trunk/conf/mjesec-test.yml

--- #YAML:1.0
# DO NOT USE TABS FOR INDENTATION OR label/value SEPARATION!!!

# encoding of this configuration file
config_encoding: 'ISO-8859-2'
# encoding in Catalyst.
catalyst_encoding: 'UTF-8'
# relative path to sites templates under Catalyst root
sites_root: 'sites'

# which indexing engine to use? (hyperestraier is default)
#use_indexer: 'hyperestraier'
use_indexer: 'hyperestraier-native'
#use_indexer: 'kinosearch'

# configuration for Hyper Estraier full text search engine
hyperestraier:
  #url: 'http://localhost:1978/node/webpac2'
  masterurl: 'http://localhost:1978'
  defaultnode: 'webpac2'
#  defaultnode: 'ps'
  defaultdepth: 1
  user: 'admin'
  passwd: 'admin'
  # don't turn this on! it will spit huge amounts of output
  #debug: 1
  #
  #
  path: 'casket/'
  # number of results on each page
  hits_on_page: 10
  # number of results to fetch for suggestion (it will fold multiple sameones)
  hits_for_suggest: 20
  #
  # options used while indexing
  #
  # which tag type to use for search engine (used while indexing)
  type: 'search'
  #

# options for pager
pager:
  # how many pages to show for navigation?
  max_pages: 20

# configuration for KinoSearch search engine library
kinosearch:
  index_path: './kinosearch/'
  database: 'unconfigured database name'
  label: 'unconfigured database label'
  encoding: 'iso-8859-2'
  # clean database before opening? (WARNING: this erases existing database)
  clean: 1
  # which field type to index?
  type: 'search'

webpac:
  # default template to use
  template: 'html_ffzg_results_short.tt'
  # path to database files
  db_path: '/data/webpac2/db'
  # path to templates used by WebPAC::Output
  template_path: '/data/webpac2/conf/output/tt'
  # default template for results
  default_template: 'html_ffzg.tt'
  # default user editable css file
  default_css: 'user.css'
  css_path: 'root/css'
  # encoding comming from webpac
  webpac_encoding: 'iso-8859-2'
  # encoding expected by Catalyst
  out_encoding: 'utf-8'
  # define different input formats (types) and perl modules to handle them
  inputs:
    isis: 'WebPAC::Input::ISIS'
    marc: 'WebPAC::Input::MARC'
    excel: 'WebPAC::Input::Excel'
    ovid: 'WebPAC::Input::Ovid'
    csv: 'WebPAC::Input::CSV'
  # define delimiters for validation
  delimiters:
    - ' ; '
    - ' : '
    - ' / '
    - ' = '
editor:
  # open this record when opening editor
  # (it will also be used to deduce default database and input)
  default_record_uri: 'ps/libri/1'

# directives after this are used when indexing using core WebPAC modules

databases:
  # This is empty database created only in Hyper Estraier to merge
  # all three databases
  'webpac2':
    name: 'Search all'
    links:
      - to: ffan
        credit: 10000
      - to: ffar
        credit: 10000
      - to: ffet
        credit: 10000
      - to: fffi 
        credit: 10000
      - to: fffo
        credit: 10000
      - to: ffge
        credit: 10000
      - to: ffhu 
        credit: 10000
      - to: ffiz
        credit: 10000
      - to: ffkf
        credit: 10000
      - to: ffkk 
        credit: 10000
      - to: fflo
        credit: 10000
      - to: fftu
        credit: 10000
      - to: ffpe   
        credit: 10000
      - to: ffpo
        credit: 10000
      - to: ffpu
        credit: 10000
      - to: ffps 
        credit: 10000
      - to: ffro   
        credit: 10000
      - to: ffsk
        credit: 10000
      - to: ffsf
        credit: 10000
      - to: ffso 
        credit: 10000
      - to: ffta 
        credit: 10000
      - to: ffsr 
        credit: 10000
      - to: ffmic 
        credit: 10000
      - to: vusp 
        credit: 10000
      - to: kem 
        credit: 10000
      - to: efzg 
        credit: 10000


