Export employee stats
Former-commit-id: cffd8ae4e74fc751c71cdaf62195ce99303268a2
--- a/.gitmodules
+++ b/.gitmodules
@@ -4,4 +4,16 @@
[submodule "couchdb/settee"]
path = couchdb/settee
url = https://github.com/inadarei/settee.git
+[submodule "lib/springy"]
+ path = lib/springy
+ url = https://github.com/dhotson/springy.git
+[submodule "lib/php-diff"]
+ path = lib/php-diff
+ url = https://github.com/chrisboulton/php-diff.git
+[submodule "lib/Requests"]
+ path = lib/Requests
+ url = https://github.com/rmccue/Requests.git
+[submodule "javascripts/flotr2"]
+ path = javascripts/flotr2
+ url = https://github.com/HumbleSoftware/Flotr2.git
--- a/about.php
+++ b/about.php
@@ -13,48 +13,51 @@
Prometheus is the agent which polls agency websites to assess compliance.
<h2> Open everything </h2>
-all documents released CC-BY 3 AU
+All documents released CC-BY 3 AU
Open source git @
<h2>Organisational Data Sources</h2>
http://www.comlaw.gov.au/Browse/Results/ByTitle/AdministrativeArrangementsOrders/Current/Ad/0 defines departments
-Agencies can be found in the Schedule to an Appropriation Bill (budget), Schedule to FMA Regulations and/or Public Service Act.
+Agencies can be found in the Schedule to an Appropriation Bill (budget), Schedule to FMA Regulations and/or Public Service Act.<br>
-http://www.finance.gov.au/publications/flipchart/docs/FMACACFlipchart.pdf summarises these. view-source:https://www.tenders.gov.au/?event=public.advancedsearch.home is great for the suspended/active status
+http://www.finance.gov.au/publications/flipchart/docs/FMACACFlipchart.pdf summarises these. view-source:https://www.tenders.gov.au/?event=public.advancedsearch.home is great for the suspended/active status<br>
+
+Fraud in gov depts by Fairfax Media http://www.smh.com.au/national/public-service-keeps-fraud-cases-private-20110923-1kpdr.html
When defining the hierachy, this system is designed towards monitoring accountablity. Thus large agencies that have registered their own ABN
and have their own accountablity mechanisms/website receive a seperate record as a child of their department.
-Some small agencies will choose to simply rely on their parent department's accountablity measures.
+Some small agencies will choose to simply rely on their parent department's accountablity measures.<br>
This flows through to organisation name and other/past names. A department that completely accounts for an agency will list that agency as an other child name.
As agencies themselves shift between departments, there may be scope for providing time ranges but typically the newest hierarchy will be the one recorded.
-A department/agency name will be the newest active name assigned to that ABN.
+A department/agency name will be the newest active name assigned to that ABN.<br>
ABN information is derived from the ABR. This is the definitive umpire about which former name should be linked to which current name.
For example "Department of Transport and Regional Services" became "Department of Infrastructure, Transport, Regional Development and Local Government" (same ABN)
however it later split into "Department of Infrastructure and Transport" (same ABN)
-and "Department of Regional Australia, Regional Development and Local Government" (new ABN).
+and "Department of Regional Australia, Regional Development and Local Government" (new ABN).<br>
Statistical information from http://www.apsc.gov.au/stateoftheservice/1011/statsbulletin/section1.html#t2total https://www.apsedii.gov.au/apsedii/CustomQueryx33.shtml
-and individual annual reports.
+and individual annual reports.<br>
-Webpage Assessment
+<h2>Webpage Assessment</h2>
Much due care has been put into correctly recording disclosure URLs. Typically the "About", "Corporate", "Publications" and "Sitemap" sections are checked at the very least.
-Occasionally it is nessicary to use a site or Google search. In several rare cases, there is a secret "Disclosure" navigation menu you can find if you find one of the mandatory publishing obligations in that category (seriously).
-Some rules about leniency:
- An empty FOI disclosure log counts, a page outlining what the FOI Act is does not.
- A disclosure log in PDF or Word format counts :(
- An empty File/Record list counts (although that's very minimalistic that you have no files, electronic or paper)
- Only a current information publication scheme page counts, not a s.9 FOI Act page or an organisation chart.
- If there isn't a page easily listing all current and past Annual Reports, the most current one (html, pdf) counts.
- Consultancy contracts might not need it's own webpage (if in Annual Report), grants/appointments might not apply to all organisations but Legal Services Expenditure (and all other obligations) does need a webpage.
+Occasionally it is nessicary to use a site or Google search. In several rare cases, there is a secret "Disclosure" navigation menu you can find if you find one of the mandatory publishing obligations in that category (seriously).<br>
+Some rules about leniency:<br>
+<ul>
+ <li>An empty FOI disclosure log counts, a page outlining what the FOI Act is does not.</li>
+ <li>A disclosure log in PDF or Word format counts :(</li>
+ <li>An empty File/Record list counts (although that's very minimalistic that you have no files, electronic or paper)</li>
+ <li>Only a current information publication scheme page counts, not a s.9 FOI Act page or an organisation chart.</li>
+ <li>If there isn't a page easily listing all current and past Annual Reports, the most current one (html, pdf) counts.</li>
+ <li>Consultancy contracts might not need it's own webpage (if in Annual Report), grants/appointments might not apply to all organisations but Legal Services Expenditure (and all other obligations) does need a webpage. </li>
<h2>Open Government Scoring</h2>
-+1 point for every true Has... attribute
--1 point for every false Has... (ie. Has Not) attribute
++1 point for every true Has... attribute<br>
+-1 point for every false Has... (ie. Has Not) attribute</br>
-Don't like this? Make your own score, suggest a better scoring mechanism.
+Don't like this? Make your own score, suggest a better scoring mechanism.</br>
<?php
include_footer();
--- /dev/null
+++ b/admin/0203apsemployees.csv
@@ -1,1 +1,75 @@
+"Agriculture, Fisheries & Forestry",3530
+Attorney-Generals,708
+Administrative Appeals Tribunal,108
+AUSTRAC,60
+Australian Crime Commission,225
+Australian Customs Service,4934
+Australian Protective Service,1124
+CrimTrac Agency,34
+Family Court of Australia,574
+Federal Court of Australia,221
+Federal Magistrates Service,46
+Federal Privacy Commissioner,32
+HREOC,79
+ITSA,227
+National Native Title Tribunal,241
+Office of the D.P.P.,403
+Office of Film & Literature Class,39
+Office of the Parl. Counsel,45
+Communications IT & the Arts,776
+Aust. Broadcasting Authority,114
+Aust. Communications Authority,391
+Aust. National Maritime Museum,90
+National Archives of Australia,363
+National Library of Australia,443
+National Museum of Australia,185
+NOIE,160
+Defence,18231
+Defence Housing Authority,661
+DEST,1464
+Australian Research Council,50
+DEWR,2055
+EOWA,11
+Australian Industrial Registry,185
+Comcare,300
+NOHSC,89
+Environment and Heritage,1271
+Australian Greenhouse Office,167
+Bureau of Meteorology,1403
+GBRMPA,153
+National Oceans Office,31
+Family and Community Services,4832
+Aust. Institute of Family Studies,39
+Centrelink,25147
+Finance and Administration,771
+Australian Electoral Comm.,729
+ComSuper,338
+Foreign Affairs & Trade,2763
+ACIAR,44
+Health & Ageing,5137
+Aust. Institute of Health & Welfare,152
+ARPANSA,126
+FSANZ,110
+PSR,25
+DIMIA,4089
+Aboriginal Hostels Ltd.,341
+ATSIC,1072
+AIATSIS,62
+Torres Strait Regional Authority,30
+Industry Tourism & Resources,2691
+Prime Minister & Cabinet,344
+ANAO,278
+Commonwealth Ombudsman,78
+Office of National Assessments,47
+APS Commission,164
+Transport & Regional Services,945
+National Capital Authority,76
+Treasury,906
+Australian Bureau of Statistics,3024
+ACCC,446
+ASIC,1048
+Australian Taxation Office,20238
+Productivity Commission,197
+Veterans Affairs,2335
+Australian War Memorial,215
--- /dev/null
+++ b/admin/0304apsemployees.csv
@@ -1,1 +1,78 @@
+"Agriculture, Fisheries and Forestry",3928
+Attorney-General's,781
+Administrative Appeals Tribunal,117
+AUSTRAC,61
+Australian Crime Commission,315
+Australian Customs Service,4736
+Australian Protective Service,1234
+CrimTrac Agency,44
+Family Court of Australia,608
+Federal Court of Australia,238
+Federal Magistrates Service,56
+Federal Privacy Commissioner,29
+HREOC,84
+ITSA,240
+National Native Title Tribunal,243
+Office of the D.P.P.,416
+Office of Film & Literature Class.,37
+Office of Parliamentary Counsel,46
+Communications IT & the Arts,607
+Aust. Broadcasting Authority,114
+Aust. Communications Authority,407
+Australian Film Commission,176
+AGIMO,95
+Aust. National Maritime Museum,93
+National Archives of Australia,386
+National Library of Australia,438
+National Museum of Australia,186
+Defence,18301
+Defence Housing Authority,679
+DEST,1684
+Australian Research Council,53
+DEWR,2171
+Australian Industrial Registry,188
+Comcare,325
+EOWA,10
+NOHSC,89
+Environment and Heritage,1276
+Australian Greenhouse Office,166
+Bureau of Meteorology,1371
+GBRMPA,162
+National Oceans Office,39
+Family and Community Services,4818
+Aust. Institute of Family Studies,38
+Centrelink,24904
+Finance and Administration,895
+Australian Electoral Comm.,765
+ComSuper,329
+Foreign Affairs and Trade,2809
+ACIAR,40
+Health and Ageing,5365
+Aust. Institute of Health & Welfare,157
+ARPANSA,117
+FSANZ,104
+National Blood Authority,33
+OPSR,27
+DIMIA,4487
+Aboriginal Hostels Ltd.,291
+AIATSIS,59
+ATSIC,24
+ATSIS,1109
+Torres Strait Regional Authority,34
+"Industry, Tourism & Resources",2878
+Prime Minister & Cabinet,372
+ANAO,287
+APS Commission,172
+Commonwealth Ombudsman,82
+Office of National Assessments,51
+Transport & Regional Services,885
+National Capital Authority,68
+Treasury,984
+Australian Bureau of Statistics,2866
+ACCC,450
+ASIC,1137
+Australian Taxation Office,20501
+Productivity Commission,183
+Veterans' Affairs,2356
+Australian War Memorial,196
--- /dev/null
+++ b/admin/0405apsemployees.csv
@@ -1,1 +1,88 @@
+"Agriculture, Fisheries & Forestry ",1297
+AQIS,3028
+Attorney-Generals ,1011
+Administrative Appeals Tribunal,154
+AUSTRAC,137
+Australian Crime Commission,412
+Australian Customs Service,5054
+Commonwealth DPP,451
+CrimTrac Agency,51
+Family Court of Australia,737
+Federal Court of Australia,403
+Federal Magistrates Service,102
+Federal Privacy Commissioner,40
+HREOC,95
+ITSA,275
+National Native Title Tribunal,265
+Office of Film & Literature Class,41
+Office of the Parl. Counsel,43
+"Communications, IT & the Arts ",804
+Aust. Broadcasting Authority,106
+Aust. Communications Authority,416
+Australian Film Commission,167
+Aust. National Maritime Museum,114
+National Archives of Australia,514
+National Library of Australia,523
+National Museum of Australia,260
+Defence ,18638
+Defence Housing Authority,685
+DEST ,2001
+Australian Research Council,71
+Questacon,211
+DEWR ,3141
+Australian Industrial Registry,202
+Comcare,384
+EOWA,28
+Environment & Heritage ,1852
+Bureau of Meteorology,1442
+GBRMPA,170
+ORER,11
+Family & Community Services ,1949
+Aust. Institute of Family Studies,66
+Finance & Administration ,1297
+Australian Electoral Comm.,808
+Commonwealth Grants Commission,40
+ComSuper,393
+Foreign Affairs & Trade ,2468
+ACIAR,47
+AusAID,615
+Health & Ageing ,4220
+Aust. Institute of Health & Welfare,215
+ARPANSA,130
+FSANZ,121
+National Blood Authority,40
+Professional Services Review,18
+Human Services ,37
+Centrelink,25317
+Child Support Agency,3238
