3 * Stefano Zacchiroli <zack@cs.unibo.it>
4 * for the HELM Team http://helm.cs.unibo.it/
6 * This file is part of HELM, an Hypertextual, Electronic
7 * Library of Mathematics, developed at the Computer Science
8 * Department, University of Bologna, Italy.
10 * HELM is free software; you can redistribute it and/or
11 * modify it under the terms of the GNU General Public License
12 * as published by the Free Software Foundation; either version 2
13 * of the License, or (at your option) any later version.
15 * HELM is distributed in the hope that it will be useful,
16 * but WITHOUT ANY WARRANTY; without even the implied warranty of
17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 * GNU General Public License for more details.
20 * You should have received a copy of the GNU General Public License
21 * along with HELM; if not, write to the Free Software
22 * Foundation, Inc., 59 Temple Place - Suite 330, Boston,
25 * For details, see the HELM World-Wide-Web page,
26 * http://helm.cs.unibo.it/
29 open Http_getter_common;;
30 open Http_getter_misc;;
31 open Http_getter_types;;
32 open Http_getter_debugger;;
37 let common_headers = [
38 "Cache-Control", "no-cache";
43 (* HTTP queries argument parsing *)
45 (* parse encoding ("format" parameter), default is Enc_normal *)
46 let parse_enc (req: Http_types.request) =
48 (match req#param "format" with
49 | "normal" -> Enc_normal
51 | s -> raise (Http_getter_bad_request ("Invalid format: " ^ s)))
52 with Http_types.Param_not_found _ -> Enc_normal
54 (* parse "patch_dtd" parameter, default is true *)
55 let parse_patch (req: Http_types.request) =
57 (match req#param "patch_dtd" with
58 | s when String.lowercase s = "yes" -> true
59 | s when String.lowercase s = "no" -> false
60 | s -> raise (Http_getter_bad_request ("Invalid patch_dtd value: " ^ s)))
61 with Http_types.Param_not_found _ -> true
63 (* parse output format ("format" parameter), no default value *)
64 let parse_output_format (req: Http_types.request) =
65 match req#param "format" with
66 | s when String.lowercase s = "txt" -> Fmt_text
67 | s when String.lowercase s = "xml" -> Fmt_xml
68 | s -> raise (Http_getter_bad_request ("Invalid /ls format: " ^ s))
70 (* parse "baseuri" format for /ls method, no default value *)
72 let parse_ls_RE = Pcre.regexp "^(\\w+):(.*)$" in
73 let trailing_slash_RE = Pcre.regexp "/+$" in
75 raise (Http_getter_bad_request ("Invalid /ls baseuri: " ^ uri))
77 fun (req: Http_types.request) ->
78 let baseuri = req#param "baseuri" in
81 Pcre.extract ~rex:parse_ls_RE
82 (Pcre.replace ~rex:trailing_slash_RE baseuri)
84 (match (subs.(1), subs.(2)) with
85 | "cic", uri -> Cic uri
86 | "theory", uri -> Theory uri
87 | _ -> wrong_uri baseuri)
88 with Not_found -> wrong_uri baseuri
91 (* global maps, shared by all threads *)
93 let cic_map = new Http_getter_map.map Http_getter_env.cic_dbm in
94 let nuprl_map = new Http_getter_map.map Http_getter_env.nuprl_dbm in
95 let rdf_map = new Http_getter_map.map Http_getter_env.rdf_dbm in
96 let xsl_map = new Http_getter_map.map Http_getter_env.xsl_dbm in
99 cic_map#close; nuprl_map#close; rdf_map#close; xsl_map#close in
100 let map_of_uri = function
