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,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:
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=""):
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&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization)
149: if highlightQuery is not None:
150: textParam1 =urllib.quote(textParam)
151: textParam1 +="&highlightQuery=%s&sn=%s"%(highlightQuery,sn)
152:
153: pagexml = self.getServerData("page-fragment.xql",textParam1)
154: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/page-fragment.xql", textParam, outputUnicode=False)
155:
156: pagedom = Parse(pagexml)
157: # plain text mode
158: if mode == "text":
159: # first div contains text
160: pagedivs = pagedom.xpath("/div")
161: if len(pagedivs) > 0:
162: pagenode = pagedivs[0]
163: links = pagenode.xpath("//a")
164: for l in links:
165: hrefNode = l.getAttributeNodeNS(None, u"href")
166: if hrefNode:
167: href= hrefNode.nodeValue
168: if href.startswith('#note-'):
169: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
170: return serializeNode(pagenode)
171: if mode == "xml":
172: # first div contains text
173: pagedivs = pagedom.xpath("/div")
174: if len(pagedivs) > 0:
175: pagenode = pagedivs[0]
176: return serializeNode(pagenode)
177: if mode == "gis":
178: # first div contains text
179: pagedivs = pagedom.xpath("/div")
180: if len(pagedivs) > 0:
181: pagenode = pagedivs[0]
182: links =pagenode.xpath("//a")
183: for l in links:
184: hrefNode =l.getAttributeNodeNS(None, u"href")
185: if hrefNode:
186: href=hrefNode.nodeValue
187: if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'):
188: hrefNode.nodeValue =href.replace('chinagis_REST','chinagis')
189: return serializeNode(pagenode)
190:
191: if mode == "pureXml":
192: # first div contains text
193: pagedivs = pagedom.xpath("/div")
194: if len(pagedivs) > 0:
195: pagenode = pagedivs[0]
196: return serializeNode(pagenode)
197: # text-with-links mode
198: if mode == "text_dict":
199: # first div contains text
200: pagedivs = pagedom.xpath("/div")
201: if len(pagedivs) > 0:
202: pagenode = pagedivs[0]
203: # check all a-tags
204: links = pagenode.xpath("//a")
205: for l in links:
206: hrefNode = l.getAttributeNodeNS(None, u"href")
207: if hrefNode:
208: # is link with href
209: href = hrefNode.nodeValue
210: if href.startswith('lt/lex.xql'):
211: # is pollux link
212: selfurl = self.absolute_url()
213: # change href
214: hrefNode.nodeValue = href.replace('lt/lex.xql','%s/template/head_main_voc'%selfurl)
215: # add target
216: l.setAttributeNS(None, 'target', '_blank')
217: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
218: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
219: if href.startswith('lt/lemma.xql'):
220: selfurl = self.absolute_url()
221: hrefNode.nodeValue = href.replace('lt/lemma.xql','%s/template/head_main_lemma'%selfurl)
222: l.setAttributeNS(None, 'target', '_blank')
223: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
224: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
225: if href.startswith('#note-'):
226: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
227: return serializeNode(pagenode)
228: return "no text here"
229:
230: def getTranslate(self, query=None, language=None):
231: """translate into another languages"""
232: data = self.getServerData("lt/lex.xql","document=&language="+str(language)+"&query="+urllib.quote(query))
233: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lex.xql","document=&language="+str(language)+"&query="+url_quote(str(query)))
234: return data
235:
236: def getLemma(self, lemma=None, language=None):
237: """simular words lemma """
238: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(lemma))
239: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(lemma)))
240: return data
241:
242: def getLemmaNew(self, query=None, language=None):
243: """simular words lemma """
244: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(query))
245: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(query)))
246: return data
247:
248: def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1):
249: """number of"""
250: docpath = docinfo['textURLPath']
251: pagesize = pageinfo['queryPageSize']
252: pn = pageinfo['searchPN']
253: query =pageinfo['query']
254: queryType =pageinfo['queryType']
255: tocSearch = 0
256: tocDiv = None
257:
258: 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))
259: #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)
260: pagedom = Parse(pagexml)
261: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
262: tocSearch = int(getTextFromNode(numdivs[0]))
263: tc=int((tocSearch/10)+1)
264: logging.debug("documentViewer (gettoc) tc: %s"%(tc))
265: return tc
266:
267: def getToc(self, mode="text", docinfo=None):
268: """loads table of contents and stores in docinfo"""
269: logging.debug("documentViewer (gettoc) mode: %s"%(mode))
270: if mode == "none":
271: return docinfo
272: if 'tocSize_%s'%mode in docinfo:
273: # cached toc
274: return docinfo
275:
276: docpath = docinfo['textURLPath']
277: # we need to set a result set size
278: pagesize = 1000
279: pn = 1
280: if mode == "text":
281: queryType = "toc"
282: else:
283: queryType = mode
284: # number of entries in toc
285: tocSize = 0
286: tocDiv = None
287:
288: pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn))
289: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql", "document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType,pagesize,pn), outputUnicode=False)
290: # post-processing downloaded xml
291: pagedom = Parse(pagexml)
292: # get number of entries
293: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
294: if len(numdivs) > 0:
295: tocSize = int(getTextFromNode(numdivs[0]))
296: docinfo['tocSize_%s'%mode] = tocSize
297: return docinfo
298:
299: def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None):
300: """returns single page from the table of contents"""
301: # TODO: this should use the cached TOC
302: if mode == "text":
303: queryType = "toc"
304: else:
305: queryType = mode
306: docpath = docinfo['textURLPath']
307: path = docinfo['textURLPath']
308: pagesize = pageinfo['tocPageSize']
309: pn = pageinfo['tocPN']
310: url = docinfo['url']
311: selfurl = self.absolute_url()
312: viewMode= pageinfo['viewMode']
313: characterNormalization = pageinfo ['characterNormalization']
314: tocMode = pageinfo['tocMode']
315: tocPN = pageinfo['tocPN']
316:
317: data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=%s"%(docpath,queryType, pagesize, pn,characterNormalization))
318: page = data.replace('page-fragment.xql?document=%s'%str(path),'%s?url=%s&viewMode=%s&tocMode=%s&tocPN=%s'%(selfurl,url, viewMode, tocMode, tocPN))
319: text = page.replace('mode=image','mode=texttool')
320: logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
321: #logging.debug("documentViewer (characterNormalization) text: %s"%(text))
322: return text
323:
324: def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
325: """change settings"""
326: self.title=title
327: self.timeout = timeout
328: self.serverUrl = serverUrl
329: if RESPONSE is not None:
330: RESPONSE.redirect('manage_main')
331:
332: # management methods
333: def manage_addMpdlXmlTextServerForm(self):
334: """Form for adding"""
335: pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self)
336: return pt()
337:
338: def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
339: """add zogiimage"""
340: newObj = MpdlXmlTextServer(id,title,serverUrl,timeout)
341: self.Destination()._setObject(id, newObj)
342: if RESPONSE is not None:
343: RESPONSE.redirect('manage_main')
344:
345:
346:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>