path: root/python/fatcat_tools/importers/wayback_static.py

#!/usr/bin/env python3

"""
Helpers to create Web Capture entities from extracted wayback content.

Works as a stand-alone script (for debugging) or as library routines.
"""

import argparse
import datetime
import hashlib
import json
import subprocess
import sys
from typing import Any, Dict, List, Optional, Tuple

import requests
from bs4 import BeautifulSoup
from fatcat_openapi_client import (
    ApiClient,
    Editgroup,
    EntityEdit,
    WebcaptureCdxLine,
    WebcaptureEntity,
    WebcaptureUrl,
)

from .common import b32_hex

CDX_API_BASE = "https://web.archive.org/cdx/search/cdx"
GWB_URL_BASE = "https://web.archive.org/web"
REQ_SESSION = requests.Session()


def parse_wbm_url(url: str) -> Tuple[str, datetime.datetime, str]:
    """Takes a wayback machine URL, and returns a tuple:

    (timestamp, datetime, original_url)
    """
    chunks = url.split("/")
    assert len(chunks) >= 6
    assert chunks[2] == "web.archive.org"
    assert chunks[3] == "web"
    return (chunks[4], parse_wbm_timestamp(chunks[4]), "/".join(chunks[5:]))


def test_parse_wbm_url() -> None:
    u = "http://web.archive.org/web/20010712114837/http://www.dlib.org/dlib/june01/reich/06reich.html"
    assert parse_wbm_url(u) == (
        "20010712114837",
        datetime.datetime(2001, 7, 12, 11, 48, 37),
        "http://www.dlib.org/dlib/june01/reich/06reich.html",
    )


def parse_wbm_timestamp(timestamp: str) -> datetime.datetime:
    """
    Takes a complete WBM timestamp string (like "20020327115625") and returns a
    python datetime object (UTC)
    """
    # strip any "im_" or "id_" suffix
    if timestamp.endswith("_"):
        timestamp = timestamp[:-3]
    # inflexible; require the full second-precision timestamp
    assert len(timestamp) == 14
    return datetime.datetime(
        year=int(timestamp[0:4]),
        month=int(timestamp[4:6]),
        day=int(timestamp[6:8]),
        hour=int(timestamp[8:10]),
        minute=int(timestamp[10:12]),
        second=int(timestamp[12:14]),
    )


def test_parse_wbm_timestamp() -> None:
    assert parse_wbm_timestamp("20010712114837") == datetime.datetime(2001, 7, 12, 11, 48, 37)


def fetch_wbm(url: str) -> bytes:
    resp = REQ_SESSION.get(url)
    resp.raise_for_status()
    assert resp.content
    return resp.content


def lookup_cdx(
    embed_url: str, verify_hashes: bool = True, cdx_output: Any = None
) -> Optional[WebcaptureCdxLine]:
    sys.stderr.write(embed_url + "\n")
    assert embed_url.startswith("/web/")
    embed_url = embed_url.split("/")
    timestamp = embed_url[2]
    if timestamp.endswith("_"):
        timestamp = timestamp[:-3]
    url = "/".join(embed_url[3:])
    # print((timestamp, url))
    resp = REQ_SESSION.get(
        CDX_API_BASE,
        params=dict(
            url=url,
            closest=timestamp,
            sort="closest",
            resolveRevisits="true",
            matchType="exact",
            limit=1,
        ),
    )
    resp.raise_for_status()
    # print(resp.url)
    if resp.content:
        hit = resp.content.decode("utf-8").split("\n")[0]
        if cdx_output:
            cdx_output.write(hit + "\n")
        cdx = hit.split(" ")
        cdx = [x if (x and x != "-") else None for x in cdx]
        webcapture_cdx = WebcaptureCdxLine(
            surt=cdx[0],
            timestamp=parse_wbm_timestamp(cdx[1]).isoformat() + "Z",
            url=cdx[2],
            mimetype=cdx[3],
            status_code=(cdx[4] and int(cdx[4])) or None,
            sha1=b32_hex(cdx[5]),
            sha256=None,
        )
        if verify_hashes:
            resp = REQ_SESSION.get(
                GWB_URL_BASE + "/{}id_/{}".format(cdx[1], webcapture_cdx.url)  # raw timestamp
            )
            resp.raise_for_status()
            assert webcapture_cdx.sha1 == hashlib.sha1(resp.content).digest().hex()
            webcapture_cdx.sha256 = hashlib.sha256(resp.content).digest().hex()
            webcapture_cdx.size = len(resp.content)
        return webcapture_cdx
    else:
        return None


def wayback_url_to_relative(url: str) -> Optional[str]:
    """
    Wayback URLs can be relative or absolute in rewritten documents. This
    function converts any form of rewritten URL to a relative (to
    web.archive.org) one, or returns None if it isn't a rewritten URL at all.
    """
    if url.startswith("https://web.archive.org/"):
        url = url[23:]
    elif url.startswith("http://web.archive.org/"):
        url = url[22:]

    if url.startswith("/web/"):
        return url
    else:
        return None


def extract_embeds(soup: BeautifulSoup) -> List[str]:

    embeds = set()

