wms_cap_parsers.py 22 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670
  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 pathlib
  16. try:
  17. from xml.etree.ElementTree import ParseError
  18. except ImportError: # < Python 2.7
  19. from xml.parsers.expat import ExpatError as ParseError
  20. import xml.etree.ElementTree as etree
  21. import grass.script as grass
  22. class BaseCapabilitiesTree(etree.ElementTree):
  23. def __init__(self, cap_file):
  24. """!Initialize xml.etree.ElementTree
  25. """
  26. is_file = False
  27. try:
  28. xml = pathlib.Path(cap_file)
  29. if xml.exists():
  30. is_file = True
  31. except OSError as exc:
  32. if exc.errno == 36: # file name too long
  33. pass
  34. else:
  35. raise
  36. if is_file:
  37. try:
  38. etree.ElementTree.__init__(self, file=cap_file)
  39. except ParseError:
  40. raise ParseError(_("Unable to parse XML file"))
  41. except IOError as error:
  42. raise ParseError(_("Unable to open XML file '%s'.\n%s\n" % (cap_file, error)))
  43. else:
  44. try:
  45. etree.ElementTree.__init__(self, element=etree.fromstring(cap_file))
  46. except ParseError:
  47. raise ParseError(_("Unable to parse XML file"))
  48. if self.getroot() is None:
  49. raise ParseError(_("Root node was not found."))
  50. class WMSXMLNsHandler:
  51. def __init__(self, caps):
  52. """!Handle XML namespaces according to WMS version of capabilities.
  53. """
  54. self.namespace = "{http://www.opengis.net/wms}"
  55. if caps.getroot().find("Service") is not None:
  56. self.use_ns = False
  57. elif caps.getroot().find(self.namespace + "Service") is not None:
  58. self.use_ns = True
  59. else:
  60. raise ParseError(_("Unable to parse capabilities file.\n\
  61. Tag <%s> was not found.") % "Service")
  62. def Ns(self, tag_name):
  63. """!Add namespace to tag_name according to version
  64. """
  65. if self.use_ns:
  66. tag_name = self.namespace + tag_name
  67. return tag_name
  68. class WMSCapabilitiesTree(BaseCapabilitiesTree):
  69. def __init__(self, cap_file, force_version=None):
  70. """!Parses WMS capabilities file.
  71. If the capabilities file cannot be parsed if it raises xml.etree.ElementTree.ParseError.
  72. The class manges inheritance in 'Layer' elements. Inherited elements
  73. are added to 'Layer' element.
  74. The class also removes elements which are in invalid form and are needed
  75. by wxGUI capabilities dialog.
  76. @param cap_file - capabilities file
  77. @param force_version - force capabilities file version (1.1.1, 1.3.0)
  78. """
  79. BaseCapabilitiesTree.__init__(self, cap_file)
  80. self.xml_ns = WMSXMLNsHandler(self)
  81. grass.debug('Checking WMS capabilities tree.', 4)
  82. if not "version" in self.getroot().attrib:
  83. raise ParseError(_("Missing version attribute root node "
  84. "in Capabilities XML file"))
  85. else:
  86. wms_version = self.getroot().attrib["version"]
  87. if wms_version == "1.3.0":
  88. self.proj_tag = "CRS"
  89. else:
  90. self.proj_tag = "SRS"
  91. if force_version is not None:
  92. if wms_version != force_version:
  93. raise ParseError(_("WMS server does not support '%s' version.") % wms_version)
  94. capability = self._find(self.getroot(), "Capability")
  95. root_layer = self._find(capability, "Layer")
  96. self._checkFormats(capability)
  97. self._checkLayerTree(root_layer)
  98. grass.debug('Check of WMS capabilities tree was finished.', 4)
  99. def _checkFormats(self, capability):
  100. """!Check if format element is defined.
