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 maps = [ cic_map; nuprl_map; rdf_map; xsl_map ]
71 let close_maps () = List.iter (fun m -> (Lazy.force m) # close) maps
72 let clear_maps () = List.iter (fun m -> (Lazy.force m) # clear) maps
73 let sync_maps () = List.iter (fun m -> (Lazy.force m) # sync) maps
75 let map_of_uri = function
76 | uri when is_cic_uri uri -> Lazy.force cic_map
77 | uri when is_nuprl_uri uri -> Lazy.force nuprl_map
78 | uri when is_rdf_uri uri -> Lazy.force rdf_map
79 | uri when is_xsl_uri uri -> Lazy.force xsl_map
80 | uri -> raise (Unresolvable_URI uri)
82 let update_from_server logger server_url = (* use global maps *)
83 Http_getter_logger.log ("Updating information from " ^ server_url);
84 let xml_url_of_uri = function
85 (* TODO missing sanity checks on server_url, e.g. it can contains $1 *)
86 | uri when (Pcre.pmatch ~rex:heading_cic_RE uri) ->
87 Pcre.replace ~rex:heading_cic_RE ~templ:server_url uri
88 | uri when (Pcre.pmatch ~rex:heading_theory_RE uri) ->
89 Pcre.replace ~rex:heading_theory_RE ~templ:server_url uri
90 | uri when (Pcre.pmatch ~rex:heading_nuprl_RE uri) ->
91 Pcre.replace ~rex:heading_nuprl_RE ~templ:server_url uri
92 | uri -> raise (Invalid_URI uri)
94 let rdf_url_of_uri = function (* TODO as above *)
95 | uri when (Pcre.pmatch ~rex:heading_rdf_cic_RE uri) ->
96 Pcre.replace ~rex:heading_rdf_cic_RE ~templ:server_url uri
97 | uri when (Pcre.pmatch ~rex:heading_rdf_theory_RE uri) ->
98 Pcre.replace ~rex:heading_rdf_theory_RE ~templ:server_url uri
99 | uri -> raise (Invalid_URI uri)
101 logger (`T ("Processing server: " ^ server_url));
103 let (xml_index, rdf_index, xsl_index) =
104 (* TODO keeps index in memory, is better to keep them on temp files? *)
105 (http_get (server_url ^ "/" ^ (Lazy.force Http_getter_env.xml_index)),
106 http_get (server_url ^ "/" ^ (Lazy.force Http_getter_env.rdf_index)),
107 http_get (server_url ^ "/" ^ (Lazy.force Http_getter_env.xsl_index)))
109 if (xml_index = None && rdf_index = None && xsl_index = None) then
110 Http_getter_logger.log (sprintf "Warning: useless server %s" server_url);
111 (match xml_index with
113 logger (`T "- Updating XML db ...");
117 | l when is_blank_line l -> () (* skip blank and commented lines *)
120 (match Pcre.split ~rex:index_line_sep_RE l with
122 assert (is_cic_uri uri || is_nuprl_uri uri) ;
123 (map_of_uri uri)#replace
124 uri ((xml_url_of_uri uri) ^ ".xml.gz")
126 assert (is_cic_uri uri || is_nuprl_uri uri) ;
127 (map_of_uri uri)#replace
128 uri ((xml_url_of_uri uri) ^ ".xml")
130 logger (`T ("Ignoring invalid line: '" ^ l));
132 with Invalid_URI uri ->
133 logger (`T ("Ignoring invalid XML URI: '" ^ l));
135 (Pcre.split ~rex:index_sep_RE xml_index); (* xml_index lines *)
136 logger (`T "All done");
139 (match rdf_index with
141 logger (`T "- Updating RDF db ...");
146 (match Pcre.split ~rex:index_line_sep_RE l with
148 (Lazy.force rdf_map) # replace uri
149 ((rdf_url_of_uri uri) ^ ".xml.gz")
151 (Lazy.force rdf_map) # replace uri
152 ((rdf_url_of_uri uri) ^ ".xml")
154 logger (`T ("Ignoring invalid line: '" ^ l));
156 with Invalid_URI uri ->
157 logger (`T ("Ignoring invalid RDF URI: '" ^ l));
159 (Pcre.split ~rex:index_sep_RE rdf_index); (* rdf_index lines *)
160 logger (`T "All done");
163 (match xsl_index with
165 logger (`T "- Updating XSLT db ...");
168 (fun l -> (Lazy.force xsl_map) # replace l (server_url ^ "/" ^ l))
169 (Pcre.split ~rex:index_sep_RE xsl_index);
170 logger (`T "All done");
173 Http_getter_logger.log "done with this server"