#  hazu:
#    name: 'HAZU'
#    input:
#      - name: bib
#        type: isis
#        path: '/data/unizg/hazu/BIB'
#        encoding: 'cp852'
#        modify_file: 'conf/modify/common.pl'
#        normalize:
#          path: 'conf/normalize/minimal.pl'
#
#  sand:
#    name: 'SAND'
#    input:
#      name: casopisi
#      type: dbf
#      path: '/data/unizg/casopisi/sys/cas2000.dbf'
#      normalize:
#        path: 'conf/normalize/common.pl'
##      mapping_path: 'conf/input/dbf/cas2000.yml'
#  
#  hidra:
#    name: 'HIDRA'
#    input:
#      - name: bib
#        type: isis
#        path: '/data/hidra/test/BIB'
#        encoding: 'cp852'
#        modify_file: 'conf/modify/common.pl'
#        normalize:
#          path: 'conf/normalize/minimal.pl'
#
#  efzg:
#    name: 'EFZG'
#    input:
#      - name: crolist
#        type: marc
#        path: '/data/unizg/drustvene/efzg/unimarc.iso'
#        encoding: 'ISO-8859-2'
#        normalize:
#          path: 'conf/normalize/minimal.pl'
#  
#  ipu:
#    name: 'Institut za povijest umjetnosti'
#    input:
#      - name: bibl
#        type: isis
#        path: '/data/unizg/humanistika/ipu/20071115/bibl/BIBL'
#        encoding: 'cp852'
#        # modify_file: 'conf/modify/common.pl'
#        normalize:
#          path: 'conf/normalize/minimal.pl'
#
#  zvonar:
#    name: 'HAZU Odsjek za povijesne znanosti'
#    input:
#      - name: libri
#        type: isis
#        path: '/data/unizg/humanistika/hazu-p/LIBRI'
#        normalize:
#          path: 'conf/normalize/minimal.pl'
#      - name: peri
#        type: isis
#        path: '/data/unizg/humanistika/hazu-p/PERI'
#        normalize:
#          path: 'conf/normalize/minimal.pl'
#
#  stross:
#    name: 'Strossmayerova galerija'
#    input:
#      - name: bib
#        type: isis
#        path: '/data/unizg/stross/BIB'
#        normalize:
#          path: 'conf/normalize/minimal.pl'
#       
#  koncar:
#    name: 'Konèar'
#    input:
#      - name: unimar
#        type: isis
#        path: '/data/tehnika/koncar/UNIMAR'
#        normalize:
#          path: 'conf/normalize/minimal.pl'
#       
#   datumi:
#     name: 'Nevaljali datumi'
#     input:
#       - name: tablica
#         type: excel
#         path: '/data/isis_data/greske/datum_unosa.xls'
#         encoding: 'windows-1250'
#         normalize:
#           - path: 'conf/normalize/common.pl'
#   
#  ffps:
#    name: 'Psihologija, Filozofski fakultet u Zagrebu'
#    code: 'ps'
#    url: 'http://www.knjiznice.ffzg.hr/psihologija'
#    input:
#      - name: libri
#        type: isis
#        path: '/backup/isis_backup/sunce2/isisdata/latest/LIBRI/'
#        encoding: 'cp852'
#        modify_file: 'conf/modify/common.pl'
#        normalize:
#          - path: 'conf/normalize/report-sorted-txt.pl'
#    output:
##      - module: 'Excel'
##        path: '/data/isis_data/ps/reports/libri-nema-jezik.xls'
#      - module: 'Sorted'
#        path: 'out/report/mfn/ps/'
##      - module: 'Excel'
##        path: '/data/isis_data/ps/reports/autori-udk.xls'