  101. """
  102. request = self._find(capability, "Request")
  103. get_map = self._find(request, "GetMap")
  104. formats = self._findall(get_map, "Format")
  105. def _checkLayerTree(self, parent_layer, first=True):
  106. """!Recursively check layer tree and manage inheritance in the tree
  107. """
  108. if first:
  109. self._initLayer(parent_layer, None)
  110. layers = parent_layer.findall((self.xml_ns.Ns("Layer")))
  111. for l in layers:
  112. self._initLayer(l, parent_layer)
  113. self._checkLayerTree(l, False)
  114. def _initLayer(self, layer, parent_layer):
  115. """Inherit elements from parent layer
  116. @param layer - <Layer> element which inherits
  117. @param parent_layer - <Layer> element which is inherited from
  118. """
  119. if parent_layer is not None:
  120. replaced_elements = [["EX_GeographicBoundingBox", "replace"],
  121. ["Attribution", "replace"],
  122. ["MinScaleDenominator", "replace"],
  123. ["MaxScaleDenominator", "replace"],
  124. ["AuthorityURL", "add"]]
  125. for element in replaced_elements:
  126. elems = layer.findall(self.xml_ns.Ns(element[0]))
  127. if len(elems) != 0 or element[1] == "add":
  128. for e in parent_layer.findall(self.xml_ns.Ns(element[0])):
  129. layer.append(e)
  130. inh_arguments = ["queryable", "cascaded", "opaque",
  131. "noSubsets", "fixedWidth", "fixedHeight"]
  132. for attr in parent_layer.attrib:
  133. if attr not in layer.attrib and attr in inh_arguments:
  134. layer.attrib[attr] = parent_layer.attrib[attr]
  135. self._inhNotSame(self.proj_tag, "element_content", layer, parent_layer)
  136. self._inhNotSame("BoundingBox", "attribute", layer, parent_layer, self.proj_tag)
  137. # remove invalid Styles
  138. styles = layer.findall(self.xml_ns.Ns('Style'))
  139. for s in styles:
  140. s_name = s.find(self.xml_ns.Ns('Name'))
  141. if s_name is None or not s_name.text:
  142. grass.debug('Removed invalid <Style> element.', 4)
  143. layer.remove(s)
  144. self._inhNotSame("Style", "child_element_content", layer, parent_layer, "Name")
  145. self._inhNotSame("Dimension", "attribute", layer, parent_layer, "name")
  146. def _inhNotSame(self, element_name, cmp_type, layer, parent_layer, add_arg=None):
  147. """Inherit elements which have unique values.
  148. @param element_name - name of inherited element
  149. @param cmp_type - 'element_content' - compared value is text of <Layer> element
  150. @param cmp_type - 'child_element_content' - compared value is text of a child of the <Layer> element
  151. @param cmp_type - 'attribute' - compared value is text of the <Layer> element attribute
  152. @param layer - <Layer> element which inherits
  153. @param parent_layer - <Layer> element which is inherited from
  154. @param add_arg - name of child element or attribute
  155. """
  156. elem = layer.findall(self.xml_ns.Ns(element_name))
  157. parent_elems = []
  158. if parent_layer is not None:
  159. parent_elems = parent_layer.findall(self.xml_ns.Ns(element_name))
  160. for par_elem in parent_elems:
  161. parent_cmp_text = None
  162. if cmp_type == "attribute":
  163. if add_arg in par_elem.attrib:
  164. parent_cmp_text = par_elem.attrib[add_arg]
  165. elif cmp_type == "element_content":
  166. parent_cmp_text = par_elem.text
  167. elif cmp_type == "child_element_content":
  168. parent_cmp = par_elem.find(self.xml_ns.Ns(add_arg))
  169. if parent_cmp is not None:
  170. parent_cmp_text = parent_cmp.text
  171. if parent_cmp_text is None:
  172. continue
  173. is_there = False
  174. for elem in elem:
  175. cmp_text = None
  176. if cmp_type == "attribute":
  177. if add_arg in elem.attrib:
  178. cmp_text = elem.attrib[add_arg]
  179. elif cmp_type == "element_content":
  180. cmp_text = elem.text
  181. elif cmp_type == "child_element_content":
  182. cmp = elem.find(self.xml_ns.Ns(add_arg))
  183. if cmp is not None:
  184. cmp_text = cmp.text
  185. if cmp_text is None or \
  186. cmp_text.lower() == parent_cmp_text.lower():
  187. is_there = True
  188. break
  189. if not is_there:
  190. layer.append(par_elem)
  191. def _find(self, etreeElement, tag):
  192. """!Find child element.
  193. If the element is not found it raises xml.etree.ElementTree.ParseError.
  194. """
  195. res = etreeElement.find(self.xml_ns.Ns(tag))
  196. if res is None:
  197. raise ParseError(_("Unable to parse capabilities file. \n\
  198. Tag <%s> was not found.") % tag)
  199. return res
  200. def _findall(self, etreeElement, tag):
  201. """!Find all children element.
  202. If no element is found it raises xml.etree.ElementTree.ParseError.
  203. """
  204. res = etreeElement.findall(self.xml_ns.Ns(tag))
  205. if not res:
  206. raise ParseError(_("Unable to parse capabilities file. \n\
  207. Tag <%s> was not found.") % tag)
  208. return res
  209. def getprojtag(self):
  210. """!Return projection tag according to version of capabilities (CRS/SRS).
