1: from OFS.Folder import Folder
2: from Products.PageTemplates.ZopePageTemplate import ZopePageTemplate
3: from Products.PageTemplates.PageTemplateFile import PageTemplateFile
4: from App.ImageFile import ImageFile
5: from AccessControl import ClassSecurityInfo
6: from AccessControl import getSecurityManager
7: from Globals import package_home
8:
9: import xml.etree.ElementTree as ET
10:
11: import os
12: import sys
13: import urllib
14: import logging
15: import math
16: import urlparse
17: import re
18: import string
19:
20: from SrvTxtUtils import getInt, getText, getHttpData, refreshingImageFileIndexHtml
21:
22: def serializeNode(node, encoding="utf-8"):
23: """returns a string containing node as XML"""
24: s = ET.tostring(node)
25:
26: # 4Suite:
27: # stream = cStringIO.StringIO()
28: # Ft.Xml.Domlette.Print(node, stream=stream, encoding=encoding)
29: # s = stream.getvalue()
30: # stream.close()
31: return s
32:
33: def browserCheck(self):
34: """check the browsers request to find out the browser type"""
35: bt = {}
36: ua = self.REQUEST.get_header("HTTP_USER_AGENT")
37: bt['ua'] = ua
38: bt['isIE'] = False
39: bt['isN4'] = False
40: bt['versFirefox']=""
41: bt['versIE']=""
42: bt['versSafariChrome']=""
43: bt['versOpera']=""
44:
45: if string.find(ua, 'MSIE') > -1:
46: bt['isIE'] = True
47: else:
48: bt['isN4'] = (string.find(ua, 'Mozilla/4.') > -1)
49: # Safari oder Chrome identification
50: try:
51: nav = ua[string.find(ua, '('):]
52: nav1=ua[string.find(ua,')'):]
53: nav2=nav1[string.find(nav1,'('):]
54: nav3=nav2[string.find(nav2,')'):]
55: ie = string.split(nav, "; ")[1]
56: ie1 =string.split(nav1, " ")[2]
57: ie2 =string.split(nav3, " ")[1]
58: ie3 =string.split(nav3, " ")[2]
59: if string.find(ie3, "Safari") >-1:
60: bt['versSafariChrome']=string.split(ie2, "/")[1]
61: except: pass
62: # IE identification
63: try:
64: nav = ua[string.find(ua, '('):]
65: ie = string.split(nav, "; ")[1]
66: if string.find(ie, "MSIE") > -1:
67: bt['versIE'] = string.split(ie, " ")[1]
68: except:pass
69: # Firefox identification
70: try:
71: nav = ua[string.find(ua, '('):]
72: nav1=ua[string.find(ua,')'):]
73: if string.find(ie1, "Firefox") >-1:
74: nav5= string.split(ie1, "/")[1]
75: logging.debug("FIREFOX: %s"%(nav5))
76: bt['versFirefox']=nav5[0:3]
77: except:pass
78: #Opera identification
79: try:
80: if string.find(ua,"Opera") >-1:
81: nav = ua[string.find(ua, '('):]
82: nav1=nav[string.find(nav,')'):]
83: bt['versOpera']=string.split(nav1,"/")[2]
84: except:pass
85:
86: bt['isMac'] = string.find(ua, 'Macintosh') > -1
87: bt['isWin'] = string.find(ua, 'Windows') > -1
88: bt['isIEWin'] = bt['isIE'] and bt['isWin']
89: bt['isIEMac'] = bt['isIE'] and bt['isMac']
90: bt['staticHTML'] = False
91:
92: return bt
93:
94: def getParentPath(path, cnt=1):
95: """returns pathname shortened by cnt"""
96: # make sure path doesn't end with /
97: path = path.rstrip('/')
98: # split by /, shorten, and reassemble
99: return '/'.join(path.split('/')[0:-cnt])
100:
101: ##
102: ## documentViewer class
103: ##
104: class documentViewer(Folder):
105: """document viewer"""
106: meta_type="Document viewer"
107:
108: security=ClassSecurityInfo()
109: manage_options=Folder.manage_options+(
110: {'label':'main config','action':'changeDocumentViewerForm'},
111: )
112:
113: metadataService = None
114: """MetaDataFolder instance"""
115:
116: # templates and forms
117: viewer_text = PageTemplateFile('zpt/viewer_text', globals())
118: viewer_images = PageTemplateFile('zpt/viewer_images', globals())
119: viewer_main = PageTemplateFile('zpt/viewer_main', globals())
120: toc_thumbs = PageTemplateFile('zpt/toc_thumbs', globals())
121: toc_text = PageTemplateFile('zpt/toc_text', globals())
122: toc_figures = PageTemplateFile('zpt/toc_figures', globals())
123: page_main_images = PageTemplateFile('zpt/page_main_images', globals())
