rusto 0.4.0__tar.gz → 0.4.1__tar.gz

{rusto-0.4.0 → rusto-0.4.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: rusto
-Version: 0.4.0
+Version: 0.4.1
 Summary: Pure Rust OCR library for Python based on RapidOCR, Indra OCR, and pure-Rust MNN (rusto-nn)
 Author-email: RustO Contributors <contact@rusto.dev>
 License-Expression: MIT
@@ -15,7 +15,7 @@ Official Python bindings for **RustO!** — a high-performance, pure-Rust OCR en
 
 ## Features
 
-- **Zero-Config Indra OCR**: `RustO.initialize()` automatically downloads and caches the default `indraocr` models on first run from Hugging Face Hub.
+- **Zero-Config PP-OCRv6 Tiny Default**: `RustO.initialize()` automatically downloads and caches the default `ppocrv6-tiny` (FP32, Apache 2.0) models on first run from Hugging Face Hub.
 - **Context Manager Support**: Deterministic native resource cleanup with `with RustO.initialize() as engine:` or `engine.close()`.
 - **Full Document Pipeline**: Neural document unwarping (`with_auto_unwarp()`), page orientation detection (`with_orientation_detection()`), semantic layout analysis, and reading-order recovery (`with_layout_reading()`).
 - **Fluent Builder API**: Chain `.as_lines()`, `.as_words()`, `.as_spatial()`, `.as_tsv()`, or `.with_output(...)` into a single `.run()` terminal method.
@@ -32,7 +32,7 @@ pip install rusto
 ```python
 from rusto import RustO
 
-# Zero-config initialization (auto-downloads Indra OCR models from Hugging Face Hub if not cached)
+# Zero-config initialization (auto-downloads PP-OCRv6 Tiny FP32 models from Hugging Face Hub if not cached)
 with RustO.initialize() as engine:
     result = (
         engine.detect_text("invoice.png")
@@ -110,7 +110,7 @@ Pre-download models into a target cache directory ahead of time (e.g., during Do
 ```python
 from rusto import download_model
 
-paths = download_model(model="indraocr", lang="latin", cache_dir="./models-cache")
+paths = download_model(model="ppocrv6-tiny", cache_dir="./models-cache")
 ```
 
 ## Input Sources & Raw RGB Buffers

{rusto-0.4.0 → rusto-0.4.1}/README.md

@@ -4,7 +4,7 @@ Official Python bindings for **RustO!** — a high-performance, pure-Rust OCR en
 
 ## Features
 
-- **Zero-Config Indra OCR**: `RustO.initialize()` automatically downloads and caches the default `indraocr` models on first run from Hugging Face Hub.
+- **Zero-Config PP-OCRv6 Tiny Default**: `RustO.initialize()` automatically downloads and caches the default `ppocrv6-tiny` (FP32, Apache 2.0) models on first run from Hugging Face Hub.
 - **Context Manager Support**: Deterministic native resource cleanup with `with RustO.initialize() as engine:` or `engine.close()`.
 - **Full Document Pipeline**: Neural document unwarping (`with_auto_unwarp()`), page orientation detection (`with_orientation_detection()`), semantic layout analysis, and reading-order recovery (`with_layout_reading()`).
 - **Fluent Builder API**: Chain `.as_lines()`, `.as_words()`, `.as_spatial()`, `.as_tsv()`, or `.with_output(...)` into a single `.run()` terminal method.
@@ -21,7 +21,7 @@ pip install rusto
 ```python
 from rusto import RustO
 
-# Zero-config initialization (auto-downloads Indra OCR models from Hugging Face Hub if not cached)
+# Zero-config initialization (auto-downloads PP-OCRv6 Tiny FP32 models from Hugging Face Hub if not cached)
 with RustO.initialize() as engine:
     result = (
         engine.detect_text("invoice.png")
@@ -99,7 +99,7 @@ Pre-download models into a target cache directory ahead of time (e.g., during Do
 ```python
 from rusto import download_model
 
-paths = download_model(model="indraocr", lang="latin", cache_dir="./models-cache")
+paths = download_model(model="ppocrv6-tiny", cache_dir="./models-cache")
 ```
 
 ## Input Sources & Raw RGB Buffers

{rusto-0.4.0 → rusto-0.4.1}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "rusto"
-version = "0.4.0"
+version = "0.4.1"
 description = "Pure Rust OCR library for Python based on RapidOCR, Indra OCR, and pure-Rust MNN (rusto-nn)"
 readme = "README.md"
 authors = [{ name = "RustO Contributors", email = "contact@rusto.dev" }]

{rusto-0.4.0 → rusto-0.4.1}/rusto/__init__.py

@@ -1,11 +1,11 @@
 """RustO - High performance, pure-Rust OCR engine for Python.
 