  211. """
  212. return self.proj_tag
  213. def getxmlnshandler(self):
  214. """!Return WMSXMLNsHandler object.
  215. """
  216. return self.xml_ns
  217. class WMTSXMLNsHandler:
  218. """!Handle XML namespaces which are used in WMTS capabilities file.
  219. """
  220. def NsWmts(self, tag):
  221. """!Add namespace.
  222. """
  223. return "{http://www.opengis.net/wmts/1.0}" + tag
  224. def NsOws(self, tag):
  225. """!Add namespace.
  226. """
  227. return "{http://www.opengis.net/ows/1.1}" + tag
  228. class WMTSCapabilitiesTree(BaseCapabilitiesTree):
  229. def __init__(self, cap_file):
  230. """!Parses WMTS capabilities file.
  231. If the capabilities file cannot be parsed it raises xml.etree.ElementTree.ParseError.
  232. The class also removes elements which are in invalid form and are needed
  233. by wxGUI capabilities dialog or for creation of GetTile request by GRASS WMS library.
  234. @param cap_file - capabilities file
  235. """
  236. BaseCapabilitiesTree.__init__(self, cap_file)
  237. self.xml_ns = WMTSXMLNsHandler()
  238. grass.debug('Checking WMTS capabilities tree.', 4)
  239. contents = self._find(self.getroot(), 'Contents', self.xml_ns.NsWmts)
  240. tile_mat_sets = self._findall(contents, 'TileMatrixSet', self.xml_ns.NsWmts)
  241. for mat_set in tile_mat_sets:
  242. if not self._checkMatSet(mat_set):
  243. grass.debug('Removed invalid <TileMatrixSet> element.', 4)
  244. contents.remove(mat_set)
  245. # are there any <TileMatrixSet> elements after the check
  246. self._findall(contents, 'TileMatrixSet', self.xml_ns.NsWmts)
  247. layers = self._findall(contents, 'Layer', self.xml_ns.NsWmts)
  248. for l in layers:
  249. if not self._checkLayer(l):
  250. grass.debug('Removed invalid <Layer> element.', 4)
  251. contents.remove(l)
  252. # are there any <Layer> elements after the check
  253. self._findall(contents, 'Layer', self.xml_ns.NsWmts)
  254. grass.debug('Check of WMTS capabilities tree was finished.', 4)
  255. def _checkMatSet(self, mat_set):
  256. """!Check <TileMatrixSet>.
  257. """
  258. mat_set_id = mat_set.find(self.xml_ns.NsOws('Identifier'))
  259. if mat_set_id is None or not mat_set_id.text:
  260. return False
  261. mat_set_srs = mat_set.find(self.xml_ns.NsOws('SupportedCRS'))
  262. if mat_set_srs is None or \
  263. not mat_set_srs.text:
  264. return False
  265. tile_mats = mat_set.findall(self.xml_ns.NsWmts('TileMatrix'))
  266. if not tile_mats:
  267. return False
  268. for t_mat in tile_mats:
  269. if not self._checkMat(t_mat):
  270. grass.debug('Removed invalid <TileMatrix> element.', 4)
  271. mat_set.remove(t_mat)
  272. tile_mats = mat_set.findall(self.xml_ns.NsWmts('TileMatrix'))
  273. if not tile_mats:
  274. return False
  275. return True
  276. def _checkMat(self, t_mat):
  277. """!Check <TileMatrix>.
  278. """
  279. def _checkElement(t_mat, e, func):
  280. element = t_mat.find(self.xml_ns.NsWmts(e))
  281. if element is None or not element.text:
  282. return False
  283. try:
  284. e = func(element.text)
  285. except ValueError:
  286. return False
  287. if e < 0:
  288. return False
  289. return True
  290. for e, func in [['ScaleDenominator', float],
  291. ['TileWidth', int],
  292. ['TileHeight', int]]:
  293. if not _checkElement(t_mat, e, func):
  294. return False
  295. tile_mat_id = t_mat.find(self.xml_ns.NsOws('Identifier'))
  296. if tile_mat_id is None or not tile_mat_id.text:
  297. return False
  298. tl_str = t_mat.find(self.xml_ns.NsWmts('TopLeftCorner'))
  299. if tl_str is None or not tl_str.text:
  300. return False
  301. tl = tl_str.text.split(' ')
  302. if len(tl) < 2:
  303. return False
  304. for t in tl:
  305. try:
  306. t = float(t)
  307. except ValueError:
  308. return False
  309. return True
  310. def _checkLayer(self, layer):
  311. """!Check <Layer> element.
