Mercurial > hg > documentViewer
annotate documentViewer.py @ 459:aabfa6124cfb elementtree
nicer comments
author | casties |
---|---|
date | Wed, 20 Jul 2011 10:22:36 +0200 |
parents | 48b135b089c8 |
children | 76bc2317146f |
rev | line source |
---|---|
0 | 1 from OFS.Folder import Folder |
2 from Products.PageTemplates.ZopePageTemplate import ZopePageTemplate | |
52 | 3 from Products.PageTemplates.PageTemplateFile import PageTemplateFile |
0 | 4 from AccessControl import ClassSecurityInfo |
32 | 5 from AccessControl import getSecurityManager |
0 | 6 from Globals import package_home |
7 | |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
8 #from Ft.Xml import EMPTY_NAMESPACE, Parse |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
9 #import Ft.Xml.Domlette |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
10 |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
11 import xml.etree.ElementTree as ET |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
12 |
0 | 13 import os.path |
31 | 14 import sys |
0 | 15 import urllib |
50
6c0f20cecc60
added evaluation of the presentation/info.xml in texttools
dwinter
parents:
49
diff
changeset
|
16 import logging |
61 | 17 import math |
46 | 18 import urlparse |
174 | 19 import re |
389 | 20 import string |
231 | 21 |
458 | 22 from SrvTxtUtils import getInt, getText, getHttpData |
23 | |
52 | 24 def logger(txt,method,txt2): |
25 """logging""" | |
26 logging.info(txt+ txt2) | |
27 | |
28 | |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
29 def serializeNode(node, encoding="utf-8"): |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
30 """returns a string containing node as XML""" |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
31 s = ET.tostring(node) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
32 |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
33 # 4Suite: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
34 # stream = cStringIO.StringIO() |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
35 # Ft.Xml.Domlette.Print(node, stream=stream, encoding=encoding) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
36 # s = stream.getvalue() |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
37 # stream.close() |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
38 return s |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
39 |
387 | 40 def browserCheck(self): |
41 """check the browsers request to find out the browser type""" | |
42 bt = {} | |
43 ua = self.REQUEST.get_header("HTTP_USER_AGENT") | |
44 bt['ua'] = ua | |
45 bt['isIE'] = False | |
46 bt['isN4'] = False | |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
47 bt['versFirefox']="" |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
48 bt['versIE']="" |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
49 bt['versSafariChrome']="" |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
50 bt['versOpera']="" |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
51 |
387 | 52 if string.find(ua, 'MSIE') > -1: |
53 bt['isIE'] = True | |
54 else: | |
55 bt['isN4'] = (string.find(ua, 'Mozilla/4.') > -1) | |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
56 # Safari oder Chrome identification |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
57 try: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
58 nav = ua[string.find(ua, '('):] |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
59 nav1=ua[string.find(ua,')'):] |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
60 nav2=nav1[string.find(nav1,'('):] |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
61 nav3=nav2[string.find(nav2,')'):] |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
62 ie = string.split(nav, "; ")[1] |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
63 ie1 =string.split(nav1, " ")[2] |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
64 ie2 =string.split(nav3, " ")[1] |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
65 ie3 =string.split(nav3, " ")[2] |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
66 if string.find(ie3, "Safari") >-1: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
67 bt['versSafariChrome']=string.split(ie2, "/")[1] |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
68 except: pass |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
69 # IE identification |
387 | 70 try: |
71 nav = ua[string.find(ua, '('):] | |
72 ie = string.split(nav, "; ")[1] | |
73 if string.find(ie, "MSIE") > -1: | |
74 bt['versIE'] = string.split(ie, " ")[1] | |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
75 except:pass |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
76 # Firefox identification |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
77 try: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
78 nav = ua[string.find(ua, '('):] |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
79 nav1=ua[string.find(ua,')'):] |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
80 if string.find(ie1, "Firefox") >-1: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
81 nav5= string.split(ie1, "/")[1] |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
82 logging.debug("FIREFOX: %s"%(nav5)) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
83 bt['versFirefox']=nav5[0:3] |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
84 except:pass |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
85 #Opera identification |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
86 try: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
87 if string.find(ua,"Opera") >-1: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
88 nav = ua[string.find(ua, '('):] |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
89 nav1=nav[string.find(nav,')'):] |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
90 bt['versOpera']=string.split(nav1,"/")[2] |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
91 except:pass |
387 | 92 |
93 bt['isMac'] = string.find(ua, 'Macintosh') > -1 | |
94 bt['isWin'] = string.find(ua, 'Windows') > -1 | |
95 bt['isIEWin'] = bt['isIE'] and bt['isWin'] | |
96 bt['isIEMac'] = bt['isIE'] and bt['isMac'] | |
97 bt['staticHTML'] = False | |
98 | |
99 return bt | |
234 | 100 |
35 | 101 def getParentDir(path): |
102 """returns pathname shortened by one""" | |
103 return '/'.join(path.split('/')[0:-1]) | |
104 | |
458 | 105 def getBibdataFromDom(dom): |
106 """returns dict with all elements from bib-tag""" | |
107 bibinfo = {} | |
108 bib = dom.find(".//meta/bib") | |
109 if bib is not None: | |
110 # put type in @type | |
111 type = bib.get('type') | |
112 bibinfo['@type'] = type | |
113 # put all subelements in dict | |
114 for e in bib: | |
115 bibinfo[e.tag] = getText(e) | |
116 | |
117 return bibinfo | |
130
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
118 |
22 | 119 ## |
120 ## documentViewer class | |
121 ## | |
122 class documentViewer(Folder): | |
0 | 123 """document viewer""" |
124 meta_type="Document viewer" | |
125 | |
126 security=ClassSecurityInfo() | |
22 | 127 manage_options=Folder.manage_options+( |
0 | 128 {'label':'main config','action':'changeDocumentViewerForm'}, |
129 ) | |
130 | |
22 | 131 # templates and forms |
132 viewer_main = PageTemplateFile('zpt/viewer_main', globals()) | |
90
6a4a72033d58
new version with new full-text infrastructure and some more changed templates
casties
parents:
84
diff
changeset
|
133 toc_thumbs = PageTemplateFile('zpt/toc_thumbs', globals()) |
6a4a72033d58
new version with new full-text infrastructure and some more changed templates
casties
parents:
84
diff
changeset
|
134 toc_text = PageTemplateFile('zpt/toc_text', globals()) |
6a4a72033d58
new version with new full-text infrastructure and some more changed templates
casties
parents:
84
diff
changeset
|
135 toc_figures = PageTemplateFile('zpt/toc_figures', globals()) |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