175 let update_from_all_servers logger () = (* use global maps *)
178 (update_from_server logger)
179 (* reverse order: 1st server is the most important one *)
180 (List.map snd (List.rev (Http_getter_env.servers ())));
183 let update_from_one_server ?(logger = fun _ -> ()) server_url =
184 update_from_server logger server_url
186 let temp_file_of_uri uri =
187 let flat_string s s' c =
188 let cs = String.copy s in
189 for i = 0 to (String.length s) - 1 do
190 if String.contains s' s.[i] then cs.[i] <- c
194 let user = try Unix.getlogin () with _ -> "" in
195 Filename.open_temp_file (user ^ flat_string uri ".-=:;!?/&" '_') ""
197 (* should we use a remote getter or not *)
200 Helm_registry.get "getter.mode" = "remote"
201 with Helm_registry.Key_not_found _ -> false
202 let getter_url () = Helm_registry.get "getter.url"
204 (* Remote interface: getter methods implemented using a remote getter *)
207 let getxslt_remote ~patch_dtd uri = not_implemented "getxslt_remote"
208 let getdtd_remote ~patch_dtd uri = not_implemented "getdtd_remote"
209 let clean_cache_remote () = not_implemented "clean_cache_remote"
210 let list_servers_remote () = not_implemented "list_servers_remote"
211 let add_server_remote ~logger ~position name =
212 not_implemented "add_server_remote"
213 let remove_server_remote ~logger position =
214 not_implemented "remove_server_remote"
215 let getalluris_remote () = not_implemented "getalluris_remote"
216 let getallrdfuris_remote () = not_implemented "getallrdfuris_remote"
217 let ls_remote lsuri = not_implemented "ls_remote"
220 let resolve_remote uri =
221 (* deliver resolve request to http_getter *)
222 let doc = ClientHTTP.get (sprintf "%sresolve?uri=%s" (getter_url ()) uri) in
223 let res = ref Unknown in
224 Pxp_ev_parser.process_entity PxpHelmConf.pxp_config (`Entry_content [])
225 (Pxp_ev_parser.create_entity_manager ~is_document:true
226 PxpHelmConf.pxp_config (Pxp_yacc.from_string doc))
228 | Pxp_types.E_start_tag ("url",["value",url],_,_) -> res := Resolved url
229 | Pxp_types.E_start_tag ("unresolvable",[],_,_) ->
230 res := Exception (Unresolvable_URI uri)
231 | Pxp_types.E_start_tag ("not_found",[],_,_) ->
232 res := Exception (Key_not_found uri)
233 | Pxp_types.E_start_tag (x,_,_,_) ->
234 res := Exception UnexpectedGetterOutput
237 | Unknown -> raise UnexpectedGetterOutput
238 | Exception e -> raise e
239 | Resolved url -> url
241 let register_remote ~uri ~url =
242 ClientHTTP.send (sprintf "%sregister?uri=%s&url=%s" (getter_url ()) uri url)
244 let unregister_remote uri =
245 ClientHTTP.send (sprintf "%sunregister?uri=%s" (getter_url ()) uri)
247 let update_remote logger () =
248 let answer = ClientHTTP.get (getter_url () ^ "update") in
252 let getxml_remote ~format ~patch_dtd uri =
253 ClientHTTP.get_and_save_to_tmp
254 (sprintf "%sgetxml?uri=%s&format=%s&patch_dtd=%s"
256 (match format with `Normal -> "normal" | `Gzipped -> "gzipped")
257 (match patch_dtd with true -> "yes" | false -> "no"))
261 let help () = Http_getter_const.usage_string (Http_getter_env.env_to_string ())
269 if is_cic_uri uri && Pcre.pmatch ~pat:"\\.univ$" uri then
271 prerr_endline "!!! E' in ~tassi !!!";
272 "file:///home/tassi/mylib" ^
273 (String.sub uri 4 ((String.length uri) - 4)) ^ ".xml.gz"
276 (map_of_uri uri)#resolve uri
278 let register ~uri ~url =
280 register_remote ~uri ~url
282 (map_of_uri uri)#add uri url
286 unregister_remote uri
289 (map_of_uri uri)#remove uri
290 with Key_not_found _ -> ()
292 let update ?(logger = fun _ -> ()) () =
294 update_remote logger ()
296 update_from_all_servers logger ()
