Annotation of documentViewer/MpdlXmlTextServer.py, revision 1.48
1.2 casties 1:
2: from OFS.SimpleItem import SimpleItem
3: from Products.PageTemplates.PageTemplateFile import PageTemplateFile
4: from Ft.Xml import EMPTY_NAMESPACE, Parse
5:
6: import sys
7: import logging
1.5 casties 8: import urllib
1.2 casties 9: import documentViewer
10: from documentViewer import getTextFromNode, serializeNode
11:
12: class MpdlXmlTextServer(SimpleItem):
13: """TextServer implementation for MPDL-XML eXist server"""
14: meta_type="MPDL-XML TextServer"
15:
16: manage_options=(
17: {'label':'Config','action':'manage_changeMpdlXmlTextServerForm'},
18: )+SimpleItem.manage_options
19:
20: manage_changeMpdlXmlTextServerForm = PageTemplateFile("zpt/manage_changeMpdlXmlTextServer", globals())
21:
1.3 casties 22: def __init__(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/", serverName=None, timeout=40):
1.2 casties 23: """constructor"""
24: self.id=id
25: self.title=title
26: self.timeout = timeout
1.3 casties 27: if serverName is None:
28: self.serverUrl = serverUrl
29: else:
30: self.serverUrl = "http://%s/mpdl/interface/"%serverName
1.2 casties 31:
32: def getHttpData(self, url, data=None):
33: """returns result from url+data HTTP request"""
34: return documentViewer.getHttpData(url,data,timeout=self.timeout)
35:
36: def getServerData(self, method, data=None):
37: """returns result from text server for method+data"""
38: url = self.serverUrl+method
39: return documentViewer.getHttpData(url,data,timeout=self.timeout)
40:
41: def getSearch(self, pn=1, pageinfo=None, docinfo=None, query=None, queryType=None, lemma=None):
42: """get search list"""
43: docpath = docinfo['textURLPath']
44: url = docinfo['url']
1.23 abukhman 45: logging.debug("documentViewer (gettoc) docpath: %s"%(docpath))
46: logging.debug("documentViewer (gettoc) url: %s"%(url))
1.2 casties 47: pagesize = pageinfo['queryPageSize']
48: pn = pageinfo['searchPN']
49: sn = pageinfo['sn']
50: highlightQuery = pageinfo['highlightQuery']
1.34 abukhman 51: query =pageinfo['query']
1.2 casties 52: queryType =pageinfo['queryType']
53: viewMode= pageinfo['viewMode']
54: tocMode = pageinfo['tocMode']
1.24 abukhman 55: characterNormalization = pageinfo['characterNormalization']
1.2 casties 56: tocPN = pageinfo['tocPN']
57: selfurl = self.absolute_url()
58:
1.41 abukhman 59: data = self.getServerData("doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&sn=%s&viewMode=%s&highlightQuery=%s"%(docpath, 'text', queryType, urllib.quote(query), pagesize, pn, sn, viewMode,urllib.quote(highlightQuery)))
1.2 casties 60: #page=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&sn=%s&viewMode=%s&highlightQuery=%s"%(docpath, 'text', queryType, query, pagesize, pn, sn, viewMode,highlightQuery) ,outputUnicode=False)
61:
62: pagexml = data.replace('?document=%s'%str(docpath),'?url=%s'%url)
63: pagedom = Parse(pagexml)
64: if (queryType=="fulltext")or(queryType=="xpath")or(queryType=="xquery")or(queryType=="fulltextMorphLemma"):
65: pagedivs = pagedom.xpath("//div[@class='queryResultPage']")
66: if len(pagedivs)>0:
67: pagenode=pagedivs[0]
68: links=pagenode.xpath("//a")
69: for l in links:
70: hrefNode = l.getAttributeNodeNS(None, u"href")
71: if hrefNode:
72: href = hrefNode.nodeValue
73: if href.startswith('page-fragment.xql'):
74: selfurl = self.absolute_url()
1.30 abukhman 75: pagexml=href.replace('mode=text','mode=texttool&viewMode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&tocMode=%s&searchPN=%s&tocPN=%s'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN))
1.2 casties 76: hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl)
77: return serializeNode(pagenode)
78: if (queryType=="fulltextMorph"):
79: pagedivs = pagedom.xpath("//div[@class='queryResult']")
80: if len(pagedivs)>0:
81: pagenode=pagedivs[0]
82: links=pagenode.xpath("//a")
83: for l in links:
84: hrefNode = l.getAttributeNodeNS(None, u"href")
85: if hrefNode:
86: href = hrefNode.nodeValue
