Browse Source

Bump DVC version

Michael Ekstrand 4 months ago
parent
commit
621efa70d0
57 changed files with 64 additions and 311 deletions
  1. 0
    1
      Dvcfile
  2. 11
    11
      bookdata/dvcpatch.py
  3. 2
    5
      data/BX-CSV-Dump.zip.dvc
  4. 1
    10
      data/BX.dvc
  5. 0
    4
      data/goodreads_book_authors.json.gz.dvc
  6. 0
    4
      data/goodreads_book_genres_initial.json.gz.dvc
  7. 0
    4
      data/goodreads_book_works.json.gz.dvc
  8. 0
    4
      data/goodreads_books.json.gz.dvc
  9. 0
    4
      data/goodreads_interactions.json.gz.dvc
  10. 1
    4
      data/loc-books.dvc
  11. 2
    5
      data/loc-listings.dvc
  12. 1
    4
      data/loc-names.dvc
  13. 2
    5
      data/ol_dump_authors.txt.gz.dvc
  14. 2
    5
      data/ol_dump_editions.txt.gz.dvc
  15. 2
    5
      data/ol_dump_works.txt.gz.dvc
  16. 1
    5
      data/ratings_Books.csv.dvc
  17. 2
    5
      data/viaf-clusters-marc21.xml.gz.dvc
  18. 1
    1
      environment.yml
  19. 1
    6
      import/az-ratings.dvc
  20. 1
    6
      import/bx-ratings.dvc
  21. 1
    6
      import/gr-authors.dvc
  22. 1
    6
      import/gr-book-genres.dvc
  23. 1
    6
      import/gr-books.dvc
  24. 1
    6
      import/gr-interactions.dvc
  25. 1
    6
      import/gr-works.dvc
  26. 1
    6
      import/loc-mds-books.dvc
  27. 1
    6
      import/loc-mds-names.dvc
  28. 1
    6
      import/ol-authors.dvc
  29. 1
    6
      import/ol-editions.dvc
  30. 1
    6
      import/ol-works.dvc
  31. 1
    6
      import/viaf.dvc
  32. 1
    6
      index/az-index.dvc
  33. 1
    6
      index/bx-index.dvc
  34. 1
    6
      index/gr-book-info.dvc
  35. 1
    6
      index/gr-index-books.dvc
  36. 1
    6
      index/gr-index-ratings.dvc
  37. 1
    6
      index/isbn-norm.dvc
  38. 1
    6
      index/loc-mds-book-info.dvc
  39. 1
    6
      index/loc-mds-extract-isbns.dvc
  40. 1
    6
      index/loc-mds-index-books.dvc
  41. 1
    6
      index/loc-mds-index-names.dvc
  42. 1
    6
      index/ol-book-info.dvc
  43. 1
    6
      index/ol-index.dvc
  44. 1
    6
      index/viaf-index.dvc
  45. 0
    3
      init.dvc
  46. 1
    6
      integrate/author-info.dvc
  47. 1
    6
      integrate/author-stats.dvc
  48. 1
    6
      integrate/cluster-stats.dvc
  49. 1
    9
      integrate/cluster.dvc
  50. 0
    1
      schemas.dvc
  51. 1
    6
      schemas/az-schema.dvc
  52. 1
    6
      schemas/bx-schema.dvc
  53. 0
    2
      schemas/common-schema.dvc
  54. 1
    6
      schemas/gr-schema.dvc
  55. 1
    6
      schemas/loc-mds-schema.dvc
  56. 1
    6
      schemas/ol-schema.dvc
  57. 1
    6
      schemas/viaf-schema.dvc

+ 0
- 1
Dvcfile

@@ -11,4 +11,3 @@ deps:
   md5: 7c052a1b981bcc763cad9abde972639e
 - path: pgstat://loc-mds-index-names
   md5: 4d8d4e061447b2d53f39262c41e60f74
-md5: 86c498eff35f23796d3ef73fff157a36

+ 11
- 11
bookdata/dvcpatch.py

@@ -7,16 +7,16 @@ import logging
 from urllib.parse import urlparse
 import hashlib
 
-from dvc.remote.base import RemoteBASE
-from dvc.output.base import OutputBase
-from dvc.dependency.base import DependencyBase
+from dvc.remote.base import BaseRemoteTree
+from dvc.output.base import BaseOutput
+from dvc.dependency.base import BaseDependency
 
 from . import tracking
 
 _log = logging.getLogger('dvc.bgpatch')
 
 
-class PGRemote(RemoteBASE):
+class PGRemoteTree(BaseRemoteTree):
     """
     PG status remote
     """
@@ -26,7 +26,7 @@ class PGRemote(RemoteBASE):
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
 
-    def get_file_checksum(self, path_info):
+    def get_file_hash(self, path_info):
         _log.debug('checksum from %s', path_info)
         status = tracking.stage_status(path_info.bucket)
         h = hashlib.md5()
@@ -50,20 +50,20 @@ class PGRemote(RemoteBASE):
         raise NotImplementedError()
 
