Sack/conf/llin.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)
# disable indexer
use_indexer: ''
#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
  #
  # if using hyperestraier-native use following:
  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'
    gutenberg: 'WebPAC::Input::Gutenberg'
    dbf: 'WebPAC::Input::DBF'
  # 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: ps
        credit: 10000
      - to: kk
        credit: 10000
      - to: jzav
        credit: 10000

  # site with alternative databases
  'hr':
    name: 'Primjer za hrvatsko suèelje'
    links:
      - to: ffps
        credit: 10000
      - to: ffkk
        credit: 5000
      - to: fffo
        credit: 10000

  # Psihologija
  'ffps':
    # all variables here will be available as config variables in template
    name: 'Knji¾nica psihologije'
    code: 'ps'
  
    # define input source
    input:
      # this will create unique name with name of database and mfn
      - name: libri
        # isis
        type: isis
    
        # full path to database
        path: '/data/isis_data/ffps-libri/LIBRI'
    
        # encoding of character set in isis data
        encoding: 'cp852'
    
        # limit number of records to read from database
        #limit: 100
    
        # define normalisation for that source
        normalize:
          # which tag to use in normalize xml for data?
          - tag: 'isis'
            # path to normalization xml
            path: 'conf/normalize/common.pl'
            #path: 'conf/normalize/isis_ffzg.yml'
          # convert input to MARC21
#          - tag: 'marc21'
#            #
#            path: 'conf/normalize/mapping.pl'
#            output: 'out/marc/ps-libri.marc'
  
      # another input database
      - name: peri
        type: isis
        path: '/data/isis_data/ffps-peri/PERI'
        encoding: 'cp852'
        #limit: 100
        normalize:
          tag: 'isis'
          path: 'conf/normalize/common.pl'

  # Komparativna
  ffkk:
    name: 'Komparativna knji¾evnost, Filozofski fakultet u Zagrebu'
  
    input:
      - name: libri
        type: isis
        path: '/data/isis_data/ffkk-libri/LIBRI'
        encoding: 'cp852'
        modify_file: 'conf/modify/common.pl'
        normalize:
#          path: 'conf/normalize/common.pl'
          path: 'conf/normalize/mapping.pl'
      - name: peri
        type: isis
        path: '/data/isis_data/ffkk-peri/PERI'
        encoding: 'cp852'
        modify_file: 'conf/modify/common.pl'
        normalize:
#          path: 'conf/normalize/common.pl'
          path: 'conf/normalize/mapping.pl'
      - name: arti
        type: isis
        path: '/data/isis_data/ffkk-arti/ARTI'
        encoding: 'cp852'
        modify_file: 'conf/modify/ffkk-arti.pl'
        normalize:
#          - path: 'conf/normalize/common.pl'
          - path: 'conf/normalize/mapping.pl'
          - path: 'conf/lookup/ffkk-arti.pl'
  
  fffo:
    name: 'Filozofski Fakultet - Fonetika'
  
    input:
      name: libri
      type: isis
      path: '/data/isis_data/fffo-libri/LIBRI'
      encoding: 'cp852'
      #limit: 100
      normalize:
        tag: 'isis'
        path: 'conf/normalize/common.pl'

  jzav:
    name: 'Jadranski zavod'
    input:
      name: knjige
      type: isis
      path: '/data/isis_data/jzav/BIB'
      encoding: 'cp852'
      #limit: 100
      normalize:
        tag: 'isis'
        path: 'conf/normalize/common.pl'
  
  efzg:
    name: 'Ekonomski fakutet u Zagrebu'
    input:
      name: knjige
      type: marc
      path: '/data/isis_data/efzg/unimarc.iso'
      encoding: 'cp852'
      # CroList recoding pairs to fix encoding problems
      recode: 'ÏC È Ïc è ÂC Æ Âc æ ò ð ÏS © Ïs ¹ ÏZ ® Ïz ¾'
      #limit: 100
      normalize:
        tag: 'isis'
        path: 'conf/normalize/common.pl'
  
