theread.me/scripts/organize-photo-highlights.py

"""One-time migration of September 2026 uploads into individual highlighted photos."""
from datetime import datetime, timedelta, timezone
from pathlib import Path
import html
import json
import re


ART = Path("_art")
UPLOAD = "/img/arts/uploads"
OLD = [
    "260cc99527c13f256d4ea39d0959f5daa942c888ea882852e73fd94f6c30b7db",
    "474ab3283ce8801ac6e59b67b65b4169167053873a5a8db4a15b840a3448266c",
    "7c832afa1b73023d8f2391ff40c221be9ffd468a1a96a6bf4900cbf95595d31d",
    "b18a2f1c7127077c801160782012cb164759f0b9dd41762205b45e4c9a07d673",
    "c53f20236f166632bff163339c8e789cc09b0d7e63c5b531b1fc20705460975b",
]

armenia = [
    (OLD[2], 9), (OLD[2], 8), (OLD[2], 7), (OLD[2], 6),
    (OLD[2], 5), (OLD[2], 4), (OLD[1], 0),
]
romania = [
    (OLD[2], 0), (OLD[2], 1), (OLD[4], 0), (OLD[4], 1),
    (OLD[4], 2), (OLD[4], 3), (OLD[4], 4), (OLD[0], 0), (OLD[3], 0),
]
ungrouped = [(OLD[2], 2), (OLD[2], 3)]
ordered = armenia + romania + ungrouped
assert len(ordered) == len(set(ordered)) == 18

collection_groups = [
    (
        "iran-analog-2024",
        datetime(2024, 10, 25, 23, 59, 59, tzinfo=timezone.utc),
        "Iran - Hamadan, Kermanshah, Tabriz",
        [
            "Taq-e-Bostan_000521270020.jpg", "Taq-e-Bostan_000521270019.jpg",
            "Taq-e-Bostan_000521270018.jpg", "Tabriz_000521270013.jpg",
            "Tabriz_000521270012.jpg", "Iran_000521270023.jpg",
            "Iran_000521270021.jpg", "Hamadan_000521270035.jpg",
            "Hamadan_000521270034.jpg", "Hamadan_000521270033.jpg",
            "Hamadan_000521270032.jpg", "Hamadan_000521270031.jpg",
            "Ganjname,-Hamadan_000521270029.jpg", "Cowboy-from-Hamadan_000521270030.jpg",
            "Bistoon_000521270027.jpg", "Bistoon_000521270026.jpg",
            "Bistoon_000521270025.jpg", "Bazaar-e-Tabriz_000521270010.jpg",
            "Bazaar-e-Tabriz_000521270009.jpg", "Bazaar-e-Tabriz_000521270008.jpg",
            ("dresden-prague-iran-2024", "Babataher.jpg"),
        ],
    ),
    (
        "kazakhstan-2024",
        datetime(2024, 10, 9, 23, 59, 59, tzinfo=timezone.utc),
        "Kazakhstan - World Nomad Games",
        [
            "x_DSC-0999.jpg", "x_DSC-0994.jpg",
            "World-Nomad-Games,-Kazakhstan_DSC-0685.jpg",
            "World-Nomad-Games,-Kazakhstan_DSC-0536.jpg",
            "World-Nomad-Games,-Kazakhstan_DSC-0472.jpg",
            "World-Nomad-Games,-Kazakhstan_DSC-0298.jpg",
            "World-Nomad-Games,-Kazakhstan_DSC-0251.jpg",
            "Kazakhstan_DSC-0870.jpg", "Kazakhstan_DSC-0865.jpg",
        ],
    ),
    (
        "dresden-prague-iran-2024",
        datetime(2024, 10, 9, 23, 50, 59, tzinfo=timezone.utc),
        "Dresden & Berlin",
        [
            "x_000512370028.jpg", "x_000512370027.jpg", "x_000512370026.jpg",
            "x_000512370024.jpg", "x_000512370022.jpg", "x_000512370013.jpg",
            "x_000512350002.jpg", "x_000512350001.jpg",
            "Let-me-tell-you-a-story_000512370015.jpg", "Dresden_000512370034.jpg",
            "Dresden_000512370033.jpg", "Dresden_000512370032.jpg",
            "Dresden_000512370019.jpg", "Dresden_000512370018.jpg",
            "Dresden_000512370017.jpg", "Dresden_000512370016.jpg",
        ],
    ),
    (
        "dresden-prague-iran-2024",
        datetime(2024, 10, 9, 23, 40, 59, tzinfo=timezone.utc),
        "Iran - Tehran & Qom",
        [
            "Home_000512360003.jpg", "Somewhere-in-Tehran_2.jpg",
            "Somewhere-in-Tehran.jpg", "Grandmother's-old-room.jpg",
        ],
    ),
    (
        "dresden-prague-iran-2024",
        datetime(2024, 10, 9, 23, 30, 59, tzinfo=timezone.utc),
        "Prague",
        [
            "x_000512370006.jpg", "x_000512370005.jpg", "x_000512370003.jpg",
            "Prague_000512370012.jpg", "Prague_000512360031.jpg",
            "Prague_000512360030.jpg", "Prague_000512360029.jpg",
            "Prague_000512360027.jpg", "Prague_000512360021.jpg",
            "Prague_000512360020.jpg", "Prague_000512360019.jpg",
        ],
    ),
    (
        "dresden-prague-iran-2024",
        datetime(2024, 10, 9, 23, 20, 59, tzinfo=timezone.utc),
        "Belfast & Tallaght",
        [
            "x_000512350014.jpg", "x_000512350012.jpg", "x_000512350011.jpg",
            "x_000512350006.jpg", "x_000512350004.jpg", "x_000512350003.jpg",
            "Bobby-Sands.jpg", "Belfast_000512350032.jpg",
            "Belfast_000512350027.jpg", "Belfast_000512350026.jpg",
            "Belfast_000512350022.jpg", "Belfast_000512350018.jpg",
        ],
    ),
]

