ENCD-6102-change-regex-lot-id-product-id (#4069)

src/encoded/schemas/antibody_lot.json

@@ -17,6 +17,7 @@
         { "$ref": "mixins.json#/submitted" },
         { "$ref": "mixins.json#/source" },
         { "$ref": "mixins.json#/url" },
+        { "$ref": "mixins.json#/lot_id" },
         { "$ref": "mixins.json#/product_id" },
         { "$ref": "mixins.json#/notes" }
 
@@ -31,7 +32,7 @@
     ],
     "properties": {
         "schema_version": {
-            "default": "9"
+            "default": "10"
         },
         "accession": {
             "accessionType": "AB"
@@ -52,12 +53,6 @@
                 "pattern": "^(ENCAB\\d{3}[A-Z]{3})$|^(TSTAB\\d{6})$"
             }
         },
-        "lot_id": {
-            "title": "Lot ID",
-            "description": "The lot identifier provided by the originating lab or vendor.",
-            "type": "string",
-            "pattern": "^(\\S+(\\s|\\S)*\\S+|\\S)$|^$"
-        },
         "lot_id_alias": {
             "title": "Lot ID aliases",
             "description": "The lot identifiers for this lot deemed to be exactly the same by the vendor.",

src/encoded/schemas/biosample.json

@@ -357,7 +357,7 @@
     },
     "properties": {
         "schema_version": {
-            "default": "25"
+            "default": "26"
         },
         "accession": {
             "accessionType": "BS"

src/encoded/schemas/changelogs/antibody_lot.md

@@ -1,5 +1,8 @@
 ## Changelog for antibody_lot.json
 
+### Schema version 10
+* Modified regex pattern for *lot_id* and *product_id* to disallow blank strings.
+
 ### Minor changes since schema version 9
 
 * Added *single cell* to *biosample_type* enums

src/encoded/schemas/changelogs/biosample.md

@@ -1,5 +1,9 @@
 ## Changelog for biosample.json
 
+### Schema version 26
+
+* Modified regex pattern for *lot_id* and *product_id* to disallow blank strings.
+
 ### Minor changes since schema version 25
 
 * Added *simple_summary* truncated version of biosample_summary as calculated property

src/encoded/schemas/changelogs/library.md

@@ -1,5 +1,8 @@
 ## Changelog for library.json
 
+### Schema version 19
+* Modified regex pattern for *lot_id* and *product_id* to disallow blank strings.
+
 ### Schema version 18
 * The *nucleic_acid_starting_quantity* property must now be formatted as a number.
 

src/encoded/schemas/changelogs/treatment.md

@@ -1,5 +1,8 @@
 ## Changelog for treatment.json
 
+### Schema version 13
+* Modified regex pattern for *product_id* to disallow blank strings.
+
 ### Schema version 12
 * Moved *stimulation* from *treatment_type* to *purpose*
 

src/encoded/schemas/library.json

@@ -223,7 +223,7 @@
     },
     "properties": {
         "schema_version": {
-            "default": "18"
+            "default": "19"
         },
         "accession": {
             "accessionType": "LB"

src/encoded/schemas/mixins.json

@@ -315,15 +315,15 @@
             "title": "Product ID",
             "description": "The product identifier provided by the originating lab or vendor.",
             "type": "string",
-            "pattern": "^(\\S+(\\s|\\S)*\\S+|\\S)$|^$"
+            "pattern": "^(\\S+(\\s|\\S)*\\S+|\\S)$"
         }
     },
     "lot_id": {
         "lot_id":{
     	    "title": "Lot ID",
     	    "description": "The lot identifier provided by the originating lab or vendor.",
     	    "type": "string",
-            "pattern": "^(\\S+(\\s|\\S)*\\S+|\\S)$|^$"
+            "pattern": "^(\\S+(\\s|\\S)*\\S+|\\S)$"
         }
     },
     "source": {

src/encoded/schemas/treatment.json

@@ -28,7 +28,7 @@
     },
     "properties": {
         "schema_version": {
-            "default": "12"
+            "default": "13"
         },
         "documents": {
             "description": "Documents that describe the treatment protocol."