136 page_main_images = PageTemplateFile('zpt/page_main_images', globals()) |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
137 page_main_double = PageTemplateFile('zpt/page_main_double', globals()) |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
138 page_main_text = PageTemplateFile('zpt/page_main_text', globals()) |
90
6a4a72033d58
new version with new full-text infrastructure and some more changed templates
casties
parents:
84
diff
changeset
|
139 page_main_text_dict = PageTemplateFile('zpt/page_main_text_dict', globals()) |
140 | 140 page_main_gis =PageTemplateFile ('zpt/page_main_gis', globals()) |
99 | 141 page_main_xml = PageTemplateFile('zpt/page_main_xml', globals()) |
404 | 142 page_main_pureXml = PageTemplateFile('zpt/page_main_pureXml', globals()) |
22 | 143 head_main = PageTemplateFile('zpt/head_main', globals()) |
144 docuviewer_css = PageTemplateFile('css/docuviewer.css', globals()) | |
57 | 145 info_xml = PageTemplateFile('zpt/info_xml', globals()) |
130
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
146 |
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
147 |
68 | 148 thumbs_main_rss = PageTemplateFile('zpt/thumbs_main_rss', globals()) |
22 | 149 security.declareProtected('View management screens','changeDocumentViewerForm') |
150 changeDocumentViewerForm = PageTemplateFile('zpt/changeDocumentViewer', globals()) | |
151 | |
0 | 152 |
95 | 153 def __init__(self,id,imageScalerUrl=None,textServerName=None,title="",digilibBaseUrl=None,thumbcols=2,thumbrows=5,authgroups="mpiwg"): |
0 | 154 """init document viewer""" |
155 self.id=id | |
156 self.title=title | |
25 | 157 self.thumbcols = thumbcols |
158 self.thumbrows = thumbrows | |
32 | 159 # authgroups is list of authorized groups (delimited by ,) |
160 self.authgroups = [s.strip().lower() for s in authgroups.split(',')] | |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
161 # create template folder so we can always use template.something |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
162 |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
163 templateFolder = Folder('template') |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
164 #self['template'] = templateFolder # Zope-2.12 style |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
165 self._setObject('template',templateFolder) # old style |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
166 try: |
130
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
167 import MpdlXmlTextServer |
132 | 168 textServer = MpdlXmlTextServer.MpdlXmlTextServer(id='fulltextclient',serverName=textServerName) |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
169 #templateFolder['fulltextclient'] = xmlRpcClient |
130
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
170 templateFolder._setObject('fulltextclient',textServer) |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
171 except Exception, e: |
130
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
172 logging.error("Unable to create MpdlXmlTextServer for fulltextclient: "+str(e)) |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
173 try: |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
174 from Products.zogiLib.zogiLib import zogiLib |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
175 zogilib = zogiLib(id="zogilib", title="zogilib for docuviewer", dlServerURL=imageScalerUrl, layout="book") |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
176 #templateFolder['zogilib'] = zogilib |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
177 templateFolder._setObject('zogilib',zogilib) |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
178 except Exception, e: |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
179 logging.error("Unable to create zogiLib for zogilib: "+str(e)) |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
180 |
130
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
181 |
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
182 # proxy text server methods to fulltextclient |
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
183 def getTextPage(self, **args): |
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
184 """get page""" |
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
185 return self.template.fulltextclient.getTextPage(**args) |
22 | 186 |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
187 def getOrigPages(self, **args): |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
188 """get page""" |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
189 return self.template.fulltextclient.getOrigPages(**args) |
396 | 190 |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
191 def getOrigPagesNorm(self, **args): |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
192 """get page""" |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
193 return self.template.fulltextclient.getOrigPagesNorm(**args) |
130
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
194 |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
195 def getQuery(self, **args): |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
196 """get query in search""" |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
197 return self.template.fulltextclient.getQuery(**args) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
198 |
130
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
199 def getSearch(self, **args): |
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
200 """get search""" |
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
201 return self.template.fulltextclient.getSearch(**args) |
256 | 202 |
203 def getGisPlaces(self, **args): | |
307 | 204 """get gis places""" |
256 | 205 return self.template.fulltextclient.getGisPlaces(**args) |
307 | 206 |
207 def getAllGisPlaces(self, **args): | |
310 | 208 """get all gis places """ |
209 return self.template.fulltextclient.getAllGisPlaces(**args) | |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
210 |
130
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
211 def getTranslate(self, **args): |
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
212 """get translate""" |
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
213 return self.template.fulltextclient.getTranslate(**args) |
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
214 |
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
215 def getLemma(self, **args): |
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
216 """get lemma""" |
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
217 return self.template.fulltextclient.getLemma(**args) |
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
218 |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
219 def getLemmaQuery(self, **args): |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
220 """get query""" |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
221 return self.template.fulltextclient.getLemmaQuery(**args) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
222 |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
223 def getLex(self, **args): |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
224 """get lex""" |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
225 return self.template.fulltextclient.getLex(**args) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
226 |
130
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
227 def getToc(self, **args): |
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
228 """get toc""" |
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
229 return self.template.fulltextclient.getToc(**args) |
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
230 |
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
231 def getTocPage(self, **args): |
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