#  ffiz:
#     name: 'Informacijske znanosti, Filozofski fakultet u Zagrebu'
#     input:
#       - name: libri
#         type: isis
#         path: '/backup/isis_backup/A129-2/isisdata/latest/LIBRI/LIBRI'
#         encoding: 'cp852'
#         modify_file: 'conf/modify/common.pl'
#         normalize:
#           - path: 'conf/normalize/report-sorted-txt.pl'
#     output:
# #      - module: 'Excel'
# #        path: '/data/isis_data/iz/reports/libri-nema-jezik.xls'
#       - module: 'Sorted'
#         path: 'out/report/mfn/iz/'
# #      - module: 'Excel'
# #        path: '/data/isis_data/iz/reports/autori-udk.xls'
# 
# 
#   casopisi:
#     name: 'Èasopisi u Knji¾nicama FF-a'
#     input:
#       - name: ff
#         type: excel
#         path: '/data/isis_data/CASOPISI/svi-casopisi-FF.xls'
#         ncoding: 'utf-8'
#         normalize:
#           - path: 'conf/normalize/ff-casopisi-xls.pl'
# #          - path: 'conf/normalize/ff-casopisi-marc.pl'
#     output:
#       - module: 'Excel'
#         path: 'out/excel/casopisi-FF-dupli-knjiznice.xls'
# 
#   fflibri:
#     name: 'Filozofski fakultet u Zagrebu'
#     input:
#       - name: marc
#         type: marc
#         path: 'out/marc/fflibri.marc'
#         encoding: 'ISO-8859-2'
#         normalize:
#           - path: 'conf/normalize/ff-libri-dupli.pl'
#     output:
#       - module: 'Excel'
#         path: 'out/excel/knjige-duplo.xls'


  ffso:
    name: 'Sociologija, Filozofski fakultet u Zagrebu'
    input:
      - name: libri
        type: isis
        path: '/backup/isis_backup/A002-3/isisdata/latest/LIBRI/LIBRI'
        encoding: 'cp852'
        #modify_file: 'conf/modify/common.pl'
        normalize:
          - path: 'conf/normalize/minimal.pl'

  items:
    name: 'Items'
    input:
      - name: inv
        type: excel
        path: '/data/isis_data/so/inv.xls'
        encoding: 'windows-1250'
        normalize:
          - path: 'conf/normalize/ff-libri-primjerci.pl'
    output:
      - module: 'Excel'
        path: 'out/excel/items.xls'

  libraryguy:
    name: 'Test set'
    input:
      - name: test
        type: marc
        path: '/data/FF/libraryguy/4marc.mrc'
        encoding: 'UTF-8'
        normalize:
          - path: 'conf/normalize/libraryguy.pl'

  franjo:
    name: 'Franjo podaci'
    input:
      - name: franjo
        type: marc
        path: '/data/FF/franjo/issnXXX.iso'
        encoding: 'cp852'
        modify_file: 'conf/modify/franjo.pl'
        normalize:
          - path: 'conf/normalize/franjo.pl'
    output:
      - module: 'Excel'
        path: '/data/FF/franjo/franjo.xls'

  posudba:
    name: 'Posudba psihologija'
    input:
      - name: psi
        type: excel
        path: '/data/FF/Book1.xls'
        normalize:
          - path: 'conf/normalize/psi-posudba.pl'


  ovid:
    name: 'Ovid export'
    input:
      - name: fulltext
        type: ovid
        path: '/data/FF/ecasopisi/ovid/cites-all.txt'
        normalize:
          - path: 'conf/normalize/ecasopisi/ovid2xls.pl'
    output:
      - module: 'Excel'
        path: '/data/FF/ecasopisi/ovid/ovid-ft.xls'

  casopisi:
    name: 'Casopisi'
    input:
      - name: text
        type: csv
        path: '/data/FF/casopisi/Casopisi-signature.txt'
        normalize:
          - path: 'conf/normalize/ffcasopisi/csv.pl'

