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: # logging.debug("documentViewer (text) text: %s"%(text))
139: #docinfo['numPages']=text('resultPage')
140: # dom = Parse(text)
141:
142: # result= dom.xpath("//result/resultPage")
143: # docinfo['numPages']=int(getTextFromNode(result[0]))
144: #logging.debug("documentViewer (characterNormalization) docinfo['numPages']: %s"%(docinfo['numPages']))
145: return docinfo
146:
147: def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None, highlightQuery=None,sn=None, viewMode=None, tocMode=None, tocPN=None, characterNormalization=""):
148: """returns single page from fulltext"""
149: docpath = docinfo['textURLPath']
150: path = docinfo['textURLPath']
151: url = docinfo['url']
152: viewMode= pageinfo['viewMode']
153: tocMode = pageinfo['tocMode']
154: characterNormalization=pageinfo['characterNormalization']
155: tocPN = pageinfo['tocPN']
156: selfurl = self.absolute_url()
157: if mode == "text_dict":
158: textmode = "textPollux"
159: else:
160: textmode = mode
161: #logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
162: textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization)
163: if highlightQuery is not None:
164: textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn)
165:
166: pagexml = self.getServerData("page-fragment.xql",textParam)
167: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/page-fragment.xql", textParam, outputUnicode=False)
168:
169: pagedom = Parse(pagexml)
170: # plain text mode
171: if mode == "text":
172: # first div contains text
173: pagedivs = pagedom.xpath("/div")
174: if len(pagedivs) > 0:
175: pagenode = pagedivs[0]
176: links = pagenode.xpath("//a")
177: for l in links:
178: hrefNode = l.getAttributeNodeNS(None, u"href")
179: if hrefNode:
180: href= hrefNode.nodeValue
181: if href.startswith('#note-'):
182: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
183: return serializeNode(pagenode)
184: if mode == "xml":
185: # first div contains text
186: pagedivs = pagedom.xpath("/div")
187: if len(pagedivs) > 0:
188: pagenode = pagedivs[0]
189: return serializeNode(pagenode)
190: if mode == "gis":
191: # first div contains text
192: pagedivs = pagedom.xpath("/div")
193: if len(pagedivs) > 0:
194: pagenode = pagedivs[0]
195: links =pagenode.xpath("//a")
196: for l in links:
197: hrefNode =l.getAttributeNodeNS(None, u"href")
198: if hrefNode:
199: href=hrefNode.nodeValue
200: if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'):
201: hrefNode.nodeValue =href.replace('chinagis_REST','chinagis')
202: l.setAttributeNS(None, 'target', '_blank')
203: return serializeNode(pagenode)
204:
205: if mode == "pureXml":
206: # first div contains text
207: pagedivs = pagedom.xpath("/div")
208: if len(pagedivs) > 0:
209: pagenode = pagedivs[0]
210: return serializeNode(pagenode)
211: # text-with-links mode
212: if mode == "text_dict":
213: # first div contains text
214: pagedivs = pagedom.xpath("/div")
215: if len(pagedivs) > 0:
216: pagenode = pagedivs[0]
217: # check all a-tags
218: links = pagenode.xpath("//a")
219: for l in links:
220: hrefNode = l.getAttributeNodeNS(None, u"href")
221: if hrefNode:
222: # is link with href
223: href = hrefNode.nodeValue
224: if href.startswith('lt/lex.xql'):
225: # is pollux link
226: selfurl = self.absolute_url()
227: # change href
228: hrefNode.nodeValue = href.replace('lt/lex.xql','%s/template/head_main_voc'%selfurl)
229: # add target
230: l.setAttributeNS(None, 'target', '_blank')
231: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
232: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
233: if href.startswith('lt/lemma.xql'):
234: selfurl = self.absolute_url()
235: hrefNode.nodeValue = href.replace('lt/lemma.xql','%s/template/head_main_lemma'%selfurl)
236: l.setAttributeNS(None, 'target', '_blank')
237: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
238: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
239: if href.startswith('#note-'):
240: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
241: return serializeNode(pagenode)
242: return "no text here"
243:
244: def getTranslate(self, query=None, language=None):
245: """translate into another languages"""
246: data = self.getServerData("lt/lex.xql","document=&language="+str(language)+"&query="+urllib.quote(query))
