|
| 1 | +import requests |
| 2 | + |
| 3 | +import fsspec |
| 4 | + |
| 5 | +from ..spec import AbstractFileSystem |
| 6 | +from ..utils import infer_storage_options |
| 7 | +from .memory import MemoryFile |
| 8 | + |
| 9 | + |
| 10 | +class GistFileSystem(AbstractFileSystem): |
| 11 | + """ |
| 12 | + Interface to files in a single GitHub Gist. |
| 13 | +
|
| 14 | + Provides read-only access to a gist's files. Gists do not contain |
| 15 | + subdirectories, so file listing is straightforward. |
| 16 | +
|
| 17 | + Parameters |
| 18 | + ---------- |
| 19 | + gist_id : str |
| 20 | + The ID of the gist you want to access (the long hex value from the URL). |
| 21 | + sha : str (optional) |
| 22 | + If provided, fetch a particular revision of the gist. If omitted, |
| 23 | + the latest revision is used. |
| 24 | + username : str (optional) |
| 25 | + GitHub username for authentication (required if token is given). |
| 26 | + token : str (optional) |
| 27 | + GitHub personal access token (required if username is given). |
| 28 | + timeout : (float, float) or float, optional |
| 29 | + Connect and read timeouts for requests (default 60s each). |
| 30 | + kwargs : dict |
| 31 | + Passed to AbstractFileSystem base class. |
| 32 | + """ |
| 33 | + |
| 34 | + protocol = "gist" |
| 35 | + gist_url = "https://api.github.com/gists/{gist_id}" |
| 36 | + gist_rev_url = "https://api.github.com/gists/{gist_id}/{sha}" |
| 37 | + |
| 38 | + def __init__( |
| 39 | + self, gist_id, sha=None, username=None, token=None, timeout=None, **kwargs |
| 40 | + ): |
| 41 | + super().__init__(**kwargs) |
| 42 | + self.gist_id = gist_id |
| 43 | + self.sha = sha # revision of the gist (optional) |
| 44 | + if (username is None) ^ (token is None): |
| 45 | + # Both or neither must be set |
| 46 | + if username or token: |
| 47 | + raise ValueError("Auth requires both username and token, or neither.") |
| 48 | + self.username = username |
| 49 | + self.token = token |
| 50 | + # Default timeouts to 60s connect/read if none provided |
| 51 | + self.timeout = timeout if timeout is not None else (60, 60) |
| 52 | + |
| 53 | + # We use a single-level "directory" cache, because a gist is essentially flat |
| 54 | + self.dircache[""] = self._fetch_file_list() |
| 55 | + |
| 56 | + @property |
| 57 | + def kw(self): |
| 58 | + """Auth parameters passed to 'requests' if we have username/token.""" |
| 59 | + if self.username is not None and self.token is not None: |
| 60 | + return {"auth": (self.username, self.token)} |
| 61 | + return {} |
| 62 | + |
| 63 | + def _fetch_gist_metadata(self): |
| 64 | + """ |
| 65 | + Fetch the JSON metadata for this gist (possibly for a specific revision). |
| 66 | + """ |
| 67 | + if self.sha: |
| 68 | + url = self.gist_rev_url.format(gist_id=self.gist_id, sha=self.sha) |
| 69 | + else: |
| 70 | + url = self.gist_url.format(gist_id=self.gist_id) |
| 71 | + |
| 72 | + r = requests.get(url, timeout=self.timeout, **self.kw) |
| 73 | + if r.status_code == 404: |
| 74 | + raise FileNotFoundError( |
| 75 | + f"Gist not found: {self.gist_id}@{self.sha or 'latest'}" |
| 76 | + ) |
| 77 | + r.raise_for_status() |
| 78 | + return r.json() |
| 79 | + |
| 80 | + def _fetch_file_list(self): |
| 81 | + """ |
| 82 | + Returns a list of dicts describing each file in the gist. These get stored |
| 83 | + in self.dircache[""]. |
| 84 | + """ |
| 85 | + meta = self._fetch_gist_metadata() |
| 86 | + files = meta.get("files", {}) |
| 87 | + out = [] |
| 88 | + for fname, finfo in files.items(): |
| 89 | + if finfo is None: |
| 90 | + # Occasionally GitHub returns a file entry with null if it was deleted |
| 91 | + continue |
| 92 | + # Build a directory entry |
| 93 | + out.append( |
| 94 | + { |
| 95 | + "name": fname, # file's name |
| 96 | + "type": "file", # gists have no subdirectories |
| 97 | + "size": finfo.get("size", 0), # file size in bytes |
| 98 | + "raw_url": finfo.get("raw_url"), |
| 99 | + } |
| 100 | + ) |
| 101 | + return out |