124: page_main_double = PageTemplateFile('zpt/page_main_double', globals())
125: page_main_text = PageTemplateFile('zpt/page_main_text', globals())
126: page_main_text_dict = PageTemplateFile('zpt/page_main_text_dict', globals())
127: page_main_gis =PageTemplateFile ('zpt/page_main_gis', globals())
128: page_main_xml = PageTemplateFile('zpt/page_main_xml', globals())
129: page_main_pureXml = PageTemplateFile('zpt/page_main_pureXml', globals())
130: head_main = PageTemplateFile('zpt/head_main', globals())
131: info_xml = PageTemplateFile('zpt/info_xml', globals())
132: docuviewer_css = ImageFile('css/docuviewer.css',globals())
133: # make ImageFile better for development
134: docuviewer_css.index_html = refreshingImageFileIndexHtml
135:
136:
137: thumbs_main_rss = PageTemplateFile('zpt/thumbs_main_rss', globals())
138:
139:
140: def __init__(self,id,imageScalerUrl=None,textServerName=None,title="",digilibBaseUrl=None,thumbcols=2,thumbrows=5,authgroups="mpiwg"):
141: """init document viewer"""
142: self.id=id
143: self.title=title
144: self.thumbcols = thumbcols
145: self.thumbrows = thumbrows
146: # authgroups is list of authorized groups (delimited by ,)
147: self.authgroups = [s.strip().lower() for s in authgroups.split(',')]
148: # create template folder so we can always use template.something
149:
150: templateFolder = Folder('template')
151: #self['template'] = templateFolder # Zope-2.12 style
152: self._setObject('template',templateFolder) # old style
153: try:
154: import MpdlXmlTextServer
155: textServer = MpdlXmlTextServer.MpdlXmlTextServer(id='fulltextclient',serverName=textServerName)
156: #templateFolder['fulltextclient'] = xmlRpcClient
157: templateFolder._setObject('fulltextclient',textServer)
158: except Exception, e:
159: logging.error("Unable to create MpdlXmlTextServer for fulltextclient: "+str(e))
160:
161: try:
162: from Products.zogiLib.zogiLib import zogiLib
163: zogilib = zogiLib(id="zogilib", title="zogilib for docuviewer", dlServerURL=imageScalerUrl, layout="book")
164: #templateFolder['zogilib'] = zogilib
165: templateFolder._setObject('zogilib',zogilib)
166: except Exception, e:
167: logging.error("Unable to create zogiLib for zogilib: "+str(e))
168:
169: try:
170: # assume MetaDataFolder instance is called metadata
171: self.metadataService = getattr(self, 'metadata')
172: except Exception, e:
173: logging.error("Unable to find MetaDataFolder 'metadata': "+str(e))
174:
175: if digilibBaseUrl is not None:
176: self.digilibBaseUrl = digilibBaseUrl
177:
178:
179: # proxy text server methods to fulltextclient
180: def getTextPage(self, **args):
181: """get page"""
182: return self.template.fulltextclient.getTextPage(**args)
183:
184: def getOrigPages(self, **args):
185: """get page"""
186: return self.template.fulltextclient.getOrigPages(**args)
187:
188: def getOrigPagesNorm(self, **args):
189: """get page"""
190: return self.template.fulltextclient.getOrigPagesNorm(**args)
191:
192: def getQuery(self, **args):
193: """get query in search"""
194: return self.template.fulltextclient.getQuery(**args)
195:
196: def getSearch(self, **args):
197: """get search"""
198: return self.template.fulltextclient.getSearch(**args)
199:
200: def getGisPlaces(self, **args):
201: """get gis places"""
202: return self.template.fulltextclient.getGisPlaces(**args)
203:
204: def getAllGisPlaces(self, **args):
205: """get all gis places """
206: return self.template.fulltextclient.getAllGisPlaces(**args)
207:
208: def getWordInfo(self, **args):
209: """get translate"""
210: return self.template.fulltextclient.getWordInfo(**args)
211:
212: def getLemma(self, **args):
213: """get lemma"""
214: return self.template.fulltextclient.getLemma(**args)
215:
216: def getLemmaQuery(self, **args):
217: """get query"""
218: return self.template.fulltextclient.getLemmaQuery(**args)