  312. """
  313. layer_id = layer.find(self.xml_ns.NsOws('Identifier'))
  314. if layer_id is None or not layer_id.text:
  315. return False
  316. mat_set_links = layer.findall(self.xml_ns.NsWmts('TileMatrixSetLink'))
  317. if not mat_set_links:
  318. return False
  319. styles = layer.findall(self.xml_ns.NsWmts('Style'))
  320. if not styles:
  321. return False
  322. for s in styles:
  323. s_name = s.find(self.xml_ns.NsOws('Identifier'))
  324. if s_name is None or not s_name.text:
  325. grass.debug('Removed invalid <Style> element.', 4)
  326. layer.remove(s_name)
  327. contents = self.getroot().find(self.xml_ns.NsWmts('Contents'))
  328. mat_sets = contents.findall(self.xml_ns.NsWmts('TileMatrixSet'))
  329. for link in mat_set_links:
  330. # <TileMatrixSetLink> does not point to existing <TileMatrixSet>
  331. if not self._checkMatSetLink(link, mat_sets):
  332. grass.debug('Removed invalid <TileMatrixSetLink> element.', 4)
  333. layer.remove(link)
  334. return True
  335. def _checkMatSetLink(self, link, mat_sets):
  336. """!Check <TileMatrixSetLink> element.
  337. """
  338. mat_set_link_id = link.find(self.xml_ns.NsWmts('TileMatrixSet')).text
  339. found = False
  340. for mat_set in mat_sets:
  341. mat_set_id = mat_set.find(self.xml_ns.NsOws('Identifier')).text
  342. if mat_set_id != mat_set_link_id:
  343. continue
  344. # the link points to existing <TileMatrixSet>
  345. found = True
  346. tile_mat_set_limits = link.find(self.xml_ns.NsWmts('TileMatrixSetLimits'))
  347. if tile_mat_set_limits is None:
  348. continue
  349. tile_mat_limits = tile_mat_set_limits.findall(self.xml_ns.NsWmts('TileMatrixLimits'))
  350. for limit in tile_mat_limits:
  351. if not self._checkMatSetLimit(limit):
  352. grass.debug('Removed invalid <TileMatrixLimits> element.', 4)
  353. tile_mat_limits.remove(limit)
  354. # are there any <TileMatrixLimits> elements after the check
  355. tile_mat_limits = tile_mat_set_limits.findall(self.xml_ns.NsWmts('TileMatrixLimits'))
  356. if not tile_mat_limits:
  357. grass.debug('Removed invalid <TileMatrixSetLimits> element.', 4)
  358. link.remove(tile_mat_set_limits)
  359. if not found:
  360. return False
  361. return True
  362. def _checkMatSetLimit(self, limit):
  363. """!Check <TileMatrixLimits> element.
  364. """
  365. limit_tile_mat = limit.find(self.xml_ns.NsWmts('TileMatrix'))
  366. if limit_tile_mat is None or not limit_tile_mat.text:
  367. return False
  368. for i in ['MinTileRow', 'MaxTileRow', 'MinTileCol', 'MaxTileCol']:
  369. i_tag = limit.find(self.xml_ns.NsWmts(i))
  370. if i_tag is None:
  371. return False
  372. try:
  373. int(i_tag.text)
  374. except ValueError:
  375. return False
  376. return True
  377. def _find(self, etreeElement, tag, ns=None):
  378. """!Find child element.
  379. If the element is not found it raises xml.etree.ElementTree.ParseError.
  380. """
  381. if not ns:
  382. res = etreeElement.find(tag)
  383. else:
  384. res = etreeElement.find(ns(tag))
  385. if res is None:
  386. raise ParseError(_("Unable to parse capabilities file. \n\
  387. Tag '%s' was not found.") % tag)
  388. return res
  389. def _findall(self, etreeElement, tag, ns=None):
  390. """!Find all children element.
  391. If no element is found it raises xml.etree.ElementTree.ParseError.
  392. """
  393. if not ns:
  394. res = etreeElement.findall(tag)
  395. else:
  396. res = etreeElement.findall(ns(tag))
  397. if not res:
  398. raise ParseError(_("Unable to parse capabilities file. \n\
  399. Tag '%s' was not found.") % tag)
  400. return res
  401. def getxmlnshandler(self):
  402. """!Return WMTSXMLNsHandler object.
  403. """
  404. return self.xml_ns
  405. class OnEarthCapabilitiesTree(BaseCapabilitiesTree):
  406. def __init__(self, cap_file):
  407. """!Parse NASA OnEarth tile service file.
