2 * Copyright (C) 2003-2004:
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/
31 open Http_getter_common
33 open Http_getter_types
35 exception Not_implemented of string
36 exception UnexpectedGetterOutput
38 (* resolve_result is needed because it is not possible to raise *)
39 (* an exception in a pxp ever-processing callback. Too bad. *)
45 type logger_callback = HelmLogger.html_tag -> unit
47 let stdout_logger tag = print_string (HelmLogger.string_of_html_tag tag)
49 let not_implemented s = raise (Not_implemented ("Http_getter." ^ s))
51 let (index_line_sep_RE, index_sep_RE, trailing_types_RE,
52 heading_cic_RE, heading_theory_RE, heading_nuprl_RE,
53 heading_rdf_cic_RE, heading_rdf_theory_RE) =
54 (Pcre.regexp "[ \t]+", Pcre.regexp "\r\n|\r|\n",
55 Pcre.regexp "\\.types$",
56 Pcre.regexp "^cic:", Pcre.regexp "^theory:", Pcre.regexp "^nuprl:",
57 Pcre.regexp "^helm:rdf.*//cic:", Pcre.regexp "^helm:rdf.*//theory:")
59 (* global maps, shared by all threads *)
62 lazy (new Http_getter_map.map (Lazy.force Http_getter_env.cic_dbm))
64 lazy (new Http_getter_map.map (Lazy.force Http_getter_env.nuprl_dbm))
66 lazy (new Http_getter_map.map (Lazy.force Http_getter_env.rdf_dbm))
68 lazy (new Http_getter_map.map (Lazy.force Http_getter_env.xsl_dbm))
70 let uri_tree = ref None
73 | None -> assert false
75 let is_prefetch_on () =
76 match !uri_tree with None -> false | Some _ -> true
79 let path = Lazy.force Http_getter_env.dump_file in
80 Tree.save_to_disk path (deref_if_some uri_tree);
81 Http_getter_md5.create_hash [
82 (Lazy.force Http_getter_env.cic_dbm_real);
86 if not (Http_getter_md5.check_hash ()) then
89 uri_tree := Some (Tree.load_from_disk
90 (Lazy.force Http_getter_env.dump_file))
92 let sync_with_map () =
93 if not (Http_getter_md5.check_hash ()) then begin
94 let tree = ref (Some Tree.empty_tree) in
95 Http_getter_logger.log "Updating cic map dump...";
96 let t = Unix.time () in
97 (Lazy.force cic_map)#iter
99 tree := Some (Tree.add_uri k (deref_if_some tree)));
101 Http_getter_logger.log
102 (sprintf "done in %.0f sec" (Unix.time () -. t));
105 Http_getter_logger.log "Cic map dump is up to date!";
106 load_tree () (* XXX TASSI: race condition here *)
109 let maps = [ cic_map; nuprl_map; rdf_map; xsl_map ]
110 let close_maps () = List.iter (fun m -> (Lazy.force m) # close) maps
111 let clear_maps () = List.iter (fun m -> (Lazy.force m) # clear) maps
113 List.iter (fun m -> (Lazy.force m) # sync) maps;
116 let map_of_uri = function
117 | uri when is_cic_uri uri -> Lazy.force cic_map
118 | uri when is_nuprl_uri uri -> Lazy.force nuprl_map
119 | uri when is_rdf_uri uri -> Lazy.force rdf_map
120 | uri when is_xsl_uri uri -> Lazy.force xsl_map
121 | uri -> raise (Unresolvable_URI uri)
123 let update_from_server logger server_url = (* use global maps *)
124 Http_getter_logger.log ("Updating information from " ^ server_url);
125 let xml_url_of_uri = function
126 (* TODO missing sanity checks on server_url, e.g. it can contains $1 *)
127 | uri when (Pcre.pmatch ~rex:heading_cic_RE uri) ->
128 Pcre.replace ~rex:heading_cic_RE ~templ:server_url uri
129 | uri when (Pcre.pmatch ~rex:heading_theory_RE uri) ->
130 Pcre.replace ~rex:heading_theory_RE ~templ:server_url uri
131 | uri when (Pcre.pmatch ~rex:heading_nuprl_RE uri) ->
132 Pcre.replace ~rex:heading_nuprl_RE ~templ:server_url uri
133 | uri -> raise (Invalid_URI uri)
135 let rdf_url_of_uri = function (* TODO as above *)
136 | uri when (Pcre.pmatch ~rex:heading_rdf_cic_RE uri) ->
137 Pcre.replace ~rex:heading_rdf_cic_RE ~templ:server_url uri
138 | uri when (Pcre.pmatch ~rex:heading_rdf_theory_RE uri) ->