  ffsfb:
    name: 'Slavenska filologija (B), Filozofski fakulteti u Zagrebu'
    input:
      name: libri
      type: isis
      path: '/data/isis_data/ffsfb-libri/'
      encoding: 'cp852'
      #limit: 10
      modify_records:
        200:
          '^a':
            ' ; ': '^k'
          '^d':
            '^e': ' : '
          '^e':
            '^d': ' = '
          '*':
            '^g': ' ; '
        210:
          '*':
            '^c': '^b'
            '^d': '^c'
            '. (': '. (^e'
            ' : ': ' :^f' 
        700:
          '^a':
            '^b': ', '
        701:
          '^a':
            '^b': ', '
        702:
          '^a':
            '^b': ', '
      normalize:
        path: 'conf/normalize/common.pl'
  fflo:
    name: 'Lingvistika, Filozofski fakultet u Zagrebu'
    input:
      name: libri
      type: isis
      path: '/data/isis_data/fflo-libri/LIBRI'
      encoding: 'cp852'
      #limit: 10
      modify_records:
        200:
          '^d':
            '^e': ' : '
          '^e':
            '^d': ' = '
          '*':
            '^g': ' ; '
            '^c': '. '
        210:
          '*':
            '^c': '^b'
            '^d': '^c'
        700:
          '^a':
            '^b': ', '
        701:
          '^a':
            '^b': ', '
        702:
          '^a':
            '^b': ', '
      normalize:
        path: 'conf/normalize/common.pl'

  gutenberg:
    name: 'Project Gutenberg archive'
    input:
      name: archive
      type: gutenberg
      path: 't/data/gutenberg-small.rdf'
      normalize:
        path: 'conf/normalize/common.pl'

  sand:
    name: 'SAND'
    input:
      name: casopisi
      type: dbf
      path: 't/data/cas2000.dbf'
      normalize:
        path: 'conf/normalize/common.pl'
#      mapping_path: 'conf/input/dbf/cas2000.yml'

  exhibit:
    name: 'SMILE Exhibit based ouptput for WebPAC'
    input:
      name: 'FFZG - Psihologija'
      type: marc
      path: 'out/marc/ffsf-peri.marc'
      encoding: 'cp852'
      normalize:
        path: 'conf/normalize/json.pl'
    output:
      module: 'JSON'
      path: 'out/exhibit/ps.js'

  webpacus:
    name: 'Webpacus'
    input:
      name: 'foobar'
      type: marc
#      path: 'out/marc/ffsf-peri.marc'
#      path: 'out/marc/ffkk-libri.marc'
      path: 'out/marc/ffiz-libri.marc'
      encoding: 'cp852'
      normalize:
        path: 'conf/normalize/webpacus.pl'
#        path: 'conf/normalize/webpacus-unimarc.pl'
    output:
      - module: 'KinoSearch'
        path: 'var/kinosearch'
      - module: 'Sorted'
        path: 'var/sorted'
      - module: 'Webpacus'
        path: '/data/Webpacus2'

  hidra:
    name: 'HIDRA'
    input:
      - name: bib
        type: isis
        path: '/data/hidra/BIB'
        encoding: 'cp852'
        #modify_file: 'conf/modify/common.pl'
        normalize:
          path: 'conf/normalize/minimal.pl' 

  perlmonks:
    name: 'PerlMonks'
    input:
      - name: xml
        module: 'WebPAC::Input::XML'
        path: '/home/dpavlin/monk-search/xml-dump'
        mungle: 'conf/mungle/perlmonks-xml.pl'
        normalize:
          path: 'conf/normalize/perlmonks.pl'
    output:
      - module: 'KinoSearch'
        path: 'var/kinosearch'
      - module: 'Sorted'
        path: 'var/sorted'
      - module: 'Webpacus'
        path: '/data/Webpacus2'

  ufo:
    name: 'UFO Reports 1998-2006 in the UK'
    input:
      - name: pdf
        module: 'WebPAC::Input::PDF'
        path: '/home/dpavlin/x/ufo/UFOReport1998.pdf'
#          - '/home/dpavlin/x/ufo/UFOReport1998.pdf'
#          - '/home/dpavlin/x/ufo/UFOReport1999.pdf'
#          - '/home/dpavlin/x/ufo/UFOReport2000.pdf'
#          - '/home/dpavlin/x/ufo/UFOReport2001.pdf'
#          - '/home/dpavlin/x/ufo/UFOReports2002WholeoftheUK.pdf'
#          - '/home/dpavlin/x/ufo/UFOReports2003WholeoftheUK.pdf'
#          - '/home/dpavlin/x/ufo/UFOReports2004WholeoftheUK.pdf'
#          - '/home/dpavlin/x/ufo/UFOReports2005WholeoftheUK.pdf'
#          - '/home/dpavlin/x/ufo/UFOReports2006WholeoftheUK.pdf'
        normalize:
          path: 'conf/normalize/ufo.pl'
    output:
      - module: 'Jifty'
        path: '/data/Webpacus2'
        model: 'Webpacus::Model::UFO'