-RustO provides fast, memory-safe text extraction powered by Indra OCR and PaddleOCR
-architectures on the pure-Rust RTen deep learning inference engine.
+RustO provides fast, memory-safe text extraction powered by PaddleOCR and Indra OCR
+architectures on pure-Rust MNN (`rusto-nn`) and RTen (`rusto-rten`) inference engines.
 
 Quick Start:
     >>> from rusto import RustO, InitializeConfig
-    >>> engine = RustO.initialize(InitializeConfig.indra())
+    >>> engine = RustO.initialize()
     >>> result = engine.detect_text("receipt.jpg").with_auto_unwarp().with_layout_reading().as_lines().run()
     >>> if result.is_STRUCTURED():
     ...     for item in result.results:
@@ -169,14 +169,24 @@ class PreRecognitionOptions(_GeneratedPreRecognitionOptions):
 
 
 class RecognitionRunOptions(_GeneratedRecognitionRunOptions):
-    """Phase 5: Recognition inference options."""
+    """Phase 5: Recognition inference options.
+
+    Args:
+        batch_size: Recognition batch size (number of text crops processed concurrently).
+        workers: Number of parallel worker threads for recognition inference (`None` = auto).
+        max_workers: Maximum worker thread count clamp for recognition inference (`None` = engine default 8).
+    """
     def __init__(
         self,
         *,
         batch_size: int | None = None,
+        workers: int | None = None,
+        max_workers: int | None = None,
     ):
         super().__init__(
             batch_size=batch_size,
+            workers=workers,
+            max_workers=max_workers,
         )
 
 
@@ -220,19 +230,32 @@ class DetectionConfig(_GeneratedDetectionConfig):
 
 
 class RecognitionConfig(_GeneratedRecognitionConfig):
-    """Text recognition model configuration."""
+    """Text recognition model configuration.
+
+    Args:
+        enabled: Whether recognition stage is enabled.
+        model_path: Filesystem path to recognition model (.rten or .onnx).
+        dict_path: Filesystem path to dictionary vocabulary file (.txt).
+        rec_batch_num: Recognition batch size. Default: 1.
+        rec_workers: Number of parallel worker threads for recognition (`None` = auto).
+        max_rec_workers: Maximum worker thread count clamp (`None` = engine default 8).
+    """
     def __init__(
         self,
         enabled: bool | None = None,
         model_path: str | None = None,
         dict_path: str | None = None,
         rec_batch_num: int | None = None,
+        rec_workers: int | None = None,
+        max_rec_workers: int | None = None,
     ):
         super().__init__(
             enabled=enabled,
             model_path=model_path,
             dict_path=dict_path,
             rec_batch_num=rec_batch_num,
+            rec_workers=rec_workers,
+            max_rec_workers=max_rec_workers,
         )
 
 
@@ -443,7 +466,7 @@ class InitializeConfig(_GeneratedInitializeConfig):
     @classmethod
     def for_model(
         cls,
-        model: str | None = "indraocr",
+        model: str | None = "ppocrv6-tiny",
         lang: str | None = None,
         model_path: str | None = None,
     ) -> InitializeConfig:
@@ -660,12 +683,18 @@ class DetectTextBuilder:
         options: RecognitionRunOptions | None = None,
         *,
         batch_size: int | None = None,
+        workers: int | None = None,
+        max_workers: int | None = None,
     ) -> DetectTextBuilder:
         """Configure recognition inference options."""
         if options is not None:
             self._recognition = options
         else:
-            self._recognition = RecognitionRunOptions(batch_size=batch_size)
+            self._recognition = RecognitionRunOptions(
+                batch_size=batch_size,
+                workers=workers,
+                max_workers=max_workers,
+            )
         return self
 
     def with_post_recognition(

{rusto-0.4.0 → rusto-0.4.1}/rusto/generated/rusto.py

@@ -1372,22 +1372,62 @@ class _UniffiFfiConverterOptionalTypeDetectionConfig(_UniffiConverterRustBuffer)
         else:
             raise InternalError("Unexpected flag byte for optional type")
 
