switch to more semantic metadata

cldf-datasets · May 8, 2019 · f36deac · f36deac
1 parent c964fd5
commit f36deac
Show file tree

Hide file tree

Showing 3 changed files with 38 additions and 20 deletions.
diff --git a/cldf/StructureDataset-metadata.json b/cldf/StructureDataset-metadata.json
@@ -7,17 +7,27 @@
     ],
     "dc:conformsTo": "http://cldf.clld.org/v1.0/terms.rdf#StructureDataset",
     "dc:source": "sources.bib",
-    "dialect": {
-        "commentPrefix": null
-    },
-    "notes": [
+    "prov:wasDerivedFrom": [
         {
-            "dc:title": "environment",
-            "properties": {
-                "glottolog_version": "v3.3.2"
-            }
+            "dc:title": "phoible/phoible-dev",
+            "dc:description": "v2014-241-g0ec7201"
+        },
+        {
+            "dc:title": "bambooforest/phoible-scripts",
+            "dc:description": "a2a55d5"
+        },
+        {
+            "dc:title": "clld/glottolog",
+            "dc:description": "v3.3.2"
         }
     ],
+    "prov:wasGeneratedBy": {
+        "dc:title": "cldf-datasets/phoible",
+        "dc:description": "v2.0-13-gc964fd5"
+    },
+    "dialect": {
+        "commentPrefix": null
+    },
     "tables": [
         {
             "dc:conformsTo": "http://cldf.clld.org/v1.0/terms.rdf#ValueTable",

diff --git a/cldf/sources.bib b/cldf/sources.bib
@@ -13698,7 +13698,7 @@ @incollection{1896_gawthorneetal1984
 }
 
 @book{1897_adelaaretal2004,
-    author = {Adelaar, Willem F. H. \\\\\\\\& Pieter C. Muysken},
+    author = {Adelaar, Willem F. H. & Pieter C. Muysken},
     address = {New York},
     bibtexkey = {Adelaaretal2004},
     date-added = {2014-07-09 13:41:25 +0000},
@@ -14822,7 +14822,7 @@ @phdthesis{1953_oliveira2005
 }
 
 @book{1954_arikapu2010,
-    author = {Arikapú, Mamoa; Nazaré Arikapú; Hein van der Voort \\\\\\\\& Ana Carolina Ferreira Alves},
+    author = {Arikapú, Mamoa; Nazaré Arikapú; Hein van der Voort & Ana Carolina Ferreira Alves},
     bibtexkey = {Arikapu2010},
     booktitle = {Cadernos de Etnolingü{'i}stica},
     date-added = {2014-07-09 13:41:25 +0000},
@@ -15571,7 +15571,7 @@ @book{1993_martins2005
 
 @incollection{1994_tellesetal2011,
     author = {Telles, Stella and Wetzels, Leo},
-    editor = {Emilio Bonvini, Joëlle Busuttil \\\\\\\\& Alain Peyraube},
+    editor = {Emilio Bonvini, Joëlle Busuttil & Alain Peyraube},
     address = {Paris},
     bibtexkey = {Tellesetal2011},
     booktitle = {Dictionnaire des langues},
@@ -15588,7 +15588,7 @@ @incollection{1994_tellesetal2011
 
 @incollection{1995_tellesetal2011,
     author = {Telles, Stella and Wetzels, Leo},
-    editor = {Emilio Bonvini, Joëlle Busuttil \\\\\\\\& Alain Peyraube},
+    editor = {Emilio Bonvini, Joëlle Busuttil & Alain Peyraube},
     address = {Paris},
     bibtexkey = {Tellesetal2011},
     booktitle = {Dictionnaire des langues},
@@ -17254,7 +17254,7 @@ @phdthesis{2083_martins2003
 }
 
 @book{2084_krivosheindecanese2000,
-    author = {Krivoshein de Canese, Natalia, \\\\\\\\& Feliciano Acosta Alcaraz},
+    author = {Krivoshein de Canese, Natalia, & Feliciano Acosta Alcaraz},
     bibtexkey = {Krivosheindecanese2000},
     date-added = {2014-07-09 13:41:25 +0000},
     date-modified = {2014-07-09 13:41:26 +0000},
@@ -17979,7 +17979,7 @@ @article{2126_olson1967
 
 @incollection{2127_muysken2000,
     author = {Muysken, Pieter},
-    editor = {van der Voort \\\\\\\\& Simon van de Kerke, Hein},
+    editor = {van der Voort & Simon van de Kerke, Hein},
     address = {Leiden},
     bibtexkey = {Muysken2000},
     date-added = {2014-07-09 13:41:25 +0000},

diff --git a/create.py b/create.py
@@ -29,12 +29,20 @@ def main(scripts, dev, glr):
     glottolog = Glottolog(glr)
 
     ds = StructureDataset.in_dir(cldf_dir)
-    ds.tablegroup.notes.append(OrderedDict([
-        ('dc:title', 'environment'),
-        ('properties', OrderedDict([
-            ('glottolog_version', git_describe(glottolog.repos)),
-        ]))
-    ]))
+
+    def describe_repos(r, org, name=None):
+        return OrderedDict([
+            ('dc:title', '{0}/{1}'.format(org, name or r.name)),
+            ('dc:description', git_describe(r))])
+
+    ds.tablegroup.common_props['prov:wasDerivedFrom'] = [
+        describe_repos(dev, 'phoible'),
+        describe_repos(scripts, 'bambooforest'),
+        describe_repos(glottolog.repos, 'clld'),
+    ]
+    ds.tablegroup.common_props['prov:wasGeneratedBy'] = describe_repos(
+        Path(__file__).parent, 'cldf-datasets', name='phoible')
+
     ds.add_columns(
         'ValueTable',
         {'name': 'Marginal', 'datatype': 'boolean'},