    # <link href="">
    for tag in soup.find_all("link", href=True):
        if tag["rel"] not in ("stylesheet",):
            continue
        url = wayback_url_to_relative(tag["href"])
        if url:
            embeds.add(url)
    # <img src="">
    for tag in soup.find_all("img", src=True):
        url = wayback_url_to_relative(tag["src"])
        if url:
            embeds.add(url)

    # <script src="">
    for tag in soup.find_all("script", src=True):
        url = wayback_url_to_relative(tag["src"])
        if url:
            embeds.add(url)

    return list(embeds)


def static_wayback_webcapture(wayback_url: str, cdx_output: Any = None) -> WebcaptureEntity:
    """
    Given a complete wayback machine capture URL, like:

        http://web.archive.org/web/20010712114837/http://www.dlib.org/dlib/june01/reich/06reich.html

    Will return a new ("bare") fatcat webcapture entity python object, with all
    the CDX entries filled in.
    """

    wbm_html = fetch_wbm(wayback_url)
    raw_timestamp, timestamp, original_url = parse_wbm_url(wayback_url)
    # with open(rewritten_path, 'r') as fp:
    #    soup = BeautifulSoup(fp, "lxml")
    soup = BeautifulSoup(wbm_html, "lxml")
    embeds = extract_embeds(soup)
    cdx_obj = lookup_cdx(
        "/web/{}/{}".format(raw_timestamp, original_url), cdx_output=cdx_output
    )
    cdx_list = [cdx_obj]
    for url in embeds:
        cdx_obj = lookup_cdx(url, cdx_output=cdx_output)
        cdx_list.append(cdx_obj)
    archive_urls = [
        WebcaptureUrl(
            rel="wayback",
            url="https://web.archive.org/web/",
        )
    ]
    wc = WebcaptureEntity(
        cdx=cdx_list,
        timestamp=timestamp.isoformat() + "Z",
        original_url=original_url,
        archive_urls=archive_urls,
        release_ids=None,
    )
    return wc


def auto_wayback_static(
    api: ApiClient, release_id: str, wayback_url: str, editgroup_id: Optional[str] = None
) -> Tuple[Optional[str], Optional[EntityEdit]]:
    """
    Returns a tuple: (editgroup_id, edit). If failed, both are None
    """

    raw_timestamp, timestamp, original_url = parse_wbm_url(wayback_url)
    git_rev = subprocess.check_output(["git", "describe", "--always"]).strip().decode("utf-8")

    release = api.get_release(release_id, expand="webcaptures")

    # check for existing webcapture with same parameters
    for wc in release.webcaptures:
        if wc.original_url == original_url and wc.timestamp.date() == timestamp.date():
            # skipping: already existed
            print(
                "release {} already had webcapture {} {}".format(
                    release_id, raw_timestamp, original_url
                )
            )
            return (None, None)

    wc = static_wayback_webcapture(wayback_url)
    assert len(wc.cdx) >= 1
    wc.release_ids = [release_id]
    if not editgroup_id:
        eg = api.create_editgroup(
            Editgroup(
                description="One-off import of static web content from wayback machine",
                extra=dict(git_rev=git_rev, agent="fatcat_tools.auto_wayback_static"),
            )
        )
        editgroup_id = eg.editgroup_id
    edit = api.create_webcapture(eg.editgroup_id, wc)
    return (editgroup_id, edit)


def main() -> None:
    parser = argparse.ArgumentParser()
    parser.add_argument("--verbose", action="store_true", help="verbose output")
    parser.add_argument("wayback_url", type=str, help="URL of wayback capture to extract from")
    parser.add_argument(
        "--json-output",
        type=argparse.FileType("w"),
        default=sys.stdout,
        help="where to write out webcapture entity (as JSON)",
    )
    parser.add_argument(
        "--cdx-output",
        type=argparse.FileType("w"),
        default=None,
        help="(optional) file to write out CDX stub",
    )

    args = parser.parse_args()

    # entity-to-JSON code; duplicate of entity_to_dict()
    api_client = ApiClient()
    wc = static_wayback_webcapture(args.wayback_url, cdx_output=args.cdx_output)
    wc_dict = api_client.sanitize_for_serialization(wc)
    print(json.dumps(wc_dict))


if __name__ == "__main__":
    main()