1#!/bin/sh
2"""": # -*-python-*-
3# https://sourceware.org/bugzilla/show_bug.cgi?id=26034
4export "BUP_ARGV_0"="$0"
5arg_i=1
6for arg in "$@"; do
7    export "BUP_ARGV_${arg_i}"="$arg"
8    shift
9    arg_i=$((arg_i + 1))
10done
11# Here to end of preamble replaced during install
12bup_python="$(dirname "$0")/../../config/bin/python" || exit $?
13exec "$bup_python" "$0"
14"""
15# end of bup preamble
16
17from __future__ import absolute_import, print_function
18from collections import namedtuple
19import mimetypes, os, posixpath, signal, stat, sys, time, urllib, webbrowser
20from binascii import hexlify
21
22sys.path[:0] = [os.path.dirname(os.path.realpath(__file__)) + '/..']
23
24from bup import compat, options, git, vfs
25from bup.helpers import (chunkyreader, debug1, format_filesize, handle_ctrl_c,
26                         log, saved_errors)
27from bup.metadata import Metadata
28from bup.path import resource_path
29from bup.repo import LocalRepo
30from bup.io import path_msg
31
32try:
33    from tornado import gen
34    from tornado.httpserver import HTTPServer
35    from tornado.ioloop import IOLoop
36    from tornado.netutil import bind_unix_socket
37    import tornado.web
38except ImportError:
39    log('error: cannot find the python "tornado" module; please install it\n')
40    sys.exit(1)
41
42
43# FIXME: right now the way hidden files are handled causes every
44# directory to be traversed twice.
45
46handle_ctrl_c()
47
48
49def http_date_from_utc_ns(utc_ns):
50    return time.strftime('%a, %d %b %Y %H:%M:%S', time.gmtime(utc_ns / 10**9))
51
52
53def _compute_breadcrumbs(path, show_hidden=False):
54    """Returns a list of breadcrumb objects for a path."""
55    breadcrumbs = []
56    breadcrumbs.append((b'[root]', b'/'))
57    path_parts = path.split(b'/')[1:-1]
58    full_path = b'/'
59    for part in path_parts:
60        full_path += part + b"/"
61        url_append = b""
62        if show_hidden:
63            url_append = b'?hidden=1'
64        breadcrumbs.append((part, full_path+url_append))
65    return breadcrumbs
66
67
68def _contains_hidden_files(repo, dir_item):
69    """Return true if the directory contains items with names other than
70    '.' and '..' that begin with '.'
71
72    """
73    for name, item in vfs.contents(repo, dir_item, want_meta=False):
74        if name in (b'.', b'..'):
75            continue
76        if name.startswith(b'.'):
77            return True
78    return False
79
80
81def _dir_contents(repo, resolution, show_hidden=False):
82    """Yield the display information for the contents of dir_item."""
83
84    url_query = b'?hidden=1' if show_hidden else b''
85
86    def display_info(name, item, resolved_item, display_name=None):
87        # link should be based on fully resolved type to avoid extra
88        # HTTP redirect.
89        link = tornado.escape.url_escape(name, plus=False)
90        if stat.S_ISDIR(vfs.item_mode(resolved_item)):
91            link += '/'
92        link = link.encode('ascii')
93
94        size = vfs.item_size(repo, item)
95        if opt.human_readable:
96            display_size = format_filesize(size)
97        else:
98            display_size = size
99
100        if not display_name:
101            mode = vfs.item_mode(item)
102            if stat.S_ISDIR(mode):
103                display_name = name + b'/'
104            elif stat.S_ISLNK(mode):
105                display_name = name + b'@'
106            else:
107                display_name = name
108
109        return display_name, link + url_query, display_size
110
111    dir_item = resolution[-1][1]
112    for name, item in vfs.contents(repo, dir_item):
113        if not show_hidden:
114            if (name not in (b'.', b'..')) and name.startswith(b'.'):
115                continue
116        if name == b'.':
117            yield display_info(name, item, item, b'.')
118            parent_item = resolution[-2][1] if len(resolution) > 1 else dir_item
119            yield display_info(b'..', parent_item, parent_item, b'..')
120            continue
121        res_item = vfs.ensure_item_has_metadata(repo, item, include_size=True)
122        yield display_info(name, item, res_item)
123
124
125class BupRequestHandler(tornado.web.RequestHandler):
126
127    def initialize(self, repo=None):
128        self.repo = repo
129
130    def decode_argument(self, value, name=None):
131        if name == 'path':
132            return value
133        return super(BupRequestHandler, self).decode_argument(value, name)
134
135    def get(self, path):
136        return self._process_request(path)
137
138    def head(self, path):
139        return self._process_request(path)
140
141    def _process_request(self, path):
142        print('Handling request for %s' % path)
143        sys.stdout.flush()
144        # Set want_meta because dir metadata won't be fetched, and if
145        # it's not a dir, then we're going to want the metadata.
146        res = vfs.resolve(self.repo, path, want_meta=True)
147        leaf_name, leaf_item = res[-1]
148        if not leaf_item:
149            self.send_error(404)
150            return
151        mode = vfs.item_mode(leaf_item)
152        if stat.S_ISDIR(mode):
153            self._list_directory(path, res)
154        else:
155            self._get_file(self.repo, path, res)
156
157    def _list_directory(self, path, resolution):
158        """Helper to produce a directory listing.
159
160        Return value is either a file object, or None (indicating an
161        error).  In either case, the headers are sent.