1	mglavica	1188	--- #YAML:1.0
2			# DO NOT USE TABS FOR INDENTATION OR label/value SEPARATION!!!
3
4			# encoding of this configuration file
5			config_encoding: 'ISO-8859-2'
6			# encoding in Catalyst.
7			catalyst_encoding: 'UTF-8'
8			# relative path to sites templates under Catalyst root
9			sites_root: 'sites'
10
11			# which indexing engine to use? (hyperestraier is default)
12			#use_indexer: 'hyperestraier'
13			use_indexer: 'hyperestraier-native'
14			#use_indexer: 'kinosearch'
15
16			# configuration for Hyper Estraier full text search engine
17			hyperestraier:
18			#url: 'http://localhost:1978/node/webpac2'
19			masterurl: 'http://localhost:1978'
20			defaultnode: 'webpac2'
21			# defaultnode: 'ps'
22			defaultdepth: 1
23			user: 'admin'
24			passwd: 'admin'
25			# don't turn this on! it will spit huge amounts of output
26			#debug: 1
27			#
28			#
29			path: 'casket/'
30			# number of results on each page
31			hits_on_page: 10
32			# number of results to fetch for suggestion (it will fold multiple sameones)
33			hits_for_suggest: 20
34			#
35			# options used while indexing
36			#
37			# which tag type to use for search engine (used while indexing)
38			type: 'search'
39			#
40
41			# options for pager
42			pager:
43			# how many pages to show for navigation?
44			max_pages: 20
45
46			# configuration for KinoSearch search engine library
47			kinosearch:
48			index_path: './kinosearch/'
49			database: 'unconfigured database name'
50			label: 'unconfigured database label'
51			encoding: 'iso-8859-2'
52			# clean database before opening? (WARNING: this erases existing database)
53			clean: 1
54			# which field type to index?
55			type: 'search'
56
57			webpac:
58			# default template to use
59			template: 'html_ffzg_results_short.tt'
60			# path to database files
61			db_path: '/data/webpac2/db'
62			# path to templates used by WebPAC::Output
63			template_path: '/data/webpac2/conf/output/tt'
64			# default template for results
65			default_template: 'html_ffzg.tt'
66			# default user editable css file
67			default_css: 'user.css'
68			css_path: 'root/css'
69			# encoding comming from webpac
70			webpac_encoding: 'iso-8859-2'
71			# encoding expected by Catalyst
72			out_encoding: 'utf-8'
73			# define different input formats (types) and perl modules to handle them
74			inputs:
75			isis: 'WebPAC::Input::ISIS'
76			marc: 'WebPAC::Input::MARC'
77			excel: 'WebPAC::Input::Excel'
78			ovid: 'WebPAC::Input::Ovid'
79			csv: 'WebPAC::Input::CSV'
80			# define delimiters for validation
81			delimiters:
82			- ' ; '
83			- ' : '
84			- ' / '
85			- ' = '
86			editor:
87			# open this record when opening editor
88			# (it will also be used to deduce default database and input)
89			default_record_uri: 'ps/libri/1'
90
91			# directives after this are used when indexing using core WebPAC modules
92
93			databases:
94			# This is empty database created only in Hyper Estraier to merge
95			# all three databases
96			'webpac2':
97			name: 'Search all'
98			links:
99			- to: ffan
100			credit: 10000
101			- to: ffar
102			credit: 10000
103			- to: ffet
104			credit: 10000
105			- to: fffi
106			credit: 10000
107			- to: fffo
108			credit: 10000
109			- to: ffge
110			credit: 10000
111			- to: ffhu
112			credit: 10000
113			- to: ffiz
114			credit: 10000
115			- to: ffkf
116			credit: 10000
117			- to: ffkk
118			credit: 10000
119			- to: fflo
120			credit: 10000
121			- to: fftu
122			credit: 10000
123			- to: ffpe
