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: pn = pageinfo['current']
150: hrefList=[]
151: myList= ""
152: text=self.getServerData("xpath.xql", "document=%s&xpath=%s&pn=%s"%(docinfo['textURLPath'],xpath,pn))
153: dom = Parse(text)
154: result = dom.xpath("//result/resultPage/place")
155: for l in result:
156: hrefNode= l.getAttributeNodeNS(None, u"id")
157: href= hrefNode.nodeValue
158: hrefList.append(href)
159: myList = ",".join(hrefList)
160: logging.debug("getGisPlaces :%s"%(myList))
161: return myList
162:
163: def getAllGisPlaces (self, docinfo=None, pageinfo=None):
164: """Show all Gis Places of whole Book """
165: xpath ='//echo:place'
166: docpath =docinfo['textURLPath']
167: url = docinfo['url']
168: selfurl =self.absolute_url()
169: pn =pageinfo['current']
170: hrefList=[]
171: myList=""
172: text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'],xpath))
173: dom =Parse(text)
174: result = dom.xpath("//result/resultPage/place")
175: for l in result:
176: hrefNode = l.getAttributeNodeNS(None, u"id")
177: href= hrefNode.nodeValue
178: hrefList.append(href)
179: myList = ",".join(hrefList)
180: logging.debug("getALLGisPlaces :%s"%(myList))
181: return myList
182:
183:
184: def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None, highlightQuery=None,sn=None, viewMode=None, tocMode=None, tocPN=None, characterNormalization=""):
185: """returns single page from fulltext"""
186: docpath = docinfo['textURLPath']
187: path = docinfo['textURLPath']
188: url = docinfo['url']
189: name = docinfo['name']
190: viewMode= pageinfo['viewMode']
191: tocMode = pageinfo['tocMode']
192: characterNormalization=pageinfo['characterNormalization']
193: tocPN = pageinfo['tocPN']
194: selfurl = self.absolute_url()
195: if mode == "text_dict":
196: textmode = "textPollux"
197: else:
198: textmode = mode
199: #logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
200: textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization)
201: if highlightQuery is not None:
202: textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn)
203:
204: pagexml = self.getServerData("page-fragment.xql",textParam)
205: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/page-fragment.xql", textParam, outputUnicode=False)
206:
207: pagedom = Parse(pagexml)
208: # plain text mode
209: if mode == "text":
210: # first div contains text
211: pagedivs = pagedom.xpath("/div")
212: if len(pagedivs) > 0:
213: pagenode = pagedivs[0]
214: links = pagenode.xpath("//a")
215: for l in links:
216: hrefNode = l.getAttributeNodeNS(None, u"href")
217: if hrefNode:
218: href= hrefNode.nodeValue
219: if href.startswith('#note-'):
220: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
221: return serializeNode(pagenode)
222: if mode == "xml":
223: # first div contains text
224: pagedivs = pagedom.xpath("/div")
225: if len(pagedivs) > 0:
226: pagenode = pagedivs[0]
227: return serializeNode(pagenode)
228: if mode == "gis":
229: # first div contains text
230: pagedivs = pagedom.xpath("/div")
231: if len(pagedivs) > 0:
232: pagenode = pagedivs[0]
233: links =pagenode.xpath("//a")
234: for l in links:
235: hrefNode =l.getAttributeNodeNS(None, u"href")
236: if hrefNode:
237: href=hrefNode.nodeValue
238: if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'):
239: hrefNode.nodeValue =href.replace('chinagis_REST/REST/db/chgis/mpdl','chinagis/REST/db/mpdl/%s'%name)
240: l.setAttributeNS(None, 'target', '_blank')
241: return serializeNode(pagenode)
242:
243: if mode == "pureXml":
244: # first div contains text
245: pagedivs = pagedom.xpath("/div")
246: if len(pagedivs) > 0:
247: pagenode = pagedivs[0]
248: return serializeNode(pagenode)
249: # text-with-links mode
250: if mode == "text_dict":
251: # first div contains text
252: pagedivs = pagedom.xpath("/div")
253: if len(pagedivs) > 0:
254: pagenode = pagedivs[0]
255: # check all a-tags
256: links = pagenode.xpath("//a")
257: for l in links:
258: hrefNode = l.getAttributeNodeNS(None, u"href")
259: if hrefNode:
260: # is link with href
261: href = hrefNode.nodeValue
262: if href.startswith('lt/lex.xql'):
263: # is pollux link
264: selfurl = self.absolute_url()
265: # change href
266: hrefNode.nodeValue = href.replace('lt/lex.xql','%s/template/head_main_voc'%selfurl)
267: # add target
268: l.setAttributeNS(None, 'target', '_blank')
269: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
270: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
271: if href.startswith('lt/lemma.xql'):
272: selfurl = self.absolute_url()
273: hrefNode.nodeValue = href.replace('lt/lemma.xql','%s/template/head_main_lemma'%selfurl)
274: l.setAttributeNS(None, 'target', '_blank')
275: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
