git-importer/lib/git_exporter.py

import logging
import os

import yaml

from lib.binary import is_binary_or_large
from lib.db import DB
from lib.git import Git
from lib.obs import OBS
from lib.proxy_sha256 import ProxySHA256
from lib.tree_builder import TreeBuilder
from lib.user import User


class GitExporter:
    def __init__(self, api_url, project, package, repodir, cachedir):
        self.obs = OBS()
        self.project = project
        self.package = package
        # TODO: Store the api url in the revision
        self.obs.change_url(api_url)
        self.proxy_sha256 = ProxySHA256(self.obs, enabled=True)
        self.git = Git(
            repodir / package,
            committer="Git OBS Bridge",
            committer_email="obsbridge@suse.de",
        ).create()
        self.state_file = os.path.join(self.git.path, ".git", "_flat_state.yaml")
        self.gc_interval = 200
        self.cachedir = cachedir

    def set_gc_interval(self, gc):
        self.gc_interval = gc

    def check_repo_state(self, flats, branch_state):
        state_data = dict()
        if os.path.exists(self.state_file):
            with open(self.state_file, "r") as f:
                state_data = yaml.safe_load(f)
                if type(state_data) != dict:
                    state_data = {}
        left_to_commit = []
        for flat in reversed(flats):
            found_state = False
            for branch in ["factory", "devel"]:
                if flat.commit.dbid == state_data.get(branch):
                    branch_state[branch] = flat.commit
                    flat.commit.git_commit = self.git.branch_head(branch)
                    logging.debug(
                        f"Found {self.git.path}'s {branch} branch in state {flat}"
                    )
                    left_to_commit = []
                    found_state = True
            if not found_state:
                left_to_commit.append(flat)
        return left_to_commit

    def export_as_git(self):
        db = DB()
        tree = TreeBuilder(db).build(self.project, self.package)
        flats = tree.as_flat_list()

        branch_state = {"factory": None, "devel": None}
        left_to_commit = self.check_repo_state(flats, branch_state)

        if not left_to_commit:
            return

        logging.info(f"Commiting into {self.git.path}")
        self.run_gc()
        users = dict()

        for flat in left_to_commit:
            if flat.commit.userid not in users:
                users[flat.commit.userid] = User(db, flat.commit.userid)
            flat.user = users[flat.commit.userid]
            logging.debug(f"USER {flat.user}")
            self.gc_cnt -= 1
            if self.gc_cnt <= 0 and self.gc_interval:
                self.run_gc()
            logging.debug(f"Committing {flat}")
            self.commit_flat(flat, branch_state)

    def run_gc(self):
        self.gc_cnt = self.gc_interval
        self.git.gc()

    def commit_file(self, flat, file, size, md5):
        # have such files been detected as text mimetype before?
        is_text = self.proxy_sha256.is_text(flat.commit.package, file.name)
        if not is_text and is_binary_or_large(file.name, size):
            file_sha256 = self.proxy_sha256.get_or_put(
                flat.commit.project,
                flat.commit.package,
                file.name,
                flat.commit.expanded_srcmd5,
                md5,
                size,
            )
            self.git.add_lfs(file.name, file_sha256["sha256"], size)
        else:
            self.obs.download(
                flat.commit.project,
                flat.commit.package,
                file.name,
                flat.commit.expanded_srcmd5,
                self.git.path,
                self.cachedir,
                file_md5=md5,
            )
            self.git.add(file)

    def branch_fits_parent1(self, flat, branch_state):
        if branch_state[flat.branch] is None:
            # everything fits nothing
            return True
        return flat.parent1 == branch_state[flat.branch]

    def commit_flat(self, flat, branch_state):
        parents = []
        self.git.checkout(flat.branch)

        if flat.parent1:
            if not self.branch_fits_parent1(flat, branch_state):
                logging.debug(f"Reset {flat.branch} onto {flat.parent1.short_string()}")
                assert flat.parent1.git_commit
                self.git.set_branch_head(flat.branch, flat.parent1.git_commit)
                self.git.checkout(flat.branch)
            parents.append(flat.parent1.git_commit)
            if flat.parent2:
                assert flat.parent2.git_commit
                parents.append(flat.parent2.git_commit)

        # create file if not existant
        self.git.add_default_lfs_gitattributes(force=False)

        to_download, to_delete = flat.commit.calc_delta(branch_state[flat.branch])
        for file in to_delete:
            self.git.remove(file)
        for file, size, md5 in to_download:
            self.commit_file(flat, file, size, md5)

        commit = self.git.commit(
            flat.user.realname,
            flat.user.email,
            flat.commit.commit_time,
            # TODO: Normalize better the commit message
            f"{flat.commit.comment}\n\n{flat.commit}",
            parents=parents,
        )
        flat.commit.git_commit = commit
        branch_state[flat.branch] = flat.commit
        with open(self.state_file, "w") as f:
            data = {}
            for branch in ["factory", "devel"]:
                commit = branch_state[branch]
                if commit:
                    data[branch] = commit.dbid
            yaml.dump(data, f)