232 """get tocpage""" |
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
233 return self.template.fulltextclient.getTocPage(**args) |
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
234 |
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
235 |
68 | 236 security.declareProtected('View','thumbs_rss') |
237 def thumbs_rss(self,mode,url,viewMode="auto",start=None,pn=1): | |
238 ''' | |
239 view it | |
240 @param mode: defines how to access the document behind url | |
241 @param url: url which contains display information | |
242 @param viewMode: if images display images, if text display text, default is images (text,images or auto) | |
243 | |
244 ''' | |
167 | 245 logging.debug("HHHHHHHHHHHHHH:load the rss") |
458 | 246 logging.debug("documentViewer (index) mode: %s url:%s start:%s pn:%s"%(mode,url,start,pn)) |
68 | 247 |
248 if not hasattr(self, 'template'): | |
249 # create template folder if it doesn't exist | |
250 self.manage_addFolder('template') | |
251 | |
252 if not self.digilibBaseUrl: | |
253 self.digilibBaseUrl = self.findDigilibUrl() or "http://nausikaa.mpiwg-berlin.mpg.de/digitallibrary" | |
254 | |
255 docinfo = self.getDocinfo(mode=mode,url=url) | |
338 | 256 #pageinfo = self.getPageinfo(start=start,current=pn,docinfo=docinfo) |
345 | 257 pageinfo = self.getPageinfo(start=start,current=pn, docinfo=docinfo) |
331 | 258 ''' ZDES ''' |
68 | 259 pt = getattr(self.template, 'thumbs_main_rss') |
260 | |
261 if viewMode=="auto": # automodus gewaehlt | |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
262 if docinfo.has_key("textURL") or docinfo.get('textURLPath',None): #texturl gesetzt und textViewer konfiguriert |
68 | 263 viewMode="text" |
264 else: | |
265 viewMode="images" | |
266 | |
267 return pt(docinfo=docinfo,pageinfo=pageinfo,viewMode=viewMode) | |
268 | |
22 | 269 security.declareProtected('View','index_html') |
405 | 270 def index_html(self,url,mode="texttool",viewMode="auto",tocMode="thumbs",start=None,pn=1,mk=None): |
22 | 271 ''' |
272 view it | |
57 | 273 @param mode: defines how to access the document behind url |
22 | 274 @param url: url which contains display information |
90
6a4a72033d58
new version with new full-text infrastructure and some more changed templates
casties
parents:
84
diff
changeset
|
275 @param viewMode: if images display images, if text display text, default is auto (text,images or auto) |
99 | 276 @param tocMode: type of 'table of contents' for navigation (thumbs, text, figures, none) |
141 | 277 @param characterNormalization type of text display (reg, norm, none) |
100 | 278 @param querySearch: type of different search modes (fulltext, fulltextMorph, xpath, xquery, ftIndex, ftIndexMorph, fulltextMorphLemma) |
22 | 279 ''' |
0 | 280 |
345 | 281 logging.debug("documentViewer (index) mode: %s url:%s start:%s pn:%s"%(mode,url,start,pn)) |
22 | 282 |
283 if not hasattr(self, 'template'): | |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
284 # this won't work |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
285 logging.error("template folder missing!") |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
286 return "ERROR: template folder missing!" |
22 | 287 |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
288 if not getattr(self, 'digilibBaseUrl', None): |
132 | 289 self.digilibBaseUrl = self.findDigilibUrl() or "http://digilib.mpiwg-berlin.mpg.de/digitallibrary" |
22 | 290 |
25 | 291 docinfo = self.getDocinfo(mode=mode,url=url) |
97 | 292 |
90
6a4a72033d58
new version with new full-text infrastructure and some more changed templates
casties
parents:
84
diff
changeset
|
293 if tocMode != "thumbs": |
6a4a72033d58
new version with new full-text infrastructure and some more changed templates
casties
parents:
84
diff
changeset
|
294 # get table of contents |
6a4a72033d58
new version with new full-text infrastructure and some more changed templates
casties
parents:
84
diff
changeset
|
295 docinfo = self.getToc(mode=tocMode, docinfo=docinfo) |
455 | 296 |
297 # auto viewMode: text_dict if text else images | |
298 if viewMode=="auto": | |
299 if docinfo.get('textURL', None) or docinfo.get('textURLPath', None): | |
300 #texturl gesetzt und textViewer konfiguriert | |
127 | 301 viewMode="text_dict" |
51
c5d3aabbf61b
textviewer now integrated, new modus auto introduced as standard for viewing
dwinter
parents:
50
diff
changeset
|
302 else: |
c5d3aabbf61b
textviewer now integrated, new modus auto introduced as standard for viewing
dwinter
parents:
50
diff
changeset
|
303 viewMode="images" |
90
6a4a72033d58
new version with new full-text infrastructure and some more changed templates
casties
parents:
84
diff
changeset
|
304 |
455 | 305 pageinfo = self.getPageinfo(start=start, current=pn, docinfo=docinfo, viewMode=viewMode, tocMode=tocMode) |
127 | 306 |
455 | 307 if viewMode != 'images' and docinfo.get('textURLPath', None): |
308 # get full text page | |
309 page = self.getTextPage(mode=viewMode, pn=pn, docinfo=docinfo, pageinfo=pageinfo) | |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
310 pageinfo['textPage'] = page |
455 | 311 |
312 # get template /template/viewer_main | |
313 pt = getattr(self.template, 'viewer_main') | |
314 # and execute with parameters | |
315 return pt(docinfo=docinfo, pageinfo=pageinfo, viewMode=viewMode, mk=self.generateMarks(mk)) | |
0 | 316 |
74 | 317 def generateMarks(self,mk): |
318 ret="" | |
90
6a4a72033d58
new version with new full-text infrastructure and some more changed templates
casties
parents:
84
diff
changeset
|
319 if mk is None: |
6a4a72033d58
new version with new full-text infrastructure and some more changed templates
casties
parents:
84
diff
changeset
|
320 return "" |
134 | 321 if not isinstance(mk, list): |
132 | 322 mk=[mk] |
74 | 323 for m in mk: |
75
9673218e155b
minorCVS: ----------------------------------------------------------------------
dwinter
parents:
74
diff
changeset
|
324 ret+="mk=%s"%m |
74 | 325 return ret |
389 | 326 |
327 | |
387 | 328 def getBrowser(self): |
329 """getBrowser the version of browser """ | |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
330 bt = browserCheck(self) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
331 logging.debug("BROWSER VERSION: %s"%(bt)) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
332 return bt |
387 | 333 |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
334 def findDigilibUrl(self): |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
335 """try to get the digilib URL from zogilib""" |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
336 url = self.template.zogilib.getDLBaseUrl() |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
337 return url |
126 | 338 |
339 def getDocumentViewerURL(self): | |
340 """returns the URL of this instance""" | |
341 return self.absolute_url() | |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
342 |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
343 def getStyle(self, idx, selected, style=""): |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
344 """returns a string with the given style and append 'sel' if path == selected.""" |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
345 #logger("documentViewer (getstyle)", logging.INFO, "idx: %s selected: %s style: %s"%(idx,selected,style)) |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
346 if idx == selected: |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
347 return style + 'sel' |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
348 else: |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
349 return style |
74 | 350 |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
351 def getLink(self, param=None, val=None, params=None, baseUrl=None, paramSep='&'): |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
