agents concepts places: add other identifiers

* Adds `identifiedBy` for `035` values with `$2` for IDREF. * Update dependencies. Co-Authored-by: Peter Weber <[email protected]>
rero · Jul 12, 2024 · e8d7b50 · e8d7b50
1 parent 0d16ac4
commit e8d7b50
Show file tree

Hide file tree

Showing 14 changed files with 349 additions and 300 deletions.
diff --git a/poetry.lock b/poetry.lock
diff --git a/rero_mef/config.py b/rero_mef/config.py
@@ -96,7 +96,7 @@
 #: Template for error pages.
 THEME_ERROR_TEMPLATE = "rero_mef/page_error.html"
 
-WEBPACKEXT_PROJECT = 'rero_mef.theme.webpack:project'
+WEBPACKEXT_PROJECT = "rero_mef.theme.webpack:project"
 
 # Email configuration
 # ===================

diff --git a/rero_mef/jsonschemas/common/agent-v0.0.1.json b/rero_mef/jsonschemas/common/agent-v0.0.1.json
@@ -86,15 +86,13 @@
           "type": {
             "enum": [
               "bf:Local",
-              "uri"
+              "uri",
+              "bf:Nbn"
             ]
           },
           "source": {
-            "enum": [
-              "GND",
-              "RERO",
-              "IDREF"
-            ]
+            "type": "string",
+            "minLength": 2
           },
           "value": {
             "type": "string",

diff --git a/rero_mef/jsonschemas/common/concept-v0.0.1.json b/rero_mef/jsonschemas/common/concept-v0.0.1.json
@@ -48,15 +48,13 @@
           "type": {
             "enum": [
               "bf:Local",
-              "uri"
+              "uri",
+              "bf:Nbn"
             ]
           },
           "source": {
-            "enum": [
-              "BNF",
-              "RERO",
-              "IDREF"
-            ]
+            "type": "string",
+            "minLength": 2
           },
           "value": {
             "type": "string",
@@ -232,7 +230,8 @@
               "type": {
                 "enum": [
                   "bf:Local",
-                  "uri"
+                  "uri",
+                  "bf:Nbn"
                 ]
               },
               "source": {

diff --git a/rero_mef/jsonschemas/common/place-v0.0.1.json b/rero_mef/jsonschemas/common/place-v0.0.1.json
@@ -47,15 +47,13 @@
           "type": {
             "enum": [
               "bf:Local",
-              "uri"
+              "uri",
+              "bf:Nbn"
             ]
           },
           "source": {
-            "enum": [
-              "BNF",
-              "RERO",
-              "IDREF"
-            ]
+            "type": "string",
+            "minLength": 2
           },
           "value": {
             "type": "string",
@@ -231,7 +229,8 @@
               "type": {
                 "enum": [
                   "bf:Local",
-                  "uri"
+                  "uri",
+                  "bf:Nbn"
                 ]
               },
               "source": {

diff --git a/rero_mef/marctojson/do_idref_agent.py b/rero_mef/marctojson/do_idref_agent.py
@@ -175,11 +175,30 @@ def trans_idref_relation_pid(self):
         if self.logger and self.verbose:
             self.logger.info("Call Function", "trans_idref_relation_pid")
         for field_035 in self.marc.get_fields("035"):
-            if field_035.get("a") and field_035.get("9") and field_035["9"] == "sudoc":
+            subfield_a = field_035.get("a")
+            if isinstance(subfield_a, list):
+                subfield_a = subfield_a[0]
+            subfield_2 = field_035.get("2")
+            if isinstance(subfield_2, list):
+                subfield_2 = subfield_2[0]
+            subfield_9 = field_035.get("9")
+            if isinstance(subfield_9, list):
+                subfield_9 = subfield_9[0]
+            if subfield_a and subfield_9 == "sudoc":
                 self.json_dict["relation_pid"] = {
                     "value": field_035["a"],
                     "type": "redirect_from",
                 }
+            elif subfield_2:
+                identified_by = self.json_dict.get("identifiedBy", [])
+                identified_by.append(
+                    {
+                        "source": subfield_2.upper(),
+                        "type": "uri" if subfield_a.startswith("http") else "bf:Nbn",
+                        "value": subfield_a,
+                    }
+                )
+                self.json_dict["identifiedBy"] = identified_by
 
     def trans_idref_gender(self):
         """Transformation gender 120 $a a:female, b: male, -:not known."""

