Annotation of documentViewer/MpdlXmlTextServer.py, revision 1.36
1.2 casties 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
1.5 casties 8: import urllib
1.2 casties 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:
1.3 casties 22: def __init__(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/", serverName=None, timeout=40):
1.2 casties 23: """constructor"""
24: self.id=id
25: self.title=title
26: self.timeout = timeout
1.3 casties 27: if serverName is None:
28: self.serverUrl = serverUrl
29: else:
30: self.serverUrl = "http://%s/mpdl/interface/"%serverName
1.2 casties 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']
1.23 abukhman 45: logging.debug("documentViewer (gettoc) docpath: %s"%(docpath))
46: logging.debug("documentViewer (gettoc) url: %s"%(url))
1.2 casties 47: pagesize = pageinfo['queryPageSize']
48: pn = pageinfo['searchPN']
49: sn = pageinfo['sn']
50: highlightQuery = pageinfo['highlightQuery']
1.34 abukhman 51: query =pageinfo['query']
1.2 casties 52: queryType =pageinfo['queryType']
53: viewMode= pageinfo['viewMode']
54: tocMode = pageinfo['tocMode']
1.24 abukhman 55: characterNormalization = pageinfo['characterNormalization']
1.2 casties 56: tocPN = pageinfo['tocPN']
57: selfurl = self.absolute_url()
58:
1.30 abukhman 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))
1.2 casties 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()
1.30 abukhman 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))
1.2 casties 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()
1.30 abukhman 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))
1.2 casties 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;")
1.6 abukhman 95: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 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
1.18 abukhman 107: hrefNode.nodeValue=href.replace('mode=text','mode=texttool&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s'%(viewMode,tocMode,tocPN,pn))
1.2 casties 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;")
1.6 abukhman 112: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 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;")
1.6 abukhman 117: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 118: return serializeNode(pagenode)
119: return "no text here"
120:
1.4 casties 121: def getNumPages(self, docinfo):
1.2 casties 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:
1.25 abukhman 133: def getTextPage(self, mode="text", pn=1, docinfo=None, pageinfo=None, highlightQuery=None,sn=None, viewMode=None, tocMode=None, tocPN=None, characterNormalization=""):
1.2 casties 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']
1.20 abukhman 140: characterNormalization=pageinfo['characterNormalization']
1.2 casties 141: tocPN = pageinfo['tocPN']
142: selfurl = self.absolute_url()
143: if mode == "text_dict":
144: textmode = "textPollux"
145: else:
146: textmode = mode
1.19 abukhman 147: #logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
1.26 abukhman 148: textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization)
1.2 casties 149: if highlightQuery is not None:
1.36 ! abukhman 150: textParam +="&highlightQuery=%s&sn=%s"%(highlightQuery,sn)
1.2 casties 151:
1.36 ! abukhman 152: pagexml = self.getServerData("page-fragment.xql",textParam)
1.2 casties 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-'):
1.27 abukhman 168: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
1.2 casties 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)
1.7 abukhman 176: if mode == "gis":
177: # first div contains text
178: pagedivs = pagedom.xpath("/div")
179: if len(pagedivs) > 0:
180: pagenode = pagedivs[0]
1.28 abukhman 181: links =pagenode.xpath("//a")
182: for l in links:
183: hrefNode =l.getAttributeNodeNS(None, u"href")
184: if hrefNode:
185: href=hrefNode.nodeValue
186: if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'):
187: hrefNode.nodeValue =href.replace('chinagis_REST','chinagis')
1.7 abukhman 188: return serializeNode(pagenode)
189:
1.2 casties 190: if mode == "pureXml":
191: # first div contains text
192: pagedivs = pagedom.xpath("/div")
193: if len(pagedivs) > 0:
194: pagenode = pagedivs[0]
195: return serializeNode(pagenode)
196: # text-with-links mode
197: if mode == "text_dict":
198: # first div contains text
199: pagedivs = pagedom.xpath("/div")
200: if len(pagedivs) > 0:
201: pagenode = pagedivs[0]
202: # check all a-tags
203: links = pagenode.xpath("//a")
204: for l in links:
205: hrefNode = l.getAttributeNodeNS(None, u"href")
206: if hrefNode:
207: # is link with href
208: href = hrefNode.nodeValue
209: if href.startswith('lt/lex.xql'):
210: # is pollux link
211: selfurl = self.absolute_url()
212: # change href
213: hrefNode.nodeValue = href.replace('lt/lex.xql','%s/template/head_main_voc'%selfurl)
214: # add target
215: l.setAttributeNS(None, 'target', '_blank')
216: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
1.6 abukhman 217: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 218: if href.startswith('lt/lemma.xql'):
219: selfurl = self.absolute_url()
220: hrefNode.nodeValue = href.replace('lt/lemma.xql','%s/template/head_main_lemma'%selfurl)
221: l.setAttributeNS(None, 'target', '_blank')
