Integrate JSONL with recent changes

src/mc_optimade/examples/direct_from_jsonl/example.jsonl

@@ -1,3 +1,4 @@
+{"x-optimade": {"api_version": "1.2.0"}}
 {"formats": ["json"], "description": "structures", "properties": {"_mcloudarchive_energy": {"description": "The total energy per atom as computed by DFT", "unit": "eV/atom", "sortable": null, "type": "float"}, "_mcloudarchive_property_b": {"description": "Alias for some more complicated property_b", "unit": null, "sortable": null, "type": "float"}, "_mcloudarchive_structure_description": {"description": "Provides a human-readable description for this particular entry_type", "unit": null, "sortable": null, "type": "string"}}, "output_fields_by_format": {"json": ["_mcloudarchive_energy", "_mcloudarchive_property_b", "_mcloudarchive_structure_description"]}}
 {"id": "structures.zip/structures/cifs/cc1a41b1-a841-4818-baf1-a6c1441dc52a.cif", "type": "structures", "links": null, "meta": null, "attributes": {"immutable_id": null, "last_modified": null, "elements": ["B", "Ir", "Mg", "Zn"], "nelements": 4, "elements_ratios": [0.2, 0.5, 0.2, 0.1], "chemical_formula_descriptive": "B4Ir10Mg4Zn2", "chemical_formula_reduced": "B2Ir5Mg2Zn", "chemical_formula_hill": null, "chemical_formula_anonymous": "A5B2C2D", "dimension_types": [1, 1, 1], "nperiodic_dimensions": 3, "lattice_vectors": [[9.4623270052342, 0.0, 0.0], [0.0, 9.4623270052342, 0.0], [0.0, 0.0, 2.9327245575729]], "cartesian_site_positions": [[3.0564738352234, 1.6746896673437002, 0.0], [7.7876373378905, 3.0564738352234, 0.0], [6.4058531699108, 7.7876373378905, 0.0], [1.6746896673437002, 6.4058531699108, 0.0], [4.7311635025671, 4.7311635025671, 0.0], [0.0, 0.0, 0.0], [8.2578546699396, 5.9356358378617, 0.0], [3.5266911673725, 8.2578546699396, 0.0], [1.2044723352946, 3.5266911673725, 0.0], [5.9356358378617, 1.2044723352946, 0.0], [4.0683890403188, 6.7751547390362, 1.4663622787864], [8.7995525428859, 7.418335768765, 1.4663622787864], [2.6871722661979005, 4.0683890403188, 1.4663622787864], [7.418335768765, 0.66277446224826, 1.4663622787864], [5.3939379649153, 2.6871722661979005, 1.4663622787864], [0.66277446224826, 2.0439912363691, 1.4663622787864], [6.7751547390362, 5.3939379649153, 1.4663622787864], [2.0439912363691, 8.7995525428859, 1.4663622787864], [4.7311635025671, 0.0, 1.4663622787864], [0.0, 4.7311635025671, 1.4663622787864]], "nsites": 20, "species": [{"name": "Ir", "chemical_symbols": ["Ir"], "concentration": [1.0], "mass": null, "original_name": null, "attached": null, "nattached": null}, {"name": "Mg", "chemical_symbols": ["Mg"], "concentration": [1.0], "mass": null, "original_name": null, "attached": null, "nattached": null}, {"name": "B", "chemical_symbols": ["B"], "concentration": [1.0], "mass": null, "original_name": null, "attached": null, "nattached": null}, {"name": "Zn", "chemical_symbols": ["Zn"], "concentration": [1.0], "mass": null, "original_name": null, "attached": null, "nattached": null}], "species_at_sites": ["Mg", "Mg", "Mg", "Mg", "Zn", "Zn", "B", "B", "B", "B", "Ir", "Ir", "Ir", "Ir", "Ir", "Ir", "Ir", "Ir", "Ir", "Ir"], "assemblies": null, "structure_features": [], "_mcloudarchive_energy": -0.45, "_mcloudarchive_property_b": 0.86, "_mcloudarchive_structure_description": "describing something else"}, "relationships": null}
 {"id": "structures.zip/structures/cifs/991bec7a-b3a8-49af-ba6d-be5afd685cd4.cif", "type": "structures", "links": null, "meta": null, "attributes": {"immutable_id": null, "last_modified": null, "elements": ["C", "Sr"], "nelements": 2, "elements_ratios": [0.5, 0.5], "chemical_formula_descriptive": "CSr", "chemical_formula_reduced": "CSr", "chemical_formula_hill": null, "chemical_formula_anonymous": "AB", "dimension_types": [1, 1, 1], "nperiodic_dimensions": 3, "lattice_vectors": [[4.006498849786306, 0.0, 0.0], [2.0032494248931525, 3.469729784148075, 0.0], [2.0032494248931525, 1.1565765947160247, 3.271292612341386]], "cartesian_site_positions": [[0.0, 0.0, 0.0], [4.006498849786305, 2.31315318943205, 1.635646306170693]], "nsites": 2, "species": [{"name": "C", "chemical_symbols": ["C"], "concentration": [1.0], "mass": null, "original_name": null, "attached": null, "nattached": null}, {"name": "Sr", "chemical_symbols": ["Sr"], "concentration": [1.0], "mass": null, "original_name": null, "attached": null, "nattached": null}], "species_at_sites": ["Sr", "C"], "assemblies": null, "structure_features": [], "_mcloudarchive_energy": -0.55, "_mcloudarchive_property_b": 1.01, "_mcloudarchive_structure_description": NaN}, "relationships": null}