 
-class PGOut(OutputBase):
-    REMOTE = PGRemote
+class PGOutput(BaseOutput):
+    TREE_CLS = PGRemoteTree
 
 
-class PGDep(DependencyBase, OutputBase):
-    REMOTE = PGRemote
+class PGDep(BaseDependency, PGOutput):
+    pass
 
 
 def patch():
     "Patch DVC to include our classes"
 
     import dvc.output, dvc.dependency, dvc.config
-    dvc.output.OUTS.append(PGOut)
-    dvc.output.OUTS_MAP['pgstat'] = PGOut
+    dvc.output.OUTS.append(PGOutput)
+    dvc.output.OUTS_MAP['pgstat'] = PGOutput
 
     dvc.dependency.DEPS.append(PGDep)
     dvc.dependency.DEP_MAP['pgstat'] = PGDep

+ 2
- 5
data/BX-CSV-Dump.zip.dvc

@@ -1,11 +1,8 @@
-md5: f1914018a914e6d2e6f9efafd3b9f9a9
-locked: true
+md5: d9cf4e55b584b38ab1f20f9577e76cf7
 deps:
 - etag: '"18e0884-3e62cb8b5be80"'
   path: http://www2.informatik.uni-freiburg.de/~cziegler/BX/BX-CSV-Dump.zip
 outs:
 - md5: 37d647ee9e18ba134ea6d78ee4fe5292
   path: BX-CSV-Dump.zip
-  cache: true
-  metric: false
-  persist: false
+frozen: true

+ 1
- 10
data/BX.dvc

@@ -1,4 +1,4 @@
-md5: 3263fde487668e0e552206f23e341771
+md5: 133918ea0060237ae6ce943420a6d5a2
 cmd: unzip BX-CSV-Dump.zip
 deps:
 - md5: 37d647ee9e18ba134ea6d78ee4fe5292
@@ -6,16 +6,7 @@ deps:
 outs:
 - md5: b34fe0534c9b846b8a45f316c60eb92b
   path: BX-Book-Ratings.csv
-  cache: true
-  metric: false
-  persist: false
 - md5: d50d59b0c40f10d37d379d9fd3fb98ac
   path: BX-Books.csv
-  cache: true
-  metric: false
-  persist: false
 - md5: 6ec3fe6463da9e149d474ed8226612a2
   path: BX-Users.csv
-  cache: true
-  metric: false
-  persist: false

+ 0
- 4
data/goodreads_book_authors.json.gz.dvc

@@ -1,7 +1,3 @@
-md5: 7ec9939e301570c59b3314ea8a8b4e7d
 outs:
 - md5: b193c3febd961fb69443b65ba05b83a7
   path: goodreads_book_authors.json.gz
-  cache: true
-  metric: false
-  persist: false

+ 0
- 4
data/goodreads_book_genres_initial.json.gz.dvc

@@ -1,7 +1,3 @@
-md5: b5c85eb5ad4421aac6b89f26a374924a
 outs:
 - md5: 99ee3d1cadd68818c3dd0ef0d2f10602
   path: goodreads_book_genres_initial.json.gz
-  cache: true
-  metric: false
-  persist: false

+ 0
- 4
data/goodreads_book_works.json.gz.dvc

@@ -1,7 +1,3 @@
-md5: f4c6c15096988d8543872be20c1bf26c
 outs:
 - md5: e80738a88d02d2b0081cd249d9b4f081
   path: goodreads_book_works.json.gz
-  cache: true
-  metric: false
-  persist: false

+ 0
- 4
data/goodreads_books.json.gz.dvc

@@ -1,7 +1,3 @@
-md5: 2c3ce46885286a8a5261e39a5d5dcab0
 outs:
 - md5: 01b40c70a00fb6aa321ee478f0fd0d6b
   path: goodreads_books.json.gz
-  cache: true
-  metric: false
-  persist: false

+ 0
- 4
data/goodreads_interactions.json.gz.dvc

@@ -1,7 +1,3 @@
-md5: 485af93cf682c58090becbc7d91a190b
 outs:
 - md5: f2d054a85f33d405a9bff6933005ba89
   path: goodreads_interactions.json.gz
-  cache: true
-  metric: false
-  persist: false

+ 1
- 4
data/loc-books.dvc

@@ -4,8 +4,5 @@ deps:
   md5: 695c2898d88df8eb50f086488a492433
 outs:
 - path: loc-books
-  cache: true
-  metric: false
-  persist: false
   md5: 1b1e7ab1d98cc81e373dfc53345d4bb7.dir
-md5: e332b4fba51532c2ac17dd38fa7f509f
+md5: a2e9f1c6e01f4140d88ce51534b77133

+ 2
- 5
data/loc-listings.dvc

@@ -1,10 +1,7 @@
-md5: 6b3d23c7e9c9a8a651a261f510bd0403
+md5: 9b1ae0e4404499438b43a7d9276ca520
 cmd: python run.py loc.list-files https://www.loc.gov/cds/downloads/MDSConnect/ data/loc-listings
 wdir: ..
 outs:
 - md5: a758253fe6f30d4adad8b208082fc517.dir
   path: data/loc-listings
-  cache: true
-  metric: false
-  persist: false
-locked: true
+frozen: true