219:
220: def getLex(self, **args):
221: """get lex"""
222: return self.template.fulltextclient.getLex(**args)
223:
224: def getToc(self, **args):
225: """get toc"""
226: return self.template.fulltextclient.getToc(**args)
227:
228: def getTocPage(self, **args):
229: """get tocpage"""
230: return self.template.fulltextclient.getTocPage(**args)
231:
232:
233: security.declareProtected('View','thumbs_rss')
234: def thumbs_rss(self,mode,url,viewMode="auto",start=None,pn=1):
235: '''
236: view it
237: @param mode: defines how to access the document behind url
238: @param url: url which contains display information
239: @param viewMode: if images display images, if text display text, default is images (text,images or auto)
240:
241: '''
242: logging.debug("HHHHHHHHHHHHHH:load the rss")
243: logging.debug("documentViewer (index) mode: %s url:%s start:%s pn:%s"%(mode,url,start,pn))
244:
245: if not hasattr(self, 'template'):
246: # create template folder if it doesn't exist
247: self.manage_addFolder('template')
248:
249: if not self.digilibBaseUrl:
250: self.digilibBaseUrl = self.findDigilibUrl() or "http://nausikaa.mpiwg-berlin.mpg.de/digitallibrary"
251:
252: docinfo = self.getDocinfo(mode=mode,url=url)
253: #pageinfo = self.getPageinfo(start=start,current=pn,docinfo=docinfo)
254: pageinfo = self.getPageinfo(start=start,current=pn, docinfo=docinfo)
255: ''' ZDES '''
256: pt = getattr(self.template, 'thumbs_main_rss')
257:
258: if viewMode=="auto": # automodus gewaehlt
259: if docinfo.has_key("textURL") or docinfo.get('textURLPath',None): #texturl gesetzt und textViewer konfiguriert
260: viewMode="text"
261: else:
262: viewMode="images"
263:
264: return pt(docinfo=docinfo,pageinfo=pageinfo,viewMode=viewMode)
265:
266:
267: security.declareProtected('View','index_html')
268: def index_html(self,url,mode="texttool",viewMode="auto",viewType=None,tocMode="thumbs",start=1,pn=1):
269: """
270: view page
271: @param url: url which contains display information
272: @param mode: defines how to access the document behind url
273: @param viewMode: 'images': display images, 'text': display text, default is 'auto'
274: @param viewType: sub-type of viewMode, e.g. 'dict' for viewMode='text'
275: @param tocMode: type of 'table of contents' for navigation (thumbs, text, figures, none)
276: """
277:
278: logging.debug("documentViewer(index_html) mode=%s url=%s viewMode=%s viewType=%s start=%s pn=%s"%(mode,url,viewMode,viewType,start,pn))
279:
280: if not hasattr(self, 'template'):
281: # this won't work
282: logging.error("template folder missing!")
283: return "ERROR: template folder missing!"
284:
285: if not getattr(self, 'digilibBaseUrl', None):
286: self.digilibBaseUrl = self.findDigilibUrl() or "http://digilib.mpiwg-berlin.mpg.de/digitallibrary"
287:
288: docinfo = self.getDocinfo(mode=mode,url=url)
289:
290: if tocMode != "thumbs":
291: # get table of contents
292: docinfo = self.getToc(mode=tocMode, docinfo=docinfo)
293:
294: # auto viewMode: text if there is a text else images
295: if viewMode=="auto":
296: if docinfo.get('textURL', None) or docinfo.get('textURLPath', None):
297: viewMode = "text"
298: viewType = "dict"
299: else:
300: viewMode = "images"
301:
302: elif viewMode == "text_dict":
303: # legacy fix
304: viewMode = "text"
305: viewType = "dict"
306:
307: # stringify viewType
308: if isinstance(viewType, list):
309: logging.debug("index_html: viewType is list:%s"%viewType)
310: viewType = ','.join([t for t in viewType if t])
311:
312: pageinfo = self.getPageinfo(start=start, current=pn, docinfo=docinfo, viewMode=viewMode, viewType=viewType, tocMode=tocMode)
313:
314: # get template /template/viewer_$viewMode
315: pt = getattr(self.template, 'viewer_%s'%viewMode, None)
316: if pt is None:
317: logging.error("No template for viewMode=%s!"%viewMode)
318: # TODO: error page?
