Datasets
Our mission is to archive all the books in the world (as well as papers, magazines, etc), and make them widely accessible. We believe that all books should be mirrored far and wide, to ensure redundancy and resiliency. This is why we’re pooling together files from a variety of sources. Some sources are completely open and can be mirrored in bulk (such as Sci-Hub). Others are closed and protective, so we try to scrape them in order to “liberate” their books. Yet others fall somewhere in between.
All our data can be torrented, and all our metadata can be generated or downloaded as ElasticSearch and MariaDB databases. The raw data can be manually explored through JSON files such as this.
Overview
Below is a quick overview of the sources of the files on Anna’s Archive.
| Source | Size | % mirrored by AA / torrents available Percentages of number of files |
Last updated |
|---|---|---|---|
|
Libgen.rs [lgrs]
Non-Fiction and Fiction
|
7,418,912 files 84.0 TB |
99.756% / 99.756% | 2024-11-04 |
|
Sci-Hub [scihub]
Via Libgen.li “scimag”
|
101,004,444 files 96.0 TB |
87.052% / 87.052% |
Sci-Hub: frozen since 2021; most available through torrents
Libgen.li: minor additions since then |
|
Libgen.li [lgli]
Excluding “scimag”
|
19,457,189 files 281.4 TB |
86.667% / 84.89%
Fiction torrents are behind (though IDs ~4-6M not torrented since they overlap with our Zlib torrents).
|
2024-09-01 |
| Z-Library [zlib] |
18,119,026 files 131.3 TB |
99.309% / 99.303% | 2024-11-05 |
| Z-Library Chinese [zlibzh] |
1,481,975 files 45.5 TB |
72.291% / 72.291%
The “Chinese” collection in Z-Library appears to be the same as our DuXiu collection, but with different MD5s. We exclude these files from torrents to avoid duplication, but still show them in our search index.
|
2024-11-05 |
| IA Controlled Digital Lending [ia] |
10,463,656 files 317.6 TB |
96.862% / 96.862%
98%+ of files are searchable.
|
2024-11-05 |
| DuXiu 读秀 [duxiu] |
3,721,900 files 193.7 TB |
100% / 100% | 2024-06-13 |
| Uploads to AA [upload] |
4,956,843 files 79.6 TB |
99.461% / 99.429% | 2024-05-27 |
| MagzDB [magzdb] |
649,485 files 17.1 TB |
94.449% / 94.118% | 2024-07-29 |
| Nexus/STC [nexusstc] |
4,778,884 files 76.0 TB |
97.047% / 97.047% | 2024-05-16 |
|
Total
Excluding duplicates
|
143,528,771 files 956.9 TB |
88.709% / 88.469% |
Since the shadow libraries often sync data from each other, there is considerable overlap between the libraries. That’s why the numbers don’t add up to the total.
The “mirrored and seeded by Anna’s Archive” percentage shows how many files we mirror ourselves. We seed those files in bulk through torrents, and make them available for direct download through partner websites.
Source libraries
Some source libraries promote the bulk sharing of their data through torrents, while others do not readily share their collection. In the latter case, Anna’s Archive tries to scrape their collections, and make them available (see our Torrents page). There are also in-between situations, for example, where source libraries are willing to share, but don’t have the resources to do so. In those cases, we also try to help out.
Below is an overview of how we interface with the different source libraries.
| Source | Metadata | Files |
|---|---|---|
| Libgen.rs [lgrs] |
✅ Daily HTTP database dumps
|
✅ Automated torrents for Non-Fiction and Fiction
👩💻 Anna’s Archive manages a collection of book cover torrents
|
| Sci-Hub / Libgen “scimag” [scihub] |
❌ Sci-Hub has frozen new files since 2021.
✅ Metadata dumps available here and here, as well as as part of the Libgen.li database (which we use)
|
|
| Libgen.li [lgli] |
✅ Quarterly HTTP database dumps
|
✅ Non-Fiction torrents are shared with Libgen.rs (and mirrored here).