+ 1
- 4
data/loc-names.dvc

@@ -4,8 +4,5 @@ deps:
   md5: c041ad93309f581e602f54bf4994ef78
 outs:
 - path: loc-names
-  cache: true
-  metric: false
-  persist: false
   md5: fc488a8775561070cced774803fe0d72.dir
-md5: 62e0127700898e8c3d56c7930d442854
+md5: eb8d044fe51899bb08dc3e9738ee5a0a

+ 2
- 5
data/ol_dump_authors.txt.gz.dvc

@@ -1,11 +1,8 @@
-md5: 228c5718410bc1d5ba7d35c3c50dcf8b
-locked: true
+md5: 347dc074a94adde3ed92d95ea589882a
 deps:
 - etag: '"5e5beea4-12b31489"'
   path: https://openlibrary.org/data/ol_dump_authors_2020-02-29.txt.gz
 outs:
 - md5: 364e02a44e9e9a572e88692fc78fef27
   path: ol_dump_authors.txt.gz
-  cache: true
-  metric: false
-  persist: false
+frozen: true

+ 2
- 5
data/ol_dump_editions.txt.gz.dvc

@@ -1,11 +1,8 @@
-md5: de1d3715fd0fee6677b32444371369bc
-locked: true
+md5: 1fb3567c86abebd6e4bebb1969ede0c2
 deps:
 - etag: '"5e5beea4-16a78f153"'
   path: https://openlibrary.org/data/ol_dump_editions_2020-02-29.txt.gz
 outs:
 - md5: e105295bf5f8025ecd7e43838ed0739c
   path: ol_dump_editions.txt.gz
-  cache: true
-  metric: false
-  persist: false
+frozen: true

+ 2
- 5
data/ol_dump_works.txt.gz.dvc

@@ -1,11 +1,8 @@
-md5: b167e336efc0357620063abcff6b214d
-locked: true
+md5: 0aac1c536ff4f7ab90dcbb69913c4138
 deps:
 - etag: '"5e5beea4-65201a7b"'
   path: https://openlibrary.org/data/ol_dump_works_2020-02-29.txt.gz
 outs:
 - md5: 84e236955e5f683adde6a677d80475a0
   path: ol_dump_works.txt.gz
-  cache: true
-  metric: false
-  persist: false
+frozen: true

+ 1
- 5
data/ratings_Books.csv.dvc

@@ -1,8 +1,4 @@
-md5: 3102d1aad1e8e7e0dcc38559bb6c8a24
-locked: true
 outs:
 - md5: 77b4a5b887e14e16b13e7788bdf70156
   path: ratings_Books.csv
-  cache: true
-  metric: false
-  persist: false
+frozen: true

+ 2
- 5
data/viaf-clusters-marc21.xml.gz.dvc

@@ -1,9 +1,6 @@
 cmd: aria2c -o viaf-clusters-marc21.xml.gz http://viaf.org/viaf/data/viaf-20191007-clusters-marc21.xml.gz
 outs:
 - path: viaf-clusters-marc21.xml.gz
-  cache: true
-  metric: false
-  persist: false
   md5: 2f1af5262584d38f7331d333dec81cc3
-md5: bb76127e27e8301ad17a6361e4690c7c
-locked: true
+md5: f78eb89cc411792885908b9c6940f166
+frozen: true

+ 1
- 1
environment.yml

@@ -25,7 +25,7 @@ dependencies:
 - requests
 - html5lib
 - gitpython
-- dvc>=0.90
+- dvc>=1.1
 - boto3
 - pip:
   - chromalog

+ 1
- 6
import/az-ratings.dvc

@@ -1,4 +1,4 @@
-md5: 343fe7fc28fb0772a52668049d480d16
+md5: 3659a42512b1273d52751f019a7ed365
 cmd: python run.py --rust pcat -t az.raw_ratings -s az-ratings -T import/az-ratings.transcript
   -D az-schema -f CSV data/ratings_Books.csv
 wdir: ..
@@ -11,10 +11,5 @@ outs:
 - path: pgstat://az-ratings
   cache: false
   md5: 08a44188987803be46be4286a2e5a5f3
-  metric: false
-  persist: false
 - md5: 4266a96f87e134dac6b60df0514a36d2
   path: import/az-ratings.transcript
-  cache: true
-  metric: false
-  persist: false

+ 1
- 6
import/bx-ratings.dvc

@@ -1,4 +1,4 @@
-md5: 683ad95cff8b14edbddf50817e231d4b
+md5: 039228477cc645453b79fb5fd70ec1f1
 cmd: python run.py bx-import -T import/bx-ratings.transcript data/BX-Book-Ratings.csv
 wdir: ..
 deps:
@@ -10,10 +10,5 @@ outs:
 - path: pgstat://bx-ratings
   cache: false
   md5: 9f1009bfdb68267b39ac94ff3d754756
-  metric: false
-  persist: false
 - md5: 511096523052e1563c4bdcc66b6ed143
   path: import/bx-ratings.transcript
-  cache: true
-  metric: false
-  persist: false

