Relative image paths in readmes are incorrect #4918

New Issue

GiteaMirror · 2025-11-02T06:07:15-06:00

GiteaMirror commented

2025-11-02 06:07:15 -06:00

Originally created by @SuperSandro2000 on GitHub (Feb 22, 2020).

Gitea version (or commit ref): 1.12.0+dev-354-gc8d1c3812
Git version: irrelevant
Operating system: Debian
Database (use [x]):
- PostgreSQL
- MySQL
- MSSQL
- SQLite
Can you reproduce the bug at https://try.gitea.io:
- Yes (provide example URL)
- No
- Not relevant

Description

When I visit https://try.gitea.io/supersandro2000/hacker-laws I get a 404 for a missing embedded image in the readme. The image is embedded like this <img width="480px" alt="Diagram: Amdahl's Law" src="./images/amdahls_law.png" /> and it resolves that to https://try.gitea.io/supersandro2000/images/amdahls_law.png which is not correct. The repository is a mirror of https://github.com/dwmkerr/hacker-laws.git .

~~When I open the readme file directly it gets resolved correctly.~~

Also this triggers a http basic auth on my private instance cause I think some api request has that activated.

Screenshots

Originally created by @SuperSandro2000 on GitHub (Feb 22, 2020). - Gitea version (or commit ref): 1.12.0+dev-354-gc8d1c3812 - Git version: irrelevant - Operating system: Debian - Database (use `[x]`): - [ ] PostgreSQL - [ ] MySQL - [ ] MSSQL - [ ] SQLite - Can you reproduce the bug at https://try.gitea.io: - [x] Yes (provide example URL) - [ ] No - [ ] Not relevant ## Description When I visit https://try.gitea.io/supersandro2000/hacker-laws I get a 404 for a missing embedded image in the readme. The image is embedded like this ``<img width="480px" alt="Diagram: Amdahl's Law" src="./images/amdahls_law.png" />`` and it resolves that to https://try.gitea.io/supersandro2000/images/amdahls_law.png which is not correct. The repository is a mirror of https://github.com/dwmkerr/hacker-laws.git . ~~When I open the [readme file directly](https://try.gitea.io/supersandro2000/hacker-laws/src/branch/master/README.md) it gets resolved correctly.~~ Also this triggers a http basic auth on my private instance cause I think some api request has that activated. ## Screenshots

GiteaMirror added the issue/duplicate label 2025-11-02 06:07:15 -06:00

GiteaMirror closed this issue

2025-11-02 06:07:15 -06:00

GiteaMirror commented

2025-11-02 06:07:16 -06:00

@silverwind commented on GitHub (Feb 22, 2020):

The issues is that the relative link resolves to https://try.gitea.io/supersandro2000/hacker-laws/src/branch/master/images/amdahls_law.png which is the /src/ page while it should use /raw/ for the actual image resource.

@silverwind commented on GitHub (Feb 22, 2020): The issues is that the relative link resolves to `https://try.gitea.io/supersandro2000/hacker-laws/src/branch/master/images/amdahls_law.png` which is the `/src/` page while it should use `/raw/` for the actual image resource.

GiteaMirror commented

2025-11-02 06:07:16 -06:00

@SuperSandro2000 commented on GitHub (Feb 22, 2020):

You are right. So it does not work either way.

@SuperSandro2000 commented on GitHub (Feb 22, 2020): You are right. So it does not work either way.

GiteaMirror commented

2025-11-02 06:07:17 -06:00

@zeripath commented on GitHub (Feb 22, 2020):

@SuperSandro2000

In fact you should not be using the ./raw endpoint you should use the ./media endpoint as that will transparently work for LFS.

Now if you want to link to an image the simplest thing is to use the ![Diagram: Amdahl's Law](images/amdahls_law.png)

That being said we can do something better with detecting image links. So we should really make these links relative to the media link.

I wonder if the /media link of a markdown file should actually just render the markdown.

@zeripath commented on GitHub (Feb 22, 2020): @SuperSandro2000 In fact you should not be using the `./raw` endpoint you should use the `./media` endpoint as that will transparently work for LFS. Now if you want to link to an image the simplest thing is to use the `![Diagram: Amdahl's Law](images/amdahls_law.png)` That being said we can do something better with detecting image links. So we should really make these links relative to the media link. I wonder if the /media link of a markdown file should actually just render the markdown.

GiteaMirror commented

2025-11-02 06:07:17 -06:00

@SuperSandro2000 commented on GitHub (Feb 22, 2020):

This is not my repo. I just mirrored it from GitHub and I get greeted with a basic auth pop up all the time when I visit it on my private instance and only when I am not logged in on the demo instance. The pop up is the only reason why I reported this as it is very annoying and disrupts the user.

@SuperSandro2000 commented on GitHub (Feb 22, 2020): This is not my repo. I just mirrored it from GitHub and I get greeted with a basic auth pop up all the time when I visit it on my private instance and only when I am not logged in on the demo instance. The pop up is the only reason why I reported this as it is very annoying and disrupts the user.

GiteaMirror commented

2025-11-02 06:07:17 -06:00

@silverwind commented on GitHub (Feb 22, 2020):

In fact you should not be using the ./raw endpoint you should use the ./media endpoint as that will transparently work for LFS.

I'd expect ./image.png links to files in the repo to resolve to the raw endpoint. If the file has a URL in LFS, it could be used instead, but generally the user should not have to care whether LFS is enabled or not so those images work everywhere.

@silverwind commented on GitHub (Feb 22, 2020): > In fact you should not be using the ./raw endpoint you should use the ./media endpoint as that will transparently work for LFS. I'd expect `./image.png` links to files in the repo to resolve to the `raw` endpoint. If the file has a URL in LFS, it could be used instead, but generally the user should not have to care whether LFS is enabled or not so those images work everywhere.

GiteaMirror commented

2025-11-02 06:07:17 -06:00

@6543 commented on GitHub (Feb 24, 2020):

this is a dublicate of #8830

@6543 commented on GitHub (Feb 24, 2020): this is a dublicate of #8830

GiteaMirror commented

2025-11-02 06:07:18 -06:00

@guillep2k commented on GitHub (Feb 25, 2020):

Close in favor of #8830

@guillep2k commented on GitHub (Feb 25, 2020): Close in favor of #8830

GiteaMirror referenced this issue

2025-11-02 14:49:10 -06:00