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, characterNormalization=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&characterNormalization=%s&highlightQuery=%s"%(docpath, 'text', queryType, urllib.quote(query), pagesize, pn, sn, viewMode,characterNormalization ,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:
182: def getPDF (self, docinfo=None, pageinfo=None):
183: """Show and Save different Pages as PDF in Options"""
184: selfurl=self.absolute_url()
185: pn=pageinfo['current']
186: viewmode =pageinfo['viewMode']
187: text = self.getServerData("page-fragment.xql", "document=%s&mode=%s&pn=%s&export=%s"(docinfo['textURLPath'],mode, pn,'pdf'))
188: dom =Parse(text)
189: logging.debug("dom :%s"%(dom))
190: return dom
191:
192:
193: def getOrigPages (self, docinfo=None, pageinfo=None):
194: """Show original page """
195: docpath = docinfo['textURLPath']
196: #url = docinfo['url']
197: selfurl = self.absolute_url()
198: pn =pageinfo['current']
199:
200: viewMode= pageinfo['viewMode']
201: text = self.getServerData("page-fragment.xql","document=%s&mode=%s&pn=%s"%(docinfo['textURLPath'], 'text', pn))
202: dom =Parse(text)
203: pagedivs = dom.xpath("//div[@class='pageNumberOrig']")
204: logging.debug("YYYYYYpagedivs :%s"%(pagedivs))
205: if len(pagedivs)>0:
206: originalPage= getTextFromNode(pagedivs[0])
207: #return docinfo['originalPage']
208: return originalPage
209:
210:
211: def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None, viewMode=None, tocMode=None, tocPN=None, characterNormalization="", highlightQuery=None, sn=None):
212: """returns single page from fulltext"""
213: docpath = docinfo['textURLPath']
214: path = docinfo['textURLPath']
215: url = docinfo['url']
216: name = docinfo['name']
217: viewMode= pageinfo['viewMode']
218: sn = pageinfo['sn']
219: highlightQuery = pageinfo['highlightQuery']
220:
221: tocMode = pageinfo['tocMode']
222: characterNormalization=pageinfo['characterNormalization']
223: tocPN = pageinfo['tocPN']
224: selfurl = self.absolute_url()
225: if mode == "text_dict":
226: textmode = "textPollux"
227: else:
228: textmode = mode
229: #logging.debug("documentViewer highlightQuery: %s"%(highlightQuery))
230: textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization)
231: if highlightQuery is not None:
232: textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn)
233: #logging.debug("documentViewer highlightQuery: %s"%(highlightQuery))
234: pagexml = self.getServerData("page-fragment.xql",textParam)
235: logging.debug("documentViewer highlightQuery: %s"%(highlightQuery))
236: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/page-fragment.xql", textParam, outputUnicode=False)
237:
238: pagedom = Parse(pagexml)
239: # plain text mode
240: if mode == "text":
241: # first div contains text
242: pagedivs = pagedom.xpath("/div")
243: if len(pagedivs) > 0:
244: pagenode = pagedivs[0]
245: links = pagenode.xpath("//a")
246: for l in links:
247: hrefNode = l.getAttributeNodeNS(None, u"href")
248: if hrefNode:
249: href= hrefNode.nodeValue
250: if href.startswith('#note-'):
251: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
252: return serializeNode(pagenode)
253: if mode == "xml":
254: # first div contains text
255: pagedivs = pagedom.xpath("/div")
256: if len(pagedivs) > 0:
257: pagenode = pagedivs[0]
258: return serializeNode(pagenode)
259: if mode == "gis":
260: # first div contains text
261: pagedivs = pagedom.xpath("/div")
262: if len(pagedivs) > 0:
263: pagenode = pagedivs[0]
264: links =pagenode.xpath("//a")
265: for l in links:
266: hrefNode =l.getAttributeNodeNS(None, u"href")
267: if hrefNode:
268: href=hrefNode.nodeValue
269: if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'):
270: hrefNode.nodeValue =href.replace('chinagis_REST/REST/db/chgis/mpdl','chinagis/REST/db/mpdl/%s'%name)
271: l.setAttributeNS(None, 'target', '_blank')
272: return serializeNode(pagenode)
273:
274: if mode == "pureXml":
275: # first div contains text
276: pagedivs = pagedom.xpath("/div")
277: if len(pagedivs) > 0:
278: pagenode = pagedivs[0]
279: return serializeNode(pagenode)
280: # text-with-links mode
281: if mode == "text_dict":
282: # first div contains text
283: pagedivs = pagedom.xpath("/div")
284: if len(pagedivs) > 0:
285: pagenode = pagedivs[0]
286: # check all a-tags
287: links = pagenode.xpath("//a")
288: for l in links:
289: hrefNode = l.getAttributeNodeNS(None, u"href")
290: if hrefNode:
291: # is link with href
292: href = hrefNode.nodeValue
293: if href.startswith('lt/lex.xql'):
294: # is pollux link
295: selfurl = self.absolute_url()
296: # change href
297: hrefNode.nodeValue = href.replace('lt/lex.xql','%s/template/head_main_voc'%selfurl)
298: # add target
299: l.setAttributeNS(None, 'target', '_blank')
300: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
301: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
302: if href.startswith('lt/lemma.xql'):
