Installation and Quick Start

Installation and Quick Start

Installing, scraping a first comic, the common flags, and priming comics for another machine to finish.

Requirements

Python 3.9+ and Chrome or Firefox (Selenium 4.6+ fetches the driver itself), or use the Docker container.

pip install -r requirements.txt

Quick start

python mirror_base.py --output "Uncompressed/MyComic" "https://example.com/comic/first-page/"

It finds the image and the next link, walks forward until there is no next page, saves each image, and writes mirror_metadata.json in the folder plus a .cbz: under a sibling CBZs/ tree when one exists beside Uncompressed/, otherwise beside the folder (details in Library Layout). Run the same command later and it resumes from where it stopped, adding only new pages. For a whole library, see Running Unattended.

Flags

Flag	Effect
-o PATH, --cbz-path PATH, --no-cbz	Where pages go; where the .cbz goes; no archive at all.
-p, --prefix; -i N	Number saved files (see below), counting from N (default 1).
--prime; --keep-index / --no-keep-index	Below.
-ej; -f; -w N	JavaScript on (off by default, for speed); Firefox instead of Chrome; wait N seconds per page.
--no-headless; -v	Show the browser window; log every step.
--check	Report which element paths match a page, and stop.
--no-direction-check, -m, -n; --no-multi-page	See Notes and Limits; Several pages on one address.

How pages are named

A page keeps the name the site gives it: the filename the server sends, else the last part of the image address without its query string. Only a name with no picture extension (/comic-image/1650564/?token=...) gets one, from the server's content type or the file's first bytes. Comics scraped when .png was added to every name keep those names; the page a run re-saves at its start switches to the new spelling rather than being held twice.

When to number pages

Use --prefix when the filename scheme changes mid-run (a date stamp becomes NAME_0421.jpg), since a prefix pins download order, which is publication order; and when a site numbers each chapter from one (/comics/1/1.png, /comics/131/1.png), where chapters would overwrite each other: a scrape stops with code 9 first. Such a site reuses a name at each chapter boundary, so the backwards check waits for a second backwards step.

The page index a scrape keeps

A scrape from page one into an empty folder records which page is which, with real names and sizes, so the comic never needs a walk. To start a comic over, delete its folder and archive and add it again; its old index is renamed aside in config/index with .replaced-<date>.

Priming comics to download elsewhere

Scraping over a network share is slow. --prime saves only the first page, follows the next link once to prove it works, writes metadata and archive, and stops:

python mirror_base.py --prime -o "D:/Comics/Uncompressed/MyComic" "https://example.com/comic/first-page/"

The metadata resumes on page two, so the next update_comics.py run on the machine holding the library fetches the rest (drop an update-now.txt to start it). No next button means the comic has one page or needs its next element added (Element Paths).