Mercurial > hg > documentViewer
annotate MpdlXmlTextServer.py @ 473:74e9e74277e9 elementtree
smaller improvements
author | casties |
---|---|
date | Fri, 05 Aug 2011 11:24:42 +0200 |
parents | 415a7026eeda |
children | 0bc4a153863a |
rev | line source |
---|---|
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
1 from OFS.SimpleItem import SimpleItem |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
2 from Products.PageTemplates.PageTemplateFile import PageTemplateFile |
455 | 3 |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
4 from Ft.Xml import EMPTY_NAMESPACE, Parse |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
5 from Ft.Xml.Domlette import NonvalidatingReader |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
6 import Ft.Xml.Domlette |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
7 import cStringIO |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
8 |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
9 import xml.etree.ElementTree as ET |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
10 |
455 | 11 import re |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
12 import logging |
134 | 13 import urllib |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
14 |
458 | 15 from SrvTxtUtils import getInt, getText, getHttpData |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
16 |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
17 def serialize(node): |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
18 """returns a string containing an XML snippet of node""" |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
19 s = ET.tostring(node, 'UTF-8') |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
20 # snip off XML declaration |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
21 if s.startswith('<?xml'): |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
22 i = s.find('?>') |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
23 return s[i+3:] |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
24 |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
25 return s |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
26 |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
27 |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
28 def getTextFromNode(node): |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
29 """get the cdata content of a node""" |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
30 if node is None: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
31 return "" |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
32 # ET: |
455 | 33 # text = node.text or "" |
34 # for e in node: | |
35 # text += gettext(e) | |
36 # if e.tail: | |
37 # text += e.tail | |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
38 |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
39 # 4Suite: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
40 nodelist=node.childNodes |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
41 text = "" |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
42 for n in nodelist: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
43 if n.nodeType == node.TEXT_NODE: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
44 text = text + n.data |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
45 |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
46 return text |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
47 |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
48 def serializeNode(node, encoding="utf-8"): |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
49 """returns a string containing node as XML""" |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
50 #s = ET.tostring(node) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
51 |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
52 # 4Suite: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
53 stream = cStringIO.StringIO() |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
54 Ft.Xml.Domlette.Print(node, stream=stream, encoding=encoding) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
55 s = stream.getvalue() |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
56 stream.close() |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
57 |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
58 return s |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
59 |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
60 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
61 class MpdlXmlTextServer(SimpleItem): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
62 """TextServer implementation for MPDL-XML eXist server""" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
63 meta_type="MPDL-XML TextServer" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
64 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
65 manage_options=( |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
66 {'label':'Config','action':'manage_changeMpdlXmlTextServerForm'}, |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
67 )+SimpleItem.manage_options |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
68 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
69 manage_changeMpdlXmlTextServerForm = PageTemplateFile("zpt/manage_changeMpdlXmlTextServer", globals()) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
70 |
455 | 71 def __init__(self,id,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de/mpdl/interface/", serverName=None, timeout=40): |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
72 """constructor""" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
73 self.id=id |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
74 self.title=title |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
75 self.timeout = timeout |
132 | 76 if serverName is None: |
77 self.serverUrl = serverUrl | |
78 else: | |
79 self.serverUrl = "http://%s/mpdl/interface/"%serverName | |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
80 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
81 def getHttpData(self, url, data=None): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
82 """returns result from url+data HTTP request""" |
458 | 83 return getHttpData(url,data,timeout=self.timeout) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
84 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