247: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lex.xql","document=&language="+str(language)+"&query="+url_quote(str(query)))
248: return data
249:
250: def getLemma(self, lemma=None, language=None):
251: """simular words lemma """
252: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(lemma))
253: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(lemma)))
254: return data
255:
256: def getLemmaNew(self, query=None, language=None):
257: """simular words lemma """
258: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(query))
259: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(query)))
260: return data
261:
262: def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1):
263: """number of"""
264: docpath = docinfo['textURLPath']
265: pagesize = pageinfo['queryPageSize']
266: pn = pageinfo['searchPN']
267: query =pageinfo['query']
268: queryType =pageinfo['queryType']
269: tocSearch = 0
270: tocDiv = None
271:
272: 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))
273: #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)
274: pagedom = Parse(pagexml)
275: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
276: tocSearch = int(getTextFromNode(numdivs[0]))
277: tc=int((tocSearch/10)+1)
278: logging.debug("documentViewer (gettoc) tc: %s"%(tc))
279: return tc
280:
281: def getToc(self, mode="text", docinfo=None):
282: """loads table of contents and stores in docinfo"""
283: logging.debug("documentViewer (gettoc) mode: %s"%(mode))
284: if mode == "none":
285: return docinfo
286: if 'tocSize_%s'%mode in docinfo:
287: # cached toc
288: return docinfo
289:
290: docpath = docinfo['textURLPath']
291: # we need to set a result set size
292: pagesize = 1000
293: pn = 1
294: if mode == "text":
295: queryType = "toc"
296: else:
297: queryType = mode
298: # number of entries in toc
299: tocSize = 0
300: tocDiv = None
301:
302: pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn))
303: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql", "document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType,pagesize,pn), outputUnicode=False)
304: # post-processing downloaded xml
305: pagedom = Parse(pagexml)
306: # get number of entries
307: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
308: if len(numdivs) > 0:
309: tocSize = int(getTextFromNode(numdivs[0]))
310: docinfo['tocSize_%s'%mode] = tocSize
311: return docinfo
312:
313: def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None):
314: """returns single page from the table of contents"""
315: # TODO: this should use the cached TOC
316: if mode == "text":
317: queryType = "toc"
318: else:
319: queryType = mode
320: docpath = docinfo['textURLPath']
321: path = docinfo['textURLPath']
322: pagesize = pageinfo['tocPageSize']
323: pn = pageinfo['tocPN']
324: url = docinfo['url']
325: selfurl = self.absolute_url()
326: viewMode= pageinfo['viewMode']
327: characterNormalization = pageinfo ['characterNormalization']
328: tocMode = pageinfo['tocMode']
329: tocPN = pageinfo['tocPN']
330:
331: data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=%s"%(docpath,queryType, pagesize, pn,characterNormalization))
332: page = data.replace('page-fragment.xql?document=%s'%str(path),'%s?url=%s&viewMode=%s&tocMode=%s&tocPN=%s'%(selfurl,url, viewMode, tocMode, tocPN))
333: text = page.replace('mode=image','mode=texttool')
334: logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
335: #logging.debug("documentViewer (characterNormalization) text: %s"%(text))
336: return text
337:
338: def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
339: """change settings"""
340: self.title=title
341: self.timeout = timeout
342: self.serverUrl = serverUrl
343: if RESPONSE is not None:
344: RESPONSE.redirect('manage_main')
345:
346: # management methods
347: def manage_addMpdlXmlTextServerForm(self):
348: """Form for adding"""
349: pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self)
350: return pt()
351:
352: def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
353: """add zogiimage"""
354: newObj = MpdlXmlTextServer(id,title,serverUrl,timeout)
355: self.Destination()._setObject(id, newObj)
356: if RESPONSE is not None:
357: RESPONSE.redirect('manage_main')
358:
359:
360:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>