src/encoded/tests/test_upgrade_antibody_lot.py

@@ -32,3 +32,12 @@ def test_antibody_lot_unique_array(upgrader, antibody_lot_4):
     value = upgrader.upgrade('antibody_lot', antibody_lot_4, current_version='4', target_version='5')
     assert len(value['purifications']) == len(set(value['purifications']))
     assert len(value['lot_id_alias']) == len(set(value['lot_id_alias']))
+
+
+def test_antibody_lot_upgrade_9_10(upgrader, antibody_lot_base):
+    antibody_lot_base['lot_id'] = ''
+    antibody_lot_base['product_id'] = ''
+    value = upgrader.upgrade('antibody_lot', antibody_lot_base, current_version='9', target_version='10')
+    assert value['schema_version'] == '10'
+    assert value['lot_id'] == 'Placeholder'
+    assert value['product_id'] == 'Placeholder'

src/encoded/tests/test_upgrade_biosample.py

@@ -339,3 +339,12 @@ def test_upgrade_biosample_24_to_25(upgrader, biosample_with_disease):
     assert value['schema_version'] == '25'
     assert isinstance(value['disease_term_id'], list)
     assert value['disease_term_id'] == ['DOID:0080600']
+
+
+def test_upgrade_biosample_25_to_26(upgrader, biosample):
+    biosample['lot_id'] = ''
+    biosample['product_id'] = ''
+    value = upgrader.upgrade('biosample', biosample, current_version='25', target_version='26')
+    assert value['schema_version'] == '26'
+    assert 'lot_id' not in value
+    assert 'product_id' not in value

src/encoded/tests/test_upgrade_library.py

@@ -135,3 +135,12 @@ def test_library_upgrade_17_to_18(upgrader, library_schema_17):
     value = upgrader.upgrade('library', library_schema_17, target_version='18')
     assert value['schema_version'] == '18'
     assert value['nucleic_acid_starting_quantity'] == 9.2
+
+
+def test_library_upgrade_18_to_19(upgrader, base_library):
+    base_library['lot_id'] = ''
+    base_library['product_id'] = ''
+    value = upgrader.upgrade('library', base_library, current_version='18', target_version='19')
+    assert value['schema_version'] == '19'
+    assert 'lot_id' not in value
+    assert 'product_id' not in value

src/encoded/tests/test_upgrade_treatment.py

@@ -71,3 +71,10 @@ def test_treatment_upgrade_11_12(upgrader, treatment_11):
     value = upgrader.upgrade('treatment', treatment_11, current_version='11', target_version='12')
     assert value['treatment_type'] == 'chemical'
     assert value['purpose'] == 'stimulation'
+
+
+def test_treatment_upgrade_12_13(upgrader, treatment_12):
+    treatment_12['product_id'] = ''
+    value = upgrader.upgrade('treatment', treatment_12, current_version='12', target_version='13')
+    assert value['schema_version'] == '13'
+    assert 'product_id' not in value

src/encoded/upgrade/antibody_lot.py

@@ -81,3 +81,12 @@ def antibody_lot_7_8(value, system):
 def antibody_lot_8_9(value, system):
     # http://redmine.encodedcc.org/issues/5041
     return
+
+
+@upgrade_step('antibody_lot', '9', '10')
+def antibody_lot_9_10(value, system):
+    # https://encodedcc.atlassian.net/browse/ENCD-6102
+    if 'lot_id' in value and value['lot_id'] == '':
+        value['lot_id'] = 'Placeholder'
+    if 'product_id' in value and value['product_id'] == '':
+        value['product_id'] = 'Placeholder'

src/encoded/upgrade/biosample.py

@@ -376,3 +376,12 @@ def biosample_24_25(value, system):
     # https://encodedcc.atlassian.net/browse/ENCD-5466
     if 'disease_term_id' in value:
         value['disease_term_id'] = [value['disease_term_id']]
+
+
+@upgrade_step('biosample', '25', '26')
+def biosample_25_26(value, system):
+    # https://encodedcc.atlassian.net/browse/ENCD-6102
+    if 'lot_id' in value and value['lot_id'] == '':
+        value.pop('lot_id')
+    if 'product_id' in value and value['product_id'] == '':
+        value.pop('product_id')

src/encoded/upgrade/library.py

@@ -327,3 +327,12 @@ def library_17_18(value, system):
         except ValueError:
             value['nucleic_acid_starting_quantity'] = 0
             value['notes'] = (notes + ' The nucleic_acid_starting_quantity of this library could not be upgraded: '+ quantity).strip()
+
+
+@upgrade_step('library', '18', '19')
+def library_18_19(value, system):
+    # https://encodedcc.atlassian.net/browse/ENCD-6102
+    if 'lot_id' in value and value['lot_id'] == '':
+        value.pop('lot_id')
+    if 'product_id' in value and value['product_id'] == '':
+        value.pop('product_id')

src/encoded/upgrade/treatment.py

@@ -105,3 +105,10 @@ def treatment_11_12(value, system):
     if value.get('treatment_type') == 'stimulation':
         value['treatment_type'] = 'chemical'
         value['purpose'] = 'stimulation'
+
+
+@upgrade_step('treatment', '12', '13')
+def treatment_12_13(value, system):
+    # https://encodedcc.atlassian.net/browse/ENCD-6102
+    if 'product_id' in value and value['product_id'] == '':
+        value.pop('product_id')