85 def getServerData(self, method, data=None): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
86 """returns result from text server for method+data""" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
87 url = self.serverUrl+method |
458 | 88 return getHttpData(url,data,timeout=self.timeout) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
89 |
455 | 90 # WTF: what does this really do? can it be integrated in getPage? |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
91 def getSearch(self, pageinfo=None, docinfo=None): |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
92 """get search list""" |
455 | 93 logging.debug("getSearch()") |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
94 docpath = docinfo['textURLPath'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
95 url = docinfo['url'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
96 pagesize = pageinfo['queryPageSize'] |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
97 pn = pageinfo.get('searchPN',1) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
98 sn = pageinfo['sn'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
99 highlightQuery = pageinfo['highlightQuery'] |
181 | 100 query =pageinfo['query'] |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
101 queryType =pageinfo['queryType'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
102 viewMode= pageinfo['viewMode'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
103 tocMode = pageinfo['tocMode'] |
169 | 104 characterNormalization = pageinfo['characterNormalization'] |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
105 #optionToggle = pageinfo['optionToggle'] |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
106 tocPN = pageinfo['tocPN'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
107 selfurl = self.absolute_url() |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
108 data = self.getServerData("doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&sn=%s&viewMode=%s&characterNormalization=%s&highlightQuery=%s"%(docpath, 'text', queryType, urllib.quote(query), pagesize, pn, sn, viewMode,characterNormalization, urllib.quote(highlightQuery))) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
109 pagexml = data.replace('?document=%s'%str(docpath),'?url=%s'%url) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
110 pagedom = Parse(pagexml) |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
111 |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
112 """ |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
113 pagedivs = pagedom.xpath("//div[@class='queryResultHits']") |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
114 if (pagedivs == pagedom.xpath("//div[@class='queryResultHits']")): |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
115 if len(pagedivs)>0: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
116 docinfo['queryResultHits'] = int(getTextFromNode(pagedivs[0])) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
117 s = getTextFromNode(pagedivs[0]) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
118 s1 = int(s)/10+1 |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
119 try: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
120 docinfo['queryResultHits'] = int(s1) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
121 logging.debug("SEARCH ENTRIES: %s"%(s1)) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
122 except: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
123 docinfo['queryResultHits'] = 0 |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
124 """ |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
125 if (queryType=="fulltext")or(queryType=="xpath")or(queryType=="xquery")or(queryType=="fulltextMorphLemma"): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
126 pagedivs = pagedom.xpath("//div[@class='queryResultPage']") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
127 if len(pagedivs)>0: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
128 pagenode=pagedivs[0] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
129 links=pagenode.xpath("//a") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
130 for l in links: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
131 hrefNode = l.getAttributeNodeNS(None, u"href") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
132 if hrefNode: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
133 href = hrefNode.nodeValue |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
134 if href.startswith('page-fragment.xql'): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
135 selfurl = self.absolute_url() |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
136 pagexml=href.replace('mode=text','mode=texttool&viewMode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&tocMode=%s&searchPN=%s&tocPN=%s&characterNormalization=%s'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN, characterNormalization)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
137 hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl) |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
138 #logging.debug("PUREXML :%s"%(serializeNode(pagenode))) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
139 return serializeNode(pagenode) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
140 if (queryType=="fulltextMorph"): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
141 pagedivs = pagedom.xpath("//div[@class='queryResult']") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
142 if len(pagedivs)>0: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
143 pagenode=pagedivs[0] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
144 links=pagenode.xpath("//a") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
145 for l in links: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
146 hrefNode = l.getAttributeNodeNS(None, u"href") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
147 if hrefNode: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
148 href = hrefNode.nodeValue |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
149 if href.startswith('page-fragment.xql'): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
150 selfurl = self.absolute_url() |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