352 """returns URL to documentviewer with parameter param set to val or from dict params""" |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
353 # copy existing request params |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
354 urlParams=self.REQUEST.form.copy() |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
355 # change single param |
25 | 356 if param is not None: |
31 | 357 if val is None: |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
358 if urlParams.has_key(param): |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
359 del urlParams[param] |
25 | 360 else: |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
361 urlParams[param] = str(val) |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
362 |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
363 # change more params |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
364 if params is not None: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
365 for k in params.keys(): |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
366 v = params[k] |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
367 if v is None: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
368 # val=None removes param |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
369 if urlParams.has_key(k): |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
370 del urlParams[k] |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
371 |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
372 else: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
373 urlParams[k] = v |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
374 |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
375 # FIXME: does this belong here? |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
376 if urlParams.get("mode", None) == "filepath": #wenn beim erst Aufruf filepath gesetzt wurde aendere das nun zu imagepath |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
377 urlParams["mode"] = "imagepath" |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
378 urlParams["url"] = getParentDir(urlParams["url"]) |
31 | 379 |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
380 # quote values and assemble into query string (not escaping '/') |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
381 ps = paramSep.join(["%s=%s"%(k,urllib.quote_plus(v,'/')) for (k, v) in urlParams.items()]) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
382 #ps = urllib.urlencode(urlParams) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
383 if baseUrl is None: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
384 baseUrl = self.REQUEST['URL1'] |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
385 |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
386 url = "%s?%s"%(baseUrl, ps) |
25 | 387 return url |
388 | |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
389 |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
390 def getLinkAmp(self, param=None, val=None, params=None, baseUrl=None): |
68 | 391 """link to documentviewer with parameter param set to val""" |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
392 return self.getLink(param, val, params, baseUrl, '&') |
81
fae97f071724
fixed problem with info.xml when url without index.meta
casties
parents:
79
diff
changeset
|
393 |
57 | 394 def getInfo_xml(self,url,mode): |
395 """returns info about the document as XML""" | |
396 | |
397 if not self.digilibBaseUrl: | |
398 self.digilibBaseUrl = self.findDigilibUrl() or "http://nausikaa.mpiwg-berlin.mpg.de/digitallibrary" | |
399 | |
400 docinfo = self.getDocinfo(mode=mode,url=url) | |
401 pt = getattr(self.template, 'info_xml') | |
402 return pt(docinfo=docinfo) | |
403 | |
394 | 404 def getOptionToggle(self, newState=None, optionName='text_options_open', initialState=True): |
405 """returns new option state""" | |
397 | 406 if not self.REQUEST.SESSION.has_key(optionName): |
394 | 407 # not in session -- initial |
408 opt = {'lastState': newState, 'state': initialState} | |
409 else: | |
397 | 410 opt = self.REQUEST.SESSION.get(optionName) |
394 | 411 if opt['lastState'] != newState: |
412 # state in session has changed -- toggle | |
413 opt['state'] = not opt['state'] | |
414 opt['lastState'] = newState | |
415 | |
416 self.REQUEST.SESSION[optionName] = opt | |
417 return opt['state'] | |
0 | 418 |
35 | 419 def isAccessible(self, docinfo): |
32 | 420 """returns if access to the resource is granted""" |
421 access = docinfo.get('accessType', None) | |
167 | 422 logging.debug("documentViewer (accessOK) access type %s"%access) |
45 | 423 if access is not None and access == 'free': |
167 | 424 logging.debug("documentViewer (accessOK) access is free") |
32 | 425 return True |
45 | 426 elif access is None or access in self.authgroups: |
35 | 427 # only local access -- only logged in users |
428 user = getSecurityManager().getUser() | |
167 | 429 logging.debug("documentViewer (accessOK) user=%s ip=%s"%(user,self.REQUEST.getClientAddr())) |
35 | 430 if user is not None: |
431 #print "user: ", user | |
432 return (user.getUserName() != "Anonymous User") | |
433 else: | |
434 return False | |
32 | 435 |
167 | 436 logging.error("documentViewer (accessOK) unknown access type %s"%access) |
32 | 437 return False |
35 | 438 |
32 | 439 |
73 | 440 def getDirinfoFromDigilib(self,path,docinfo=None,cut=0): |
29 | 441 """gibt param von dlInfo aus""" |
31 | 442 if docinfo is None: |
443 docinfo = {} | |
73 | 444 |
445 for x in range(cut): | |
454 | 446 path=getParentDir(path) |
78 | 447 |
40 | 448 infoUrl=self.digilibBaseUrl+"/dirInfo-xml.jsp?mo=dir&fn="+path |
29 | 449 |
167 | 450 logging.debug("documentViewer (getparamfromdigilib) dirInfo from %s"%(infoUrl)) |
29 | 451 |
130
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
452 txt = getHttpData(infoUrl) |
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
453 if txt is None: |
40 | 454 raise IOError("Unable to get dir-info from %s"%(infoUrl)) |
130
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
455 |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
456 dom = ET.fromstring(txt) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
457 #dom = Parse(txt) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
458 size=getText(dom.find("size")) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
459 #sizes=dom.xpath("//dir/size") |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
460 logging.debug("documentViewer (getparamfromdigilib) dirInfo:size=%s"%size) |
29 | 461 |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
462 if size: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
463 docinfo['numPages'] = int(size) |
31 | 464 else: |
465 docinfo['numPages'] = 0 | |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
466 |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
467 # TODO: produce and keep list of image names and numbers |
31 | 468 |
469 return docinfo | |
32 | 470 |
174 | 471 def getIndexMetaPath(self,url): |
472 """gib nur den Pfad zurueck""" | |
473 regexp = re.compile(r".*(experimental|permanent)/(.*)") | |
474 regpath = regexp.match(url) | |
475 if (regpath==None): | |
476 return "" | |
224 | 477 logging.debug("(getDomFromIndexMeta): URLXAXA: %s"%regpath.group(2)) |
174 | 478 return ("/mpiwg/online/"+regpath.group(1)+"/"+regpath.group(2)) |
479 | |
225 | 480 |
481 | |
174 | 482 def getIndexMetaUrl(self,url): |
483 """returns utr of index.meta document at url""" | |
484 | |
39
1dd90aabd366
added retry when reading index meta from texter applet
casties
parents:
38
diff
changeset
|
485 metaUrl = None |
35 | 486 if url.startswith("http://"): |
487 # real URL | |
39
1dd90aabd366
added retry when reading index meta from texter applet
casties
parents:
38
diff
changeset
|
488 metaUrl = url |
35 | 489 else: |
490 # online path | |
491 server=self.digilibBaseUrl+"/servlet/Texter?fn=" | |
