A couple of refactors

To resolve circular dependencies and also a couple of bug fixes

kraken/lib/arrow_dataset.py

@@ -29,10 +29,9 @@
 from typing import Optional, List, Union, Callable, Tuple, Dict
 from multiprocessing import Pool
 from kraken.containers import Segmentation
-from kraken.lib import functional_im_transforms as F_t
 from kraken.lib.segmentation import extract_polygons
 from kraken.lib.xml import XMLPage
-from kraken.lib.util import is_bitonal, make_printable
+from kraken.lib import util
 from kraken.lib.exceptions import KrakenInputException
 from os import PathLike
 
@@ -47,7 +46,7 @@ def _extract_line(xml_record, skip_empty_lines: bool = True):
         im = Image.open(xml_record.imagename)
     except (FileNotFoundError, UnidentifiedImageError):
         return lines, None, None
-    if is_bitonal(im):
+    if util.is_bitonal(im):
         im = im.convert('1')
     recs = xml_record.lines.values()
     for idx, rec in enumerate(recs):
@@ -81,7 +80,7 @@ def _extract_path_line(xml_record, skip_empty_lines: bool = True):
         return [], None, None
     if not xml_record['lines'][0]['text'] and skip_empty_lines:
         return [], None, None
-    if is_bitonal(im):
+    if util.is_bitonal(im):
         im = im.convert('1')
     fp = io.BytesIO()
     im.save(fp, format='png')
@@ -91,9 +90,9 @@ def _extract_path_line(xml_record, skip_empty_lines: bool = True):
 
 def parse_path(path: Union[str, PathLike],
                suffix: str = '.gt.txt',
-               split=F_t.default_split,
+               split=util.default_split,
                skip_empty_lines: bool = True):
-    with open(F_t.suffix_split(path, split=split, suffix=suffix), 'r', encoding='utf-8') as fp:
+    with open(util.suffix_split(path, split=split, suffix=suffix), 'r', encoding='utf-8') as fp:
         gt = fp.read().strip('\n\r')
         if not gt and skip_empty_lines:
             raise KrakenInputException(f'No text for ground truth line {path}.')
@@ -197,7 +196,7 @@ def build_binary_dataset(files: Optional[List[Union[str, PathLike, Dict]]] = Non
     callback(0, num_lines)
 
     for k, v in sorted(alphabet.items(), key=lambda x: x[1], reverse=True):
-        char = make_printable(k)
+        char = util.make_printable(k)
         if char == k:
             char = '\t' + char
         logger.info(f'{char}\t{v}')

kraken/lib/dataset/recognition.py

@@ -32,13 +32,11 @@
 from typing import List, Tuple, Callable, Optional, Any, Union, Literal
 
 from kraken.containers import BaselineLine, BBoxLine, Segmentation
-from kraken.lib.util import is_bitonal
+from kraken.lib import util
 from kraken.lib.codec import PytorchCodec
 from kraken.lib.segmentation import extract_polygons
 from kraken.lib.exceptions import KrakenInputException, KrakenEncodeException
 
-from kraken.lib import functional_im_transforms as F_t
-
 __all__ = ['DefaultAugmenter',
            'ArrowIPCRecognitionDataset',
            'PolygonGTDataset',
@@ -124,14 +122,14 @@ def __init__(self,
         self.seg_type = None
         # built text transformations
         if normalization:
-            self.text_transforms.append(partial(F_t.text_normalize, normalization=normalization))
+            self.text_transforms.append(partial(util.text_normalize, normalization=normalization))
         if whitespace_normalization:
-            self.text_transforms.append(F_t.text_whitespace_normalize)
+            self.text_transforms.append(util.text_whitespace_normalize)
         if reorder:
             if reorder in ('L', 'R'):
-                self.text_transforms.append(partial(F_t.text_reorder, base_dir=reorder))
+                self.text_transforms.append(partial(util.text_reorder, base_dir=reorder))
             else:
-                self.text_transforms.append(F_t.text_reorder)
+                self.text_transforms.append(util.text_reorder)
         if augmentation:
             self.aug = DefaultAugmenter()
 
