refactor: file repo

Author: nalgeon

README.md

@@ -42,16 +42,17 @@ Transliterate using specified schema:
 Or pick schema by name
 
 ```python
->>> schema = iuliia.Schemas.get("wikipedia")
->>> schema.translate(source)
+>>> iuliia.schemas.has("wikipedia")
+True
+>>> schema = iuliia.schemas.get("wikipedia")
+>>> schema.translate("Юлия Щеглова")
 'Yuliya Shcheglova'
 ```
 
 List all supported schemas:
 
 ```python
-import iuliia
-for name, schema in iuliia.Schemas.items():
+for name, schema in iuliia.schemas.items():
     print("{0:<20}{1}".format(name, schema.description))
 ```
 

iuliia/__init__.py

@@ -3,13 +3,14 @@
 """
 
 from .engine import translate
+from .repo import FileRepo
 from .schema import Schema
-from .schemas import Schemas
 
 __all__ = [
     "translate",
+    "FileRepo",
     "Schema",
-    "Schemas",
+    "schemas",
     "ALA_LC",
     "ALA_LC_ALT",
     "BGN_PCGN",
@@ -40,31 +41,33 @@
     "YANDEX_MONEY",
 ]
 
-ALA_LC = Schemas.ala_lc.value
-ALA_LC_ALT = Schemas.ala_lc_alt.value
-BGN_PCGN = Schemas.bgn_pcgn.value
-BGN_PCGN_ALT = Schemas.bgn_pcgn_alt.value
-BS_2979 = Schemas.bs_2979.value
-BS_2979_ALT = Schemas.bs_2979_alt.value
-GOST_16876 = Schemas.gost_16876.value
-GOST_16876_ALT = Schemas.gost_16876_alt.value
-GOST_52290 = Schemas.gost_52290.value
-GOST_52535 = Schemas.gost_52535.value
-GOST_7034 = Schemas.gost_7034.value
-GOST_779 = Schemas.gost_779.value
-GOST_779_ALT = Schemas.gost_779_alt.value
-ICAO_DOC_9303 = Schemas.icao_doc_9303.value
-ISO_9_1954 = Schemas.iso_9_1954.value
-ISO_9_1968 = Schemas.iso_9_1968.value
-ISO_9_1968_ALT = Schemas.iso_9_1968_alt.value
-MOSMETRO = Schemas.mosmetro.value
-MVD_310 = Schemas.mvd_310.value
-MVD_310_FR = Schemas.mvd_310_fr.value
-MVD_782 = Schemas.mvd_782.value
-SCIENTIFIC = Schemas.scientific.value
-TELEGRAM = Schemas.telegram.value
-UNGEGN_1987 = Schemas.ungegn_1987.value
-UZ = Schemas.uz.value
-WIKIPEDIA = Schemas.wikipedia.value
-YANDEX_MAPS = Schemas.yandex_maps.value
-YANDEX_MONEY = Schemas.yandex_money.value
+schemas = FileRepo()
+
+ALA_LC: Schema = schemas.get("ala_lc")
+ALA_LC_ALT: Schema = schemas.get("ala_lc_alt")
+BGN_PCGN: Schema = schemas.get("bgn_pcgn")
+BGN_PCGN_ALT: Schema = schemas.get("bgn_pcgn_alt")
+BS_2979: Schema = schemas.get("bs_2979")
+BS_2979_ALT: Schema = schemas.get("bs_2979_alt")
+GOST_16876: Schema = schemas.get("gost_16876")
+GOST_16876_ALT: Schema = schemas.get("gost_16876_alt")
+GOST_52290: Schema = schemas.get("gost_52290")
+GOST_52535: Schema = schemas.get("gost_52535")
+GOST_7034: Schema = schemas.get("gost_7034")
+GOST_779: Schema = schemas.get("gost_779")
+GOST_779_ALT: Schema = schemas.get("gost_779_alt")
+ICAO_DOC_9303: Schema = schemas.get("icao_doc_9303")
+ISO_9_1954: Schema = schemas.get("iso_9_1954")
+ISO_9_1968: Schema = schemas.get("iso_9_1968")
+ISO_9_1968_ALT: Schema = schemas.get("iso_9_1968_alt")
+MOSMETRO: Schema = schemas.get("mosmetro")
+MVD_310: Schema = schemas.get("mvd_310")
+MVD_310_FR: Schema = schemas.get("mvd_310_fr")
+MVD_782: Schema = schemas.get("mvd_782")
+SCIENTIFIC: Schema = schemas.get("scientific")
+TELEGRAM: Schema = schemas.get("telegram")
+UNGEGN_1987: Schema = schemas.get("ungegn_1987")
+UZ: Schema = schemas.get("uz")
+WIKIPEDIA: Schema = schemas.get("wikipedia")
+YANDEX_MAPS: Schema = schemas.get("yandex_maps")
+YANDEX_MONEY: Schema = schemas.get("yandex_money")