139 Pcre.replace ~rex:heading_rdf_theory_RE ~templ:server_url uri
140 | uri -> raise (Invalid_URI uri)
142 logger (`T ("Processing server: " ^ server_url));
144 let (xml_index, rdf_index, xsl_index) =
145 (* TODO keeps index in memory, is better to keep them on temp files? *)
146 (http_get (server_url ^ "/" ^ (Lazy.force Http_getter_env.xml_index)),
147 http_get (server_url ^ "/" ^ (Lazy.force Http_getter_env.rdf_index)),
148 http_get (server_url ^ "/" ^ (Lazy.force Http_getter_env.xsl_index)))
150 if (xml_index = None && rdf_index = None && xsl_index = None) then
151 Http_getter_logger.log (sprintf "Warning: useless server %s" server_url);
152 (match xml_index with
154 logger (`T "- Updating XML db ...");
158 | l when is_blank_line l -> () (* skip blank and commented lines *)
161 (match Pcre.split ~rex:index_line_sep_RE l with
163 assert (is_cic_uri uri || is_nuprl_uri uri) ;
164 (map_of_uri uri)#replace
165 uri ((xml_url_of_uri uri) ^ ".xml.gz")
167 assert (is_cic_uri uri || is_nuprl_uri uri) ;
168 (map_of_uri uri)#replace
169 uri ((xml_url_of_uri uri) ^ ".xml")
171 logger (`T ("Ignoring invalid line: '" ^ l));
173 with Invalid_URI uri ->
174 logger (`T ("Ignoring invalid XML URI: '" ^ l));
176 (Pcre.split ~rex:index_sep_RE xml_index); (* xml_index lines *)
177 logger (`T "All done");
180 (match rdf_index with
182 logger (`T "- Updating RDF db ...");
187 (match Pcre.split ~rex:index_line_sep_RE l with
189 (Lazy.force rdf_map) # replace uri
190 ((rdf_url_of_uri uri) ^ ".xml.gz")
192 (Lazy.force rdf_map) # replace uri
193 ((rdf_url_of_uri uri) ^ ".xml")
195 logger (`T ("Ignoring invalid line: '" ^ l));
197 with Invalid_URI uri ->
198 logger (`T ("Ignoring invalid RDF URI: '" ^ l));
200 (Pcre.split ~rex:index_sep_RE rdf_index); (* rdf_index lines *)
201 logger (`T "All done");
204 (match xsl_index with
206 logger (`T "- Updating XSLT db ...");
209 (fun l -> (Lazy.force xsl_map) # replace l (server_url ^ "/" ^ l))
210 (Pcre.split ~rex:index_sep_RE xsl_index);
211 logger (`T "All done");
214 Http_getter_logger.log "done with this server"
216 let update_from_all_servers logger () = (* use global maps *)
219 (update_from_server logger)
220 (* reverse order: 1st server is the most important one *)
221 (List.map snd (List.rev (Http_getter_env.servers ())));
224 let update_from_one_server ?(logger = fun _ -> ()) server_url =
225 update_from_server logger server_url
227 let temp_file_of_uri uri =
228 let flat_string s s' c =
229 let cs = String.copy s in
230 for i = 0 to (String.length s) - 1 do
231 if String.contains s' s.[i] then cs.[i] <- c
235 let user = try Unix.getlogin () with _ -> "" in
236 Filename.open_temp_file (user ^ flat_string uri ".-=:;!?/&" '_') ""
238 (* should we use a remote getter or not *)
241 Helm_registry.get "getter.mode" = "remote"
242 with Helm_registry.Key_not_found _ -> false
243 let getter_url () = Helm_registry.get "getter.url"
245 (* Remote interface: getter methods implemented using a remote getter *)
248 let getxslt_remote ~patch_dtd uri = not_implemented "getxslt_remote"
249 let getdtd_remote ~patch_dtd uri = not_implemented "getdtd_remote"
250 let clean_cache_remote () = not_implemented "clean_cache_remote"
251 let list_servers_remote () = not_implemented "list_servers_remote"
252 let add_server_remote ~logger ~position name =
253 not_implemented "add_server_remote"
254 let remove_server_remote ~logger position =
255 not_implemented "remove_server_remote"
256 let getalluris_remote () = not_implemented "getalluris_remote"
257 let getallrdfuris_remote () = not_implemented "getallrdfuris_remote"