222: l.setAttributeNS(None, 'onClick',"popupWin = window.open(this.href, 'contacts', 'location,width=500,height=600,top=180, left=700, scrollbars=1'); return false;")
1.6 abukhman 223: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 224: if href.startswith('#note-'):
1.19 abukhman 225: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
1.2 casties 226: return serializeNode(pagenode)
227: return "no text here"
228:
229: def getTranslate(self, query=None, language=None):
230: """translate into another languages"""
1.5 casties 231: data = self.getServerData("lt/lex.xql","document=&language="+str(language)+"&query="+urllib.quote(query))
1.2 casties 232: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lex.xql","document=&language="+str(language)+"&query="+url_quote(str(query)))
233: return data
234:
235: def getLemma(self, lemma=None, language=None):
236: """simular words lemma """
1.5 casties 237: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(lemma))
1.2 casties 238: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(lemma)))
239: return data
240:
241: def getLemmaNew(self, query=None, language=None):
242: """simular words lemma """
1.5 casties 243: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(query))
1.2 casties 244: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(query)))
245: return data
1.28 abukhman 246:
1.2 casties 247: def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1):
248: """number of"""
249: docpath = docinfo['textURLPath']
250: pagesize = pageinfo['queryPageSize']
251: pn = pageinfo['searchPN']
1.34 abukhman 252: query =pageinfo['query']
1.2 casties 253: queryType =pageinfo['queryType']
254: tocSearch = 0
255: tocDiv = None
256:
1.32 abukhman 257: 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))
1.2 casties 258: #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)
259: pagedom = Parse(pagexml)
260: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
261: tocSearch = int(getTextFromNode(numdivs[0]))
262: tc=int((tocSearch/10)+1)
1.23 abukhman 263: logging.debug("documentViewer (gettoc) tc: %s"%(tc))
1.2 casties 264: return tc
265:
266: def getToc(self, mode="text", docinfo=None):
267: """loads table of contents and stores in docinfo"""
1.23 abukhman 268: logging.debug("documentViewer (gettoc) mode: %s"%(mode))
1.2 casties 269: if mode == "none":
270: return docinfo
271: if 'tocSize_%s'%mode in docinfo:
272: # cached toc
273: return docinfo
274:
275: docpath = docinfo['textURLPath']
276: # we need to set a result set size
277: pagesize = 1000
278: pn = 1
279: if mode == "text":
280: queryType = "toc"
281: else:
282: queryType = mode
283: # number of entries in toc
284: tocSize = 0
285: tocDiv = None
286:
287: pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn))
288: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql", "document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType,pagesize,pn), outputUnicode=False)
289: # post-processing downloaded xml
290: pagedom = Parse(pagexml)
291: # get number of entries
292: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
293: if len(numdivs) > 0:
294: tocSize = int(getTextFromNode(numdivs[0]))
295: docinfo['tocSize_%s'%mode] = tocSize
296: return docinfo
297:
298: def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None):
299: """returns single page from the table of contents"""
300: # TODO: this should use the cached TOC
301: if mode == "text":
302: queryType = "toc"
303: else:
304: queryType = mode
305: docpath = docinfo['textURLPath']
306: path = docinfo['textURLPath']
307: pagesize = pageinfo['tocPageSize']
308: pn = pageinfo['tocPN']
309: url = docinfo['url']
310: selfurl = self.absolute_url()
311: viewMode= pageinfo['viewMode']
1.26 abukhman 312: characterNormalization = pageinfo ['characterNormalization']
1.2 casties 313: tocMode = pageinfo['tocMode']
314: tocPN = pageinfo['tocPN']
315:
1.23 abukhman 316: data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=%s"%(docpath,queryType, pagesize, pn,characterNormalization))
317: page = data.replace('page-fragment.xql?document=%s'%str(path),'%s?url=%s&viewMode=%s&tocMode=%s&tocPN=%s'%(selfurl,url, viewMode, tocMode, tocPN))
1.2 casties 318: text = page.replace('mode=image','mode=texttool')
1.21 abukhman 319: logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
1.19 abukhman 320: #logging.debug("documentViewer (characterNormalization) text: %s"%(text))
1.2 casties 321: return text
322:
323: def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
324: """change settings"""
325: self.title=title
326: self.timeout = timeout
327: self.serverUrl = serverUrl
328: if RESPONSE is not None:
329: RESPONSE.redirect('manage_main')
330:
331: # management methods
332: def manage_addMpdlXmlTextServerForm(self):
333: """Form for adding"""
334: pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self)
335: return pt()
336:
337: def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
338: """add zogiimage"""
339: newObj = MpdlXmlTextServer(id,title,serverUrl,timeout)
340: self.Destination()._setObject(id, newObj)
341: if RESPONSE is not None:
342: RESPONSE.redirect('manage_main')
343:
344:
1.4 casties 345:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>