ocr-stream 2.3.8__tar.gz → 2.4__tar.gz

{ocr_stream-2.3.8 → ocr_stream-2.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ocr_stream
-Version: 2.3.8
+Version: 2.4
 Summary: Extrair textos de documentos digitalizados e imagens.
 Author: Bruno Chaves
 Requires-Python: >=3.11
@@ -12,7 +12,7 @@ Requires-Dist: pandas
 Requires-Dist: requests>=2.32.4
 Requires-Dist: opencv-python
 Requires-Dist: pytesseract
-Requires-Dist: convert_stream>=2.3.8
+Requires-Dist: convert_stream>=2.4
 Requires-Dist: soup_files>=1.4
 
 # ocrlib

{ocr_stream-2.3.8 → ocr_stream-2.4}/ocr_stream/__init__.py

@@ -9,9 +9,9 @@ from ocr_stream.recognize import RecognizeImage, RecognizePdf
 from convert_stream import DocumentPdf, PageDocumentPdf
 from convert_stream.image import ImageObject, CollectionImages
 from convert_stream.mod_types import (
-    ColumnsTable, ColumnBody, HeadCell, HeadValues, create_void_dict_map
+    ColumnsTable, ColumnBody, HeadCell, HeadValues, DictTextTable
 )
-from time import sleep
+
 from convert_stream.text import FindText
 from convert_stream.pdf import ConvertPdfToImages, CollectionPagePdf
 import pandas as pd
@@ -122,7 +122,7 @@ class SearchableTextImage(object):
     """
     def __init__(self, silent: bool = False):
         self.silent = silent
-        self.elements: dict[str, ColumnBody] = create_void_dict_map()
+        self.elements: DictTextTable = DictTextTable.create_void_dict()
 
     def __repr__(self):
         return f'SearchableTextPdf: {self.elements}'

{ocr_stream-2.3.8 → ocr_stream-2.4}/ocr_stream/modules/__init__.py

@@ -8,11 +8,13 @@ class LibOcr(Enum):
 
     PYTESSERACT = "pytesseract"
     PYOCR = "pyocr"
+    EASYOCR = "easyocr"
     NOT_IMPLEMENTED = "not_implemented"
 
 
 MOD_PYTESSERACT: bool = False
 MOD_PYOCR: bool = False
+MOD_EASYOCR: bool = False
 
 try:
     from pytesseract import pytesseract, Output
@@ -27,16 +29,24 @@ try:
     import pyocr.tesseract
     MOD_PYOCR = True
 except Exception as err:
-    print(err)
+    #print(err)
     pyocr = object
 
 
-if MOD_PYTESSERACT and MOD_PYOCR:
-    ModuleOcr: TypeAlias = Union[pyocr, pytesseract]
+try:
+    import easyocr
+    MOD_EASYOCR = True
+except Exception as err:
+    easyocr = object
+
+if MOD_PYTESSERACT and MOD_PYOCR and MOD_EASYOCR:
+    ModuleOcr: TypeAlias = Union[esasyocr, pytesseract, pyocr]
     DEFAULT_LIB_OCR = LibOcr.PYTESSERACT
 elif MOD_PYTESSERACT:
     ModuleOcr: TypeAlias = Union[pytesseract]
     DEFAULT_LIB_OCR = LibOcr.PYTESSERACT
+elif MOD_EASYOCR:
+    ModuleOcr: TypeAlias = Union[easyocr]
 elif MOD_PYOCR:
     ModuleOcr: TypeAlias = Union[pyocr]
     DEFAULT_LIB_OCR = LibOcr.PYOCR
@@ -44,5 +54,3 @@ else:
     ModuleOcr: TypeAlias = None
     DEFAULT_LIB_OCR = LibOcr.NOT_IMPLEMENTED
 
-
-__all__ = ['LibOcr', 'DEFAULT_LIB_OCR']

{ocr_stream-2.3.8 → ocr_stream-2.4}/ocr_stream/ocr/__init__.py

@@ -13,16 +13,10 @@ from convert_stream.image import ImageObject
 from convert_stream.mod_types import MetaDataItem, MetaDataFile
 from convert_stream.mod_types.table_types import ColumnBody
 from ocr_stream.modules import (
-    LibOcr, DEFAULT_LIB_OCR, pytesseract, Output, pyocr,
+    LibOcr, DEFAULT_LIB_OCR, pytesseract, Output, pyocr, easyocr
 )
 from ocr_stream.bin_tess import BinTesseract
 
-try:
-    import pyocr.tesseract
-except Exception as err:
-    print(err)
-    pyocr = None
-
 
 class TextRecognized(object):
     """
@@ -142,6 +136,78 @@ class IPytesseract(ABCOcrTesseract):
         return TextRecognized(bt)
 