diff --git a/rero_mef/marctojson/do_idref_concepts.py b/rero_mef/marctojson/do_idref_concepts.py
@@ -104,13 +104,31 @@ def trans_idref_relation_pid(self):
         """Transformation old pids 035 $a $9 = sudoc."""
         if self.logger and self.verbose:
             self.logger.info("Call Function", "trans_idref_relation_pid")
-        fields_035 = self.marc.get_fields("035")
-        for field_035 in fields_035:
-            if field_035.get("a") and field_035.get("9") and field_035["9"] == "sudoc":
+        for field_035 in self.marc.get_fields("035"):
+            subfield_a = field_035.get("a")
+            if isinstance(subfield_a, list):
+                subfield_a = subfield_a[0]
+            subfield_2 = field_035.get("2")
+            if isinstance(subfield_2, list):
+                subfield_2 = subfield_2[0]
+            subfield_9 = field_035.get("9")
+            if isinstance(subfield_9, list):
+                subfield_9 = subfield_9[0]
+            if subfield_a and subfield_9 == "sudoc":
                 self.json_dict["relation_pid"] = {
                     "value": field_035["a"],
                     "type": "redirect_from",
                 }
+            elif subfield_2:
+                identified_by = self.json_dict.get("identifiedBy", [])
+                identified_by.append(
+                    {
+                        "source": subfield_2.upper(),
+                        "type": "uri" if subfield_a.startswith("http") else "bf:Nbn",
+                        "value": subfield_a,
+                    }
+                )
+                self.json_dict["identifiedBy"] = identified_by
 
     def trans_idref_deleted(self):
         """Transformation deleted leader 5 == d."""

diff --git a/rero_mef/marctojson/do_idref_places.py b/rero_mef/marctojson/do_idref_places.py
@@ -91,13 +91,31 @@ def trans_idref_relation_pid(self):
         """Transformation old pids 035 $a $9 = sudoc."""
         if self.logger and self.verbose:
             self.logger.info("Call Function", "trans_idref_relation_pid")
-        fields_035 = self.marc.get_fields("035")
-        for field_035 in fields_035:
-            if field_035.get("a") and field_035.get("9") and field_035["9"] == "sudoc":
+        for field_035 in self.marc.get_fields("035"):
+            subfield_a = field_035.get("a")
+            if isinstance(subfield_a, list):
+                subfield_a = subfield_a[0]
+            subfield_2 = field_035.get("2")
+            if isinstance(subfield_2, list):
+                subfield_2 = subfield_2[0]
+            subfield_9 = field_035.get("9")
+            if isinstance(subfield_9, list):
+                subfield_9 = subfield_9[0]
+            if subfield_a and subfield_9 == "sudoc":
                 self.json_dict["relation_pid"] = {
                     "value": field_035["a"],
                     "type": "redirect_from",
                 }
+            elif subfield_2:
+                identified_by = self.json_dict.get("identifiedBy", [])
+                identified_by.append(
+                    {
+                        "source": subfield_2.upper(),
+                        "type": "uri" if subfield_a.startswith("http") else "bf:Nbn",
+                        "value": subfield_a,
+                    }
+                )
+                self.json_dict["identifiedBy"] = identified_by
 
     def trans_idref_deleted(self):
         """Transformation deleted leader 5 == d."""

diff --git a/scripts/test b/scripts/test
@@ -88,7 +88,7 @@ msg=$(safety check -o text ${safety_exceptions}) || {
   exit 1
 }
 info_msg "Test black:"
-black tests rero_mef --diff
+black tests rero_mef tests --check
 info_msg "Test pydocstyle:"
 pydocstyle rero_mef tests docs
 info_msg "Test isort:"

diff --git a/tests/unit/agents/examples/xml_minimal_record.xml b/tests/unit/agents/examples/xml_minimal_record.xml
@@ -1,10 +1,7 @@
 
         <record>
             <leader>00589nx  a2200193   45  </leader> 
-        <datafield ind1=" " ind2=" " tag="710">
-            <subfield code="a">Paul</subfield>
-            <subfield code="b">VI</subfield>
-            <subfield code="c">pape</subfield>
-            <subfield code="d">1897-1978</subfield>
+        <datafield tag="102" ind1=" " ind2=" ">
+            <subfield code="a">DE</subfield>
         </datafield>
-     </record>
+    </record>
diff --git a/tests/unit/agents/test_agent_idref_transformation.py b/tests/unit/agents/test_agent_idref_transformation.py
@@ -60,11 +60,23 @@ def test_idref_relation_pid():
             <subfield code="C">VIAF</subfield>
             <subfield code="d">20200302</subfield>
         </datafield>
+        <datafield tag="035" ind1=" " ind2=" ">
+            <subfield code="a">A003915957</subfield>
+            <subfield code="2">RERO</subfield>
+        </datafield>
     """
     trans = trans_prep("idref", xml_part_to_add)
     trans.trans_idref_relation_pid()
     assert trans.json == {
-        "relation_pid": {"value": "027630501", "type": "redirect_from"}
+        "identifiedBy": [
+            {
+                "source": "VIAF",
+                "type": "uri",
+                "value": "http://viaf.org/viaf/124265140",
+            },
+            {"source": "RERO", "type": "bf:Nbn", "value": "A003915957"},
+        ],
+        "relation_pid": {"value": "027630501", "type": "redirect_from"},
     }
 