+CRS Australia,1998
+Immigration ,5769
+Aboriginal Hostels Ltd.,458
+AIATSIS,116
+Migration Review Tribunal,132
+Refugee Review Tribunal,176
+Torres Strait Regional Authority,48
+"Industry, Tourism & Resources ",1906
+Geoscience Australia,635
+IP Australia,821
+NOPSA,28
+Prime Minister & Cabinet ,402
+ANAO,313
+Australian Public Service Commission,211
+Commonwealth Ombudsman,113
+National Water Commission,27
+Office of National Assessments,93
+Transport & Regional Services ,1178
+National Capital Authority,91
+Treasury ,885
+Australian Bureau of Statistics,3095
+ACCC,509
+AOFM,33
+ASIC,1494
+Australian Taxation Office,23417
+National Competition Council,11
+Productivity Commission,200
+Royal Australian Mint,117
+Veterans Affairs ,2655
+Australian War Memorial,310
--- /dev/null
+++ b/admin/0506apsemployees.csv
@@ -1,1 +1,90 @@
+"Agriculture, Fisheries & Forestry",1498
+AQIS,3289
+Attorney-Generals,1226
+Administrative Appeals Tribunal,156
+AUSTRAC,141
+Australian Crime Commission,425
+Australian Customs Service,5574
+CrimTrac Agency,53
+Family Court of Australia,765
+Federal Court of Australia,412
+Federal Magistrates Service,112
+Federal Privacy Commissioner,44
+HREOC,102
+ITSA,270
+National Native Title Tribunal,244
+Commonwealth DPP,469
+Office of Film & Literature Class,49
+Office of the Parl. Counsel,46
+"Communications, IT & the Arts",833
+Australian Communications & Media Authority,497
+Australian Film Commission,255
+Australian National Maritime Museum,107
+Australian Sports Anti-Doping Authority,46
+National Archives of Australia,467
+National Library of Australia,499
+National Museum of Australia,301
+Defence,19791
+Defence Housing Authority,699
+DEST,2359
+AIATSIS,128
+Australian Research Council,69
+Questacon,230
+DEWR,3853
+Australian Building and Construction Commission,85
+Australian Fair Pay Commission,11
+Australian Industrial Registry,175
+Comcare,379
+EOWA,30
+Office of Workplace Services,188
+Environment & Heritage,2060
+Bureau of Meteorology,1404
+GBRMPA,172
+ORER,11
+FaCSIA,2589
+Aboriginal Hostels Ltd.,489
+Australian Institute of Family Studies,53
+Torres Strait Regional Authority,72
+Finance & Administration,1409
+Australian Electoral Commission,790
+Commonwealth Grants Commission,41
+ComSuper,435
+Foreign Affairs & Trade,2551
+ACIAR,50
+AusAID,706
+Health & Ageing,4301
+Australian Institute of Health & Welfare,203
+ARPANSA,136
+FSANZ,125
+National Blood Authority,52
+Professional Services Review,14
+Human Services,5545
+Centrelink,26549
+Medicare Australia,5393
+Immigration,6341
+Migration Review Tribunal,103
+Refugee Review Tribunal,161
+"Industry, Tourism & Resources",1979
+Geoscience Australia,669
+IP AUSTRALIA,888
+NOPSA,34
+Prime Minister & Cabinet,590
+ANAO,308
+Australian Public Service Commission,253
+Commonwealth Ombudsman,141
+National Water Commission,45
+Office of National Assessments,122
+Transport & Regional Services,1327
+National Capital Authority,95
+Treasury,915
+Australian Bureau of Statistics,3511
+ACCC,582
+AOFM,39
+ASIC,1540
+Australian Taxation Office,22568
+National Competition Council,9
+Productivity Commission,205
+Royal Australian Mint,145
+Veterans Affairs,2521
+Australian War Memorial,316
--- /dev/null
+++ b/admin/0607apsemployees.csv
@@ -1,1 +1,92 @@
+"Agriculture, Fisheries & Forestry",4951
+Attorney-Generals,1504
+Administrative Appeals Tribunal,162
+AUSTRAC,247
+ACLEI,6
+Australian Crime Commission,554
+Australian Customs Service,5904
+CrimTrac Agency,81
+Family Court of Australia,727
+Federal Court of Australia,401
+Federal Magistrates Court,160
+Federal Privacy Commissioner,58
+HREOC,116
+ITSA,285
+National Native Title Tribunal,244
+Commonwealth DPP,497
+Office of the Parl. Counsel,48
+"Communications, IT & the Arts",914
+ACMA,552
+Australian Film Commission,244
+Aust. National Maritime Museum,109
+ASADA,55
+National Archives of Australia,465
+National Library of Australia,521
+National Museum of Australia,282
+Defence,21177
+Defence Housing Australia,699
+DEST,2470
+AIATSIS,114
+Australian Research Council,73
+Questacon,231
+DEWR,3862
+ABCC,128
+Australian Fair Pay Commission,28
+Australian Industrial Registry,175
+Comcare,426
+EOWA,21
+Office of Workplace Services,303
+Environment & Water Resources,2292
+Bureau of Meteorology,1451
+GBRMPA,171
+ORER,11
+FaCSIA,3099
+Aboriginal Hostels Ltd.,491
+Aust. Institute of Family Studies,59
+Torres Strait Regional Authority,76
+Finance & Administration,1514
+Australian Electoral Commission,886
+Commonwealth Grants Commission,41
+ComSuper,595
+Future Fund Management Agency,12
+Foreign Affairs & Trade,2784
+ACIAR,52
+AusAID,775
+Austrade,515
+Health & Ageing,4830
+Aust. Institute of Health & Welfare,211
+ARPANSA,134
+Cancer Australia,26
+FSANZ,131
+National Blood Authority,50
+NHMRC,221
+Professional Services Review,16
+Human Services,6433
+Centrelink,27296
+Medicare Australia,5959
+Immigration,6788
+MRT/RRT,304
+"Industry, Tourism & Resources",2075
+Geoscience Australia,682
+IP AUSTRALIA,913
+NOPSA,42
+Prime Minister & Cabinet,711
+ANAO,330
+Australian Public Service Commission,250
+Commonwealth Ombudsman,153
+National Water Commission,32
+Office of National Assessments,133
+Transport & Regional Services,1330
+National Capital Authority,106
+Treasury,996
+Australian Bureau of Statistics,3167
+ACCC,647
+AOFM,33
+ASIC,1664
+Australian Taxation Office,23110
+National Competition Council,9
+Productivity Commission,207
+Royal Australian Mint,181
+Veterans' Affairs,2401
+Australian War Memorial,293
--- /dev/null
+++ b/admin/0708apsemployees.csv
@@ -1,1 +1,98 @@
+"Agriculture, Fisheries & Forestry",4785
+APVMA,130
+Export Wheat Commission,15
+Attorney-General's,1420
+Administrative Appeals Tribunal,114
+ACLEI,11
+AUSTRAC,332
+Australian Crime Commission,495
+Australian Customs Service,6106
+CrimTrac Agency,138
+Family Court of Australia,558
+Federal Court of Australia,261
+Federal Magistrates Court,141
+HREOC,93
+ITSA,270
+National Capital Authority,47
+National Native Title Tribunal,217
+Commonwealth DPP,509
+Office of the Parl. Counsel,50
+DBCDE,559
+ACMA,569
+Climate Change,218
+ORER,10
+Defence,20819
+Defence Housing Australia,565
+DEEWR,5619
+ABCC,137
+Australian Fair Pay Commission,25
+Australian Industrial Registry,157
+AIATSIS,48
+Comcare,469
+Office of the Workplace Ombudsman,365
+Workplace Authority,610
+"Environment, Water, Heritage and the Arts",2333
+Australian Film Commission,256
+Aust. National Maritime Museum,105
+Bureau of Meteorology,1265
+GBRMPA,184
+National Library of Australia,444
+National Museum of Australia,226
+National Water Commission,29
+FaHCSIA,2810
+Aboriginal Hostels Ltd.,362
+EOWA,17
+Torres Strait Regional Authority,36
+Finance & Deregulation,1349
+Australian Electoral Comm.,697
+ComSuper,533
+Future Fund Management Agency,38
+National Archives of Australia,390
+Foreign Affairs & Trade,2665
+ACIAR,32
+AusAID,801
+AUSTRADE,485
+Health & Ageing,4445
+Aust. Institute of Health & Welfare,228
+Australian Sports Anti-Doping Authority,60
+ARPANSA,132
+Cancer Australia,22
+FSANZ,118
+National Blood Authority,44
+NHMRC,209
+Private Health Insurance Ombudsman,11
+Professional Services Review,19
+Human Services,5951
+Centrelink,25416
+Medicare Australia,5593
+Immigration,6495
+RRT and MRT,270
+Infrastructure,1225
+DIISR,1656
+Australian Research Council,87
+IP AUSTRALIA,968
+Questacon,82
+Prime Minister & Cabinet,483
+Aust. Institute of Family Studies,42
+ANAO,307
+Australian Public Service Commission,201
+Commonwealth Ombudsman,155
+Federal Privacy Commissioner,61
+OIGIS,9
+Office of National Assessments,96
+"Resources, Energy and Tourism",329
+Geoscience Australia,611
+NOPSA,49
+Treasury,973
+Australian Bureau of Statistics,2858
+ACCC,658
+AOFM,30
+ASIC,1451
+Australian Taxation Office,22486
+Commonwealth Grants Commission,39
+National Competition Council,7
+Productivity Commission,179
+Royal Australian Mint,146
+Veterans' Affairs,2250
+Australian War Memorial,228
--- /dev/null
+++ b/admin/0809apsemployees.csv
@@ -1,1 +1,103 @@
+"Agriculture, Fisheries & Forestry",5168
+Australian Fisheries Management Authority,260
+APVMA,161
+Wheat Exports Australia,17
+Attorney-General's,1663
+Administrative Appeals Tribunal,155
+ACLEI,18
+AUSTRAC,343
+Australian Crime Commission,527
+Australian Customs and Border Protection Service,6031
+CrimTrac Agency,187
+Family Court of Australia,669
+Federal Court of Australia,376
+Federal Magistrates Court,213
+HREOC,122
+ITSA,294
+National Capital Authority,58
+National Native Title Tribunal,246
+Commonwealth DPP,614
+Office of Parliamentary Counsel,50
+DBCDE,651
+ACMA,624
+Climate Change,415
+ORER,21
+Defence,21152
+Defence Housing Australia,728
+DEEWR,5996
+ABCC,156
+Australian Fair Pay Commission Secretariat,29
+Australian Industrial Registry,203
+Comcare,570
+Office of the Workplace Ombudsman,484
+Workplace Authority,470
+"Environment, Water, Heritage and the Arts",3392
+Australian National Maritime Museum,119
+Bureau of Meteorology,1545
+GBRMPA,241
+Murray Darling Basin Authority,235
+National Library of Australia,513
+National Museum of Australia,295
+National Film and Sound Archive,235
+National Water Commission,48
+Screen Australia,78
+FaHCSIA,3455
+Aboriginal Hostels Ltd.,519
+EOWA,31
+Torres Strait Regional Authority,77
+Finance & Deregulation,1706
+Australian Electoral Commission,805
+ComSuper,574
+Future Fund Management Agency,57
+National Archives of Australia,506
+Foreign Affairs & Trade,2984
+ACIAR,48
+AusAID,981
+AUSTRADE,524
+Health & Ageing,4912
+Australian Institute of Health & Welfare,265
+AOTDTA,15
+Australian Sports Anti-Doping Authority,60
+ARPANSA,152
+Cancer Australia,25
+FSANZ,123
+National Blood Authority,42
+NHMRC,238
+Private Health Insurance Ombudsman,10
+Professional Services Review,30
+Human Services,6620
+Centrelink,27973
+Medicare Australia,5844
+Immigration & Citizenship,7052
+MRT & RRT,273
+Infrastructure,1279
+DIISR,1843
+AIATSIS,119
+Australian Research Council,107
+IP AUSTRALIA,1049
+Questacon,233
+Prime Minister & Cabinet,647
+Australian Institute of Family Studies,75
+ANAO,365
+Australian Public Service Commission,241
+Commonwealth Ombudsman,168
+Federal Privacy Commissioner,64
+OIGIS,13
+Office of National Assessments,148
+Old Parliament House,88
+"Resources, Energy and Tourism",401
+Geoscience Australia,745
+NOPSA,54
+Treasury,1103
+Australian Bureau of Statistics,2871
+ACCC,787
+AOFM,36
+ASIC,1882
+Australian Taxation Office,23505
+Commonwealth Grants Commission,46
+National Competition Council,9
+Productivity Commission,183
+Royal Australian Mint,197
+Veterans' Affairs,2165
+Australian War Memorial,313
--- /dev/null
+++ b/admin/0910apsemployees.csv
@@ -1,1 +1,103 @@
+National Competition Council,9
+Private Health Insurance Ombudsman,12
+OIGIS,12
+Wheat Exports Australia,13
+Cancer Australia,19
+ACLEI,20
+ORER,25
+Professional Services Review,30
+AOTDTA,31
+AOFM,35
+EOWA,36
+Commonwealth Grants Commission,40
+National Blood Authority,46
+ACIAR,51
+NOPSA,51
+Office of the Parl. Counsel,54
+National Water Commission,57
+Office of the Privacy Commissioner,58
+National Capital Authority,60
+ASADA,69
+Future Fund Management Agency,70
+Aust. Institute of Family Studies,76
+Torres Strait Regional Authority,82
+Old Parliament House,95
+Screen Australia,102
+Australian Transport Safety Bureau,107
+Safe Work Australia,109
+Australian Research Council,109
+Human Rights Commission,124
+AIATSIS,126
+FSANZ,130
+Aust. National Maritime Museum,132
+Office of National Assessments,146
+ABCC,151
+ARPANSA,152
+APVMA,155
+Commonwealth Ombudsman,156
+Administrative Appeals Tribunal,165
+Royal Australian Mint,171
+CrimTrac Agency,185
+Federal Magistrates Court,186
+Productivity Commission,196
+Questacon,218
+National Native Title Tribunal,223
+AFMA,225