258 let ls_remote lsuri = not_implemented "ls_remote"
261 let resolve_remote uri =
262 (* deliver resolve request to http_getter *)
263 let doc = ClientHTTP.get (sprintf "%sresolve?uri=%s" (getter_url ()) uri) in
264 let res = ref Unknown in
265 Pxp_ev_parser.process_entity PxpHelmConf.pxp_config (`Entry_content [])
266 (Pxp_ev_parser.create_entity_manager ~is_document:true
267 PxpHelmConf.pxp_config (Pxp_yacc.from_string doc))
269 | Pxp_types.E_start_tag ("url",["value",url],_,_) -> res := Resolved url
270 | Pxp_types.E_start_tag ("unresolvable",[],_,_) ->
271 res := Exception (Unresolvable_URI uri)
272 | Pxp_types.E_start_tag ("not_found",[],_,_) ->
273 res := Exception (Key_not_found uri)
274 | Pxp_types.E_start_tag (x,_,_,_) ->
275 res := Exception UnexpectedGetterOutput
278 | Unknown -> raise UnexpectedGetterOutput
279 | Exception e -> raise e
280 | Resolved url -> url
282 let register_remote ~uri ~url =
283 ClientHTTP.send (sprintf "%sregister?uri=%s&url=%s" (getter_url ()) uri url)
285 let unregister_remote uri =
286 ClientHTTP.send (sprintf "%sunregister?uri=%s" (getter_url ()) uri)
288 let update_remote logger () =
289 let answer = ClientHTTP.get (getter_url () ^ "update") in
293 let getxml_remote ~format ~patch_dtd uri =
295 sprintf "%sgetxml?uri=%s&format=%s&patch_dtd=%s"
297 (match format with `Normal -> "normal" | `Gzipped -> "gz")
298 (match patch_dtd with true -> "yes" | false -> "no")
300 ClientHTTP.get_and_save_to_tmp uri
304 let help () = Http_getter_const.usage_string (Http_getter_env.env_to_string ())
310 (map_of_uri uri)#resolve uri
312 let register ~uri ~url =
314 register_remote ~uri ~url
317 (map_of_uri uri)#add uri url;
318 if is_prefetch_on () then
319 uri_tree := Some (Tree.add_uri uri (deref_if_some uri_tree))
324 unregister_remote uri
328 (map_of_uri uri)#remove uri;
329 if is_prefetch_on () then
330 uri_tree := Some (Tree.remove_uri uri (deref_if_some uri_tree))
332 with Key_not_found _ -> ()
334 let update ?(logger = fun _ -> ()) () =
336 update_remote logger ()
338 update_from_all_servers logger ()
340 let getxml ?(format = `Gzipped) ?(patch_dtd = false) uri =
342 getxml_remote ~format ~patch_dtd uri
344 Http_getter_logger.log ~level:2 ("getxml: " ^ uri);
345 let url = resolve uri in
346 Http_getter_logger.log ~level:2 ("resolved_uri: " ^ url) ;
347 let (fname, outchan) = temp_file_of_uri uri in
348 Http_getter_cache.respond_xml ~via_http:false ~enc:format ~patch:patch_dtd
354 let getxslt ?(patch_dtd = true) uri =
356 getxslt_remote ~patch_dtd uri
359 let url = resolve uri in
360 let (fname, outchan) = temp_file_of_uri uri in
361 Http_getter_cache.respond_xsl ~via_http:false ~url ~patch:patch_dtd outchan;
366 let getdtd ?(patch_dtd = true) uri =
368 getdtd_remote ~patch_dtd uri
370 let url = Lazy.force Http_getter_env.dtd_dir ^ "/" ^ uri in
371 let (fname, outchan) = temp_file_of_uri uri in
372 Http_getter_cache.respond_dtd ~via_http:false ~url ~patch:patch_dtd outchan;
379 clean_cache_remote ()
381 Http_getter_cache.clean ()
383 let list_servers () =
385 list_servers_remote ()
387 Http_getter_env.servers ()
389 let add_server ?(logger = fun _ -> ()) ?(position = 0) name =
391 add_server_remote ~logger ~position name
393 if position = 0 then begin
394 Http_getter_env.add_server ~position:0 name;
395 update_from_one_server ~logger name (* quick update (new server only) *)
396 end else if position > 0 then begin
397 Http_getter_env.add_server ~position name;
399 end else (* already checked by parse_position *)
403 let has_server position = List.mem_assoc position (Http_getter_env.servers ())
