2 * Copyright (C) 2003, HELM Team.
4 * This file is part of HELM, an Hypertextual, Electronic
5 * Library of Mathematics, developed at the Computer Science
6 * Department, University of Bologna, Italy.
8 * HELM is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU General Public License
10 * as published by the Free Software Foundation; either version 2
11 * of the License, or (at your option) any later version.
13 * HELM is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 * GNU General Public License for more details.
18 * You should have received a copy of the GNU General Public License
19 * along with HELM; if not, write to the Free Software
20 * Foundation, Inc., 59 Temple Place - Suite 330, Boston,
23 * For details, see the HELM World-Wide-Web page,
24 * http://cs.unibo.it/helm/.
27 open Http_getter_common;;
28 open Http_getter_misc;;
29 open Http_getter_types;;
30 open Http_getter_debugger;;
35 let common_headers = [
36 "Cache-Control", "no-cache";
41 (* HTTP queries argument parsing *)
43 let parse_enc (req: Http_types.request) =
45 (match req#param "format" with
46 | "normal" -> Enc_normal
48 | s -> raise (Http_getter_bad_request ("Invalid format: " ^ s)))
49 with Http_types.Param_not_found _ -> Enc_normal
51 let parse_patch_dtd (req: Http_types.request) =
52 match req#param "patch_dtd" with
53 | s when String.lowercase s = "yes" -> true
54 | s when String.lowercase s = "no" -> false
55 | s -> raise (Http_getter_bad_request ("Invalid patch_dtd value: " ^ s))
57 let parse_output_format (req: Http_types.request) =
58 match req#param "format" with
59 | s when String.lowercase s = "txt" -> Fmt_text
60 | s when String.lowercase s = "xml" -> Fmt_xml
61 | s -> raise (Http_getter_bad_request ("Invalid /ls format: " ^ s))
64 let parse_ls_RE = Pcre.regexp "^(\\w+):(.*)$" in
65 let trailing_slash_RE = Pcre.regexp "/+$" in
66 fun (req: Http_types.request) ->
67 let baseuri = req#param "baseuri" in
69 Pcre.extract ~rex:parse_ls_RE
70 (Pcre.replace ~rex:trailing_slash_RE baseuri)
72 match (subs.(1), subs.(2)) with
73 | "cic", uri -> Cic uri
74 | "theory", uri -> Theory uri
75 | _ -> raise (Http_getter_bad_request ("Invalid /ls baseuri: " ^ baseuri))
78 (* global maps, shared by all threads *)
80 let xml_map = new Http_getter_map.map Http_getter_env.xml_dbm in
81 let rdf_map = new Http_getter_map.map Http_getter_env.rdf_dbm in
82 let xsl_map = new Http_getter_map.map Http_getter_env.xsl_dbm in
84 let save_maps () = xml_map#close; rdf_map#close; xsl_map#close in
85 let map_of_uri = function
86 | uri when is_xml_uri uri -> xml_map
87 | uri when is_rdf_uri uri -> rdf_map
88 | uri when is_xsl_uri uri -> xsl_map
89 | uri -> raise (Http_getter_unresolvable_URI uri)
93 (map_of_uri uri)#resolve uri
94 with Http_getter_map.Key_not_found _ ->
95 raise (Http_getter_unresolvable_URI uri)
97 let register uri = (map_of_uri uri )#add uri in
98 let return_all_foo_uris map doctype filter outchan =
99 (** return all URIs contained in 'map' which satisfy predicate 'filter'; URIs
100 are written in an XMLish format ('doctype' is the XML doctype) onto 'outchan'
102 Http_daemon.send_basic_headers ~code:200 outchan;
103 Http_daemon.send_header "Content-Type" "text/xml" outchan;
104 Http_daemon.send_headers common_headers outchan;
105 Http_daemon.send_CRLF outchan;
109 "<?xml version=\"1.0\" encoding=\"ISO-8859-1\"?>
110 <!DOCTYPE %s SYSTEM \"%s/getdtd?uri=%s.dtd\">
115 Http_getter_env.my_own_url
121 output_string outchan (sprintf "\t<uri value=\"%s\" />\n" uri));
122 output_string outchan (sprintf "</%s>\n" doctype)
124 let return_all_xml_uris = return_all_foo_uris xml_map "alluris" in
125 let return_all_rdf_uris = return_all_foo_uris rdf_map "allrdfuris" in
127 let (++) (oldann, oldtypes, oldbody) (newann, newtypes, newbody) =
128 ((if newann > oldann then newann else oldann),
129 (if newtypes > oldtypes then newtypes else oldtypes),
130 (if newbody > oldbody then newbody else oldbody))