151 pagexml=href.replace('mode=text','mode=texttool&viewMode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&tocMode=%s&searchPN=%s&tocPN=%s&characterNormalization=%s'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN,characterNormalization)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
152 hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
153 if href.startswith('../lt/lemma.xql'): |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
154 hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_query'%(selfurl)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
155 l.setAttributeNS(None, 'target', '_blank') |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
156 l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;") |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
157 l.setAttributeNS(None, 'onDblclick', 'popupWin.focus();') |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
158 pagedivs = pagedom.xpath("//div[@class='queryResultMorphExpansion']") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
159 return serializeNode(pagenode) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
160 if (queryType=="ftIndex")or(queryType=="ftIndexMorph"): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
161 pagedivs= pagedom.xpath("//div[@class='queryResultPage']") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
162 if len(pagedivs)>0: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
163 pagenode=pagedivs[0] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
164 links=pagenode.xpath("//a") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
165 for l in links: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
166 hrefNode = l.getAttributeNodeNS(None, u"href") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
167 if hrefNode: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
168 href = hrefNode.nodeValue |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
169 hrefNode.nodeValue=href.replace('mode=text','mode=texttool&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s&characterNormalization=%s'%(viewMode,tocMode,tocPN,pn,characterNormalization)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
170 if href.startswith('../lt/lex.xql'): |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
171 hrefNode.nodeValue = href.replace('../lt/lex.xql','%s/template/head_main_lex'%selfurl) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
172 l.setAttributeNS(None, 'target', '_blank') |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
173 l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;") |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
174 l.setAttributeNS(None, 'onDblclick', 'popupWin.focus();') |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
175 if href.startswith('../lt/lemma.xql'): |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
176 hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_lemma'%(selfurl)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
177 l.setAttributeNS(None, 'target', '_blank') |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
178 l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;") |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
179 l.setAttributeNS(None, 'onDblclick', 'popupWin.focus();') |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
180 return serializeNode(pagenode) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
181 return "no text here" |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
182 |
253 | 183 def getGisPlaces(self, docinfo=None, pageinfo=None): |
207 | 184 """ Show all Gis Places of whole Page""" |
264 | 185 xpath='//place' |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
186 docpath = docinfo.get('textURLPath',None) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
187 if not docpath: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
188 return None |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
189 |
253 | 190 url = docinfo['url'] |
191 selfurl = self.absolute_url() | |
257 | 192 pn = pageinfo['current'] |
291 | 193 hrefList=[] |
306 | 194 myList= "" |
264 | 195 text=self.getServerData("xpath.xql", "document=%s&xpath=%s&pn=%s"%(docinfo['textURLPath'],xpath,pn)) |
455 | 196 dom = ET.fromstring(text) |
197 result = dom.findall(".//result/resultPage/place") | |
236 | 198 for l in result: |
455 | 199 href = l.get("id") |
292 | 200 hrefList.append(href) |
455 | 201 # WTF: what does this do? |
309 | 202 myList = ",".join(hrefList) |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
203 #logging.debug("getGisPlaces :%s"%(myList)) |
307 | 204 return myList |
205 | |
206 def getAllGisPlaces (self, docinfo=None, pageinfo=None): | |
207 """Show all Gis Places of whole Book """ | |
208 xpath ='//echo:place' | |
209 docpath =docinfo['textURLPath'] | |
210 url = docinfo['url'] | |
211 selfurl =self.absolute_url() | |
212 pn =pageinfo['current'] | |
213 hrefList=[] | |
214 myList="" | |
215 text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'],xpath)) | |
455 | 216 dom = ET.fromstring(text) |
217 result = dom.findall(".//result/resultPage/place") | |
395 | 218 |
307 | 219 for l in result: |
455 | 220 href = l.get("id") |
307 | 221 hrefList.append(href) |
455 | 222 # WTF: what does this do? |
300 | 223 myList = ",".join(hrefList) |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
224 #logging.debug("getALLGisPlaces :%s"%(myList)) |
309 | 225 return myList |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
226 |
455 | 227 def processPageInfo(self, dom, docinfo, pageinfo): |
228 """processes page info divs from dom and stores in docinfo and pageinfo""" | |
469 | 229 # assume first second level div is pageMeta |
230 alldivs = dom.find("div") | |
473 | 231 |
232 if alldivs is None or alldivs.get('class', '') != 'pageMeta': | |
233 logging.error("processPageInfo: pageMeta div not found!") | |
234 return | |
235 | |
455 | 236 for div in alldivs: |
237 dc = div.get('class') | |
238 | |
239 # pageNumberOrig | |
469 | 240 if dc == 'pageNumberOrig': |
455 | 241 pageinfo['pageNumberOrig'] = div.text |
242 | |
243 # pageNumberOrigNorm | |
244 elif dc == 'pageNumberOrigNorm': | |
245 pageinfo['pageNumberOrigNorm'] = div.text | |
246 | |
247 # pageNumberOrigNorm | |
248 elif dc == 'countFigureEntries': | |
469 | 249 docinfo['numFigureEntries'] = getInt(div.text) |
455 | 250 |
251 # pageNumberOrigNorm | |
252 elif dc == 'countTocEntries': | |
253 # WTF: s1 = int(s)/30+1 | |
469 | 254 docinfo['numTocEntries'] = getInt(div.text) |