319: return "No template for viewMode=%s!"%viewMode
320:
321: # and execute with parameters
322: return pt(docinfo=docinfo, pageinfo=pageinfo)
323:
324: def generateMarks(self,mk):
325: ret=""
326: if mk is None:
327: return ""
328: if not isinstance(mk, list):
329: mk=[mk]
330: for m in mk:
331: ret+="mk=%s"%m
332: return ret
333:
334:
335: def getBrowser(self):
336: """getBrowser the version of browser """
337: bt = browserCheck(self)
338: logging.debug("BROWSER VERSION: %s"%(bt))
339: return bt
340:
341: def findDigilibUrl(self):
342: """try to get the digilib URL from zogilib"""
343: url = self.template.zogilib.getDLBaseUrl()
344: return url
345:
346: def getScalerUrl(self, fn=None, pn=None, dw=100, dh=100, docinfo=None):
347: """returns URL to digilib Scaler with params"""
348: url = None
349: if docinfo is not None:
350: url = docinfo.get('imageURL', None)
351:
352: if url is None:
353: url = "%s/servlet/Scaler?"%self.digilibBaseUrl
354: if fn is None and docinfo is not None:
355: fn = docinfo.get('imagePath','')
356:
357: url += "fn=%s"%fn
358:
359: if pn:
360: url += "&pn=%s"%pn
361:
362: url += "&dw=%s&dh=%s"%(dw,dh)
363: return url
364:
365: def getDocumentViewerURL(self):
366: """returns the URL of this instance"""
367: return self.absolute_url()
368:
369: def getStyle(self, idx, selected, style=""):
370: """returns a string with the given style and append 'sel' if idx == selected."""
371: #logger("documentViewer (getstyle)", logging.INFO, "idx: %s selected: %s style: %s"%(idx,selected,style))
372: if idx == selected:
373: return style + 'sel'
374: else:
375: return style
376:
377: def getParams(self, param=None, val=None, params=None, duplicates=None):
378: """returns dict with URL parameters.
379:
380: Takes URL parameters and additionally param=val or dict params.
381: Deletes key if value is None."""
382: # copy existing request params
383: newParams=self.REQUEST.form.copy()
384: # change single param
385: if param is not None:
386: if val is None:
387: if newParams.has_key(param):
388: del newParams[param]
389: else:
390: newParams[param] = str(val)
391:
392: # change more params
393: if params is not None:
394: for (k, v) in params.items():
395: if v is None:
396: # val=None removes param
397: if newParams.has_key(k):
398: del newParams[k]
399:
400: else:
401: newParams[k] = v
402:
403: if duplicates:
404: # eliminate lists (coming from duplicate keys)
405: for (k,v) in newParams.items():
406: if isinstance(v, list):
407: if duplicates == 'comma':
408: # make comma-separated list of non-empty entries
409: newParams[k] = ','.join([t for t in v if t])
410: elif duplicates == 'first':
411: # take first non-empty entry
412: newParams[k] = [t for t in v if t][0]
413:
414: return newParams
415:
416: def getLink(self, param=None, val=None, params=None, baseUrl=None, paramSep='&', duplicates='comma'):
417: """returns URL to documentviewer with parameter param set to val or from dict params"""
418: urlParams = self.getParams(param=param, val=val, params=params, duplicates=duplicates)
419: # quote values and assemble into query string (not escaping '/')
420: ps = paramSep.join(["%s=%s"%(k,urllib.quote_plus(unicode(v),'/')) for (k, v) in urlParams.items()])
421: if baseUrl is None:
422: baseUrl = self.getDocumentViewerURL()
423:
424: url = "%s?%s"%(baseUrl, ps)
425: return url
426:
427: def getLinkAmp(self, param=None, val=None, params=None, baseUrl=None, duplicates='comma'):
428: """link to documentviewer with parameter param set to val"""
429: return self.getLink(param=param, val=val, params=params, baseUrl=baseUrl, paramSep='&', duplicates=duplicates)
430:
431:
432: def getInfo_xml(self,url,mode):
433: """returns info about the document as XML"""
434: if not self.digilibBaseUrl:
435: self.digilibBaseUrl = self.findDigilibUrl() or "http://nausikaa.mpiwg-berlin.mpg.de/digitallibrary"
436:
437: docinfo = self.getDocinfo(mode=mode,url=url)
438: pt = getattr(self.template, 'info_xml')
439: return pt(docinfo=docinfo)
440:
441: def isAccessible(self, docinfo):
442: """returns if access to the resource is granted"""
443: access = docinfo.get('accessType', None)
444: logging.debug("documentViewer (accessOK) access type %s"%access)
445: if access == 'free':