133 Pcre.regexp "^([^.]*\\.[^.]*)((\\.body)|(\\.types))?(\\.ann)?"
135 let (types_RE, types_ann_RE, body_RE, body_ann_RE) =
136 (Pcre.regexp "\\.types", Pcre.regexp "\\.types.ann",
137 Pcre.regexp "\\.body", Pcre.regexp "\\.body.ann")
139 let (slash_RE, til_slash_RE, no_slashes_RE) =
140 (Pcre.regexp "/", Pcre.regexp "^.*/", Pcre.regexp "^[^/]*$")
142 fun lsuri fmt outchan ->
145 (match lsuri with Cic p -> ("cic:" ^ p) | Theory p -> ("theory:" ^ p))
147 let (dir_RE, obj_RE) =
148 (Pcre.regexp (pat ^ "/"), Pcre.regexp (pat ^ "(\\.|$)"))
150 let dirs = ref StringSet.empty in
151 let objs = Hashtbl.create 17 in
153 dirs := StringSet.add (List.hd (Pcre.split ~rex:slash_RE d)) !dirs
156 let basepart = Pcre.replace ~rex:basepart_RE ~templ:"$1" o in
159 Hashtbl.find objs basepart
160 with Not_found -> (false, No, No) (* no ann, no types no body *)
164 | s when Pcre.pmatch ~rex:types_RE s -> (false, Yes, No)
165 | s when Pcre.pmatch ~rex:types_ann_RE s -> (true, Ann, No)
166 | s when Pcre.pmatch ~rex:body_RE s -> (false, No, Yes)
167 | s when Pcre.pmatch ~rex:body_ann_RE s -> (true, No, Ann)
168 | s -> (false, No, No)
170 Hashtbl.replace objs basepart (oldflags ++ newflags)
172 xml_map#iter (* BLEARGH Dbm module lacks support for fold-like functions *)
175 | uri when Pcre.pmatch ~rex:dir_RE uri -> (* directory hit *)
176 let localpart = Pcre.replace ~rex:dir_RE uri in
177 if Pcre.pmatch ~rex:no_slashes_RE localpart then
181 | uri when Pcre.pmatch ~rex:obj_RE uri -> (* file hit *)
182 store_obj (Pcre.replace ~rex:til_slash_RE uri)
183 | uri -> () (* miss *));
188 (fun s d -> sprintf "%sdir, %s\n" s d) ""
189 (StringSet.elements !dirs)) ^
190 (Http_getter_misc.hashtbl_sorted_fold
191 (fun uri (annflag, typesflag, bodyflag) cont ->
192 sprintf "%sobject, %s, <%s,%s,%s>\n"
193 cont uri (if annflag then "YES" else "NO")
194 (string_of_ls_flag typesflag) (string_of_ls_flag bodyflag))
198 ~headers:(("Content-Type", "text/plain") :: common_headers)
203 "<?xml version=\"1.0\" encoding=\"ISO-8859-1\"?>
204 <!DOCTYPE ls SYSTEM \"%s/getdtd?uri=ls.dtd\">
210 Http_getter_env.my_own_url
215 (fun d -> "<section>" ^ d ^ "</section>")
216 (StringSet.elements !dirs))) ^ "\n" ^
217 (Http_getter_misc.hashtbl_sorted_fold
218 (fun uri (annflag, typesflag, bodyflag) cont ->
220 "%s<object name=\"%s\">
221 \t<ann value=\"%s\" />
222 \t<types value=\"%s\" />
223 \t<body value=\"%s\" />
226 cont uri (if annflag then "YES" else "NO")
227 (string_of_ls_flag typesflag)
228 (string_of_ls_flag bodyflag))
232 ~headers:(("Content-Type", "text/xml") :: common_headers)
235 let (index_line_sep_RE, index_sep_RE, trailing_types_RE,
236 heading_cic_RE, heading_theory_RE,
237 heading_rdf_cic_RE, heading_rdf_theory_RE) =
238 (Pcre.regexp "[ \t]+", Pcre.regexp "\n+", Pcre.regexp "\\.types$",
239 Pcre.regexp "^cic:", Pcre.regexp "^theory:",