src/mc_optimade/mc_optimade/config.py

@@ -52,7 +52,7 @@ class ParsedFiles(BaseModel):
         description="The path to an archive or file to be unzipped/decompressed."
     )
 
-    matches: list[str] = Field(
+    matches: Optional[list[str]] = Field(
         description="A list of matches to be used to filter the file contents. Each match can use simple '*' wildcard syntax.",
         examples=[["structures/*.cif", "relaxed-structures/1.cif"]],
     )
@@ -78,10 +78,11 @@ class EntryConfig(BaseModel):
 
     @validator("entry_type")
     def check_optimade_entry_type(cls, v):
-        if v not in ("structures", "references") and not v.startswith("_"):
-            raise ValueError(
-                f"OPTIMADE entry type must be either 'structures', 'references', or contain a custom prefix, not {v}"
-            )
+        if not isinstance(v, JSONLConfig):
+            if v not in ("structures", "references") and not v.startswith("_"):
+                raise ValueError(
+                    f"OPTIMADE entry type must be either 'structures', 'references', or contain a custom prefix, not {v}"
+                )
 
         return v
 
@@ -127,10 +128,11 @@ class Config(BaseModel):
 
     @validator("entries")
     def check_one_entry_per_type(cls, v):
-        if len({e.entry_type for e in v}) != len(v):
-            raise ValueError(
-                "Each entry type must be listed only once in the config file."
-            )
+        if not isinstance(v, JSONLConfig):
+            if len({e.entry_type for e in v}) != len(v):
+                raise ValueError(
+                    "Each entry type must be listed only once in the config file."
+                )
         return v
 
     @staticmethod

src/mc_optimade/mc_optimade/convert.py

@@ -66,8 +66,8 @@ def convert_archive(archive_path: Path) -> Path:
     # if the config specifies just a JSON-L, then extract any archives
     # and return the JSONL path
     if isinstance(mc_config.entries, JSONLConfig):
-        if mc_config.entries.archive_file is not None:
-            inflate_archive(Path(archive_path), Path(mc_config.entries.archive_file))
+        if mc_config.entries.file is not None:
+            inflate_archive(Path(archive_path), Path(mc_config.entries.file))
         return Path(archive_path) / mc_config.entries.jsonl_path
 
     # first, decompress any provided data paths