40 | 492 metaUrl=server+url.replace("/mpiwg/online","") |
35 | 493 if not metaUrl.endswith("index.meta"): |
494 metaUrl += "/index.meta" | |
174 | 495 |
496 return metaUrl | |
497 | |
498 def getDomFromIndexMeta(self, url): | |
499 """get dom from index meta""" | |
500 dom = None | |
501 metaUrl = self.getIndexMetaUrl(url) | |
39
1dd90aabd366
added retry when reading index meta from texter applet
casties
parents:
38
diff
changeset
|
502 |
174 | 503 logging.debug("(getDomFromIndexMeta): METAURL: %s"%metaUrl) |
130
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
504 txt=getHttpData(metaUrl) |
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
505 if txt is None: |
39
1dd90aabd366
added retry when reading index meta from texter applet
casties
parents:
38
diff
changeset
|
506 raise IOError("Unable to read index meta from %s"%(url)) |
130
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
507 |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
508 dom = ET.fromstring(txt) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
509 #dom = Parse(txt) |
35 | 510 return dom |
50
6c0f20cecc60
added evaluation of the presentation/info.xml in texttools
dwinter
parents:
49
diff
changeset
|
511 |
6c0f20cecc60
added evaluation of the presentation/info.xml in texttools
dwinter
parents:
49
diff
changeset
|
512 def getPresentationInfoXML(self, url): |
6c0f20cecc60
added evaluation of the presentation/info.xml in texttools
dwinter
parents:
49
diff
changeset
|
513 """returns dom of info.xml document at url""" |
6c0f20cecc60
added evaluation of the presentation/info.xml in texttools
dwinter
parents:
49
diff
changeset
|
514 dom = None |
6c0f20cecc60
added evaluation of the presentation/info.xml in texttools
dwinter
parents:
49
diff
changeset
|
515 metaUrl = None |
6c0f20cecc60
added evaluation of the presentation/info.xml in texttools
dwinter
parents:
49
diff
changeset
|
516 if url.startswith("http://"): |
6c0f20cecc60
added evaluation of the presentation/info.xml in texttools
dwinter
parents:
49
diff
changeset
|
517 # real URL |
6c0f20cecc60
added evaluation of the presentation/info.xml in texttools
dwinter
parents:
49
diff
changeset
|
518 metaUrl = url |
6c0f20cecc60
added evaluation of the presentation/info.xml in texttools
dwinter
parents:
49
diff
changeset
|
519 else: |
6c0f20cecc60
added evaluation of the presentation/info.xml in texttools
dwinter
parents:
49
diff
changeset
|
520 # online path |
6c0f20cecc60
added evaluation of the presentation/info.xml in texttools
dwinter
parents:
49
diff
changeset
|
521 server=self.digilibBaseUrl+"/servlet/Texter?fn=" |
6c0f20cecc60
added evaluation of the presentation/info.xml in texttools
dwinter
parents:
49
diff
changeset
|
522 metaUrl=server+url.replace("/mpiwg/online","") |
6c0f20cecc60
added evaluation of the presentation/info.xml in texttools
dwinter
parents:
49
diff
changeset
|
523 |
130
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
524 txt=getHttpData(metaUrl) |
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
525 if txt is None: |
50
6c0f20cecc60
added evaluation of the presentation/info.xml in texttools
dwinter
parents:
49
diff
changeset
|
526 raise IOError("Unable to read infoXMLfrom %s"%(url)) |
130
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
527 |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
528 dom = ET.fromstring(txt) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
529 #dom = Parse(txt) |
50
6c0f20cecc60
added evaluation of the presentation/info.xml in texttools
dwinter
parents:
49
diff
changeset
|
530 return dom |
35 | 531 |
532 | |
70 | 533 def getAuthinfoFromIndexMeta(self,path,docinfo=None,dom=None,cut=0): |
35 | 534 """gets authorization info from the index.meta file at path or given by dom""" |
167 | 535 logging.debug("documentViewer (getauthinfofromindexmeta) path: %s"%(path)) |
32 | 536 |
537 access = None | |
538 | |
539 if docinfo is None: | |
540 docinfo = {} | |
541 | |
542 if dom is None: | |
78 | 543 for x in range(cut): |
70 | 544 path=getParentDir(path) |
174 | 545 dom = self.getDomFromIndexMeta(path) |
46 | 546 |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
547 acc = dom.find(".//access-conditions/access") |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
548 if acc is not None: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
549 acctype = acc.get('type') |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
550 #acctype = dom.xpath("//access-conditions/access/@type") |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
551 if acctype: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
552 access=acctype |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
553 if access in ['group', 'institution']: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
554 access = dom.find(".//access-conditions/access/name").text.lower() |
32 | 555 |
556 docinfo['accessType'] = access | |
557 return docinfo | |
29 | 558 |
32 | 559 |
70 | 560 def getBibinfoFromIndexMeta(self,path,docinfo=None,dom=None,cut=0): |
35 | 561 """gets bibliographical info from the index.meta file at path or given by dom""" |
167 | 562 logging.debug("documentViewer (getbibinfofromindexmeta) path: %s"%(path)) |
20 | 563 |
22 | 564 if docinfo is None: |
565 docinfo = {} | |
78 | 566 |
22 | 567 if dom is None: |
78 | 568 for x in range(cut): |
70 | 569 path=getParentDir(path) |
174 | 570 dom = self.getDomFromIndexMeta(path) |
571 | |
572 docinfo['indexMetaPath']=self.getIndexMetaPath(path); | |
79
df6952ac93e9
bug in getDocInforFromImagePath, relative lage der index.meta zu path war falsch.
dwinter
parents:
78
diff
changeset
|
573 |
167 | 574 logging.debug("documentViewer (getbibinfofromindexmeta cutted) path: %s"%(path)) |
459 | 575 # put all raw bib fields in dict "bib" |
458 | 576 bib = getBibdataFromDom(dom) |
577 docinfo['bib'] = bib | |
59 | 578 |
579 # extract some fields (author, title, year) according to their mapping | |
25 | 580 metaData=self.metadata.main.meta.bib |
458 | 581 bibtype=bib.get("@type") |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
582 #bibtype=dom.xpath("//bib/@type") |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
583 if not bibtype: |
25 | 584 bibtype="generic" |
59 | 585 |
458 | 586 bibtype=bibtype.replace("-"," ") # wrong types in index meta "-" instead of " " (not wrong! ROC) |
59 | 587 docinfo['bib_type'] = bibtype |
25 | 588 bibmap=metaData.generateMappingForType(bibtype) |
174 | 589 logging.debug("documentViewer (getbibinfofromindexmeta) bibmap:"+repr(bibmap)) |
590 logging.debug("documentViewer (getbibinfofromindexmeta) bibtype:"+repr(bibtype)) | |
32 | 591 # if there is no mapping bibmap is empty (mapping sometimes has empty fields) |
457 | 592 if len(bibmap) > 0 and bibmap.get('author',None) or bibmap.get('title',None): |
63 | 593 try: |
458 | 594 docinfo['author']=bib.get(bibmap['author'][0]) |
63 | 595 except: pass |
596 try: | |
458 | 597 docinfo['title']=bib.get(bibmap['title'][0]) |
63 | 598 except: pass |
599 try: | |
458 | 600 docinfo['year']=bib.get(bibmap['year'][0]) |
63 | 601 except: pass |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
602 |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
603 # ROC: why is this here? |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
604 # logging.debug("documentViewer (getbibinfofromindexmeta) using mapping for %s"%bibtype) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
605 # try: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
606 # docinfo['lang']=getTextFromNode(dom.find(".//bib/lang")[0]) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
607 # except: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
608 # docinfo['lang']='' |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
609 # try: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
610 # docinfo['city']=getTextFromNode(dom.find(".//bib/city")[0]) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