240 Pcre.regexp "^helm:rdf.*//cic:", Pcre.regexp "^helm:rdf.*//theory:")
242 let update_from_server logmsg server_url = (* use global maps *)
243 debug_print ("Updating information from " ^ server_url);
244 let xml_url_of_uri = function
245 (* TODO missing sanity checks on server_url, e.g. it can contains $1 *)
246 | uri when (Pcre.pmatch ~rex:heading_cic_RE uri) ->
247 Pcre.replace ~rex:heading_cic_RE ~templ:server_url uri
248 | uri when (Pcre.pmatch ~rex:heading_theory_RE uri) ->
249 Pcre.replace ~rex:heading_theory_RE ~templ:server_url uri
250 | uri -> raise (Http_getter_invalid_URI uri)
252 let rdf_url_of_uri = function (* TODO as above *)
253 | uri when (Pcre.pmatch ~rex:heading_rdf_cic_RE uri) ->
254 Pcre.replace ~rex:heading_rdf_cic_RE ~templ:server_url uri
255 | uri when (Pcre.pmatch ~rex:heading_rdf_theory_RE uri) ->
256 Pcre.replace ~rex:heading_rdf_theory_RE ~templ:server_url uri
257 | uri -> raise (Http_getter_invalid_URI uri)
259 let log = ref (logmsg ^ "Processing server: " ^ server_url ^ "<br />\n") in
260 let (xml_index, rdf_index, xsl_index) =
261 (* TODO keeps index in memory, is better to keep them on temp files? *)
262 (http_get (server_url ^ "/" ^ Http_getter_env.xml_index),
263 http_get (server_url ^ "/" ^ Http_getter_env.rdf_index),
264 http_get (server_url ^ "/" ^ Http_getter_env.xsl_index))
266 if (xml_index = None && rdf_index = None && xsl_index = None) then
267 debug_print (sprintf "Warning: useless server %s" server_url);
268 (match xml_index with
270 (log := !log ^ "Updating XML db ...<br />\n";
274 (match Pcre.split ~rex:index_line_sep_RE l with
275 | [uri; "gz"] -> xml_map#add uri ((xml_url_of_uri uri) ^ ".xml.gz")
276 | [uri] -> xml_map#add uri ((xml_url_of_uri uri) ^ ".xml")
277 | _ -> log := !log ^ "Ignoring invalid line: " ^ l ^ "<br />\n")
278 with Http_getter_invalid_URI uri ->
279 log := !log ^ "Ignoring invalid XML URI: " ^ uri ^ "<br />\n")
280 (Pcre.split ~rex:index_sep_RE xml_index)) (* xml_index lines *)
282 (match rdf_index with
284 (log := !log ^ "Updating RDF db ...<br />\n";
288 (match Pcre.split ~rex:index_line_sep_RE l with
289 | [uri; "gz"] -> rdf_map#add uri ((rdf_url_of_uri uri) ^ ".xml.gz")
290 | [uri] -> rdf_map#add uri ((rdf_url_of_uri uri) ^ ".xml")
291 | _ -> log := !log ^ "Ignoring invalid line: " ^ l ^ "<br />\n")
292 with Http_getter_invalid_URI uri ->
293 log := !log ^ "Ignoring invalid RDF URI: " ^ uri ^ "<br />\n")
294 (Pcre.split ~rex:index_sep_RE rdf_index)) (* rdf_index lines *)
296 (match xsl_index with
298 (log := !log ^ "Updating XSLT db ...<br />\n";
300 (fun l -> xsl_map#add l (server_url ^ "/" ^ l))
301 (Pcre.split ~rex:index_sep_RE xsl_index);
302 log := !log ^ "All done!<br />\n")
304 debug_print "done with this server";
310 let callback (req: Http_types.request) outchan =
