mmar-mapi 1.0.19__py3-none-any.whl → 1.0.21__py3-none-any.whl

mmar_mapi/api.py

@@ -71,13 +71,18 @@ class TextExtractorAPI:
         raise NotImplementedError
 
 
-PageRange = Annotated[tuple[int, int], AfterValidator(lambda rng: rng[0] <= rng[1])]
+def _validate_page_range(v: tuple[int, int]) -> tuple[int, int]:
+    if v[0] < 1 or v[1] < v[0]:
+        raise ValueError("Invalid page range: start must be ≥ 1 and end must be ≥ start.")
+    return v
+
+
+PageRange = Annotated[tuple[int, int], AfterValidator(_validate_page_range)]
 ForceOCR = StrEnum("ForceOCR", ["ENABLED", "DISABLED", "AUTO"])
 OutputType = StrEnum("OutputType", ["RAW", "PLAIN", "MARKDOWN"])
 
 
-class DocExtractionSpec(BaseModel):
-    page_range: PageRange | None = None
+class ExtractionEngineSpec(BaseModel):
     output_type: OutputType = OutputType.MARKDOWN
     force_ocr: ForceOCR = ForceOCR.AUTO
     do_ocr: bool = False
@@ -88,19 +93,28 @@ class DocExtractionSpec(BaseModel):
     generate_page_images: bool = False
     images_scale: float = 2.0
 
+
+class DocExtractionSpec(BaseModel):
+    page_range: PageRange | None = None
+    engine: ExtractionEngineSpec = ExtractionEngineSpec()
+
     def _update(self, **update):
         return self.model_copy(update=update)
 
+    def _update_engine(self, **engine_update):
+        return self._update(engine=self.engine.model_copy(update=engine_update))
+
     # fmt: off
-    def with_output_type_raw(self): return self._update(output_type=OutputType.RAW)
-    def with_output_type_plain(self): return self._update(output_type=OutputType.PLAIN)
-    def with_ocr(self): return self._update(do_ocr=True)
-    def with_tables(self): return self._update(do_table_structure=True, do_cell_matching=True)
-    def with_images(self): return self._update(do_image_extraction=True)
-    def with_annotations(self): return self._update(do_annotations=True)
-    def with_force_ocr_enabled(self): return self._update(force_ocr=ForceOCR.ENABLED)
-    def with_force_ocr_disabled(self): return self._update(force_ocr=ForceOCR.DISABLED)
-    def with_page_images(self): return self._update(generate_page_images=True)
+    def with_output_type_raw(self): return self._update_engine(output_type=OutputType.RAW)
+    def with_output_type_plain(self): return self._update_engine(output_type=OutputType.PLAIN)
+    def with_ocr(self): return self._update_engine(do_ocr=True)
+    def with_tables(self): return self._update_engine(do_table_structure=True, do_cell_matching=True)
+    def with_images(self): return self._update_engine(do_image_extraction=True)
+    def with_annotations(self): return self._update_engine(do_annotations=True)
+    def with_force_ocr_enabled(self): return self._update_engine(force_ocr=ForceOCR.ENABLED)
+    def with_force_ocr_disabled(self): return self._update_engine(force_ocr=ForceOCR.DISABLED)
+    def with_page_images(self): return self._update_engine(generate_page_images=True)
+
     def with_page_range(self, page_range: PageRange): return self._update(page_range=page_range)
     # fmt: on
 
@@ -130,7 +144,7 @@ class ExtractedPageImage(ExtractedImage):
 
 
 class DocExtractionOutput(BaseModel):
-    config: DocExtractionSpec
+    spec: DocExtractionSpec
     text: str = ""
     tables: list[ExtractedTable] = []
     pictures: list[ExtractedPicture] = []
@@ -138,6 +152,6 @@ class DocExtractionOutput(BaseModel):
 
 
 class DocumentExtractorAPI:
-    def extract(self, *, resource_id: ResourceId, spec: DocExtractionSpec) -> ResourceId:
+    def extract(self, *, resource_id: ResourceId, spec: DocExtractionSpec) -> ResourceId | None:
         """returns file with DocExtractionOutput"""
         raise NotImplementedError

mmar_mapi/file_storage.py

@@ -96,6 +96,9 @@ class FileStorage:
         res = self.download_text(resource_id).split("\n")
         return res
 