1	dpavlin	872	--- #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			# disable indexer
13			use_indexer: ''
14			#use_indexer: 'hyperestraier'
15			#use_indexer: 'hyperestraier-native'
16			#use_indexer: 'kinosearch'
17
18			# configuration for Hyper Estraier full text search engine
19			hyperestraier:
20			#url: 'http://localhost:1978/node/webpac2'
21			masterurl: 'http://localhost:1978'
22			defaultnode: 'webpac2'
23			# defaultnode: 'ps'
24			defaultdepth: 1
25			user: 'admin'
26			passwd: 'admin'
27			# don't turn this on! it will spit huge amounts of output
28			#debug: 1
29			#
30			# if using hyperestraier-native use following:
31			path: 'casket/'
32			#
33			# number of results on each page
34			hits_on_page: 10
35			# number of results to fetch for suggestion (it will fold multiple sameones)
36			hits_for_suggest: 20
37			#
38			# options used while indexing
39			#
40			# which tag type to use for search engine (used while indexing)
41			type: 'search'
42			#
43
44			# options for pager
45			pager:
46			# how many pages to show for navigation?
47			max_pages: 20
48
49			# configuration for KinoSearch search engine library
50			kinosearch:
51			index_path: './kinosearch/'
52			database: 'unconfigured database name'
53			label: 'unconfigured database label'
54			encoding: 'iso-8859-2'
55			# clean database before opening? (WARNING: this erases existing database)
56			clean: 1
57			# which field type to index?
58			type: 'search'
59
60			webpac:
61			# default template to use
62			template: 'html_ffzg_results_short.tt'
63			# path to database files
64			db_path: '/data/webpac2/db'
65			# path to templates used by WebPAC::Output
66			template_path: '/data/webpac2/conf/output/tt'
67			# default template for results
68			default_template: 'html_ffzg.tt'
69			# default user editable css file
70			default_css: 'user.css'
71			css_path: 'root/css'
72			# encoding comming from webpac
73			webpac_encoding: 'iso-8859-2'
74			# encoding expected by Catalyst
75			out_encoding: 'utf-8'
76			# define different input formats (types) and perl modules to handle them
77			inputs:
78			isis: 'WebPAC::Input::ISIS'
79			marc: 'WebPAC::Input::MARC'
80			excel: 'WebPAC::Input::Excel'
81			gutenberg: 'WebPAC::Input::Gutenberg'
82			dbf: 'WebPAC::Input::DBF'
83			# define delimiters for validation
84			delimiters:
85			- ' ; '
86			- ' : '
87			- ' / '
88			- ' \. '
89
90			editor:
91			# open this record when opening editor
92			# (it will also be used to deduce default database and input)
93			default_record_uri: 'ps/libri/1'
94
95			# directives after this are used when indexing using core WebPAC modules
96
97			databases:
98			# This is empty database created only in Hyper Estraier to merge
99			# all three databases
100			'webpac2':
101			name: 'Search all'
102			links:
103			- to: ps
104			credit: 10000
105			- to: kk
106			credit: 10000
107			- to: jzav
108			credit: 10000
109
110			# site with alternative databases
111			'hr':
112			name: 'Primjer za hrvatsko suèelje'
113			links:
114			- to: ffps
115			credit: 10000
116			- to: ffkk
117			credit: 5000
118			- to: fffo
119			credit: 10000
120
121			# Psihologija
122			'ffps':
123			# all variables here will be available as config variables in template
124			name: 'Knji¾nica psihologije'
125			code: 'ps'
126
127			# define input source
128			input:
129			# this will create unique name with name of database and mfn
130			- name: libri
131			# isis
132			type: isis
133