+National Film and Sound Archive,232
+GBRMPA,237
+Australian Public Service Commission,247
+NHMRC,249
+MRT & RRT,273
+Australian War Memorial,291
+Fair Work Australia,299
+National Museum of Australia,312
+Murray Darling Basin Authority,314
+AUSTRAC,318
+ITSA,321
+Federal Court of Australia,360
+ANAO,367
+Aust. Institute of Health & Welfare,383
+"Resources, Energy and Tourism",454
+National Archives of Australia,504
+National Library of Australia,540
+Australian Crime Commission,543
+Aboriginal Hostels Ltd.,556
+Commonwealth DPP,558
+AUSTRADE,563
+ComSuper,587
+Comcare,593
+Defence Housing Australia,630
+Family Court of Australia,660
+ACMA,663
+Prime Minister & Cabinet,680
+DBCDE,715
+Geoscience Australia,723
+ACCC,804
+Australian Electoral Commission,922
+Office of the Fair Work Ombudsman,969
+Climate Change and Energy Efficiency,1019
+AusAID,1037
+IP AUSTRALIA,1050
+Infrastructure,1105
+Treasury,1133
+Bureau of Meteorology,1656
+Attorney-General's,1707
+Finance & Deregulation,1841
+DIISR,2072
+ASIC,2080
+Veterans' Affairs,2144
+"Environment, Water, Heritage and the Arts",3115
+Foreign Affairs & Trade,3160
+Australian Bureau of Statistics,3332
+FaHCSIA,3685
+"Agriculture, Fisheries & Forestry",4892
+Health & Ageing,5232
+Medicare Australia,5695
+Customs,5940
+DEEWR,6054
+Human Services,6607
+Immigration & Citizenship,7014
+Defence,21458
+Australian Taxation Office,24070
+Centrelink,27312
--- /dev/null
+++ b/admin/1011apsemployees.csv
@@ -1,1 +1,98 @@
+"Agriculture, Fisheries and Forestry",5219
+Australian Fisheries Management Authority,227
+APVMA,165
+Wheat Exports Australia,13
+Attorney-General's,1618
+Administrative Appeals Tribunal,162
+ACLEI,27
+AUSTRAC,332
+Australian Crime Commission,635
+Aust. Customs & Border Prot. Service,5767
+Aust. Human Rights Commission,139
+Commonwealth DPP,514
+CrimTrac Agency,181
+Family Court of Australia,670
+Federal Court of Australia,362
+Federal Magistrates Court,195
+ITSA,344
+National Native Title Tribunal,176
+Office of the Parl. Counsel,55
+DBCDE,700
+ACMA,662
+Climate Change and Energy Efficiency,984
+ORER,44
+Defence,22481
+Defence Housing Australia,614
+DEEWR,5468
+ABCC,150
+Comcare,667
+Fair Work Australia,310
+Office of the Fair Work Ombudsman,955
+Safe Work Australia,121
+FaHCSIA,3500
+Aboriginal Hostels Ltd.,563
+EOWA,34
+Torres Strait Regional Authority,107
+Finance and Deregulation,1964
+Australian Electoral Commission,913
+ComSuper,545
+Future Fund Management Agency,80
+Foreign Affairs and Trade,3354
+ACIAR,50
+AusAID,1245
+AUSTRADE,546
+Health and Ageing,5396
+Aust. Institute of Health and Welfare,393
+AOTDTA,33
+ARPANSA,162
+Cancer Australia,22
+FSANZ,133
+National Blood Authority,50
+NHMRC,248
+Private Health Insurance Ombudsman,14
+Professional Services Review,30
+Human Services,6443
+Centrelink,25225
+Medicare Australia,5842
+Immigration and Citizenship,7903
+MRT RRT,283
+Infrastructure,1001
+Australian Transport Safety Bureau,109
+DIISR,3594
+AIATSIS,136
+Australian Research Council,112
+Prime Minister and Cabinet,1140
+Aust. Institute of Family Studies,81
+ANAO,360
+Aust. National Maritime Museum,127
+Australian Public Service Commission,332
+Australian Sports Anti-Doping Authority,67
+Commonwealth Ombudsman,183
+National Archives of Australia,494
+National Library of Australia,494
+National Museum of Australia,287
+National Film and Sound Archive,235
+Office of the Australian Information Commissioner,86
+OIGIS,13
+Office of National Assessments,148
+Old Parliament House,92
+Screen Australia,35
+Regional Australia,336
+National Capital Authority,69
+"Resources, Energy and Tourism",1262
+NOPSA,60
+"Sustainability, Environment, Water, Population and Communities",2949
+Bureau of Meteorology,1775
+GBRMPA,266
+Murray Darling Basin Authority,314
+National Water Commission,74
+Treasury,1397
+Australian Bureau of Statistics,3540
+ACCC,825
+ASIC,1905
+Australian Taxation Office,24902
+National Competition Council,9
+Productivity Commission,203
+Veterans' Affairs,2146
+Australian War Memorial,302
--- /dev/null
+++ b/admin/cacfma.csv
@@ -1,1 +1,191 @@
+AAF Company,82?008?629?490
+Aboriginal Hostels Limited ,47?008?504?587
+Administrative Appeals Tribunal,90?680?970?626
+Aged Care Standards and Accreditation Agency Ltd,64?079?618?652
+Airservices Australia ,59?698?720?886
+Albury-Wodonga Development Corporation ,71?893?478?442
+Anindilyakwa Land Council ,45?175?406?445
+Army and Air Force Canteen Service ,69?289?134?420
+ASC Pty Ltd ,64?008?605?034
+Attorney-General's Department,92?661?124?436
+Australia Business Arts Foundation Ltd ,88?072?479?835
+Australia Council,38?392?626?187
+Australian Agency for International Development (AusAID),62?921?558?838
+Australian Broadcasting Corporation,52?429?278?345
+Australian Bureau of Statistics,26?331?428?522
+Australian Centre for International Agricultural Research (ACIAR),34?864?955?427
+Australian Commission for Law Enforcement Integrity (ACLEI),78?796?734?093
+Australian Commission on Safety and Quality in Health Care,97250687371
+Australian Communications and Media Authority (ACMA),55?386?169?386
+Australian Competition and Consumer Commission,94?410?483?623
+Australian Crime Commission,11?259?448?410
+"Australian Curriculum, Assessment and Reporting Authority ",54?735?928?084
+Australian Customs and Border Protection Service,66?015?286?036
+Australian Electoral Commission,21?133?285?851
+Australian Federal Police,17?864?931?143
+"Australian Film, Television and Radio School",19?892?732?021
+Australian Fisheries Management Authority,81?098?497?517
+Australian Government Solicitor,69?405?937?639
+Australian Hearing Services ,80?308?797?003
+Australian Human Rights Commission,47?996?232?602
+Australian Industry Development,55?085?059?559
+Australian Institute for Teaching and School Leadership Limited,17?117?362?740
+Australian Institute of Aboriginal and Torres Strait Islander Studies,62?020?533?641
+Australian Institute of Criminology,63257175248
+Australian Institute of Family Studies (AIFS),64?001?053?079
+Australian Institute of Health and Welfare ,16?515?245?497
+Australian Institute of Marine Science,78?961?616?230
+Australian Law Reform Commission,88913413914
+Australian Learning and Teaching Council Limited ,30?109?826?628
+Australian Maritime Safety Authority,65?377?938?320
+Australian Military Forces Relief Trust Fund ,52?168?913?646
+Australian National Audit Office ,33?020?645?631
+Australian National Maritime Museum,35?023?590?988
+Australian National Preventive Health Agency (ANPHA),33?965?140?953
+Australian National University,52?234?063?906
+Australian Nuclear Science and Technology Organisation ,47?956?969?590
+Australian Office of Financial Management (AOFM),13?059?525?039
+Australian Pesticides and Veterinary Medicines Authority (APVMA),19?495?043?447
+Australian Postal Corporation,28?864?970?579
+Australian Prudential Regulation Authority (APRA),79?635?582?658
+Australian Public Service Commission (APS Commission),99?470?863?260
+Australian Radiation Protection and Nuclear Safety Agency (ARPANSA),61?321?195?155
+Australian Rail Track Corporation Limited ,75?081?455?754
+Australian Reinsurance Pool Corporation,74?807?136?872
+Australian Research Council,35?201?451?156
+Australian River Co. Limited,94?008?654?206
+Australian Secret Intelligence Service,49?667?785?014
+Australian Securities and Investments Commission,86?768?265?615
+Australian Security Intelligence Organisation,37?467?566?201
+Australian Skills Quality Authority (National Vocational Education and Training Regulator),72581678650
+Australian Solar Institute Limited ,65138300688
+Australian Sports Anti-Doping Authority (ASADA),91?592?527?503
+Australian Sports Commission,67374695240
+Australian Sports Foundation Limited ,27?008?613?858
+Australian Strategic Policy Institute Limited ,77?097?369?045
+Australian Taxation Office,51?824?753?556
+Australian Trade Commission (Austrade),11?764?698?227
+Australian Transaction Reports and Analysis Centre (AUSTRAC),32?770?513?371
+Australian Transport Safety Bureau (ATSB),86?267?354?017
+Australian War Memorial ,64?909?221?257
+Bundanon Trust,72?058?829?217
+Bureau of Meteorology,92?637?533?532
+Cancer Australia,21?075?951?918
+Central Land Council,71?979?619?393
+Civil Aviation Safety Authority,44?808?014?470
+Coal Mining Industry (Long Service Leave Funding) Corporation,12?039?670?644
+Comcare ,41?640?788?304
+Commonwealth Grants Commission,64?703?642?210
+Commonwealth Scientific and Industrial Research Organisation,41?687?119?230
+Commonwealth Superannuation Corporation ,48882817243
+ComSuper,77?310?752?950
+Corporations and Markets Advisory Committee (CAMAC),41?574?479?010
+Cotton Research and Development Corporation,71?054?238?316
+CrimTrac Agency,17?193?904?699
+Defence Housing Australia,72?968?504?934
+"Department of Agriculture, Fisheries and Forestry ",24?113?085?695
+"Department of Broadband, Communications and the Digital Economy",51?491?646?726
+Department of Climate Change and Energy Efficiency,50?182?626?845
+"Department of Education, Employment and Workplace Relations",63?578?775?294
+"Department of Families, Housing, Community Services and Indigenous Affairs",36?342?015?855
+Department of Finance and Deregulation,61?970?632?495
+Department of Foreign Affairs and Trade,47?065?634?525
+Department of Health and Ageing,83?605?426?759
+Department of Human Services,90?794?605?008
+Department of Immigration and Citizenship,33?380?054?835
+Department of Infrastructure and Transport,86?267?354?017
+"Department of Innovation, Industry, Science and Research",74?599?608?295
+Department of Parliamentary Services,52?997?141?147
+"Department of Regional Australia, Regional Development and Local Government",37?862?725?624
+"Department of Resources, Energy and Tourism",46?252?861?927
+"Department of Sustainability, Environment, Water, Population and Communities",34?190?894?983
+Department of the House of Representatives,18?526?287?740
+Department of the Prime Minister and Cabinet,18?108?001?191
+Department of the Senate,23?991?641?527
+Department of the Treasury,92?802?414?793
+Department of Veterans' Affairs,23?964?290?824
+Director of National Parks ,13?051?694?963
+Equal Opportunity for Women in the Workplace Agency,47?641?643?874
+Export Finance and Insurance Corporation,96?874?024?697
+Fair Work Australia (FWA),93?614?579?199
+Family Court of Australia,63?684?208?971
+Federal Court of Australia,49?110?847?399
+Federal Magistrates Court of Australia,60?265?617?271
+Fisheries Research and Development Corporation,74?311?094?913
+Food Standards Australia New Zealand,20?537?066?246
+Future Fund Management Agency,53?156?699?293
+General Practice Education and Training Limited,95?095?433?140
+Geoscience Australia,80?091?799?039
+Grains Research and Development Corporation ,55?611?223?291
+Grape and Wine Research and Development Corporation,72?618?007?571
+Great Barrier Reef Marine Park Authority,12?949?356?885
+Health Workforce Australia,21?295?050?589
+HIH Claims Support Limited,92?096?857?635
+IIF Investments Pty Limited,55?082?153?884
+Indigenous Business Australia,25?192?932?833
+Indigenous Land Corporation,59?912?679?254
+Insolvency and Trustee Service Australia (ITSA),63?384?330?717
+Inspector-General of Taxation,51?248?702?319
+Interim Independent Hospital Pricing Authority,27598959960
+IP Australia,38?113?072?755
+Low Carbon Australia Limited,63?097?727?968
+Medibank Private Limited ,47?080?890?259
+Migration Review Tribunal and Refugee Review Tribunal ,50?760?799?564
+Murray-Darling Basin Authority,13?679?821?382
+National Archives of Australia,36?889?228?992
+National Australia Day Council Limited ,76?050?300?626
+National Blood Authority,87?361?602?478
+National Breast and Ovarian Cancer Centre,85?094?118?902