+    def get_path(self, resource_id: ResourceId | None) -> Path | None:
+        return self._get_path(resource_id)
+
     def _get_path(self, resource_id: ResourceId | None) -> Path | None:
         if not resource_id:
             return None

{mmar_mapi-1.0.19.dist-info → mmar_mapi-1.0.21.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mmar-mapi
-Version: 1.0.19
+Version: 1.0.21
 Summary: Common pure/IO utilities for multi-modal architectures team
 Keywords: 
 Author: Eugene Tagin

{mmar_mapi-1.0.19.dist-info → mmar_mapi-1.0.21.dist-info}/RECORD

@@ -1,7 +1,7 @@
 mmar_mapi/__init__.py,sha256=9Q5xsrj26uUnn7ZWvvJUvdVIuzC2oCIeNB4dEoqjF-o,1256
-mmar_mapi/api.py,sha256=R9v-1QQWocj5OjNk70T4XnEUTBYGujlwBFurbodiBZA,4373
+mmar_mapi/api.py,sha256=8e_C3sfzX67bLDhgFJlUovIlhYN1Lw9ip5qCRpCCVVM,4869
 mmar_mapi/decorators_maybe_lru_cache.py,sha256=eO2I6t1fHLUNRABClK1c8EZzHAmCeSK6O-hbJGb2c9E,444
-mmar_mapi/file_storage.py,sha256=kxh2DcKY1M9MMb-U03doDYmowHH9VoGYetqBubIJhLI,4937
+mmar_mapi/file_storage.py,sha256=xJU59HmXFsfc53XALdx53IwyqV_k4218AzzXq1Q65Js,5052
 mmar_mapi/models/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 mmar_mapi/models/base.py,sha256=mKtXV2x51XVj7W-et9tjGcPMDUUUMelW-BywMgFc2p0,411
 mmar_mapi/models/chat.py,sha256=-XilkiderIOFG1oSKRDG9NDOEN21sBpbTPHUrqVPjc4,15400
@@ -13,7 +13,7 @@ mmar_mapi/type_union.py,sha256=diwmzcnbqkpGFckPHNw9o8zyQ955mOGNvhTlcBJ0RMI,1905
 mmar_mapi/utils.py,sha256=FlW9n-84xz2zSHsahHzJ3Y4Wu5mjpFer6t9z6PF6lS0,488
 mmar_mapi/utils_import.py,sha256=pUyMFd8SItTxBKI-GO9JhRmy43jG_OQlUPr8QCBOSwg,1682
 mmar_mapi/xml_parser.py,sha256=VvLIX_XCZao9i0qqpTVx8nx0vbFXSe8pEbdJdXnj97g,568
-mmar_mapi-1.0.19.dist-info/licenses/LICENSE,sha256=2A90w8WjhOgQXnFuUijKJYazaqZ4_NTokYb9Po4y-9k,1061
-mmar_mapi-1.0.19.dist-info/WHEEL,sha256=-neZj6nU9KAMg2CnCY6T3w8J53nx1kFGw_9HfoSzM60,79
-mmar_mapi-1.0.19.dist-info/METADATA,sha256=iFhcn0K4RZtc0xJSmsQM1fBbrbG1XvmMTeuC0XSdAHc,944
-mmar_mapi-1.0.19.dist-info/RECORD,,
+mmar_mapi-1.0.21.dist-info/licenses/LICENSE,sha256=2A90w8WjhOgQXnFuUijKJYazaqZ4_NTokYb9Po4y-9k,1061
+mmar_mapi-1.0.21.dist-info/WHEEL,sha256=-neZj6nU9KAMg2CnCY6T3w8J53nx1kFGw_9HfoSzM60,79
+mmar_mapi-1.0.21.dist-info/METADATA,sha256=EFNEhMdQS1WL5gCxZFSqaqBGdQQRriWcWh9aPFkCm2k,944
+mmar_mapi-1.0.21.dist-info/RECORD,,