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
43 type logger_callback = HelmLogger.html_tag -> unit
45 let stdout_logger tag = print_string (HelmLogger.string_of_html_tag tag)
47 let not_implemented s = raise (Not_implemented ("Http_getter." ^ s))
49 let index_line_sep_RE = Pcre.regexp "[ \t]+"
50 let index_sep_RE = Pcre.regexp "\r\n|\r|\n"
51 let trailing_types_RE = Pcre.regexp "\\.types$"
52 let heading_cic_RE = Pcre.regexp "^cic:"
53 let heading_theory_RE = Pcre.regexp "^theory:"
54 let heading_nuprl_RE = Pcre.regexp "^nuprl:"
55 let heading_rdf_cic_RE = Pcre.regexp "^helm:rdf.*//cic:"
56 let heading_rdf_theory_RE = Pcre.regexp "^helm:rdf.*//theory:"
58 (* global maps, shared by all threads *)
61 lazy (new Http_getter_map.map (Lazy.force Http_getter_env.cic_dbm))
63 lazy (new Http_getter_map.map (Lazy.force Http_getter_env.nuprl_dbm))
65 lazy (new Http_getter_map.map (Lazy.force Http_getter_env.rdf_dbm))
67 lazy (new Http_getter_map.map (Lazy.force Http_getter_env.xsl_dbm))
69 let uri_tree = ref None
72 | None -> assert false
74 let is_prefetch_on () =
75 match !uri_tree with None -> false | Some _ -> true
78 let path = Lazy.force Http_getter_env.dump_file in
79 Tree.save_to_disk path (deref_if_some uri_tree);
80 Http_getter_md5.create_hash [
81 (Lazy.force Http_getter_env.cic_dbm_real);
85 if not (Http_getter_md5.check_hash ()) then
88 uri_tree := Some (Tree.load_from_disk
89 (Lazy.force Http_getter_env.dump_file))
91 let sync_with_map () =
92 if not (Http_getter_md5.check_hash ()) then begin
93 let tree = ref (Some Tree.empty_tree) in
94 Http_getter_logger.log "Updating cic map dump...";
95 let t = Unix.time () in
96 (Lazy.force cic_map)#iter
98 tree := Some (Tree.add_uri k (deref_if_some tree)));
100 Http_getter_logger.log
101 (sprintf "done in %.0f sec" (Unix.time () -. t));
104 Http_getter_logger.log "Cic map dump is up to date!";
105 load_tree () (* XXX TASSI: race condition here *)
108 let maps = [ cic_map; nuprl_map; rdf_map; xsl_map ]
109 let close_maps () = List.iter (fun m -> (Lazy.force m) # close) maps
110 let clear_maps () = List.iter (fun m -> (Lazy.force m) # clear) maps
112 List.iter (fun m -> (Lazy.force m) # sync) maps;
115 let map_of_uri = function
116 | uri when is_cic_uri uri -> Lazy.force cic_map
117 | uri when is_nuprl_uri uri -> Lazy.force nuprl_map
118 | uri when is_rdf_uri uri -> Lazy.force rdf_map
119 | uri when is_xsl_uri uri -> Lazy.force xsl_map
120 | uri -> raise (Unresolvable_URI uri)
122 let update_from_server logger server_url = (* use global maps *)
123 Http_getter_logger.log ("Updating information from " ^ server_url);
124 let xml_url_of_uri = function
125 (* TODO missing sanity checks on server_url, e.g. it can contains $1 *)
126 | uri when (Pcre.pmatch ~rex:heading_cic_RE uri) ->
127 Pcre.replace ~rex:heading_cic_RE ~templ:server_url uri
128 | uri when (Pcre.pmatch ~rex:heading_theory_RE uri) ->
129 Pcre.replace ~rex:heading_theory_RE ~templ:server_url uri
130 | uri when (Pcre.pmatch ~rex:heading_nuprl_RE uri) ->
131 Pcre.replace ~rex:heading_nuprl_RE ~templ:server_url uri
132 | uri -> raise (Invalid_URI uri)
134 let rdf_url_of_uri = function (* TODO as above *)
135 | uri when (Pcre.pmatch ~rex:heading_rdf_cic_RE uri) ->
136 Pcre.replace ~rex:heading_rdf_cic_RE ~templ:server_url uri
137 | uri when (Pcre.pmatch ~rex:heading_rdf_theory_RE uri) ->
138 Pcre.replace ~rex:heading_rdf_theory_RE ~templ:server_url uri
139 | uri -> raise (Invalid_URI uri)
141 logger (`T ("Processing server: " ^ server_url));