+National Capital Authority,75?149?374?427
+National Competition Council ,56?552?760?098
+National Film and Sound Archive,41?251?017?588
+National Gallery of Australia,27?855?975?449
+National Health and Medical Research Council (NHMRC),88?601?010?284
+National Library of Australia ,28?346?858?075
+National Museum of Australia ,70?592?297?967
+National Native Title Tribunal,70?238?042?351
+National Offshore Petroleum Safety Authority (NOPSA),22?385?178?289
+National Water Commission ,94?364?176?431
+NBN Co Limited,86?136?533?741
+Northern Land Council,56?327?515?336
+Office of National Assessments,87?904?367?991
+Office of Parliamentary Counsel,41?425?630?817
+Office of the Auditing and Assurance Standards Board ,80?959?780?601
+Office of the Australian Accounting Standards Board (AASB),92?702?019?575
+Office of the Australian Building and Construction Commissioner,68?003?725?098
+Office of the Australian Information Commissioner ,85249230937
+Office of the Commonwealth Ombudsman,53?003?678?148
+Office of the Director of Public Prosecutions,41?036?606?436
+Office of the Fair Work Ombudsman,71?141?751?477
+Office of the Inspector-General of Intelligence and Security,67?332?668?643
+Office of the Official Secretary to the Governor-General,67?582?329?284
+Office of the Renewable Energy Regulator,68?574?011?917
+Old Parliament House,30?620?774?963
+Organ and Tissue Authority (Australian Organ and Tissue Donation and Transplantation Authority),56?253?405?315
+Outback Stores Pty Ltd ,63120661234
+Private Health Insurance Administration Council ,50?831?782?014
+Private Health Insurance Ombudsman,61?673?137?709
+Productivity Commission,78?094?372?050
+Professional Services Review Scheme,45?307?308?260
+RAAF Welfare Recreational Company ,45?008?499?303
+Reserve Bank of Australia,50?008?559?486
+Royal Australian Air Force Veterans' Residences Trust Fund ,40?594?141?285
+Royal Australian Air Force Welfare Trust Fund ,24?616?803?717
+Royal Australian Mint,45?852?104?259
+Royal Australian Navy Central Canteens Board,50?616?294?781
+Royal Australian Navy Relief Trust Fund ,49?934?525?476
+Rural Industries Research and Development Corporation,25?203?754?319
+Safe Work Australia,81?840?374?163
+Screen Australia ,46?741?353?180
+"Seafarers Safety, Rehabilitation and Compensation Authority (Seacare Authority)",32?745?854?352
+Special Broadcasting Service Corporation,91?314?398?574
+Sugar Research and Development Corporation,41?343?997?980
+Sydney Harbour Federation Trust,14?178?614?905
+Tertiary Education Quality and Standards Agency,50658250012
+Tiwi Land Council,86?106?441?085
+Torres Strait Regional Authority,57?155?285?807
+Tourism Australia ,99?657?548?712
+Wheat Exports Australia,40?485?918?341
+Wine Australia Corporation ,59?728?300?326
+Wreck Bay Aboriginal Community Council,62?564?797?956
--- /dev/null
+++ b/admin/exportAll.csv.php
@@ -1,1 +1,63 @@
+<?php
+include_once("../include/common.inc.php");
+
+
+setlocale(LC_CTYPE, 'C');
+
+$headers = Array("#id", "name", "request_email", "short_name", "notes", "publication_scheme", "home_page", "tag_string");
+
+$db = $server->get_db('disclosr-agencies');
+$headers = Array();
+ try {
+ $rows = $db->get_view("app", "fieldNames?group=true", null, true)->rows;
+
+ $dataValues = Array();
+ foreach ($rows as $row) {
+ $headers[] = $row->key;
+ }
+} catch (SetteeRestClientException $e) {
+ setteErrorHandler($e);
+}
+
+$fp = fopen('php://output', 'w');
+if ($fp && $db) {
+ header('Content-Type: text/csv; charset=utf-8');
+ header('Content-Disposition: attachment; filename="export.' . date("c") . '.csv"');
+ header('Pragma: no-cache');
+ header('Expires: 0');
+ fputcsv($fp, $headers);
+ try {
+ $agencies = $db->get_view("app", "byCanonicalName", null, true)->rows;
+ //print_r($rows);
+ foreach ($agencies as $agency) {
+ // print_r($agency);
+
+ if ( !isset($agency->value->status)) {
+ $row = Array();
+ $agencyArray = object_to_array($agency->value);
+ foreach ($headers as $fieldName) {
+ if (isset($agencyArray[$fieldName])) {
+ if (is_array($agencyArray[$fieldName])) {
+ $row[] = implode(";",$agencyArray[$fieldName]);
+ } else {
+ $row[] = $agencyArray[$fieldName];
+ }
+ } else {
+ $row[] = "";
+ }
+ }
+
+ fputcsv($fp, array_values($row));
+
+
+ }
+ }
+ } catch (SetteeRestClientException $e) {
+ setteErrorHandler($e);
+ }
+
+ die;
+}
+?>
+
--- /dev/null
+++ b/admin/exportEmployees.csv.php
@@ -1,1 +1,60 @@
+<?php
+include_once("../include/common.inc.php");
+
+
+setlocale(LC_CTYPE, 'C');
+
+$headers = Array("name");
+
+$db = $server->get_db('disclosr-agencies');
+try {
+ $rows = $db->get_view("app", "all", null, true)->rows;
+
+ $dataValues = Array();
+ foreach ($rows as $row) {
+ if (isset($row->value->statistics->employees)) {
+
+ $headers = array_unique(array_merge($headers, array_keys(object_to_array($row->value->statistics->employees))));
+ }
+ }
+} catch (SetteeRestClientException $e) {
+ setteErrorHandler($e);
+}
+
+$fp = fopen('php://output', 'w');
+if ($fp && $db) {
+ header('Content-Type: text/csv; charset=utf-8');
+ header('Content-Disposition: attachment; filename="export.employeestats.' . date("c") . '.csv"');
+ header('Pragma: no-cache');
+ header('Expires: 0');
+ fputcsv($fp, $headers);
+ try {
+ $agencies = $db->get_view("app", "all", null, true)->rows;
+ //print_r($agencies);
+ foreach ($agencies as $agency) {
+
+ if (isset($agency->value->statistics->employees)) {
+ $row = Array();
+ $agencyEmployeesArray = object_to_array($agency->value->statistics->employees);
+ foreach ($headers as $fieldName) {
+ if ($fieldName == "name") {
+ $row[] = $agency->value->name;
+ } else if (isset($agencyEmployeesArray[$fieldName])) {
+ $row[] = $agencyEmployeesArray[$fieldName]["value"];
+ } else {
+ $row[] = 0;
+ }
+ }
+
+ fputcsv($fp, array_values($row));
+ }
+ }
+ } catch (SetteeRestClientException $e) {
+ setteErrorHandler($e);
+ }
+
+ die;
+}
+?>
+
--- /dev/null
+++ b/admin/importAPSCEmployees.php
@@ -1,1 +1,55 @@
+<?php
+require_once '../include/common.inc.php';
+
+$db = $server->get_db('disclosr-agencies');
+$rows = $db->get_view("app", "byName")->rows;
+$nametoid = Array();
+$sums = Array();
+foreach ($rows as $row) {
+ $nametoid[trim($row->key)] = $row->value;
+}
+$employeeCSVs = Array("2002-2003" => "0203apsemployees.csv",
+ "2003-2004" => "0304apsemployees.csv",
+ "2004-2005" => "0405apsemployees.csv",
+ "2005-2006" => "0506apsemployees.csv",
+ "2006-2007" => "0607apsemployees.csv",
+ "2007-2008" => "0708apsemployees.csv",
+ "2008-2009" => "0809apsemployees.csv",
+ "2009-2010" => "0910apsemployees.csv",
+ "2010-2011" => "1011apsemployees.csv"
+);
+foreach ($employeeCSVs as $timePeriod => $employeeCSV) {
+ echo $employeeCSV . "<br>" . PHP_EOL;
+ $row = 1;
+ if (($handle = fopen($employeeCSV, "r")) !== FALSE) {
+ while (($data = fgetcsv($handle, 1000, ",")) !== FALSE) {
+ //print_r($data);
+ $name = trim($data[0]);
+ if (isset($nametoid[$name])) {
+ $id = $nametoid[$name];
+ //echo $id . "<br>" . PHP_EOL;
+ @$sums[$id][$timePeriod] += $data[1];
+ } else {
+ echo "<br>ERROR NAME MISSING FROM ID LIST<br><bR>" . PHP_EOL;
+
+ die();
+
+ }
+ }
+ fclose($handle);
+ }
+}
+foreach ($sums as $id => $sum) {
+ echo $id. "<br>" . PHP_EOL;
+ $doc = $db->get($id);
+ // print_r($doc);
+ if (isset($doc->statistics)) $doc->statistics = Array();
+ foreach ($sum as $timePeriod => $value) {
+ $doc->statistics["employees"][$timePeriod] = Array("value"=>$value, "source"=>"http://apsc.gov.au/stateoftheservice/");
+ }
+ $db->save($doc);
+}
+// employees: timeperiod, source = apsc state of service, value
+?>
+
--- /dev/null
+++ b/admin/importContractDashboardAgencies.php
@@ -1,1 +1,34 @@
+<?php
+require_once '../include/common.inc.php';
+try {
+ $server->create_db('disclosr-agencies');
+} catch (SetteeRestClientException $e) {
+ setteErrorHandler($e);
+}
+$db = $server->get_db('disclosr-agencies');
+createAgencyDesignDoc();
+$conn = new PDO("pgsql:dbname=contractDashboard;user=postgres;password=snmc;host=localhost");
+$namesQ = 'select agency.abn, string_agg("agencyName",\'|\') as names from agency inner join agency_nametoabn on agency.abn::text = agency_nametoabn.abn group by agency.abn;';
+$abntonames = Array();
+foreach ($conn->query($namesQ) as $row) {
+ $abntonames[$row['abn']] = explode("|", $row['names']);
+}
+$result = $conn->query("select * from agency");
+while ($agency = $result->fetch(PDO::FETCH_ASSOC)) {
+ $agency['_id'] = md5($agency['abn']);
+ $agency['otherNames'] = $abntonames[$agency['abn']];
+ if (sizeof($abntonames[$agency['abn']]) == 1)
+ $agency['name'] = $abntonames[$agency['abn']][0];
+ $agency["lastScraped"] = "1/1/1970";
+ $agency["scrapeDepth"] = 1;
+ try {
+ $doc = $db->save($agency);
+ //print_r($doc);
+ echo $agency['abn'] . " imported \n<br>";
+ } catch (SetteeRestClientException $e) {
+ setteErrorHandler($e);
+ }
+}
+?>
+
--- /dev/null
+++ b/admin/importDirectoryUUIDs.php
@@ -1,1 +1,13 @@
+<?php
+// http://gold.gov.au/reports/department/index.xml
+require_once '../include/common.inc.php';
+try {
+ $server->create_db('disclosr-agencies');
+} catch (SetteeRestClientException $e) {
+ setteErrorHandler($e);
+}
+$db = $server->get_db('disclosr-agencies');
+createAgencyDesignDoc();
+?>
+
--- /dev/null
+++ b/admin/importGov2RegisterRSSFacebookTwitter.php
@@ -1,1 +1,16 @@
+<?php
+require_once '../include/common.inc.php';
+try {
+ $server->create_db('disclosr-agencies');
+} catch (SetteeRestClientException $e) {
+ setteErrorHandler($e);
+}
+$db = $server->get_db('disclosr-agencies');
+createAgencyDesignDoc();
+
+// twitter https://docs.google.com/spreadsheet/fm?id=tsJVd9EYoAjbl014y3qMgWg.03918275400592898296.8568379511161083736&hl=en&fmcmd=5&gid=0
+// RSS https://docs.google.com/spreadsheet/fm?id=tbqjwIDHKHiVYF_glJ93GgA.03918275400592898296.8789688748524615194&authkey=CJDP-uQG&hl=en_GB&fmcmd=5&gid=0
+// facebook https://docs.google.com/spreadsheet/fm?id=tkcqoo9wrgzNWmoANuVhsBw.03918275400592898296.3040387705062056060&authkey=CKzl7r0I&hl=en_GB&fmcmd=5&gid=0
+?>
+
--- /dev/null
+++ b/admin/refreshDesignDoc.php
@@ -1,1 +1,7 @@
+<?php
+require_once '../include/common.inc.php';
+$db = $server->get_db('disclosr-agencies');
+createAgencyDesignDoc();
+?>
+
--- /dev/null
+++ b/admin/resolveConflicts.php
@@ -1,1 +1,43 @@
+<?php
+include_once('../include/common.inc.php');
+include_header();
+// Include the diff class
+echo '<STYLE TYPE="text/css">
+<!--
+ @import url(../lib/php-diff/example/styles.css);
+-->
+</STYLE>
+ ';
+require_once dirname(__FILE__) . '/../lib/php-diff/lib/Diff.php';
+// Generate a side by side diff
+require_once dirname(__FILE__) . '/../lib/php-diff/lib/Diff/Renderer/Html/SideBySide.php';
+$renderer = new Diff_Renderer_Html_SideBySide;
+
+
+
+$db = $server->get_db('disclosr-agencies');
+$docs = Array();
+try {
+ $rows = $db->get_view("app", "getConflicts")->rows;
+ //print_r($rows);
+ foreach ($rows as $row) {
+ echo '<h2>' . $row->id . '</h2>';
+ echo "Comparing " . $row->value[0] . " and " . $row->value[1];
+ $docA = explode(",", json_encode($db->get($row->id . "?rev=" . $row->value[0])));
+ $docB = explode(",", json_encode($db->get($row->id . "?rev=" . $row->value[1])));
+ // Options for generating the diff
+ $options = array(
+ //'ignoreWhitespace' => true,
+ //'ignoreCase' => true,
+ );
+
+ // Initialize the diff class
+ $diff = new Diff($docA, $docB, $options);
+ echo $diff->Render($renderer);
+ }
+} catch (SetteeRestClientException $e) {
+ setteErrorHandler($e);
+}
+include_footer();
+?>
--- /dev/null
+++ b/admin/verify.php
@@ -1,1 +1,56 @@
+<?php
+include_once('../include/common.inc.php');
+include_header();
+
+$db = $server->get_db('disclosr-agencies');
+$docs = Array();
+try {
+ $rows = $db->get_view("app", "byABN")->rows;
+ //print_r($rows);