| 102 | + |
| 103 | + @classmethod |
| 104 | + def _strip_protocol(cls, path): |
| 105 | + """ |
| 106 | + Remove 'gist://' from the path, if present. |
| 107 | + """ |
| 108 | + # The default infer_storage_options can handle gist://username:token@id/file |
| 109 | + # or gist://id/file, but let's ensure we handle a normal usage too. |
| 110 | + # We'll just strip the protocol prefix if it exists. |
| 111 | + path = infer_storage_options(path).get("path", path) |
| 112 | + return path.lstrip("/") |
| 113 | + |
| 114 | + @staticmethod |
| 115 | + def _get_kwargs_from_urls(path): |
| 116 | + """ |
| 117 | + Parse 'gist://' style URLs into GistFileSystem constructor kwargs. |
| 118 | + For example: |
| 119 | + gist://:TOKEN@<gist_id>/file.txt |
| 120 | + gist://username:TOKEN@<gist_id>/file.txt |
| 121 | + """ |
| 122 | + so = infer_storage_options(path) |
| 123 | + out = {} |
| 124 | + if "username" in so and so["username"]: |
| 125 | + out["username"] = so["username"] |
| 126 | + if "password" in so and so["password"]: |
| 127 | + out["token"] = so["password"] |
| 128 | + if "host" in so and so["host"]: |
| 129 | + # We interpret 'host' as the gist ID |
| 130 | + out["gist_id"] = so["host"] |
| 131 | + return out |
| 132 | + |
| 133 | + def ls(self, path="", detail=False, **kwargs): |
| 134 | + """ |
| 135 | + List files in the gist. Gists are single-level, so any 'path' is basically |
| 136 | + the filename, or empty for all files. |
| 137 | +
|
| 138 | + Parameters |
| 139 | + ---------- |
| 140 | + path : str, optional |
| 141 | + The filename to list. If empty, returns all files in the gist. |
| 142 | + detail : bool, default False |
| 143 | + If True, return a list of dicts; if False, return a list of filenames. |
| 144 | + """ |
| 145 | + path = self._strip_protocol(path or "") |
| 146 | + # If path is empty, return all |
| 147 | + if path == "": |
| 148 | + results = self.dircache[""] |
| 149 | + else: |
| 150 | + # We want just the single file with this name |
| 151 | + all_files = self.dircache[""] |
| 152 | + results = [f for f in all_files if f["name"] == path] |
| 153 | + if not results: |
| 154 | + raise FileNotFoundError(path) |
| 155 | + if detail: |
| 156 | + return results |
| 157 | + else: |
| 158 | + return sorted(f["name"] for f in results) |
| 159 | + |
| 160 | + def _open(self, path, mode="rb", block_size=None, **kwargs): |
| 161 | + """ |
| 162 | + Read a single file from the gist. |
| 163 | + """ |
| 164 | + if mode != "rb": |
| 165 | + raise NotImplementedError("GitHub Gist FS is read-only (no write).") |
| 166 | + |
| 167 | + path = self._strip_protocol(path) |
| 168 | + # Find the file entry in our dircache |
| 169 | + matches = [f for f in self.dircache[""] if f["name"] == path] |
| 170 | + if not matches: |
| 171 | + raise FileNotFoundError(path) |
| 172 | + finfo = matches[0] |
| 173 | + |
| 174 | + raw_url = finfo.get("raw_url") |
| 175 | + if not raw_url: |
| 176 | + raise FileNotFoundError(f"No raw_url for file: {path}") |
| 177 | + |
| 178 | + r = requests.get(raw_url, timeout=self.timeout, **self.kw) |
| 179 | + if r.status_code == 404: |
| 180 | + raise FileNotFoundError(path) |
| 181 | + r.raise_for_status() |
| 182 | + return MemoryFile(path, None, r.content) |
| 183 | + |
| 184 | + def cat(self, path, recursive=False, on_error="raise", **kwargs): |
| 185 | + """ |
| 186 | + Return {path: contents} for the given file or files. If 'recursive' is True, |
| 187 | + and path is empty, returns all files in the gist. |
| 188 | + """ |
| 189 | + paths = self.expand_path(path, recursive=recursive) |
| 190 | + out = {} |
| 191 | + for p in paths: |
| 192 | + try: |
| 193 | + with self.open(p, "rb") as f: |
| 194 | + out[p] = f.read() |
| 195 | + except FileNotFoundError as e: |
| 196 | + if on_error == "raise": |
| 197 | + raise e |
| 198 | + elif on_error == "omit": |
| 199 | + pass # skip |
| 200 | + else: |
| 201 | + out[p] = e |
| 202 | + return out |
0 commit comments