Return page content as a Python dict of images and text spans.
(self, cb=None, sort=False)
| 16469 | return lines |
| 16470 | |
| 16471 | def extractDICT(self, cb=None, sort=False) -> dict: |
| 16472 | """Return page content as a Python dict of images and text spans.""" |
| 16473 | val = self._textpage_dict(raw=False) |
| 16474 | if cb is not None: |
| 16475 | val["width"] = cb.width |
| 16476 | val["height"] = cb.height |
| 16477 | if sort: |
| 16478 | blocks = val["blocks"] |
| 16479 | blocks.sort(key=lambda b: (b["bbox"][3], b["bbox"][0])) |
| 16480 | val["blocks"] = blocks |
| 16481 | return val |
| 16482 | |
| 16483 | def extractHTML(self) -> str: |
| 16484 | """Return page content as a HTML string.""" |