+ foreach ($rows as $row) {
+ $docs["a" . $row->key] = $row->value;
+ }
+} catch (SetteeRestClientException $e) {
+ setteErrorHandler($e);
+}
+//print_r($docs);
+$row = 1;
+if (($handle = fopen("cacfma.csv", "r")) !== FALSE) {
+ while (($data = fgetcsv($handle, 1000, ",")) !== FALSE) {
+ $row++;
+ echo $data[0] . " " . str_replace("?", "", $data[1]) . "<br />\n";
+ $name = $data[0];
+ $abn = trim(str_replace("?", "", $data[1]));
+ $aabn = "a".$abn;
+ if (isset($docs[$aabn])) {
+ echo "Existing agency ABN detected<br>";
+ if (!in_array($name, object_to_array($docs[$aabn]->otherNames)) && $name != $docs[$aabn]->name) {
+ $docs[$aabn]->otherNames[] = $name;
+ try {
+ $docs[$aabn] = $db->save($docs[$aabn]);
+ //print_r($doc);
+ echo $abn . " additional names imported \n<br>";
+ } catch (SetteeRestClientException $e) {
+ setteErrorHandler($e);
+ }
+ }
+ } else {
+ echo "New agency ABN detected<br>";
+ $agency['_id'] = md5($aabn);
+ $agency['name'] = $name;
+ $agency["abn"] = $abn;
+ try {
+ $doc = $db->save($agency);
+ print_r($doc);
+ echo $abn . " imported \n<br>";
+ } catch (SetteeRestClientException $e) {
+ setteErrorHandler($e);
+ }
+ }
+ echo "<hr>";
+ }
+ fclose($handle);
+}
+include_footer();
+?>
--- /dev/null
+++ b/alaveteli/exportAgencies.csv.php
@@ -1,1 +1,107 @@
+<?php
+include_once("../include/common.inc.php");
+
+function shortName($name) {
+ $name = trim($name);
+ if (strstr($name,"Minister ") || strstr($name,"Treasurer") || strstr($name,"Parliamentary Secretary")) {
+ $badWords = Array ("Assisting the Prime Minister on","Assisting on"," the "," of "," for "," on "," and "," to ",","," ","'","`");
+ return str_replace($badWords,"",$name);
+ }
+
+ else {
+ $out = Array();
+ preg_match_all('/[A-Z]/', $name, $out);
+ return implode("", $out[0]);
+ }
+}
+
+setlocale(LC_CTYPE, 'C');
+
+$headers = Array("#id", "name", "request_email", "short_name", "notes", "publication_scheme", "home_page", "tag_string");
+
+$db = $server->get_db('disclosr-agencies');
+
+$tag = Array();
+try {
+ $rows = $db->get_view("app", "byDeptStateName", null, true)->rows;
+ //print_r($rows);
+ foreach ($rows as $row) {
+ $tag[$row->id] = phrase_to_tag(dept_to_portfolio($row->key));
+ }
+} catch (SetteeRestClientException $e) {
+ setteErrorHandler($e);
+ die();
+}
+
+$foiEmail = Array();
+try {
+ $rows = $db->get_view("app", "foiEmails", null, true)->rows;
+ //print_r($rows);
+ foreach ($rows as $row) {
+ $foiEmail[$row->key] = $row->value;
+ }
+} catch (SetteeRestClientException $e) {
+ setteErrorHandler($e);
+ die();
+}
+
+$fp = fopen('php://output', 'w');
+if ($fp && $db) {
+ header('Content-Type: text/csv; charset=utf-8');
+ header('Content-Disposition: attachment; filename="export.' . date("c") . '.csv"');
+ header('Pragma: no-cache');
+ header('Expires: 0');
+ fputcsv($fp, $headers);
+ try {
+ $agencies = $db->get_view("app", "byCanonicalName", null, true)->rows;
+ //print_r($rows);
+ foreach ($agencies as $agency) {
+ // print_r($agency);
+
+ if (isset($agency->value->foiEmail) && $agency->value->foiEmail != "null" && !isset($agency->value->status)) {
+ $row = Array();
+ $row["#id"] = $agency->id;
+ $row["name"] = trim($agency->value->name);
+ if (isset($agency->value->foiEmail)) {
+ $row["request_email"] = $agency->value->foiEmail;
+ } else {
+ if ($agency->value->orgType == "FMA-DepartmentOfState") {
+ $row["request_email"] = "foi@" . GetDomain($agency->value->website);
+ } else {
+ $row["request_email"] = $foiEmail[$agency->value->parentOrg];
+ }
+ }
+ if (isset($agency->value->shortName)) {
+ $row["short_name"] = $agency->value->shortName;
+ } else {
+ $row["short_name"] = shortName($agency->value->name);
+ }
+ $row["notes"] = "";
+ $row["publication_scheme"] = (isset($agency->value->infoPublicationSchemeURL) ? $agency->value->infoPublicationSchemeURL : "");
+ $row["home_page"] = (isset($agency->value->website) ? $agency->value->website : "");
+ if ($agency->value->orgType == "FMA-DepartmentOfState") {
+ $row["tag_string"] = $tag[$agency->value->_id] . " " . $agency->value->orgType;
+ } else {
+ $row["tag_string"] = $tag[$agency->value->parentOrg] . " " . $agency->value->orgType;
+ }
+
+ fputcsv($fp, array_values($row));
+
+ if (isset($agency->value->foiBodies)) {
+ foreach ($agency->value->foiBodies as $foiBody) {
+ $row['name'] = iconv("UTF-8", "ASCII//TRANSLIT",$foiBody);
+ $row["short_name"] = shortName($foiBody);
+ fputcsv($fp, array_values($row));
+ }
+ }
+ }
+ }
+ } catch (SetteeRestClientException $e) {
+ setteErrorHandler($e);
+ }
+
+ die;
+}
+?>
+
--- /dev/null
+++ b/alaveteli/exportCategories.rb.php
@@ -1,1 +1,24 @@
+<?php
+include_once("../include/common.inc.php");
+setlocale(LC_CTYPE, 'C');
+header('Content-Type: text/csv');
+header('Content-Disposition: attachment; filename="public_body_categories_en.rb"');
+header('Pragma: no-cache');
+header('Expires: 0');
+echo 'PublicBodyCategories.add(:en, [' . PHP_EOL;
+echo ' "Portfolios",' . PHP_EOL;
+$db = $server->get_db('disclosr-agencies');
+
+try {
+ $rows = $db->get_view("app", "byDeptStateName", null, true)->rows;
+ //print_r($rows);
+ foreach ($rows as $row) {
+ echo ' [ "' . phrase_to_tag(dept_to_portfolio($row->key)) . '","' . dept_to_portfolio($row->key) . '","part of the ' . dept_to_portfolio($row->key) . ' portfolio" ],' . PHP_EOL;
+ }
+} catch (SetteeRestClientException $e) {
+ setteErrorHandler($e);
+}
+echo '])';
+?>
+
--- /dev/null
+++ b/charts.php
@@ -1,1 +1,82 @@
+<?php
+include_once('include/common.inc.php');
+include_header();
+$db = $server->get_db('disclosr-agencies');
+?>
+<div class="foundation-header">
+ <h1><a href="about.php">Charts</a></h1>
+ <h4 class="subheader">Lorem ipsum.</h4>
+</div>
+<div id="placeholder" style="width:900px;height:500px;"></div>
+<script id="source">
+ window.onload = function() {
+ $(document).ready(function() {
+ var d1 = [];
+ var labels = [];
+ <?php
+ try {
+ $rows = $db->get_view("app", "scoreHas?group=true", null, true)->rows;
+
+ $dataValues = Array();
+ foreach ($rows as $row) {
+ $dataValues[$row->value] = $row->key;
+ }
+ $i = 0;
+ ksort($dataValues);
+ foreach ($dataValues as $value => $key) {
+
+ echo " d1.push([$i, $value]);" . PHP_EOL;
+ echo " labels.push('$key');" . PHP_EOL;
+ $i++;
+ }
+ } catch (SetteeRestClientException $e) {
+ setteErrorHandler($e);
+ }
+ ?>
+ function trackformatter(obj) {
+ if (labels[Math.floor(obj.x)]) {
+ return (labels[Math.floor(obj.x)])+"="+obj.y;
+
+ } else {
+ return "";
+ }
+ }
+ function tickformatter(val, axis) {
+ if (labels[Math.floor(val)]) {
+ return '<p style="margin-top:8em;-webkit-transform:rotate(-90deg);">'+(labels[Math.floor(val)])+"</b>";
+
+ } else {
+ return "";
+ }
+ }
+ Flotr.draw(document.getElementById("placeholder"), [ {data: d1}], {
+ HtmlText: true,
+ bars : {
+ show : true
+ },
+ mouse : {
+ track : true,
+ relative : true,
+ trackFormatter: trackformatter
+ },yaxis: {
+ min:0
+ },
+ xaxis: {
+
+ minorTickFreq: 0.6,
+ noTicks : 19,
+ tickFormatter: tickformatter
+ }
+ });
+
+
+
+ });
+ };
+
+</script>
+
+<?php
+include_footer();
+?>
--- /dev/null
+++ b/couchdb/SetteeDatabase.class.php
@@ -1,1 +1,306 @@
-
+<?php
+
+/**
+* Databaase class.
+*/
+class SetteeDatabase {
+
+ /**
+ * Base URL of the CouchDB REST API
+ */
+ private $conn_url;
+
+ /**
+ * HTTP REST Client instance
+ */
+ protected $rest_client;
+
+ /**
+ * Name of the database
+ */
+ private $dbname;
+
+ /**
+ * Default constructor
+ */
+ function __construct($conn_url, $dbname) {
+ $this->conn_url = $conn_url;
+ $this->dbname = $dbname;
+ $this->rest_client = SetteeRestClient::get_instance($this->conn_url);
+ }
+
+
+ /**
+ * Get UUID from CouchDB
+ *
+ * @return
+ * CouchDB-generated UUID string
+ *
+ */
+ function gen_uuid() {
+ $ret = $this->rest_client->http_get('_uuids');
+ return $ret['decoded']->uuids[0]; // should never be empty at this point, so no checking
+ }
+
+ /**
+ * Create or update a document database
+ *
+ * @param $document
+ * PHP object, a PHP associative array, or a JSON String representing the document to be saved. PHP Objects and arrays are JSON-encoded automatically.
+ *
+ * <p>If $document has a an "_id" property set, it will be used as document's unique id (even for "create" operation).
+ * If "_id" is missing, CouchDB will be used to generate a UUID.
+ *
+ * <p>If $document has a "_rev" property (revision), document will be updated, rather than creating a new document.
+ * You have to provide "_rev" if you want to update an existing document, otherwise operation will be assumed to be
+ * one of creation and you will get a duplicate document exception from CouchDB. Also, you may not provide "_rev" but
+ * not provide "_id" since that is an invalid input.
+ *
+ * @param $allowRevAutoDetection
+ * Default: false. When true and _rev is missing from the document, save() function will auto-detect latest revision
+ * for a document and use it. This option is "false" by default because it involves an extra http HEAD request and
+ * therefore can make save() operation slightly slower if such auto-detection is not required.
+ *
+ * @return
+ * document object with the database id (uuid) and revision attached;
+ *
+ * @throws SetteeCreateDatabaseException
+ */
+ function save($document, $allowRevAutoDetection = false) {
+ if (is_string($document)) {
+ $document = json_decode($document);
+ }
+
+ // Allow passing of $document as an array (for syntactic simplicity and also because in JSON world it does not matter)
+ if(is_array($document)) {
+ $document = (object) $document;
+ }
+
+ if (empty($document->_id) && empty($document->_rev)) {
+ $id = $this->gen_uuid();
+ }
+ elseif (empty($document->_id) && !empty($document->_rev)) {
+ throw new SetteeWrongInputException("Error: You can not save a document with a revision provided, but missing id");
+ }
+ else {
+ $id = $document->_id;
+
+ if ($allowRevAutoDetection) {
+ try {
+ $rev = $this->get_rev($id);
+ } catch (SetteeRestClientException $e) {
+ // auto-detection may fail legitimately, if a document has never been saved before (new doc), so skipping error
+ }
+ if (!empty($rev)) {
+ $document->_rev = $rev;
+ }
+ }
+ }
+
+ $full_uri = $this->dbname . "/" . $this->safe_urlencode($id);
+ $document_json = json_encode($document, JSON_NUMERIC_CHECK);
+
+ $ret = $this->rest_client->http_put($full_uri, $document_json);
+
+ $document->_id = $ret['decoded']->id;
+ $document->_rev = $ret['decoded']->rev;
+
+ return $document;
+ }
+
+ /**
+ * @param $doc
+ * @param $name
+ * @param $content
+ * Content of the attachment in a string-buffer format. This function will automatically base64-encode content for
+ * you, so you don't have to do it.
+ * @param $mime_type
+ * Optional. Will be auto-detected if not provided
+ * @return void
+ */
+ public function add_attachment($doc, $name, $content, $mime_type = null) {
+ if (empty($doc->_attachments) || !is_object($doc->_attachments)) {
+ $doc->_attachments = new stdClass();
+ }
+
+ if (empty($mime_type)) {
+ $mime_type = $this->rest_client->content_mime_type($content);
+ }
+
+ $doc->_attachments->$name = new stdClass();
+ $doc->_attachments->$name->content_type = $mime_type;
+ $doc->_attachments->$name->data = base64_encode($content);
+ }
+
+ /**
+ * @param $doc
+ * @param $name
+ * @param $file
+ * Full path to a file (e.g. as returned by PHP's realpath function).
+ * @param $mime_type
+ * Optional. Will be auto-detected if not provided
+ * @return void
+ */
+ public function add_attachment_file($doc, $name, $file, $mime_type = null) {
+ $content = file_get_contents($file);
+ $this->add_attachment($doc, $name, $content, $mime_type);
+ }
+
+ /**
+ *
+ * Retrieve a document from CouchDB
+ *
+ * @throws SetteeWrongInputException
+ *
+ * @param $id
+ * Unique ID (usually: UUID) of the document to be retrieved.
