1 # Copyright (C) 2021 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # parent class for LeiImport, LeiConvert
5 package PublicInbox::LeiInput;
9 use PublicInbox::Spawn qw(which popen_rd);
12 # https://www.iana.org/assignments/imap-jmap-keywords/imap-jmap-keywords.xhtml
13 our @KW = (qw(seen answered flagged draft), # widely-compatible
14 qw(forwarded), # IMAP + Maildir
15 qw(phishing junk notjunk)); # rarely supported
17 # note: RFC 8621 states "Users may add arbitrary keywords to an Email",
18 # but is it good idea? Stick to the system and reserved ones, for now.
19 # The widely-compatible ones map to IMAP system flags, Maildir flags
20 # and mbox Status/X-Status headers.
21 my %KW = map { $_ => 1 } @KW;
22 my $L_MAX = 244; # Xapian term limit - length('L')
24 # RFC 8621, sec 2 (Mailboxes) a "label" for us is a JMAP Mailbox "name"
25 # "Servers MAY reject names that violate server policy"
29 length($label) >= $L_MAX and
30 return "`$label' too long (must be <= $L_MAX)";
31 $label =~ m{\A[a-z0-9_](?:[a-z0-9_\-\./\@,]*[a-z0-9])?\z}i ?
32 undef : "`$label' is invalid";
36 $KW{$kw} ? undef : <<EOM;
37 `$kw' is not one of: `seen', `flagged', `answered', `draft'
38 `junk', `notjunk', `phishing' or `forwarded'
43 sub check_input_format ($;$) {
44 my ($lei, $files) = @_;
45 my $opt_key = 'in-format';
46 my $fmt = $lei->{opt}->{$opt_key};
48 my $err = $files ? "regular file(s):\n@$files" : '--stdin';
49 return $lei->fail("--$opt_key unset for $err");
51 return 1 if $fmt eq 'eml';
52 require PublicInbox::MboxLock if $files;
53 require PublicInbox::MboxReader;
54 PublicInbox::MboxReader->reads($fmt) or
55 return $lei->fail("--$opt_key=$fmt unrecognized");
59 # import a single file handle of $name
60 # Subclass must define ->input_eml_cb and ->input_mbox_cb
62 my ($self, $ifmt, $fh, $name, @args) = @_;
64 my $buf = do { local $/; <$fh> } //
65 return $self->{lei}->child_error(1 << 8, <<"");
66 error reading $name: $!
68 # mutt pipes single RFC822 messages with a "From " line,
69 # but no Content-Length or "From " escaping.
70 # "git format-patch" also generates such files by default.
71 $buf =~ s/\A[\r\n]*From [^\r\n]*\r?\n//s;
72 $self->input_eml_cb(PublicInbox::Eml->new(\$buf), @args);
74 # prepare_inputs already validated $ifmt
75 my $cb = PublicInbox::MboxReader->reads($ifmt) //
76 die "BUG: bad fmt=$ifmt";
77 $cb->(undef, $fh, $self->can('input_mbox_cb'), $self, @args);
81 # handles mboxrd endpoints described in Documentation/design_notes.txt
82 sub handle_http_input ($$@) {
83 my ($self, $url, @args) = @_;
84 my $lei = $self->{lei} or die 'BUG: {lei} missing';
85 my $curl_opt = delete $self->{"-curl-$url"} or
86 die("BUG: $url curl options not prepared");
87 my $uri = pop @$curl_opt;
88 my $curl = PublicInbox::LeiCurl->new($lei, $self->{curl}) or return;
89 push @$curl, '-s', @$curl_opt;
90 my $cmd = $curl->for_uri($lei, $uri);
92 my $rdr = { 2 => $lei->{2}, pgid => 0 };
93 my ($fh, $pid) = popen_rd($cmd, undef, $rdr);
94 grep(/\A--compressed\z/, @$curl) or
95 $fh = IO::Uncompress::Gunzip->new($fh, MultiStream => 1);
97 PublicInbox::MboxReader->mboxrd($fh,
98 $self->can('input_mbox_cb'),
104 $lei->child_error($? || 1, "@$cmd failed".$err ? " $err" : '');
108 my ($self, $input, @args) = @_;
