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: pagesize = pageinfo['queryPageSize']
46: pn = pageinfo['searchPN']
47: sn = pageinfo['sn']
48: highlightQuery = pageinfo['highlightQuery']
49: query =pageinfo['query']
50: queryType =pageinfo['queryType']
51: viewMode= pageinfo['viewMode']
52: tocMode = pageinfo['tocMode']
53: characterNormalization = pageinfo['characterNormalization']
54: tocPN = pageinfo['tocPN']
55: selfurl = self.absolute_url()
56:
57: 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)))
58: #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)
59:
60: pagexml = data.replace('?document=%s'%str(docpath),'?url=%s'%url)
61: pagedom = Parse(pagexml)
62: if (queryType=="fulltext")or(queryType=="xpath")or(queryType=="xquery")or(queryType=="fulltextMorphLemma"):
63: pagedivs = pagedom.xpath("//div[@class='queryResultPage']")
64: if len(pagedivs)>0:
65: pagenode=pagedivs[0]
66: links=pagenode.xpath("//a")
67: for l in links:
68: hrefNode = l.getAttributeNodeNS(None, u"href")
69: if hrefNode:
70: href = hrefNode.nodeValue
71: if href.startswith('page-fragment.xql'):
72: selfurl = self.absolute_url()
73: 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))
74: hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl)
75: return serializeNode(pagenode)
76: if (queryType=="fulltextMorph"):
77: pagedivs = pagedom.xpath("//div[@class='queryResult']")
78: if len(pagedivs)>0:
79: pagenode=pagedivs[0]
80: links=pagenode.xpath("//a")
81: for l in links:
82: hrefNode = l.getAttributeNodeNS(None, u"href")
83: if hrefNode:
84: href = hrefNode.nodeValue
85: if href.startswith('page-fragment.xql'):
86: selfurl = self.absolute_url()
87: 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))
88: hrefNode.nodeValue = pagexml.replace('page-fragment.xql','%s'%selfurl)
89: if href.startswith('../lt/lemma.xql'):
90: hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_lemma_New'%(selfurl))
91: l.setAttributeNS(None, 'target', '_blank')
92: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;")
93: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
94: pagedivs = pagedom.xpath("//div[@class='queryResultMorphExpansion']")
95: return serializeNode(pagenode)
96: if (queryType=="ftIndex")or(queryType=="ftIndexMorph"):
97: pagedivs= pagedom.xpath("//div[@class='queryResultPage']")
98: if len(pagedivs)>0:
99: pagenode=pagedivs[0]
100: links=pagenode.xpath("//a")
101: for l in links:
102: hrefNode = l.getAttributeNodeNS(None, u"href")
103: if hrefNode:
104: href = hrefNode.nodeValue
105: hrefNode.nodeValue=href.replace('mode=text','mode=texttool&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s'%(viewMode,tocMode,tocPN,pn))
106: if href.startswith('../lt/lex.xql'):
107: hrefNode.nodeValue = href.replace('../lt/lex.xql','%s/template/head_main_voc'%selfurl)
108: l.setAttributeNS(None, 'target', '_blank')
109: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;")
110: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
111: if href.startswith('../lt/lemma.xql'):
112: hrefNode.nodeValue = href.replace('../lt/lemma.xql','%s/template/head_main_lemma'%selfurl)
113: l.setAttributeNS(None, 'target', '_blank')
114: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=400, scrollbars=1'); return false;")
115: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
116: return serializeNode(pagenode)
117: return "no text here"
118:
119: """def getNumPages(self, docinfo):
120: ""get list of pages from fulltext and put in docinfo""
121: if 'numPages' in docinfo:
122: # already there
123: return docinfo
124: xquery = '//pb'
125: text = self.getServerData("xquery.xql","document=%s&xquery=%s"%(docinfo['textURLPath'],xquery))
126: docinfo['numPages'] = text.count("<pb ")
127: return docinfo
128: """
129: def getNumTextPages (self, docinfo):
130: """get list of pages from fulltext (texts without images) and put in docinfo"""
131: if 'numPages' in docinfo:
132: # allredy there
133: return docinfo
134: xpath ='/count(//pb)'
135: text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'], xpath))
136: dom = Parse(text)
137: result= dom.xpath("//result/resultPage")
138: docinfo['numPages']=int(getTextFromNode(result[0]))
139: return docinfo
140:
141: def getGisPlaces(self, docinfo=None, pageinfo=None):
142: """ Show all Gis Places of whole Page"""
143: xpath='//place'
144: docpath = docinfo['textURLPath']
145: url = docinfo['url']
146: selfurl = self.absolute_url()
147: pn = pageinfo['current']
148: hrefList=[]
149: myList= ""
150: text=self.getServerData("xpath.xql", "document=%s&xpath=%s&pn=%s"%(docinfo['textURLPath'],xpath,pn))
151: dom = Parse(text)
152: result = dom.xpath("//result/resultPage/place")
153: for l in result:
154: hrefNode= l.getAttributeNodeNS(None, u"id")
155: href= hrefNode.nodeValue
