class TitleAndAuthorInfo(JsonSerializable):
pdf = traits.Unicode()
pdf_sha1 = traits.Unicode()
image_path = traits.Unicode()
title_bounding_box = traits.Instance(BoxClass)
title_text = traits.Unicode()
authors = traits.List(traits.Instance(AuthorInfo))
class Figure(JsonSerializable):
figure_boundary = traits.Instance(BoxClass)
caption_boundary = traits.Instance(BoxClass)
caption_text = traits.Unicode()
name = traits.Unicode()
page = traits.Int()
figure_type = traits.Unicode()
dpi = traits.Int()
page_width = traits.Int()
page_height = traits.Int()
# URI to cropped image of the figure
uri = traits.Unicode(
default_value=None, allow_none=True)
def page_size(self) -> Tuple[int, int]:
return self.page_height, self.page_width
@staticmethod
def from_pf_ann(ann: dict, target_page_size: Tuple[int, int]) -> 'Figure':
"""Convert an annotation in the pdffigures format"""
cur_page_size = ann['page_height'], ann['page_width']
if cur_page_size[0] is None:
cur_page_size = [
d * DEFAULT_INFERENCE_DPI / ann['dpi'] for d in target_page_size
]
return Figure(
figure_boundary=BoxClass.from_tuple(ann['region_bb'])
.resize_by_page(cur_page_size, target_page_size),
caption_boundary=BoxClass.from_tuple(ann['caption_bb'])
.resize_by_page(cur_page_size, target_page_size),
caption_text=ann['caption'],
name=ann['name'],
page=ann['page'],
figure_type=ann['figure_type'],
page_width=target_page_size[
1
],
page_height=target_page_size[
0
]
)
@staticmethod
def from_pf_output(res: dict, target_dpi=DEFAULT_INFERENCE_DPI) -> 'Figure':
"""Convert a pdffigures output figure to a Figure object"""
scale_factor = target_dpi / 72
return Figure(
figure_boundary=BoxClass.from_dict(res['regionBoundary']
).rescale(scale_factor),
caption_boundary=BoxClass.from_dict(res['captionBoundary'])
.rescale(scale_factor),
caption_text=res['caption'],
name=res['name'],
page=res['page'],
figure_type=res['figType']
)
class PdfDetectionResult(JsonSerializable):
pdf = traits.Unicode()
figures = traits.List(traits.Instance(Figure))
dpi = traits.Int()
raw_detected_boxes = traits.List(
traits.List(traits.Instance(BoxClass)), allow_none=True
) # type: Optional[List[List[BoxClass]]]
raw_pdffigures_output = traits.Dict(
traits.Any(), allow_none=True
) # type: Optional[dict]
error = traits.Unicode(
default_value=None, allow_none=True
) # type: Optional[str]
class CaptionOnly(JsonSerializable):
caption_boundary = traits.Instance(BoxClass)
caption_text = traits.Unicode()
name = traits.Unicode()
page = traits.Int()
figure_type = traits.Unicode()
dpi = traits.Int()
class PubmedMatchedFigure(config.JsonSerializable):
"""
Contains data on a figure extracted from a PMC paper via caption matching with the included nxml file.
"""
fig_im = traits.Instance(np.ndarray)
page_image_name = traits.Unicode()
caption = traits.Unicode()
name = traits.Unicode()
matched_caption = traits.Unicode()
html_page = traits.Unicode()
start_pos = traits.Int()
end_pos = traits.Int()
pdf = traits.Unicode()
page_num = traits.Int()
class AuthorInfo(JsonSerializable):
bounding_box = traits.Instance(BoxClass)
name = traits.Unicode()