134			# full path to database
135			path: '/data/isis_data/ffps-libri/LIBRI'
136
137			# encoding of character set in isis data
138			encoding: 'cp852'
139
140			# limit number of records to read from database
141			#limit: 100
142
143			# define normalisation for that source
144			normalize:
145			# which tag to use in normalize xml for data?
146			- tag: 'isis'
147			# path to normalization xml
148			path: 'conf/normalize/common.pl'
149			#path: 'conf/normalize/isis_ffzg.yml'
150			# convert input to MARC21
151			# - tag: 'marc21'
152			# #
153			# path: 'conf/normalize/mapping.pl'
154			# output: 'out/marc/ps-libri.marc'
155
156			# another input database
157			- name: peri
158			type: isis
159			path: '/data/isis_data/ffps-peri/PERI'
160			encoding: 'cp852'
161			#limit: 100
162			normalize:
163			tag: 'isis'
164			path: 'conf/normalize/common.pl'
165
166			# Komparativna
167			ffkk:
168			name: 'Komparativna knji¾evnost, Filozofski fakultet u Zagrebu'
169
170			input:
171			- name: libri
172			type: isis
173			path: '/data/isis_data/ffkk-libri/LIBRI'
174			encoding: 'cp852'
175			modify_file: 'conf/modify/common.pl'
176			normalize:
177			# path: 'conf/normalize/common.pl'
178			path: 'conf/normalize/mapping.pl'
179			- name: peri
180			type: isis
181			path: '/data/isis_data/ffkk-peri/PERI'
182			encoding: 'cp852'
183			modify_file: 'conf/modify/common.pl'
184			normalize:
185			# path: 'conf/normalize/common.pl'
186			path: 'conf/normalize/mapping.pl'
187			- name: arti
188			type: isis
189			path: '/data/isis_data/ffkk-arti/ARTI'
190			encoding: 'cp852'
191			modify_file: 'conf/modify/ffkk-arti.pl'
192			normalize:
193			# - path: 'conf/normalize/common.pl'
194			- path: 'conf/normalize/mapping.pl'
195			- path: 'conf/lookup/ffkk-arti.pl'
196
197			fffo:
198			name: 'Filozofski Fakultet - Fonetika'
199
200			input:
201			name: libri
202			type: isis
203			path: '/data/isis_data/fffo-libri/LIBRI'
204			encoding: 'cp852'
205			#limit: 100
206			normalize:
207			tag: 'isis'
208			path: 'conf/normalize/common.pl'
209
210			jzav:
211			name: 'Jadranski zavod'
212			input:
213			name: knjige
214			type: isis
215			path: '/data/isis_data/jzav/BIB'
216			encoding: 'cp852'
217			#limit: 100
218			normalize:
219			tag: 'isis'
220			path: 'conf/normalize/common.pl'
221
222			efzg:
223			name: 'Ekonomski fakutet u Zagrebu'
224			input:
225			name: knjige
226			type: marc
227			path: '/data/isis_data/efzg/unimarc.iso'
228			encoding: 'cp852'
229			# CroList recoding pairs to fix encoding problems
230			recode: 'ÏC È Ïc è ÂC Æ Âc æ ò ð ÏS © Ïs ¹ ÏZ ® Ïz ¾'
231			#limit: 100
232			normalize:
233			tag: 'isis'
234			path: 'conf/normalize/common.pl'
235
236			ffsfb:
237			name: 'Slavenska filologija (B), Filozofski fakulteti u Zagrebu'
238			input:
239			name: libri
240			type: isis
241			path: '/data/isis_data/ffsfb-libri/'
242			encoding: 'cp852'
243			#limit: 10
244			modify_records:
245			200:
246			'^a':
247			' ; ': '^k'
248			'^d':
249			'^e': ' : '
250			'^e':
251			'^d': ' = '
252			'*':
253			'^g': ' ; '
254			210:
255			'*':
256			'^c': '^b'
257			'^d': '^c'
258			'. (': '. (^e'
259			' : ': ' :^f'
260			700:
261			'^a':
262			'^b': ', '
263			701:
264			'^a':
265			'^b': ', '
266			702:
267			'^a':
268			'^b': ', '
269			normalize:
270			path: 'conf/normalize/common.pl'
271			fflo:
272			name: 'Lingvistika, Filozofski fakultet u Zagrebu'
273			input:
274			name: libri
275			type: isis
276			path: '/data/isis_data/fflo-libri/LIBRI'