143 let (xml_index, rdf_index, xsl_index) =
144 (* TODO keeps index in memory, is better to keep them on temp files? *)
145 (http_get (server_url ^ "/" ^ (Lazy.force Http_getter_env.xml_index)),
146 http_get (server_url ^ "/" ^ (Lazy.force Http_getter_env.rdf_index)),
147 http_get (server_url ^ "/" ^ (Lazy.force Http_getter_env.xsl_index)))
149 if (xml_index = None && rdf_index = None && xsl_index = None) then
150 Http_getter_logger.log (sprintf "Warning: useless server %s" server_url);
151 (match xml_index with
153 logger (`T "- Updating XML db ...");
157 | l when is_blank_line l -> () (* skip blank and commented lines *)
160 (match Pcre.split ~rex:index_line_sep_RE l with
162 assert (is_cic_uri uri || is_nuprl_uri uri) ;
163 (map_of_uri uri)#replace
164 uri ((xml_url_of_uri uri) ^ ".xml.gz")
166 assert (is_cic_uri uri || is_nuprl_uri uri) ;
167 (map_of_uri uri)#replace
168 uri ((xml_url_of_uri uri) ^ ".xml")
170 logger (`T ("Ignoring invalid line: '" ^ l));
172 with Invalid_URI uri ->
173 logger (`T ("Ignoring invalid XML URI: '" ^ l));
175 (Pcre.split ~rex:index_sep_RE xml_index); (* xml_index lines *)
176 logger (`T "All done");
179 (match rdf_index with
181 logger (`T "- Updating RDF db ...");
186 (match Pcre.split ~rex:index_line_sep_RE l with
188 (Lazy.force rdf_map) # replace uri
189 ((rdf_url_of_uri uri) ^ ".xml.gz")
191 (Lazy.force rdf_map) # replace uri
192 ((rdf_url_of_uri uri) ^ ".xml")
194 logger (`T ("Ignoring invalid line: '" ^ l));
196 with Invalid_URI uri ->
197 logger (`T ("Ignoring invalid RDF URI: '" ^ l));
199 (Pcre.split ~rex:index_sep_RE rdf_index); (* rdf_index lines *)
200 logger (`T "All done");
203 (match xsl_index with
205 logger (`T "- Updating XSLT db ...");
208 (fun l -> (Lazy.force xsl_map) # replace l (server_url ^ "/" ^ l))
209 (Pcre.split ~rex:index_sep_RE xsl_index);
210 logger (`T "All done");
213 Http_getter_logger.log "done with this server"
215 let update_from_all_servers logger () = (* use global maps *)
218 (update_from_server logger)
219 (* reverse order: 1st server is the most important one *)
220 (List.map snd (List.rev (Http_getter_env.servers ())));
223 let update_from_one_server ?(logger = fun _ -> ()) server_url =
224 update_from_server logger server_url
226 let temp_file_of_uri uri =
227 let flat_string s s' c =
228 let cs = String.copy s in
229 for i = 0 to (String.length s) - 1 do
230 if String.contains s' s.[i] then cs.[i] <- c
234 let user = try Unix.getlogin () with _ -> "" in
235 Filename.open_temp_file (user ^ flat_string uri ".-=:;!?/&" '_') ""
237 (* should we use a remote getter or not *)
240 Helm_registry.get "getter.mode" = "remote"
241 with Helm_registry.Key_not_found _ -> false
242 let getter_url () = Helm_registry.get "getter.url"
244 (* Remote interface: getter methods implemented using a remote getter *)
247 let getxslt_remote ~patch_dtd uri = not_implemented "getxslt_remote"
248 let getdtd_remote ~patch_dtd uri = not_implemented "getdtd_remote"
249 let clean_cache_remote () = not_implemented "clean_cache_remote"
250 let list_servers_remote () = not_implemented "list_servers_remote"
251 let add_server_remote ~logger ~position name =
252 not_implemented "add_server_remote"
253 let remove_server_remote ~logger position =
254 not_implemented "remove_server_remote"
255 let getalluris_remote () = not_implemented "getalluris_remote"
256 let getallrdfuris_remote () = not_implemented "getallrdfuris_remote"
257 let ls_remote lsuri = not_implemented "ls_remote"
260 let resolve_remote uri =