298 let getxml ?(format = `Normal) ?(patch_dtd = true) uri =
300 getxml_remote ~format ~patch_dtd uri
302 Http_getter_logger.log ~level:2 ("getxml: " ^ uri);
303 let url = resolve uri in
304 Http_getter_logger.log ~level:2 ("resolved_uri: " ^ url) ;
305 let (fname, outchan) = temp_file_of_uri uri in
306 Http_getter_cache.respond_xml ~via_http:false ~enc:format ~patch:patch_dtd
312 let getxslt ?(patch_dtd = true) uri =
314 getxslt_remote ~patch_dtd uri
317 let url = resolve uri in
318 let (fname, outchan) = temp_file_of_uri uri in
319 Http_getter_cache.respond_xsl ~via_http:false ~url ~patch:patch_dtd outchan;
324 let getdtd ?(patch_dtd = true) uri =
326 getdtd_remote ~patch_dtd uri
328 let url = Lazy.force Http_getter_env.dtd_dir ^ "/" ^ uri in
329 let (fname, outchan) = temp_file_of_uri uri in
330 Http_getter_cache.respond_dtd ~via_http:false ~url ~patch:patch_dtd outchan;
337 clean_cache_remote ()
339 Http_getter_cache.clean ()
341 let list_servers () =
343 list_servers_remote ()
345 Http_getter_env.servers ()
347 let add_server ?(logger = fun _ -> ()) ?(position = 0) name =
349 add_server_remote ~logger ~position name
351 if position = 0 then begin
352 Http_getter_env.add_server ~position:0 name;
353 update_from_one_server ~logger name (* quick update (new server only) *)
354 end else if position > 0 then begin
355 Http_getter_env.add_server ~position name;
357 end else (* already checked by parse_position *)
361 let has_server position = List.mem_assoc position (Http_getter_env.servers ())
363 let remove_server ?(logger = fun _ -> ()) position =
365 remove_server_remote ~logger ()
369 List.assoc position (Http_getter_env.servers ())
371 raise (Invalid_argument (sprintf "no server with position %d" position))
373 Http_getter_env.remove_server position;
377 let return_uris map filter =
379 map#iter (fun uri _ -> if filter uri then uris := uri :: !uris);
387 (Pcre.pmatch ~rex:heading_cic_RE uri) &&
388 not (Pcre.pmatch ~rex:trailing_types_RE uri)
390 return_uris (Lazy.force cic_map) filter
392 let getallrdfuris classs =
394 getallrdfuris_remote ()
397 let base = "^helm:rdf:www\\.cs\\.unibo\\.it/helm/rdf/" in
399 | `Forward -> (fun uri -> Pcre.pmatch ~pat:(base ^ "forward") uri)
400 | `Backward -> (fun uri -> Pcre.pmatch ~pat:(base ^ "backward") uri)
402 return_uris (Lazy.force rdf_map) filter
405 let (++) (oldann, oldtypes, oldbody, oldtree)
406 (newann, newtypes, newbody, newtree) =
407 ((if newann > oldann then newann else oldann),
408 (if newtypes > oldtypes then newtypes else oldtypes),
409 (if newbody > oldbody then newbody else oldbody),
410 (if newtree > oldtree then newtree else oldtree))
414 "^([^.]*\\.[^.]*)((\\.body)|(\\.proof_tree)|(\\.types))?(\\.ann)?$"
416 let (types_RE, types_ann_RE, body_RE, body_ann_RE,
417 proof_tree_RE, proof_tree_ann_RE, trailing_slash_RE, theory_RE) =
418 (Pcre.regexp "\\.types$", Pcre.regexp "\\.types\\.ann$",
419 Pcre.regexp "\\.body$", Pcre.regexp "\\.body\\.ann$",
420 Pcre.regexp "\\.proof_tree$", Pcre.regexp "\\.proof_tree\\.ann$",
421 Pcre.regexp "/$", Pcre.regexp "\\.theory$")
423 let (slash_RE, til_slash_RE, no_slashes_RE) =
424 (Pcre.regexp "/", Pcre.regexp "^.*/", Pcre.regexp "^[^/]*$")
430 let looking_for_dir = Pcre.pmatch ~rex:trailing_slash_RE regexp in
431 let pat = Pcre.replace ~rex:trailing_slash_RE ("^" ^ regexp) in
432 let (dir_RE, dir_RE2, obj_RE, orig_theory_RE) =
433 Pcre.regexp (pat ^ "/"), Pcre.regexp (pat ^ "/[^/]*"),
434 Pcre.regexp (pat ^ "(\\.|$)"), Pcre.regexp (pat ^ ".theory$")
436 let dirs = ref StringSet.empty in