277			encoding: 'cp852'
278			#limit: 10
279			modify_records:
280			200:
281			'^d':
282			'^e': ' : '
283			'^e':
284			'^d': ' = '
285			'*':
286			'^g': ' ; '
287			'^c': '. '
288			210:
289			'*':
290			'^c': '^b'
291			'^d': '^c'
292			700:
293			'^a':
294			'^b': ', '
295			701:
296			'^a':
297			'^b': ', '
298			702:
299			'^a':
300			'^b': ', '
301			normalize:
302			path: 'conf/normalize/common.pl'
303
304			gutenberg:
305			name: 'Project Gutenberg archive'
306			input:
307			name: archive
308			type: gutenberg
309			path: 't/data/gutenberg-small.rdf'
310			normalize:
311			path: 'conf/normalize/common.pl'
312
313			sand:
314	dpavlin	877	name: 'SAND'
315	dpavlin	872	input:
316			name: casopisi
317			type: dbf
318			path: 't/data/cas2000.dbf'
319			normalize:
320			path: 'conf/normalize/common.pl'
321	dpavlin	877	# mapping_path: 'conf/input/dbf/cas2000.yml'
322	dpavlin	882
323			exhibit:
324			name: 'SMILE Exhibit based ouptput for WebPAC'
325			input:
326			name: 'FFZG - Psihologija'
327			type: marc
328	dpavlin	918	path: 'out/marc/ffsf-peri.marc'
329	dpavlin	882	encoding: 'cp852'
330			normalize:
331			path: 'conf/normalize/json.pl'
332			output:
333			module: 'JSON'
334			path: 'out/exhibit/ps.js'
335
336	dpavlin	918	webpacus:
337			name: 'Webpacus'
338			input:
339			name: 'foobar'
340			type: marc
341	dpavlin	932	# path: 'out/marc/ffsf-peri.marc'
342	dpavlin	1007	# path: 'out/marc/ffkk-libri.marc'
343			path: 'out/marc/ffiz-libri.marc'
344	dpavlin	918	encoding: 'cp852'
345			normalize:
346			path: 'conf/normalize/webpacus.pl'
347	dpavlin	1007	# path: 'conf/normalize/webpacus-unimarc.pl'
348	dpavlin	918	output:
349	dpavlin	921	- module: 'KinoSearch'
350			path: 'var/kinosearch'
351			- module: 'Sorted'
352			path: 'var/sorted'
353	dpavlin	932	- module: 'Webpacus'
354			path: '/data/Webpacus2'
355	dpavlin	918
356	dpavlin	965	hidra:
357			name: 'HIDRA'
358			input:
359			- name: bib
360			type: isis
361			path: '/data/hidra/BIB'
362			encoding: 'cp852'
363			#modify_file: 'conf/modify/common.pl'
364			normalize:
365			path: 'conf/normalize/minimal.pl'
366
367	dpavlin	971	perlmonks:
368			name: 'PerlMonks'
369			input:
370			- name: xml
371			module: 'WebPAC::Input::XML'
372			path: '/home/dpavlin/monk-search/xml-dump'
373	dpavlin	991	mungle: 'conf/mungle/perlmonks-xml.pl'
374	dpavlin	971	normalize:
375	dpavlin	991	path: 'conf/normalize/perlmonks.pl'
376	dpavlin	971	output:
377			- module: 'KinoSearch'
378			path: 'var/kinosearch'
379	dpavlin	1009	- module: 'Sorted'
380			path: 'var/sorted'
381	dpavlin	971	- module: 'Webpacus'
382			path: '/data/Webpacus2'
383
384	dpavlin	1060	ufo:
385			name: 'UFO Reports 1998-2006 in the UK'
386			input:
387			- name: pdf
388			module: 'WebPAC::Input::PDF'
389			path: '/home/dpavlin/x/ufo/UFOReport1998.pdf'
390			# - '/home/dpavlin/x/ufo/UFOReport1998.pdf'
391			# - '/home/dpavlin/x/ufo/UFOReport1999.pdf'
392			# - '/home/dpavlin/x/ufo/UFOReport2000.pdf'
393			# - '/home/dpavlin/x/ufo/UFOReport2001.pdf'
394			# - '/home/dpavlin/x/ufo/UFOReports2002WholeoftheUK.pdf'
395			# - '/home/dpavlin/x/ufo/UFOReports2003WholeoftheUK.pdf'
396			# - '/home/dpavlin/x/ufo/UFOReports2004WholeoftheUK.pdf'
397			# - '/home/dpavlin/x/ufo/UFOReports2005WholeoftheUK.pdf'
398			# - '/home/dpavlin/x/ufo/UFOReports2006WholeoftheUK.pdf'
399			normalize:
400			path: 'conf/normalize/ufo.pl'
401			output:
402			- module: 'Jifty'
403			path: '/data/Webpacus2'
404			model: 'Webpacus::Model::UFO'
405