261 (* deliver resolve request to http_getter *)
263 Http_getter_wget.get (sprintf "%sresolve?uri=%s" (getter_url ()) uri)
265 let res = ref Unknown in
266 let start_element tag attrs =
270 res := Resolved (List.assoc "value" attrs)
271 with Not_found -> ())
272 | "unresolvable" -> res := Exception (Unresolvable_URI uri)
273 | "not_found" -> res := Exception (Key_not_found uri)
277 XmlPushParser.default_callbacks with
278 XmlPushParser.start_element = Some start_element
280 let xml_parser = XmlPushParser.create_parser callbacks in
281 XmlPushParser.parse xml_parser (`String doc);
282 XmlPushParser.final xml_parser;
284 | Unknown -> raise UnexpectedGetterOutput
285 | Exception e -> raise e
286 | Resolved url -> url
288 let register_remote ~uri ~url =
289 Http_getter_wget.send
290 (sprintf "%sregister?uri=%s&url=%s" (getter_url ()) uri url)
292 let unregister_remote uri =
293 Http_getter_wget.send (sprintf "%sunregister?uri=%s" (getter_url ()) uri)
295 let update_remote logger () =
296 let answer = Http_getter_wget.get (getter_url () ^ "update") in
300 let getxml_remote ~format ~patch_dtd uri =
302 sprintf "%sgetxml?uri=%s&format=%s&patch_dtd=%s"
304 (match format with `Normal -> "normal" | `Gzipped -> "gz")
305 (match patch_dtd with true -> "yes" | false -> "no")
307 Http_getter_wget.get_and_save_to_tmp uri
311 let help () = Http_getter_const.usage_string (Http_getter_env.env_to_string ())
317 (map_of_uri uri)#resolve uri
319 let register ~uri ~url =
321 register_remote ~uri ~url
324 (map_of_uri uri)#add uri url;
325 if is_prefetch_on () then
326 uri_tree := Some (Tree.add_uri uri (deref_if_some uri_tree))
331 unregister_remote uri
335 (map_of_uri uri)#remove uri;
336 if is_prefetch_on () then
337 uri_tree := Some (Tree.remove_uri uri (deref_if_some uri_tree))
339 with Key_not_found _ -> ()
341 let update ?(logger = fun _ -> ()) () =
343 update_remote logger ()
345 update_from_all_servers logger ()
347 let getxml ?(format = `Gzipped) ?(patch_dtd = false) uri =
349 getxml_remote ~format ~patch_dtd uri
351 Http_getter_logger.log ~level:2 ("getxml: " ^ uri);
352 let url = resolve uri in
353 Http_getter_logger.log ~level:2 ("resolved_uri: " ^ url) ;
354 let (fname, outchan) = temp_file_of_uri uri in
355 Http_getter_cache.respond_xml ~via_http:false ~enc:format ~patch:patch_dtd
361 let getxslt ?(patch_dtd = true) uri =
363 getxslt_remote ~patch_dtd uri
366 let url = resolve uri in
367 let (fname, outchan) = temp_file_of_uri uri in
368 Http_getter_cache.respond_xsl ~via_http:false ~url ~patch:patch_dtd outchan;
373 let getdtd ?(patch_dtd = true) uri =
375 getdtd_remote ~patch_dtd uri
377 let url = Lazy.force Http_getter_env.dtd_dir ^ "/" ^ uri in
378 let (fname, outchan) = temp_file_of_uri uri in
379 Http_getter_cache.respond_dtd ~via_http:false ~url ~patch:patch_dtd outchan;
386 clean_cache_remote ()
388 Http_getter_cache.clean ()
390 let list_servers () =
392 list_servers_remote ()
394 Http_getter_env.servers ()
396 let add_server ?(logger = fun _ -> ()) ?(position = 0) name =
398 add_server_remote ~logger ~position name
400 if position = 0 then begin
401 Http_getter_env.add_server ~position:0 name;
402 update_from_one_server ~logger name (* quick update (new server only) *)
403 end else if position > 0 then begin
404 Http_getter_env.add_server ~position name;
406 end else (* already checked by parse_position *)
410 let has_server position = List.mem_assoc position (Http_getter_env.servers ())
412 let remove_server ?(logger = fun _ -> ()) position =
414 remove_server_remote ~logger ()