124			credit: 10000
125			- to: ffpo
126			credit: 10000
127			- to: ffpu
128			credit: 10000
129			- to: ffps
130			credit: 10000
131			- to: ffro
132			credit: 10000
133			- to: ffsk
134			credit: 10000
135			- to: ffsf
136			credit: 10000
137			- to: ffso
138			credit: 10000
139			- to: ffta
140			credit: 10000
141			- to: ffsr
142			credit: 10000
143			- to: ffmic
144			credit: 10000
145			- to: vusp
146			credit: 10000
147			- to: kem
148			credit: 10000
149			- to: efzg
150			credit: 10000
151
152
153			# hazu:
154			# name: 'HAZU'
155			# input:
156			# - name: bib
157			# type: isis
158			# path: '/data/unizg/hazu/BIB'
159			# encoding: 'cp852'
160			# modify_file: 'conf/modify/common.pl'
161			# normalize:
162			# path: 'conf/normalize/minimal.pl'
163			#
164			# sand:
165			# name: 'SAND'
166			# input:
167			# name: casopisi
168			# type: dbf
169			# path: '/data/unizg/casopisi/sys/cas2000.dbf'
170			# normalize:
171			# path: 'conf/normalize/common.pl'
172			## mapping_path: 'conf/input/dbf/cas2000.yml'
173			#
174			# hidra:
175			# name: 'HIDRA'
176			# input:
177			# - name: bib
178			# type: isis
179			# path: '/data/hidra/test/BIB'
180			# encoding: 'cp852'
181			# modify_file: 'conf/modify/common.pl'
182			# normalize:
183			# path: 'conf/normalize/minimal.pl'
184			#
185			# efzg:
186			# name: 'EFZG'
187			# input:
188			# - name: crolist
189			# type: marc
190			# path: '/data/unizg/drustvene/efzg/unimarc.iso'
191			# encoding: 'ISO-8859-2'
192			# normalize:
193			# path: 'conf/normalize/minimal.pl'
194			#
195			# ipu:
196			# name: 'Institut za povijest umjetnosti'
197			# input:
198			# - name: bibl
199			# type: isis
200			# path: '/data/unizg/humanistika/ipu/20071115/bibl/BIBL'
201			# encoding: 'cp852'
202			# # modify_file: 'conf/modify/common.pl'
203			# normalize:
204			# path: 'conf/normalize/minimal.pl'
205			#
206			# zvonar:
207			# name: 'HAZU Odsjek za povijesne znanosti'
208			# input:
209			# - name: libri
210			# type: isis
211			# path: '/data/unizg/humanistika/hazu-p/LIBRI'
212			# normalize:
213			# path: 'conf/normalize/minimal.pl'
214			# - name: peri
215			# type: isis
216			# path: '/data/unizg/humanistika/hazu-p/PERI'
217			# normalize:
218			# path: 'conf/normalize/minimal.pl'
219			#
220			# stross:
221			# name: 'Strossmayerova galerija'
222			# input:
223			# - name: bib
224			# type: isis
225			# path: '/data/unizg/stross/BIB'
226			# normalize:
227			# path: 'conf/normalize/minimal.pl'
228			#
229			# koncar:
230			# name: 'Konèar'
231			# input:
232			# - name: unimar
233			# type: isis
234			# path: '/data/tehnika/koncar/UNIMAR'
235			# normalize:
236			# path: 'conf/normalize/minimal.pl'
237			#
238			# datumi:
239			# name: 'Nevaljali datumi'
240			# input:
241			# - name: tablica
242			# type: excel
243			# path: '/data/isis_data/greske/datum_unosa.xls'
244			# encoding: 'windows-1250'
245			# normalize:
246			# - path: 'conf/normalize/common.pl'
247			#
248			# ffps:
249			# name: 'Psihologija, Filozofski fakultet u Zagrebu'
250			# code: 'ps'
251			# url: 'http://www.knjiznice.ffzg.hr/psihologija'
252			# input:
253			# - name: libri
254			# type: isis
255			# path: '/backup/isis_backup/sunce2/isisdata/latest/LIBRI/'
256			# encoding: 'cp852'
257			# modify_file: 'conf/modify/common.pl'
258			# normalize:
259			# - path: 'conf/normalize/report-sorted-txt.pl'
260			# output:
261			## - module: 'Excel'
262			## path: '/data/isis_data/ps/reports/libri-nema-jezik.xls'
