Annotation of documentViewer/MpdlXmlTextServer.py, revision 1.49
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:
1.41 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,urllib.quote(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
1.43 abukhman 125: return docinfo
1.2 casties 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
1.43 abukhman 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)'
1.45 abukhman 137: text=self.getServerData("xpath.xql", "document=%s&xpath=%s"%(docinfo['textURLPath'], xpath))
1.44 abukhman 138: logging.debug("documentViewer (text) text: %s"%(text))
1.48 abukhman 139: #docinfo['numPages']=text('resultPage')
1.49 ! abukhman 140: pagedom = Parse(text)
! 141: docinfo['numPages']= pagedom.xpath("//resultPage")
1.48 abukhman 142: #logging.debug("documentViewer (characterNormalization) docinfo['numPages']: %s"%(docinfo['numPages']))
1.43 abukhman 143: return docinfo
1.2 casties 144:
1.25 abukhman 145: 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 146: """returns single page from fulltext"""
147: docpath = docinfo['textURLPath']
148: path = docinfo['textURLPath']
149: url = docinfo['url']
150: viewMode= pageinfo['viewMode']
151: tocMode = pageinfo['tocMode']
1.20 abukhman 152: characterNormalization=pageinfo['characterNormalization']
1.2 casties 153: tocPN = pageinfo['tocPN']
154: selfurl = self.absolute_url()
155: if mode == "text_dict":
156: textmode = "textPollux"
157: else:
158: textmode = mode
1.19 abukhman 159: #logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
1.26 abukhman 160: textParam = "document=%s&mode=%s&pn=%s&characterNormalization=%s"%(docpath,textmode,pn,characterNormalization)
1.2 casties 161: if highlightQuery is not None:
1.40 abukhman 162: textParam +="&highlightQuery=%s&sn=%s"%(urllib.quote(highlightQuery),sn)
1.2 casties 163:
1.38 abukhman 164: pagexml = self.getServerData("page-fragment.xql",textParam)
1.2 casties 165: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/page-fragment.xql", textParam, outputUnicode=False)
166:
1.39 abukhman 167: pagedom = Parse(pagexml)
1.2 casties 168: # plain text mode
169: if mode == "text":
170: # first div contains text
171: pagedivs = pagedom.xpath("/div")
172: if len(pagedivs) > 0:
173: pagenode = pagedivs[0]
174: links = pagenode.xpath("//a")
175: for l in links:
176: hrefNode = l.getAttributeNodeNS(None, u"href")
177: if hrefNode:
178: href= hrefNode.nodeValue
179: if href.startswith('#note-'):
1.27 abukhman 180: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
1.2 casties 181: return serializeNode(pagenode)
182: if mode == "xml":
183: # first div contains text
184: pagedivs = pagedom.xpath("/div")
185: if len(pagedivs) > 0:
186: pagenode = pagedivs[0]
187: return serializeNode(pagenode)
1.7 abukhman 188: if mode == "gis":
189: # first div contains text
190: pagedivs = pagedom.xpath("/div")
191: if len(pagedivs) > 0:
192: pagenode = pagedivs[0]
1.28 abukhman 193: links =pagenode.xpath("//a")
194: for l in links:
195: hrefNode =l.getAttributeNodeNS(None, u"href")
196: if hrefNode:
197: href=hrefNode.nodeValue
198: if href.startswith('http://chinagis.mpiwg-berlin.mpg.de'):
199: hrefNode.nodeValue =href.replace('chinagis_REST','chinagis')
1.42 abukhman 200: l.setAttributeNS(None, 'target', '_blank')
1.7 abukhman 201: return serializeNode(pagenode)
202:
1.2 casties 203: if mode == "pureXml":
204: # first div contains text
205: pagedivs = pagedom.xpath("/div")
206: if len(pagedivs) > 0:
207: pagenode = pagedivs[0]
208: return serializeNode(pagenode)
209: # text-with-links mode
210: if mode == "text_dict":
211: # first div contains text
212: pagedivs = pagedom.xpath("/div")
213: if len(pagedivs) > 0:
214: pagenode = pagedivs[0]
215: # check all a-tags
216: links = pagenode.xpath("//a")
217: for l in links:
218: hrefNode = l.getAttributeNodeNS(None, u"href")
219: if hrefNode:
220: # is link with href
221: href = hrefNode.nodeValue
222: if href.startswith('lt/lex.xql'):
223: # is pollux link
224: selfurl = self.absolute_url()
225: # change href
226: hrefNode.nodeValue = href.replace('lt/lex.xql','%s/template/head_main_voc'%selfurl)
227: # add target
228: l.setAttributeNS(None, 'target', '_blank')
229: 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 230: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 231: if href.startswith('lt/lemma.xql'):
232: selfurl = self.absolute_url()
233: hrefNode.nodeValue = href.replace('lt/lemma.xql','%s/template/head_main_lemma'%selfurl)
234: l.setAttributeNS(None, 'target', '_blank')
235: 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 236: l.setAttributeNS(None, 'onClick', 'popupWin.focus();')
1.2 casties 237: if href.startswith('#note-'):
1.19 abukhman 238: hrefNode.nodeValue = href.replace('#note-',"?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&pn=%s#note-"%(url,viewMode,tocMode,tocPN,pn))
1.2 casties 239: return serializeNode(pagenode)
240: return "no text here"
241:
242: def getTranslate(self, query=None, language=None):
