Bug: incomplete link extraction

[kevinhu]

I'm using pdftext to extract links from a PDF, but I'm encountering issues where the anchor text is cut off.

For example, I've been using this PDF from the latest Anthropic funding attachment, which I've linked below. After running dictionary_output, links such as "Claude 3.7 Sonnet" appear to only have a URL attached to the last word in the link rather than the entire span.

If you open the PDF, you can confirm that the entire text is linked rather than only the last word.

{
    "bbox": [
        304.00433349609375,
        733.3996200561523,
        343.6524658203125,
        745.5496215820312
    ],
    "text": "Claude",
    "rotation": 0.0,
    "font": {
        "name": "DAAAAA+TiemposText-Regular",
        "flags": 524292,
        "size": 18.0,
        "weight": 724
    },
    "char_start_idx": 750,
    "char_end_idx": 755,
    "url": ""
},
{
    "bbox": [
        346.7061767578125,
        742.655029296875,
        346.7061767578125,
        742.655029296875
    ],
    "text": " ",
    "rotation": 0.0,
    "font": {
        "name": "",
        "flags": 0,
        "size": 1.0,
        "weight": -1
    },
    "char_start_idx": 756,
    "char_end_idx": 756,
    "url": ""
},
{
    "bbox": [
        346.251220703125,
        733.3996200561523,
        361.2065124511719,
        745.5496215820312
    ],
    "text": "3.7",
    "rotation": 0.0,
    "font": {
        "name": "DAAAAA+TiemposText-Regular",
        "flags": 524292,
        "size": 18.0,
        "weight": 724
    },
    "char_start_idx": 757,
    "char_end_idx": 759,
    "url": ""
},
{
    "bbox": [
        364.1081237792969,
        742.655029296875,
        364.1081237792969,
        742.655029296875
    ],
    "text": " ",
    "rotation": 0.0,
    "font": {
        "name": "",
        "flags": 0,
        "size": 1.0,
        "weight": -1
    },
    "char_start_idx": 760,
    "char_end_idx": 760,
    "url": ""
},
{
    "bbox": [
        363.8052978515625,
        733.3996200561523,
        403.8422546386719,
        745.5496215820312
    ],
    "text": "Sonnet",
    "rotation": 0.0,
    "font": {
        "name": "DAAAAA+TiemposText-Regular",
        "flags": 524292,
        "size": 18.0,
        "weight": 724
    },
    "char_start_idx": 761,
    "char_end_idx": 766,
    "url": "https://www.anthropic.com/news/claude-3-7-sonnet"
}

anthropic.pdf