@@ -310,14 +308,14 @@ def __init__(self,
         self.seg_type = 'baselines'
         # built text transformations
         if normalization:
-            self.text_transforms.append(partial(F_t.text_normalize, normalization=normalization))
+            self.text_transforms.append(partial(util.text_normalize, normalization=normalization))
         if whitespace_normalization:
-            self.text_transforms.append(F_t.text_whitespace_normalize)
+            self.text_transforms.append(util.text_whitespace_normalize)
         if reorder:
             if reorder in ('L', 'R'):
-                self.text_transforms.append(partial(F_t.text_reorder, base_dir=reorder))
+                self.text_transforms.append(partial(util.text_reorder, base_dir=reorder))
             else:
-                self.text_transforms.append(F_t.text_reorder)
+                self.text_transforms.append(util.text_reorder)
         if augmentation:
             self.aug = DefaultAugmenter()
 
@@ -430,7 +428,7 @@ def __getitem__(self, index: int) -> Tuple[torch.Tensor, torch.Tensor]:
                 im_mode = 'RGB'
             elif im.shape[0] == 1:
                 im_mode = 'L'
-            if is_bitonal(im):
+            if util.is_bitonal(im):
                 im_mode = '1'
 
             if im_mode > self.im_mode:
@@ -499,14 +497,14 @@ def __init__(self,
         self.seg_type = 'bbox'
         # built text transformations
         if normalization:
-            self.text_transforms.append(partial(F_t.text_normalize, normalization=normalization))
+            self.text_transforms.append(partial(util.text_normalize, normalization=normalization))
         if whitespace_normalization:
-            self.text_transforms.append(F_t.text_whitespace_normalize)
+            self.text_transforms.append(util.text_whitespace_normalize)
         if reorder:
             if reorder in ('L', 'R'):
-                self.text_transforms.append(partial(F_t.text_reorder, base_dir=reorder))
+                self.text_transforms.append(partial(util.text_reorder, base_dir=reorder))
             else:
-                self.text_transforms.append(F_t.text_reorder)
+                self.text_transforms.append(util.text_reorder)
         if augmentation:
             self.aug = DefaultAugmenter()
 
@@ -608,7 +606,7 @@ def __getitem__(self, index: int) -> Tuple[torch.Tensor, torch.Tensor]:
                 im_mode = 'RGB'
             elif im.shape[0] == 1:
                 im_mode = 'L'
-            if is_bitonal(im):
+            if util.is_bitonal(im):
                 im_mode = '1'
             if im_mode > self.im_mode:
                 logger.info(f'Upgrading "im_mode" from {self.im_mode} to {im_mode}')

kraken/lib/functional_im_transforms.py

@@ -17,16 +17,11 @@
 facilitate pickling.
 """
 import torch
-import regex
-import unicodedata
-import bidi.algorithm as bd
 
-from os import PathLike
-from pathlib import Path
 from PIL import Image
 from PIL.Image import Resampling
 
-from typing import Tuple, Optional, Callable, Any, Union
+from typing import Tuple, Any
 
 from kraken.binarization import nlbin
 from kraken.lib.lineest import dewarp, CenterNormalizer
@@ -77,26 +72,3 @@ def _fixed_resize(img: Image.Image, size: Tuple[int, int], interpolation: int =
         ow = int(w * oh/h)
     img = img.resize((ow, oh), interpolation)
     return img
-
-
-def text_normalize(text: str, normalization: str) -> str:
-    return unicodedata.normalize(normalization, text)
-
-
-def text_whitespace_normalize(text: str) -> str:
-    return regex.sub(r'\s', ' ', text).strip()
-
-
-def text_reorder(text: str, base_dir: Optional[str] = None) -> str:
-    return bd.get_display(text, base_dir=base_dir)
-
-
-def default_split(x: Union[PathLike, str]) -> str:
-    x = Path(x)
-    while x.suffixes:
-        x = x.with_suffix('')
-    return str(x)
-
-
-def suffix_split(x: Union[PathLike, str], split: Callable[[Union[PathLike, str]], str], suffix: str) -> str:
-    return split(x) + suffix

kraken/lib/segmentation.py

@@ -40,14 +40,14 @@
 from skimage.morphology import skeletonize
 from skimage.transform import PiecewiseAffineTransform, SimilarityTransform, AffineTransform, warp
 
