contents, 'Layer', self.xml_ns.NsWmts)
  241. grass.debug('Check of WMTS capabilities tree was finished.', 4)
  242. def _checkMatSet(self, mat_set):
  243. """!Check <TileMatrixSet>.
  244. """
  245. mat_set_id = mat_set.find(self.xml_ns.NsOws('Identifier'))
  246. if mat_set_id is None or not mat_set_id.text:
  247. return False
  248. mat_set_srs = mat_set.find(self.xml_ns.NsOws('SupportedCRS'))
  249. if mat_set_srs is None or \
  250. not mat_set_srs.text:
  251. return False
  252. tile_mats = mat_set.findall(self.xml_ns.NsWmts('TileMatrix'))
  253. if not tile_mats:
  254. return False
  255. for t_mat in tile_mats:
  256. if not self._checkMat(t_mat):
  257. grass.debug('Removed invalid <TileMatrix> element.', 4)
  258. mat_set.remove(t_mat)
  259. tile_mats = mat_set.findall(self.xml_ns.NsWmts('TileMatrix'))
  260. if not tile_mats:
  261. return False
  262. return True
  263. def _checkMat(self, t_mat):
  264. """!Check <TileMatrix>.
  265. """
  266. def _checkElement(t_mat, e, func):
  267. element = t_mat.find(self.xml_ns.NsWmts(e))
  268. if element is None or not element.text:
  269. return False
  270. try:
  271. e = func(element.text)
  272. except ValueError:
  273. return False
  274. if e < 0:
  275. return False
  276. return True
  277. for e, func in [['ScaleDenominator', float],
  278. ['TileWidth', int],
  279. ['TileHeight', int]]:
  280. if not _checkElement(t_mat, e, func):
  281. return False
  282. tile_mat_id = t_mat.find(self.xml_ns.NsOws('Identifier'))
  283. if tile_mat_id is None or not tile_mat_id.text:
  284. return False
  285. tl_str = t_mat.find(self.xml_ns.NsWmts('TopLeftCorner'))
  286. if tl_str is None or not tl_str.text:
  287. return False
  288. tl = tl_str.text.split(' ')
  289. if len(tl) < 2:
  290. return False
  291. for t in tl:
  292. try:
  293. t = float(t)
  294. except ValueError:
  295. return False
  296. return True
  297. def _checkLayer(self, layer):
  298. """!Check <Layer> element.
  299. """
  300. layer_id = layer.find(self.xml_ns.NsOws('Identifier'))
  301. if layer_id is None or not layer_id.text:
  302. return False
  303. mat_set_links = layer.findall(self.xml_ns.NsWmts('TileMatrixSetLink'))
  304. if not mat_set_links:
  305. return False
  306. styles = layer.findall(self.xml_ns.NsWmts('Style'))
  307. if not styles:
  308. return False
  309. for s in styles:
  310. s_name = s.find(self.xml_ns.NsOws('Identifier'))
  311. if s_name is None or not s_name.text:
  312. grass.debug('Removed invalid <Style> element.', 4)
  313. layer.remove(s_name)
  314. contents = self.getroot().find(self.xml_ns.NsWmts('Contents'))
  315. mat_sets = contents.findall(self.xml_ns.NsWmts('TileMatrixSet'))
  316. for link in mat_set_links:
  317. # <TileMatrixSetLink> does not point to existing <TileMatrixSet>
  318. if not self._checkMatSetLink(link, mat_sets):
  319. grass.debug('Removed invalid <TileMatrixSetLink> element.', 4)
  320. layer.remove(link)
  321. return True
  322. def _checkMatSetLink(self, link, mat_sets):
  323. """!Check <TileMatrixSetLink> element.
  324. """
  325. mat_set_link_id = link.find(self.xml_ns.NsWmts('TileMatrixSet')).text
  326. found = False
  327. for mat_set in mat_sets:
  328. mat_set_id = mat_set.find(self.xml_ns.NsOws('Identifier')).text
  329. if mat_set_id != mat_set_link_id:
  330. continue
  331. # the link points to existing <TileMatrixSet>
  332. found = True
  333. tile_mat_set_limits = link.find(self.xml_ns.NsWmts('TileMatrixSetLimits'))
  334. if tile_mat_set_limits is None:
  335. continue
  336. tile_mat_limits = tile_mat_set_limits.findall(self.xml_ns.NsWmts('TileMatrixLimits'))
  337. for limit in tile_mat_limits:
  338. if not self._checkMatSetLimit(limit):
  339. grass.debug('Removed invalid <TileMatrixLimits> element.', 4)
  340. tile_mat_limits.remove(limit)
  341. # are there any <TileMatrixLimits> elements after the check
  342. tile_mat_limits = tile_mat_set_limits.findall(self.xml_ns.NsWmts('TileMatrixLimits'))
  343. if not tile_mat_limits:
  344. grass.debug('Removed invalid <TileMatrixSetLimits> element.', 4)
  345. link.remove(tile_mat_set_limits)
  346. if not found:
  347. return False
  348. return True
  349. def _checkMatSetLimit(self, limit):
  350. """!Check <TileMatrixLimits> element.
  351. """
  352. limit_tile_mat = limit.find(self.xml_ns.NsWmts('TileMatrix'))
  353. if limit_tile_mat is None or not limit_tile_mat.text:
  354. return False
  355. for i in ['MinTileRow', 'MaxTileRow', 'MinTileCol', 'MaxTileCol']:
  356. i_tag = limit.find(self.xml_ns.NsWmts(i))
  357. if i_tag is None:
  358. return False
  359. try:
  360. int(i_tag.text)
  361. except ValueError:
  362. return False
  363. return True
  364. def _find(self, etreeElement, tag, ns=None):
  365. """!Find child element.