87: if href.startswith('page-fragment.xql'):
88: selfurl = self.absolute_url()
1.30 abukhman 89: pagexml=href.replace('mode=text','mode=texttool&viewMode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s&tocMode=%s&searchPN=%s&tocPN=%s'%(viewMode,queryType,urllib.quote(query),pagesize,pn,tocMode,pn,tocPN))
1.2 casties 90: hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl)
91: if href.startswith('../lt/lemma.xql'):
92: hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_lemma_New'%(selfurl))
93: l.setAttributeNS(None, 'target', '_blank')
94: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;")
1.6 abukhman 95: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 96: pagedivs = pagedom.xpath("//div[@class='queryResultMorphExpansion']")
97: return serializeNode(pagenode)
98: if (queryType=="ftIndex")or(queryType=="ftIndexMorph"):
99: pagedivs= pagedom.xpath("//div[@class='queryResultPage']")
100: if len(pagedivs)>0:
101: pagenode=pagedivs[0]
102: links=pagenode.xpath("//a")
103: for l in links:
104: hrefNode = l.getAttributeNodeNS(None, u"href")
105: if hrefNode:
106: href = hrefNode.nodeValue
1.18 abukhman 107: hrefNode.nodeValue=href.replace('mode=text','mode=texttool&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s'%(viewMode,tocMode,tocPN,pn))
1.2 casties 108: if href.startswith('../lt/lex.xql'):
109: hrefNode.nodeValue = href.replace('../lt/lex.xql','%s/template/head_main_voc'%selfurl)
110: l.setAttributeNS(None, 'target', '_blank')
111: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;")
1.6 abukhman 112: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 113: if href.startswith('../lt/lemma.xql'):
114: hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_lemma'%selfurl)
115: l.setAttributeNS(None, 'target', '_blank')
116: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;")
1.6 abukhman 117: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 118: return serializeNode(pagenode)
119: return "no text here"
120:
1.4 casties 121: def getNumPages(self, docinfo):
1.2 casties 122: """get list of pages from fulltext and put in docinfo"""
123: if 'numPages' in docinfo:
124: # already there
1.43 abukhman 125: return docinfo
1.2 casties 126: xquery = '//pb'
127: text = self.getServerData("xquery.xql","document=%s&xquery=%s"%(docinfo['textURLPath'],xquery))
128: docinfo['numPages'] = text.count("<pb ")
129: return docinfo
1.43 abukhman 130:
131: def getNumTextPages (self, docinfo):
132: """get list of pages from fulltext (texts without images) and put in docinfo"""
133: if 'numPages' in docinfo:
134: # allredy there
135: return docinfo
136: xpath ='/count(//pb)'
1.45 abukhman 137: text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'], xpath))
1.44 abukhman 138: logging.debug("documentViewer (text) text: %s"%(text))
1.48 ! abukhman 139: #docinfo['numPages']=text('resultPage')
! 140: docinfo['numPages']= pagedom.text("//resultPage")
! 141: #logging.debug("documentViewer (characterNormalization) docinfo['numPages']: %s"%(docinfo['numPages']))
1.43 abukhman 142: return docinfo
1.2 casties 143:
1.25 abukhman 144: def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None, highlightQuery=None,sn=None, viewMode=None, tocMode=None, tocPN=None, characterNormalization=""):
1.2 casties 145: """returns single page from fulltext"""
146: docpath = docinfo['textURLPath']
147: path = docinfo['textURLPath']
148: url = docinfo['url']
149: viewMode= pageinfo['viewMode']
150: tocMode = pageinfo['tocMode']
1.20 abukhman 151: characterNormalization=pageinfo['characterNormalization']
1.2 casties 152: tocPN = pageinfo['tocPN']
153: selfurl = self.absolute_url()
154: if mode == "text_dict":
155: textmode = "textPollux"
156: else:
157: textmode = mode
1.19 abukhman 158: #logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
1.26 abukhman 159: textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization)
1.2 casties 160: if highlightQuery is not None:
1.40 abukhman 161: textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn)
1.2 casties 162:
1.38 abukhman 163: pagexml = self.getServerData("page-fragment.xql",textParam)
1.2 casties 164: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/page-fragment.xql", textParam, outputUnicode=False)
165:
1.39 abukhman 166: pagedom = Parse(pagexml)
1.2 casties 167: # plain text mode
168: if mode == "text":
169: # first div contains text
170: pagedivs = pagedom.xpath("/div")
171: if len(pagedivs) > 0:
172: pagenode = pagedivs[0]
173: links = pagenode.xpath("//a")
174: for l in links:
175: hrefNode = l.getAttributeNodeNS(None, u"href")
176: if hrefNode:
177: href= hrefNode.nodeValue
178: if href.startswith('#note-'):
1.27 abukhman 179: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
1.2 casties 180: return serializeNode(pagenode)
181: if mode == "xml":
182: # first div contains text
183: pagedivs = pagedom.xpath("/div")
184: if len(pagedivs) > 0:
185: pagenode = pagedivs[0]
186: return serializeNode(pagenode)
1.7 abukhman 187: if mode == "gis":
188: # first div contains text
189: pagedivs = pagedom.xpath("/div")
190: if len(pagedivs) > 0:
191: pagenode = pagedivs[0]
1.28 abukhman 192: links =pagenode.xpath("//a")
193: for l in links:
194: hrefNode =l.getAttributeNodeNS(None, u"href")
195: if hrefNode:
196: href=hrefNode.nodeValue
197: if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'):
198: hrefNode.nodeValue =href.replace('chinagis_REST','chinagis')
1.42 abukhman 199: l.setAttributeNS(None, 'target', '_blank')
1.7 abukhman 200: return serializeNode(pagenode)
201:
1.2 casties 202: if mode == "pureXml":
203: # first div contains text
204: pagedivs = pagedom.xpath("/div")
205: if len(pagedivs) > 0:
206: pagenode = pagedivs[0]
207: return serializeNode(pagenode)
208: # text-with-links mode
209: if mode == "text_dict":
210: # first div contains text
211: pagedivs = pagedom.xpath("/div")
212: if len(pagedivs) > 0:
213: pagenode = pagedivs[0]
214: # check all a-tags
215: links = pagenode.xpath("//a")
216: for l in links:
217: hrefNode = l.getAttributeNodeNS(None, u"href")
218: if hrefNode:
219: # is link with href
220: href = hrefNode.nodeValue
221: if href.startswith('lt/lex.xql'):
222: # is pollux link
223: selfurl = self.absolute_url()
224: # change href
225: hrefNode.nodeValue = href.replace('lt/lex.xql','%s/template/head_main_voc'%selfurl)
226: # add target
227: l.setAttributeNS(None, 'target', '_blank')
228: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
1.6 abukhman 229: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 230: if href.startswith('lt/lemma.xql'):
231: selfurl = self.absolute_url()
232: hrefNode.nodeValue = href.replace('lt/lemma.xql','%s/template/head_main_lemma'%selfurl)
233: l.setAttributeNS(None, 'target', '_blank')
234: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
1.6 abukhman 235: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 236: if href.startswith('#note-'):
1.19 abukhman 237: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
1.2 casties 238: return serializeNode(pagenode)
239: return "no text here"
240:
241: def getTranslate(self, query=None, language=None):
242: """translate into another languages"""
1.5 casties 243: data = self.getServerData("lt/lex.xql","document=&language="+str(language)+"&query="+urllib.quote(query))
1.2 casties 244: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lex.xql","document=&language="+str(language)+"&query="+url_quote(str(query)))
245: return data
246:
247: def getLemma(self, lemma=None, language=None):
248: """simular words lemma """
1.5 casties 249: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(lemma))
1.2 casties 250: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(lemma)))
251: return data
252:
253: def getLemmaNew(self, query=None, language=None):
254: """simular words lemma """
1.5 casties 255: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(query))
1.2 casties 256: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(query)))