+ * @return
+ * database document in PHP object format.
+ */
+ function get($id) {
+ if (empty($id)) {
+ throw new SetteeWrongInputException("Error: Can't retrieve a document without a uuid.");
+ }
+
+ $full_uri = $this->dbname . "/" . $this->safe_urlencode($id);
+$full_uri = str_replace("%3Frev%3D","?rev=",$full_uri);
+ $ret = $this->rest_client->http_get($full_uri);
+ return $ret['decoded'];
+ }
+
+ /**
+ *
+ * Get the latest revision of a document with document id: $id in CouchDB.
+ *
+ * @throws SetteeWrongInputException
+ *
+ * @param $id
+ * Unique ID (usually: UUID) of the document to be retrieved.
+ * @return
+ * database document in PHP object format.
+ */
+ function get_rev($id) {
+ if (empty($id)) {
+ throw new SetteeWrongInputException("Error: Can't query a document without a uuid.");
+ }
+
+ $full_uri = $this->dbname . "/" . $this->safe_urlencode($id);
+ $headers = $this->rest_client->http_head($full_uri);
+ if (empty($headers['Etag'])) {
+ throw new SetteeRestClientException("Error: could not retrieve revision. Server unexpectedly returned empty Etag");
+ }
+ $etag = str_replace('"', '', $headers['Etag']);
+ return $etag;
+ }
+
+ /**
+ * Delete a document
+ *
+ * @param $document
+ * a PHP object or JSON representation of the document that has _id and _rev fields.
+ *
+ * @return void
+ */
+ function delete($document) {
+ if (!is_object($document)) {
+ $document = json_decode($document);
+ }
+
+ $full_uri = $this->dbname . "/" . $this->safe_urlencode($document->_id) . "?rev=" . $document->_rev;
+ $this->rest_client->http_delete($full_uri);
+ }
+
+
+ /*----------------- View-related functions --------------*/
+
+ /**
+ * Create a new view or update an existing one.
+ *
+ * @param $design_doc
+ * @param $view_name
+ * @param $map_src
+ * Source code of the map function in Javascript
+ * @param $reduce_src
+ * Source code of the reduce function in Javascript (optional)
+ * @return void
+ */
+ function save_view($design_doc, $view_name, $map_src, $reduce_src = null) {
+ $obj = new stdClass();
+ $obj->_id = "_design/" . urlencode($design_doc);
+ $view_name = urlencode($view_name);
+ $obj->views->$view_name->map = $map_src;
+ if (!empty($reduce_src)) {
+ $obj->views->$view_name->reduce = $reduce_src;
+ }
+
+ // allow safe updates (even if slightly slower due to extra: rev-detection check).
+ return $this->save($obj, true);
+ }
+
+ /**
+ * Create a new view or update an existing one.
+ *
+ * @param $design_doc
+ * @param $view_name
+ * @param $key
+ * key parameter to a view. Can be a single value or an array (for a range). If passed an array, function assumes
+ * that first element is startkey, second: endkey.
+ * @param $descending
+ * return results in descending order. Please don't forget that if you are using a startkey/endkey, when you change
+ * order you also need to swap startkey and endkey values!
+ *
+ * @return void
+ */
+ function get_view($design_doc, $view_name, $key = null, $descending = false) {
+ $id = "_design/" . urlencode($design_doc);
+ $view_name = urlencode($view_name);
+ $id .= "/_view/$view_name";
+
+ $data = array();
+ if (!empty($key)) {
+ if (is_string($key)) {
+ $data = "key=" . '"' . $key . '"';
+ }
+ elseif (is_array($key)) {
+ list($startkey, $endkey) = $key;
+ $data = "startkey=" . '"' . $startkey . '"&' . "endkey=" . '"' . $endkey . '"';
+ }
+
+ if ($descending) {
+ $data .= "&descending=true";
+ }
+ }
+
+
+
+ if (empty($id)) {
+ throw new SetteeWrongInputException("Error: Can't retrieve a document without a uuid.");
+ }
+
+ $full_uri = $this->dbname . "/" . $this->safe_urlencode($id);
+$full_uri = str_replace("%253Fgroup%253Dtrue","?group=true",$full_uri);
+ $ret = $this->rest_client->http_get($full_uri, $data);
+ return $ret['decoded'];
+
+ }
+
+ /**
+ * @param $id
+ * @return
+ * return a properly url-encoded id.
+ */
+ private function safe_urlencode($id) {
+ //-- System views like _design can have "/" in their URLs.
+ $id = rawurlencode($id);
+ if (substr($id, 0, 1) == '_') {
+ $id = str_replace('%2F', '/', $id);
+ }
+ return $id;
+ }
+
+ /** Getter for a database name */
+ function get_name() {
+ return $this->dbname;
+ }
+
+}
--- /dev/null
+++ b/couchdb/couchdb-lucene-run.bat.txt
@@ -1,1 +1,18 @@
+@echo off
+set CLASSPATH="../conf;"
+Echo Cleared classpath
+
+set JAVA_OPTS="-server -Xmx1g"
+Echo set Java Opts %JAVA_OPTS%
+
+set CLASS=com.github.rnewson.couchdb.lucene.Main
+Echo set couchdb-lucene class %CLASS%
+
+setlocal ENABLEDELAYEDEXPANSION
+if defined CLASSPATH (set CLASSPATH=%CLASSPATH%;.) else (set CLASSPATH=.)
+FOR /R ..\lib %%G IN (*.jar) DO set CLASSPATH=!CLASSPATH!;%%G
+Echo The Classpath definition is %CLASSPATH%
+
+java "%JAVA_OPTS%" -cp %CLASSPATH% %CLASS%
+
--- a/couchdb/settee
+++ b/couchdb/settee
--- a/getAgency.php
+++ b/getAgency.php
@@ -4,16 +4,36 @@
include_header();
function displayValue($key, $value, $mode) {
+ global $db, $schemas;
if ($mode == "view") {
+
+ echo "<tr>";
+
+ echo "<td>" . $schemas['agency']["properties"][$key]['x-title'] . "<br><small>" . $schemas['agency']["properties"][$key]['description'] . "</small></td><td>";
if (is_array($value)) {
- echo "<tr><td>$key</td><td><ol>";
+ echo "<ol>";
foreach ($value as $subkey => $subvalue) {
- echo "<li>$subvalue</li>";
+ if (isset($schemas['agency']["properties"][$key]['x-property'])) {
+ echo '<li property="' . $schemas['agency']["properties"][$key]['x-property'] . '">';
+ } else {
+ echo "<li>";
+ }
+ echo "$subvalue</li>";
}
echo "</ol></td></tr>";
} else {
- echo "<tr><td>$key</td><td>$value</td></tr>";
+ if (isset($schemas['agency']["properties"][$key]['x-property'])) {
+ echo '<span property="' . $schemas['agency']["properties"][$key]['x-property'] . '">';
+ } else {
+ echo "<span>";
+ }
+ if ((strpos($key, "URL") > 0 || $key == 'website') && $value != "") {
+ echo "<a href='$value'>view</a></span>";
+ } else {
+ echo "$value</span>";
+ }
}
+ echo "</td></tr>";
}
if ($mode == "edit") {
if (is_array($value)) {
@@ -30,15 +50,21 @@
} else {
if (strpos($key, "_") === 0) {
echo"<input type='hidden' id='$key' name='$key' value='$value'/>";
- } if (strpos($key, "has") === 0) {
- echo "<label for='$key'><input type='checkbox' id='$key' name='$key' checked='$value'> $key</label>";
- } else {
+ } else if ($key == "parentOrg") {
+ echo "<label for='$key'>$key</label><select id='$key' name='$key'><option value=''> Select... </option>";
+ $rows = $db->get_view("app", "byDeptStateName")->rows;
+ //print_r($rows);
+ foreach ($rows as $row) {
+ echo "<option value='{$row->value}'" . (($row->value == $value) ? "SELECTED" : "") . " >" . str_replace("Department of ", "", $row->key) . "</option>";
+ }
+ echo" </select>";
+ } else {
echo "<label>$key</label><input class='input-text' type='text' id='$key' name='$key' value='$value'/>";
- if ((strpos($key,"URL") > 0 || $key == 'website')&& $value != "") {
- echo "<a href='$value'>view</a>";
+ if ((strpos($key, "URL") > 0 || $key == 'website') && $value != "") {
+ echo "<a href='$value'>view</a>";
}
if ($key == 'abn') {
- echo "<a href='http://www.abr.business.gov.au/SearchByAbn.aspx?SearchText=$value'>view abn</a>";
+ echo "<a href='http://www.abr.business.gov.au/SearchByAbn.aspx?SearchText=$value'>view abn</a>";
}
}
}
@@ -47,10 +73,20 @@
}
function addDefaultFields($row) {
- $defaultFields = Array("name");
+ global $schemas;
+ $defaultFields = array_keys($schemas['agency']['properties']);
foreach ($defaultFields as $defaultField) {
- if (!isset($row[$defaultField]))
- $row[$defaultField] = "";
+ if (!isset($row[$defaultField])) {
+ if ($schemas['agency']['properties'][$defaultField]['type'] == "string") {
+
+ $row[$defaultField] = "";
+
+ }
+ if ($schemas['agency']['properties'][$defaultField]['type'] == "array") {
+
+ $row[$defaultField] = Array("");
+ }
+ }
}
return $row;
}
@@ -66,20 +102,33 @@
//print_r($row);
if (sizeof($_POST) > 0) {
//print_r($_POST);
+ foreach ($_POST as $postkey => $postvalue) {
+ if ($postvalue == "") {
+ unset($_POST[$postkey]);
+ }
+ if (is_array($postvalue) && count($postvalue) == 1 && $postvalue[0] == "") {
+ unset($_POST[$postkey]);
+ }
+ }
if (isset($_POST['_id']) && $db->get_rev($_POST['_id']) == $_POST['_rev']) {
echo "Edited version was latest version, continue saving";
$newdoc = $_POST;
$newdoc['metadata']['lastModified'] = time();
$row = $db->save($newdoc);
} else {
- echo "ALERT doc revised by someone else while editing.";
+ echo "ALERT doc revised by someone else while editing. Document not saved.";
}
}
$mode = "edit";
- $row = addDefaultFields(object_to_array($row));
+ if ($mode == "edit") {
+ $row = addDefaultFields(object_to_array($row));
+ } else {
+ $row = object_to_array($row);
+ }
+
if ($mode == "view") {
- echo '<table width="100%">';
+ echo '<div typeof="schema:GovernmentOrganisation" about="#' . $row['_id'] . '"><table width="100%">';
echo '<tr> <td colspan="2"><h3>' . $row['name'] . "</h3></td></tr>";
echo "<tr><th>Field Name</th><th>Field Value</th></tr>";
}
@@ -104,32 +153,44 @@
};
</script>
<form id="editform" class="nice" method="post">
- <?php
+ <?php
+ }
+ foreach ($row as $key => $value) {
+ echo displayValue($key, $value, $mode);
+ }
+ if ($mode == "view") {
+ echo "</table></div>";
+ }
+ if ($mode == "edit") {
+ echo '<input id="submitbutton" type="submit"/></form>';
+ }
+} else {
+
+ try {
+ /* $rows = $db->get_view("app", "showNamesABNs")->rows;
+ //print_r($rows);
+ foreach ($rows as $row) {
+ // print_r($row);
+ echo '<li><a href="getAgency.php?id=' . $row->key . '">' .
+ (isset($row->value->name) && $row->value->name != "" ? $row->value->name : "NO NAME " . $row->value->abn)
+ . '</a></li>';
+ } */
+ $rows = $db->get_view("app", "byName")->rows;
+ //print_r($rows);
+echo '<ul>';
+ foreach ($rows as $row) {
+ // print_r($row);
+ echo '<li typeof="schema:GovernmentOrganisation foaf:Organization" about="getAgency.php?id=' . $row->value . '">
+<a href="getAgency.php?id=' . $row->value . '" rel="schema:url foaf:page" property="schema:name foaf:name">' .
+ $row->key
+ . '</a></li>';
}
- foreach ($row as $key => $value) {
- echo displayValue($key, $value, $mode);
- }
- if ($mode == "view") {
- echo "</table>";
- }
- if ($mode == "edit") {
- echo '<input id="submitbutton" type="submit"/></form>';
- }
- } else {
+echo "</ul>";
+ } catch (SetteeRestClientException $e) {
+ setteErrorHandler($e);
+ }
+}
+include_footer();
+?>
- try {
- $rows = $db->get_view("app", "showNamesABNs")->rows;
- //print_r($rows);
- foreach ($rows as $row) {
- // print_r($row);
- echo '<li><a href="getAgency.php?id=' . $row->key . '">' .