+ 1
- 6
import/gr-authors.dvc

@@ -12,11 +12,6 @@ outs:
 - path: pgstat://gr-authors
   cache: false
   md5: b7701864036091016af9123c02b5d7ed
-  metric: false
-  persist: false
 - path: import/gr-authors.transcript
-  cache: true
-  metric: false
-  persist: false
   md5: 40acfcc81a4be3363d61febed1716f00
-md5: 4141333ba8de2ec2b0a7b3c48d50e3a1
+md5: d40df1ed5cbf385bd9f721a3a601e76c

+ 1
- 6
import/gr-book-genres.dvc

@@ -12,11 +12,6 @@ outs:
 - path: pgstat://gr-book-genres
   cache: false
   md5: 0c77c736582157805e8248a97b5f7037
-  metric: false
-  persist: false
 - path: import/gr-book-genres.transcript
-  cache: true
-  metric: false
-  persist: false
   md5: 5cd443ca86c79bef13040a97aa135eb3
-md5: 6585ccd5fddaac688da54bf804bcbb16
+md5: 21e97b98d657e26e73527f4db20c5d31

+ 1
- 6
import/gr-books.dvc

@@ -12,11 +12,6 @@ outs:
 - path: pgstat://gr-books
   cache: false
   md5: a0dcde1044f2c61895def1a9523be067
-  metric: false
-  persist: false
 - path: import/gr-books.transcript
-  cache: true
-  metric: false
-  persist: false
   md5: f493bf8e51b22ccdf258b07141c0a79d
-md5: aff267468fbbc899d2d9e17487d3282b
+md5: 3a6a5f8f12f4720e1a2e629cbc62001e

+ 1
- 6
import/gr-interactions.dvc

@@ -12,11 +12,6 @@ outs:
 - path: pgstat://gr-interactions
   cache: false
   md5: acf7d8bdbf506c81f6c3ea4eded3d702
-  metric: false
-  persist: false
 - path: import/gr-interactions.transcript
-  cache: true
-  metric: false
-  persist: false
   md5: 83ef4771b7c84a5730ba94fe8580a261
-md5: 797211e75a7ea96b6d4d5de89f5e82ef
+md5: 992aa377a425cdd4d8d4aba5f77119bc

+ 1
- 6
import/gr-works.dvc

@@ -12,11 +12,6 @@ outs:
 - path: pgstat://gr-works
   cache: false
   md5: a223984989927e62d8aea9230810d6bb
-  metric: false
-  persist: false
 - path: import/gr-works.transcript
-  cache: true
-  metric: false
-  persist: false
   md5: 2ff4ab63f7c9043b7526f85f9bb36028
-md5: c290cd956d9bed639ae2b493c6d28fdc
+md5: b500dc8755f32e5e2f3302dc5a4069aa

+ 1
- 6
import/loc-mds-books.dvc

@@ -1,4 +1,4 @@
-md5: ed37f6ac8cd504e6a01b1b7f72de9840
+md5: c05dd43881b82bbeffb23a7b6d3e1e96
 cmd: python run.py --rust parse-marc --db-schema locmds -t book_marc_field --truncate
   --stage loc-mds-books -D loc-mds-schema --transcript import/loc-mds-books.transcript
   --src-dir data/loc-books --src-prefix BooksAll.2016
@@ -12,10 +12,5 @@ outs:
 - path: pgstat://loc-mds-books
   cache: false
   md5: f6e0026b4d4fe4bac7056c7fe0491259
-  metric: false
-  persist: false
 - md5: 19b0a011c9053361278a3d812518b229
   path: import/loc-mds-books.transcript
-  cache: true
-  metric: false
-  persist: false

+ 1
- 6
import/loc-mds-names.dvc

@@ -1,4 +1,4 @@
-md5: 6565d7569b910b4571428067c278e563
+md5: a83ec7b726726e26f5084b80bf894c9f
 cmd: python run.py --rust parse-marc --db-schema locmds -t name_marc_field --truncate
   --stage loc-mds-names -D loc-mds-schema --transcript import/loc-mds-names.transcript
   --src-dir data/loc-names --src-prefix Names.2016
@@ -12,10 +12,5 @@ outs:
 - path: pgstat://loc-mds-names
   cache: false
   md5: 5d6d486b5b3acde6b3a9ce64d0fe794b
-  metric: false
-  persist: false
 - md5: 14ce7449e200f45a93058a6fdac918ec
   path: import/loc-mds-names.transcript
-  cache: true
-  metric: false
-  persist: false