109 my $lei = $self->{lei};
110 my $ifmt = lc($lei->{opt}->{'in-format'} // '');
112 if ($input =~ m!\Aimaps?://!i) {
113 $lei->{net}->imap_each($input, $self->can('input_net_cb'),
116 } elsif ($input =~ m!\A(?:nntps?|s?news)://!i) {
117 $lei->{net}->nntp_each($input, $self->can('input_net_cb'),
120 } elsif ($input =~ m!\Ahttps?://!i) {
121 handle_http_input($self, $input, @args);
124 if ($input =~ s!\A([a-z0-9]+):!!i) {
126 } elsif ($input =~ /\.(?:patch|eml)\z/i) {
129 my $devfd = $lei->path_to_fd($input) // return;
131 $self->input_fh($ifmt, $lei->{$devfd}, $input, @args);
132 } elsif (-f $input && $ifmt eq 'eml') {
133 open my $fh, '<', $input or
134 return $lei->fail("open($input): $!");
135 $self->input_fh($ifmt, $fh, $input, @args);
137 my $m = $lei->{opt}->{'lock'} //
138 PublicInbox::MboxLock->defaults;
139 my $mbl = PublicInbox::MboxLock->acq($input, 0, $m);
140 my $zsfx = PublicInbox::MboxReader::zsfx($input);
142 my $in = delete $mbl->{fh};
144 PublicInbox::MboxReader::zsfxcat($in, $zsfx, $lei);
146 local $PublicInbox::DS::in_loop = 0 if $zsfx; # dwaitpid
147 $self->input_fh($ifmt, $mbl->{fh}, $input, @args);
148 } elsif (-d _ && (-d "$input/cur" || -d "$input/new")) {
149 return $lei->fail(<<EOM) if $ifmt && $ifmt ne 'maildir';
150 $input appears to a be a maildir, not $ifmt
152 PublicInbox::MdirReader->new->maildir_each_eml($input,
153 $self->can('input_maildir_cb'),
156 $lei->fail("$input unsupported (TODO)");
160 sub bad_http ($$;$) {
161 my ($lei, $url, $alt) = @_;
162 my $x = $alt ? "did you mean <$alt>?" : 'download and import manually';
163 $lei->fail("E: <$url> not recognized, $x");
166 sub prepare_http_input ($$$) {
167 my ($self, $lei, $url) = @_;
169 require PublicInbox::MboxReader;
170 require PublicInbox::LeiCurl;
171 require IO::Uncompress::Gunzip;
172 $self->{curl} //= which('curl') or
173 return $lei->fail("curl missing for <$url>");
174 my $uri = URI->new($url);
175 my $path = $uri->path;
176 my %qf = $uri->query_form;
178 if ($path =~ m!/(?:t\.mbox\.gz|all\.mbox\.gz)\z!) {
180 } elsif ($path =~ m!/raw\z!) {
181 push @curl_opt, '--compressed';
182 # convert search query to mboxrd request since they require POST
183 # this is only intended for PublicInbox::WWW, and will false-positive
184 # on many other search engines... oh well
185 } elsif (defined $qf{'q'}) {
187 $uri->query_form(\%qf);
188 push @curl_opt, '-d', '';
189 $$uri ne $url and $lei->qerr(<<"");
190 # <$url> rewritten to <$$uri> with HTTP POST
192 # try to provide hints for /$INBOX/$MSGID/T/ and /$INBOX/
193 } elsif ($path =~ s!/[tT]/\z!/t.mbox.gz! ||
194 $path =~ s!/t\.atom\z!/t.mbox.gz! ||
195 $path =~ s!/([^/]+\@[^/]+)/\z!/$1/raw!) {
197 return bad_http($lei, $url, $$uri);
199 return bad_http($lei, $url);
201 $self->{"-curl-$url"} = [ @curl_opt, $uri ]; # for handle_http_input
204 sub prepare_inputs { # returns undef on error
205 my ($self, $lei, $inputs) = @_;
206 my $in_fmt = $lei->{opt}->{'in-format'};
207 my $sync = $lei->{opt}->{sync} ? {} : undef; # using LeiMailSync
208 if ($lei->{opt}->{stdin}) {
210 $lei->fail("--stdin and @$inputs do not mix");
211 check_input_format($lei) or return;
212 push @$inputs, '/dev/stdin';
213 push @{$sync->{no}}, '/dev/stdin' if $sync;
