--- trunk/conf/dipl/config.yml 2009/08/19 14:05:30 1273 +++ trunk/conf/dipl/config.yml 2009/08/19 15:59:04 1274 @@ -77,6 +77,7 @@ excel: 'WebPAC::Input::Excel' dbf: 'WebPAC::Input::DBF' isi: 'WebPAC::Input::ISI' + csv: 'WebPAC::Input::CSV' # define delimiters for validation delimiters: - ' ; ' @@ -107,7 +108,7 @@ # path: '/data/FF/citati/proba.xls' # encoding: 'windows-1250' # normalize: -# path: 'conf/isi/isi2xls-citirani.pl' +# path: 'conf/dipl/isi2xls-citirani.pl' # output: # - module: 'Excel' # path: '/data/FF/citiari/citirani.xls' @@ -121,32 +122,32 @@ encoding: 'ISO-8859-1' normalize: # - path: 'conf/normalize/isi/isi2xls.pl' - # - path: 'conf/isi/isi2sorted.pl' - - path: 'conf/isi/tables.pl' + # - path: 'conf/dipl/isi2sorted.pl' + - path: 'conf/dipl/tables.pl' - name: arambasicl type: isi path: '/data/FF/citati/citing/arambasicl.txt' encoding: 'ISO-8859-1' normalize: # - path: 'conf/normalize/isi/isi2xls.pl' - # - path: 'conf/isi/isi2sorted.pl' - - path: 'conf/isi/tables.pl' + # - path: 'conf/dipl/isi2sorted.pl' + - path: 'conf/dipl/tables.pl' - name: bratkod type: isi path: '/data/FF/citati/citing/bratkod.txt' encoding: 'ISO-8859-1' normalize: # - path: 'conf/normalize/isi/isi2xls.pl' - # - path: 'conf/isi/isi2sorted.pl' - - path: 'conf/isi/tables.pl' + # - path: 'conf/dipl/isi2sorted.pl' + - path: 'conf/dipl/tables.pl' - name: buskov type: isi path: '/data/FF/citati/citing/buskov.txt' encoding: 'ISO-8859-1' normalize: # - path: 'conf/normalize/isi/isi2xls.pl' - # - path: 'conf/isi/isi2sorted.pl' - - path: 'conf/isi/tables.pl' + # - path: 'conf/dipl/isi2sorted.pl' + - path: 'conf/dipl/tables.pl' skip: 1 - name: butkovica.txt type: isi @@ -154,8 +155,8 @@ encoding: 'ISO-8859-1' normalize: # - path: 'conf/normalize/isi/isi2xls.pl' - # - path: 'conf/isi/isi2sorted.pl' - - path: 'conf/isi/tables.pl' + # - path: 'conf/dipl/isi2sorted.pl' + - path: 'conf/dipl/tables.pl' skip: 1 - name: corkalobiruskid.txt type: isi @@ -163,8 +164,8 @@ encoding: 'ISO-8859-1' normalize: # - path: 'conf/normalize/isi/isi2xls.pl' - # - path: 'conf/isi/isi2sorted.pl' - - path: 'conf/isi/tables.pl' + # - path: 'conf/dipl/isi2sorted.pl' + - path: 'conf/dipl/tables.pl' skip: 1 - name: hrend type: isi @@ -172,16 +173,16 @@ encoding: 'ISO-8859-1' normalize: # - path: 'conf/normalize/isi/isi2xls.pl' - # - path: 'conf/isi/isi2sorted.pl' - - path: 'conf/isi/tables.pl' + # - path: 'conf/dipl/isi2sorted.pl' + - path: 'conf/dipl/tables.pl' - name: hromatkoi type: isi path: '/data/FF/citati/citing/hromatkoi.txt' encoding: 'ISO-8859-1' normalize: # - path: 'conf/normalize/isi/isi2xls.pl' - # - path: 'conf/isi/isi2sorted.pl' - - path: 'conf/isi/tables.pl' + # - path: 