418 List.assoc position (Http_getter_env.servers ())
420 raise (Invalid_argument (sprintf "no server with position %d" position))
422 Http_getter_env.remove_server position;
426 let return_uris map filter =
428 map#iter (fun uri _ -> if filter uri then uris := uri :: !uris);
436 (Pcre.pmatch ~rex:heading_cic_RE uri)
437 (* && not (Pcre.pmatch ~rex:trailing_types_RE uri) *)
439 return_uris (Lazy.force cic_map) filter
441 let getallrdfuris classs =
443 getallrdfuris_remote ()
446 let base = "^helm:rdf:www\\.cs\\.unibo\\.it/helm/rdf/" in
448 | `Forward -> (fun uri -> Pcre.pmatch ~pat:(base ^ "forward") uri)
449 | `Backward -> (fun uri -> Pcre.pmatch ~pat:(base ^ "backward") uri)
451 return_uris (Lazy.force rdf_map) filter
454 let (++) (oldann, oldtypes, oldbody, oldtree)
455 (newann, newtypes, newbody, newtree) =
456 ((if newann > oldann then newann else oldann),
457 (if newtypes > oldtypes then newtypes else oldtypes),
458 (if newbody > oldbody then newbody else oldbody),
459 (if newtree > oldtree then newtree else oldtree))
461 let (types_RE, types_ann_RE, body_RE, body_ann_RE,
462 proof_tree_RE, proof_tree_ann_RE, trailing_slash_RE, theory_RE) =
463 (Pcre.regexp "\\.types$", Pcre.regexp "\\.types\\.ann$",
464 Pcre.regexp "\\.body$", Pcre.regexp "\\.body\\.ann$",
465 Pcre.regexp "\\.proof_tree$", Pcre.regexp "\\.proof_tree\\.ann$",
466 Pcre.regexp "/$", Pcre.regexp "\\.theory$")
470 "^([^.]*\\.[^.]*)((\\.body)|(\\.proof_tree)|(\\.types))?(\\.ann)?$"
472 let (slash_RE, til_slash_RE, no_slashes_RE) =
473 (Pcre.regexp "/", Pcre.regexp "^.*/", Pcre.regexp "^[^/]*$")
474 let fix_regexp_RE = Pcre.regexp ("^" ^ (Pcre.quote "(cic|theory)"))
480 let looking_for_dir = Pcre.pmatch ~rex:trailing_slash_RE regexp in
481 let pat = Pcre.replace ~rex:trailing_slash_RE ("^" ^ regexp) in
482 let (dir_RE, dir_local_RE, obj_RE, first_comp_RE) =
483 Pcre.regexp (pat ^ "/"), Pcre.regexp "[^/]+/[^/]*",
484 Pcre.regexp (pat ^ "(\\.|$)"), Pcre.regexp "/.*"
486 let exists_theory regexp =
488 Pcre.replace ~rex:fix_regexp_RE ~templ:"theory" regexp ^ "index.theory"
491 ignore (resolve theory);
493 with Key_not_found _ -> false
495 let toplevel_theory =
496 match List.rev (Pcre.split ~rex:slash_RE pat) with
497 | dir :: _ -> Some (dir ^ ".theory")
500 let dirs = ref StringSet.empty in
501 let objs = Hashtbl.create 17 in
503 dirs := StringSet.add (List.hd (Pcre.split ~rex:slash_RE d)) !dirs
506 let basepart = Pcre.replace ~rex:basepart_RE ~templ:"$1" o in
507 let no_flags = false, No, No, No in
510 Hashtbl.find objs basepart
511 with Not_found -> (* no ann, no types, no body, no proof tree *)
516 | s when Pcre.pmatch ~rex:types_RE s -> (false, Yes, No, No)
517 | s when Pcre.pmatch ~rex:types_ann_RE s -> (true, Ann, No, No)
518 | s when Pcre.pmatch ~rex:body_RE s -> (false, No, Yes, No)
519 | s when Pcre.pmatch ~rex:body_ann_RE s -> (true, No, Ann, No)
520 | s when Pcre.pmatch ~rex:proof_tree_RE s -> (false, No, No, Yes)
521 | s when Pcre.pmatch ~rex:proof_tree_ann_RE s -> (true, No, No, Ann)
524 Hashtbl.replace objs basepart (oldflags ++ newflags)
526 (* Variables used in backward compatibility code to map
527 theory:/path/t.theory into theory:/path/t/index.theory
528 when cic:/path/t/ exists *)
529 let the_candidate_for_remapping =
530 (* CSC: Here I am making a strong assumption: the pattern
531 can be only of the form [^:]*:/path where path is
532 NOT a regular expression *)
533 "theory:" ^ Pcre.replace ~rex:(Pcre.regexp "[^:]*:") pat
535 let index_not_generated_yet = ref true in
