|
|
| version 1.227, 2011/02/18 11:21:46 | version 1.228, 2011/02/24 13:55:28 |
|---|---|
| Line 183 class MpdlXmlTextServer(SimpleItem): | Line 183 class MpdlXmlTextServer(SimpleItem): |
| url = docinfo.get('url',None) | url = docinfo.get('url',None) |
| name = docinfo.get('name',None) | name = docinfo.get('name',None) |
| pn =pageinfo['current'] | pn =pageinfo['current'] |
| #viewMode= pageinfo['viewMode'] | |
| sn = pageinfo['sn'] | sn = pageinfo['sn'] |
| highlightQuery = pageinfo['highlightQuery'] | highlightQuery = pageinfo['highlightQuery'] |
| #mode = pageinfo ['viewMode'] | #mode = pageinfo ['viewMode'] |
| Line 209 class MpdlXmlTextServer(SimpleItem): | Line 207 class MpdlXmlTextServer(SimpleItem): |
| if pagedivs == dom.xpath("//div[@class='pageNumberOrig']"): | if pagedivs == dom.xpath("//div[@class='pageNumberOrig']"): |
| if len(pagedivs)>0: | if len(pagedivs)>0: |
| docinfo['pageNumberOrig']= getTextFromNode(pagedivs[0]) | docinfo['pageNumberOrig']= getTextFromNode(pagedivs[0]) |
| logging.debug("ORIGINAL PAGES: %s"%(docinfo['pageNumberOrig'])) | #logging.debug("ORIGINAL PAGES: %s"%(docinfo['pageNumberOrig'])) |
| #original Pages Norm | #original Pages Norm |
| pagedivs = dom.xpath("//div[@class='pageNumberOrigNorm']") | pagedivs = dom.xpath("//div[@class='pageNumberOrigNorm']") |
| if pagedivs == dom.xpath("//div[@class='pageNumberOrigNorm']"): | if pagedivs == dom.xpath("//div[@class='pageNumberOrigNorm']"): |
| if len(pagedivs)>0: | if len(pagedivs)>0: |
| docinfo['pageNumberOrigNorm']= getTextFromNode(pagedivs[0]) | docinfo['pageNumberOrigNorm']= getTextFromNode(pagedivs[0]) |
| logging.debug("ORIGINAL PAGES: %s"%(docinfo['pageNumberOrigNorm'])) | #logging.debug("ORIGINAL PAGES: %s"%(docinfo['pageNumberOrigNorm'])) |
| #figureEntries | #figureEntries |
| Line 228 class MpdlXmlTextServer(SimpleItem): | Line 226 class MpdlXmlTextServer(SimpleItem): |
| if s=='0': | if s=='0': |
| try: | try: |
| docinfo['countFigureEntries'] = int(s) | docinfo['countFigureEntries'] = int(s) |
| logging.debug("FIGURE ENTRIES: %s"%(s)) | #logging.debug("FIGURE ENTRIES: %s"%(s)) |
| except: | except: |
| docinfo['countFigureEntries'] = 0 | docinfo['countFigureEntries'] = 0 |
| else: | else: |
| s1 = int(s)/30+1 | s1 = int(s)/30+1 |
| try: | try: |
| docinfo['countFigureEntries'] = int(s1) | docinfo['countFigureEntries'] = int(s1) |
| logging.debug("FIGURE ENTRIES: %s"%(s1)) | #logging.debug("FIGURE ENTRIES: %s"%(s1)) |
| except: | except: |
| docinfo['countFigureEntries'] = 0 | docinfo['countFigureEntries'] = 0 |
| Line 247 class MpdlXmlTextServer(SimpleItem): | Line 245 class MpdlXmlTextServer(SimpleItem): |
| s = getTextFromNode(pagedivs[0]) | s = getTextFromNode(pagedivs[0]) |
| try: | try: |
| docinfo['countPlaces'] = int(s) | docinfo['countPlaces'] = int(s) |
| logging.debug("PLACES HERE: %s"%(s)) | #logging.debug("PLACES HERE: %s"%(s)) |
| except: | except: |
| docinfo['countPlaces'] = 0 | docinfo['countPlaces'] = 0 |
| Line 260 class MpdlXmlTextServer(SimpleItem): | Line 258 class MpdlXmlTextServer(SimpleItem): |
| if s=='0': | if s=='0': |
| try: | try: |
| docinfo['countTocEntries'] = int(s) | docinfo['countTocEntries'] = int(s) |