+class _UniffiFfiConverterUInt32(_UniffiConverterPrimitiveInt):
+    CLASS_NAME = "u32"
+    VALUE_MIN = 0
+    VALUE_MAX = 2**32
+
+    @staticmethod
+    def read(buf):
+        return buf.read_u32()
+
+    @staticmethod
+    def write(value, buf):
+        buf.write_u32(value)
+
+class _UniffiFfiConverterOptionalUInt32(_UniffiConverterRustBuffer):
+    @classmethod
+    def check_lower(cls, value):
+        if value is not None:
+            _UniffiFfiConverterUInt32.check_lower(value)
+
+    @classmethod
+    def write(cls, value, buf):
+        if value is None:
+            buf.write_u8(0)
+            return
+
+        buf.write_u8(1)
+        _UniffiFfiConverterUInt32.write(value, buf)
+
+    @classmethod
+    def read(cls, buf):
+        flag = buf.read_u8()
+        if flag == 0:
+            return None
+        elif flag == 1:
+            return _UniffiFfiConverterUInt32.read(buf)
+        else:
+            raise InternalError("Unexpected flag byte for optional type")
+
 @dataclass
 class RecognitionConfig:
     """
     Text recognition model configuration.
 """
-    def __init__(self, *, enabled:typing.Optional[bool], model_path:typing.Optional[str], dict_path:typing.Optional[str], rec_batch_num:typing.Optional[int]):
+    def __init__(self, *, enabled:typing.Optional[bool], model_path:typing.Optional[str], dict_path:typing.Optional[str], rec_batch_num:typing.Optional[int], rec_workers:typing.Optional[int], max_rec_workers:typing.Optional[int]):
         self.enabled = enabled
         self.model_path = model_path
         self.dict_path = dict_path
         self.rec_batch_num = rec_batch_num
+        self.rec_workers = rec_workers
+        self.max_rec_workers = max_rec_workers
         
         
 
     
     def __str__(self):
-        return "RecognitionConfig(enabled={}, model_path={}, dict_path={}, rec_batch_num={})".format(self.enabled, self.model_path, self.dict_path, self.rec_batch_num)
+        return "RecognitionConfig(enabled={}, model_path={}, dict_path={}, rec_batch_num={}, rec_workers={}, max_rec_workers={})".format(self.enabled, self.model_path, self.dict_path, self.rec_batch_num, self.rec_workers, self.max_rec_workers)
     def __eq__(self, other):
         if self.enabled != other.enabled:
             return False
@@ -1397,6 +1437,10 @@ class RecognitionConfig:
             return False
         if self.rec_batch_num != other.rec_batch_num:
             return False
+        if self.rec_workers != other.rec_workers:
+            return False
+        if self.max_rec_workers != other.max_rec_workers:
+            return False
         return True
 
 class _UniffiFfiConverterTypeRecognitionConfig(_UniffiConverterRustBuffer):
@@ -1407,6 +1451,8 @@ class _UniffiFfiConverterTypeRecognitionConfig(_UniffiConverterRustBuffer):
             model_path=_UniffiFfiConverterOptionalString.read(buf),
             dict_path=_UniffiFfiConverterOptionalString.read(buf),
             rec_batch_num=_UniffiFfiConverterOptionalInt32.read(buf),
+            rec_workers=_UniffiFfiConverterOptionalUInt32.read(buf),
+            max_rec_workers=_UniffiFfiConverterOptionalUInt32.read(buf),
         )
 
     @staticmethod
@@ -1415,6 +1461,8 @@ class _UniffiFfiConverterTypeRecognitionConfig(_UniffiConverterRustBuffer):
         _UniffiFfiConverterOptionalString.check_lower(value.model_path)
         _UniffiFfiConverterOptionalString.check_lower(value.dict_path)
         _UniffiFfiConverterOptionalInt32.check_lower(value.rec_batch_num)
+        _UniffiFfiConverterOptionalUInt32.check_lower(value.rec_workers)
+        _UniffiFfiConverterOptionalUInt32.check_lower(value.max_rec_workers)
 