257: return data
1.28 abukhman 258:
1.2 casties 259: def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1):
260: """number of"""
261: docpath = docinfo['textURLPath']
262: pagesize = pageinfo['queryPageSize']
263: pn = pageinfo['searchPN']
1.34 abukhman 264: query =pageinfo['query']
1.2 casties 265: queryType =pageinfo['queryType']
266: tocSearch = 0
267: tocDiv = None
268:
1.32 abukhman 269: 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))
1.2 casties 270: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath, 'text', queryType, query, pagesize, pn) ,outputUnicode=False)
271: pagedom = Parse(pagexml)
272: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
273: tocSearch = int(getTextFromNode(numdivs[0]))
274: tc=int((tocSearch/10)+1)
1.23 abukhman 275: logging.debug("documentViewer (gettoc) tc: %s"%(tc))
1.2 casties 276: return tc
277:
278: def getToc(self, mode="text", docinfo=None):
279: """loads table of contents and stores in docinfo"""
1.23 abukhman 280: logging.debug("documentViewer (gettoc) mode: %s"%(mode))
1.2 casties 281: if mode == "none":
282: return docinfo
283: if 'tocSize_%s'%mode in docinfo:
284: # cached toc
285: return docinfo
286:
287: docpath = docinfo['textURLPath']
288: # we need to set a result set size
289: pagesize = 1000
290: pn = 1
291: if mode == "text":
292: queryType = "toc"
293: else:
294: queryType = mode
295: # number of entries in toc
296: tocSize = 0
297: tocDiv = None
298:
299: pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn))
300: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql", "document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType,pagesize,pn), outputUnicode=False)
301: # post-processing downloaded xml
302: pagedom = Parse(pagexml)
303: # get number of entries
304: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
305: if len(numdivs) > 0:
306: tocSize = int(getTextFromNode(numdivs[0]))
307: docinfo['tocSize_%s'%mode] = tocSize
308: return docinfo
309:
310: def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None):
311: """returns single page from the table of contents"""
312: # TODO: this should use the cached TOC
313: if mode == "text":
314: queryType = "toc"
315: else:
316: queryType = mode
317: docpath = docinfo['textURLPath']
318: path = docinfo['textURLPath']
319: pagesize = pageinfo['tocPageSize']
320: pn = pageinfo['tocPN']
321: url = docinfo['url']
322: selfurl = self.absolute_url()
323: viewMode= pageinfo['viewMode']
1.26 abukhman 324: characterNormalization = pageinfo ['characterNormalization']
1.2 casties 325: tocMode = pageinfo['tocMode']
326: tocPN = pageinfo['tocPN']
327:
1.23 abukhman 328: data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=%s"%(docpath,queryType, pagesize, pn,characterNormalization))
329: page = data.replace('page-fragment.xql?document=%s'%str(path),'%s?url=%s&viewMode=%s&tocMode=%s&tocPN=%s'%(selfurl,url, viewMode, tocMode, tocPN))
1.2 casties 330: text = page.replace('mode=image','mode=texttool')
1.21 abukhman 331: logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
1.19 abukhman 332: #logging.debug("documentViewer (characterNormalization) text: %s"%(text))
1.2 casties 333: return text
334:
335: def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
336: """change settings"""
337: self.title=title
338: self.timeout = timeout
339: self.serverUrl = serverUrl
340: if RESPONSE is not None:
341: RESPONSE.redirect('manage_main')
342:
343: # management methods
344: def manage_addMpdlXmlTextServerForm(self):
345: """Form for adding"""
346: pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self)
347: return pt()
348:
349: def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
350: """add zogiimage"""
351: newObj = MpdlXmlTextServer(id,title,serverUrl,timeout)
352: self.Destination()._setObject(id, newObj)
353: if RESPONSE is not None:
354: RESPONSE.redirect('manage_main')
355:
356:
1.4 casties 357:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>