263			# - module: 'Sorted'
264			# path: 'out/report/mfn/ps/'
265			## - module: 'Excel'
266			## path: '/data/isis_data/ps/reports/autori-udk.xls'
267
268
269			# ffiz:
270			# name: 'Informacijske znanosti, Filozofski fakultet u Zagrebu'
271			# input:
272			# - name: libri
273			# type: isis
274			# path: '/backup/isis_backup/A129-2/isisdata/latest/LIBRI/LIBRI'
275			# encoding: 'cp852'
276			# modify_file: 'conf/modify/common.pl'
277			# normalize:
278			# - path: 'conf/normalize/report-sorted-txt.pl'
279			# output:
280			# # - module: 'Excel'
281			# # path: '/data/isis_data/iz/reports/libri-nema-jezik.xls'
282			# - module: 'Sorted'
283			# path: 'out/report/mfn/iz/'
284			# # - module: 'Excel'
285			# # path: '/data/isis_data/iz/reports/autori-udk.xls'
286			#
287			#
288			# casopisi:
289			# name: 'Èasopisi u Knji¾nicama FF-a'
290			# input:
291			# - name: ff
292			# type: excel
293			# path: '/data/isis_data/CASOPISI/svi-casopisi-FF.xls'
294			# ncoding: 'utf-8'
295			# normalize:
296			# - path: 'conf/normalize/ff-casopisi-xls.pl'
297			# # - path: 'conf/normalize/ff-casopisi-marc.pl'
298			# output:
299			# - module: 'Excel'
300			# path: 'out/excel/casopisi-FF-dupli-knjiznice.xls'
301			#
302			# fflibri:
303			# name: 'Filozofski fakultet u Zagrebu'
304			# input:
305			# - name: marc
306			# type: marc
307			# path: 'out/marc/fflibri.marc'
308			# encoding: 'ISO-8859-2'
309			# normalize:
310			# - path: 'conf/normalize/ff-libri-dupli.pl'
311			# output:
312			# - module: 'Excel'
313			# path: 'out/excel/knjige-duplo.xls'
314
315
316			ffso:
317			name: 'Sociologija, Filozofski fakultet u Zagrebu'
318			input:
319			- name: libri
320			type: isis
321			path: '/backup/isis_backup/A002-3/isisdata/latest/LIBRI/LIBRI'
322			encoding: 'cp852'
323			#modify_file: 'conf/modify/common.pl'
324			normalize:
325			- path: 'conf/normalize/minimal.pl'
326
327			items:
328			name: 'Items'
329			input:
330			- name: inv
331			type: excel
332			path: '/data/isis_data/so/inv.xls'
333			encoding: 'windows-1250'
334			normalize:
335			- path: 'conf/normalize/ff-libri-primjerci.pl'
336			output:
337			- module: 'Excel'
338			path: 'out/excel/items.xls'
339
340			libraryguy:
341			name: 'Test set'
342			input:
343			- name: test
344			type: marc
345			path: '/data/FF/libraryguy/4marc.mrc'
346			encoding: 'UTF-8'
347			normalize:
348			- path: 'conf/normalize/libraryguy.pl'
349
350			franjo:
351			name: 'Franjo podaci'
352			input:
353			- name: franjo
354			type: marc
355			path: '/data/FF/franjo/issnXXX.iso'
356			encoding: 'cp852'
357			modify_file: 'conf/modify/franjo.pl'
358			normalize:
359			- path: 'conf/normalize/franjo.pl'
360			output:
361			- module: 'Excel'
362			path: '/data/FF/franjo/franjo.xls'
363
364			posudba:
365			name: 'Posudba psihologija'
366			input:
367			- name: psi
368			type: excel
369			path: '/data/FF/Book1.xls'
370			normalize:
371			- path: 'conf/normalize/psi-posudba.pl'
372
373
374			ovid:
375			name: 'Ovid export'
376			input:
377			- name: fulltext
378			type: ovid
379			path: '/data/FF/ecasopisi/ovid/cites-all.txt'
380			normalize:
381			- path: 'conf/normalize/ecasopisi/ovid2xls.pl'
382			output:
383			- module: 'Excel'
384			path: '/data/FF/ecasopisi/ovid/ovid-ft.xls'
385
386			casopisi:
387			name: 'Casopisi'
388			input:
389			- name: text
390			type: csv
391			path: '/data/FF/casopisi/Casopisi-signature.txt'
392			normalize:
393			- path: 'conf/normalize/ffcasopisi/csv.pl'
394