 
+# ======================================================================#
+# Modulo easyocr
+# ======================================================================#
+class IEasyOcr(ABCOcrTesseract):
+
+    def __init__(self, bin_tess: BinTesseract, *, lang: str = 'pt', tess_data_dir: str = None):
+        super().__init__(bin_tess, lang=lang, tess_data_dir=tess_data_dir)
+        self.reader: easyocr.Reader = easyocr.Reader([self.lang], gpu=True)
+
+    def to_string(self, img: Union[File, ImageObject]) -> str:
+        if isinstance(img, File):
+            img = ImageObject(img)
+        result = self.reader.readtext(img.to_opencv())
+        text = '\n'.join([res[1] for res in result])
+        return text
+
+    def to_reconize(self, img: Union[File, ImageObject]) -> TextRecognized:
+        import cv2
+        import numpy as np
+        from PIL import Image
+        from io import BytesIO
+        from reportlab.pdfgen import canvas
+        from reportlab.lib.pagesizes import A4
+        from reportlab.lib.utils import ImageReader
+
+        # 1️⃣ Converter o objeto de entrada para imagem OpenCV
+        if isinstance(img, File):
+            img = ImageObject(img)
+        img_cv = img.to_opencv()
+        img_h, img_w = img_cv.shape[:2]
+
+        # 2️⃣ Reconhecer o texto com EasyOCR
+        results = self.reader.readtext(img_cv)
+
+        # 3️⃣ Converter a imagem para PDF (imagem de fundo)
+        buffer = BytesIO()
+        pdf = canvas.Canvas(buffer, pagesize=(img_w, img_h))
+
+        # Desenha a imagem original como fundo
+        img_pil = Image.fromarray(cv2.cvtColor(img_cv, cv2.COLOR_BGR2RGB))
+        pdf.drawImage(ImageReader(img_pil), 0, 0, width=img_w, height=img_h)
+
+        # 4️⃣ Adiciona o texto OCR como camada "invisível" sobre a imagem
+        pdf.setFont("Helvetica", 8)
+        pdf.setFillColorRGB(1, 1, 1, alpha=0)  # texto invisível
+
+        for (bbox, text, conf) in results:
+            if not text.strip():
+                continue
+            # Coordenadas do bounding box (EasyOCR retorna 4 pontos)
+            (x1, y1), (x2, y2), (x3, y3), (x4, y4) = bbox
+
+            # Média da posição vertical
+            avg_y = (y1 + y2 + y3 + y4) / 4
+
+            # Posição invertida (PDF tem origem no canto inferior esquerdo)
+            y_pdf = img_h - avg_y
+
+            # Largura estimada
+            text_width = abs(x2 - x1)
+            pdf.drawString(x1, y_pdf, text)
+
+        pdf.showPage()
+        pdf.save()
+        buffer.seek(0)
+
+        # 5️⃣ Retorna como TextRecognized (bytes do PDF com texto embutido)
+        text_rec = TextRecognized(buffer.read())
+        text_rec.metadata = img.metadata if hasattr(img, "metadata") else MetaDataFile()
+        return text_rec
+
+
 # ======================================================================#
 # Modulo OCR pyocr
 # ======================================================================#
@@ -206,6 +272,10 @@ class TesseractOcr(object):
             md = IPytesseract(bin_tess, lang=lang, tess_data_dir=tess_data_dir)
         elif lib_ocr == LibOcr.PYOCR:
             md = IPyOcr(bin_tess, lang=lang, tess_data_dir=tess_data_dir)
+        elif lib_ocr == LibOcr.EASYOCR:
+            if lang is None:
+                lang = 'pt'
+            md = IEasyOcr(bin_tess, lang=lang, tess_data_dir=tess_data_dir)
         else:
             raise ValueError(f'{lib_ocr} is not supported')
         return cls(md)

{ocr_stream-2.3.8 → ocr_stream-2.4}/ocr_stream.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ocr_stream
-Version: 2.3.8
+Version: 2.4
 Summary: Extrair textos de documentos digitalizados e imagens.
 Author: Bruno Chaves
 Requires-Python: >=3.11
@@ -12,7 +12,7 @@ Requires-Dist: pandas
 Requires-Dist: requests>=2.32.4
 Requires-Dist: opencv-python
 Requires-Dist: pytesseract
-Requires-Dist: convert_stream>=2.3.8
+Requires-Dist: convert_stream>=2.4
 Requires-Dist: soup_files>=1.4
 
 # ocrlib

{ocr_stream-2.3.8 → ocr_stream-2.4}/ocr_stream.egg-info/requires.txt

@@ -5,5 +5,5 @@ pandas
 requests>=2.32.4
 opencv-python
 pytesseract
-convert_stream>=2.3.8
+convert_stream>=2.4
 soup_files>=1.4

{ocr_stream-2.3.8 → ocr_stream-2.4}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "ocr_stream"
-version = "2.3.8"
+version = "2.4"
 authors = [
   { name="Bruno Chaves"},
 ]
@@ -21,7 +21,7 @@ dependencies = [
     "requests>=2.32.4",
     "opencv-python",
     "pytesseract",
-    "convert_stream>=2.3.8",
+    "convert_stream>=2.4",
     "soup_files>=1.4",
 ]