255 | |
256 # pageHeaderTitle | |
257 elif dc == 'pageHeaderTitle': | |
258 docinfo['pageHeaderTitle'] = div.text | |
455 | 259 |
260 # numTextPages | |
261 elif dc == 'countPages': | |
458 | 262 np = getInt(div.text) |
455 | 263 if np > 0: |
264 docinfo['numTextPages'] = np | |
265 if docinfo.get('numPages', 0) == 0: | |
465 | 266 # seems to be text-only - update page count |
267 docinfo['numPages'] = np | |
455 | 268 pageinfo['end'] = min(pageinfo['end'], np) |
269 pageinfo['numgroups'] = int(np / pageinfo['groupsize']) | |
270 if np % pageinfo['groupsize'] > 0: | |
271 pageinfo['numgroups'] += 1 | |
473 | 272 |
273 #logging.debug("processPageInfo: pageinfo=%s"%repr(pageinfo)) | |
455 | 274 return |
275 | |
388 | 276 |
471 | 277 def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None): |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
278 """returns single page from fulltext""" |
455 | 279 logging.debug("getTextPage mode=%s, pn=%s"%(mode,pn)) |
280 # check for cached text -- but this shouldn't be called twice | |
281 if pageinfo.has_key('textPage'): | |
282 logging.debug("getTextPage: using cached text") | |
283 return pageinfo['textPage'] | |
284 | |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
285 docpath = docinfo['textURLPath'] |
455 | 286 # just checking |
287 if pageinfo['current'] != pn: | |
288 logging.warning("getTextPage: current!=pn!") | |
289 | |
290 # stuff for constructing full urls | |
291 url = docinfo['url'] | |
292 urlmode = docinfo['mode'] | |
293 sn = pageinfo.get('sn', None) | |
294 highlightQuery = pageinfo.get('highlightQuery', None) | |
295 tocMode = pageinfo.get('tocMode', None) | |
296 tocPN = pageinfo.get('tocPN',None) | |
297 characterNormalization = pageinfo.get('characterNormalization', None) | |
298 selfurl = docinfo['viewerUrl'] | |
299 | |
473 | 300 if mode == "dict" or mode == "text_dict": |
301 # dict is called textPollux in the backend | |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
302 textmode = "textPollux" |
471 | 303 elif not mode: |
304 # default is text | |
305 textmode = "text" | |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
306 else: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
307 textmode = mode |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
308 |
373 | 309 textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization) |
455 | 310 if highlightQuery: |
376 | 311 textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn) |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
312 |
455 | 313 # fetch the page |
185 | 314 pagexml = self.getServerData("page-fragment.xql",textParam) |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
315 dom = ET.fromstring(pagexml) |
455 | 316 # extract additional info |
317 self.processPageInfo(dom, docinfo, pageinfo) | |
318 # page content is in <div class="pageContent"> | |
319 pagediv = None | |
320 # ElementTree 1.2 in Python 2.6 can't do div[@class='pageContent'] | |
469 | 321 # so we look at the second level divs |
322 alldivs = dom.findall("div") | |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
323 for div in alldivs: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
324 dc = div.get('class') |
455 | 325 # page content div |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
326 if dc == 'pageContent': |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
327 pagediv = div |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
328 break |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
329 |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
330 # plain text mode |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
331 if mode == "text": |
473 | 332 if pagediv is not None: |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
333 links = pagediv.findall(".//a") |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
334 for l in links: |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
335 href = l.get('href') |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
336 if href and href.startswith('#note-'): |
455 | 337 href = href.replace('#note-',"?mode=%s&url=%s&viewMode=text&tocMode=%s&tocPN=%s&pn=%s#note-"%(urlmode,url,tocMode,tocPN,pn)) |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
338 l.set('href', href) |
455 | 339 |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
340 return serialize(pagediv) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
341 |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
342 # text-with-links mode |
455 | 343 elif mode == "text_dict": |
473 | 344 if pagediv is not None: |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
345 # check all a-tags |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
346 links = pagediv.findall(".//a") |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
347 for l in links: |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
348 href = l.get('href') |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
349 |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
350 if href: |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
351 # is link with href |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
352 if href.startswith('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/lt/wordInfo.xql'): |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
353 # is pollux link |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
354 selfurl = self.absolute_url() |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
355 # change href |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
356 l.set('href', href.replace('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/lt/wordInfo.xql','%s/head_main_voc'%selfurl)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
357 # add target |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
358 l.set('target', '_blank') |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
359 |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
360 if href.startswith('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/lt/lemma.xql'): |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
361 selfurl = self.absolute_url() |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
362 l.set('href', href.replace('http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/lt/lemma.xql','%s/head_main_lemma'%selfurl)) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
