cocina-models 0.121.0 → 0.122.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: cfcb2c2a393845b1c9129ab48f5fc7340beb8067936694798e800a5935bf1ac9
-  data.tar.gz: f9dc80c520ddecb670a5df20273c635146f75d75025132402d4d1329090e29fd
+  metadata.gz: a7ba361f4c17d3165dce9e5042739493896f798fe9a215ce20054b6f2c739c11
+  data.tar.gz: 5ca9b3f5efdd6ab255dfe1f48b19be012638986f7126af70547d701bd5059b01
 SHA512:
-  metadata.gz: 2f7ccadc053ba81117275251401432ec92c65fefd86da94c25e7c862f8680c868d3f392a84b3966fb48609a5c383abd8dc90b8a51e457fd2fe8d3792371bd4a6
-  data.tar.gz: c1355245a0571f8933155c9de04b38226b55b96a3ffdb899af3d2af553e30f53ae7986a41e6293339a90cc2a017fcd6eec273fcd1ce64df57b26cbbb5f161193
+  metadata.gz: cfd5524aa8e1ce0dc3510855b1c1cb9cadd875878192c71ecfb9fdd02fe74457e00a9a3766eefd3f399b2b3a539a013eb63e929b7b1a46b58fedb789ef170d6c
+  data.tar.gz: 27d345e4bd13c181165a943a696ecbfe44c36fce1fcd1db4324f13c8d97fce029b953a36030fc9a71586d95a6458622300937990069b57fcae0baeed20398be3

data/AGENTS.md

@@ -22,10 +22,10 @@ Run `pv --version`. If pv is not installed, tell the user:
 
 ### Output format (always apply)
 
-Every jq query produced by this skill **must output a CSV line** using `@csv`. The **first field must always be the external identifier** (`externalIdentifier`). Additional fields follow based on the user's query. Example:
+Every jq query produced by this skill **must output a CSV line** using `@csv`. The **first field must always be the external identifier** (`externalIdentifier`). Additional fields follow based on the user's query. By default, multiple values should be joined by " | ". Example:
 
 ```
-"druid:bc123df4567","some value","another value"
+"druid:bc123df4567","some value | another value"
 ```
 
 Use `[.externalIdentifier, ...] | @csv` as the output expression. Apply this constraint automatically — do not ask the user whether to include the external identifier.

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    cocina-models (0.121.0)
+    cocina-models (0.122.0)
       activesupport
       deprecation
       dry-struct (~> 1.0)
@@ -220,7 +220,7 @@ CHECKSUMS
   bigdecimal (4.1.2) sha256=53d217666027eab4280346fba98e7d5b66baaae1b9c3c1c0ffe89d48188a3fbd
   builder (3.3.0) sha256=497918d2f9dca528fdca4b88d84e4ef4387256d984b8154e9d5d3fe5a9c8835f
   bundler (4.0.13) sha256=19f08be7f27022cf0b89f27da0b044ae075e8270a9ef44ad248a932614e1ca3b
-  cocina-models (0.121.0)
+  cocina-models (0.122.0)
   concurrent-ruby (1.3.6) sha256=6b56837e1e7e5292f9864f34b69c5a2cbc75c0cf5338f1ce9903d10fa762d5ab
   connection_pool (3.0.2) sha256=33fff5ba71a12d2aa26cb72b1db8bba2a1a01823559fb01d29eb74c286e62e0a
   csv (3.3.5) sha256=6e5134ac3383ef728b7f02725d9872934f523cb40b961479f69cf3afa6c8e73f

data/bin/enhance-report-csv

@@ -67,18 +67,16 @@ def extract_fields(doc)
 end
 
 def build_output(solr, rows, batch_size)
-  extra_col_count = (rows.first&.size || 1) - 1
-  extra_headers = extra_col_count.times.map { |i| "col#{i + 2}" }
-
   CSV.generate do |out|
-    out << (['druid'] + FIELD_HEADERS.values + extra_headers)
+    out << (%w[druid value] + FIELD_HEADERS.values)
 
     rows.each_slice(batch_size) do |batch|
       docs = fetch_solr_docs(solr, batch.map { |row| row[0] })
 
       batch.each do |row|
         druid = row[0]
-        out << ([druid] + extract_fields(docs[druid]) + row[1..])
+        value = row.size > 1 ? row[1..].join('|') : ''
+        out << ([druid, value] + extract_fields(docs[druid]))
       end
     end
   end

data/bin/validate-data

@@ -19,6 +19,7 @@ require 'cocina/models'
 require 'json'
 require 'ruby-progressbar'
 require 'optparse'
+require 'csv'
 