 

diff --git a/tests/unit/concepts/examples/xml_minimal_record.xml b/tests/unit/concepts/examples/xml_minimal_record.xml
@@ -1,55 +1,24 @@
 
         <record>
             <leader>00589nx  a2200193   45  </leader> 
-        <datafield tag="670" ind1=" " ind2=" ">
-          <subfield code="a">
-            Grand Larousse universel (art. : Livre)
-          </subfield>
+        <datafield tag="035" ind1=" " ind2=" ">
+            <subfield code="a">027630501</subfield>
+            <subfield code="9">sudoc</subfield>
         </datafield>
-        <datafield tag="675" ind1=" " ind2=" ">
-          <subfield code="a">
-            Laval RVM (en ligne), 2004-11-23
-          </subfield>
+        <datafield tag="035" ind1=" " ind2=" ">
+            <subfield code="a">frBN001940328</subfield>
         </datafield>
-        <datafield tag="680" ind1=" " ind2=" ">
-          <subfield code="a">
-            Mers profondément engagées dans la masse des continents
-        </subfield>
+        <datafield tag="035" ind1=" " ind2=" ">
+            <subfield code="a">frBN000000089</subfield>
         </datafield>
-        <datafield tag="667" ind1=" " ind2=" ">
-          <subfield code="a">Note interne</subfield>
+        <datafield tag="035" ind1=" " ind2=" ">
+            <subfield code="a">FRBNF118620892</subfield>
+            <subfield code="z">FRBNF11862089</subfield>
         </datafield>
-        <datafield tag="260" ind1=" " ind2="9">
-          <subfield code="a">
-            Voir le descripteur Opposition (science politique)
-        </subfield>
-        </datafield>
-        <datafield tag="260" ind1=" " ind2="9">
-          <subfield code="a">
-            Combiner un des descripteurs Mouvements contestataires
-        </subfield>
-        </datafield>
-        <datafield tag="260" ind1=" " ind2=" ">
-          <subfield code="a">
-            Voir les vedettes : Mouvements contestataires ; Opposition
-        </subfield>
-        </datafield>
-        <datafield tag="260" ind1=" " ind2=" ">
-          <subfield code="a">
-            Voir les vedettes du type : Antifascisme ; Mouvements
-          </subfield>
-        </datafield>
-        <datafield tag="260" ind1=" " ind2=" ">
-          <subfield code="a">
-            Voir aux mouvements d'opposition particuliers, par ex. : Combat
-          </subfield>
-        </datafield>
-        <datafield tag="360" ind1=" " ind2=" ">
-          <subfield code="a">
-            Voir aussi aux mers et océans particuliers
-        </subfield>
-        </datafield>
-        <datafield tag="016" ind1=" " ind2=" ">
-          <subfield code="9">VF3, NC3, NC30</subfield>
+        <datafield tag="035" ind1=" " ind2=" ">
+            <subfield code="a">http://viaf.org/viaf/124265140</subfield>
+            <subfield code="2">VIAF</subfield>
+            <subfield code="C">VIAF</subfield>
+            <subfield code="d">20200302</subfield>
         </datafield>
     </record>
diff --git a/tests/unit/concepts/test_concept_idref_transformation.py b/tests/unit/concepts/test_concept_idref_transformation.py
@@ -443,5 +443,8 @@ def test_idref_relation_pid():
     trans = trans_prep(Transformation, "concepts", xml_part_to_add)
     trans.trans_idref_relation_pid()
     assert trans.json == {
-        "relation_pid": {"value": "027630501", "type": "redirect_from"}
+        "identifiedBy": [
+            {"source": "VIAF", "type": "uri", "value": "http://viaf.org/viaf/124265140"}
+        ],
+        "relation_pid": {"value": "027630501", "type": "redirect_from"},
     }
diff --git a/tests/unit/places/test_places_idref_transformation.py b/tests/unit/places/test_places_idref_transformation.py
@@ -330,5 +330,8 @@ def test_idref_relation_pid():
     trans = trans_prep(Transformation, "places", xml_part_to_add)
     trans.trans_idref_relation_pid()
     assert trans.json == {
-        "relation_pid": {"value": "027630501", "type": "redirect_from"}
+        "identifiedBy": [
+            {"source": "VIAF", "type": "uri", "value": "http://viaf.org/viaf/124265140"}
+        ],
+        "relation_pid": {"value": "027630501", "type": "redirect_from"},
     }