-from typing import List, Tuple, Union, Dict, Sequence, Optional, Literal
-
-from kraken.containers import Segmentation
+from typing import List, Tuple, Union, Dict, Sequence, Optional, Literal, TYPE_CHECKING
 
 from kraken.lib import default_specs
 from kraken.lib.vgsl import TorchVGSLModel
 from kraken.lib.exceptions import KrakenInputException
 
+if TYPE_CHECKING:
+    from kraken.containers import Segmentation
 
 logger = logging.getLogger('kraken')
 
@@ -1032,7 +1032,7 @@ def compute_polygon_section(baseline: Sequence[Tuple[int, int]],
     return tuple(o)
 
 
-def extract_polygons(im: Image.Image, bounds: Segmentation) -> Image.Image:
+def extract_polygons(im: Image.Image, bounds: "Segmentation") -> Image.Image:
     """
     Yields the subimages of image im defined in the list of bounding polygons
     with baselines preserving order.

kraken/lib/util.py

@@ -6,15 +6,18 @@
 import torch
 import unicodedata
 import numpy as np
+import bidi.algorithm as bd
+import regex
 
+from os import PathLike
+from pathlib import Path
 from PIL import Image
 from os import PathLike
 
 from typing import Union, Callable, Optional, Literal
 
-from kraken.lib import functional_im_transforms as F_t
 from kraken.containers import BBoxLine
-from kraken.exceptions import KrakenInputException
+from kraken.lib.exceptions import KrakenInputException
 
 __all__ = ['pil2array', 'array2pil', 'is_bitonal', 'make_printable', 'get_im_str', 'parse_gt_path']
 
@@ -62,6 +65,29 @@ def get_im_str(im: Image.Image) -> str:
     return im.filename if hasattr(im, 'filename') else str(im)
 
 
+def text_normalize(text: str, normalization: str) -> str:
+    return unicodedata.normalize(normalization, text)
+
+
+def text_whitespace_normalize(text: str) -> str:
+    return regex.sub(r'\s', ' ', text).strip()
+
+
+def text_reorder(text: str, base_dir: Optional[str] = None) -> str:
+    return bd.get_display(text, base_dir=base_dir)
+
+
+def default_split(x: Union[PathLike, str]) -> str:
+    x = Path(x)
+    while x.suffixes:
+        x = x.with_suffix('')
+    return str(x)
+
+
+def suffix_split(x: Union[PathLike, str], split: Callable[[Union[PathLike, str]], str], suffix: str) -> str:
+    return split(x) + suffix
+
+
 def is_printable(char: str) -> bool:
     """
     Determines if a chode point is printable/visible when printed.
@@ -103,7 +129,7 @@ def make_printable(char: str) -> str:
 
 def parse_gt_path(path: Union[str, PathLike],
                   suffix: str = '.gt.txt',
-                  split: Callable[[Union[PathLike, str]], str] = F_t.default_split,
+                  split: Callable[[Union[PathLike, str]], str] = default_split,
                   skip_empty_lines: bool = True,
                   base_dir: Optional[Literal['L', 'R']] = None,
                   text_direction: Literal['horizontal-lr', 'horizontal-rl', 'vertical-lr', 'vertical-rl'] = 'horizontal-lr') -> BBoxLine:
@@ -128,7 +154,7 @@ def parse_gt_path(path: Union[str, PathLike],
 
     gt = ''
     try:
-        with open(F_t.suffix_split(path, split=split, suffix=suffix), 'r', encoding='utf-8') as fp:
+        with open(suffix_split(path, split=split, suffix=suffix), 'r', encoding='utf-8') as fp:
             gt = fp.read().strip('\n\r')
     except FileNotFoundError:
         if not skip_empty_lines:

kraken/templates/alto

@@ -78,9 +78,9 @@
         <OtherTag DESCRIPTION="region type" ID="TYPE_{{ loop.index }}" TYPE={{ type }} LABEL="{{ label }}"/>
     {% endfor %}
     </Tags>
-    {% if len(page.line_orders) > 0 %}
+    {% if page.line_orders | length > 0 %}
     <ReadingOrder>
-        {% if len(page.line_orders) == 1 %}
+        {% if page.line_orders | length == 1 %}
         <OrderedGroup ID="ro_0">
            {% for id in page.line_orders[0] %}
            <ElementRef ID="o_{{ loop.index }}" REF="{{ id }}"/>