+ 1
- 6
import/ol-authors.dvc

@@ -1,4 +1,4 @@
-md5: 19cfc3b62480708a68fb703520939227
+md5: 41b12fb82a71b892c79c766e60213349
 cmd: python run.py --rust import-json -T import/ol-authors.transcript --stage ol-authors
   -D ol-schema --truncate import/ol-authors.toml data/ol_dump_authors.txt.gz
 wdir: ..
@@ -13,10 +13,5 @@ outs:
 - path: pgstat://ol-authors
   cache: false
   md5: 456954970c9a56193680bb9399ac9164
-  metric: false
-  persist: false
 - md5: 50e5bedc17cba4c5ceb8eef3d86fc307
   path: import/ol-authors.transcript
-  cache: true
-  metric: false
-  persist: false

+ 1
- 6
import/ol-editions.dvc

@@ -1,4 +1,4 @@
-md5: 89599918b5648a3bc5011375826102ff
+md5: 24dbe71ecb56b10c6ba6c621fc734aad
 cmd: python run.py --rust import-json -T import/ol-editions.transcript --stage ol-editions
   -D ol-schema --truncate import/ol-editions.toml data/ol_dump_editions.txt.gz
 wdir: ..
@@ -13,10 +13,5 @@ outs:
 - path: pgstat://ol-editions
   cache: false
   md5: 4c0b955dcf06c319b8dd9c1057c3c056
-  metric: false
-  persist: false
 - md5: c2c6951a052ebebe5f51ceb79f324162
   path: import/ol-editions.transcript
-  cache: true
-  metric: false
-  persist: false

+ 1
- 6
import/ol-works.dvc

@@ -1,4 +1,4 @@
-md5: cb0ebc399a4a8f0dcde413f449a0d9fd
+md5: 401afdb0f51cafd634e1980cce30cb96
 cmd: python run.py --rust import-json -T import/ol-works.transcript --stage ol-works
   -D ol-schema --truncate import/ol-works.toml data/ol_dump_works.txt.gz
 wdir: ..
@@ -13,10 +13,5 @@ outs:
 - path: pgstat://ol-works
   cache: false
   md5: c84b6ccf5f89e6d8faf00cc21b4d5566
-  metric: false
-  persist: false
 - md5: 6345885738815d2a51d331f7aba9e792
   path: import/ol-works.transcript
-  cache: true
-  metric: false
-  persist: false

+ 1
- 6
import/viaf.dvc

@@ -1,4 +1,4 @@
-md5: cc05ef70321cdb2f481b172251eccd50
+md5: 3e58afead501a712551bf345111e0d02
 cmd: python run.py --rust parse-marc --db-schema viaf -t marc_field --truncate --stage
   viaf -D viaf-schema --transcript import/viaf.transcript --line-mode data/viaf-clusters-marc21.xml.gz
 wdir: ..
@@ -11,10 +11,5 @@ outs:
 - path: pgstat://viaf
   cache: false
   md5: 55fb31f4aaa86a2fd921c59c797c09a8
-  metric: false
-  persist: false
 - md5: 326fe03b3007b0dbe02ff43c3da7e6cb
   path: import/viaf.transcript
-  cache: true
-  metric: false
-  persist: false

+ 1
- 6
index/az-index.dvc

@@ -1,4 +1,4 @@
-md5: db3b43e5dfc246741e54e4c891214d5b
+md5: a773a007075da53c462331462a2ff0b3
 cmd: python ../run.py sql-script az-index.sql
 deps:
 - path: az-index.sql
@@ -11,10 +11,5 @@ outs:
 - path: pgstat://az-index
   cache: false
   md5: b845cbb28bd6735919d6ef20c069b2d1
-  metric: false
-  persist: false
 - path: az-index.transcript
-  cache: true
-  metric: false
-  persist: false
   md5: 11bcbcd2d11a8a121b22a9a799312434

+ 1
- 6
index/bx-index.dvc

@@ -1,4 +1,4 @@
-md5: d38329d720915efe0ed4281cb4dd691e
+md5: 45d9ca358eb4531acc994d0e22d17ae7
 cmd: python ../run.py sql-script bx-index.sql
 deps:
 - path: bx-index.sql
@@ -11,10 +11,5 @@ outs:
 - path: pgstat://bx-index
   cache: false
   md5: 7e6211a90148898456ab9dada9353cc2
-  metric: false
-  persist: false
 - path: bx-index.transcript
-  cache: true
-  metric: false
-  persist: false
   md5: 10895cefc67a2d442b8d834d8399d05b

+ 1
- 6
index/gr-book-info.dvc

@@ -1,4 +1,4 @@
-md5: 7a392b4a783feaf7a4b2d9ce0ab89c2b
+md5: 060b4fd5f5cfb58f177d29ee4cf60306
 cmd: python ../run.py sql-script gr-book-info.sql
 deps:
 - path: gr-book-info.sql
@@ -15,10 +15,5 @@ outs:
 - path: pgstat://gr-book-info
   cache: false
   md5: b18c7337da454ea778442904c2aba99c
-  metric: false
-  persist: false
 - path: gr-book-info.transcript
-  cache: true
-  metric: false
-  persist: false
   md5: f2532c9b91cd7d5e461ac7dae0b43efa