215 my $net = $lei->{net}; # NetWriter may be created by l2m
217 # e.g. Maildir:/home/user/Mail/ or imaps://example.com/INBOX
218 for my $input (@$inputs) {
219 my $input_path = $input;
220 if ($input =~ m!\A(?:imaps?|nntps?|s?news)://!i) {
221 require PublicInbox::NetReader;
222 $net //= PublicInbox::NetReader->new;
223 $net->add_url($input);
225 if ($input =~ m!\Aimaps?://!) {
226 push @{$sync->{ok}}, $input;
228 push @{$sync->{no}}, $input;
231 } elsif ($input_path =~ m!\Ahttps?://!i) {
232 prepare_http_input($self, $lei, $input_path) or return;
233 } elsif ($input_path =~ s/\A([a-z0-9]+)://is) {
235 if (($in_fmt // $ifmt) ne $ifmt) {
236 return $lei->fail(<<"");
237 --in-format=$in_fmt and `$ifmt:' conflict
241 if ($ifmt =~ /\A(?:maildir|mh)\z/i) {
242 push @{$sync->{ok}}, $input;
244 push @{$sync->{no}}, $input;
247 my $devfd = $lei->path_to_fd($input_path) // return;
248 if ($devfd >= 0 || (-f $input_path || -p _)) {
249 require PublicInbox::MboxLock;
250 require PublicInbox::MboxReader;
251 PublicInbox::MboxReader->reads($ifmt) or return
252 $lei->fail("$ifmt not supported");
253 } elsif (-d $input_path) {
254 require PublicInbox::MdirReader;
255 $ifmt eq 'maildir' or return
256 $lei->fail("$ifmt not supported");
257 $input = $lei->abs_path($input) if $sync;
259 return $lei->fail("Unable to handle $input");
261 } elsif ($input =~ /\.(eml|patch)\z/i && -f $input) {
262 lc($in_fmt//'eml') eq 'eml' or return $lei->fail(<<"");
263 $input is `eml', not --in-format=$in_fmt
265 require PublicInbox::Eml;
266 push @{$sync->{no}}, $input if $sync;
268 my $devfd = $lei->path_to_fd($input) // return;
269 if ($devfd >= 0 || -f $input || -p _) {
270 push @{$sync->{no}}, $input if $sync;
272 } elsif (-d $input) {
274 $input = $lei->abs_path($input);
275 push @{$sync->{ok}}, $input;
279 return $lei->fail("Unable to handle $input")
283 if (@f) { check_input_format($lei, \@f) or return }
284 if (@d) { # TODO: check for MH vs Maildir, here
285 require PublicInbox::MdirReader;
287 if ($sync && $sync->{no}) {
288 return $lei->fail(<<"") if !$sync->{ok};
289 --sync specified but no inputs support it
291 # non-fatal if some inputs support support sync
292 $lei->err("# --sync will only be used for @{$sync->{ok}}");
293 $lei->err("# --sync is not supported for: @{$sync->{no}}");
296 if (my $err = $net->errors) {
297 return $lei->fail($err);
299 $net->{quiet} = $lei->{opt}->{quiet};
300 require PublicInbox::LeiAuth;
301 $lei->{auth} //= PublicInbox::LeiAuth->new;
302 $lei->{net} //= $net;
304 $self->{inputs} = $inputs;
309 for my $input (@{$self->{inputs}}) {
310 $self->input_path_url($input);
312 my $wait = $self->{lei}->{sto}->ipc_do('done') if $self->{lei}->{sto};
315 sub input_only_atfork_child {
317 my $lei = $self->{lei};
318 $lei->_lei_atfork_child;
319 PublicInbox::IPC::ipc_atfork_child($self);
320 $lei->{auth}->do_auth_atfork($self) if $lei->{auth};
324 # like Getopt::Long, but for +kw:FOO and -kw:FOO to prepare
325 # for update_xvmd -> update_vmd
326 sub vmd_mod_extract {
331 if ($x =~ /\A(\+|\-)(kw|L):(.+)\z/) {
332 my ($op, $pfx, $val) = ($1, $2, $3);
333 if (my $err = $ERR{$pfx}->($val)) {
334 push @{$vmd_mod->{err}}, $err;
335 } else { # set "+kw", "+L", "-L", "-kw"
336 push @{$vmd_mod->{$op.$pfx}}, $val;