path: root/git-gmi/git.py

from pygit2 import *
from hurry.filesize import size, alternative
from datetime import datetime, timedelta
import dateutil.parser
from pathlib import Path
import os
import shutil
import mimetypes
from const import *
from config import *

mimetypes.add_type("text/gemini", ".gmi")
mimetypes.add_type("text/gemini", ".gemini")


def convert_filesize(bytes: int) -> str:
    # convert filesize in bytes to a human-friendly format
    return size(bytes, system=alternative)


class GitGmiRepo:
    def __init__(self, name: str, path: str):
        self.name = name
        self.path = path
        self.cache_dir = Path(CACHE_DIR) / name
        self._init_cache()
        try:
            self.repo = Repository(path)
        except GitError:
            raise FileNotFoundError(f"Error: no such repo: {name}")

    def _init_cache(self):
        try:
            os.mkdir(self.cache_dir)
        except FileExistsError:
            pass

    def _read_cache(self, req: list) -> str:
        # req is what the user requests after the repo name,
        # like ["tree", "master", "src"]
        # which points to a file called tree_master_src.gmi
        # file content:
        # 20 text/gemini
        # [body - page content]
        # [newline]
        # cached at:
        # [iso timestamp]
        fn = "_".join(req) + ".gmi"
        try:
            with open(self.cache_dir / fn) as f:
                response = f.read()
                f.close()
                created_at = dateutil.parser.isoparse(response.splitlines()[-1])
                if datetime.now() - created_at < timedelta(seconds=CACHE_TTL):
                    # cache valid
                    # response will include the timestamp
                    return response
        except FileNotFoundError:
            pass

        return None

    def _write_cache(self, req: list, resp: str):
        # write resp into cache, appended with timestamp
        fn = "_".join(req) + ".gmi"
        try:
            f = open(self.cache_dir / fn, "x")
        except FileExistsError:
            f = open(self.cache_dir / fn, "w")
        f.write(resp + "\ncached at:\n" + datetime.now().isoformat())

    def _flush_cache(self):
        try:
            shutil.rmtree(self.cache_dir)
        except FileNotFoundError:
            pass

    def _generate_header(self):
        # global "header" to display above all views (except raw files)
        header = (
            f"# {self.name}\n"
            f"=> {CGI_PATH} {GIT_GMI_SITE_TITLE}\n"
            f"=> {CGI_PATH}{self.name}/summary summary\n"
            f"=> {CGI_PATH}{self.name}/tree/{MAIN_BRANCH}/ tree\n"
            f"=> {CGI_PATH}{self.name}/log log\n"
            f"=> {CGI_PATH}{self.name}/refs refs\n\n"
        )
        return header

    def view_summary(self) -> str:
        cached = self._read_cache(["summary"])
        if cached is not None:
            return cached

        response = f"{STATUS_SUCCESS} {META_GEMINI}\r\n" + self._generate_header()
        # show 3 recent commits
        recent_commits = self._get_commit_log()[:3]
        for cmt in recent_commits:
            time = str(datetime.utcfromtimestamp(cmt["time"])) + " UTC"
            response += (
                f"### {cmt['short_id']} - {cmt['author']} - {time}\n"
                f"{cmt['msg'].splitlines()[0]}\n\n"
            )  # TODO: link to commit view
        # find and display readme(.*)
        tree = self._get_tree(MAIN_BRANCH)
        trls = self._list_tree(tree)
        found_readme = False
        for item in trls:
            if (
                item["type"] == "file"
                and item["name"].lower().split(".")[0] == ("readme")
                and not found_readme
            ):
                found_readme = True
                response += (
                    f"## {item['name']} | {convert_filesize(item['size'])}\n"
                    f"{item['blob'].data.decode('utf-8')}"
                )
        if not found_readme:
            response += "## No readme found."

