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=%sxpath=%s"%(docinfo['textURLPath'], xpath))
138: logging.debug("documentViewer (text) text: %s"%(text))
139: docinfo['numPages']=text.xpath
140: #logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
141: return docinfo
142:
143: def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None, highlightQuery=None,sn=None, viewMode=None, tocMode=None, tocPN=None, characterNormalization=""):
144: """returns single page from fulltext"""
145: docpath = docinfo['textURLPath']
146: path = docinfo['textURLPath']
147: url = docinfo['url']
148: viewMode= pageinfo['viewMode']
149: tocMode = pageinfo['tocMode']
150: characterNormalization=pageinfo['characterNormalization']
151: tocPN = pageinfo['tocPN']
152: selfurl = self.absolute_url()
153: if mode == "text_dict":
154: textmode = "textPollux"
155: else:
156: textmode = mode
157: #logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
158: textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization)
159: if highlightQuery is not None:
160: textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn)
161:
162: pagexml = self.getServerData("page-fragment.xql",textParam)
163: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/page-fragment.xql", textParam, outputUnicode=False)
164:
165: pagedom = Parse(pagexml)
166: # plain text mode
167: if mode == "text":
168: # first div contains text
169: pagedivs = pagedom.xpath("/div")
170: if len(pagedivs) > 0:
171: pagenode = pagedivs[0]
172: links = pagenode.xpath("//a")
173: for l in links:
174: hrefNode = l.getAttributeNodeNS(None, u"href")
175: if hrefNode:
176: href= hrefNode.nodeValue
177: if href.startswith('#note-'):
178: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
179: return serializeNode(pagenode)
180: if mode == "xml":
181: # first div contains text
182: pagedivs = pagedom.xpath("/div")
183: if len(pagedivs) > 0:
184: pagenode = pagedivs[0]
185: return serializeNode(pagenode)
186: if mode == "gis":
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('http://chinagis.mpiwg-berlin.mpg.de'):
197: hrefNode.nodeValue =href.replace('chinagis_REST','chinagis')
198: l.setAttributeNS(None, 'target', '_blank')
199: return serializeNode(pagenode)
200:
201: if mode == "pureXml":
202: # first div contains text
203: pagedivs = pagedom.xpath("/div")
204: if len(pagedivs) > 0:
205: pagenode = pagedivs[0]
206: return serializeNode(pagenode)
207: # text-with-links mode
208: if mode == "text_dict":
209: # first div contains text
210: pagedivs = pagedom.xpath("/div")
211: if len(pagedivs) > 0:
212: pagenode = pagedivs[0]
213: # check all a-tags
214: links = pagenode.xpath("//a")
215: for l in links:
216: hrefNode = l.getAttributeNodeNS(None, u"href")
217: if hrefNode:
218: # is link with href
219: href = hrefNode.nodeValue
220: if href.startswith('lt/lex.xql'):
221: # is pollux link
222: selfurl = self.absolute_url()
223: # change href
224: hrefNode.nodeValue = href.replace('lt/lex.xql','%s/template/head_main_voc'%selfurl)
225: # add target
226: l.setAttributeNS(None, 'target', '_blank')
227: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
228: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
229: if href.startswith('lt/lemma.xql'):
230: selfurl = self.absolute_url()
231: hrefNode.nodeValue = href.replace('lt/lemma.xql','%s/template/head_main_lemma'%selfurl)
232: l.setAttributeNS(None, 'target', '_blank')
233: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
234: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
235: if href.startswith('#note-'):
236: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
237: return serializeNode(pagenode)
238: return "no text here"
239:
240: def getTranslate(self, query=None, language=None):
241: """translate into another languages"""
242: data = self.getServerData("lt/lex.xql","document=&language="+str(language)+"&query="+urllib.quote(query))
243: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lex.xql","document=&language="+str(language)+"&query="+url_quote(str(query)))
244: return data
245:
246: def getLemma(self, lemma=None, language=None):
247: """simular words lemma """
248: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(lemma))
249: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(lemma)))
250: return data
251:
252: def getLemmaNew(self, query=None, language=None):
253: """simular words lemma """
254: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(query))
255: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(query)))
256: return data
257:
258: def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1):
259: """number of"""
260: docpath = docinfo['textURLPath']
261: pagesize = pageinfo['queryPageSize']
262: pn = pageinfo['searchPN']
263: query =pageinfo['query']
264: queryType =pageinfo['queryType']
265: tocSearch = 0
266: tocDiv = None
267:
268: 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))
269: #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)
270: pagedom = Parse(pagexml)
271: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
272: tocSearch = int(getTextFromNode(numdivs[0]))
273: tc=int((tocSearch/10)+1)
274: logging.debug("documentViewer (gettoc) tc: %s"%(tc))
275: return tc
276:
277: def getToc(self, mode="text", docinfo=None):
278: """loads table of contents and stores in docinfo"""
279: logging.debug("documentViewer (gettoc) mode: %s"%(mode))
280: if mode == "none":
281: return docinfo
282: if 'tocSize_%s'%mode in docinfo:
283: # cached toc
284: return docinfo
285:
286: docpath = docinfo['textURLPath']
287: # we need to set a result set size
288: pagesize = 1000
289: pn = 1
290: if mode == "text":
291: queryType = "toc"
292: else:
293: queryType = mode
294: # number of entries in toc
295: tocSize = 0
296: tocDiv = None
297:
298: pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn))
299: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql", "document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType,pagesize,pn), outputUnicode=False)
300: # post-processing downloaded xml
301: pagedom = Parse(pagexml)
302: # get number of entries
303: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
304: if len(numdivs) > 0:
305: tocSize = int(getTextFromNode(numdivs[0]))
306: docinfo['tocSize_%s'%mode] = tocSize
307: return docinfo
308:
309: def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None):
310: """returns single page from the table of contents"""
311: # TODO: this should use the cached TOC
312: if mode == "text":
313: queryType = "toc"
314: else:
315: queryType = mode
316: docpath = docinfo['textURLPath']
317: path = docinfo['textURLPath']
318: pagesize = pageinfo['tocPageSize']
319: pn = pageinfo['tocPN']
320: url = docinfo['url']
321: selfurl = self.absolute_url()
322: viewMode= pageinfo['viewMode']
323: characterNormalization = pageinfo ['characterNormalization']
324: tocMode = pageinfo['tocMode']
325: tocPN = pageinfo['tocPN']
326:
327: data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=%s"%(docpath,queryType, pagesize, pn,characterNormalization))
328: page = data.replace('page-fragment.xql?document=%s'%str(path),'%s?url=%s&viewMode=%s&tocMode=%s&tocPN=%s'%(selfurl,url, viewMode, tocMode, tocPN))
329: text = page.replace('mode=image','mode=texttool')
330: logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
331: #logging.debug("documentViewer (characterNormalization) text: %s"%(text))
332: return text
333:
334: def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
335: """change settings"""
336: self.title=title
337: self.timeout = timeout
338: self.serverUrl = serverUrl
339: if RESPONSE is not None:
340: RESPONSE.redirect('manage_main')
341:
342: # management methods
343: def manage_addMpdlXmlTextServerForm(self):
344: """Form for adding"""
345: pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self)
346: return pt()
347:
348: def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
349: """add zogiimage"""
350: newObj = MpdlXmlTextServer(id,title,serverUrl,timeout)
351: self.Destination()._setObject(id, newObj)
352: if RESPONSE is not None:
353: RESPONSE.redirect('manage_main')
354:
355:
356:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>