363 l.set('target', '_blank') |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
364 l.set('onclick',"popupWin = window.open(this.href, 'InfoWindow', 'menubar=no, location,width=500,height=600,top=180, left=700, toolbar=no, scrollbars=1'); return false;") |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
365 l.set('ondblclick', 'popupWin.focus();') |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
366 |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
367 if href.startswith('#note-'): |
455 | 368 l.set('href', href.replace('#note-',"?mode=%s&url=%s&viewMode=text_dict&tocMode=%s&tocPN=%s&pn=%s#note-"%(urlmode,url,tocMode,tocPN,pn))) |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
369 |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
370 return serialize(pagediv) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
371 |
455 | 372 # xml mode |
373 elif mode == "xml": | |
473 | 374 if pagediv is not None: |
455 | 375 return serialize(pagediv) |
376 | |
377 # pureXml mode | |
378 elif mode == "pureXml": | |
473 | 379 if pagediv is not None: |
455 | 380 return serialize(pagediv) |
381 | |
382 # gis mode | |
383 elif mode == "gis": | |
384 name = docinfo['name'] | |
473 | 385 if pagediv is not None: |
455 | 386 # check all a-tags |
387 links = pagediv.findall(".//a") | |
388 for l in links: | |
389 href = l.get('href') | |
390 if href: | |
391 if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'): | |
392 l.set('href', href.replace('chinagis_REST/REST/db/chgis/mpdl','chinagis/REST/db/mpdl/%s'%name)) | |
393 l.set('target', '_blank') | |
394 | |
395 return serialize(pagediv) | |
396 | |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
397 return "no text here" |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
398 |
455 | 399 # WTF: is this needed? |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
400 def getOrigPages(self, docinfo=None, pageinfo=None): |
455 | 401 logging.debug("CALLED: getOrigPages!") |
402 if not pageinfo.has_key('pageNumberOrig'): | |
403 logging.warning("getOrigPages: not in pageinfo!") | |
404 return None | |
405 | |
406 return pageinfo['pageNumberOrig'] | |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
407 |
455 | 408 # WTF: is this needed? |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
409 def getOrigPagesNorm(self, docinfo=None, pageinfo=None): |
455 | 410 logging.debug("CALLED: getOrigPagesNorm!") |
411 if not pageinfo.has_key('pageNumberOrigNorm'): | |
412 logging.warning("getOrigPagesNorm: not in pageinfo!") | |
413 return None | |
414 | |
415 return pageinfo['pageNumberOrigNorm'] | |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
416 |
455 | 417 # TODO: should be getWordInfo |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
418 def getTranslate(self, word=None, language=None): |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
419 """translate into another languages""" |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
420 data = self.getServerData("lt/wordInfo.xql","language="+str(language)+"&word="+urllib.quote(word)+"&output=html") |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
421 return data |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
422 |
455 | 423 # WTF: what does this do? |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
424 def getLemma(self, lemma=None, language=None): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
425 """simular words lemma """ |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
426 data = self.getServerData("lt/lemma.xql","language="+str(language)+"&lemma="+urllib.quote(lemma)+"&output=html") |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
427 return data |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
428 |
455 | 429 # WTF: what does this do? |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
430 def getLemmaQuery(self, query=None, language=None): |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
431 """simular words lemma """ |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
432 data = self.getServerData("lt/lemma.xql","language="+str(language)+"&query="+urllib.quote(query)+"&output=html") |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
433 return data |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
434 |
455 | 435 # WTF: what does this do? |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
436 def getLex(self, query=None, language=None): |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
437 #simular words lemma |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
438 data = self.getServerData("lt/lex.xql","document=&language="+str(language)+"&query="+urllib.quote(query)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
439 return data |
455 | 440 |
441 # WTF: what does this do? | |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
442 def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1): |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
443 #number of |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
444 docpath = docinfo['textURLPath'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
445 pagesize = pageinfo['queryPageSize'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
446 pn = pageinfo['searchPN'] |
181 | 447 query =pageinfo['query'] |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
448 queryType =pageinfo['queryType'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
449 tocSearch = 0 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
450 tocDiv = None |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
451 |
179 | 452 pagexml = self.getServerData("doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath, 'text', queryType, urllib.quote(query), pagesize, pn)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
453 pagedom = Parse(pagexml) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
454 numdivs = pagedom.xpath("//div[@class='queryResultHits']") |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
455 tocSearch = int(getTextFromNode(numdivs[0])) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
456 tc=int((tocSearch/10)+1) |
402 | 457 return tc |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
458 |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
459 def getToc(self, mode="text", docinfo=None): |
455 | 460 """loads table of contents and stores XML in docinfo""" |