611 # except: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
612 # docinfo['city']='' |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
613 # try: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
614 # docinfo['number_of_pages']=getTextFromNode(dom.find(".//bib/number_of_pages")[0]) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
615 # except: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
616 # docinfo['number_of_pages']='' |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
617 # try: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
618 # docinfo['series_volume']=getTextFromNode(dom.find(".//bib/series_volume")[0]) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
619 # except: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
620 # docinfo['series_volume']='' |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
621 # try: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
622 # docinfo['number_of_volumes']=getTextFromNode(dom.find(".//bib/number_of_volumes")[0]) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
623 # except: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
624 # docinfo['number_of_volumes']='' |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
625 # try: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
626 # docinfo['translator']=getTextFromNode(dom.find(".//bib/translator")[0]) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
627 # except: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
628 # docinfo['translator']='' |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
629 # try: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
630 # docinfo['edition']=getTextFromNode(dom.find(".//bib/edition")[0]) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
631 # except: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
632 # docinfo['edition']='' |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
633 # try: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
634 # docinfo['series_author']=getTextFromNode(dom.find(".//bib/series_author")[0]) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
635 # except: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
636 # docinfo['series_author']='' |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
637 # try: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
638 # docinfo['publisher']=getTextFromNode(dom.find(".//bib/publisher")[0]) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
639 # except: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
640 # docinfo['publisher']='' |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
641 # try: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
642 # docinfo['series_title']=getTextFromNode(dom.find(".//bib/series_title")[0]) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
643 # except: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
644 # docinfo['series_title']='' |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
645 # try: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
646 # docinfo['isbn_issn']=getTextFromNode(dom.find(".//bib/isbn_issn")[0]) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
647 # except: |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
648 # docinfo['isbn_issn']='' |
22 | 649 return docinfo |
83 | 650 |
454 | 651 |
652 # TODO: is this needed? | |
218 | 653 def getNameFromIndexMeta(self,path,docinfo=None,dom=None,cut=0): |
654 """gets name info from the index.meta file at path or given by dom""" | |
655 if docinfo is None: | |
656 docinfo = {} | |
657 | |
658 if dom is None: | |
659 for x in range(cut): | |
660 path=getParentDir(path) | |
661 dom = self.getDomFromIndexMeta(path) | |
330 | 662 |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
663 docinfo['name']=getText(dom.find("name")) |
230 | 664 logging.debug("documentViewer docinfo[name] %s"%docinfo['name']) |
218 | 665 return docinfo |
83 | 666 |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
667 def getDocinfoFromTextTool(self, url, dom=None, docinfo=None): |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
668 """parse texttool tag in index meta""" |
167 | 669 logging.debug("documentViewer (getdocinfofromtexttool) url: %s" % (url)) |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
670 if docinfo is None: |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
671 docinfo = {} |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
672 if docinfo.get('lang', None) is None: |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
673 docinfo['lang'] = '' # default keine Sprache gesetzt |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
674 if dom is None: |
174 | 675 dom = self.getDomFromIndexMeta(url) |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
676 |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
677 archivePath = None |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
678 archiveName = None |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
679 |
454 | 680 archiveName = getText(dom.find("name")) |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
681 if not archiveName: |
130
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
682 logging.warning("documentViewer (getdocinfofromtexttool) resource/name missing in: %s" % (url)) |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
683 |
454 | 684 archivePath = getText(dom.find("archive-path")) |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
685 if archivePath: |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
686 # clean up archive path |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
687 if archivePath[0] != '/': |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
688 archivePath = '/' + archivePath |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
689 if archiveName and (not archivePath.endswith(archiveName)): |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
690 archivePath += "/" + archiveName |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
691 else: |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
692 # try to get archive-path from url |
167 | 693 logging.warning("documentViewer (getdocinfofromtexttool) resource/archive-path missing in: %s" % (url)) |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
694 if (not url.startswith('http')): |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
695 archivePath = url.replace('index.meta', '') |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
696 |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
697 if archivePath is None: |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
698 # we balk without archive-path |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
699 raise IOError("Missing archive-path (for text-tool) in %s" % (url)) |
22 | 700 |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
701 imageDir = getText(dom.find(".//texttool/image")) |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
702 |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
703 if not imageDir: |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
704 # we balk with no image tag / not necessary anymore because textmode is now standard |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
705 #raise IOError("No text-tool info in %s"%(url)) |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
706 imageDir = "" |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
707 #xquery="//pb" |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
708 docinfo['imagePath'] = "" # keine Bilder |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
709 docinfo['imageURL'] = "" |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
710 |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
711 if imageDir and archivePath: |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