  408. If the file cannot be parsed it raises xml.etree.ElementTree.ParseError.
  409. The class also removes elements which are in invalid form and are needed
  410. by wxGUI capabilities dialog or for creation of GetMap request by GRASS WMS library.
  411. @param cap_file - capabilities file
  412. """
  413. BaseCapabilitiesTree.__init__(self, cap_file)
  414. grass.debug('Checking OnEarth capabilities tree.', 4)
  415. self._checkLayerTree(self.getroot())
  416. grass.debug('Check if OnEarth capabilities tree was finished.', 4)
  417. def _checkLayerTree(self, parent_layer, first=True):
  418. """!Recursively check layer tree.
  419. """
  420. if first:
  421. tiled_patterns = self._find(parent_layer, 'TiledPatterns')
  422. layers = tiled_patterns.findall('TiledGroup')
  423. layers += tiled_patterns.findall('TiledGroups')
  424. parent_layer = tiled_patterns
  425. else:
  426. layers = parent_layer.findall('TiledGroup')
  427. layers += parent_layer.findall('TiledGroups')
  428. for l in layers:
  429. if not self._checkLayer(l):
  430. grass.debug(('Removed invalid <%s> element.' % l.tag), 4)
  431. parent_layer.remove(l)
  432. if l.tag == 'TiledGroups':
  433. self._checkLayerTree(l, False)
  434. def _find(self, etreeElement, tag):
  435. """!Find child element.
  436. If the element is not found it raises xml.etree.ElementTree.ParseError.
  437. """
  438. res = etreeElement.find(tag)
  439. if res is None:
  440. raise ParseError(_("Unable to parse tile service file. \n\
  441. Tag <%s> was not found.") % tag)
  442. return res
  443. def _checkLayer(self, layer):
  444. """!Check <TiledGroup>/<TiledGroups> elements.
  445. """
  446. if layer.tag == 'TiledGroups':
  447. return True
  448. name = layer.find('Name')
  449. if name is None or not name.text:
  450. return False
  451. t_patts = layer.findall('TilePattern')
  452. for patt in t_patts:
  453. urls = self._getUrls(patt)
  454. for url in urls:
  455. if not self.gettilepatternurldata(url):
  456. urls.remove(url)
  457. # check if there are any valid urls
  458. if not urls:
  459. grass.debug('<TilePattern> was removed. It has no valid url.', 4)
  460. layer.remove(patt)
  461. patt.text = '\n'.join(urls)
  462. t_patts = layer.findall('TilePattern')
  463. if not t_patts:
  464. return False
  465. return True
  466. def _getUrls(self, tile_pattern):
  467. """!Get all urls from tile pattern.
  468. """
  469. urls = []
  470. if tile_pattern.text is not None:
  471. tile_patt_lines = tile_pattern.text.split('\n')
  472. for line in tile_patt_lines:
  473. if 'request=GetMap' in line:
  474. urls.append(line.strip())
  475. return urls
  476. def gettilepatternurldata(self, url):
  477. """!Parse url string in Tile Pattern.
  478. """
  479. par_url = bbox = width = height = None
  480. bbox_idxs = self.geturlparamidxs(url, "bbox=")
  481. if bbox_idxs is None:
  482. return None
  483. par_url = [url[:bbox_idxs[0] - 1], url[bbox_idxs[1]:]]
  484. bbox = url[bbox_idxs[0] + len('bbox='): bbox_idxs[1]]
  485. bbox_list = bbox.split(',')
  486. if len(bbox_list) < 4:
  487. return None
  488. try:
  489. bbox = list(map(float, bbox.split(',')))
  490. except ValueError:
  491. return None
  492. width_idxs = self.geturlparamidxs(url, "width=")
  493. if width_idxs is None:
  494. return None
  495. try:
  496. width = int(url[width_idxs[0] + len('width='): width_idxs[1]])
  497. except ValueError:
  498. return None
  499. height_idxs = self.geturlparamidxs(url, "height=")
  500. if height_idxs is None:
  501. return None
  502. try:
  503. height = int(url[height_idxs[0] + len('height='): height_idxs[1]])
  504. except ValueError:
  505. return None
  506. if height < 0 or width < 0:
  507. return None
  508. return par_url, bbox, width, height
  509. def geturlparamidxs(self, params_str, param_key):
  510. """!Find start and end index of parameter and it's value in url string
  511. """
  512. start_i = params_str.lower().find(param_key)
  513. if start_i < 0:
  514. return None
  515. end_i = params_str.find("&", start_i)
  516. if end_i < 0:
  517. end_i = len(params_str)
  518. return (start_i, end_i)