446: logging.debug("documentViewer (accessOK) access is free")
447: return True
448:
449: elif access is None or access in self.authgroups:
450: # only local access -- only logged in users
451: user = getSecurityManager().getUser()
452: logging.debug("documentViewer (accessOK) user=%s ip=%s"%(user,self.REQUEST.getClientAddr()))
453: if user is not None:
454: #print "user: ", user
455: return (user.getUserName() != "Anonymous User")
456: else:
457: return False
458:
459: logging.error("documentViewer (accessOK) unknown access type %s"%access)
460: return False
461:
462:
463:
464: def getDocinfo(self, mode, url):
465: """returns docinfo depending on mode"""
466: logging.debug("getDocinfo: mode=%s, url=%s"%(mode,url))
467: # look for cached docinfo in session
468: if self.REQUEST.SESSION.has_key('docinfo'):
469: docinfo = self.REQUEST.SESSION['docinfo']
470: # check if its still current
471: if docinfo is not None and docinfo.get('mode', None) == mode and docinfo.get('url', None) == url:
472: logging.debug("getDocinfo: docinfo in session. keys=%s"%docinfo.keys())
473: return docinfo
474:
475: # new docinfo
476: docinfo = {'mode': mode, 'url': url}
477: # add self url
478: docinfo['viewerUrl'] = self.getDocumentViewerURL()
479: docinfo['digilibBaseUrl'] = self.digilibBaseUrl
480: # get index.meta DOM
481: docUrl = None
482: metaDom = None
483: if mode=="texttool":
484: # url points to document dir or index.meta
485: metaDom = self.metadataService.getDomFromPathOrUrl(url)
486: docUrl = url.replace('/index.meta', '')
487: if metaDom is None:
488: raise IOError("Unable to find index.meta for mode=texttool!")
489:
490: elif mode=="imagepath":
491: # url points to folder with images, index.meta optional
492: # asssume index.meta in parent dir
493: docUrl = getParentPath(url)
494: metaDom = self.metadataService.getDomFromPathOrUrl(docUrl)
495:
496: elif mode=="filepath":
497: # url points to image file, index.meta optional
498: # asssume index.meta is two path segments up
499: docUrl = getParentPath(url, 2)
500: metaDom = self.metadataService.getDomFromPathOrUrl(docUrl)
501:
502: else:
503: logging.error("documentViewer (getdocinfo) unknown mode: %s!"%mode)
504: raise ValueError("Unknown mode %s! Has to be one of 'texttool','imagepath','filepath'."%(mode))
505:
506: docinfo['documentUrl'] = docUrl
507: # process index.meta contents
508: if metaDom is not None and metaDom.tag == 'resource':
509: # document directory name and path
510: resource = self.metadataService.getResourceData(dom=metaDom)
511: if resource:
512: docinfo = self.getDocinfoFromResource(docinfo, resource)
513:
514: # texttool info
515: texttool = self.metadataService.getTexttoolData(dom=metaDom)
516: if texttool:
517: docinfo = self.getDocinfoFromTexttool(docinfo, texttool)
518:
519: # bib info
520: bib = self.metadataService.getBibData(dom=metaDom)
521: if bib:
522: docinfo = self.getDocinfoFromBib(docinfo, bib)
523: else:
524: # no bib - try info.xml
525: docinfo = self.getDocinfoFromPresentationInfoXml(docinfo)
526:
527: # auth info
528: access = self.metadataService.getAccessData(dom=metaDom)
529: if access:
530: docinfo = self.getDocinfoFromAccess(docinfo, access)
531:
532: # attribution info
533: attribution = self.metadataService.getAttributionData(dom=metaDom)
534: if attribution:
535: logging.debug("getDocinfo: attribution=%s"%repr(attribution))
536: docinfo['attribution'] = attribution
537: #docinfo = self.getDocinfoFromAccess(docinfo, access)
538:
539: # copyright info
540: copyright = self.metadataService.getCopyrightData(dom=metaDom)
541: if copyright:
542: logging.debug("getDocinfo: copyright=%s"%repr(copyright))
543: docinfo['copyright'] = copyright
544: #docinfo = self.getDocinfoFromAccess(docinfo, access)
545:
546: # image path
547: if mode != 'texttool':
548: # override image path from texttool with url
549: docinfo['imagePath'] = url.replace('/mpiwg/online/', '', 1)
550:
551: # number of images from digilib
552: if docinfo.get('imagePath', None):
553: docinfo['imageURL'] = self.digilibBaseUrl + "/servlet/Scaler?fn=" + docinfo['imagePath']
554: docinfo = self.getDocinfoFromDigilib(docinfo, docinfo['imagePath'])
555:
556: logging.debug("documentViewer (getdocinfo) docinfo: keys=%s"%docinfo.keys())