405 let remove_server ?(logger = fun _ -> ()) position =
407 remove_server_remote ~logger ()
411 List.assoc position (Http_getter_env.servers ())
413 raise (Invalid_argument (sprintf "no server with position %d" position))
415 Http_getter_env.remove_server position;
419 let return_uris map filter =
421 map#iter (fun uri _ -> if filter uri then uris := uri :: !uris);
429 (Pcre.pmatch ~rex:heading_cic_RE uri)
430 (* && not (Pcre.pmatch ~rex:trailing_types_RE uri) *)
432 return_uris (Lazy.force cic_map) filter
434 let getallrdfuris classs =
436 getallrdfuris_remote ()
439 let base = "^helm:rdf:www\\.cs\\.unibo\\.it/helm/rdf/" in
441 | `Forward -> (fun uri -> Pcre.pmatch ~pat:(base ^ "forward") uri)
442 | `Backward -> (fun uri -> Pcre.pmatch ~pat:(base ^ "backward") uri)
444 return_uris (Lazy.force rdf_map) filter
447 let (++) (oldann, oldtypes, oldbody, oldtree)
448 (newann, newtypes, newbody, newtree) =
449 ((if newann > oldann then newann else oldann),
450 (if newtypes > oldtypes then newtypes else oldtypes),
451 (if newbody > oldbody then newbody else oldbody),
452 (if newtree > oldtree then newtree else oldtree))
454 let (types_RE, types_ann_RE, body_RE, body_ann_RE,
455 proof_tree_RE, proof_tree_ann_RE, trailing_slash_RE, theory_RE) =
456 (Pcre.regexp "\\.types$", Pcre.regexp "\\.types\\.ann$",
457 Pcre.regexp "\\.body$", Pcre.regexp "\\.body\\.ann$",
458 Pcre.regexp "\\.proof_tree$", Pcre.regexp "\\.proof_tree\\.ann$",
459 Pcre.regexp "/$", Pcre.regexp "\\.theory$")
463 "^([^.]*\\.[^.]*)((\\.body)|(\\.proof_tree)|(\\.types))?(\\.ann)?$"
465 let (slash_RE, til_slash_RE, no_slashes_RE) =
466 (Pcre.regexp "/", Pcre.regexp "^.*/", Pcre.regexp "^[^/]*$")
467 let fix_regexp_RE = Pcre.regexp ("^" ^ (Pcre.quote "(cic|theory)"))
473 let looking_for_dir = Pcre.pmatch ~rex:trailing_slash_RE regexp in
474 let pat = Pcre.replace ~rex:trailing_slash_RE ("^" ^ regexp) in
475 let (dir_RE, dir_local_RE, obj_RE, first_comp_RE) =
476 Pcre.regexp (pat ^ "/"), Pcre.regexp "[^/]+/[^/]*",
477 Pcre.regexp (pat ^ "(\\.|$)"), Pcre.regexp "/.*"
479 let exists_theory regexp =
481 Pcre.replace ~rex:fix_regexp_RE ~templ:"theory" regexp ^ "index.theory"
484 ignore (resolve theory);
486 with Key_not_found _ -> false
488 let toplevel_theory =
489 match List.rev (Pcre.split ~rex:slash_RE pat) with
490 | dir :: _ -> Some (dir ^ ".theory")
493 let dirs = ref StringSet.empty in
494 let objs = Hashtbl.create 17 in
496 dirs := StringSet.add (List.hd (Pcre.split ~rex:slash_RE d)) !dirs
499 let basepart = Pcre.replace ~rex:basepart_RE ~templ:"$1" o in
500 let no_flags = false, No, No, No in
503 Hashtbl.find objs basepart
504 with Not_found -> (* no ann, no types, no body, no proof tree *)
509 | s when Pcre.pmatch ~rex:types_RE s -> (false, Yes, No, No)
510 | s when Pcre.pmatch ~rex:types_ann_RE s -> (true, Ann, No, No)
511 | s when Pcre.pmatch ~rex:body_RE s -> (false, No, Yes, No)
512 | s when Pcre.pmatch ~rex:body_ann_RE s -> (true, No, Ann, No)
513 | s when Pcre.pmatch ~rex:proof_tree_RE s -> (false, No, No, Yes)
514 | s when Pcre.pmatch ~rex:proof_tree_ann_RE s -> (true, No, No, Ann)
517 Hashtbl.replace objs basepart (oldflags ++ newflags)
519 (* Variables used in backward compatibility code to map
520 theory:/path/t.theory into theory:/path/t/index.theory
521 when cic:/path/t/ exists *)
522 let the_candidate_for_remapping =
523 (* CSC: Here I am making a strong assumption: the pattern
524 can be only of the form [^:]*:/path where path is
525 NOT a regular expression *)
526 "theory:" ^ Pcre.replace ~rex:(Pcre.regexp "[^:]*:") pat