243: """translate into another languages"""
1.5 casties 244: data = self.getServerData("lt/lex.xql","document=&language="+str(language)+"&query="+urllib.quote(query))
1.2 casties 245: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lex.xql","document=&language="+str(language)+"&query="+url_quote(str(query)))
246: return data
247:
248: def getLemma(self, lemma=None, language=None):
249: """simular words lemma """
1.5 casties 250: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(lemma))
1.2 casties 251: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(lemma)))
252: return data
253:
254: def getLemmaNew(self, query=None, language=None):
255: """simular words lemma """
1.5 casties 256: data = self.getServerData("lt/lemma.xql","document=&language="+str(language)+"&lemma="+urllib.quote(query))
1.2 casties 257: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/lt/lemma.xql","document=&language="+str(language)+"&lemma="+url_quote(str(query)))
258: return data
1.28 abukhman 259:
1.2 casties 260: def getQuery (self, docinfo=None, pageinfo=None, query=None, queryType=None, pn=1):
261: """number of"""
262: docpath = docinfo['textURLPath']
263: pagesize = pageinfo['queryPageSize']
264: pn = pageinfo['searchPN']
1.34 abukhman 265: query =pageinfo['query']
1.2 casties 266: queryType =pageinfo['queryType']
267: tocSearch = 0
268: tocDiv = None
269:
1.32 abukhman 270: 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 271: #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)
272: pagedom = Parse(pagexml)
273: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
274: tocSearch = int(getTextFromNode(numdivs[0]))
275: tc=int((tocSearch/10)+1)
1.23 abukhman 276: logging.debug("documentViewer (gettoc) tc: %s"%(tc))
1.2 casties 277: return tc
278:
279: def getToc(self, mode="text", docinfo=None):
280: """loads table of contents and stores in docinfo"""
1.23 abukhman 281: logging.debug("documentViewer (gettoc) mode: %s"%(mode))
1.2 casties 282: if mode == "none":
283: return docinfo
284: if 'tocSize_%s'%mode in docinfo:
285: # cached toc
286: return docinfo
287:
288: docpath = docinfo['textURLPath']
289: # we need to set a result set size
290: pagesize = 1000
291: pn = 1
292: if mode == "text":
293: queryType = "toc"
294: else:
295: queryType = mode
296: # number of entries in toc
297: tocSize = 0
298: tocDiv = None
299:
300: pagexml = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType, pagesize, pn))
301: #pagexml=self.template.fulltextclient.eval("/mpdl/interface/doc-query.xql", "document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s"%(docpath,queryType,pagesize,pn), outputUnicode=False)
302: # post-processing downloaded xml
303: pagedom = Parse(pagexml)
304: # get number of entries
305: numdivs = pagedom.xpath("//div[@class='queryResultHits']")
306: if len(numdivs) > 0:
307: tocSize = int(getTextFromNode(numdivs[0]))
308: docinfo['tocSize_%s'%mode] = tocSize
309: return docinfo
310:
311: def getTocPage(self, mode="text", pn=1, pageinfo=None, docinfo=None):
312: """returns single page from the table of contents"""
313: # TODO: this should use the cached TOC
314: if mode == "text":
315: queryType = "toc"
316: else:
317: queryType = mode
318: docpath = docinfo['textURLPath']
319: path = docinfo['textURLPath']
320: pagesize = pageinfo['tocPageSize']
321: pn = pageinfo['tocPN']
322: url = docinfo['url']
323: selfurl = self.absolute_url()
324: viewMode= pageinfo['viewMode']
1.26 abukhman 325: characterNormalization = pageinfo ['characterNormalization']
1.2 casties 326: tocMode = pageinfo['tocMode']
327: tocPN = pageinfo['tocPN']
328:
1.23 abukhman 329: data = self.getServerData("doc-query.xql","document=%s&queryType=%s&queryResultPageSize=%s&queryResultPN=%s&characterNormalization=%s"%(docpath,queryType, pagesize, pn,characterNormalization))
330: 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 331: text = page.replace('mode=image','mode=texttool')
1.21 abukhman 332: logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization))
1.19 abukhman 333: #logging.debug("documentViewer (characterNormalization) text: %s"%(text))
1.2 casties 334: return text
335:
336: def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
337: """change settings"""
338: self.title=title
339: self.timeout = timeout
340: self.serverUrl = serverUrl
341: if RESPONSE is not None:
342: RESPONSE.redirect('manage_main')
343:
344: # management methods
345: def manage_addMpdlXmlTextServerForm(self):
346: """Form for adding"""
347: pt = PageTemplateFile("zpt/manage_addMpdlXmlTextServer", globals()).__of__(self)
348: return pt()
349:
350: def manage_addMpdlXmlTextServer(self,id,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None):
351: """add zogiimage"""
352: newObj = MpdlXmlTextServer(id,title,serverUrl,timeout)
353: self.Destination()._setObject(id, newObj)
354: if RESPONSE is not None:
355: RESPONSE.redirect('manage_main')
356:
357:
1.4 casties 358:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>