461 logging.debug("getToc mode=%s"%mode) | |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
462 if mode == "none": |
455 | 463 return docinfo |
464 | |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
465 if 'tocSize_%s'%mode in docinfo: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
466 # cached toc |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
467 return docinfo |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
468 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
469 docpath = docinfo['textURLPath'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
470 # we need to set a result set size |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
471 pagesize = 1000 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
472 pn = 1 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
473 if mode == "text": |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
474 queryType = "toc" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
475 else: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
476 queryType = mode |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
477 # number of entries in toc |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
478 tocSize = 0 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
479 tocDiv = None |
455 | 480 # fetch full toc |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
481 pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn)) |
455 | 482 dom = ET.fromstring(pagexml) |
483 # page content is in <div class="queryResultPage"> | |
484 pagediv = None | |
485 # ElementTree 1.2 in Python 2.6 can't do div[@class='queryResultPage'] | |
486 alldivs = dom.findall("div") | |
487 for div in alldivs: | |
488 dc = div.get('class') | |
489 # page content div | |
490 if dc == 'queryResultPage': | |
491 pagediv = div | |
492 | |
493 elif dc == 'queryResultHits': | |
458 | 494 docinfo['tocSize_%s'%mode] = getInt(div.text) |
455 | 495 |
496 if pagediv: | |
497 # store XML in docinfo | |
498 docinfo['tocXML_%s'%mode] = ET.tostring(pagediv, 'UTF-8') | |
499 | |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
500 return docinfo |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
501 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
502 def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
503 """returns single page from the table of contents""" |
455 | 504 logging.debug("getTocPage mode=%s, pn=%s"%(mode,pn)) |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
505 if mode == "text": |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
506 queryType = "toc" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
507 else: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
508 queryType = mode |
455 | 509 |
510 # check for cached TOC | |
511 if not docinfo.has_key('tocXML_%s'%mode): | |
512 self.getToc(mode=mode, docinfo=docinfo) | |
513 | |
514 tocxml = docinfo.get('tocXML_%s'%mode, None) | |
515 if not tocxml: | |
516 logging.error("getTocPage: unable to find tocXML") | |
517 return "No ToC" | |
518 | |
519 pagesize = int(pageinfo['tocPageSize']) | |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
520 url = docinfo['url'] |
455 | 521 urlmode = docinfo['mode'] |
522 selfurl = docinfo['viewerUrl'] | |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
523 viewMode= pageinfo['viewMode'] |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
524 tocMode = pageinfo['tocMode'] |
455 | 525 tocPN = int(pageinfo['tocPN']) |
456 | 526 pn = tocPN |
455 | 527 |
528 fulltoc = ET.fromstring(tocxml) | |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
529 |
455 | 530 if fulltoc: |
531 # paginate | |
456 | 532 start = (pn - 1) * pagesize * 2 |
533 len = pagesize * 2 | |
534 del fulltoc[:start] | |
535 del fulltoc[len:] | |
455 | 536 tocdivs = fulltoc |
537 | |
538 # check all a-tags | |
539 links = tocdivs.findall(".//a") | |
540 for l in links: | |
541 href = l.get('href') | |
542 if href: | |
543 # take pn from href | |
544 m = re.match(r'page-fragment\.xql.*pn=(\d+)', href) | |
545 if m is not None: | |
546 # and create new url | |
547 l.set('href', '%s?mode=%s&url=%s&viewMode=%s&pn=%s&tocMode=%s&tocPN=%s'%(selfurl, urlmode, url, viewMode, m.group(1), tocMode, tocPN)) | |
548 else: | |
549 logging.warning("getTocPage: Problem with link=%s"%href) | |
550 | |
551 return serialize(tocdivs) | |
552 | |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
553 |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
554 def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None): |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
555 """change settings""" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
556 self.title=title |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
557 self.timeout = timeout |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
558 self.serverUrl = serverUrl |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
559 if RESPONSE is not None: |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
560 RESPONSE.redirect('manage_main') |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
561 |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
562 # management methods |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
563 def manage_addMpdlXmlTextServerForm(self): |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
564 """Form for adding""" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
565 pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
566 return pt() |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
567 |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
568 def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None): |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
407
diff
changeset
|
569 #def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-text.mpiwg-berlin.mpg.de:30030/mpdl/interface/",timeout=40,RESPONSE=None): |
129
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
570 """add zogiimage""" |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
571 newObj = MpdlXmlTextServer(id,title,serverUrl,timeout) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
572 self.Destination()._setObject(id, newObj) |
9404b6c37920
more modular version with separate object MpdlXmlTextServer
casties
parents:
diff
changeset
|
573 if RESPONSE is not None: |
455 | 574 RESPONSE.redirect('manage_main') |
575 | |
576 |