        self._write_cache(["summary"], response)

        return response

    def _get_commit_log(self) -> list:
        # returns useful info from commit log.
        repo = self.repo
        commits = list(repo.walk(repo[repo.head.target].id, GIT_SORT_TIME))
        log = [
            {
                "id": str(cmt.id),  # hex SHA-1 hash
                "short_id": str(cmt.short_id),  # short version of the above
                "author": cmt.author.name,  # author's display name
                "time": cmt.commit_time,  # unix timestamp
                "msg": cmt.message,  # full commit message
            }
            for cmt in commits
        ]

        return log  # reverse chronical order

    def view_log(self) -> str:
        cached = self._read_cache(["log"])
        if cached is not None:
            return cached
        response = f"{STATUS_SUCCESS} {META_GEMINI}\r\n" + self._generate_header()
        log = self._get_commit_log()
        for cmt in log:
            # looks like "2020-06-06 04:51:21 UTC"
            time = str(datetime.utcfromtimestamp(cmt["time"])) + " UTC"
            response += (
                f"## {cmt['short_id']} - {cmt['author']} - {time}\n"
                f"=> commit/{cmt['id']} view diff\n"
                f"=> tree/{cmt['id']}/ view tree\n"
                f"{cmt['msg']}\n\n"
            )
        self._write_cache(["log"], response)
        return response

    def _get_commit(self, commit_str) -> dict:
        try:
            commit = self.repo.revparse_single(commit_str)
            diff = self.repo.diff(commit.parents[0], commit)
            return {
                "id": commit.id,
                "author": commit.author.name,
                "time": commit.commit_time,
                "msg": commit.message,
                "patch": diff.patch,
            }
        except ValueError:
            raise FileNotFoundError(f"Error: no such commit: {commit_str}")

    def view_commit(self, commit_str) -> str:
        cached = self._read_cache(["commit", commit_str])
        if cached is not None:
            return cached
        commit = self._get_commit(commit_str)
        response = (
            f"{STATUS_SUCCESS} {META_GEMINI}\r\n"
            + self._generate_header()
            + f"{commit['id']} - {commit['author']} - {commit['time']}\n"
            + commit["msg"]
            + "\n"
            + f"=> {CGI_PATH}{self.name}/tree/{commit['id']}/ view tree\n"
            + f"=> {commit_str}?raw view raw\n"
            + "\n```\n"
            + commit["patch"]
            + "\n```"
        )
        self._write_cache(["commit", commit_str], response)
        return response

    def view_raw_commit(self, commit_str) -> str:
        commit = self.get_commit(commit_str)
        response = f"{STATUS_SUCCESS} {META_PLAINTEXT}\r\n" + commit["patch"]
        return response

    def _get_refs(self) -> list:
        refs = self.repo.listall_reference_objects()
        return [
            {
                "name": ref.name,
                "shorthand": ref.shorthand,
                "target": ref.target,
                "type": ref.type,
            }
            for ref in refs
        ]

    def view_refs(self) -> str:
        cached = self._read_cache(["refs"])
        if cached is not None:
            return cached
        response = f"{STATUS_SUCCESS} {META_GEMINI}\r\n" + self._generate_header()
        refs = self._get_refs()
        for ref in refs:
            # HACK: filter out refs with slashes as remote branches
            if ref["shorthand"].find("/") == -1:
                response += (
                    f"## {ref['shorthand']}\n=> tree/{ref['shorthand']}/ view tree\n\n"
                )
        self._write_cache(["refs"], response)
        return response

    @classmethod
    def _parse_recursive_tree(cls, tree: Tree) -> list:
        # recursively replace all Trees with a list of Blobs inside it,
        # bundled with the Tree's name as a tuple,
        # e.g. [('src', [blob0, blob1]), otherblob].
        tree_list = list(tree)
        for idx, item in enumerate(tree_list):
            if isinstance(item, Tree):
                tree_list[idx] = (item.name, cls._parse_recursive_tree(tree_list[idx]))

        return tree_list

    def _get_tree(self, revision_str: str) -> list:
        # returns a recursive list of Blob objects
        try:
            revision = self.repo.revparse_single(revision_str)
            if isinstance(revision, Commit):
                # top level tree; may contain sub-trees
                return self._parse_recursive_tree(revision.tree)
            elif isinstance(revision, Tag):
                return self._parse_recursive_tree(revision.get_object().tree)
        except ValueError:
            raise FileNotFoundError(f"Error: no such tree: {revision_str}")
            return None