557: #logging.debug("documentViewer (getdocinfo) docinfo: %s"%docinfo)
558: # store in session
559: self.REQUEST.SESSION['docinfo'] = docinfo
560: return docinfo
561:
562: def getDocinfoFromResource(self, docinfo, resource):
563: """reads contents of resource element into docinfo"""
564: docName = resource.get('name', None)
565: docinfo['documentName'] = docName
566: docPath = resource.get('archive-path', None)
567: if docPath:
568: # clean up document path
569: if docPath[0] != '/':
570: docPath = '/' + docPath
571:
572: if docName and (not docPath.endswith(docName)):
573: docPath += "/" + docName
574:
575: else:
576: # use docUrl as docPath
577: docUrl = docinfo['documentURL']
578: if not docUrl.startswith('http:'):
579: docPath = docUrl
580: if docPath:
581: # fix URLs starting with /mpiwg/online
582: docPath = docPath.replace('/mpiwg/online', '', 1)
583:
584: docinfo['documentPath'] = docPath
585: return docinfo
586:
587: def getDocinfoFromTexttool(self, docinfo, texttool):
588: """reads contents of texttool element into docinfo"""
589: # image dir
590: imageDir = texttool.get('image', None)
591: docPath = docinfo.get('documentPath', None)
592: if imageDir and docPath:
593: #print "image: ", imageDir, " archivepath: ", archivePath
594: imageDir = os.path.join(docPath, imageDir)
595: imageDir = imageDir.replace('/mpiwg/online', '', 1)
596: docinfo['imagePath'] = imageDir
597:
598: # old style text URL
599: textUrl = texttool.get('text', None)
600: if textUrl and docPath:
601: if urlparse.urlparse(textUrl)[0] == "": #keine url
602: textUrl = os.path.join(docPath, textUrl)
603:
604: docinfo['textURL'] = textUrl
605:
606: # new style text-url-path
607: textUrl = texttool.get('text-url-path', None)
608: if textUrl:
609: docinfo['textURLPath'] = textUrl
610:
611: # page flow
612: docinfo['pageFlow'] = texttool.get('page-flow', 'ltr')
613:
614: # odd pages are left
615: docinfo['oddPage'] = texttool.get('odd-scan-position', 'left')
616:
617: # number of title page (0: not defined)
618: docinfo['titlePage'] = texttool.get('title-scan-no', 0)
619:
620: # old presentation stuff
621: presentation = texttool.get('presentation', None)
622: if presentation and docPath:
623: if presentation.startswith('http:'):
624: docinfo['presentationUrl'] = presentation
625: else:
626: docinfo['presentationUrl'] = os.path.join(docPath, presentation)
627:
628:
629: return docinfo
630:
631: def getDocinfoFromBib(self, docinfo, bib):
632: """reads contents of bib element into docinfo"""
633: logging.debug("getDocinfoFromBib bib=%s"%repr(bib))
634: # put all raw bib fields in dict "bib"
635: docinfo['bib'] = bib
636: bibtype = bib.get('@type', None)
637: docinfo['bibType'] = bibtype
638: # also store DC metadata for convenience
639: dc = self.metadataService.getDCMappedData(bib)
640: docinfo['creator'] = dc.get('creator',None)
641: docinfo['title'] = dc.get('title',None)
642: docinfo['date'] = dc.get('date',None)
643: return docinfo
644:
645: def getDocinfoFromAccess(self, docinfo, acc):
646: """reads contents of access element into docinfo"""
647: #TODO: also read resource type
648: logging.debug("getDocinfoFromAccess acc=%s"%repr(acc))
649: try:
650: acctype = acc['@attr']['type']
651: if acctype:
652: access=acctype
653: if access in ['group', 'institution']:
654: access = acc['name'].lower()
655:
656: docinfo['accessType'] = access
657:
658: except:
659: pass
660:
661: return docinfo
662:
663: def getDocinfoFromDigilib(self, docinfo, path):
664: infoUrl=self.digilibBaseUrl+"/dirInfo-xml.jsp?mo=dir&fn="+path
665: # fetch data
666: txt = getHttpData(infoUrl)
667: if not txt:
668: logging.error("Unable to get dir-info from %s"%(infoUrl))
669: return docinfo
670:
671: dom = ET.fromstring(txt)
672: size = getText(dom.find("size"))
673: logging.debug("getDocinfoFromDigilib: size=%s"%size)
674: if size:
675: docinfo['numPages'] = int(size)
676: else:
677: docinfo['numPages'] = 0
678:
679: # TODO: produce and keep list of image names and numbers
680: return docinfo
681:
682:
683: def getDocinfoFromPresentationInfoXml(self,docinfo):
684: """gets DC-like bibliographical information from the presentation entry in texttools"""
685: url = docinfo.get('presentationUrl', None)
686: if not url:
687: logging.error("getDocinfoFromPresentation: no URL!")
