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