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
8: import urllib
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:
22: def __init__(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/", serverName=None, timeout=40):
23: """constructor"""
24: self.id=id
25: self.title=title
26: self.timeout = timeout
27: if serverName is None:
28: self.serverUrl = serverUrl
29: else:
30: self.serverUrl = "http://%s/mpdl/interface/"%serverName
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']
45: logging.debug("documentViewer (gettoc) docpath: %s"%(docpath))
46: logging.debug("documentViewer (gettoc) url: %s"%(url))
47: pagesize = pageinfo['queryPageSize']
48: pn = pageinfo['searchPN']
49: sn = pageinfo['sn']
50: highlightQuery = pageinfo['highlightQuery']
51: query =pageinfo['query']
52: queryType =pageinfo['queryType']
53: viewMode= pageinfo['viewMode']
54: tocMode = pageinfo['tocMode']
55: characterNormalization = pageinfo['characterNormalization']
56: tocPN = pageinfo['tocPN']
57: selfurl = self.absolute_url()
58:
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)))
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()
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))
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()
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))
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;")
95: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
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
107: hrefNode.nodeValue=href.replace('mode=text','mode=texttool&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s'%(viewMode,tocMode,tocPN,pn))
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;")
112: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
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;")
117: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
118: return serializeNode(pagenode)
119: return "no text here"
120:
121: """def getNumPages(self, docinfo):
122: ""get list of pages from fulltext and put in docinfo""
123: if 'numPages' in docinfo:
124: # already there
125: return docinfo
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
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)'
137: text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'], xpath))
138: dom = Parse(text)
139: result= dom.xpath("//result/resultPage")
140: docinfo['numPages']=int(getTextFromNode(result[0]))
141: return docinfo
142:
143: def getGisPlaces(self, docinfo=None, pageinfo=None):
144: """ Show all Gis Places of whole Page"""
145: xpath='//place'
146: docpath = docinfo['textURLPath']
147: url = docinfo['url']
148: selfurl = self.absolute_url()
149: pn = pageinfo['current']
150: hrefList=[]
151:
152: text=self.getServerData("xpath.xql", "document=%s&xpath=%s&pn=%s"%(docinfo['textURLPath'],xpath,pn))
153: dom = Parse(text)
154: result = dom.xpath("//result/resultPage/place")
155: for l in result:
156: hrefNode= l.getAttributeNodeNS(None, u"id")
157: href= hrefNode.nodeValue
158: hrefList.add(href)
159: #href=href.href
160: #href = ",".join(href)
161: logging.debug("getGisPlaces :%s"%(hrefList))
162:
163: #return ";".join(["%s=%s" % (href, href) for href, href in params.items()])
164: #logging.debug("getGisPlaces :%s"%(href))
165: #if href.startswith('id='):
166: #hrefNode.nodeValue = href.replace('id=',"?")
167: #logging.debug("documentViewer getGisPlaces (characterNormalization) hrefNode.nodeValue:%s"%(hrefNode.nodeValue))
168: #logging.debug("getGisPlaces :%s"%(href))
169: return hrefList
170: #return "no text here"
171:
172: def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None, highlightQuery=None,sn=None, viewMode=None, tocMode=None, tocPN=None, characterNormalization=""):
173: """returns single page from fulltext"""
174: docpath = docinfo['textURLPath']
175: path = docinfo['textURLPath']
176: url = docinfo['url']
177: name = docinfo['name']
178: viewMode= pageinfo['viewMode']
179: tocMode = pageinfo['tocMode']
180: characterNormalization=pageinfo['characterNormalization']
181: tocPN = pageinfo['tocPN']
182: selfurl = self.absolute_url()
183: if mode == "text_dict":
184: textmode = "textPollux"
185: else:
186: textmode = mode
187: #logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
188: textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization)
189: if highlightQuery is not None:
190: textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn)
191:
192: pagexml = self.getServerData("page-fragment.xql",textParam)
193: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/page-fragment.xql", textParam, outputUnicode=False)
194:
195: pagedom = Parse(pagexml)
196: # plain text mode
197: if mode == "text":
198: # first div contains text
199: pagedivs = pagedom.xpath("/div")
200: if len(pagedivs) > 0:
201: pagenode = pagedivs[0]
202: links = pagenode.xpath("//a")
203: for l in links:
204: hrefNode = l.getAttributeNodeNS(None, u"href")
205: if hrefNode:
206: href= hrefNode.nodeValue
207: if href.startswith('#note-'):
208: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
209: return serializeNode(pagenode)
210: if mode == "xml":
211: # first div contains text
212: pagedivs = pagedom.xpath("/div")
213: if len(pagedivs) > 0:
214: pagenode = pagedivs[0]
215: return serializeNode(pagenode)
216: if mode == "gis":
217: # first div contains text
218: pagedivs = pagedom.xpath("/div")
219: if len(pagedivs) > 0:
220: pagenode = pagedivs[0]
221: links =pagenode.xpath("//a")
222: for l in links:
223: hrefNode =l.getAttributeNodeNS(None, u"href")
224: if hrefNode:
225: href=hrefNode.nodeValue
226: if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'):
227: hrefNode.nodeValue =href.replace('chinagis_REST/REST/db/chgis/mpdl','chinagis/REST/db/mpdl/%s'%name)
228: l.setAttributeNS(None, 'target', '_blank')
229: return serializeNode(pagenode)
230:
231: if mode == "pureXml":
232: # first div contains text