688: return docinfo
689:
690: dom = None
691: metaUrl = None
692: if url.startswith("http://"):
693: # real URL
694: metaUrl = url
695: else:
696: # online path
697:
698: server=self.digilibBaseUrl+"/servlet/Texter?fn="
699: metaUrl=server+url
700:
701: txt=getHttpData(metaUrl)
702: if txt is None:
703: logging.error("Unable to read info.xml from %s"%(url))
704: return docinfo
705:
706: dom = ET.fromstring(txt)
707: docinfo['creator']=getText(dom.find(".//author"))
708: docinfo['title']=getText(dom.find(".//title"))
709: docinfo['date']=getText(dom.find(".//date"))
710: return docinfo
711:
712:
713: def getPageinfo(self, current=None, start=None, rows=None, cols=None, docinfo=None, viewMode=None, viewType=None, tocMode=None):
714: """returns pageinfo with the given parameters"""
715: logging.debug("getPageInfo(current=%s, start=%s, rows=%s, cols=%s, viewMode=%s, viewType=%s, tocMode=%s)"%(current,start,rows,cols,viewMode,viewType,tocMode))
716: pageinfo = {}
717: pageinfo['viewMode'] = viewMode
718: pageinfo['viewType'] = viewType
719: pageinfo['tocMode'] = tocMode
720:
721: current = getInt(current)
722: pageinfo['current'] = current
723: pageinfo['pn'] = current
724: rows = int(rows or self.thumbrows)
725: pageinfo['rows'] = rows
726: cols = int(cols or self.thumbcols)
727: pageinfo['cols'] = cols
728: grpsize = cols * rows
729: pageinfo['groupsize'] = grpsize
730: # is start is empty use one around current
731: start = getInt(start, default=(math.ceil(float(current)/float(grpsize))*grpsize-(grpsize-1)))
732: # int(current / grpsize) * grpsize +1))
733: pageinfo['start'] = start
734:
735: np = int(docinfo.get('numPages', 0))
736: if np == 0:
737: # numPages unknown - maybe we can get it from text page
738: if docinfo.get('textURLPath', None):
739: # cache text page as well
740: pageinfo['textPage'] = self.getTextPage(mode=viewType, pn=current, docinfo=docinfo, pageinfo=pageinfo)
741: np = int(docinfo.get('numPages', 0))
742:
743: pageinfo['numgroups'] = int(np / grpsize)
744: if np % grpsize > 0:
745: pageinfo['numgroups'] += 1
746:
747: pageFlowLtr = docinfo.get('pageFlow', 'ltr') != 'rtl'
748: oddScanLeft = docinfo.get('oddPage', 'left') != 'right'
749: # add zeroth page for two columns
750: pageZero = (cols == 2 and (pageFlowLtr != oddScanLeft))
751: pageinfo['pageZero'] = pageZero
752: pageinfo['pageBatch'] = self.getPageBatch(start=start, rows=rows, cols=cols, pageFlowLtr=pageFlowLtr, pageZero=pageZero, minIdx=1, maxIdx=np)
753:
754: # TODO: do we need this here?
