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