- (isset($row->value->name) && $row->value->name != "" ? $row->value->name : "NO NAME " . $row->value->abn)
- . '</a></li>';
- }
- } catch (SetteeRestClientException $e) {
- setteErrorHandler($e);
- }
- }
- include_footer();
- ?>
--- /dev/null
+++ b/graph.php
@@ -1,1 +1,93 @@
+<?php
+include_once('include/common.inc.php');
+//include_header();
+$format = "html";
+if (isset($_REQUEST['format'])) {
+ $format = $_REQUEST['format'];
+}
+function add_node($id, $label) {
+ global $format;
+ if ($format == "html") {
+ echo "nodes[\"$id\"] = graph.newNode({label: \"$label\"});" . PHP_EOL;
+ }
+ if ($format == "dot" && $label != "") {
+ echo "$id [label=\"$label\"];". PHP_EOL;
+ }
+}
+
+function add_edge($from, $to, $color) {
+ global $format;
+ if ($format == "html") {
+ echo "graph.newEdge(nodes[\"$from\"], nodes['$to'], {color: '$color'});" . PHP_EOL;
+ }
+ if ($format == "dot") {
+ echo "$from -> $to ".($color != ""? "[color=$color]":"").";". PHP_EOL;
+ }
+}
+
+if ($format == "html") {
+ ?>
+ <script src="http://ajax.googleapis.com/ajax/libs/jquery/1.3.2/jquery.min.js"></script>
+ <script src="lib/springy/springy.js"></script>
+ <script src="lib/springy/springyui.js"></script>
+ <script>
+ var graph = new Graph();
+ var nodes = [];
+ <?php
+}
+if ($format == "dot") {
+ echo 'digraph g {'. PHP_EOL;
+}
+$db = $server->get_db('disclosr-agencies');
+ add_node("fedg","Federal Government - Commonwealth of Australia");
+try {
+ $rows = $db->get_view("app", "byCanonicalName", null, true)->rows;
+//print_r($rows);
+ foreach ($rows as $row) {
+ add_node($row->id, $row->key);
+ }
+} catch (SetteeRestClientException $e) {
+ setteErrorHandler($e);
+}
+
+try {
+ $rows = $db->get_view("app", "byDeptStateName", null, true)->rows;
+//print_r($rows);
+ foreach ($rows as $row) {
+ add_edge("fedg", $row->value, 'yellow');
+ }
+} catch (SetteeRestClientException $e) {
+ setteErrorHandler($e);
+}
+
+try {
+ $rows = $db->get_view("app", "parentOrgs", null, true)->rows;
+// print_r($rows);
+ foreach ($rows as $row) {
+ add_edge($row->key, $row->value, 'blue');
+ }
+} catch (SetteeRestClientException $e) {
+ setteErrorHandler($e);
+}
+if ($format == "html") {
+ ?>
+ window.onload = function() {
+ $(document).ready(function() {
+ var springy = $('#springydemo').springy({
+ graph: graph
+ });
+ });
+ };
+ </script>
+
+ <canvas id="springydemo" width="1260" height="680" />
+ <?php
+}
+if ($format == "dot") {
+ echo "}";
+}
+//include_footer();
+?>
+
+
--- a/import.php
+++ /dev/null
@@ -1,34 +1,1 @@
-<?php
-require_once 'include/common.inc.php';
-try {
- $server->create_db('disclosr-agencies');
-} catch (SetteeRestClientException $e) {
- setteErrorHandler($e);
-}
-$db = $server->get_db('disclosr-agencies');
-createAgencyDesignDoc();
-$conn = new PDO("pgsql:dbname=contractDashboard;user=postgres;password=snmc;host=localhost");
-$namesQ = 'select agency.abn, string_agg("agencyName",\'|\') as names from agency inner join agency_nametoabn on agency.abn::text = agency_nametoabn.abn group by agency.abn;';
-$abntonames = Array();
-foreach ($conn->query($namesQ) as $row) {
- $abntonames[$row['abn']] = explode("|", $row['names']);
-}
-$result = $conn->query("select * from agency");
-while ($agency = $result->fetch(PDO::FETCH_ASSOC)) {
- $agency['_id'] = md5($agency['abn']);
- $agency['otherNames'] = $abntonames[$agency['abn']];
- if (sizeof($abntonames[$agency['abn']]) == 1)
- $agency['name'] = $abntonames[$agency['abn']][0];
- $agency["lastScraped"] = "1/1/1970";
- $agency["scrapeDepth"] = 1;
- try {
- $doc = $db->save($agency);
- //print_r($doc);
- echo $agency['abn'] . " imported \n<br>";
- } catch (SetteeRestClientException $e) {
- setteErrorHandler($e);
- }
-}
-?>
-
--- a/include/common.inc.php
+++ b/include/common.inc.php
@@ -1,8 +1,19 @@
<?php
+
+date_default_timezone_set("Australia/Sydney");
+
+$basePath = "";
+if (strstr($_SERVER['PHP_SELF'], "alaveteli/")
+ || strstr($_SERVER['PHP_SELF'], "admin/")
+ || strstr($_SERVER['PHP_SELF'], "lib/")
+ || strstr($_SERVER['PHP_SELF'], "include/"))
+ $basePath = "../";
include_once ('couchdb.inc.php');
include_once ('template.inc.php');
+require_once $basePath.'lib/Requests/library/Requests.php';
+Requests::register_autoloader();
# Convert a stdClass to an Array. http://www.php.net/manual/en/language.types.object.php#102735
function object_to_array(stdClass $Class) {
@@ -33,7 +44,23 @@
# Typecast to (object) will automatically convert array -> stdClass
return (object) $array;
}
-?>
+function dept_to_portfolio($deptName) {
+ return trim(str_replace("Department of", "", str_replace("Department of the", "Department of", $deptName)));
+}
+function phrase_to_tag ($phrase) {
+ return str_replace(" ","_",str_replace("'","",str_replace(",","",strtolower($phrase))));
+}
+function GetDomain($url)
+{
+$nowww = ereg_replace('www\.','',$url);
+$domain = parse_url($nowww);
+if(!empty($domain["host"]))
+ {
+ return $domain["host"];
+ } else
+ {
+ return $domain["path"];
+ }
+}
-
--- a/include/couchdb.inc.php
+++ b/include/couchdb.inc.php
@@ -1,55 +1,165 @@
<?php
-include "schemas/schemas.inc.php";
+include $basePath . "schemas/schemas.inc.php";
+
+require ($basePath . 'couchdb/settee/src/settee.php');
+
+function createDocumentsDesignDoc() {
+ /*"views": {
+ "web_server": {
+ "map": "function(doc) {\n emit(doc.web_server, 1);\n}",
+ "reduce": "function (key, values, rereduce) {\n return sum(values);\n}"
+ },
+ "byAgency": {
+ "map": "function(doc) {\n emit(doc.agencyID, 1);\n}",
+ "reduce": "function (key, values, rereduce) {\n return sum(values);\n}"
+ },
+ "byURL": {
+ "map": "function(doc) {\n emit(doc.url, doc);\n}"
+ },
+ "agency": {
+ "map": "function(doc) {\n emit(doc.agencyID, doc);\n}"
+ },
+ "byWebServer": {
+ "map": "function(doc) {\n emit(doc.web_server, doc);\n}"
+ }
+ }*/
+}
function createAgencyDesignDoc() {
global $db;
$obj = new stdClass();
$obj->_id = "_design/" . urlencode("app");
$obj->language = "javascript";
+ $obj->views->all->map = "function(doc) { emit(doc._id, doc); };";
$obj->views->byABN->map = "function(doc) { emit(doc.abn, doc); };";
- $obj->views->byName->map = "function(doc) { emit(doc.name, doc);
+ $obj->views->byCanonicalName->map = "function(doc) {
+ if (doc.parentOrg || doc.orgType == 'FMA-DepartmentOfState') {
+ emit(doc.name, doc);
+ }
+};";
+ $obj->views->byDeptStateName->map = "function(doc) {
+ if (doc.orgType == 'FMA-DepartmentOfState') {
+ emit(doc.name, doc._id);
+ }
+};";
+ $obj->views->parentOrgs->map = "function(doc) {
+ if (doc.parentOrg) {
+ emit(doc._id, doc.parentOrg);
+ }
+};";
+ $obj->views->byName->map = 'function(doc) {
+ if (typeof(doc["status"]) == "undefined" || doc["status"] != "suspended") {
+ emit(doc.name, doc._id);
+if (typeof(doc.shortName) != "undefined" && doc.shortName != doc.name) {
+ emit(doc.shortName, doc._id);
+}
for (name in doc.otherNames) {
-if (doc.otherNames[name] != '' && doc.otherNames[name] != doc.name) {
- emit(doc.otherNames[name], doc);
+if (doc.otherNames[name] != "" && doc.otherNames[name] != doc.name) {
+ emit(doc.otherNames[name], doc._id);
}
}
+ for (name in doc.foiBodies) {
+if (doc.foiBodies[name] != "" && doc.foiBodies[name] != doc.name) {
+ emit(doc.foiBodies[name], doc._id);
+}
+ }
+ }
+};';
+
+ $obj->views->foiEmails->map = "function(doc) {
+ emit(doc._id, doc.foiEmail);
};";
+
$obj->views->byLastModified->map = "function(doc) { emit(doc.metadata.lastModified, doc); }";
$obj->views->getActive->map = 'function(doc) { if (doc.status == "active") { emit(doc._id, doc); } };';
$obj->views->getSuspended->map = 'function(doc) { if (doc.status == "suspended") { emit(doc._id, doc); } };';
- $obj->views->getScrapeRequired->map = "function(doc) { emit(doc.abn, doc); };";
+ $obj->views->getScrapeRequired->map = "function(doc) {
+
+var lastScrape = Date.parse(doc.metadata.lastScraped);
+
+var today = new Date();
+
+if (!lastScrape || lastScrape.getTime() + 1000 != today.getTime()) {
+ emit(doc._id, doc);
+}
+
+};";
$obj->views->showNamesABNs->map = "function(doc) { emit(doc._id, {name: doc.name, abn: doc.abn}); };";
+ $obj->views->getConflicts->map = "function(doc) {
+ if (doc._conflicts) {
+ emit(null, [doc._rev].concat(doc._conflicts));
+ }
+}";
// http://stackoverflow.com/questions/646628/javascript-startswith
- $obj->views->score->map = 'if(!String.prototype.startsWith){
+ $obj->views->scoreHas->map = 'if(!String.prototype.startsWith){
String.prototype.startsWith = function (str) {
return !this.indexOf(str);
}
}
-
+if(!String.prototype.endsWith){
+ String.prototype.endsWith = function(suffix) {
+ return this.indexOf(suffix, this.length - suffix.length) !== -1;
+ };
+}
function(doc) {
-count = 0;
if (typeof(doc["status"]) == "undefined" || doc["status"] != "suspended") {
for(var propName in doc) {
- if(typeof(doc[propName]) != "undefined" && propName.startsWith("l")) {
- count++
+ if(typeof(doc[propName]) != "undefined" && (propName.startsWith("has") || propName.endsWith("URL"))) {
+ emit(propName, 1);
}
}
- emit(count+doc._id, {id:doc._id, name: doc.name, score:count});
+ emit("total", 1);
}
}';
-
+ $obj->views->scoreHas->map = 'if(!String.prototype.startsWith){
+ String.prototype.startsWith = function (str) {
+ return !this.indexOf(str);
+ }
+}
+if(!String.prototype.endsWith){
+ String.prototype.endsWith = function(suffix) {
+ return this.indexOf(suffix, this.length - suffix.length) !== -1;
+ };
+}
+function(doc) {
+if (typeof(doc["status"]) == "undefined" || doc["status"] != "suspended") {
+for(var propName in doc) {
+ if(typeof(doc[propName]) != "undefined" && (propName.startsWith("has") || propName.endsWith("URL"))) {
+ emit(propName, 1);
+ }
+}
+ emit("total", 1);
+ }
+}';
+ $obj->views->scoreHas->reduce = 'function (key, values, rereduce) {
+ return sum(values);
+}';
+ $obj->views->fieldNames->map = '
+function(doc) {
+for(var propName in doc) {
+ emit(propName, doc._id);
+ }
+
+}';
+ $obj->views->fieldNames->reduce = 'function (key, values, rereduce) {
+ return values.length;
+}';
// allow safe updates (even if slightly slower due to extra: rev-detection check).