755: pageinfo['characterNormalization'] = self.REQUEST.get('characterNormalization','reg')
756: pageinfo['query'] = self.REQUEST.get('query','')
757: pageinfo['queryType'] = self.REQUEST.get('queryType','')
758: pageinfo['querySearch'] =self.REQUEST.get('querySearch', 'fulltext')
759: pageinfo['highlightQuery'] = self.REQUEST.get('highlightQuery','')
760: pageinfo['tocPageSize'] = getInt(self.REQUEST.get('tocPageSize', 30))
761: pageinfo['queryPageSize'] = getInt(self.REQUEST.get('queryPageSize', 10))
762: pageinfo['tocPN'] = getInt(self.REQUEST.get('tocPN', '1'))
763: pageinfo['searchPN'] = getInt(self.REQUEST.get('searchPN','1'))
764:
765: # limit tocPN
766: if 'tocSize_%s'%tocMode in docinfo:
767: tocSize = docinfo['tocSize_%s'%tocMode]
768: tocPageSize = pageinfo['tocPageSize']
769: # cached toc
770: if tocSize%tocPageSize>0:
771: tocPages=tocSize/tocPageSize+1
772: else:
773: tocPages=tocSize/tocPageSize
774:
775: pageinfo['tocPN'] = min(tocPages,pageinfo['tocPN'])
776:
777: return pageinfo
778:
779:
780: def getPageBatch(self, start=1, rows=10, cols=2, pageFlowLtr=True, pageZero=False, minIdx=1, maxIdx=0):
781: """returns dict with array of page informations for one screenfull of thumbnails"""
782: batch = {}
783: grpsize = rows * cols
784: if maxIdx == 0:
785: maxIdx = start + grpsize
786:
787: nb = int(math.ceil(maxIdx / float(grpsize)))
788: # list of all batch start and end points
789: batches = []
790: if pageZero:
791: ofs = 0
792: else:
793: ofs = 1
794:
795: for i in range(nb):
796: s = i * grpsize + ofs
797: e = min((i + 1) * grpsize + ofs - 1, maxIdx)
798: batches.append({'start':s, 'end':e})
799:
800: batch['batches'] = batches
801:
802: pages = []
803: if pageZero and start == 1:
804: # correct beginning
805: idx = 0
806: else:
807: idx = start
808:
809: for r in range(rows):
810: row = []
811: for c in range(cols):
812: if idx < minIdx or idx > maxIdx:
813: page = {'idx':None}
814: else:
815: page = {'idx':idx}
816:
817: idx += 1
818: if pageFlowLtr:
819: row.append(page)
820: else:
821: row.insert(0, page)
822:
823: pages.append(row)
824:
825: if start > 1:
826: batch['prevStart'] = max(start - grpsize, 1)
827: else:
828: batch['prevStart'] = None
829:
830: if start + grpsize < maxIdx:
831: batch['nextStart'] = start + grpsize
832: else:
833: batch['nextStart'] = None
834:
835: batch['pages'] = pages
836: return batch
837:
838: def getBatch(self, start=1, size=10, end=0, data=None, fullData=True):
839: """returns dict with information for one screenfull of data."""
840: batch = {}
841: if end == 0:
842: end = start + size
843:
844: nb = int(math.ceil(end / float(size)))
845: # list of all batch start and end points
846: batches = []
847: for i in range(nb):
848: s = i * size + 1
849: e = min((i + 1) * size, end)
850: batches.append({'start':s, 'end':e})
851:
852: batch['batches'] = batches
853: # list of elements in this batch
854: this = []
855: j = 0
856: for i in range(start, min(start+size, end)):
857: if data:
858: if fullData:
859: d = data[i]
860: else:
861: d = data[j]
862: j += 1
863:
864: else:
865: d = i+1
866:
867: this.append(d)
868:
869: batch['this'] = this
870: if start > 1:
871: batch['prevStart'] = max(start - size, 1)
872: else:
873: batch['prevStart'] = None
874:
875: if start + size < end:
876: batch['nextStart'] = start + size
877: else:
878: batch['nextStart'] = None
879:
880: return batch
881:
882:
883: security.declareProtected('View management screens','changeDocumentViewerForm')
884: changeDocumentViewerForm = PageTemplateFile('zpt/changeDocumentViewer', globals())
885:
886: def changeDocumentViewer(self,title="",digilibBaseUrl=None,thumbrows=2,thumbcols=5,authgroups='mpiwg',RESPONSE=None):
887: """init document viewer"""
888: self.title=title
889: self.digilibBaseUrl = digilibBaseUrl
890: self.thumbrows = thumbrows
891: self.thumbcols = thumbcols
892: self.authgroups = [s.strip().lower() for s in authgroups.split(',')]
893: try:
894: # assume MetaDataFolder instance is called metadata
895: self.metadataService = getattr(self, 'metadata')
896: except Exception, e:
897: logging.error("Unable to find MetaDataFolder 'metadata': "+str(e))
898:
899: if RESPONSE is not None:
900: RESPONSE.redirect('manage_main')
901:
902: def manage_AddDocumentViewerForm(self):
903: """add the viewer form"""
904: pt=PageTemplateFile('zpt/addDocumentViewer', globals()).__of__(self)
905: return pt()
906:
907: def manage_AddDocumentViewer(self,id,imageScalerUrl="",textServerName="",title="",RESPONSE=None):
908: """add the viewer"""
909: newObj=documentViewer(id,imageScalerUrl=imageScalerUrl,title=title,textServerName=textServerName)
910: self._setObject(id,newObj)
911:
912: if RESPONSE is not None:
913: RESPONSE.redirect('manage_main')
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>