101 | uri when is_cic_uri uri -> cic_map
102 | uri when is_nuprl_uri uri -> nuprl_map
103 | uri when is_rdf_uri uri -> rdf_map
104 | uri when is_xsl_uri uri -> xsl_map
105 | uri -> raise (Http_getter_unresolvable_URI uri)
109 (map_of_uri uri)#resolve uri
110 with Http_getter_map.Key_not_found _ ->
111 raise (Http_getter_unresolvable_URI uri)
114 (* Warning: this fail if uri is already registered *)
115 (map_of_uri uri)#add uri
117 let return_all_foo_uris map doctype filter outchan =
118 (** return all URIs contained in 'map' which satisfy predicate 'filter'; URIs
119 are written in an XMLish format ('doctype' is the XML doctype) onto 'outchan'
121 Http_daemon.send_basic_headers ~code:200 outchan;
122 Http_daemon.send_header "Content-Type" "text/xml" outchan;
123 Http_daemon.send_headers common_headers outchan;
124 Http_daemon.send_CRLF outchan;
128 "<?xml version=\"1.0\" encoding=\"ISO-8859-1\"?>
129 <!DOCTYPE %s SYSTEM \"%s/getdtd?uri=%s.dtd\">
134 Http_getter_env.my_own_url
140 output_string outchan (sprintf "\t<uri value=\"%s\" />\n" uri));
141 output_string outchan (sprintf "</%s>\n" doctype)
143 let return_all_xml_uris = return_all_foo_uris cic_map "alluris" in
144 let return_all_rdf_uris = return_all_foo_uris rdf_map "allrdfuris" in
146 let (++) (oldann, oldtypes, oldbody, oldtree)
147 (newann, newtypes, newbody, newtree) =
148 ((if newann > oldann then newann else oldann),
149 (if newtypes > oldtypes then newtypes else oldtypes),
150 (if newbody > oldbody then newbody else oldbody),
151 (if newtree > oldtree then newtree else oldtree))
155 "^([^.]*\\.[^.]*)((\\.body)|(\\.proof_tree)|(\\.types))?(\\.ann)?$"
157 let (types_RE, types_ann_RE, body_RE, body_ann_RE,
158 proof_tree_RE, proof_tree_ann_RE) =
159 (Pcre.regexp "\\.types$", Pcre.regexp "\\.types\\.ann$",
160 Pcre.regexp "\\.body$", Pcre.regexp "\\.body\\.ann$",
161 Pcre.regexp "\\.proof_tree$", Pcre.regexp "\\.proof_tree\\.ann$")
163 let (slash_RE, til_slash_RE, no_slashes_RE) =
164 (Pcre.regexp "/", Pcre.regexp "^.*/", Pcre.regexp "^[^/]*$")
166 fun lsuri fmt outchan ->
169 (match lsuri with Cic p -> ("cic:" ^ p) | Theory p -> ("theory:" ^ p))
171 let (dir_RE, obj_RE) =
172 (Pcre.regexp (pat ^ "/"), Pcre.regexp (pat ^ "(\\.|$)"))
174 let dirs = ref StringSet.empty in
175 let objs = Hashtbl.create 17 in
177 dirs := StringSet.add (List.hd (Pcre.split ~rex:slash_RE d)) !dirs
180 let basepart = Pcre.replace ~rex:basepart_RE ~templ:"$1" o in
181 let no_flags = false, No, No, No in
184 Hashtbl.find objs basepart
185 with Not_found -> (* no ann, no types, no body, no proof tree *)
190 | s when Pcre.pmatch ~rex:types_RE s -> (false, Yes, No, No)
191 | s when Pcre.pmatch ~rex:types_ann_RE s -> (true, Ann, No, No)
192 | s when Pcre.pmatch ~rex:body_RE s -> (false, No, Yes, No)
193 | s when Pcre.pmatch ~rex:body_ann_RE s -> (true, No, Ann, No)
194 | s when Pcre.pmatch ~rex:proof_tree_RE s -> (false, No, No, Yes)
195 | s when Pcre.pmatch ~rex:proof_tree_ann_RE s -> (true, No, No, Ann)
198 Hashtbl.replace objs basepart (oldflags ++ newflags)
200 cic_map#iter (* BLEARGH Dbm module lacks support for fold-like functions *)
203 | uri when Pcre.pmatch ~rex:dir_RE uri -> (* directory hit *)