312 debug_print ("Connection from " ^ req#clientAddr);
313 debug_print ("Received request: " ^ req#path);
315 | "/help" -> return_html_msg Http_getter_const.usage_string outchan
316 | "/getxml" | "/getxslt" | "/getdtd" | "/resolve" | "/register" ->
317 (let uri = req#param "uri" in (* common parameter *)
320 let enc = parse_enc req in
322 try parse_patch_dtd req with Http_types.Param_not_found _ -> true
324 Http_getter_cache.respond_xml
325 ~url:(resolve uri) ~uri ~enc ~patch_dtd outchan
328 try parse_patch_dtd req with Http_types.Param_not_found _ -> true
330 Http_getter_cache.respond_xsl ~url:(resolve uri) ~patch_dtd outchan
333 try parse_patch_dtd req with Http_types.Param_not_found _ -> true
335 Http_getter_cache.respond_dtd
336 ~patch_dtd ~url:(Http_getter_env.dtd_dir ^ "/" ^ uri) outchan
340 (sprintf "<url value=\"%s\" />\n" (resolve uri))
342 with Http_getter_unresolvable_URI uri ->
343 return_xml_msg "<unresolved />\n" outchan)
345 let url = req#param "url" in
347 return_html_msg "Register done" outchan
350 (xml_map#clear; rdf_map#clear; xsl_map#clear;
354 "" (* initial logmsg: empty *)
355 (* reverse order: 1st server is the most important one *)
356 (List.rev Http_getter_env.servers)
358 xml_map#sync; rdf_map#sync; xsl_map#sync;
359 return_html_msg log outchan)
363 (Pcre.pmatch ~rex:heading_cic_RE uri) &&
364 not (Pcre.pmatch ~rex:trailing_types_RE uri))
366 | "/getallrdfuris" ->
367 (let classs = req#param "class" in
370 let base = "^helm:rdf:www\\.cs\\.unibo\\.it/helm/rdf/" in
372 | ("forward" as c) | ("backward" as c) ->
373 (fun uri -> Pcre.pmatch ~pat:(base ^ c) uri)
374 | c -> raise (Http_getter_invalid_RDF_class c)
376 return_all_rdf_uris filter outchan
377 with Http_getter_invalid_RDF_class c ->
378 raise (Http_getter_bad_request ("Invalid RDF class: " ^ c)))
379 | "/ls" -> return_ls (parse_ls_uri req) (parse_output_format req) outchan
381 Http_daemon.respond ~body:Http_getter_const.empty_xml outchan
383 Http_daemon.respond_error ~status:(`Client_error `Bad_request) outchan);
384 debug_print "Done!\n"
386 | Http_types.Param_not_found attr_name ->
387 return_400 (sprintf "Parameter '%s' is missing" attr_name) outchan
388 | Http_getter_bad_request msg -> return_html_error msg outchan
389 | Http_getter_internal_error msg -> return_html_internal_error msg outchan
390 | Shell.Subprocess_error l ->
391 return_html_internal_error
392 (String.concat "<br />\n"
395 sprintf "Command '%s' returned %s"
396 cmd (string_of_proc_status code))
401 ("Uncaught exception: " ^ (Printexc.to_string exc))
405 (* daemon initialization *)
408 Http_getter_env.dump_env ();
409 Unix.putenv "http_proxy" "";
411 Sys.catch_break true;
414 ~timeout:(Some 600) ~port:Http_getter_env.port ~mode:`Thread callback
415 with Sys.Break -> () (* 'save_maps' already registered with 'at_exit' *)