712 #print "image: ", imageDir, " archivepath: ", archivePath |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
713 imageDir = os.path.join(archivePath, imageDir) |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
714 imageDir = imageDir.replace("/mpiwg/online", '') |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
715 docinfo = self.getDirinfoFromDigilib(imageDir, docinfo=docinfo) |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
716 docinfo['imagePath'] = imageDir |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
717 |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
718 docinfo['imageURL'] = self.digilibBaseUrl + "/servlet/Scaler?fn=" + imageDir |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
719 |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
720 viewerUrl = getText(dom.find(".//texttool/digiliburlprefix")) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
721 if viewerUrl: |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
722 docinfo['viewerURL'] = viewerUrl |
130
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
723 |
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
724 # old style text URL |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
725 textUrl = getText(dom.find(".//texttool/text")) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
726 if textUrl: |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
727 if urlparse.urlparse(textUrl)[0] == "": #keine url |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
728 textUrl = os.path.join(archivePath, textUrl) |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
729 # fix URLs starting with /mpiwg/online |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
730 if textUrl.startswith("/mpiwg/online"): |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
731 textUrl = textUrl.replace("/mpiwg/online", '', 1) |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
732 |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
733 docinfo['textURL'] = textUrl |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
734 |
130
5c779d7b5f71
more modular version with separate object MpdlXmlTextServer
casties
parents:
128
diff
changeset
|
735 # new style text-url-path |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
736 textUrl = getText(dom.find(".//texttool/text-url-path")) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
737 if textUrl: |
102 | 738 docinfo['textURLPath'] = textUrl |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
739 textUrlkurz = string.split(textUrl, ".")[0] |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
740 docinfo['textURLPathkurz'] = textUrlkurz |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
741 #if not docinfo['imagePath']: |
102 | 742 # text-only, no page images |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
743 #docinfo = self.getNumTextPages(docinfo) |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
744 |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
745 |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
746 presentationUrl = getText(dom.find(".//texttool/presentation")) |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
747 docinfo = self.getBibinfoFromIndexMeta(url, docinfo=docinfo, dom=dom) # get info von bib tag |
454 | 748 # TODO: is this needed here? |
228 | 749 docinfo = self.getNameFromIndexMeta(url, docinfo=docinfo, dom=dom) |
386 | 750 |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
751 |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
752 if presentationUrl: # ueberschreibe diese durch presentation informationen |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
753 # presentation url ergiebt sich ersetzen von index.meta in der url der fuer die Metadaten |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
754 # durch den relativen Pfad auf die presentation infos |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
755 presentationPath = presentationUrl |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
756 if url.endswith("index.meta"): |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
757 presentationUrl = url.replace('index.meta', presentationPath) |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
758 else: |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
759 presentationUrl = url + "/" + presentationPath |
102 | 760 |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
761 docinfo = self.getBibinfoFromTextToolPresentation(presentationUrl, docinfo=docinfo, dom=dom) |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
762 |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
763 docinfo = self.getAuthinfoFromIndexMeta(url, docinfo=docinfo, dom=dom) # get access info |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
764 |
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
765 return docinfo |
22 | 766 |
50
6c0f20cecc60
added evaluation of the presentation/info.xml in texttools
dwinter
parents:
49
diff
changeset
|
767 |
6c0f20cecc60
added evaluation of the presentation/info.xml in texttools
dwinter
parents:
49
diff
changeset
|
768 def getBibinfoFromTextToolPresentation(self,url,docinfo=None,dom=None): |
6c0f20cecc60
added evaluation of the presentation/info.xml in texttools
dwinter
parents:
49
diff
changeset
|
769 """gets the bibliographical information from the preseantion entry in texttools |
6c0f20cecc60
added evaluation of the presentation/info.xml in texttools
dwinter
parents:
49
diff
changeset
|
770 """ |
6c0f20cecc60
added evaluation of the presentation/info.xml in texttools
dwinter
parents:
49
diff
changeset
|
771 dom=self.getPresentationInfoXML(url) |
454 | 772 docinfo['author']=getText(dom.find(".//author")) |
773 docinfo['title']=getText(dom.find(".//title")) | |
774 docinfo['year']=getText(dom.find(".//date")) | |
50
6c0f20cecc60
added evaluation of the presentation/info.xml in texttools
dwinter
parents:
49
diff
changeset
|
775 return docinfo |
6c0f20cecc60
added evaluation of the presentation/info.xml in texttools
dwinter
parents:
49
diff
changeset
|
776 |
70 | 777 def getDocinfoFromImagePath(self,path,docinfo=None,cut=0): |
22 | 778 """path ist the path to the images it assumes that the index.meta file is one level higher.""" |
167 | 779 logging.debug("documentViewer (getdocinfofromimagepath) path: %s"%(path)) |
22 | 780 if docinfo is None: |
781 docinfo = {} | |
29 | 782 path=path.replace("/mpiwg/online","") |
22 | 783 docinfo['imagePath'] = path |
73 | 784 docinfo=self.getDirinfoFromDigilib(path,docinfo=docinfo,cut=cut) |
78 | 785 |
79
df6952ac93e9
bug in getDocInforFromImagePath, relative lage der index.meta zu path war falsch.
dwinter
parents:
78
diff
changeset
|
786 pathorig=path |
78 | 787 for x in range(cut): |
788 path=getParentDir(path) | |
167 | 789 logging.debug("documentViewer (getdocinfofromimagepath) PATH:"+path) |
31 | 790 imageUrl=self.digilibBaseUrl+"/servlet/Scaler?fn="+path |
22 | 791 docinfo['imageURL'] = imageUrl |
792 | |
79
df6952ac93e9
bug in getDocInforFromImagePath, relative lage der index.meta zu path war falsch.
dwinter
parents:
78
diff
changeset
|
793 #path ist the path to the images it assumes that the index.meta file is one level higher. |
df6952ac93e9
bug in getDocInforFromImagePath, relative lage der index.meta zu path war falsch.
dwinter
parents:
78
diff
changeset
|
794 docinfo = self.getBibinfoFromIndexMeta(pathorig,docinfo=docinfo,cut=cut+1) |
df6952ac93e9
bug in getDocInforFromImagePath, relative lage der index.meta zu path war falsch.
dwinter
parents:
78
diff
changeset
|
795 docinfo = self.getAuthinfoFromIndexMeta(pathorig,docinfo=docinfo,cut=cut+1) |
22 | 796 return docinfo |
20 | 797 |
22 | 798 |
799 def getDocinfo(self, mode, url): | |
800 """returns docinfo depending on mode""" | |
167 | 801 logging.debug("documentViewer (getdocinfo) mode: %s, url: %s"%(mode,url)) |