204 let localpart = Pcre.replace ~rex:dir_RE uri in
205 if Pcre.pmatch ~rex:no_slashes_RE localpart then
209 | uri when Pcre.pmatch ~rex:obj_RE uri -> (* file hit *)
210 store_obj (Pcre.replace ~rex:til_slash_RE uri)
211 | uri -> () (* miss *));
216 (fun s d -> sprintf "%sdir, %s\n" s d) ""
217 (StringSet.elements !dirs)) ^
218 (Http_getter_misc.hashtbl_sorted_fold
219 (fun uri (annflag, typesflag, bodyflag, treeflag) cont ->
220 sprintf "%sobject, %s, <%s,%s,%s,%s>\n"
221 cont uri (if annflag then "YES" else "NO")
222 (string_of_ls_flag typesflag)
223 (string_of_ls_flag bodyflag)
224 (string_of_ls_flag treeflag))
228 ~headers:(("Content-Type", "text/plain") :: common_headers)
233 "<?xml version=\"1.0\" encoding=\"ISO-8859-1\"?>
234 <!DOCTYPE ls SYSTEM \"%s/getdtd?uri=ls.dtd\">
240 Http_getter_env.my_own_url
245 (fun d -> "<section>" ^ d ^ "</section>")
246 (StringSet.elements !dirs))) ^ "\n" ^
247 (Http_getter_misc.hashtbl_sorted_fold
248 (fun uri (annflag, typesflag, bodyflag, treeflag) cont ->
250 "%s<object name=\"%s\">
251 \t<ann value=\"%s\" />
252 \t<types value=\"%s\" />
253 \t<body value=\"%s\" />
254 \t<proof_tree value=\"%s\" />
257 cont uri (if annflag then "YES" else "NO")
258 (string_of_ls_flag typesflag)
259 (string_of_ls_flag bodyflag)
260 (string_of_ls_flag treeflag))
264 ~headers:(("Content-Type", "text/xml") :: common_headers)
267 let (index_line_sep_RE, index_sep_RE, trailing_types_RE,
268 heading_cic_RE, heading_theory_RE, heading_nuprl_RE,
269 heading_rdf_cic_RE, heading_rdf_theory_RE) =
270 (Pcre.regexp "[ \t]+", Pcre.regexp "\r\n|\r|\n",
271 Pcre.regexp "\\.types$",
272 Pcre.regexp "^cic:", Pcre.regexp "^theory:", Pcre.regexp "^nuprl:",
273 Pcre.regexp "^helm:rdf.*//cic:", Pcre.regexp "^helm:rdf.*//theory:")
275 let update_from_server logmsg server_url = (* use global maps *)
276 debug_print ("Updating information from " ^ server_url);
277 let xml_url_of_uri = function
278 (* TODO missing sanity checks on server_url, e.g. it can contains $1 *)
279 | uri when (Pcre.pmatch ~rex:heading_cic_RE uri) ->
280 Pcre.replace ~rex:heading_cic_RE ~templ:server_url uri
281 | uri when (Pcre.pmatch ~rex:heading_theory_RE uri) ->
282 Pcre.replace ~rex:heading_theory_RE ~templ:server_url uri
283 | uri when (Pcre.pmatch ~rex:heading_nuprl_RE uri) ->
284 Pcre.replace ~rex:heading_nuprl_RE ~templ:server_url uri
285 | uri -> raise (Http_getter_invalid_URI uri)
287 let rdf_url_of_uri = function (* TODO as above *)
288 | uri when (Pcre.pmatch ~rex:heading_rdf_cic_RE uri) ->
289 Pcre.replace ~rex:heading_rdf_cic_RE ~templ:server_url uri
290 | uri when (Pcre.pmatch ~rex:heading_rdf_theory_RE uri) ->
291 Pcre.replace ~rex:heading_rdf_theory_RE ~templ:server_url uri
292 | uri -> raise (Http_getter_invalid_URI uri)
294 let log = ref (logmsg ^ "Processing server: " ^ server_url ^ "<br />\n") in
295 let (xml_index, rdf_index, xsl_index) =
296 (* TODO keeps index in memory, is better to keep them on temp files? *)
297 (http_get (server_url ^ "/" ^ Http_getter_env.xml_index),
298 http_get (server_url ^ "/" ^ Http_getter_env.rdf_index),
299 http_get (server_url ^ "/" ^ Http_getter_env.xsl_index))
301 if (xml_index = None && rdf_index = None && xsl_index = None) then