156: hrefList.append(href)
157: myList = ",".join(hrefList)
158: logging.debug("getGisPlaces :%s"%(myList))
159: return myList
160:
161: def getAllGisPlaces (self, docinfo=None, pageinfo=None):
162: """Show all Gis Places of whole Book """
163: xpath ='//echo:place'
164: docpath =docinfo['textURLPath']
165: url = docinfo['url']
166: selfurl =self.absolute_url()
167: pn =pageinfo['current']
168: hrefList=[]
169: myList=""
170: text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'],xpath))
171: dom =Parse(text)
172: result = dom.xpath("//result/resultPage/place")
173: for l in result:
174: hrefNode = l.getAttributeNodeNS(None, u"id")
175: href= hrefNode.nodeValue
176: hrefList.append(href)
177: myList = ",".join(hrefList)
178: logging.debug("getALLGisPlaces :%s"%(myList))
179: return myList
180:
181: def getOrigPages (self, docinfo=None, pageinfo=None):
182: """Show original page """
183: docpath = docinfo['textURLPath']
184: #url = docinfo['url']
185: selfurl = self.absolute_url()
186: pn =pageinfo['current']
187: viewMode= pageinfo['viewMode']
188: text = self.getServerData("page-fragment.xql","document=%s&mode=%s&pn=%s"%(docinfo['textURLPath'], mode, pn))
189: dom =Parse(text)
190: pagedivs = dom.xpath("//div[@class='pageNumberOrig']")
191: logging.debug("getOrigPages :%s"%(pagedivs))
192: return "pagedivs"
193:
194: def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None, highlightQuery=None,sn=None, viewMode=None, tocMode=None, tocPN=None, characterNormalization=""):
195: """returns single page from fulltext"""
196: docpath = docinfo['textURLPath']
197: path = docinfo['textURLPath']
198: url = docinfo['url']
199: name = docinfo['name']
200: viewMode= pageinfo['viewMode']
201: tocMode = pageinfo['tocMode']
202: characterNormalization=pageinfo['characterNormalization']
203: tocPN = pageinfo['tocPN']
204: selfurl = self.absolute_url()
205: if mode == "text_dict":
206: textmode = "textPollux"
207: else:
208: textmode = mode
209: #logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
210: textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization)
211: if highlightQuery is not None:
212: textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn)
213:
214: pagexml = self.getServerData("page-fragment.xql",textParam)
215: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/page-fragment.xql", textParam, outputUnicode=False)
216:
217: pagedom = Parse(pagexml)
218: # plain text mode
219: if mode == "text":
220: # first div contains text
221: pagedivs = pagedom.xpath("/div")
222: if len(pagedivs) > 0:
223: pagenode = pagedivs[0]
224: links = pagenode.xpath("//a")
225: for l in links:
226: hrefNode = l.getAttributeNodeNS(None, u"href")
227: if hrefNode:
228: href= hrefNode.nodeValue
229: if href.startswith('#note-'):
230: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
231: return serializeNode(pagenode)
232: if mode == "xml":
233: # first div contains text
234: pagedivs = pagedom.xpath("/div")
235: if len(pagedivs) > 0:
236: pagenode = pagedivs[0]
237: return serializeNode(pagenode)
238: if mode == "gis":
239: # first div contains text
240: pagedivs = pagedom.xpath("/div")
241: if len(pagedivs) > 0:
242: pagenode = pagedivs[0]
243: links =pagenode.xpath("//a")
244: for l in links:
245: hrefNode =l.getAttributeNodeNS(None, u"href")
246: if hrefNode:
247: href=hrefNode.nodeValue
248: if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'):
249: hrefNode.nodeValue =href.replace('chinagis_REST/REST/db/chgis/mpdl','chinagis/REST/db/mpdl/%s'%name)
250: l.setAttributeNS(None, 'target', '_blank')
251: return serializeNode(pagenode)
252:
253: if mode == "pureXml":
254: # first div contains text
255: pagedivs = pagedom.xpath("/div")
256: if len(pagedivs) > 0:
257: pagenode = pagedivs[0]
258: return serializeNode(pagenode)
259: # text-with-links mode
260: if mode == "text_dict":
261: # first div contains text
262: pagedivs = pagedom.xpath("/div")
263: if len(pagedivs) > 0:
264: pagenode = pagedivs[0]
265: # check all a-tags
266: links = pagenode.xpath("//a")
267: for l in links:
268: hrefNode = l.getAttributeNodeNS(None, u"href")
269: if hrefNode:
270: # is link with href
271: href = hrefNode.nodeValue
272: if href.startswith('lt/lex.xql'):
273: # is pollux link
274: selfurl = self.absolute_url()
275: # change href
276: hrefNode.nodeValue = href.replace('lt/lex.xql','%s/template/head_main_voc'%selfurl)
277: # add target
278: l.setAttributeNS(None, 'target', '_blank')
279: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
280: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
281: if href.startswith('lt/lemma.xql'):
282: selfurl = self.absolute_url()
283: hrefNode.nodeValue = href.replace('lt/lemma.xql','%s/template/head_main_lemma'%selfurl)
284: l.setAttributeNS(None, 'target', '_blank')