    @staticmethod
    def _list_tree(tree_list: list, location=[]) -> list:
        # tree_list is the output of _parse_recursive_tree(<tree>);
        # location is which dir you are viewing, represented path-like
        # in a list, e.g. ['src', 'static', 'css'] => 'src/static/css',
        # which this method will cd into and display to the visitor.
        # when there is no such dir, raises FileNotFoundError.
        trls = tree_list
        for loc in location:
            found = False
            for item in trls:
                if isinstance(item, tuple) and item[0] == loc:
                    trls = item[1]
                    found = True
                    break
            if not found:
                raise FileNotFoundError(
                    f"Error: no such directory: {'/'.join(location)}"
                )

        contents = []
        for item in trls:
            if isinstance(item, tuple):
                # was originally a Tree; structure: ('dir_name', [list_of_blobs])
                contents.append(
                    {
                        "type": "dir",
                        "name": item[0],
                        "items": len(item[1]),  # number of objects in dir
                    }
                )

            elif isinstance(item, Blob):
                contents.append(
                    {
                        "type": "file",
                        "name": item.name,
                        "blob": item,
                        "size": item.size,  # size in bytes
                    }
                )

        return contents

    def view_tree(self, branch: str, location=[]) -> str:
        # actual Gemini response
        # consists of a header and a body
        cached = self._read_cache(["tree", branch] + location)
        if cached is not None:
            return cached

        tree = self._get_tree(branch)
        contents = self._list_tree(tree, location)
        items = len(contents)
        response = (
            f"{STATUS_SUCCESS} {META_GEMINI}\r\n"
            + self._generate_header()
            + f"## {self.name}{'/' if location else ''}{'/'.join(location)}/"
            f" | {items} {'items' if items > 1 else 'item'}\n\n"
        )
        for item in contents:
            if item["type"] == "dir":
                response += (
                    f"=> {item['name']}/ {item['name']}/ | {item['items']} items\n"
                )
            elif item["type"] == "file":
                response += f"=> {item['name']} {item['name']} | {convert_filesize(item['size'])}\n"
        self._write_cache(["tree", branch] + location, response)
        return response

    def _get_blob(self, commit_str: str, location=[]) -> Blob:
        # returns a specific Blob object
        # location: just like that of _list_tree, but the last element
        # is the filename
        try:
            tree = self._get_tree(commit_str)
            trls = self._list_tree(tree, location[:-1])
            for item in trls:
                if item["type"] == "file" and item["name"] == location[-1]:
                    return item["blob"]
            raise FileNotFoundError(f"Error: no such file: {'/'.join(location)}")
        except FileNotFoundError:
            raise FileNotFoundError(f"Error: No such tree: {'/'.join(location[:-1])}")

    def view_blob(self, branch: str, location=[]) -> str:
        cached = self._read_cache(["tree", branch] + location)
        if cached is not None:
            return cached
        blob = self._get_blob(branch, location)
        response = (
            f"{STATUS_SUCCESS} {META_GEMINI}\r\n"
            + self._generate_header()
            + f"## {self.name}/{'/'.join(location)} | {convert_filesize(blob.size)}\n\n"
            f"=> {blob.name}?raw view raw\n\n"
            f"```\n"
        )
        response += blob.data.decode("utf-8") + "\n```"
        self._write_cache(["tree", branch] + location, response)
        return response

    def view_raw_blob(self, branch: str, location=[]) -> str:
        blob = self._get_blob(branch, location)
        # if mimetypes can't make out the type, set it to plaintext
        guessed_mimetype = mimetypes.guess_type(blob.name)[0] or META_PLAINTEXT
        response = f"{STATUS_SUCCESS} {guessed_mimetype}\r\n"
        response += blob.data.decode("utf-8")
        return response