302 debug_print (sprintf "Warning: useless server %s" server_url);
303 (match xml_index with
305 (log := !log ^ "Updating XML db ...<br />\n";
308 | l when is_blank_line l -> () (* skip blank and commented lines *)
311 (match Pcre.split ~rex:index_line_sep_RE l with
313 assert (is_cic_uri uri || is_nuprl_uri uri) ;
314 (map_of_uri uri)#replace
315 uri ((xml_url_of_uri uri) ^ ".xml.gz")
317 assert (is_cic_uri uri || is_nuprl_uri uri) ;
318 (map_of_uri uri)#replace
319 uri ((xml_url_of_uri uri) ^ ".xml")
321 log := !log ^ "Ignoring invalid line: '" ^ l ^ "'<br />\n")
322 with Http_getter_invalid_URI uri ->
323 log := !log ^ "Ignoring invalid XML URI: '" ^ uri ^ "'<br />\n")
324 (Pcre.split ~rex:index_sep_RE xml_index)) (* xml_index lines *)
326 (match rdf_index with
328 (log := !log ^ "Updating RDF db ...<br />\n";
332 (match Pcre.split ~rex:index_line_sep_RE l with
334 rdf_map#replace uri ((rdf_url_of_uri uri) ^ ".xml.gz")
335 | [uri] -> rdf_map#replace uri ((rdf_url_of_uri uri) ^ ".xml")
336 | _ -> log := !log ^ "Ignoring invalid line: " ^ l ^ "<br />\n")
337 with Http_getter_invalid_URI uri ->
338 log := !log ^ "Ignoring invalid RDF URI: " ^ uri ^ "<br />\n")
339 (Pcre.split ~rex:index_sep_RE rdf_index)) (* rdf_index lines *)
341 (match xsl_index with
343 (log := !log ^ "Updating XSLT db ...<br />\n";
345 (fun l -> xsl_map#replace l (server_url ^ "/" ^ l))
346 (Pcre.split ~rex:index_sep_RE xsl_index);
347 log := !log ^ "All done!<br />\n")
349 debug_print "done with this server";
352 let update_from_all_servers () = (* use global maps *)
353 cic_map#clear; nuprl_map#clear; rdf_map#clear; xsl_map#clear;
357 "" (* initial logmsg: empty *)
358 (* reverse order: 1st server is the most important one *)
359 (List.rev !Http_getter_env.servers)
361 cic_map#sync; nuprl_map#sync; rdf_map#sync; xsl_map#sync;
367 let callback (req: Http_types.request) outchan =
369 debug_print ("Connection from " ^ req#clientAddr);
370 debug_print ("Received request: " ^ req#path);
374 (Http_getter_const.usage_string (Http_getter_env.env_to_string ()))
376 | "/getxml" | "/getxslt" | "/getdtd" | "/resolve" | "/register" ->
377 (let uri = req#param "uri" in (* common parameter *)
380 let enc = parse_enc req in
381 let patch = parse_patch req in
382 Http_getter_cache.respond_xml
383 ~url:(resolve uri) ~uri ~enc ~patch outchan
385 let patch = parse_patch req in
386 Http_getter_cache.respond_xsl ~url:(resolve uri) ~patch outchan
388 let patch = parse_patch req in
389 Http_getter_cache.respond_dtd
390 ~patch ~url:(Http_getter_env.dtd_dir ^ "/" ^ uri) outchan
394 (sprintf "<url value=\"%s\" />\n" (resolve uri))
396 with Http_getter_unresolvable_URI uri ->
397 return_xml_raw "<unresolved />\n" outchan)
399 let url = req#param "url" in
401 return_html_msg "Register done" outchan
404 Http_getter_cache.clean ();
405 return_html_msg "Done." outchan
407 Http_getter_env.reload (); (* reload servers list from servers file *)
408 let log = update_from_all_servers () in
409 return_html_msg log outchan
412 (sprintf "<html><body><table>\n%s\n</table></body></html>"
416 fun s -> incr i; sprintf "<tr><td>%d</td><td>%s</td></tr>" !i s)