285: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
286: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
287: if href.startswith('#note-'):
288: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
289: return serializeNode(pagenode)
290: return "no text here"
291:
292: def getTranslate(self, query=None, language=None):
293: """translate into another languages"""
294: data = self.getServerData("lt/lex.xql","document=&language="+str(language)+"&query="+urllib.quote(query))
295: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lex.xql","document=&language="+str(language)+"&query="+url_quote(str(query)))
296: return data
297:
298: def getLemma(self, lemma=None, language=None):
299: """simular words lemma """
300: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(lemma))
301: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(lemma)))
302: return data
303:
304: def getLemmaNew(self, query=None, language=None):
305: """simular words lemma """
306: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(query))
307: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(query)))
308: return data
309:
310: def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1):
311: """number of"""
312: docpath = docinfo['textURLPath']
313: pagesize = pageinfo['queryPageSize']
314: pn = pageinfo['searchPN']
315: query =pageinfo['query']
316: queryType =pageinfo['queryType']
317: tocSearch = 0
318: tocDiv = None
319:
320: 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))
321: #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)
322: pagedom = Parse(pagexml)
323: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
324: tocSearch = int(getTextFromNode(numdivs[0]))
325: tc=int((tocSearch/10)+1)
326: logging.debug("documentViewer (gettoc) tc: %s"%(tc))
327: return tc
328:
329: def getToc(self, mode="text", docinfo=None):
330: """loads table of contents and stores in docinfo"""
331: logging.debug("documentViewer (gettoc) mode: %s"%(mode))
332: if mode == "none":
333: return docinfo
334: if 'tocSize_%s'%mode in docinfo:
335: # cached toc
336: return docinfo
337:
338: docpath = docinfo['textURLPath']
339: # we need to set a result set size
340: pagesize = 1000
341: pn = 1
342: if mode == "text":
343: queryType = "toc"
344: else:
345: queryType = mode
346: # number of entries in toc
347: tocSize = 0
348: tocDiv = None
349:
350: pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn))
351: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql", "document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType,pagesize,pn), outputUnicode=False)
352: # post-processing downloaded xml
353: pagedom = Parse(pagexml)
354: # get number of entries
355: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
356: if len(numdivs) > 0:
357: tocSize = int(getTextFromNode(numdivs[0]))
358: docinfo['tocSize_%s'%mode] = tocSize
359: return docinfo
360:
361: def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None):
362: """returns single page from the table of contents"""
363: # TODO: this should use the cached TOC
364: if mode == "text":
365: queryType = "toc"
366: else:
367: queryType = mode
368: docpath = docinfo['textURLPath']
369: path = docinfo['textURLPath']
370: pagesize = pageinfo['tocPageSize']
371: pn = pageinfo['tocPN']
372: url = docinfo['url']
373: selfurl = self.absolute_url()
374: viewMode= pageinfo['viewMode']
375: characterNormalization = pageinfo ['characterNormalization']
376: tocMode = pageinfo['tocMode']
377: tocPN = pageinfo['tocPN']
378:
379: data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=%s"%(docpath,queryType, pagesize, pn,characterNormalization))
380: page = data.replace('page-fragment.xql?document=%s'%str(path),'%s?url=%s&viewMode=%s&tocMode=%s&tocPN=%s'%(selfurl,url, viewMode, tocMode, tocPN))
381: text = page.replace('mode=image','mode=texttool')
382: logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
383: #logging.debug("documentViewer (characterNormalization) text: %s"%(text))
384: return text
385:
386: def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
387: """change settings"""
388: self.title=title
389: self.timeout = timeout
390: self.serverUrl = serverUrl
391: if RESPONSE is not None:
392: RESPONSE.redirect('manage_main')
393:
394: # management methods
395: def manage_addMpdlXmlTextServerForm(self):
396: """Form for adding"""
397: pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self)
398: return pt()
399:
400: def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
401: """add zogiimage"""
402: newObj = MpdlXmlTextServer(id,title,serverUrl,timeout)
403: self.Destination()._setObject(id, newObj)
404: if RESPONSE is not None:
405: RESPONSE.redirect('manage_main')
406:
407:
408:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>