276: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
277: if href.startswith('#note-'):
278: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
279: return serializeNode(pagenode)
280: return "no text here"
281:
282: def getTranslate(self, query=None, language=None):
283: """translate into another languages"""
284: data = self.getServerData("lt/lex.xql","document=&language="+str(language)+"&query="+urllib.quote(query))
285: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lex.xql","document=&language="+str(language)+"&query="+url_quote(str(query)))
286: return data
287:
288: def getLemma(self, lemma=None, language=None):
289: """simular words lemma """
290: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(lemma))
291: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(lemma)))
292: return data
293:
294: def getLemmaNew(self, query=None, language=None):
295: """simular words lemma """
296: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(query))
297: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(query)))
298: return data
299:
300: def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1):
301: """number of"""
302: docpath = docinfo['textURLPath']
303: pagesize = pageinfo['queryPageSize']
304: pn = pageinfo['searchPN']
305: query =pageinfo['query']
306: queryType =pageinfo['queryType']
307: tocSearch = 0
308: tocDiv = None
309:
310: 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))
311: #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)
312: pagedom = Parse(pagexml)
313: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
314: tocSearch = int(getTextFromNode(numdivs[0]))
315: tc=int((tocSearch/10)+1)
316: logging.debug("documentViewer (gettoc) tc: %s"%(tc))
317: return tc
318:
319: def getToc(self, mode="text", docinfo=None):
320: """loads table of contents and stores in docinfo"""
321: logging.debug("documentViewer (gettoc) mode: %s"%(mode))
322: if mode == "none":
323: return docinfo
324: if 'tocSize_%s'%mode in docinfo:
325: # cached toc
326: return docinfo
327:
328: docpath = docinfo['textURLPath']
329: # we need to set a result set size
330: pagesize = 1000
331: pn = 1
332: if mode == "text":
333: queryType = "toc"
334: else:
335: queryType = mode
336: # number of entries in toc
337: tocSize = 0
338: tocDiv = None
339:
340: pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn))
341: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql", "document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType,pagesize,pn), outputUnicode=False)
342: # post-processing downloaded xml
343: pagedom = Parse(pagexml)
344: # get number of entries
345: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
346: if len(numdivs) > 0:
347: tocSize = int(getTextFromNode(numdivs[0]))
348: docinfo['tocSize_%s'%mode] = tocSize
349: return docinfo
350:
351: def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None):
352: """returns single page from the table of contents"""
353: # TODO: this should use the cached TOC
354: if mode == "text":
355: queryType = "toc"
356: else:
357: queryType = mode
358: docpath = docinfo['textURLPath']
359: path = docinfo['textURLPath']
360: pagesize = pageinfo['tocPageSize']
361: pn = pageinfo['tocPN']
362: url = docinfo['url']
363: selfurl = self.absolute_url()
364: viewMode= pageinfo['viewMode']
365: characterNormalization = pageinfo ['characterNormalization']
366: tocMode = pageinfo['tocMode']
367: tocPN = pageinfo['tocPN']
368:
369: data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=%s"%(docpath,queryType, pagesize, pn,characterNormalization))
370: page = data.replace('page-fragment.xql?document=%s'%str(path),'%s?url=%s&viewMode=%s&tocMode=%s&tocPN=%s'%(selfurl,url, viewMode, tocMode, tocPN))
371: text = page.replace('mode=image','mode=texttool')
372: logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
373: #logging.debug("documentViewer (characterNormalization) text: %s"%(text))
374: return text
375:
376: def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
377: """change settings"""
378: self.title=title
379: self.timeout = timeout
380: self.serverUrl = serverUrl
381: if RESPONSE is not None:
382: RESPONSE.redirect('manage_main')
383:
384: # management methods
385: def manage_addMpdlXmlTextServerForm(self):
386: """Form for adding"""
387: pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self)
388: return pt()
389:
390: def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
391: """add zogiimage"""
392: newObj = MpdlXmlTextServer(id,title,serverUrl,timeout)
393: self.Destination()._setObject(id, newObj)
394: if RESPONSE is not None:
395: RESPONSE.redirect('manage_main')
396:
397:
398:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>