536 let valid_candidates = ref [] in
537 let candidates_found = ref [] in
539 (*(Lazy.force cic_map) # iter*)
541 (* depending on prefetch *)
542 let if_prefetch if_is if_not =
543 if is_prefetch_on() then if_is else if_not
546 let iter_on_right_thing = if_prefetch
547 (fun f -> List.iter (fun k -> f k "")
548 (Tree.ls_path regexp (deref_if_some uri_tree)))
549 (fun f -> (Lazy.force cic_map) # iter f)
551 let calculate_localpart = if_prefetch
553 (fun uri -> Pcre.replace ~rex:dir_RE uri)
555 let check_if_x_RE = if_prefetch
556 (fun x_RE uri -> true)
557 (fun x_RE uri -> Pcre.pmatch ~rex:x_RE uri)
559 let check_if_dir_RE = check_if_x_RE dir_RE in
560 let check_if_obj_RE = check_if_x_RE obj_RE in
564 (* we work in two ways:
565 * 1 iteration on the whole map
568 * Since in the first case 'key' is a complete uri, while
569 * in the second case it is only the subtree rooted in the
570 * query regex, we must relay only on the localpath.
573 * query::= cic:/aaa/bbb/
575 * answer1 ::= the whole map
577 * aswer2 ::= [ "ccc/"; "c1.something"] where
578 * cic:/aaa/bbb/ccc/ and cic:/aaa/bbb/c1.something
579 * are the (partials) uri that matched query
581 * after the localpath extracion we have more uris in the first case,
582 * but at least the are all rooted in the same node.
584 * the Tree.get_frontier may be changed to return the same stuff as
585 * the map iteration+localpath extraction does, but I hope it is not
589 | uri when looking_for_dir && check_if_dir_RE uri ->
591 let localpart = calculate_localpart uri in
592 if Pcre.pmatch ~rex:no_slashes_RE localpart then
594 (* Backward compatibility code to map
595 theory:/path/t.theory into theory:/path/t/index.theory
596 when cic:/path/t/ exists *)
597 if Pcre.pmatch ~rex:theory_RE localpart then
598 candidates_found := localpart :: !candidates_found
604 store_dir localpart ;
605 if Pcre.pmatch localpart ~rex:dir_local_RE then
608 Pcre.replace ~rex:first_comp_RE localpart ^ ".theory"
610 if not (List.mem valid !valid_candidates) then
611 valid_candidates := valid::!valid_candidates
614 | uri when (not looking_for_dir) && check_if_obj_RE uri ->
616 store_obj (Pcre.replace ~rex:til_slash_RE uri)
620 if !index_not_generated_yet &&
621 Pcre.pmatch ~rex:orig_theory_RE uri
623 (index_not_generated_yet := false ;
624 store_obj "index.theory"));
626 if exists_theory regexp then store_obj "index.theory";
629 if not (List.mem localpart !valid_candidates) then
631 ) !candidates_found ;
632 let ls_items = ref [] in
633 StringSet.iter (fun dir -> ls_items := Ls_section dir :: !ls_items) !dirs;
634 Http_getter_misc.hashtbl_sorted_iter
635 (fun uri (annflag, typesflag, bodyflag, treeflag) ->
638 uri = uri; ann = annflag;
639 types = typesflag; body = bodyflag; proof_tree = treeflag
645 (* Shorthands from now on *)
647 let getxml' uri = getxml (UriManager.string_of_uri uri)
648 let resolve' uri = resolve (UriManager.string_of_uri uri)
649 let register' uri url = register ~uri:(UriManager.string_of_uri uri) ~url
650 let unregister' uri = unregister (UriManager.string_of_uri uri)
652 let sync_dump_file () =
653 if is_prefetch_on () then
657 Http_getter_logger.set_log_level
658 (Helm_registry.get_opt_default Helm_registry.int ~default:1
660 Http_getter_logger.set_log_file
661 (Helm_registry.get_opt Helm_registry.string "getter.log_file");
662 Http_getter_env.reload ();
663 let is_prefetch_set =
664 Helm_registry.get_opt_default Helm_registry.bool ~default:false
667 if is_prefetch_set then
668 (* ignore (Thread.create sync_with_map ()) *)