 # Parse command line options
 def parse_options # rubocop:disable Metrics/MethodLength
@@ -104,15 +105,16 @@ def worker_process(reader) # rubocop:disable Metrics/MethodLength
     # Process each line in the batch
     batch.each do |line_num, line_content|
       json = JSON.parse(line_content)
+      druid = json['externalIdentifier']
       Cocina::Models.build(json)
     rescue JSON::ParserError => e
       errors << { line: line_num, error: "JSON Parse Error: #{e.message}" }
     rescue Cocina::Models::ValidationError => e
-      errors << { line: line_num, error: "Validation Error: #{json['externalIdentifier']} - #{e.message}" }
+      errors << { line: line_num, druid:, error: "Validation Error: #{e.message}" }
     rescue Cocina::Models::UnknownTypeError => e
-      errors << { line: line_num, error: "Unknown Type Error: #{e.message}" }
+      errors << { line: line_num, druid:, error: "Unknown Type Error: #{e.message}" }
     rescue StandardError => e
-      errors << { line: line_num, error: "Error: #{e.class} - #{e.message}" }
+      errors << { line: line_num, druid:, error: "Error: #{e.class} - #{e.message}" }
     end
   end
 
@@ -245,15 +247,25 @@ def print_summary(total_lines, errors, elapsed_time) # rubocop:disable Metrics/M
 
   return unless errors.any?
 
+  sorted_errors = errors.sort_by { |e| e[:line] }
+
   puts "\n"
   puts 'Error details:'
   puts '-' * 80
   # Sort errors by line number for better readability
-  errors.sort_by { |e| e[:line] }.each do |error|
-    puts "Line #{error[:line]}: #{error[:error]}"
+  sorted_errors.each do |error|
+    puts "Line #{error[:line]}: #{error[:druid]} - #{error[:error]}"
   end
   puts "\n"
-  puts "Line numbers with errors: #{errors.sort_by { |e| e[:line] }.map { |e| e[:line] }.join(', ')}"
+  puts "Line numbers with errors: #{sorted_errors.map { |e| e[:line] }.join(', ')}"
+
+  CSV.open('validate-data-errors.csv', 'w') do |csv|
+    csv << %w[line druid error]
+    sorted_errors.each do |error|
+      csv << [error[:line], error[:druid], error[:error]]
+    end
+  end
+  puts 'Errors written to validate-data-errors.csv'
 end
 
 # Main execution

data/lib/cocina/models/mapping/from_mods/name_builder.rb

@@ -60,7 +60,7 @@ module Cocina
 
           # build non-parallel, single name
           def build_name(name_node)
-            return { type: 'unspecified others' } if name_node.xpath('mods:etal', mods: Description::DESC_METADATA_NS).present?
+            return nil if name_node.xpath('mods:etal', mods: Description::DESC_METADATA_NS).present?
 
             name_parts = build_name_parts(name_node)
             # If there are no name parts, do not map the name

data/lib/cocina/models/version.rb

@@ -2,6 +2,6 @@
 
 module Cocina
   module Models
-    VERSION = '0.121.0'
+    VERSION = '0.122.0'
   end
 end