return $db->save($obj, true);
}
-require ('couchdb/settee/src/settee.php');
+if (php_uname('n') == "vanille") {
+$serverAddr = 'http://192.168.178.21:5984/';
+
+} else
+if (php_uname('n') == "KYUUBEY") {
-$server = new SetteeServer('http://192.168.1.8:5984');
-
+ $serverAddr = 'http://192.168.1.148:5984/';
+} else {
+ $serverAddr = 'http://127.0.0.1:5984/';
+}
+ $server = new SetteeServer($serverAddr);
function setteErrorHandler($e) {
echo $e->getMessage() . "<br>" . PHP_EOL;
}
-
-?>
-
--- a/include/template.inc.php
+++ b/include/template.inc.php
@@ -1,6 +1,7 @@
<?php
function include_header() {
+ global $basePath;
?>
<!DOCTYPE html>
@@ -18,11 +19,11 @@
<title>Disclosr</title>
<!-- Included CSS Files -->
- <link rel="stylesheet" href="stylesheets/foundation.css">
- <link rel="stylesheet" href="stylesheets/app.css">
+ <link rel="stylesheet" href="<?php echo $basePath ?>stylesheets/foundation.css">
+ <link rel="stylesheet" href="<?php echo $basePath ?>stylesheets/app.css">
<!--[if lt IE 9]>
- <link rel="stylesheet" href="stylesheets/ie.css">
+ <link rel="stylesheet" href="<?php echo $basePath ?>stylesheets/ie.css">
<![endif]-->
@@ -32,7 +33,7 @@
<![endif]-->
</head>
- <body>
+ <body xmlns:schema="http://schema.org/" xmlns:foaf="http://xmlns.com/foaf/0.1/">
<!-- navBar -->
<div id="navbar" class="container">
@@ -43,7 +44,7 @@
<div class="eight columns hide-on-phones">
<strong class="right">
<a href="getAgency.php">Agencies</a>
- <a href="about.php">About/FAQ</a>
+ <a href="about.php">About/FAQ</a>
</strong>
</div>
</div>
@@ -54,7 +55,10 @@
<div class="container">
<?php }
- function include_footer() { ?>
+ function include_footer() {
+ global $basePath;
+ ?>
+
</div>
<!-- container -->
@@ -62,14 +66,14 @@
<!-- Included JS Files -->
- <script src="javascripts/foundation.js"></script>
- <script src="javascripts/app.js"></script>
- <script src="http://code.jquery.com/jquery-1.7.1.min.js"></script>
+ <script src="<?php echo $basePath; ?>javascripts/foundation.js"></script>
+ <script src="<?php echo $basePath; ?>javascripts/app.js"></script>
+ <script src="http://code.jquery.com/jquery-1.7.1.min.js"></script>
+
+ <script type="text/javascript" src="javascripts/flotr2/flotr2.js"></script>
</body>
</html>
-<?php }
+<?php }
-?>
-
--- a/javascripts/app.js
+++ b/javascripts/app.js
@@ -43,7 +43,7 @@
/* PLACEHOLDER FOR FORMS ------------- */
/* Remove this and jquery.placeholder.min.js if you don't need :) */
- $('input, textarea').placeholder();
+ //$('input, textarea').placeholder();
--- /dev/null
+++ b/javascripts/flotr2
--- /dev/null
+++ b/lib/Requests
--- /dev/null
+++ b/lib/php-diff
--- /dev/null
+++ b/lib/springy
--- a/robots.txt
+++ b/robots.txt
@@ -2,3 +2,4 @@
# www.google.com/support/webmasters/bin/answer.py?hl=en&answer=156449
User-agent: *
+Disallow: /admin/
--- a/schemas/agency.json.php
+++ b/schemas/agency.json.php
@@ -4,16 +4,57 @@
"description" => "Representation of government agency and online transparency measures",
"type" => "object",
"properties" => Array(
- "name" => Array("type" => "string", "required" => true, "x-title"=> "Agency Name", "description" => "Agency Name, most recent and broadest"),
- "othernames" => Array("type" => "array", "required" => true, "x-title" => "Agency Past/Other Names", "description" => "Agency Names",
+ "name" => Array("type" => "string", "required" => true, "x-property" => "schema:name foaf:name", "x-title" => "Name", "description" => "Name, most recent and broadest"),
+ "shortName" => Array("type" => "string", "required" => false, "x-title" => "Short Name", "description" => "Name shortened, usually to an acronym"),
+ "foiEmail" => Array("type" => "string", "required" => false, "x-title" => "FOI Contact Email", "description" => "FOI contact email if not foi@"),
+ "sameAs" => Array("type" => "array", "required" => false, "x-property"=>"owl:sameAs","x-title" => "Same As", "description" => "Same as other URLs/URIs for this entity",
+ "items" => Array("type" => "string")),
+ "otherNames" => Array("type" => "array", "required" => true, "x-title" => "Past/Other Names", "description" => "Other names for organisation",
+ "items" => Array("type" => "string")),
+ "foiBodies" => Array("type" => "array", "required" => true, "x-title" => "FOI Bodies","x-property"=>"schema:members foaf:knows", "description" => "Organisational units within this agency that are subject to FOI Act but are not autonomous",
+ "items" => Array("type" => "string")),
+ "orgType" => Array("type" => "string", "required" => true, "x-title" => "Organisation Type", "description" => "Org type based on legal formation via FMA/CAC legislation etc."),
+ "parentOrg" => Array("type" => "string", "required" => true, "x-title" => "Parent Organisation", "description" => "Parent organisation, usually a department of state"),
+ "website" => Array("type" => "string", "required" => true, "x-title" => "Website", "x-property" => "schema:url foaf:homepage", "description" => "Website URL"),
+ "abn" => Array("type" => "string", "required" => true, "x-title" => "Australian Business Number", "description" => "ABN from business register"),
+ "contractListURL" => Array("type" => "string", "required" => true, "x-title" => "Contract Listing", "description" => "Departmental and agency contracts, <a href='http://www.aph.gov.au/senate/pubs/standing_orders/d05.htm'>mandated by the Senate</a>"),
+ "grantsReportingURL" => Array("type" => "string", "required" => true, "x-title" => "Grants Awarded",
+ "description" => "Departmental and agency grants <a href='http://www.aph.gov.au/senate/pubs/standing_orders/d05.htm'>mandated by the Senate</a> and <a href='http://www.finance.gov.au/publications/fmg-series/23-commonwealth-grant-guidelines.html'>Commonwealth grants guidelines</a> "),
+ "annualReportURL" => Array("type" => "string", "required" => true, "x-title" => "Annual Report(s)", "description" => ""),
+ "consultanciesURL" => Array("type" => "string", "required" => true, "x-title" => "Consultants Hired", "description" => ""),
+ "legalExpenditureURL" => Array("type" => "string", "required" => true, "x-title" => "Legal Services Expenditure", "description" => "Legal Services Expenditure mandated by Legal Services Directions 2005"),
+ "recordsListURL" => Array("type" => "string", "required" => true, "x-title" => "Files/Records Held", "description" => "Indexed lists of departmental and agency files, <a href='http://www.aph.gov.au/senate/pubs/standing_orders/d05.htm'>mandated by the Senate</a>"),
+ "FOIDocumentsURL" => Array("type" => "string", "required" => true, "x-title" => "FOI Documents Released", "description" => "FOI Disclosure Log URL"),
+ "FOIDocumentsRSSURL" => Array("type" => "string", "required" => false, "x-title" => "RSS Feed of FOI Documents Released", "description" => "FOI Disclosure Log in RSS format"),
+ "hasFOIPDF" => Array("type" => "array", "required" => false, "x-title" => "Has FOI Documents Released in PDF", "description" => "FOI Disclosure Log contains any PDFs",
+ "items" => Array("type" => "string")),
+ "infoPublicationSchemeURL" => Array("type" => "string", "required" => true, "x-title" => "Information Publication Scheme", "description" => ""),
+ "appointmentsURL" => Array("type" => "string", "required" => true, "x-title" => "Agency Appointments/Boards", "description" => "Departmental and agency appointments and vacancies , <a href='http://www.aph.gov.au/senate/pubs/standing_orders/d05.htm'>mandated by the Senate</a>"),
+ "advertisingURL" => Array("type" => "string", "required" => true, "x-title" => "Approved Advertising Campaigns", "description" => " Agency advertising and public information projects, <a href='http://www.aph.gov.au/senate/pubs/standing_orders/d05.htm'>mandated by the Senate</a> "),
+ "hasRSS" => Array("type" => "array", "required" => true, "x-title" => "Has RSS", "description" => ""),
+ "hasMailingList" => Array("type" => "array", "required" => true, "x-title" => "Has Mailing List", "description" => "",
+ "items" => Array("type" => "string")),
+ "hasTwitter" => Array("type" => "array", "required" => true, "x-title" => "Has Twitter", "description" => "",
+ "items" => Array("type" => "string")),
+ "hasFacebook" => Array("type" => "array", "required" => true, "x-title" => "Has Facebook", "description" => "",
+ "items" => Array("type" => "string")),
+ "hasYouTube" => Array("type" => "array", "required" => true, "x-title" => "Has YouTube", "description" => "",
+ "items" => Array("type" => "string")),
+ "hasFlickr" => Array("type" => "array", "required" => true, "x-title" => "Has Flickr", "description" => "",
+ "items" => Array("type" => "string")),
+ "hasCCBY" => Array("type" => "array", "required" => true, "x-title" => "Has CC-BY", "description" => "Has any page licenced Creative Commons - Attribution",
+ "items" => Array("type" => "string")),
+ "hasRestrictiveLicence" => Array("type" => "array","required" => true, "x-title" => "Has Restrictive Licence", "description" => "Has any page licenced under terms more restrictive than Crown Copyright",
+ "items" => Array("type" => "string")),
+ "hasCrownCopyright" => Array("type" => "array", "required" => true, "x-title" => "Has Standard Crown Copyright licence", "description" => "Has any page still licenced under the former Commonwealth Copyright Administration",
"items" => Array("type" => "string")),
),
- /*"org":{"type":"object",
- "properties":{
- "organizationName":{"type":"string"},
- "organizationUnit":{"type":"string"}},
- }
- }*/
+ /* "org":{"type":"object",
+ "properties":{
+ "organizationName":{"type":"string"},
+ "organizationUnit":{"type":"string"}},
+ }
+ } */
);
?>
--- /dev/null
+++ b/scrape.py
@@ -1,1 +1,198 @@
+#http://packages.python.org/CouchDB/client.html
+import couchdb
+import urllib2
+from BeautifulSoup import BeautifulSoup
+import re
+import hashlib
+from urlparse import urljoin
+import time
+import os
+import mimetypes
+import re
+import urllib
+import urlparse
+def canonurl(url):
+ r"""Return the canonical, ASCII-encoded form of a UTF-8 encoded URL, or ''
+ if the URL looks invalid.
+ >>> canonurl('\xe2\x9e\xa1.ws') # tinyarro.ws
+ 'http://xn--hgi.ws/'
+ """
+ # strip spaces at the ends and ensure it's prefixed with 'scheme://'
+ url = url.strip()
+ if not url:
+ return ''
+ if not urlparse.urlsplit(url).scheme:
+ url = 'http://' + url
+
+ # turn it into Unicode
+ #try:
+ # url = unicode(url, 'utf-8')
+ #except UnicodeDecodeError:
+ # return '' # bad UTF-8 chars in URL
+
+ # parse the URL into its components
+ parsed = urlparse.urlsplit(url)
+ scheme, netloc, path, query, fragment = parsed
+
+ # ensure scheme is a letter followed by letters, digits, and '+-.' chars
+ if not re.match(r'[a-z][-+.a-z0-9]*$', scheme, flags=re.I):
+ return ''
+ scheme = str(scheme)
+
+ # ensure domain and port are valid, eg: sub.domain.<1-to-6-TLD-chars>[:port]
+ match = re.match(r'(.+\.[a-z0-9]{1,6})(:\d{1,5})?$', netloc, flags=re.I)
+ if not match:
+ return ''
+ domain, port = match.groups()
+ netloc = domain + (port if port else '')
+ netloc = netloc.encode('idna')
+
+ # ensure path is valid and convert Unicode chars to %-encoded
+ if not path:
+ path = '/' # eg: 'http://google.com' -> 'http://google.com/'
+ path = urllib.quote(urllib.unquote(path.encode('utf-8')), safe='/;')
+
+ # ensure query is valid
+ query = urllib.quote(urllib.unquote(query.encode('utf-8')), safe='=&?/')
+
+ # ensure fragment is valid
+ fragment = urllib.quote(urllib.unquote(fragment.encode('utf-8')))
+
+ # piece it all back together, truncating it to a maximum of 4KB
+ url = urlparse.urlunsplit((scheme, netloc, path, query, fragment))
+ return url[:4096]
+
+#http://diveintopython.org/http_web_services/etags.html
+class NotModifiedHandler(urllib2.BaseHandler):
+ def http_error_304(self, req, fp, code, message, headers):
+ addinfourl = urllib2.addinfourl(fp, headers, req.get_full_url())
+ addinfourl.code = code
+ return addinfourl
+
+def fetchURL(docsdb, url, fieldName, agencyID, scrape_again=True):
+ url = canonurl(url)
+ hash = hashlib.md5(url).hexdigest().encode("utf-8")
+ req = urllib2.Request(url)
+ print "Fetching %s" % url
+ if url.startswith("mailto") or url.startswith("javascript") or url.startswith("#") or url == None or url == "":
+ print "Not a valid HTTP url"
+ return (None,None)
+ doc = docsdb.get(hash)
+ if doc == None:
+ doc = {'_id': hash, 'agencyID': agencyID, 'url': url, 'fieldName':fieldName}
+ else:
+ if (('page_scraped' in doc) and (time.time() - doc['page_scraped']) < 999999):
+ print "Uh oh, trying to scrape URL again too soon!"
+ last_attachment_fname = doc["_attachments"].keys()[-1]
+ last_attachment = docsdb.get_attachment(doc,last_attachment_fname)
+ return (doc['mime_type'],last_attachment)
+ if scrape_again == False:
+ print "Not scraping this URL again as requested"
+ return (None,None)
+
+ time.sleep(3) # wait 3 seconds to give webserver time to recover
+
+ #if there is a previous version stored in couchdb, load caching helper tags
+ if doc.has_key('etag'):
+ req.add_header("If-None-Match", doc['etag'])
+ if doc.has_key('last_modified'):
+ req.add_header("If-Modified-Since", doc['last_modified'])
+
+ opener = urllib2.build_opener(NotModifiedHandler())
+ try:
+ url_handle = opener.open(req)
+ headers = url_handle.info() # the addinfourls have the .info() too
+ doc['etag'] = headers.getheader("ETag")
+ doc['last_modified'] = headers.getheader("Last-Modified")
+ doc['date'] = headers.getheader("Date")
+ doc['page_scraped'] = time.time()
+ doc['web_server'] = headers.getheader("Server")
+ doc['powered_by'] = headers.getheader("X-Powered-By")
+ doc['file_size'] = headers.getheader("Content-Length")
+ content_type = headers.getheader("Content-Type")
+ if content_type != None:
+ doc['mime_type'] = content_type.split(";")[0]
+ else:
+ (type,encoding) = mimetypes.guess_type(url)
+ doc['mime_type'] = type
+ if hasattr(url_handle, 'code'):
+ if url_handle.code == 304:
+ print "the web page has not been modified"
+ return (None,None)
+ else:
+ content = url_handle.read()
+ docsdb.save(doc)
+ doc = docsdb.get(hash) # need to get a _rev
+ docsdb.put_attachment(doc, content, str(time.time())+"-"+os.path.b