22 | 802 # look for cached docinfo in session |
51
c5d3aabbf61b
textviewer now integrated, new modus auto introduced as standard for viewing
dwinter
parents:
50
diff
changeset
|
803 if self.REQUEST.SESSION.has_key('docinfo'): |
22 | 804 docinfo = self.REQUEST.SESSION['docinfo'] |
805 # check if its still current | |
806 if docinfo is not None and docinfo.get('mode') == mode and docinfo.get('url') == url: | |
455 | 807 logging.debug("documentViewer (getdocinfo) docinfo in session. keys=%s"%docinfo.keys()) |
22 | 808 return docinfo |
455 | 809 |
22 | 810 # new docinfo |
811 docinfo = {'mode': mode, 'url': url} | |
455 | 812 # add self url |
813 docinfo['viewerUrl'] = self.getDocumentViewerURL() | |
814 if mode=="texttool": | |
815 # index.meta with texttool information | |
22 | 816 docinfo = self.getDocinfoFromTextTool(url, docinfo=docinfo) |
817 elif mode=="imagepath": | |
455 | 818 # folder with images, index.meta optional |
22 | 819 docinfo = self.getDocinfoFromImagePath(url, docinfo=docinfo) |
70 | 820 elif mode=="filepath": |
455 | 821 # filename |
75
9673218e155b
minorCVS: ----------------------------------------------------------------------
dwinter
parents:
74
diff
changeset
|
822 docinfo = self.getDocinfoFromImagePath(url, docinfo=docinfo,cut=1) |
22 | 823 else: |
167 | 824 logging.error("documentViewer (getdocinfo) unknown mode: %s!"%mode) |
90
6a4a72033d58
new version with new full-text infrastructure and some more changed templates
casties
parents:
84
diff
changeset
|
825 raise ValueError("Unknown mode %s! Has to be one of 'texttool','imagepath','filepath'."%(mode)) |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
826 |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
827 # FIXME: fake texturlpath |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
828 if not docinfo.has_key('textURLPath'): |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
829 docinfo['textURLPath'] = None |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
830 |
458 | 831 logging.debug("documentViewer (getdocinfo) docinfo: keys=%s"%docinfo.keys()) |
832 #logging.debug("documentViewer (getdocinfo) docinfo: %s"%docinfo) | |
22 | 833 self.REQUEST.SESSION['docinfo'] = docinfo |
834 return docinfo | |
128 | 835 |
405 | 836 def getPageinfo(self, current, start=None, rows=None, cols=None, docinfo=None, viewMode=None, tocMode=None): |
22 | 837 """returns pageinfo with the given parameters""" |
838 pageinfo = {} | |
25 | 839 current = getInt(current) |
338 | 840 |
25 | 841 pageinfo['current'] = current |
842 rows = int(rows or self.thumbrows) | |
843 pageinfo['rows'] = rows | |
844 cols = int(cols or self.thumbcols) | |
845 pageinfo['cols'] = cols | |
846 grpsize = cols * rows | |
847 pageinfo['groupsize'] = grpsize | |
61 | 848 start = getInt(start, default=(math.ceil(float(current)/float(grpsize))*grpsize-(grpsize-1))) |
849 # int(current / grpsize) * grpsize +1)) | |
22 | 850 pageinfo['start'] = start |
25 | 851 pageinfo['end'] = start + grpsize |
90
6a4a72033d58
new version with new full-text infrastructure and some more changed templates
casties
parents:
84
diff
changeset
|
852 if (docinfo is not None) and ('numPages' in docinfo): |
25 | 853 np = int(docinfo['numPages']) |
854 pageinfo['end'] = min(pageinfo['end'], np) | |
855 pageinfo['numgroups'] = int(np / grpsize) | |
856 if np % grpsize > 0: | |
128 | 857 pageinfo['numgroups'] += 1 |
90
6a4a72033d58
new version with new full-text infrastructure and some more changed templates
casties
parents:
84
diff
changeset
|
858 pageinfo['viewMode'] = viewMode |
6a4a72033d58
new version with new full-text infrastructure and some more changed templates
casties
parents:
84
diff
changeset
|
859 pageinfo['tocMode'] = tocMode |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
860 pageinfo['characterNormalization'] = self.REQUEST.get('characterNormalization','reg') |
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
861 #pageinfo['optionToggle'] = self.REQUEST.get('optionToggle','1') |
398 | 862 pageinfo['query'] = self.REQUEST.get('query','') |
384 | 863 pageinfo['queryType'] = self.REQUEST.get('queryType','') |
95 | 864 pageinfo['querySearch'] =self.REQUEST.get('querySearch', 'fulltext') |
99 | 865 pageinfo['textPN'] = self.REQUEST.get('textPN','1') |
384 | 866 pageinfo['highlightQuery'] = self.REQUEST.get('highlightQuery','') |
95 | 867 pageinfo['tocPageSize'] = self.REQUEST.get('tocPageSize', '30') |
105 | 868 pageinfo['queryPageSize'] =self.REQUEST.get('queryPageSize', '10') |
453
beb7ccb92564
first version using elementtree instead of 4suite xml
casties
parents:
405
diff
changeset
|
869 pageinfo['tocPN'] = self.REQUEST.get('tocPN', '1') |
99 | 870 toc = int (pageinfo['tocPN']) |
871 pageinfo['textPages'] =int (toc) | |
158 | 872 |
99 | 873 if 'tocSize_%s'%tocMode in docinfo: |
874 tocSize = int(docinfo['tocSize_%s'%tocMode]) | |
875 tocPageSize = int(pageinfo['tocPageSize']) | |
128 | 876 # cached toc |
99 | 877 if tocSize%tocPageSize>0: |
878 tocPages=tocSize/tocPageSize+1 | |
879 else: | |
880 tocPages=tocSize/tocPageSize | |
128 | 881 pageinfo['tocPN'] = min (tocPages,toc) |
95 | 882 pageinfo['searchPN'] =self.REQUEST.get('searchPN','1') |
112 | 883 pageinfo['sn'] =self.REQUEST.get('sn','') |
22 | 884 return pageinfo |
885 | |
128 | 886 def changeDocumentViewer(self,title="",digilibBaseUrl=None,thumbrows=2,thumbcols=5,authgroups='mpiwg',RESPONSE=None): |
22 | 887 """init document viewer""" |
888 self.title=title | |
889 self.digilibBaseUrl = digilibBaseUrl | |
25 | 890 self.thumbrows = thumbrows |
891 self.thumbcols = thumbcols | |
32 | 892 self.authgroups = [s.strip().lower() for s in authgroups.split(',')] |
22 | 893 if RESPONSE is not None: |
894 RESPONSE.redirect('manage_main') | |
0 | 895 |
896 def manage_AddDocumentViewerForm(self): | |
897 """add the viewer form""" | |
22 | 898 pt=PageTemplateFile('zpt/addDocumentViewer', globals()).__of__(self) |
0 | 899 return pt() |
900 | |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
901 def manage_AddDocumentViewer(self,id,imageScalerUrl="",textServerName="",title="",RESPONSE=None): |
0 | 902 """add the viewer""" |
84
a6e4f9b6729a
first version with new full-text infrastructure and slightly changed templates
casties
parents:
83
diff
changeset
|
903 newObj=documentViewer(id,imageScalerUrl=imageScalerUrl,title=title,textServerName=textServerName) |
0 | 904 self._setObject(id,newObj) |
905 | |
906 if RESPONSE is not None: | |
907 RESPONSE.redirect('manage_main') | |
22 | 908 |
909 ## DocumentViewerTemplate class | |
910 class DocumentViewerTemplate(ZopePageTemplate): | |
911 """Template for document viewer""" | |
912 meta_type="DocumentViewer Template" | |
913 | |
914 | |
915 def manage_addDocumentViewerTemplateForm(self): | |
916 """Form for adding""" | |
917 pt=PageTemplateFile('zpt/addDocumentViewerTemplate', globals()).__of__(self) | |
918 return pt() | |
919 | |
920 def manage_addDocumentViewerTemplate(self, id='viewer_main', title=None, text=None, | |
921 REQUEST=None, submit=None): | |
922 "Add a Page Template with optional file content." | |
923 | |
924 self._setObject(id, DocumentViewerTemplate(id)) | |
925 ob = getattr(self, id) | |
53
f4e0af8c281d
NEW - # 44: ECHO - vollst?ndige bibliographische Angabe
dwinter
parents:
52
diff
changeset
|
926 txt=file(os.path.join(package_home(globals()),'zpt/viewer_main.zpt'),'r').read() |
167 | 927 logging.info("txt %s:"%txt) |
53
f4e0af8c281d
NEW - # 44: ECHO - vollst?ndige bibliographische Angabe
dwinter
parents:
52
diff
changeset
|
928 ob.pt_edit(txt,"text/html") |
22 | 929 if title: |
930 ob.pt_setTitle(title) | |
931 try: | |
932 u = self.DestinationURL() | |
933 except AttributeError: | |
934 u = REQUEST['URL1'] | |
935 | |
936 u = "%s/%s" % (u, urllib.quote(id)) | |
937 REQUEST.RESPONSE.redirect(u+'/manage_main') | |
938 return '' | |
939 | |
940 | |
41 | 941 |