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