528 let index_not_generated_yet = ref true in
529 let valid_candidates = ref [] in
530 let candidates_found = ref [] in
532 (*(Lazy.force cic_map) # iter*)
534 (* depending on prefetch *)
535 let if_prefetch if_is if_not =
536 if is_prefetch_on() then if_is else if_not
539 let iter_on_right_thing = if_prefetch
540 (fun f -> List.iter (fun k -> f k "")
541 (Tree.ls_path regexp (deref_if_some uri_tree)))
542 (fun f -> (Lazy.force cic_map) # iter f)
544 let calculate_localpart = if_prefetch
546 (fun uri -> Pcre.replace ~rex:dir_RE uri)
548 let check_if_x_RE = if_prefetch
549 (fun x_RE uri -> true)
550 (fun x_RE uri -> Pcre.pmatch ~rex:x_RE uri)
552 let check_if_dir_RE = check_if_x_RE dir_RE in
553 let check_if_obj_RE = check_if_x_RE obj_RE in
557 (* we work in two ways:
558 * 1 iteration on the whole map
561 * Since in the first case 'key' is a complete uri, while
562 * in the second case it is only the subtree rooted in the
563 * query regex, we must relay only on the localpath.
566 * query::= cic:/aaa/bbb/
568 * answer1 ::= the whole map
570 * aswer2 ::= [ "ccc/"; "c1.something"] where
571 * cic:/aaa/bbb/ccc/ and cic:/aaa/bbb/c1.something
572 * are the (partials) uri that matched query
574 * after the localpath extracion we have more uris in the first case,
575 * but at least the are all rooted in the same node.
577 * the Tree.get_frontier may be changed to return the same stuff as
578 * the map iteration+localpath extraction does, but I hope it is not
582 | uri when looking_for_dir && check_if_dir_RE uri ->
584 let localpart = calculate_localpart uri in
585 if Pcre.pmatch ~rex:no_slashes_RE localpart then
587 (* Backward compatibility code to map
588 theory:/path/t.theory into theory:/path/t/index.theory
589 when cic:/path/t/ exists *)
590 if Pcre.pmatch ~rex:theory_RE localpart then
591 candidates_found := localpart :: !candidates_found
597 store_dir localpart ;
598 if Pcre.pmatch localpart ~rex:dir_local_RE then
601 Pcre.replace ~rex:first_comp_RE localpart ^ ".theory"
603 if not (List.mem valid !valid_candidates) then
604 valid_candidates := valid::!valid_candidates
607 | uri when (not looking_for_dir) && check_if_obj_RE uri ->
609 store_obj (Pcre.replace ~rex:til_slash_RE uri)
613 if !index_not_generated_yet &&
614 Pcre.pmatch ~rex:orig_theory_RE uri
616 (index_not_generated_yet := false ;
617 store_obj "index.theory"));
619 if exists_theory regexp then store_obj "index.theory";
622 if not (List.mem localpart !valid_candidates) then
624 ) !candidates_found ;
625 let ls_items = ref [] in
626 StringSet.iter (fun dir -> ls_items := Ls_section dir :: !ls_items) !dirs;
627 Http_getter_misc.hashtbl_sorted_iter
628 (fun uri (annflag, typesflag, bodyflag, treeflag) ->
631 uri = uri; ann = annflag;
632 types = typesflag; body = bodyflag; proof_tree = treeflag
638 (* Shorthands from now on *)
640 let getxml' uri = getxml (UriManager.string_of_uri uri)
641 let resolve' uri = resolve (UriManager.string_of_uri uri)
642 let register' uri url = register ~uri:(UriManager.string_of_uri uri) ~url
643 let unregister' uri = unregister (UriManager.string_of_uri uri)
645 let sync_dump_file () =
646 if is_prefetch_on () then
650 Http_getter_logger.set_log_level
651 (Helm_registry.get_opt_default Helm_registry.get_int 1 "getter.log_level");
652 Http_getter_logger.set_log_file
653 (Helm_registry.get_opt Helm_registry.get_string "getter.log_file");
654 Http_getter_env.reload ();
655 let is_prefetch_set =
656 Helm_registry.get_opt_default Helm_registry.get_bool false "getter.prefetch"
658 if is_prefetch_set then
659 (* ignore (Thread.create sync_with_map ()) *)