162        """
163        if not path.endswith(b'/') and len(path) > 0:
164            print('Redirecting from %s to %s' % (path_msg(path), path_msg(path + b'/')))
165            return self.redirect(path + b'/', permanent=True)
166
167        hidden_arg = self.request.arguments.get('hidden', [0])[-1]
168        try:
169            show_hidden = int(hidden_arg)
170        except ValueError as e:
171            show_hidden = False
172
173        self.render(
174            'list-directory.html',
175            path=path,
176            breadcrumbs=_compute_breadcrumbs(path, show_hidden),
177            files_hidden=_contains_hidden_files(self.repo, resolution[-1][1]),
178            hidden_shown=show_hidden,
179            dir_contents=_dir_contents(self.repo, resolution,
180                                       show_hidden=show_hidden))
181
182    @gen.coroutine
183    def _get_file(self, repo, path, resolved):
184        """Process a request on a file.
185
186        Return value is either a file object, or None (indicating an error).
187        In either case, the headers are sent.
188        """
189        file_item = resolved[-1][1]
190        file_item = vfs.augment_item_meta(repo, file_item, include_size=True)
191        meta = file_item.meta
192        ctype = self._guess_type(path)
193        self.set_header("Last-Modified", http_date_from_utc_ns(meta.mtime))
194        self.set_header("Content-Type", ctype)
195
196        self.set_header("Content-Length", str(meta.size))
197        assert len(file_item.oid) == 20
198        self.set_header("Etag", hexlify(file_item.oid))
199        if self.request.method != 'HEAD':
200            with vfs.fopen(self.repo, file_item) as f:
201                it = chunkyreader(f)
202                for blob in chunkyreader(f):
203                    self.write(blob)
204        raise gen.Return()
205
206    def _guess_type(self, path):
207        """Guess the type of a file.
208
209        Argument is a PATH (a filename).
210
211        Return value is a string of the form type/subtype,
212        usable for a MIME Content-type header.
213
214        The default implementation looks the file's extension
215        up in the table self.extensions_map, using application/octet-stream
216        as a default; however it would be permissible (if
217        slow) to look inside the data to make a better guess.
218        """
219        base, ext = posixpath.splitext(path)
220        if ext in self.extensions_map:
221            return self.extensions_map[ext]
222        ext = ext.lower()
223        if ext in self.extensions_map:
224            return self.extensions_map[ext]
225        else:
226            return self.extensions_map['']
227
228    if not mimetypes.inited:
229        mimetypes.init() # try to read system mime.types
230    extensions_map = mimetypes.types_map.copy()
231    extensions_map.update({
232        '': 'text/plain', # Default
233        '.py': 'text/plain',
234        '.c': 'text/plain',
235        '.h': 'text/plain',
236        })
237
238
239io_loop = None
240
241def handle_sigterm(signum, frame):
242    global io_loop
243    debug1('\nbup-web: signal %d received\n' % signum)
244    log('Shutdown requested\n')
245    if not io_loop:
246        sys.exit(0)
247    io_loop.stop()
248
249
250signal.signal(signal.SIGTERM, handle_sigterm)
251
252UnixAddress = namedtuple('UnixAddress', ['path'])
253InetAddress = namedtuple('InetAddress', ['host', 'port'])
254
255optspec = """
256bup web [[hostname]:port]
257bup web unix://path
258--
259human-readable    display human readable file sizes (i.e. 3.9K, 4.7M)
260browser           show repository in default browser (incompatible with unix://)
261"""
262o = options.Options(optspec)
263opt, flags, extra = o.parse(compat.argv[1:])
264
265if len(extra) > 1:
266    o.fatal("at most one argument expected")
267
268if len(extra) == 0:
269    address = InetAddress(host='127.0.0.1', port=8080)
270else:
271    bind_url = extra[0]
272    if bind_url.startswith('unix://'):
273        address = UnixAddress(path=bind_url[len('unix://'):])
274    else:
275        addr_parts = extra[0].split(':', 1)
276        if len(addr_parts) == 1:
277            host = '127.0.0.1'
278            port = addr_parts[0]
279        else:
280            host, port = addr_parts
281        try:
282            port = int(port)
283        except (TypeError, ValueError) as ex:
284            o.fatal('port must be an integer, not %r' % port)
285        address = InetAddress(host=host, port=port)
286
287git.check_repo_or_die()
288
289settings = dict(
290    debug = 1,
291    template_path = resource_path(b'web').decode('utf-8'),
292    static_path = resource_path(b'web/static').decode('utf-8'),
293)
294
295# Disable buffering on stdout, for debug messages
296try:
297    sys.stdout._line_buffering = True
298except AttributeError:
299    sys.stdout = os.fdopen(sys.stdout.fileno(), 'w', 0)
300
301application = tornado.web.Application([
302    (r"(?P<path>/.*)", BupRequestHandler, dict(repo=LocalRepo())),
303], **settings)
304
305http_server = HTTPServer(application)
306io_loop_pending = IOLoop.instance()
307
308if isinstance(address, InetAddress):
309    sockets = tornado.netutil.bind_sockets(address.port, address.host)
310    http_server.add_sockets(sockets)
311    print('Serving HTTP on %s:%d...' % sockets[0].getsockname())
312    if opt.browser:
313        browser_addr = 'http://' + address[0] + ':' + str(address[1])
314        io_loop_pending.add_callback(lambda : webbrowser.open(browser_addr))
315elif isinstance(address, UnixAddress):
316    unix_socket = bind_unix_socket(address.path)
317    http_server.add_socket(unix_socket)
318    print('Serving HTTP on filesystem socket %r' % address.path)
319else:
320    log('error: unexpected address %r', address)
321    sys.exit(1)
322
323io_loop = io_loop_pending
324io_loop.start()
325
326if saved_errors:
327    log('WARNING: %d errors encountered while saving.\n' % len(saved_errors))
328    sys.exit(1)
329