233: pagedivs = pagedom.xpath("/div")
234: if len(pagedivs) > 0:
235: pagenode = pagedivs[0]
236: return serializeNode(pagenode)
237: # text-with-links mode
238: if mode == "text_dict":
239: # first div contains text
240: pagedivs = pagedom.xpath("/div")
241: if len(pagedivs) > 0:
242: pagenode = pagedivs[0]
243: # check all a-tags
244: links = pagenode.xpath("//a")
245: for l in links:
246: hrefNode = l.getAttributeNodeNS(None, u"href")
247: if hrefNode:
248: # is link with href
249: href = hrefNode.nodeValue
250: if href.startswith('lt/lex.xql'):
251: # is pollux link
252: selfurl = self.absolute_url()
253: # change href
254: hrefNode.nodeValue = href.replace('lt/lex.xql','%s/template/head_main_voc'%selfurl)
255: # add target
256: l.setAttributeNS(None, 'target', '_blank')
257: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
258: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
259: if href.startswith('lt/lemma.xql'):
260: selfurl = self.absolute_url()
261: hrefNode.nodeValue = href.replace('lt/lemma.xql','%s/template/head_main_lemma'%selfurl)
262: l.setAttributeNS(None, 'target', '_blank')
263: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
264: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
265: if href.startswith('#note-'):
266: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
267: return serializeNode(pagenode)
268: return "no text here"
269:
270: def getTranslate(self, query=None, language=None):
271: """translate into another languages"""
272: data = self.getServerData("lt/lex.xql","document=&language="+str(language)+"&query="+urllib.quote(query))
273: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lex.xql","document=&language="+str(language)+"&query="+url_quote(str(query)))
274: return data
275:
276: def getLemma(self, lemma=None, language=None):
277: """simular words lemma """
278: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(lemma))
279: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(lemma)))
280: return data
281:
282: def getLemmaNew(self, query=None, language=None):
283: """simular words lemma """
284: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(query))
285: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(query)))
286: return data
287:
288: def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1):
289: """number of"""
290: docpath = docinfo['textURLPath']
291: pagesize = pageinfo['queryPageSize']
292: pn = pageinfo['searchPN']
293: query =pageinfo['query']
294: queryType =pageinfo['queryType']
295: tocSearch = 0
296: tocDiv = None
297:
298: 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))
299: #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)
300: pagedom = Parse(pagexml)
301: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
302: tocSearch = int(getTextFromNode(numdivs[0]))
303: tc=int((tocSearch/10)+1)
304: logging.debug("documentViewer (gettoc) tc: %s"%(tc))
305: return tc
306:
307: def getToc(self, mode="text", docinfo=None):
308: """loads table of contents and stores in docinfo"""
309: logging.debug("documentViewer (gettoc) mode: %s"%(mode))
310: if mode == "none":
311: return docinfo
312: if 'tocSize_%s'%mode in docinfo:
313: # cached toc
314: return docinfo
315:
316: docpath = docinfo['textURLPath']
317: # we need to set a result set size
318: pagesize = 1000
319: pn = 1
320: if mode == "text":
321: queryType = "toc"
322: else:
323: queryType = mode
324: # number of entries in toc
325: tocSize = 0
326: tocDiv = None
327:
328: pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn))
329: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql", "document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType,pagesize,pn), outputUnicode=False)
330: # post-processing downloaded xml
331: pagedom = Parse(pagexml)
332: # get number of entries
333: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
334: if len(numdivs) > 0:
335: tocSize = int(getTextFromNode(numdivs[0]))
336: docinfo['tocSize_%s'%mode] = tocSize
337: return docinfo
338:
339: def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None):
340: """returns single page from the table of contents"""
341: # TODO: this should use the cached TOC
342: if mode == "text":
343: queryType = "toc"
344: else:
345: queryType = mode
346: docpath = docinfo['textURLPath']
347: path = docinfo['textURLPath']
348: pagesize = pageinfo['tocPageSize']
349: pn = pageinfo['tocPN']
350: url = docinfo['url']
351: selfurl = self.absolute_url()
352: viewMode= pageinfo['viewMode']
353: characterNormalization = pageinfo ['characterNormalization']
354: tocMode = pageinfo['tocMode']
355: tocPN = pageinfo['tocPN']
356:
357: data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=%s"%(docpath,queryType, pagesize, pn,characterNormalization))
358: page = data.replace('page-fragment.xql?document=%s'%str(path),'%s?url=%s&viewMode=%s&tocMode=%s&tocPN=%s'%(selfurl,url, viewMode, tocMode, tocPN))
359: text = page.replace('mode=image','mode=texttool')
360: logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
361: #logging.debug("documentViewer (characterNormalization) text: %s"%(text))
362: return text
363:
364: def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
365: """change settings"""
366: self.title=title
367: self.timeout = timeout
368: self.serverUrl = serverUrl
369: if RESPONSE is not None:
370: RESPONSE.redirect('manage_main')
371:
372: # management methods
373: def manage_addMpdlXmlTextServerForm(self):
374: """Form for adding"""
375: pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self)
376: return pt()
377:
378: def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
379: """add zogiimage"""
380: newObj = MpdlXmlTextServer(id,title,serverUrl,timeout)
381: self.Destination()._setObject(id, newObj)
382: if RESPONSE is not None:
383: RESPONSE.redirect('manage_main')
384:
385:
386:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>