+ 1
- 6
index/gr-index-books.dvc

@@ -1,4 +1,4 @@
-md5: ce86609dc53e7da984e5ad4aa0863f0d
+md5: 037e6708b0443cf379111c026ee82c9c
 cmd: python ../run.py sql-script gr-index-books.sql
 deps:
 - path: gr-index-books.sql
@@ -15,10 +15,5 @@ outs:
 - path: pgstat://gr-index-books
   cache: false
   md5: 3fe0d48fd17efd5e670f6b2649fc2286
-  metric: false
-  persist: false
 - path: gr-index-books.transcript
-  cache: true
-  metric: false
-  persist: false
   md5: 39cfd15b59141412e4fb699be6b12762

+ 1
- 6
index/gr-index-ratings.dvc

@@ -1,4 +1,4 @@
-md5: 7b20dfba82a6ff6a7b6e59244c81ee49
+md5: e2b85332d16a2008da20430458678f95
 cmd: python ../run.py sql-script gr-index-ratings.sql
 deps:
 - path: gr-index-ratings.sql
@@ -13,10 +13,5 @@ outs:
 - path: pgstat://gr-index-ratings
   cache: false
   md5: 447f8a6cf359e5d8ae5736cf9863192f
-  metric: false
-  persist: false
 - path: gr-index-ratings.transcript
-  cache: true
-  metric: false
-  persist: false
   md5: d33fdfeca1ebee91120851434a42960c

+ 1
- 6
index/isbn-norm.dvc

@@ -11,12 +11,7 @@ deps:
 outs:
 - path: isbn-norm.transcript
   md5: 7c99a01ec9d598650ba83fd2102a8882
-  cache: true
-  metric: false
-  persist: false
 - path: pgstat://isbn-norm
   cache: false
   md5: 5f281f436639bdcb85383e0b334183de
-  metric: false
-  persist: false
-md5: c140dd077dfa8cec4c44f63fbc266ea2
+md5: 9b2c9efe08035eb145268a483f31fcc0

+ 1
- 6
index/loc-mds-book-info.dvc

@@ -1,4 +1,4 @@
-md5: e5b68041e0c434614f707abfc54cf745
+md5: 1d8bafd8d9c8f39c816b5c3f5cdad2ac
 cmd: python ../run.py sql-script loc-mds-book-info.sql
 deps:
 - path: loc-mds-book-info.sql
@@ -9,10 +9,5 @@ outs:
 - path: pgstat://loc-mds-book-info
   cache: false
   md5: ae5b3dfa240bfa1de37f585efbc9c55b
-  metric: false
-  persist: false
 - path: loc-mds-book-info.transcript
-  cache: true
-  metric: false
-  persist: false
   md5: 83e7b6b2931146f417ea4b84b76614d2

+ 1
- 6
index/loc-mds-extract-isbns.dvc

@@ -8,11 +8,6 @@ outs:
 - path: pgstat://loc-mds-extract-isbns
   cache: false
   md5: c4ceff988a5b8a7c15ca00c0dbc4ec59
-  metric: false
-  persist: false
 - path: loc-mds-extract-isbns.transcript
   md5: e08255fdca28f536f2aa8f2d065a26cf
-  cache: true
-  metric: false
-  persist: false
-md5: c802e190e0783e771f514ddb7aed5f12
+md5: fc540924aeac9e4c27470fa0fb6a55e2

+ 1
- 6
index/loc-mds-index-books.dvc

@@ -1,4 +1,4 @@
-md5: 05b74972708103f90ede481ffc1827cf
+md5: 86763b31a3e9a207bf684166307c7b58
 cmd: python ../run.py sql-script loc-mds-index-books.sql
 deps:
 - path: loc-mds-index-books.sql
@@ -11,10 +11,5 @@ outs:
 - path: pgstat://loc-mds-index-books
   cache: false
   md5: abdf7eecd1861c7318b15a2b32435204
-  metric: false
-  persist: false
 - path: loc-mds-index-books.transcript
-  cache: true
-  metric: false
-  persist: false
   md5: 0af5f4abe6b224d06ac8a2950d22ee61

+ 1
- 6
index/loc-mds-index-names.dvc

@@ -1,4 +1,4 @@
-md5: ce3177fd415581b963aed51a795ff8e3
+md5: d077fb83494125748281da178d079e89
 cmd: python ../run.py sql-script loc-mds-index-names.sql
 deps:
 - path: loc-mds-index-names.sql
@@ -9,10 +9,5 @@ outs:
 - path: pgstat://loc-mds-index-names
   cache: false
   md5: 4d8d4e061447b2d53f39262c41e60f74
-  metric: false
-  persist: false
 - path: loc-mds-index-names.transcript
-  cache: true
-  metric: false
-  persist: false
   md5: 0d07d44aa3486e6aab6aa3ee67b0a5b3

