Vous ne pouvez pas sélectionner plus de 25 sujets Les noms de sujets doivent commencer par une lettre ou un nombre, peuvent contenir des tirets ('-') et peuvent comporter jusqu'à 35 caractères.
 
 
 
 

287 lignes
9.5 KiB

  1. # cython: language_level=2
  2. """
  3. SAX-based adapter to copy trees from/to the Python standard library.
  4. Use the `ElementTreeContentHandler` class to build an ElementTree from
  5. SAX events.
  6. Use the `ElementTreeProducer` class or the `saxify()` function to fire
  7. the SAX events of an ElementTree against a SAX ContentHandler.
  8. See https://lxml.de/sax.html
  9. """
  10. from xml.sax.handler import ContentHandler
  11. from lxml import etree
  12. from lxml.etree import ElementTree, SubElement
  13. from lxml.etree import Comment, ProcessingInstruction
  14. try:
  15. from types import GenericAlias as _GenericAlias
  16. except ImportError:
  17. # Python 3.8 - we only need this as return value from "__class_getitem__"
  18. def _GenericAlias(cls, item):
  19. return f"{cls.__name__}[{item.__name__}]"
  20. class SaxError(etree.LxmlError):
  21. """General SAX error.
  22. """
  23. def _getNsTag(tag):
  24. if tag[0] == '{':
  25. return tuple(tag[1:].split('}', 1))
  26. else:
  27. return None, tag
  28. class ElementTreeContentHandler(ContentHandler):
  29. """Build an lxml ElementTree from SAX events.
  30. """
  31. def __init__(self, makeelement=None):
  32. ContentHandler.__init__(self)
  33. self._root = None
  34. self._root_siblings = []
  35. self._element_stack = []
  36. self._default_ns = None
  37. self._ns_mapping = { None : [None] }
  38. self._new_mappings = {}
  39. if makeelement is None:
  40. makeelement = etree.Element
  41. self._makeelement = makeelement
  42. def _get_etree(self):
  43. "Contains the generated ElementTree after parsing is finished."
  44. return ElementTree(self._root)
  45. etree = property(_get_etree, doc=_get_etree.__doc__)
  46. def setDocumentLocator(self, locator):
  47. pass
  48. def startDocument(self):
  49. pass
  50. def endDocument(self):
  51. pass
  52. def startPrefixMapping(self, prefix, uri):
  53. self._new_mappings[prefix] = uri
  54. try:
  55. self._ns_mapping[prefix].append(uri)
  56. except KeyError:
  57. self._ns_mapping[prefix] = [uri]
  58. if prefix is None:
  59. self._default_ns = uri
  60. def endPrefixMapping(self, prefix):
  61. ns_uri_list = self._ns_mapping[prefix]
  62. ns_uri_list.pop()
  63. if prefix is None:
  64. self._default_ns = ns_uri_list[-1]
  65. def _buildTag(self, ns_name_tuple):
  66. ns_uri, local_name = ns_name_tuple
  67. if ns_uri:
  68. el_tag = "{%s}%s" % ns_name_tuple
  69. elif self._default_ns:
  70. el_tag = "{%s}%s" % (self._default_ns, local_name)
  71. else:
  72. el_tag = local_name
  73. return el_tag
  74. def startElementNS(self, ns_name, qname, attributes=None):
  75. el_name = self._buildTag(ns_name)
  76. if attributes:
  77. attrs = {}
  78. try:
  79. iter_attributes = attributes.iteritems()
  80. except AttributeError:
  81. iter_attributes = attributes.items()
  82. for name_tuple, value in iter_attributes:
  83. if name_tuple[0]:
  84. attr_name = "{%s}%s" % name_tuple
  85. else:
  86. attr_name = name_tuple[1]
  87. attrs[attr_name] = value
  88. else:
  89. attrs = None
  90. element_stack = self._element_stack
  91. if self._root is None:
  92. element = self._root = \
  93. self._makeelement(el_name, attrs, self._new_mappings)
  94. if self._root_siblings and hasattr(element, 'addprevious'):
  95. for sibling in self._root_siblings:
  96. element.addprevious(sibling)
  97. del self._root_siblings[:]
  98. else:
  99. element = SubElement(element_stack[-1], el_name,
  100. attrs, self._new_mappings)
  101. element_stack.append(element)
  102. self._new_mappings.clear()
  103. def processingInstruction(self, target, data):
  104. pi = ProcessingInstruction(target, data)
  105. if self._root is None:
  106. self._root_siblings.append(pi)
  107. else:
  108. self._element_stack[-1].append(pi)
  109. def endElementNS(self, ns_name, qname):
  110. element = self._element_stack.pop()
  111. el_tag = self._buildTag(ns_name)
  112. if el_tag != element.tag:
  113. raise SaxError("Unexpected element closed: " + el_tag)
  114. def startElement(self, name, attributes=None):
  115. if attributes:
  116. attributes = {(None, k): v for k, v in attributes.items()}
  117. self.startElementNS((None, name), name, attributes)
  118. def endElement(self, name):
  119. self.endElementNS((None, name), name)
  120. def characters(self, data):
  121. last_element = self._element_stack[-1]
  122. try:
  123. # if there already is a child element, we must append to its tail
  124. last_element = last_element[-1]
  125. last_element.tail = (last_element.tail or '') + data
  126. except IndexError:
  127. # otherwise: append to the text
  128. last_element.text = (last_element.text or '') + data
  129. ignorableWhitespace = characters
  130. # Allow subscripting sax.ElementTreeContentHandler in type annotions (PEP 560)
  131. def __class_getitem__(cls, item):
  132. return _GenericAlias(cls, item)
  133. class ElementTreeProducer:
  134. """Produces SAX events for an element and children.
  135. """
  136. def __init__(self, element_or_tree, content_handler):
  137. try:
  138. element = element_or_tree.getroot()
  139. except AttributeError:
  140. element = element_or_tree
  141. self._element = element
  142. self._content_handler = content_handler
  143. from xml.sax.xmlreader import AttributesNSImpl as attr_class
  144. self._attr_class = attr_class
  145. self._empty_attributes = attr_class({}, {})
  146. def saxify(self):
  147. self._content_handler.startDocument()
  148. element = self._element
  149. if hasattr(element, 'getprevious'):
  150. siblings = []
  151. sibling = element.getprevious()
  152. while getattr(sibling, 'tag', None) is ProcessingInstruction:
  153. siblings.append(sibling)
  154. sibling = sibling.getprevious()
  155. for sibling in siblings[::-1]:
  156. self._recursive_saxify(sibling, {})
  157. self._recursive_saxify(element, {})
  158. if hasattr(element, 'getnext'):
  159. sibling = element.getnext()
  160. while getattr(sibling, 'tag', None) is ProcessingInstruction:
  161. self._recursive_saxify(sibling, {})
  162. sibling = sibling.getnext()
  163. self._content_handler.endDocument()
  164. def _recursive_saxify(self, element, parent_nsmap):
  165. content_handler = self._content_handler
  166. tag = element.tag
  167. if tag is Comment or tag is ProcessingInstruction:
  168. if tag is ProcessingInstruction:
  169. content_handler.processingInstruction(
  170. element.target, element.text)
  171. tail = element.tail
  172. if tail:
  173. content_handler.characters(tail)
  174. return
  175. element_nsmap = element.nsmap
  176. new_prefixes = []
  177. if element_nsmap != parent_nsmap:
  178. # There have been updates to the namespace
  179. for prefix, ns_uri in element_nsmap.items():
  180. if parent_nsmap.get(prefix) != ns_uri:
  181. new_prefixes.append( (prefix, ns_uri) )
  182. attribs = element.items()
  183. if attribs:
  184. attr_values = {}
  185. attr_qnames = {}
  186. for attr_ns_name, value in attribs:
  187. attr_ns_tuple = _getNsTag(attr_ns_name)
  188. attr_values[attr_ns_tuple] = value
  189. attr_qnames[attr_ns_tuple] = self._build_qname(
  190. attr_ns_tuple[0], attr_ns_tuple[1], element_nsmap,
  191. preferred_prefix=None, is_attribute=True)
  192. sax_attributes = self._attr_class(attr_values, attr_qnames)
  193. else:
  194. sax_attributes = self._empty_attributes
  195. ns_uri, local_name = _getNsTag(tag)
  196. qname = self._build_qname(
  197. ns_uri, local_name, element_nsmap, element.prefix, is_attribute=False)
  198. for prefix, uri in new_prefixes:
  199. content_handler.startPrefixMapping(prefix, uri)
  200. content_handler.startElementNS(
  201. (ns_uri, local_name), qname, sax_attributes)
  202. text = element.text
  203. if text:
  204. content_handler.characters(text)
  205. for child in element:
  206. self._recursive_saxify(child, element_nsmap)
  207. content_handler.endElementNS((ns_uri, local_name), qname)
  208. for prefix, uri in new_prefixes:
  209. content_handler.endPrefixMapping(prefix)
  210. tail = element.tail
  211. if tail:
  212. content_handler.characters(tail)
  213. def _build_qname(self, ns_uri, local_name, nsmap, preferred_prefix, is_attribute):
  214. if ns_uri is None:
  215. return local_name
  216. if not is_attribute and nsmap.get(preferred_prefix) == ns_uri:
  217. prefix = preferred_prefix
  218. else:
  219. # Pick the first matching prefix, in alphabetical order.
  220. candidates = [
  221. pfx for (pfx, uri) in nsmap.items()
  222. if pfx is not None and uri == ns_uri
  223. ]
  224. prefix = (
  225. candidates[0] if len(candidates) == 1
  226. else min(candidates) if candidates
  227. else None
  228. )
  229. if prefix is None:
  230. # Default namespace
  231. return local_name
  232. return prefix + ':' + local_name
  233. def saxify(element_or_tree, content_handler):
  234. """One-shot helper to generate SAX events from an XML tree and fire
  235. them against a SAX ContentHandler.
  236. """
  237. return ElementTreeProducer(element_or_tree, content_handler).saxify()