fix function names

Author: sebastianMindee

mindee/image_operations/__init__.py

@@ -0,0 +1 @@
+from mindee.image_operations.image_compressor import compress_image

mindee/pdf/__init__.py

@@ -3,4 +3,5 @@
 from mindee.pdf.pdf_utils import (
     extract_text_from_pdf,
     has_source_text,
+    lerp,
 )

mindee/pdf/pdf_compressor.py

@@ -14,6 +14,7 @@
 from mindee.pdf.pdf_utils import (
     extract_text_from_pdf,
     has_source_text,
+    lerp,
 )
 
 logger = logging.getLogger(__name__)
@@ -61,15 +62,15 @@ def compress_pdf(
         extract_text_from_pdf(pdf_bytes) if not disable_source_text else None
     )
 
-    compressed_pages = compress_pdf_pages(pdf_bytes, image_quality)
+    compressed_pages = _compress_pdf_pages(pdf_bytes, image_quality)
 
     if not compressed_pages:
         logger.warning(
             "Could not compress PDF to a smaller size. Returning original PDF."
         )
         return pdf_bytes
 
-    out_pdf = collect_images_as_pdf(
+    out_pdf = _collect_images_as_pdf(
         [compressed_page_image[0] for compressed_page_image in compressed_pages]
     )
 
@@ -83,7 +84,7 @@ def compress_pdf(
     return out_buffer.read()
 
 
-def compress_pdf_pages(
+def _compress_pdf_pages(
     pdf_data: bytes,
     image_quality: int,
 ) -> Optional[List[Tuple[bytes, int, int]]]:
@@ -98,10 +99,10 @@ def compress_pdf_pages(
     image_quality_loop = image_quality
 
     while image_quality_loop >= MIN_QUALITY:
-        compressed_pages = compress_pages_with_quality(pdf_data, image_quality_loop)
+        compressed_pages = _compress_pages_with_quality(pdf_data, image_quality_loop)
         total_compressed_size = sum(len(page) for page in compressed_pages)
 
-        if is_compression_successful(
+        if _is_compression_successful(
             total_compressed_size, original_size, image_quality
         ):
             return compressed_pages
@@ -146,7 +147,7 @@ def add_text_to_pdf_page(  # type: ignore
         pdfium_c.FPDFPage_GenerateContent(page.raw)
 
 
-def compress_pages_with_quality(
+def _compress_pages_with_quality(
     pdf_data: bytes,
     image_quality: int,
 ) -> List[Tuple[bytes, int, int]]:
@@ -160,15 +161,15 @@ def compress_pages_with_quality(
     pdf_document = pdfium.PdfDocument(pdf_data)
     compressed_pages = []
     for page in pdf_document:
-        rasterized_page = rasterize_page(page, image_quality)
+        rasterized_page = _rasterize_page(page, image_quality)
         compressed_image = compress_image(rasterized_page, image_quality)
         image = Image.open(io.BytesIO(compressed_image))
         compressed_pages.append((compressed_image, image.size[0], image.size[1]))
 
     return compressed_pages
 
 
-def is_compression_successful(
+def _is_compression_successful(
     total_compressed_size: int, original_size: int, image_quality: int
 ) -> bool:
     """
@@ -183,7 +184,7 @@ def is_compression_successful(
     return total_compressed_size + total_compressed_size * overhead < original_size
 
 
-def rasterize_page(  # type: ignore
+def _rasterize_page(  # type: ignore
     page: pdfium.PdfPage,
     quality: int = 85,
 ) -> bytes:
@@ -200,19 +201,7 @@ def rasterize_page(  # type: ignore
     return buffer.getvalue()
 
 
-def lerp(start: float, end: float, t: float) -> float:
-    """
-    Performs linear interpolation between two numbers.
-
-    :param start: The starting value.
-    :param end: The ending value.
-    :param t: The interpolation factor (0 to 1).
-    :return: The interpolated value.
-    """
-    return start * (1 - t) + end * t
-
-
-def collect_images_as_pdf(image_list: List[bytes]) -> pdfium.PdfDocument:  # type: ignore
+def _collect_images_as_pdf(image_list: List[bytes]) -> pdfium.PdfDocument:  # type: ignore
     """
     Converts a list of JPEG images into pages in a PdfDocument.
 

mindee/pdf/pdf_utils.py

@@ -37,12 +37,12 @@ def extract_text_from_pdf(pdf_bytes: bytes) -> List[List[PDFCharData]]:
     char_data_list: List[List[PDFCharData]] = []
 
     for i, page in enumerate(pdf):
-        char_data_list.append(process_page(page, i, pdfium_lock))
+        char_data_list.append(_process_page(page, i, pdfium_lock))
 
     return char_data_list
 
 
-def process_page(page, page_id: int, pdfium_lock: RLock) -> List[PDFCharData]:
+def _process_page(page, page_id: int, pdfium_lock: RLock) -> List[PDFCharData]:
     """
     Processes a single page of the PDF.
 