  366. If the element is not found it raises xml.etree.ElementTree.ParseError.
  367. """
  368. if not ns:
  369. res = etreeElement.find(tag)
  370. else:
  371. res = etreeElement.find(ns(tag))
  372. if res is None:
  373. raise ParseError(_("Unable to parse capabilities file. \n\
  374. Tag '%s' was not found.") % tag)
  375. return res
  376. def _findall(self, etreeElement, tag, ns=None):
  377. """!Find all children element.
  378. If no element is found it raises xml.etree.ElementTree.ParseError.
  379. """
  380. if not ns:
  381. res = etreeElement.findall(tag)
  382. else:
  383. res = etreeElement.findall(ns(tag))
  384. if not res:
  385. raise ParseError(_("Unable to parse capabilities file. \n\
  386. Tag '%s' was not found.") % tag)
  387. return res
  388. def getxmlnshandler(self):
  389. """!Return WMTSXMLNsHandler object.
  390. """
  391. return self.xml_ns
  392. class OnEarthCapabilitiesTree(BaseCapabilitiesTree):
  393. def __init__(self, cap_file):
  394. """!Parse NASA OnEarth tile service file.
  395. If the file cannot be parsed it raises xml.etree.ElementTree.ParseError.
  396. The class also removes elements which are in invalid form and are needed
  397. by wxGUI capabilities dialog or for creation of GetMap request by GRASS WMS library.
  398. @param cap_file - capabilities file
  399. """
  400. BaseCapabilitiesTree.__init__(self, cap_file)
  401. grass.debug('Checking OnEarth capabilities tree.', 4)
  402. self._checkLayerTree(self.getroot())
  403. grass.debug('Check if OnEarth capabilities tree was finished.', 4)
  404. def _checkLayerTree(self, parent_layer, first=True):
  405. """!Recursively check layer tree.
  406. """
  407. if first:
  408. tiled_patterns = self._find(parent_layer, 'TiledPatterns')
  409. layers = tiled_patterns.findall('TiledGroup')
  410. layers += tiled_patterns.findall('TiledGroups')
  411. parent_layer = tiled_patterns
  412. else:
  413. layers = parent_layer.findall('TiledGroup')
  414. layers += parent_layer.findall('TiledGroups')
  415. for l in layers:
  416. if not self._checkLayer(l):
  417. grass.debug(('Removed invalid <%s> element.' % l.tag), 4)
  418. parent_layer.remove(l)
  419. if l.tag == 'TiledGroups':
  420. self._checkLayerTree(l, False)
  421. def _find(self, etreeElement, tag):
  422. """!Find child element.
  423. If the element is not found it raises xml.etree.ElementTree.ParseError.
  424. """
  425. res = etreeElement.find(tag)
  426. if res is None:
  427. raise ParseError(_("Unable to parse tile service file. \n\
  428. Tag <%s> was not found.") % tag)
  429. return res
  430. def _checkLayer(self, layer):
  431. """!Check <TiledGroup>/<TiledGroups> elements.
  432. """
  433. if layer.tag == 'TiledGroups':
  434. return True
  435. name = layer.find('Name')
  436. if name is None or not name.text:
  437. return False
  438. t_patts = layer.findall('TilePattern')
  439. for patt in t_patts:
  440. urls = self._getUrls(patt)
  441. for url in urls:
  442. if not self.gettilepatternurldata(url):
  443. urls.remove(url)
  444. # check if there are any valid urls
  445. if not urls:
  446. grass.debug('<TilePattern> was removed. It has no valid url.', 4)
  447. layer.remove(patt)
  448. patt.text = '\n'.join(urls)
  449. t_patts = layer.findall('TilePattern')
  450. if not t_patts:
  451. return False
  452. return True
  453. def _getUrls(self, tile_pattern):
  454. """!Get all urls from tile pattern.
  455. """
  456. urls = []
  457. if tile_pattern.text is not None:
  458. tile_patt_lines = tile_pattern.text.split('\n')
  459. for line in tile_patt_lines:
  460. if 'request=GetMap' in line:
  461. urls.append(line.strip())
  462. return urls
  463. def gettilepatternurldata(self, url):
  464. """!Parse url string in Tile Pattern.
  465. """
  466. par_url = bbox = width = height = None
  467. bbox_idxs = self.geturlparamidxs(url, "bbox=")
  468. if bbox_idxs is None:
  469. return None
  470. par_url = [url[:bbox_idxs[0] - 1], url[bbox_idxs[1]:]]
  471. bbox = url[bbox_idxs[0] + len('bbox='): bbox_idxs[1]]
  472. bbox_list = bbox.split(',')
  473. if len(bbox_list) < 4:
  474. return None
  475. try:
  476. bbox = list(map(float, bbox.split(',')))
  477. except ValueError:
  478. return None
  479. width_idxs = self.geturlparamidxs(url, "width=")
  480. if width_idxs is None:
  481. return None
  482. try:
  483. width = int(url[width_idxs[0] + len('width='): width_idxs[1]])
  484. except ValueError:
  485. return None
  486. height_idxs = self.geturlparamidxs(url, "height=")
  487. if height_idxs is None:
  488. return None
  489. try:
  490. height = int(url[height_idxs[0] + len('height='): height_idxs[1]])
  491. except ValueError:
  492. return None
  493. if height < 0 or width < 0:
  494. return None
  495. return par_url, bbox, width, height
  496. def geturlparamidxs(self, params_str, param_key):
  497. """!Find start and end index of parameter and it's value in url string
  498. """
  499. start_i = params_str.lower().find(param_key)
  500. if start_i < 0:
  501. return None
  502. end_i = params_str.find("&", start_i)
  503. if end_i < 0:
  504. end_i = len(params_str)
  505. return (start_i, end_i)