| logging.debug("TEXT ENTRIES: %s"%(s)) | #logging.debug("TEXT ENTRIES: %s"%(s)) |
| except: | except: |
| docinfo['countTocEntries'] = 0 | docinfo['countTocEntries'] = 0 |
| else: | else: |
| s1 = int(s)/30+1 | s1 = int(s)/30+1 |
| try: | try: |
| docinfo['countTocEntries'] = int(s1) | docinfo['countTocEntries'] = int(s1) |
| logging.debug("TEXT ENTRIES: %s"%(s1)) | #logging.debug("TEXT ENTRIES: %s"%(s1)) |
| except: | except: |
| docinfo['countTocEntries'] = 0 | docinfo['countTocEntries'] = 0 |
| Line 280 class MpdlXmlTextServer(SimpleItem): | Line 278 class MpdlXmlTextServer(SimpleItem): |
| try: | try: |
| docinfo['numPages'] = int(s) | docinfo['numPages'] = int(s) |
| logging.debug("PAGE NUMBER: %s"%(s)) | #logging.debug("PAGE NUMBER: %s"%(s)) |
| np = docinfo['numPages'] | np = docinfo['numPages'] |
| pageinfo['end'] = min(pageinfo['end'], np) | pageinfo['end'] = min(pageinfo['end'], np) |
| Line 408 class MpdlXmlTextServer(SimpleItem): | Line 406 class MpdlXmlTextServer(SimpleItem): |
| pagedom = Parse(pagexml) | pagedom = Parse(pagexml) |
| numdivs = pagedom.xpath("//div[@class='queryResultHits']") | numdivs = pagedom.xpath("//div[@class='queryResultHits']") |
| tocSearch = int(getTextFromNode(numdivs[0])) | tocSearch = int(getTextFromNode(numdivs[0])) |
| logging.debug("documentViewer (gettoc) tocSearch: %s"%(tocSearch)) | #logging.debug("documentViewer (gettoc) tocSearch: %s"%(tocSearch)) |
| tc=int((tocSearch/10)+1) | tc=int((tocSearch/10)+1) |
| logging.debug("documentViewer (gettoc) tc: %s"%(tc)) | #logging.debug("documentViewer (gettoc) tc: %s"%(tc)) |
| return tc | return tc |
| def getToc(self, mode="text", docinfo=None): | def getToc(self, mode="text", docinfo=None): |
| """loads table of contents and stores in docinfo""" | """loads table of contents and stores in docinfo""" |
| logging.debug("documentViewer (gettoc) mode: %s"%(mode)) | #logging.debug("documentViewer (gettoc) mode: %s"%(mode)) |
| if mode == "none": | if mode == "none": |
| return docinfo | return docinfo |
| if 'tocSize_%s'%mode in docinfo: | if 'tocSize_%s'%mode in docinfo: |
| Line 468 class MpdlXmlTextServer(SimpleItem): | Line 466 class MpdlXmlTextServer(SimpleItem): |
| page = data.replace('page-fragment.xql?document=%s'%str(path),'%s?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&optionToggle=1'%(selfurl,url, viewMode, tocMode, tocPN)) | page = data.replace('page-fragment.xql?document=%s'%str(path),'%s?url=%s&viewMode=%s&tocMode=%s&tocPN=%s&optionToggle=1'%(selfurl,url, viewMode, tocMode, tocPN)) |
| text = page.replace('mode=image','mode=texttool') | text = page.replace('mode=image','mode=texttool') |
| #logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization)) | #logging.debug("documentViewer (characterNormalization) characterNormalization: %s"%(characterNormalization)) |
| logging.debug("TEXT %s"%(text)) | #logging.debug("TEXT %s"%(text)) |
| return text | return text |
| def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None): | def manage_changeMpdlXmlTextServer(self,title="",serverUrl="http://mpdl-proto.mpiwg-berlin.mpg.de/mpdl/interface/",timeout=40,RESPONSE=None): |