[image] How to generate a Dockerfile from an image?

Is it possible to generate a Dockerfile from an image? I want to know for two reasons:

1. I can download images from the repository but would like to see the recipe that generated them.

2. I like the idea of saving snapshots, but once I am done it would be nice to have a structured format to review what was done.

To understand how a docker image was built, use the docker history --no-trunc command.

You can build a docker file from an image, but it will not contain everything you would want to fully understand how the image was generated. Reasonably what you can extract is the MAINTAINER, ENV, EXPOSE, VOLUME, WORKDIR, ENTRYPOINT, CMD, and ONBUILD parts of the dockerfile.

The following script should work for you:

#!/bin/bash
docker history --no-trunc "$1" | \
sed -n -e 's,.*/bin/sh -c #(nop) \(MAINTAINER .*[^ ]\) *0 B,\1,p' | \
head -1
docker inspect --format='{{range $e := .Config.Env}}
ENV {{$e}}
{{end}}{{range $e,$v := .Config.ExposedPorts}}
EXPOSE {{$e}}
{{end}}{{range $e,$v := .Config.Volumes}}
VOLUME {{$e}}
{{end}}{{with .Config.User}}USER {{.}}{{end}}
{{with .Config.WorkingDir}}WORKDIR {{.}}{{end}}
{{with .Config.Entrypoint}}ENTRYPOINT {{json .}}{{end}}
{{with .Config.Cmd}}CMD {{json .}}{{end}}
{{with .Config.OnBuild}}ONBUILD {{json .}}{{end}}' "$1"

I use this as part of a script to rebuild running containers as images: https://github.com/docbill/docker-scripts/blob/master/docker-rebase

The Dockerfile is mainly useful if you want to be able to repackage an image.

The thing to keep in mind, is a docker image can actually just be the tar backup of a real or virtual machine. I have made several docker images this way. Even the build history shows me importing a huge tar file as the first step in creating the image...

I somehow absolutely missed the actual command in the accepted answer, so here it is again, bit more visible in its own paragraph, to see how many people are like me

$ docker history --no-trunc <IMAGE_ID>

docker pull chenzj/dfimage


alias dfimage="docker run -v /var/run/docker.sock:/var/run/docker.sock --rm chenzj/dfimage"

dfimage image_id

Below is ouput of dfimage command:

$ dfimage 0f1947a021ce

FROM node:8
WORKDIR /usr/src/app

COPY file:e76d2e84545dedbe901b7b7b0c8d2c9733baa07cc821054efec48f623e29218c in ./
RUN /bin/sh -c npm install
COPY dir:a89a4894689a38cbf3895fdc0870878272bb9e09268149a87a6974a274b2184a in .

EXPOSE 8080
CMD ["npm" "start"]

A bash solution :

docker history --no-trunc $argv  | tac | tr -s ' ' | cut -d " " -f 5- | sed 's,^/bin/sh -c #(nop) ,,g' | sed 's,^/bin/sh -c,RUN,g' | sed 's, && ,\n  & ,g' | sed 's,\s*[0-9]*[\.]*[0-9]*\s*[kMG]*B\s*$,,g' | head -n -1

Step by step explanations:

tac : reverse the file
tr -s ' '                                       trim multiple whitespaces into 1
cut -d " " -f 5-                                remove the first fields (until X months/years ago)
sed 's,^/bin/sh -c #(nop) ,,g'                  remove /bin/sh calls for ENV,LABEL...
sed 's,^/bin/sh -c,RUN,g'                       remove /bin/sh calls for RUN
sed 's, && ,\n  & ,g'                           pretty print multi command lines following Docker best practices
sed 's,\s*[0-9]*[\.]*[0-9]*\s*[kMG]*B\s*$,,g'      remove layer size information
head -n -1                                      remove last line ("SIZE COMMENT" in this case)

Example:

 ~ ? dih ubuntu:18.04
ADD file:28c0771e44ff530dba3f237024acc38e8ec9293d60f0e44c8c78536c12f13a0b in /
RUN set -xe
   &&  echo '#!/bin/sh' > /usr/sbin/policy-rc.d
   &&  echo 'exit 101' >> /usr/sbin/policy-rc.d
   &&  chmod +x /usr/sbin/policy-rc.d
   &&  dpkg-divert --local --rename --add /sbin/initctl
   &&  cp -a /usr/sbin/policy-rc.d /sbin/initctl
   &&  sed -i 's/^exit.*/exit 0/' /sbin/initctl
   &&  echo 'force-unsafe-io' > /etc/dpkg/dpkg.cfg.d/docker-apt-speedup
   &&  echo 'DPkg::Post-Invoke { "rm -f /var/cache/apt/archives/*.deb /var/cache/apt/archives/partial/*.deb /var/cache/apt/*.bin || true"; };' > /etc/apt/apt.conf.d/docker-clean
   &&  echo 'APT::Update::Post-Invoke { "rm -f /var/cache/apt/archives/*.deb /var/cache/apt/archives/partial/*.deb /var/cache/apt/*.bin || true"; };' >> /etc/apt/apt.conf.d/docker-clean
   &&  echo 'Dir::Cache::pkgcache ""; Dir::Cache::srcpkgcache "";' >> /etc/apt/apt.conf.d/docker-clean
   &&  echo 'Acquire::Languages "none";' > /etc/apt/apt.conf.d/docker-no-languages
   &&  echo 'Acquire::GzipIndexes "true"; Acquire::CompressionTypes::Order:: "gz";' > /etc/apt/apt.conf.d/docker-gzip-indexes
   &&  echo 'Apt::AutoRemove::SuggestsImportant "false";' > /etc/apt/apt.conf.d/docker-autoremove-suggests
RUN rm -rf /var/lib/apt/lists/*
RUN sed -i 's/^#\s*\(deb.*universe\)$/\1/g' /etc/apt/sources.list
RUN mkdir -p /run/systemd
   &&  echo 'docker' > /run/systemd/container
CMD ["/bin/bash"]

It is not possible at this point (unless the author of the image explicitly included the Dockerfile).

However, it is definitely something useful! There are two things that will help to obtain this feature.

1. Trusted builds (detailed in this docker-dev discussion
2. More detailed metadata in the successive images produced by the build process. In the long run, the metadata should indicate which build command produced the image, which means that it will be possible to reconstruct the Dockerfile from a sequence of images.

This is derived from @fallino's answer, with some adjustments and simplifications by using the output format option for docker history. Since macOS and Gnu/Linux have different command-line utilities, a different version is necessary for Mac. If you only need one or the other, you can just use those lines.

#!/bin/bash
case "$OSTYPE" in
    linux*)
        docker history --no-trunc --format "{{.CreatedBy}}" $1 | # extract information from layers
        tac                                                    | # reverse the file
        sed 's,^\(|3.*\)\?/bin/\(ba\)\?sh -c,RUN,'             | # change /bin/(ba)?sh calls to RUN
        sed 's,^RUN #(nop) *,,'                                | # remove RUN #(nop) calls for ENV,LABEL...
        sed 's,  *&&  *, \\\n \&\& ,g'                           # pretty print multi command lines following Docker best practices
    ;;
    darwin*)
        docker history --no-trunc --format "{{.CreatedBy}}" $1 | # extract information from layers
        tail -r                                                | # reverse the file
        sed -E 's,^(\|3.*)?/bin/(ba)?sh -c,RUN,'               | # change /bin/(ba)?sh calls to RUN
        sed 's,^RUN #(nop) *,,'                                | # remove RUN #(nop) calls for ENV,LABEL...
        sed $'s,  *&&  *, \\\ \\\n \&\& ,g'                      # pretty print multi command lines following Docker best practices
    ;;
    *)
        echo "unknown OSTYPE: $OSTYPE"
    ;;
esac

Update Dec 2018 to BMW's answer

chenzj/dfimage - as described on hub.docker.com regenerates Dockerfile from other images. So you can use it as follows:

docker pull chenzj/dfimage
alias dfimage="docker run -v /var/run/docker.sock:/var/run/docker.sock --rm chenzj/dfimage"
dfimage IMAGE_ID > Dockerfile