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, query, pagesize, pn, sn, viewMode,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,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,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:
127: xquery = '//pb'
128: text = self.getServerData("xquery.xql","document=%s&xquery=%s"%(docinfo['textURLPath'],xquery))
129: #text = self.template.fulltextclient.eval("/mpdl/interface/xquery.xql", "document=%s&xquery=%s"%(docinfo['textURLPath'],xquery))
130: docinfo['numPages'] = text.count("<pb ")
131: return docinfo
132:
133: def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None, highlightQuery=None,sn=None, viewMode=None, tocMode=None, tocPN=None, characterNormalization=None):
134: """returns single page from fulltext"""
135: docpath = docinfo['textURLPath']
136: path = docinfo['textURLPath']
137: url = docinfo['url']
138: viewMode= pageinfo['viewMode']
139: tocMode = pageinfo['tocMode']
140: characterNormalization=pageinfo['characterNormalization']
141: tocPN = pageinfo['tocPN']
142: selfurl = self.absolute_url()
143: if mode == "text_dict":
144: textmode = "textPollux"
145: else:
146: textmode = mode
147: #logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
148: textParam = "document=%s&mode=%s&pn=%s"%(docpath,textmode,pn)
149: if highlightQuery is not None:
150: textParam +="&highlightQuery=%s&sn=%s"%(highlightQuery,sn)
151:
152: pagexml = self.getServerData("page-fragment.xql",textParam)
153: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/page-fragment.xql", textParam, outputUnicode=False)
154:
155: pagedom = Parse(pagexml)
156: # plain text mode
157: if mode == "text":
158: # first div contains text
159: pagedivs = pagedom.xpath("/div")
160: if len(pagedivs) > 0:
161: pagenode = pagedivs[0]
162: links = pagenode.xpath("//a")
163: for l in links:
164: hrefNode = l.getAttributeNodeNS(None, u"href")
165: if hrefNode:
166: href= hrefNode.nodeValue
167: if href.startswith('#note-'):
168: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s&characterNormalization=%s#note-"%(url,viewMode,tocMode,tocPN,pn,characterNormalization))
169: return serializeNode(pagenode)
170: if mode == "xml":
171: # first div contains text
172: pagedivs = pagedom.xpath("/div")
173: if len(pagedivs) > 0:
174: pagenode = pagedivs[0]
175: return serializeNode(pagenode)
176: if mode == "gis":
177: # first div contains text
178: pagedivs = pagedom.xpath("/div")
179: if len(pagedivs) > 0:
180: pagenode = pagedivs[0]
181: return serializeNode(pagenode)
182:
183: if mode == "pureXml":
184: # first div contains text
185: pagedivs = pagedom.xpath("/div")
186: if len(pagedivs) > 0:
187: pagenode = pagedivs[0]
188: return serializeNode(pagenode)
189: # text-with-links mode
190: if mode == "text_dict":
191: # first div contains text
192: pagedivs = pagedom.xpath("/div")
193: if len(pagedivs) > 0:
194: pagenode = pagedivs[0]
195: # check all a-tags
196: links = pagenode.xpath("//a")
197: for l in links:
198: hrefNode = l.getAttributeNodeNS(None, u"href")
199: if hrefNode:
200: # is link with href
201: href = hrefNode.nodeValue
202: if href.startswith('lt/lex.xql'):
203: # is pollux link
204: selfurl = self.absolute_url()
205: # change href
206: hrefNode.nodeValue = href.replace('lt/lex.xql','%s/template/head_main_voc'%selfurl)
207: # add target
208: l.setAttributeNS(None, 'target', '_blank')
209: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
210: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
211: if href.startswith('lt/lemma.xql'):
212: selfurl = self.absolute_url()
213: hrefNode.nodeValue = href.replace('lt/lemma.xql','%s/template/head_main_lemma'%selfurl)
214: l.setAttributeNS(None, 'target', '_blank')
215: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
216: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
217: if href.startswith('#note-'):
218: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
219: return serializeNode(pagenode)
220: return "no text here"
221:
222: def getTranslate(self, query=None, language=None):
223: """translate into another languages"""
224: data = self.getServerData("lt/lex.xql","document=&language="+str(language)+"&query="+urllib.quote(query))
225: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lex.xql","document=&language="+str(language)+"&query="+url_quote(str(query)))