+ 1
- 6
index/ol-book-info.dvc

@@ -1,4 +1,4 @@
-md5: 29d331ac152cb9121ee23dcd787e8f49
+md5: cd6d77e011eebc4ed117118d9ce62246
 cmd: python ../run.py sql-script ol-book-info.sql
 deps:
 - md5: 3ef46b2a3e878a00e3a5ad7a31aa5a1f
@@ -9,10 +9,5 @@ outs:
 - path: pgstat://ol-book-info
   cache: false
   md5: 8ce3b1230c306ab53d019453bd336927
-  metric: false
-  persist: false
 - md5: 5fb342deef98c301027b1807d96d04d5
   path: ol-book-info.transcript
-  cache: true
-  metric: false
-  persist: false

+ 1
- 6
index/ol-index.dvc

@@ -1,4 +1,4 @@
-md5: 73002db403c0e76ee3a14e160c96ebfb
+md5: 2e2feff642cc7eff15b04adcf55f43f2
 cmd: python ../run.py sql-script ol-index.sql
 deps:
 - md5: 627e1e62981c643104bdd6bec34fab7d
@@ -13,10 +13,5 @@ outs:
 - path: pgstat://ol-index
   cache: false
   md5: 431be74ffa0928a66c3ec3084f8d3640
-  metric: false
-  persist: false
 - md5: 5528fa7ed03102ecefdf92ace25a13d0
   path: ol-index.transcript
-  cache: true
-  metric: false
-  persist: false

+ 1
- 6
index/viaf-index.dvc

@@ -1,4 +1,4 @@
-md5: bdbab3fd1cab579ed0f2e851bb2b4b9c
+md5: ba6816f26ddfa2bbadc7f941d9980ea2
 cmd: python ../run.py sql-script viaf-index.sql
 deps:
 - path: viaf-index.sql
@@ -9,10 +9,5 @@ outs:
 - path: pgstat://viaf-index
   cache: false
   md5: 7598c00689dd89a355a1538d9855ff60
-  metric: false
-  persist: false
 - path: viaf-index.transcript
-  cache: true
-  metric: false
-  persist: false
   md5: a2a7c0283e2dddd99d87139e596512b6

+ 0
- 3
init.dvc

@@ -1,7 +1,4 @@
-md5: 021bbeb95259e4ed9fb03979106ae4d2
 outs:
 - path: pgstat://init
   cache: false
-  metric: false
-  persist: false
   md5: 3266ce83643ca4202ccea4ee2c25b66e

+ 1
- 6
integrate/author-info.dvc

@@ -1,4 +1,4 @@
-md5: 48c0d17782f6d90ff38c7600a8bb46c0
+md5: 460f6988cb4a3d43cd3846b268cdf7d3
 cmd: python ../run.py sql-script author-info.sql
 deps:
 - path: pgstat://loc-mds-index-books
@@ -17,10 +17,5 @@ outs:
 - path: pgstat://author-info
   cache: false
   md5: 1276cdd4578b084c3b77103d7320b310
-  metric: false
-  persist: false
 - path: author-info.transcript
-  cache: true
-  metric: false
-  persist: false
   md5: 37efee138ac7c170958be11b074070ac

+ 1
- 6
integrate/author-stats.dvc

@@ -1,4 +1,4 @@
-md5: ff42eb4ffe7b3d27ceda99df6ac1ff5a
+md5: d6b4f8617938f087c8f79044394873b6
 cmd: python ../run.py sql-script author-stats.sql
 deps:
 - path: author-stats.sql
@@ -15,10 +15,5 @@ outs:
 - path: pgstat://author-stats
   cache: false
   md5: 5554974d660e2b104184535f00107ec3
-  metric: false
-  persist: false
 - path: author-stats.transcript
-  cache: true
-  metric: false
-  persist: false
   md5: 0a1929d0eb953520c312fefc79de7b94

+ 1
- 6
integrate/cluster-stats.dvc

@@ -1,4 +1,4 @@
-md5: d1376058ef76910cfd1003abf07e6801
+md5: 08416c8aaf70acbd864c422a22da71ec
 cmd: python ../run.py sql-script cluster-stats.sql
 deps:
 - path: cluster-stats.sql
@@ -11,10 +11,5 @@ outs:
 - path: pgstat://cluster-stats
   cache: false
   md5: 7c052a1b981bcc763cad9abde972639e
-  metric: false
-  persist: false
 - path: cluster-stats.transcript
-  cache: true
-  metric: false
-  persist: false
   md5: 4267da9b87d2805e56d23400bb3a8839

+ 1
- 9
integrate/cluster.dvc

@@ -1,4 +1,4 @@
-md5: fd7558c7ed423dce6ea2caea62cde657
+md5: 48a9c700f1940a1caa3f71c3b3f132fc
 cmd: python run.py cluster -T integrate/cluster.transcript
 wdir: ..
 deps:
@@ -12,15 +12,7 @@ outs:
 - path: pgstat://cluster
   cache: false
   md5: abac8ffbe1d4b0e33b39320bdfd7974d
-  metric: false
-  persist: false
 - md5: 4d94cdee39942f7030917a5e2c0d2ac1
   path: integrate/cluster.transcript
-  cache: true
-  metric: false
-  persist: false
 - path: data/id-graph.gt
   md5: 3efd63de8b1191c60ce1e88d12eb5fca
-  cache: true
-  metric: false
-  persist: false