303: selfurl = self.absolute_url()
304: hrefNode.nodeValue = href.replace('lt/lemma.xql','%s/template/head_main_lemma'%selfurl)
305: l.setAttributeNS(None, 'target', '_blank')
306: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
307: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
308: if href.startswith('#note-'):
309: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
310: return serializeNode(pagenode)
311: return "no text here"
312:
313: def getTranslate(self, query=None, language=None):
314: """translate into another languages"""
315: data = self.getServerData("lt/lex.xql","document=&language="+str(language)+"&query="+urllib.quote(query))
316: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lex.xql","document=&language="+str(language)+"&query="+url_quote(str(query)))
317: return data
318:
319: def getLemma(self, lemma=None, language=None):
320: """simular words lemma """
321: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(lemma))
322: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(lemma)))
323: return data
324:
325: def getLemmaNew(self, query=None, language=None):
326: """simular words lemma """
327: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(query))
328: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(query)))
329: return data
330:
331: def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1):
332: """number of"""
333: docpath = docinfo['textURLPath']
334: pagesize = pageinfo['queryPageSize']
335: pn = pageinfo['searchPN']
336: query =pageinfo['query']
337: queryType =pageinfo['queryType']
338: tocSearch = 0
339: tocDiv = None
340:
341: 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))
342: #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)
343: pagedom = Parse(pagexml)
344: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
345: tocSearch = int(getTextFromNode(numdivs[0]))
346: tc=int((tocSearch/10)+1)
347: logging.debug("documentViewer (gettoc) tc: %s"%(tc))
348: return tc
349:
350: def getToc(self, mode="text", docinfo=None):
351: """loads table of contents and stores in docinfo"""
352: logging.debug("documentViewer (gettoc) mode: %s"%(mode))
353: if mode == "none":
354: return docinfo
355: if 'tocSize_%s'%mode in docinfo:
356: # cached toc
357: return docinfo
358:
359: docpath = docinfo['textURLPath']
360: # we need to set a result set size
361: pagesize = 1000
362: pn = 1
363: if mode == "text":
364: queryType = "toc"
365: else:
366: queryType = mode
367: # number of entries in toc
368: tocSize = 0
369: tocDiv = None
370:
371: pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn))
372: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql", "document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType,pagesize,pn), outputUnicode=False)
373: # post-processing downloaded xml
374: pagedom = Parse(pagexml)
375: # get number of entries
376: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
377: if len(numdivs) > 0:
378: tocSize = int(getTextFromNode(numdivs[0]))
379: docinfo['tocSize_%s'%mode] = tocSize
380: return docinfo
381:
382: def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None):
383: """returns single page from the table of contents"""
384: # TODO: this should use the cached TOC
385: if mode == "text":
386: queryType = "toc"
387: else:
388: queryType = mode
389: docpath = docinfo['textURLPath']
390: path = docinfo['textURLPath']
391: pagesize = pageinfo['tocPageSize']
392: pn = pageinfo['tocPN']
393: url = docinfo['url']
394: selfurl = self.absolute_url()
395: viewMode= pageinfo['viewMode']
396: characterNormalization = pageinfo ['characterNormalization']
397: tocMode = pageinfo['tocMode']
398: tocPN = pageinfo['tocPN']
399:
400: data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=%s"%(docpath,queryType, pagesize, pn,characterNormalization))
401: page = data.replace('page-fragment.xql?document=%s'%str(path),'%s?url=%s&viewMode=%s&tocMode=%s&tocPN=%s'%(selfurl,url, viewMode, tocMode, tocPN))
402: text = page.replace('mode=image','mode=texttool')
403: logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
404: #logging.debug("documentViewer (characterNormalization) text: %s"%(text))
405: return text
406:
407: def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
408: """change settings"""
409: self.title=title
410: self.timeout = timeout
411: self.serverUrl = serverUrl
412: if RESPONSE is not None:
413: RESPONSE.redirect('manage_main')
414:
415: # management methods
416: def manage_addMpdlXmlTextServerForm(self):
417: """Form for adding"""
418: pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self)
419: return pt()
420:
421: def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
422: """add zogiimage"""
423: newObj = MpdlXmlTextServer(id,title,serverUrl,timeout)
424: self.Destination()._setObject(id, newObj)
425: if RESPONSE is not None:
426: RESPONSE.redirect('manage_main')
427:
428:
429:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>