'conf/dipl/isi2sorted.pl' + - path: 'conf/dipl/tables.pl' skip: 1 - name: hudekknezevicj type: isi @@ -189,32 +190,32 @@ encoding: 'ISO-8859-1' normalize: # - path: 'conf/normalize/isi/isi2xls.pl' - # - path: 'conf/isi/isi2sorted.pl' - - path: 'conf/isi/tables.pl' + # - path: 'conf/dipl/isi2sorted.pl' + - path: 'conf/dipl/tables.pl' - name: kaliternalipovcanlj type: isi path: '/data/FF/citati/citing/kaliternalipovcanlj.txt' encoding: 'ISO-8859-1' normalize: # - path: 'conf/normalize/isi/isi2xls.pl' - # - path: 'conf/isi/isi2sorted.pl' - - path: 'conf/isi/tables.pl' + # - path: 'conf/dipl/isi2sorted.pl' + - path: 'conf/dipl/tables.pl' - name: kardumi type: isi path: '/data/FF/citati/citing/kardumi.txt' encoding: 'ISO-8859-1' normalize: # - path: 'conf/normalize/isi/isi2xls.pl' - # - path: 'conf/isi/isi2sorted.pl' - - path: 'conf/isi/tables.pl' + # - path: 'conf/dipl/isi2sorted.pl' + - path: 'conf/dipl/tables.pl' - name: kerestesg type: isi path: '/data/FF/citati/citing/kerestesg.txt' encoding: 'ISO-8859-1' normalize: # - path: 'conf/normalize/isi/isi2xls.pl' - # - path: 'conf/isi/isi2sorted.pl' - - path: 'conf/isi/tables.pl' + # - path: 'conf/dipl/isi2sorted.pl' + - path: 'conf/dipl/tables.pl' skip: 1 - name: knezovicz type: isi @@ -222,8 +223,8 @@ encoding: 'ISO-8859-1' normalize: # - path: 'conf/normalize/isi/isi2xls.pl' - # - path: 'conf/isi/isi2sorted.pl' - - path: 'conf/isi/tables.pl' + # - path: 'conf/dipl/isi2sorted.pl' + - path: 'conf/dipl/tables.pl' skip: 1 - name: kolesaricv type: isi @@ -231,8 +232,8 @@ encoding: 'ISO-8859-1' normalize: # - path: 'conf/normalize/isi/isi2xls.pl' - # - path: 'conf/isi/isi2sorted.pl' - - path: 'conf/isi/tables.pl' + # - path: 'conf/dipl/isi2sorted.pl' + - path: 'conf/dipl/tables.pl' skip: 1 - name: kuterovacjagodicg type: isi @@ -240,40 +241,40 @@ encoding: 'ISO-8859-1' normalize: # - path: 'conf/normalize/isi/isi2xls.pl' - # - path: 'conf/isi/isi2sorted.pl' - - path: 'conf/isi/tables.pl' + # - path: 'conf/dipl/isi2sorted.pl' + - path: 'conf/dipl/tables.pl' - name: lackovicgrgink type: isi path: '/data/FF/citati/citing/lackovicgrgink.txt' encoding: 'ISO-8859-1' normalize: # - path: 'conf/normalize/isi/isi2xls.pl' - # - path: 'conf/isi/isi2sorted.pl' - - path: 'conf/isi/tables.pl' + # - path: 'conf/dipl/isi2sorted.pl' + - path: 'conf/dipl/tables.pl' - name: lamzaposavecv type: isi path: '/data/FF/citati/citing/lamzaposavecv.txt' encoding: 'ISO-8859-1' normalize: # - path: 'conf/normalize/isi/isi2xls.pl' - # - path: 'conf/isi/isi2sorted.pl' - - path: 'conf/isi/tables.pl' + # - path: 'conf/dipl/isi2sorted.pl' + - path: 'conf/dipl/tables.pl' - name: manenicai type: isi path: '/data/FF/citati/citing/manenicai.txt' encoding: 'ISO-8859-1' normalize: # - path: 'conf/normalize/isi/isi2xls.pl' - # - path: 'conf/isi/isi2sorted.pl' - - path: 'conf/isi/tables.pl' + # - path: 'conf/dipl/isi2sorted.pl' + - path: 'conf/dipl/tables.pl' - name: maslicsersicd type: isi path: '/data/FF/citati/citing/maslicsersicd.txt' encoding: 'ISO-8859-1' normalize: # - path: 'conf/normalize/isi/isi2xls.pl' - # - path: 'conf/isi/isi2sorted.pl' - - path: 'conf/isi/tables.pl' + # - path: 'conf/dipl/isi2sorted.pl' + - path: 'conf/dipl/tables.pl' skip: 1 - name: radosevicvidacekb type: isi @@ -281,24 +282,24 @@ encoding: 'ISO-8859-1' normalize: # - path: 'conf/normalize/isi/isi2xls.pl' - # - path: 'conf/isi/isi2sorted.pl' - - path: 'conf/isi/tables.pl' + # - path: 'conf/dipl/isi2sorted.pl' + - path: 'conf/dipl/tables.pl' - name: sverkob type: isi path: '/data/FF/citati/citing/sverkob.txt' encoding: 'ISO-8859-1' normalize: # - path: 'conf/normalize/isi/isi2xls.pl' - # - path: 'conf/isi/isi2sorted.pl' - - path: 'conf/isi/tables.pl' + # - path: 'conf/dipl/isi2sorted.pl' + - path: 'conf/dipl/tables.pl' - name: galicz type: isi path: '/data/FF/citati/citing/galicz.txt' encoding: 'ISO-8859-1' normalize: # - path: 'conf/normalize/isi/isi2xls.pl' - # - path: 'conf/isi/isi2sorted.pl' - - path: 'conf/isi/tables.pl' + # - path: 'conf/dipl/isi2sorted.pl' + - path: 'conf/dipl/tables.pl' skip: 1 - name: tadinacm type: isi @@ -306,8 +307,8 @@ encoding: 'ISO-8859-1' normalize: # - path: 'conf/normalize/isi/isi2xls.pl' - # - path: 'conf/isi/isi2sorted.pl' - - path: 'conf/isi/tables.pl' + # - path: 'conf/dipl/isi2sorted.pl' + - path: 'conf/dipl/tables.pl' - name: radovi type: excel path: '/data/FF/citati/citirani_radovi.xls' @@ -315,14 +316,27 @@ # encoding: 'ISO-8859-1' normalize: # - path: 'conf/normalize/isi/isi2xls.pl' - # - path: 'conf/isi/isi2sorted.pl' - - path: 'conf/isi/citirani-radovi.pl' + # - path: 'conf/dipl/isi2sorted.pl' + - path: 'conf/dipl/citirani-radovi.pl' - name: cropsy type: isi path: '/data/FF/citati/croatia-psychol.txt' normalize: - - path: 'conf/isi/tables-cropsy.pl' + - path: 'conf/dipl/tables-cropsy.pl' skip: 0 + - name: jcr + type: csv + path: "/data/FF/citati/jcr_recs.txt" + normalize: + - path: 'conf/dipl/tables-jcr.pl' + - name: izbori + type: excel + path: "/data/FF/citati/izbori.xls" + worksheet: podaci + from: 1 + to: 886 + normalize: + - path: 'conf/dipl/tables-izbori.pl' output: # - module: 'Excel' # path: '/data/FF/citati/data.xls' @@ -332,5 +346,5 @@ # url: 'http://193.198.212.57:5984' - module: 'DBI' dsn: 'dbi:Pg:dbname=dipl' - schema: 'conf/isi/schema.sql' + schema: 'conf/dipl/schema.sql' # table: 'citirani'