🙃 Fiction collection has diverged but still has torrents, though not updated since 2022 (we do have direct downloads).
👩💻 Anna’s Archive and Libgen.li collaboratively manage collections of comic books and magazines.
❌ No torrents for Russian fiction and standard documents collections.
|
| Z-Library [zlib/zlibzh] |
👩💻 Anna’s Archive and Z-Library collaboratively manage a collection of Z-Library metadata and Z-Library files
|
|
| IA Controlled Digital Lending [ia] |
✅ Some metadata available through Open Library database dumps, but those don’t cover the entire IA collection
❌ No easily accessible metadata dumps available for their entire collection
👩💻 Anna’s Archive manages a collection of IA metadata
|
❌ Files only available for borrowing on a limited basis, with various access restrictions
👩💻 Anna’s Archive manages a collection of IA files
|
| DuXiu 读秀 [duxiu] |
✅ Various metadata databases scattered around the Chinese internet; though often paid databases
❌ No easily accessible metadata dumps available for their entire collection.
👩💻 Anna’s Archive manages a collection of DuXiu metadata
|
✅ Various file databases scattered around the Chinese internet; though often paid databases
❌ Most files only accessible using premium BaiduYun accounts; slow downloading speeds.
👩💻 Anna’s Archive manages a collection of DuXiu files
|
| Uploads to AA [uploads] |
Various smaller or one-off sources. We encourage people to upload to other shadow libraries first, but sometimes people have collections that are too big for others to sort through, though not big enough to warrant their own category.
|
|
| MagzDB [magzdb] |
❌ Appears defunct since July 2023.
❌ No easily accessible metadata dumps available for their entire collection.
👩💻 Anna’s Archive manages a collection of MagzDB metadata.
|
✅ Since MagzDB was a fork from Libgen.li magazines, a large part is covered by those torrents.
❌ No official torrents from MagzDB for their unique files.
👩💻 Anna’s Archive manages a collection of magzdb files as part of our upload collection (the ones with “magzdb” in the filename).
|
| Nexus/STC [nexusstc] |
✅ Summa database available through IPFS, though can be slow to download or directly interact with.
👩💻 Anna’s Archive manages a collection of Nexus/STC metadata, through this code.
|
✅ Data can be replicated through Iroh.
❌ No mirroring by Anna’s Archive or partner servers yet.
|
Metadata-only sources
We also enrich our collection with metadata-only sources, which we can match to files, e.g. using ISBN numbers or other fields. Below is an overview of those. Again, some of these sources are completely open, while for others we have to scrape them.
Our inspiration for collecting metadata is Aaron Swartz’ goal of “one web page for every book ever published”, for which he created Open Library. That project has done well, but our unique position allows us to get metadata that they can’t. Another inspiration was our desire to know how many books there are in the world, so we can calculate how many books we still have left to save.
Note that in metadata search, we show the original records. We don’t do any merging of records.
| Source | Metadata | Last updated |
|---|---|---|
| OpenLibrary [ol] |
✅ Monthly database dumps
|
2024-08-29 |
| OCLC (WorldCat) [oclc] |
❌ Not available directly in bulk, protected against scraping
👩💻 Anna’s Archive manages a collection of OCLC (WorldCat) metadata
|
2023-10-01 |
| Other metadata scrapes |
👩💻 Anna’s Archive manages scrapes of metadata from other sources.
|
Varies |
Unified database
We combine all the above sources into one unified database that we use to serve this website. This unified database is not available directly, but since Anna’s Archive is fully open source, it can be fairly easily generated or downloaded as ElasticSearch and MariaDB databases. The scripts on that page will automatically download all the requisite metadata from the sources mentioned above.
If you’d like to explore our data before running those scripts locally, you can look at our JSON files, which link further to other JSON files. This file is a good starting point.