226: return data
227:
228: def getLemma(self, lemma=None, language=None):
229: """simular words lemma """
230: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(lemma))
231: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(lemma)))
232: return data
233:
234: def getLemmaNew(self, query=None, language=None):
235: """simular words lemma """
236: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(query))
237: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(query)))
238: return data
239:
240: def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1):
241: """number of"""
242: docpath = docinfo['textURLPath']
243: pagesize = pageinfo['queryPageSize']
244: pn = pageinfo['searchPN']
245: query =pageinfo['query']
246: queryType =pageinfo['queryType']
247: tocSearch = 0
248: tocDiv = None
249:
250: pagexml = self.getServerData("doc-query.xql","document=%s&mode=%s&queryType=%s&query=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath, 'text', queryType, query, pagesize, pn))
251: #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)
252: pagedom = Parse(pagexml)
253: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
254: tocSearch = int(getTextFromNode(numdivs[0]))
255: tc=int((tocSearch/10)+1)
256: logging.debug("documentViewer (gettoc) tc: %s"%(tc))
257: return tc
258:
259: def getToc(self, mode="text", docinfo=None):
260: """loads table of contents and stores in docinfo"""
261: logging.debug("documentViewer (gettoc) mode: %s"%(mode))
262: if mode == "none":
263: return docinfo
264: if 'tocSize_%s'%mode in docinfo:
265: # cached toc
266: return docinfo
267:
268: docpath = docinfo['textURLPath']
269: # we need to set a result set size
270: pagesize = 1000
271: pn = 1
272: if mode == "text":
273: queryType = "toc"
274: else:
275: queryType = mode
276: # number of entries in toc
277: tocSize = 0
278: tocDiv = None
279:
280: pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn))
281: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql", "document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType,pagesize,pn), outputUnicode=False)
282: # post-processing downloaded xml
283: pagedom = Parse(pagexml)
284: # get number of entries
285: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
286: if len(numdivs) > 0:
287: tocSize = int(getTextFromNode(numdivs[0]))
288: docinfo['tocSize_%s'%mode] = tocSize
289: return docinfo
290:
291: def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None):
292: """returns single page from the table of contents"""
293: # TODO: this should use the cached TOC
294: if mode == "text":
295: queryType = "toc"
296: else:
297: queryType = mode
298: docpath = docinfo['textURLPath']
299: path = docinfo['textURLPath']
300: pagesize = pageinfo['tocPageSize']
301: pn = pageinfo['tocPN']
302: url = docinfo['url']
303: selfurl = self.absolute_url()
304: viewMode= pageinfo['viewMode']
305: #characterNormalization =pageinfo ['characterNormalization']
306: tocMode = pageinfo['tocMode']
307: tocPN = pageinfo['tocPN']
308:
309: data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn))
310: page = data.replace('page-fragment.xql?document=%s'%str(path),'%s?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&characterNormalization=%s'%(selfurl,url, viewMode, tocMode, tocPN,characterNormalization))
311: text = page.replace('mode=image','mode=texttool')
312: #logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
313: #logging.debug("documentViewer (characterNormalization) text: %s"%(text))
314: return text
315:
316: def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
317: """change settings"""
318: self.title=title
319: self.timeout = timeout
320: self.serverUrl = serverUrl
321: if RESPONSE is not None:
322: RESPONSE.redirect('manage_main')
323:
324: # management methods
325: def manage_addMpdlXmlTextServerForm(self):
326: """Form for adding"""
327: pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self)
328: return pt()
329:
330: def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
331: """add zogiimage"""
332: newObj = MpdlXmlTextServer(id,title,serverUrl,timeout)
333: self.Destination()._setObject(id, newObj)
334: if RESPONSE is not None:
335: RESPONSE.redirect('manage_main')
336:
337:
338:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>