--- documentViewer/documentViewer.py 2010/06/16 18:27:04 1.69.2.3 +++ documentViewer/documentViewer.py 2010/08/12 11:45:59 1.80 @@ -2,32 +2,20 @@ from OFS.Folder import Folder from Products.PageTemplates.ZopePageTemplate import ZopePageTemplate from Products.PageTemplates.PageTemplateFile import PageTemplateFile -from Products.PythonScripts.standard import url_quote from AccessControl import ClassSecurityInfo from AccessControl import getSecurityManager from Globals import package_home -from Ft.Xml.Domlette import NonvalidatingReader -from Ft.Xml.Domlette import PrettyPrint, Print from Ft.Xml import EMPTY_NAMESPACE, Parse - -from xml.dom.minidom import parse, parseString - -from extraFunction import * - - -import Ft.Xml.XPath -import cStringIO -import xmlrpclib +import Ft.Xml.Domlette import os.path import sys -import cgi import urllib +import urllib2 import logging import math - import urlparse -from types import * +import cStringIO def logger(txt,method,txt2): """logging""" @@ -55,7 +43,7 @@ def getTextFromNode(nodename): def serializeNode(node, encoding='utf-8'): """returns a string containing node as XML""" buf = cStringIO.StringIO() - Print(node, stream=buf, encoding=encoding) + Ft.Xml.Domlette.Print(node, stream=buf, encoding=encoding) s = buf.getvalue() buf.close() return s @@ -84,7 +72,7 @@ def getHttpData(url, data=None, num_trie if sys.version_info < (2, 6): # set timeout on socket -- ugly :-( import socket - socket.setdefaulttimeout(timeout) + socket.setdefaulttimeout(float(timeout)) response = urllib2.urlopen(url) else: response = urllib2.urlopen(url,timeout=float(timeout)) @@ -131,6 +119,7 @@ class documentViewer(Folder): page_main_images = PageTemplateFile('zpt/page_main_images', globals()) page_main_text = PageTemplateFile('zpt/page_main_text', globals()) page_main_text_dict = PageTemplateFile('zpt/page_main_text_dict', globals()) + page_main_gis =PageTemplateFile ('zpt/page_main_gis', globals()) page_main_xml = PageTemplateFile('zpt/page_main_xml', globals()) head_main = PageTemplateFile('zpt/head_main', globals()) docuviewer_css = PageTemplateFile('css/docuviewer.css', globals()) @@ -157,7 +146,7 @@ class documentViewer(Folder): self._setObject('template',templateFolder) # old style try: import MpdlXmlTextServer - textServer = MpdlXmlTextServer(id='fulltextclient') + textServer = MpdlXmlTextServer.MpdlXmlTextServer(id='fulltextclient',serverName=textServerName) #templateFolder['fulltextclient'] = xmlRpcClient templateFolder._setObject('fulltextclient',textServer) except Exception, e: @@ -184,9 +173,9 @@ class documentViewer(Folder): """get search""" return self.template.fulltextclient.getSearch(**args) - def getNumPages(self, **args): + def getNumPages(self, docinfo): """get numpages""" - return self.template.fulltextclient.getNumPages(**args) + return self.template.fulltextclient.getNumPages(docinfo) def getTranslate(self, **args): """get translate""" @@ -229,7 +218,7 @@ class documentViewer(Folder): pt = getattr(self.template, 'thumbs_main_rss') if viewMode=="auto": # automodus gewaehlt - if docinfo.get("textURL",'') and self.textViewerUrl: #texturl gesetzt und textViewer konfiguriert + if docinfo.has_key("textURL") or docinfo.has_key('textURLPath'): #texturl gesetzt und textViewer konfiguriert viewMode="text" else: viewMode="images" @@ -237,13 +226,14 @@ class documentViewer(Folder): return pt(docinfo=docinfo,pageinfo=pageinfo,viewMode=viewMode) security.declareProtected('View','index_html') - def index_html(self,url,mode="texttool",viewMode="auto",tocMode="thumbs",start=None,pn=1,mk=None, query=None, querySearch=None): + def index_html(self,url,mode="texttool",viewMode="auto",tocMode="thumbs",start=None,pn=1,mk=None, query=None, querySearch=None, characterNormalization=None): ''' view