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