--- a/web/ldt/ldt_utils/projectindexer.py Thu Oct 14 12:17:31 2010 +0200
+++ b/web/ldt/ldt_utils/projectindexer.py Fri Oct 15 12:36:43 2010 +0200
@@ -4,15 +4,10 @@
import shutil
import ldt.utils.xml
from ldt import settings
-import xml
-import xml.dom
-import xml.dom.minidom
-import xml.dom.ext
-import xml.xpath
import lucene
from ldt.ldt_utils import STORE
from ldt.ldt_utils import ANALYZER
-
+import lxml.etree
def Property(func):
return property(**func())
@@ -47,68 +42,67 @@
def index_project(self, project):
# ldt.utils.log.debug("Indexing project : "+str(project.ldt_id))
- ldt=project.ldt
- doc = xml.dom.minidom.parseString(ldt.encode( "utf-8" ))
+ ldt=project.ldt
+ doc = lxml.etree.fromstring(ldt.encode( "utf-8" ))
self.__writer.deleteDocuments(lucene.Term("ldt_id", project.ldt_id))
- con = xml.xpath.Context.Context(doc, 1, 1, None)
- res = xml.xpath.Evaluate("/iri/annotations/content", context=con)
+ res = doc.xpath("/iri/annotations/content")
project.ldt.encode( "utf-8 " )
for content in res:
- contentId = content.getAttribute("id")
+ contentId = content.get("id")
- res =xml.xpath.Evaluate("ensemble", content)
+ res =content.xpath("ensemble")
for ensemble in res:
- ensembleId = ensemble.getAttribute("id")
+ ensembleId = ensemble.get("id")
- for decoupageNode in ensemble.childNodes:
+ for decoupageNode in ensemble.getchildren():
# ldt.utils.log.debug("Indexing project decoupage : "+ repr(decoupageNode.nodeType) + " in " + repr(self.decoupage_blacklist))
- if decoupageNode.nodeType != xml.dom.Node.ELEMENT_NODE or decoupageNode.tagName != "decoupage" or decoupageNode.getAttribute("id") in self.decoupage_blacklist:
+ if decoupageNode.tag != "decoupage" or decoupageNode.get("id") in self.decoupage_blacklist:
continue
- decoupId = decoupageNode.getAttribute("id")
- res = xml.xpath.Evaluate("elements/element", decoupageNode)
+ decoupId = decoupageNode.get("id")
+ res = decoupageNode.xpath("elements/element")
for elementNode in res:
doc = lucene.Document()
- elementId = elementNode.getAttribute("id")
- tags = elementNode.getAttribute("tags")
+ elementId = elementNode.get("id")
+ tags = elementNode.get("tags")
if tags is not None:
tags.replace(",", ";")
if tags is None or len(tags) == 0:
tags = ""
- restagnode = xml.xpath.Evaluate("tag/text()", elementNode)
+ restagnode = elementNode.xpath("tag/text()")
for tagnode in restagnode:
- tags = tags + " ; " + tagnode.data
+ tags = tags + " ; " + tagnode.text()
if tags is None or len(tags) == 0:
tags = ""
- restagnode = xml.xpath.Evaluate("tags/tag/text()", elementNode)
+ restagnode = elementNode.xpath("tags/tag/text()")
for tagnode in restagnode:
- tags = tags + " ; " + tagnode.data
+ tags = tags + " ; " + tagnode.text()
title = ""
- for txtRes in xml.xpath.Evaluate("title/text()", elementNode):
- title = title + txtRes.data
+ for txtRes in elementNode.xpath("title/text()"):
+ title = title + txtRes.text()
abstract = ""
- for txtRes in xml.xpath.Evaluate("abstract/text()", elementNode):
- abstract = abstract + txtRes.data
+ for txtRes in elementNode.xpath("abstract/text()"):
+ abstract = abstract + txtRes.text()
- doc.add(lucene.Field("ldt_id", project.ldt_id, lucene.Field.Store.YES, lucene.Field.Index.UN_TOKENIZED))
- doc.add(lucene.Field("iri_id", contentId, lucene.Field.Store.YES, lucene.Field.Index.UN_TOKENIZED))
+ doc.add(lucene.Field("ldt_id", project.ldt_id, lucene.Field.Store.YES, lucene.Field.Index.NOT_ANALYZED))
+ doc.add(lucene.Field("iri_id", contentId, lucene.Field.Store.YES, lucene.Field.Index.NOT_ANALYZED))
doc.add(lucene.Field("ensemble_id", ensembleId, lucene.Field.Store.YES, lucene.Field.Index.NO))
doc.add(lucene.Field("decoupage_id", decoupId, lucene.Field.Store.YES, lucene.Field.Index.NO))
doc.add(lucene.Field("element_id", elementId, lucene.Field.Store.YES, lucene.Field.Index.NO))
- doc.add(lucene.Field("tags", tags, lucene.Field.Store.NO, lucene.Field.Index.TOKENIZED))
- doc.add(lucene.Field("title", title, lucene.Field.Store.NO, lucene.Field.Index.TOKENIZED))
- doc.add(lucene.Field("abstract", abstract, lucene.Field.Store.NO, lucene.Field.Index.TOKENIZED))
- doc.add(lucene.Field("all", " ".join([tags, title, abstract]), lucene.Field.Store.NO, lucene.Field.Index.TOKENIZED))
+ doc.add(lucene.Field("tags", tags, lucene.Field.Store.NO, lucene.Field.Index.ANALYZED))
+ doc.add(lucene.Field("title", title, lucene.Field.Store.NO, lucene.Field.Index.ANALYZED))
+ doc.add(lucene.Field("abstract", abstract, lucene.Field.Store.NO, lucene.Field.Index.ANALYZED))
+ doc.add(lucene.Field("all", " ".join([tags, title, abstract]), lucene.Field.Store.NO, lucene.Field.Index.ANALYZED))
self.__writer.addDocument(doc)
- self.__writer.flush()
+ self.__writer.commit()