it @param mode: defines how to access the document behind url @param url: url which contains display information @param viewMode: if images display images, if text display text, default is auto (text,images or auto) @param tocMode: type of 'table of contents' for navigation (thumbs, text, figures, none) + @param characterNormalization type of text display (reg, norm, none) @param querySearch: type of different search modes (fulltext, fulltextMorph, xpath, xquery, ftIndex, ftIndexMorph, fulltextMorphLemma) ''' @@ -255,17 +245,16 @@ class documentViewer(Folder): return "ERROR: template folder missing!" if not getattr(self, 'digilibBaseUrl', None): - self.digilibBaseUrl = self.findDigilibUrl() or "http://nausikaa.mpiwg-berlin.mpg.de/digitallibrary" + self.digilibBaseUrl = self.findDigilibUrl() or "http://digilib.mpiwg-berlin.mpg.de/digitallibrary" docinfo = self.getDocinfo(mode=mode,url=url) - if tocMode != "thumbs": # get table of contents docinfo = self.getToc(mode=tocMode, docinfo=docinfo) if viewMode=="auto": # automodus gewaehlt - if docinfo.get("textURL",''): #texturl gesetzt und textViewer konfiguriert + if docinfo.has_key('textURL') or docinfo.has_key('textURLPath'): #texturl gesetzt und textViewer konfiguriert viewMode="text_dict" else: viewMode="images" @@ -279,8 +268,8 @@ class documentViewer(Folder): ret="" if mk is None: return "" - if type(mk) is not ListType: - mk=[mk] + if not isinstance(mk, list): + mk=[mk] for m in mk: ret+="mk=%s"%m return ret @@ -318,7 +307,8 @@ class documentViewer(Folder): params["url"] = getParentDir(params["url"]) # quote values and assemble into query string - ps = "&".join(["%s=%s"%(k,urllib.quote(v)) for (k, v) in params.items()]) + #ps = "&".join(["%s=%s"%(k,urllib.quote(v)) for (k, v) in params.items()]) + ps = urllib.urlencode(params) url=self.REQUEST['URL1']+"?"+ps return url @@ -359,13 +349,14 @@ class documentViewer(Folder): elif access is None or access in self.authgroups: # only local access -- only logged in users user = getSecurityManager().getUser() + logging.debug("documentViewer (accessOK) user=%s ip=%s"%(user,self.REQUEST.getClientAddr())) if user is not None: #print "user: ", user return (user.getUserName() != "Anonymous User") else: return False - logging.debug("documentViewer (accessOK) unknown access type %s"%access) + logging.error("documentViewer (accessOK) unknown access type %s"%access) return False @@ -580,7 +571,8 @@ class documentViewer(Folder): if viewerUrls and (len(viewerUrls) > 0): viewerUrl = getTextFromNode(viewerUrls[0]) docinfo['viewerURL'] = viewerUrl - + + # old style text URL textUrls = dom.xpath("//texttool/text") if textUrls and (len(textUrls) > 0): textUrl = getTextFromNode(textUrls[0]) @@ -592,13 +584,14 @@ class documentViewer(Folder): docinfo['textURL'] = textUrl + # new style text-url-path textUrls = dom.xpath("//texttool/text-url-path") if textUrls and (len(textUrls) > 0): textUrl = getTextFromNode(textUrls[0]) docinfo['textURLPath'] = textUrl if not docinfo['imagePath']: # text-only, no page images - docinfo = self.getNumPages(docinfo) #im moment einfach auf eins setzen, navigation ueber die thumbs geht natuerlich nicht + docinfo = self.getNumPages(docinfo) presentationUrls = dom.xpath("//texttool/presentation") docinfo = self.getBibinfoFromIndexMeta(url, docinfo=docinfo, dom=dom) # get info von bib tag @@ -708,6 +701,7 @@ class documentViewer(Folder): pageinfo['numgroups'] += 1 pageinfo['viewMode'] = viewMode pageinfo['tocMode'] = tocMode + pageinfo['characterNormalization'] = self.REQUEST.get ('characterNormalization', 'norm') pageinfo['query'] = self.REQUEST.get('query',' ') pageinfo['queryType'] = self.REQUEST.get('queryType',' ') pageinfo['querySearch'] =self.REQUEST.get('querySearch', 'fulltext')