@@ -59,7 +59,7 @@ def process_page(page, page_id: int, pdfium_lock: RLock) -> List[PDFCharData]:
         count_chars = pdfium_c.FPDFText_CountChars(text_handler)
 
     for i in range(count_chars):
-        concatenated_chars = process_char(
+        concatenated_chars = _process_char(
             i, text_handler, page, pdfium_lock, internal_height, internal_width, page_id
         )
         for concatenated_char in concatenated_chars:
@@ -70,7 +70,7 @@ def process_page(page, page_id: int, pdfium_lock: RLock) -> List[PDFCharData]:
     return char_data_list
 
 
-def process_char(
+def _process_char(
     i: int,
     text_handler,
     page,
@@ -91,21 +91,21 @@ def process_char(
     :param page_id: ID of the page the character was found on.
     :return: List of character data for a page.
     """
-    char_info = get_char_info(i, text_handler, pdfium_lock)
+    char_info = _get_char_info(i, text_handler, pdfium_lock)
     if not char_info:
         return []
-    char_box = get_char_box(i, text_handler, pdfium_lock)
-    rotation = get_page_rotation(page, pdfium_lock)
+    char_box = _get_char_box(i, text_handler, pdfium_lock)
+    rotation = _get_page_rotation(page, pdfium_lock)
 
-    adjusted_box = adjust_char_box(char_box, rotation, internal_height, internal_width)
+    adjusted_box = _adjust_char_box(char_box, rotation, internal_height, internal_width)
     char_data_list: List[PDFCharData] = []
     for c in char_info["char"] or " ":
         if c in (
             "\n",
             "\r",
         ):  # Removes duplicated carriage returns in the PDF due to weird extraction.
             # IDK how to make this better, and neither does Claude, GPT4 nor GPT-o1, so I'm leaving this weird check.
-            next_char_info = get_char_info(i + 1, text_handler, pdfium_lock)
+            next_char_info = _get_char_info(i + 1, text_handler, pdfium_lock)
             if not next_char_info or next_char_info["char"] in ("\n", "\r"):
                 continue
 
@@ -128,7 +128,7 @@ def process_char(
     return char_data_list
 
 
-def get_char_info(i: int, text_handler, pdfium_lock: RLock) -> dict:
+def _get_char_info(i: int, text_handler, pdfium_lock: RLock) -> dict:
     """
     Retrieves information about a specific character.
 
@@ -145,8 +145,8 @@ def get_char_info(i: int, text_handler, pdfium_lock: RLock) -> dict:
         if unicode_char == 0xFF:
             return {}
         char = chr(unicode_char)
-        font_name = get_font_name(text_handler, i)
-        font_flags = get_font_flags(text_handler, i)
+        font_name = _get_font_name(text_handler, i)
+        font_flags = _get_font_flags(text_handler, i)
         font_size = pdfium_c.FPDFText_GetFontSize(text_handler, i)
         font_weight = pdfium_c.FPDFText_GetFontWeight(text_handler, i)
         _ = pdfium_c.FPDFText_GetStrokeColor(
@@ -167,7 +167,7 @@ def get_char_info(i: int, text_handler, pdfium_lock: RLock) -> dict:
     }
 
 
-def get_font_name(text_handler, i: int) -> str:
+def _get_font_name(text_handler, i: int) -> str:
     """
     Retrieves the font name for a specific character.
 
@@ -186,7 +186,7 @@ def get_font_name(text_handler, i: int) -> str:
     )
 
 
-def get_font_flags(text_handler, i: int) -> int:
+def _get_font_flags(text_handler, i: int) -> int:
     """
     Retrieves the font flags for a specific character.
 
@@ -199,7 +199,7 @@ def get_font_flags(text_handler, i: int) -> int:
     return flags.value
 
 
-def get_char_box(
+def _get_char_box(
     i: int, text_handler, pdfium_lock: RLock
 ) -> Tuple[float, float, float, float]:
     """
@@ -218,7 +218,7 @@ def get_char_box(
     return left.value, right.value, bottom.value, top.value
 
 
-def get_page_rotation(page, pdfium_lock: RLock) -> int:
+def _get_page_rotation(page, pdfium_lock: RLock) -> int:
     """
     Retrieves the rotation value for a specific page.
 
@@ -232,7 +232,7 @@ def get_page_rotation(page, pdfium_lock: RLock) -> int:
         )
 
 
-def adjust_char_box(
+def _adjust_char_box(
     char_box: Tuple[float, float, float, float],
     rotation: int,
     internal_height: float,
@@ -263,3 +263,15 @@ def adjust_char_box(
             internal_height - left,
         )
     return left, right, top, bottom
+
+
+def lerp(start: float, end: float, t: float) -> float:
+    """
+    Performs linear interpolation between two numbers.
+
+    :param start: The starting value.
+    :param end: The ending value.
+    :param t: The interpolation factor (0 to 1).
+    :return: The interpolated value.
+    """
+    return start * (1 - t) + end * t