+ 0
- 1
schemas.dvc

@@ -14,4 +14,3 @@ deps:
   md5: bda41944c7f3ff2a207edb2c8a83c9fc
 - path: pgstat://viaf-schema
   md5: 4ee5de53afb5dfc1e1740a8667887cc0
-md5: 75935162b5fc55e490db86a9f8134e34

+ 1
- 6
schemas/az-schema.dvc

@@ -1,4 +1,4 @@
-md5: ba1f2802a9aba937cb7da0ec26a6e13e
+md5: 4d3c680addb3d33a3e0eea3164453325
 cmd: python ../run.py sql-script az-schema.sql
 deps:
 - md5: 0b449168158af401f2a7f399e7783329
@@ -8,11 +8,6 @@ deps:
 outs:
 - md5: f383fbf300e2d8ce2cb1dafd8b755255
   path: az-schema.transcript
-  cache: true
-  metric: false
-  persist: false
 - path: pgstat://az-schema
   cache: false
   md5: 01a7ee13889d967ed8ac69bfdf592d80
-  metric: false
-  persist: false

+ 1
- 6
schemas/bx-schema.dvc

@@ -1,4 +1,4 @@
-md5: bba3046887ed15cbcb75778fdeab447d
+md5: 915929e075ee0a220499da8f303626e9
 cmd: python ../run.py sql-script bx-schema.sql
 deps:
 - md5: 1cf832f06d45b03b47dcdb26ee501cfb
@@ -9,10 +9,5 @@ outs:
 - path: pgstat://bx-schema
   cache: false
   md5: 48cd11df72cf2bb28d5a2ff58f02d972
-  metric: false
-  persist: false
 - md5: 81c22063517d757610f0c7bb852608b0
   path: bx-schema.transcript
-  cache: true
-  metric: false
-  persist: false

+ 0
- 2
schemas/common-schema.dvc

@@ -9,5 +9,3 @@ outs:
 - md5: f12f086695a243c125731397d4a31bf7
   path: pgstat://common-schema
   cache: false
-  metric: false
-  persist: false

+ 1
- 6
schemas/gr-schema.dvc

@@ -1,4 +1,4 @@
-md5: 5ee4e68f8d2598d3da238335ead6080d
+md5: d3f0924035f47c3c04fb43915c72adee
 cmd: python ../run.py sql-script gr-schema.sql
 deps:
 - md5: 5f42d1564474a09ff40719f26bc55e95
@@ -9,10 +9,5 @@ outs:
 - path: pgstat://gr-schema
   cache: false
   md5: bda41944c7f3ff2a207edb2c8a83c9fc
-  metric: false
-  persist: false
 - md5: 4e539f7fb068c1c322c0b9f1eed125f5
   path: gr-schema.transcript
-  cache: true
-  metric: false
-  persist: false

+ 1
- 6
schemas/loc-mds-schema.dvc

@@ -1,4 +1,4 @@
-md5: 55cde8d364fbc8502fd2b3b480187605
+md5: fccea8081d6ec1d44d9e187c4b3bf66a
 cmd: python ../run.py sql-script loc-mds-schema.sql
 deps:
 - md5: 89549eac0d9232917eb62681b8aeaf1e
@@ -9,10 +9,5 @@ outs:
 - path: pgstat://loc-mds-schema
   cache: false
   md5: e63399b7692987ecd6b579066e5bd35e
-  metric: false
-  persist: false
 - md5: d06218293061d6ef25f71c581c11bbcd
   path: loc-mds-schema.transcript
-  cache: true
-  metric: false
-  persist: false

+ 1
- 6
schemas/ol-schema.dvc

@@ -1,4 +1,4 @@
-md5: 3a98c166e35a5ce03d8f5ab33a9b60c4
+md5: 0b9c6b42b40ffd49c263e7ef253d5696
 cmd: python ../run.py sql-script ol-schema.sql
 deps:
 - md5: 72d62c2f5ebe02dd86660a47e10a31fb
@@ -9,10 +9,5 @@ outs:
 - path: pgstat://ol-schema
   cache: false
   md5: ea3a792c4941083707b2835f737ada05
-  metric: false
-  persist: false
 - md5: 838db779e203e4d362e36d2401eb1880
   path: ol-schema.transcript
-  cache: true
-  metric: false
-  persist: false

+ 1
- 6
schemas/viaf-schema.dvc

@@ -1,4 +1,4 @@
-md5: 30f38e9b8afebfa18e57d12944e13b1c
+md5: 740eee012127b0952b8b411db174928a
 cmd: python ../run.py sql-script viaf-schema.sql
 deps:
 - md5: 94f75a2abb26382c235fdfef269559e8
@@ -9,10 +9,5 @@ outs:
 - path: pgstat://viaf-schema
   cache: false
   md5: 4ee5de53afb5dfc1e1740a8667887cc0
-  metric: false
-  persist: false
 - md5: abca653eed931eb353d157000e9ccc26
   path: viaf-schema.transcript
-  cache: true
-  metric: false
-  persist: false