417 !Http_getter_env.servers)))
420 let name = req#param "url" in
424 let res = int_of_string (req#param "position") in
426 raise (Failure "int_of_string");
428 with Failure "int_of_string" ->
429 raise (Http_getter_bad_request
430 (sprintf "position must be a non negative integer (%s given)"
431 (req#param "position")))
433 if position = 0 then (* fallback to default value *)
434 raise (Http_types.Param_not_found "foo")
435 else if position > 0 then begin (* add server and update all *)
436 Http_getter_env.add_server ~position name;
437 let log = update_from_all_servers () in
439 (sprintf "Added server %s in position %d)<br />\n%s"
442 end else (* position < 0 *) (* error! *)
443 assert false (* already checked above *)
444 with Http_types.Param_not_found _ -> (* add as 1st server by default *)
445 Http_getter_env.add_server ~position:0 name;
446 let log = update_from_server (* quick update (new server only) *)
447 (sprintf "Added server %s in head position<br />\n" name) name
449 return_html_msg log outchan)
450 | "/remove_server" ->
453 let res = int_of_string (req#param "position") in
455 raise (Failure "int_of_string");
457 with Failure "int_of_string" ->
458 raise (Http_getter_bad_request
459 (sprintf "position must be a non negative integer (%s given)"
460 (req#param "position")))
464 List.nth !Http_getter_env.servers position
465 with Failure "nth" ->
466 raise (Http_getter_bad_request
467 (sprintf "no server with position %d" position))
469 Http_getter_env.remove_server position;
470 let log = update_from_all_servers () in
472 (sprintf "Removed server %s (position %d)<br />\n%s"
473 server_name position log)
478 (Pcre.pmatch ~rex:heading_cic_RE uri) &&
479 not (Pcre.pmatch ~rex:trailing_types_RE uri))
481 | "/getallrdfuris" ->
482 (let classs = req#param "class" in
485 let base = "^helm:rdf:www\\.cs\\.unibo\\.it/helm/rdf/" in
487 | ("forward" as c) | ("backward" as c) ->
488 (fun uri -> Pcre.pmatch ~pat:(base ^ c) uri)
489 | c -> raise (Http_getter_invalid_RDF_class c)
491 return_all_rdf_uris filter outchan
492 with Http_getter_invalid_RDF_class c ->
493 raise (Http_getter_bad_request ("Invalid RDF class: " ^ c)))
494 | "/ls" -> return_ls (parse_ls_uri req) (parse_output_format req) outchan
496 Http_daemon.respond ~body:Http_getter_const.empty_xml outchan
498 Http_daemon.respond_error ~status:(`Client_error `Bad_request) outchan);
499 debug_print "Done!\n"
501 | Http_types.Param_not_found attr_name ->
502 return_400 (sprintf "Parameter '%s' is missing" attr_name) outchan
503 | Http_getter_bad_request msg -> return_html_error msg outchan
504 | Http_getter_internal_error msg -> return_html_internal_error msg outchan
505 | Shell.Subprocess_error l ->
506 return_html_internal_error
507 (String.concat "<br />\n"
510 sprintf "Command '%s' returned %s"
511 cmd (string_of_proc_status code))
516 ("Uncaught exception: " ^ (Printexc.to_string exc))
520 (* daemon initialization *)
523 print_string (Http_getter_env.env_to_string ());
525 Unix.putenv "http_proxy" "";
527 Sys.catch_break true;
530 ~timeout:(Some 600) ~port:Http_getter_env.port ~mode:`Thread callback
531 with Sys.Break -> () (* 'save_maps' already registered with 'at_exit' *)