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