data/schema.json

@@ -507,7 +507,28 @@
           "type": "string"
         }
       },
-      "unevaluatedProperties": false
+      "unevaluatedProperties": false,
+      "anyOf": [
+        {
+          "required": ["valueAt"]
+        },
+        {
+          "required": ["name"],
+          "properties": {
+            "name": {
+              "minItems": 1
+            }
+          }
+        },
+        {
+          "required": ["identifier"],
+          "properties": {
+            "identifier": {
+              "minItems": 1
+            }
+          }
+        }
+      ]
     },
     "ControlledDigitalLendingAccess": {
       "type": "object",
@@ -900,7 +921,57 @@
           }
         }
       },
-      "unevaluatedProperties": false
+      "unevaluatedProperties": false,
+      "anyOf": [
+        {
+          "required": ["url"],
+          "properties": {
+            "url": {
+              "minItems": 1
+            }
+          }
+        },
+        {
+          "required": ["physicalLocation"],
+          "properties": {
+            "physicalLocation": {
+              "minItems": 1
+            }
+          }
+        },
+        {
+          "required": ["digitalLocation"],
+          "properties": {
+            "digitalLocation": {
+              "minItems": 1
+            }
+          }
+        },
+        {
+          "required": ["accessContact"],
+          "properties": {
+            "accessContact": {
+              "minItems": 1
+            }
+          }
+        },
+        {
+          "required": ["digitalRepository"],
+          "properties": {
+            "digitalRepository": {
+              "minItems": 1
+            }
+          }
+        },
+        {
+          "required": ["note"],
+          "properties": {
+            "note": {
+              "minItems": 1
+            }
+          }
+        }
+      ]
     },
     "DescriptiveAdminMetadata": {
       "description": "Information about this resource description.",
@@ -1108,7 +1179,25 @@
           }
         }
       },
-      "unevaluatedProperties": false
+      "unevaluatedProperties": false,
+      "anyOf": [
+        {
+          "required": ["form"],
+          "properties": {
+            "form": {
+              "minItems": 1
+            }
+          }
+        },
+        {
+          "required": ["subject"],
+          "properties": {
+            "subject": {
+              "minItems": 1
+            }
+          }
+        }
+      ]
     },
     "DescriptiveGroupedValue": {
       "description": "Value model for a set of descriptive elements grouped together in an unstructured way.",
@@ -1437,7 +1526,65 @@
           }
         }
       ],
-      "unevaluatedProperties": false
+      "unevaluatedProperties": false,
+      "anyOf": [
+        {
+          "required": ["date"],
+          "properties": {
+            "date": {
+              "minItems": 1
+            }
+          }
+        },
+        {
+          "required": ["contributor"],
+          "properties": {
+            "contributor": {
+              "minItems": 1
+            }
+          }
+        },
+        {
+          "required": ["location"],
+          "properties": {
+            "location": {
+              "minItems": 1
+            }
+          }
+        },
+        {
+          "required": ["identifier"],
+          "properties": {
+            "identifier": {
+              "minItems": 1
+            }
+          }
+        },
+        {
+          "required": ["note"],
+          "properties": {
+            "note": {
+              "minItems": 1
+            }
+          }
+        },
+        {
+          "required": ["structuredValue"],
+          "properties": {
+            "structuredValue": {
+              "minItems": 1
+            }
+          }
+        },
+        {
+          "required": ["parallelEvent"],
+          "properties": {
+            "parallelEvent": {
+              "minItems": 1
+            }
+          }
+        }        
+      ]
     },
     "File": {
       "description": "Binaries that are the basis of what our domain manages. Binaries here do not include metadata files generated for the domain's own management purposes.",
@@ -1550,6 +1697,17 @@
         "sdrPreserve",
         "shelve"
       ],
+      "if": {
+        "properties": {
+          "shelve": { "const": true }
+        }
+      },
+      "then": {
+        "anyOf": [
+          { "properties": { "sdrPreserve": { "const": true } } },
+          { "properties": { "publish": { "const": true } } }
+        ]
+      },
       "unevaluatedProperties": false
     },
     "FileSet": {
@@ -1798,7 +1956,61 @@
           "$ref": "#/$defs/DescriptiveValueLanguage"
         }
       },
-      "unevaluatedProperties": false
+      "unevaluatedProperties": false,
+      "anyOf": [
+        {
+          "required": ["value"]
+        },
+        {
+          "required": ["code"]
+        },
+        {
+          "required": ["uri"]
+        },
+        {
+          "required": ["note"],
+          "properties": {
+            "note": {
+              "minItems": 1
+            }
+          }
+        },
+        {
+          "required": ["script"],
+          "properties": {
+            "script": {
+              "minItems": 1
+            }
+          }
+        },
+        {
+          "required": ["valueAt"]
+        },
+        {
+          "required": ["structuredValue"],
+          "properties": {
+            "structuredValue": {
+              "minItems": 1
+            }
+          }
+        },
+        {
+          "required": ["parallelValue"],
+          "properties": {
+            "parallelValue": {
+              "minItems": 1
+            }
+          }
+        },
+        {
+          "required": ["groupedValue"],
+          "properties": {
+            "groupedValue": {
+              "minItems": 1
+            }
+          }
+        }
+      ]
     },
     "LanguageTag": {
       "description": "BCP 47 language tag: https://www.rfc-editor.org/rfc/rfc4646.txt -- other applications (like media players) expect language codes of this format, see e.g. https://videojs.com/guides/text-tracks/#srclang",

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: cocina-models
 version: !ruby/object:Gem::Version
-  version: 0.121.0
+  version: 0.122.0
 platform: ruby
 authors:
 - Justin Coyne