iuliia/__main__.py

@@ -11,14 +11,14 @@ def main():
     if len(sys.argv) < 3:
         print("usage: iuliia SCHEMA SOURCE")
         print("Supported schemas:")
-        print("\n".join(iuliia.Schemas.names()))
+        print("\n".join(iuliia.schemas.names()))
         sys.exit(1)
 
     schema_name = sys.argv[1]
-    schema = iuliia.Schemas.get(schema_name)
+    schema = iuliia.schemas.get(schema_name)
     if schema is None:
         print(f"Schema '{schema_name}' does not exist. Supported schemas:")
-        print("\n".join(iuliia.Schemas.names()))
+        print("\n".join(iuliia.schemas.names()))
         sys.exit(1)
 
     source = sys.argv[2]

iuliia/lazy.py

@@ -0,0 +1,106 @@
+"""
+File-based schema repository.
+"""
+
+import json
+from pathlib import Path
+from .schema import Schema, TranslitSchema
+from .validator import Validator
+
+SCHEMA_DIR = Path(__file__).parent / "schemas"
+
+
+class FileSchema:
+    """
+    File-based transliteration schema.
+    Translates Cyrillic text into Latin using a given set of rules (mappings).
+    Lazy loads schema from JSON file as needed.
+    """
+
+    def __init__(self, name: str, path: str | Path):
+        self.name = name
+        self.path = Path(path)
+        self._schema: TranslitSchema | None = None
+
+    @property
+    def description(self) -> str | None:
+        """Schema description."""
+        if self._schema is None:
+            self._schema = self._load()
+        return self._schema.description
+
+    @property
+    def samples(self) -> list[list[str]]:
+        """Schema samples."""
+        if self._schema is None:
+            self._schema = self._load()
+        return self._schema.samples
+
+    def translate(self, source: str) -> str:
+        """
+        Translate source Cyrillic string into Latin.
+        Translates the source string word by word.
+        """
+        if self._schema is None:
+            self._schema = self._load()
+        return self._schema.translate(source)
+
+    def _load(self) -> TranslitSchema:
+        """Load schema by its name."""
+        if not self.path.exists():
+            raise ValueError(f"Schema path does not exist: {self.path}")
+
+        # Load and validate schema definition.
+        defn = {}
+        with open(self.path, encoding="utf-8") as file:
+            defn = json.load(file)
+        Validator(defn).run()
+
+        return TranslitSchema(
+            name=defn.get("name", ""),
+            description=defn.get("description"),
+            mapping=defn.get("mapping", {}),
+            prev_mapping=defn.get("prev_mapping"),
+            next_mapping=defn.get("next_mapping"),
+            ending_mapping=defn.get("ending_mapping"),
+            samples=defn.get("samples"),
+        )
+
+    def __str__(self):
+        return self.name
+
+    def __repr__(self):
+        return f"{self.__class__.__name__}('{self.name}')"
+
+
+class FileRepo:
+    """
+    File-based transliteration schema repository.
+    """
+
+    def __init__(self, base_dir: str | Path = SCHEMA_DIR):
+        self.base_dir = base_dir
+        self.schemas = {}
+        for path in Path(base_dir).iterdir():
+            if not path.suffix == ".json":
+                continue
+            name = path.stem
+            self.schemas[name] = FileSchema(name, path)
+
+    def has(self, name: str) -> bool:
+        """Check if schema exists."""
+        return name in self.schemas
+
+    def get(self, name: str) -> Schema:
+        """Get schema by name."""
+        if name not in self.schemas:
+            raise ValueError(f"Schema not found: {name}")
+        return self.schemas[name]
+
+    def names(self) -> list[str]:
+        """Return schema names sorted by name."""
+        return sorted(self.schemas.keys())
+
+    def items(self) -> list[tuple[str, Schema]]:
+        """Return (name, schema) tuples sorted by name."""
+        return sorted(self.schemas.items(), key=lambda item: item[0])

iuliia/repo.py

@@ -12,6 +12,18 @@ class Schema(Protocol):
     using a given set of rules (mappings).
     """
 
+    @property
+    def name(self) -> str | None:
+        """Schema name."""
+
+    @property
+    def description(self) -> str | None:
+        """Schema description."""
+
+    @property
+    def samples(self) -> list[list[str]]:
+        """Schema samples."""
+
     def translate(self, source: str) -> str:
         """
         Translate source Cyrillic string into Latin.