path: root/post-update

#!/usr/bin/env python3
# Copyright (C) 2019 Robin Krahl <robin.krahl@ireas.org>
# SPDX-License-Identifier: MIT

"""
Git post-update hook that triggers builds on builds.sr.ht for all pushed
references.

The hook tries to find build manifests in the `.build` directory or the
`.build.yml` file.  It triggers builds.sr.ht for each manifest and for each
reference that has been pushed.  (Multiple references pointing to the same
commit are treated as one reference.)

This hook requires the following Git configuration (`git config`):
- `sr.ht.token`: a valid access token, see [meta.sr.ht OAuth API][]
- `sr.ht.url`: the URL of the builds.sr.ht instance (defaults to builds.sr.ht)
- `sr.ht.user`: the name of the user running the builds

You can customize the build manifest before it is submitted by changing the
`customize_manifest`.  For example, you can add triggers based on the branch
being pushed to:

```
def customize_manifest(repo, manifest, commit, refs):
    manifest.setdefault('triggers', [])
    if "master" in refs:
        manifest["triggers"].append({
            "action": "email",
            "condition": "failure",
            "to": "~example/test-dev@lists.sr.ht"
         })
    else:
        manifest["triggers"].append({
            "action": "email",
            "condition": "failure",
            "to": commit.author.email,
         })
    return manifest
```

Requirements:
- Python 3 with
  - pygit2
  - pyyaml
  - requests

TODO:
- Add timeout for HTTP requests.
- Better source handling (currently, the commit is only set for the first
  source).
- Better access code handling.

[meta.sr.ht OAuth API]: https://man.sr.ht/meta.sr.ht/oauth-api.md
"""


import html
import os
import os.path
import pygit2
import requests
import string
import sys
import yaml


def customize_manifest(repo, manifest, commit, refs):
    """
    Customize the *manifest* based on the current *commit* and the *refs* that
    have been pushed to the *repo*.
    """
    return manifest


def get_config(repo):
    """
    Read and return the sr.ht configuration from the *repo*.
    """
    config = {"url": "https://builds.sr.ht"}
    for e in repo.config:
        # breaking change in pygit2 0.26.4
        if isinstance(e, str):
            (key, value) = (e, repo.config[e])
        else:
            (key, value) = (e.name, e.value)

        if key == "sr.ht.token":
            config["token"] = value
        elif key == "sr.ht.url":
            config["url"] = value
        elif key == "sr.ht.user":
            config["user"] = value

    missing = set()
    if not "token" in config:
        missing.add("token")
    if not "user" in config:
        missing.add("user")
    if missing:
        missing = ["sr.ht." + x for x in missing]
        print("Missing configuration values: {}".format(", ".join(missing)))
        return None
    return config


def get_commit(repo, obj):
    """
    Return the commit in *repo* that *obj* points to or *None* if *obj* is
    neither a commit nor a tag.
    """
    if isinstance(obj, pygit2.Commit):
        return obj
    elif isinstance(obj, pygit2.Tag):
        return repo.get(obj.target)
    else:
        return None


def get_commit_summary(commit):
    """
    Return the summary (i. e. first line) for the *commit*.
    """
    return commit.message.split("\n", 1)[0]


def get_repo_name():
    """
    Return the name of the Git repository we are currently working on, based on
    the name of the current working directory.
    """
    name = os.path.basename(os.getcwd())
    if name.endswith(".git"):
        name = name[:-4]
    return name


def parse_refs(repo, refs):
    """
    Parse the given *refs* as revisions in *repo* and return a tuple containing
    a set of commits described by *refs* (with unique IDs) and a dictionary of
    refs by commit ID.
    """
    commits = set()
    commit_ids = set()
    refs_by_commit = {}
    for ref in refs:
        try:
            obj = repo.revparse_single(ref)
        except KeyError:
            continue
        commit = get_commit(repo, obj)
        if not commit:
            continue
        if commit.id not in commit_ids:
            refs_by_commit[commit.id] = set()
            commit_ids.add(commit.id)
            commits.add(commit)
        if ref.startswith("refs/"):
            ref = repo.lookup_reference(ref).shorthand
        else:
            ref = ref[7:]
        refs_by_commit[commit.id].add(ref)
    return (commits, refs_by_commit)


def find_manifests(repo, tree):
    """
    Return a dictionary of build manifests in the *tree* in *repo* by file
    name.  This function checks the `.build.yml` file and the `.builds`
    directory.
    """
    manifests = {}
    if '.builds' in tree:
        entry = tree['.builds']
        if entry.type == 'tree':
            subtree = repo[entry.id]
            for subentry in subtree:
                if subentry.type == 'blob':
                    name = subentry.name
                    manifests[name] = subentry.id
    if '.build.yml' in tree:
        entry = tree['.build.yml']
        if entry.type == 'blob':
            manifests[entry.name] = entry.id
    
    for name in manifests:
        manifests[name] = repo[manifests[name]].data.decode()
    return manifests


def prepare_manifest(repo, text, commit, refs):
    """
    Prepare the manifest *text* for submission.  Currently, this only appends
    `#<commit>` to the first source.  Then it calls customize_manifest to allow
    user customization.
    """
    data = yaml.safe_load(text)
    if "sources" in data:
        data["sources"][0] += "#" + str(commit.id)
    return customize_manifest(repo, data, commit, refs)


def build_note(commit, refs):
    """
    Return the note to use when submitting a build.
    """
    return "{}\n\n{} – [{}](mailto:{}) – {} ".format(
            html.escape(get_commit_summary(commit)),
            str(commit.id)[:7],
            commit.author.name,
            commit.author.email,
            ", ".join(refs))


def submit_build(config, manifest, name, commit, refs):
    """
    Send an API query to builds.sr.ht to submit the a build *manifest* and
    return the response data or *None* if the query failed.
    """
    data = {}
    data["manifest"] = yaml.dump(manifest, default_flow_style=False)
    data["note"] = build_note(commit, refs)
    data["tags"] = [get_repo_name()] + [name] if name else []
    headers = {"Authorization": "token {}".format(config["token"])}
    result = requests.post("{}/api/jobs".format(config["url"]), json=data, headers=headers)
    if result.status_code != 200:
        print("API request failed with status code {}".format(request.status_code))
        return None
    result = result.json()
    if "errors" in result:
        print("API error: {}".format(result["errors"]))
        return None
    return result


def trigger_builds(config, repo, commit, refs):
    """
    Check the *commit* (that has been infered from *refs*) for build manifests
    and trigger corresponding builds on builds.sr.ht.
    """
    manifests = find_manifests(repo, commit.tree)
    for name in manifests:
        manifest = prepare_manifest(repo, manifests[name], commit, refs)
        result = submit_build(config, manifest, name, commit, refs)
        suffix = "{} ".format(name) if name != ".build.yml" else ""
        suffix += "for {}".format(", ".join(refs))
        if result:
            job_id = result["id"]
            print("Build started: {}/~{}/job/{}/ {}".format(config["url"], config["user"], job_id, suffix))
        else:
            print("Could not start build job {}".format(suffix))


def exec_hook():
    """
    Execute the Git post-update hook that triggers builds on builds.sr.ht for
    all pushed references.
    """
    if len(sys.argv) < 2:
        return
    refs = sys.argv[1:]

    repo = pygit2.Repository('.')
    config = get_config(repo)
    if not config:
        return

    (commits, refs) = parse_refs(repo, refs)
    for commit in commits:
        trigger_builds(config, repo, commit, refs[commit.id])


if __name__ == '__main__':
    exec_hook()