anchors = {armenia[0]: "Armenia", romania[0]: "Romania"}
captions = {(OLD[2], 7): "Transmission of Flesh by Erfan Ashourioun"}
first_for_job = {}
for photo in ordered:
    first_for_job.setdefault(photo[0], photo)

for job in OLD:
    batch = next(ART.glob(f"*-photo-{job}.md"), None)
    if batch:
        batch.unlink()

start = datetime(2026, 9, 19, 12, 29, 33, tzinfo=timezone.utc)
for position, (job, index) in enumerate(ordered):
    title = f"photo-{job}-{index:02}"
    if (job, index) == first_for_job[job]:
        title = f"photo-{job}"
    date = start - timedelta(seconds=position)
    lines = ["---", "layout: post", f'title: "{title}"',
             f'date: "{date.isoformat()}"', "categories: art"]
    if (job, index) in anchors:
        lines.append(f'anchor: "{anchors[(job, index)]}"')
    lines.extend(["---", ""])
    if (job, index) == first_for_job[job] and (job, index) not in anchors:
        lines.extend([f'<div id="photo-{job}"></div>', ""])
    caption = captions.get((job, index), "")
    lines.append(f'<p><img src="{UPLOAD}/{job}/{index:02}.jpg" '
                 f'alt="{html.escape(caption, quote=True)}" loading="lazy"></p>')
    if caption:
        lines.append(f'<span class="image-details">{html.escape(caption)}</span>')
    target = ART / f"2026-09-19-photo-{job}-{index:02}.md"
    target.write_text("\n".join(lines) + "\n")

for folder, group_start, anchor, filenames in collection_groups:
    for position, member in enumerate(filenames):
        member_folder, filename = member if isinstance(member, tuple) else (folder, member)
        image = f"/img/arts/{member_folder}/{filename}"
        matches = [path for path in ART.glob("*.md") if image in path.read_text()]
        assert len(matches) == 1, (image, matches)
        path = matches[0]
        content = path.read_text()
        date = group_start - timedelta(seconds=position)
        content = re.sub(r"^date:.*$", f'date: "{date.isoformat()}"', content,
                         count=1, flags=re.MULTILINE)
        content = re.sub(r"^anchor:.*\n", "", content, flags=re.MULTILINE)
        if position == 0:
            content = content.replace("categories: art\n", "categories: art\nanchor: " +
                                      json.dumps(anchor) + "\n", 1)
        path.write_text(content)

print("Organized all configured photo highlights into contiguous groups")