437 let objs = Hashtbl.create 17 in
439 dirs := StringSet.add (List.hd (Pcre.split ~rex:slash_RE d)) !dirs
442 let basepart = Pcre.replace ~rex:basepart_RE ~templ:"$1" o in
443 let no_flags = false, No, No, No in
446 Hashtbl.find objs basepart
447 with Not_found -> (* no ann, no types, no body, no proof tree *)
452 | s when Pcre.pmatch ~rex:types_RE s -> (false, Yes, No, No)
453 | s when Pcre.pmatch ~rex:types_ann_RE s -> (true, Ann, No, No)
454 | s when Pcre.pmatch ~rex:body_RE s -> (false, No, Yes, No)
455 | s when Pcre.pmatch ~rex:body_ann_RE s -> (true, No, Ann, No)
456 | s when Pcre.pmatch ~rex:proof_tree_RE s -> (false, No, No, Yes)
457 | s when Pcre.pmatch ~rex:proof_tree_ann_RE s -> (true, No, No, Ann)
460 Hashtbl.replace objs basepart (oldflags ++ newflags)
462 (* Variables used in backward compatibility code to map
463 theory:/path/t.theory into theory:/path/t/index.theory
464 when cic:/path/t/ exists *)
465 let the_candidate_for_remapping =
466 (* CSC: Here I am making a strong assumption: the pattern
467 can be only of the form [^:]*:/path where path is
468 NOT a regular expression *)
469 "theory:" ^ Pcre.replace ~rex:(Pcre.regexp "[^:]*:") pat
471 let index_not_generated_yet = ref true in
472 let valid_candidates = ref [] in
473 let candidates_found = ref [] in
474 (Lazy.force cic_map) # iter
475 (* BLEARGH Dbm module lacks support for fold-like functions *)
478 | uri when looking_for_dir && Pcre.pmatch ~rex:dir_RE uri ->
480 let localpart = Pcre.replace ~rex:dir_RE uri in
481 if Pcre.pmatch ~rex:no_slashes_RE localpart then
483 (* Backward compatibility code to map
484 theory:/path/t.theory into theory:/path/t/index.theory
485 when cic:/path/t/ exists *)
486 if Pcre.pmatch ~rex:theory_RE localpart then
487 candidates_found := (uri,localpart) :: !candidates_found
494 store_dir localpart ;
495 if String.sub uri 0 3 = "cic" then
496 let dir_found = ref "" in
498 Pcre.substitute_first ~rex:dir_RE2
499 ~subst:(fun s -> dir_found := s; "") uri in
501 "theory" ^ String.sub !dir_found 3
502 (String.length !dir_found - 3) ^ ".theory" in
504 prerr_endline ("### " ^ uri ^ " ==> " ^ !dir_found ^ " ==> " ^ dir);
506 if not (List.mem dir !valid_candidates) then
507 valid_candidates := dir::!valid_candidates
509 | uri when (not looking_for_dir) && Pcre.pmatch ~rex:obj_RE uri ->
511 store_obj (Pcre.replace ~rex:til_slash_RE uri)
513 if !index_not_generated_yet &&
514 Pcre.pmatch ~rex:orig_theory_RE uri
516 (index_not_generated_yet := false ;
517 store_obj "index.theory"));
519 prerr_endline ("@@@ " ^ String.concat " " !valid_candidates);
520 prerr_endline ("!!! " ^ String.concat " " (List.map fst !candidates_found));
523 (fun (uri,localpart) ->
524 if not (List.mem uri !valid_candidates) then
526 ) !candidates_found ;
527 let ls_items = ref [] in
528 StringSet.iter (fun dir -> ls_items := Ls_section dir :: !ls_items) !dirs;
529 Http_getter_misc.hashtbl_sorted_iter
530 (fun uri (annflag, typesflag, bodyflag, treeflag) ->
533 uri = uri; ann = annflag;
534 types = typesflag; body = bodyflag; proof_tree = treeflag
540 (* Shorthands from now on *)
542 let getxml' uri = getxml (UriManager.string_of_uri uri)
543 let resolve' uri = resolve (UriManager.string_of_uri uri)
544 let register' uri url = register ~uri:(UriManager.string_of_uri uri) ~url
546 Http_getter_logger.set_log_level
547 (Helm_registry.get_opt_default Helm_registry.get_int 1 "getter.log_level");
548 Http_getter_logger.set_log_file
549 (Helm_registry.get_opt Helm_registry.get_string "getter.log_file");
550 Http_getter_env.reload ()