     @staticmethod
     def write(value, buf):
@@ -1422,6 +1470,8 @@ class _UniffiFfiConverterTypeRecognitionConfig(_UniffiConverterRustBuffer):
         _UniffiFfiConverterOptionalString.write(value.model_path, buf)
         _UniffiFfiConverterOptionalString.write(value.dict_path, buf)
         _UniffiFfiConverterOptionalInt32.write(value.rec_batch_num, buf)
+        _UniffiFfiConverterOptionalUInt32.write(value.rec_workers, buf)
+        _UniffiFfiConverterOptionalUInt32.write(value.max_rec_workers, buf)
 
 class _UniffiFfiConverterOptionalTypeRecognitionConfig(_UniffiConverterRustBuffer):
     @classmethod
@@ -1871,44 +1921,6 @@ class _UniffiFfiConverterTypeOutputGranularity(_UniffiConverterRustBuffer):
 
 
 
-class _UniffiFfiConverterUInt32(_UniffiConverterPrimitiveInt):
-    CLASS_NAME = "u32"
-    VALUE_MIN = 0
-    VALUE_MAX = 2**32
-
-    @staticmethod
-    def read(buf):
-        return buf.read_u32()
-
-    @staticmethod
-    def write(value, buf):
-        buf.write_u32(value)
-
-class _UniffiFfiConverterOptionalUInt32(_UniffiConverterRustBuffer):
-    @classmethod
-    def check_lower(cls, value):
-        if value is not None:
-            _UniffiFfiConverterUInt32.check_lower(value)
-
-    @classmethod
-    def write(cls, value, buf):
-        if value is None:
-            buf.write_u8(0)
-            return
-
-        buf.write_u8(1)
-        _UniffiFfiConverterUInt32.write(value, buf)
-
-    @classmethod
-    def read(cls, buf):
-        flag = buf.read_u8()
-        if flag == 0:
-            return None
-        elif flag == 1:
-            return _UniffiFfiConverterUInt32.read(buf)
-        else:
-            raise InternalError("Unexpected flag byte for optional type")
-
 @dataclass
 class PreDetectionOptions:
     """
@@ -2191,17 +2203,23 @@ class RecognitionRunOptions:
     """
     Text recognition inference parameters.
 """
-    def __init__(self, *, batch_size:typing.Optional[int]):
+    def __init__(self, *, batch_size:typing.Optional[int], workers:typing.Optional[int], max_workers:typing.Optional[int]):
         self.batch_size = batch_size
+        self.workers = workers
+        self.max_workers = max_workers
         
         
 
     
     def __str__(self):
-        return "RecognitionRunOptions(batch_size={})".format(self.batch_size)
+        return "RecognitionRunOptions(batch_size={}, workers={}, max_workers={})".format(self.batch_size, self.workers, self.max_workers)
     def __eq__(self, other):
         if self.batch_size != other.batch_size:
             return False
+        if self.workers != other.workers:
+            return False
+        if self.max_workers != other.max_workers:
+            return False
         return True
 
 class _UniffiFfiConverterTypeRecognitionRunOptions(_UniffiConverterRustBuffer):
@@ -2209,15 +2227,21 @@ class _UniffiFfiConverterTypeRecognitionRunOptions(_UniffiConverterRustBuffer):
     def read(buf):
         return RecognitionRunOptions(
             batch_size=_UniffiFfiConverterOptionalInt32.read(buf),
+            workers=_UniffiFfiConverterOptionalUInt32.read(buf),
+            max_workers=_UniffiFfiConverterOptionalUInt32.read(buf),
         )
 
     @staticmethod
     def check_lower(value):
         _UniffiFfiConverterOptionalInt32.check_lower(value.batch_size)
+        _UniffiFfiConverterOptionalUInt32.check_lower(value.workers)
+        _UniffiFfiConverterOptionalUInt32.check_lower(value.max_workers)
 
     @staticmethod
     def write(value, buf):
         _UniffiFfiConverterOptionalInt32.write(value.batch_size, buf)
+        _UniffiFfiConverterOptionalUInt32.write(value.workers, buf)
+        _UniffiFfiConverterOptionalUInt32.write(value.max_workers, buf)
 
 class _UniffiFfiConverterOptionalTypeRecognitionRunOptions(_UniffiConverterRustBuffer):
     @classmethod

{rusto-0.4.0 → rusto-0.4.1}/rusto.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: rusto
-Version: 0.4.0
+Version: 0.4.1
 Summary: Pure Rust OCR library for Python based on RapidOCR, Indra OCR, and pure-Rust MNN (rusto-nn)
 Author-email: RustO Contributors <contact@rusto.dev>
 License-Expression: MIT
@@ -15,7 +15,7 @@ Official Python bindings for **RustO!** — a high-performance, pure-Rust OCR en
 
 ## Features
 
-- **Zero-Config Indra OCR**: `RustO.initialize()` automatically downloads and caches the default `indraocr` models on first run from Hugging Face Hub.
+- **Zero-Config PP-OCRv6 Tiny Default**: `RustO.initialize()` automatically downloads and caches the default `ppocrv6-tiny` (FP32, Apache 2.0) models on first run from Hugging Face Hub.
 - **Context Manager Support**: Deterministic native resource cleanup with `with RustO.initialize() as engine:` or `engine.close()`.
 - **Full Document Pipeline**: Neural document unwarping (`with_auto_unwarp()`), page orientation detection (`with_orientation_detection()`), semantic layout analysis, and reading-order recovery (`with_layout_reading()`).
 - **Fluent Builder API**: Chain `.as_lines()`, `.as_words()`, `.as_spatial()`, `.as_tsv()`, or `.with_output(...)` into a single `.run()` terminal method.
@@ -32,7 +32,7 @@ pip install rusto
 ```python
 from rusto import RustO
 
-# Zero-config initialization (auto-downloads Indra OCR models from Hugging Face Hub if not cached)
+# Zero-config initialization (auto-downloads PP-OCRv6 Tiny FP32 models from Hugging Face Hub if not cached)
 with RustO.initialize() as engine:
     result = (
         engine.detect_text("invoice.png")
@@ -110,7 +110,7 @@ Pre-download models into a target cache directory ahead of time (e.g., during Do
 ```python
 from rusto import download_model
 
-paths = download_model(model="indraocr", lang="latin", cache_dir="./models-cache")
+paths = download_model(model="ppocrv6-tiny", cache_dir="./models-cache")
 ```
 
 ## Input Sources & Raw RGB Buffers

{rusto-0.4.0 → rusto-0.4.1}/tests/test_rusto.py

@@ -2,6 +2,7 @@ import unittest
 import sys
 import os
 import json
+import re
 from pathlib import Path
 
 sys.path.insert(0, os.path.abspath(os.path.join(os.path.dirname(__file__), "..")))
@@ -133,7 +134,10 @@ class TestRustOPythonBinding(unittest.TestCase):
         self.assertTrue(callable(RustO.download_model))
         self.assertTrue(callable(version))
         self.assertTrue(callable(is_ready))
-        self.assertEqual(version(), "0.4.0")
+        pyproject = Path(__file__).resolve().parents[1] / "pyproject.toml"
+        m = re.search(r'^version\s*=\s*"([^"]+)"', pyproject.read_text(encoding="utf-8"), re.MULTILINE)
+        self.assertIsNotNone(m)
+        self.assertEqual(version(), m.group(1))
 
     def test_chronam_ad_page_with_layout(self):
         root = Path(__file__).resolve().parents[3]
@@ -158,7 +162,7 @@ class TestRustOPythonBinding(unittest.TestCase):
         )
         res = engine.detect_text(ImageSource.PATH(str(img_path)), opts)
         self.assertTrue(res.is_STRUCTURED())
-        self.assertEqual(len(res.results), 249)
+        self.assertEqual(len(res.results), 243)
         self.assertEqual(res.results[0].text, "All Orders in our")
         self.assertIsNotNone(res.results[0].region_id)
         self.assertIsNotNone(res.results[0].layout_type)
@@ -186,7 +190,7 @@ class TestRustOPythonBinding(unittest.TestCase):
             engine.detect_text(str(img_path))
             .with_layout_reading(True)
             .with_pre_detection(descreen_strength=0.1)
-            .with_recognition(batch_size=6)
+            .with_recognition(batch_size=6, workers=2, max_workers=4)
             .with_post_recognition(score_threshold=0.5)
             .as_lines()
         )