X-Git-Url: http://matita.cs.unibo.it/gitweb/?a=blobdiff_plain;f=helm%2Fhttp_getter%2Fhttp_getter.ml;h=0792d30b52a8a58ada989efe6570d3f01ab30dcd;hb=1c7fb836e2af4f2f3d18afd0396701f2094265ff;hp=0644fcc7d11000003aa8808c770cdebc1bf083ee;hpb=b6856f3de223e50ebeb3c26fd098ab128a5d661f;p=helm.git diff --git a/helm/http_getter/http_getter.ml b/helm/http_getter/http_getter.ml index 0644fcc7d..0792d30b5 100644 --- a/helm/http_getter/http_getter.ml +++ b/helm/http_getter/http_getter.ml @@ -1,5 +1,7 @@ (* - * Copyright (C) 2000, HELM Team. + * Copyright (C) 2003: + * Stefano Zacchiroli + * for the HELM Team http://helm.cs.unibo.it/ * * This file is part of HELM, an Hypertextual, Electronic * Library of Mathematics, developed at the Computer Science @@ -21,33 +23,26 @@ * MA 02111-1307, USA. * * For details, see the HELM World-Wide-Web page, - * http://cs.unibo.it/helm/. + * http://helm.cs.unibo.it/ *) -(* TODO optimization: precompile almost all regexp *) - open Http_getter_common;; +open Http_getter_misc;; open Http_getter_types;; +open Http_getter_debugger;; open Printf;; - (* debugging settings *) -let debug = true;; -let debug_print s = if debug then prerr_endline ("[HTTP-Getter] " ^ s);; -let http_debug = false;; -Http_common.debug := http_debug;; + (* constants *) -let http_get url = -(* debug_print ("Downloading URL: " ^ url); *) - try - Some (Http_client.Convenience.http_get url) - with Http_client.Http_error (code, _) -> None -(* - (debug_print - (sprintf "Failed to download %s, HTTP response was %d" url code); - None) -*) -;; +let common_headers = [ + "Cache-Control", "no-cache"; + "Pragma", "no-cache"; + "Expires", "0" +] + + (* HTTP queries argument parsing *) + (* parse encoding ("format" parameter), default is Enc_normal *) let parse_enc (req: Http_types.request) = try (match req#param "format" with @@ -56,44 +51,76 @@ let parse_enc (req: Http_types.request) = | s -> raise (Http_getter_bad_request ("Invalid format: " ^ s))) with Http_types.Param_not_found _ -> Enc_normal ;; -let parse_patch_dtd (req: Http_types.request) = - match req#param "patch_dtd" with - | s when String.lowercase s = "yes" -> true - | s when String.lowercase s = "no" -> false - | s -> raise (Http_getter_bad_request ("Invalid patch_dtd value: " ^ s)) + (* parse "patch_dtd" parameter, default is true *) +let parse_patch (req: Http_types.request) = + try + (match req#param "patch_dtd" with + | s when String.lowercase s = "yes" -> true + | s when String.lowercase s = "no" -> false + | s -> raise (Http_getter_bad_request ("Invalid patch_dtd value: " ^ s))) + with Http_types.Param_not_found _ -> true ;; + (* parse output format ("format" parameter), no default value *) let parse_output_format (req: Http_types.request) = match req#param "format" with | s when String.lowercase s = "txt" -> Fmt_text | s when String.lowercase s = "xml" -> Fmt_xml | s -> raise (Http_getter_bad_request ("Invalid /ls format: " ^ s)) ;; -let parse_ls_uri (req: Http_types.request) = - let baseuri = req#param "baseuri" in - let subs = - Pcre.extract ~pat:"^(\\w+):(.*)$" (Pcre.replace ~pat:"/+$" baseuri) + (* parse "baseuri" format for /ls method, no default value *) +let parse_ls_uri = + let parse_ls_RE = Pcre.regexp "^(\\w+):(.*)$" in + let trailing_slash_RE = Pcre.regexp "/+$" in + let wrong_uri uri = + raise (Http_getter_bad_request ("Invalid /ls baseuri: " ^ uri)) in - match (subs.(1), subs.(2)) with - | "cic", uri -> Cic uri - | "theory", uri -> Theory uri - | _ -> raise (Http_getter_bad_request ("Invalid /ls baseuri: " ^ baseuri)) + fun (req: Http_types.request) -> + let baseuri = req#param "baseuri" in + try + let subs = + Pcre.extract ~rex:parse_ls_RE + (Pcre.replace ~rex:trailing_slash_RE baseuri) + in + (match (subs.(1), subs.(2)) with + | "cic", uri -> Cic uri + | "theory", uri -> Theory uri + | _ -> wrong_uri baseuri) + with Not_found -> wrong_uri baseuri ;; -let xml_map = new Http_getter_map.map Http_getter_env.xml_dbm in + (* global maps, shared by all threads *) + +let cic_map = new Http_getter_map.map Http_getter_env.cic_dbm in +let nuprl_map = new Http_getter_map.map Http_getter_env.nuprl_dbm in let rdf_map = new Http_getter_map.map Http_getter_env.rdf_dbm in let xsl_map = new Http_getter_map.map Http_getter_env.xsl_dbm in +let save_maps () = + cic_map#close; nuprl_map#close; rdf_map#close; xsl_map#close in let map_of_uri = function - | uri when is_xml_uri uri -> xml_map + | uri when is_cic_uri uri -> cic_map + | uri when is_nuprl_uri uri -> nuprl_map | uri when is_rdf_uri uri -> rdf_map | uri when is_xsl_uri uri -> xsl_map | uri -> raise (Http_getter_unresolvable_URI uri) in -let resolve uri = (map_of_uri uri)#resolve uri in -let register uri = (map_of_uri uri )#add uri in +let resolve uri = + try + (map_of_uri uri)#resolve uri + with Http_getter_map.Key_not_found _ -> + raise (Http_getter_unresolvable_URI uri) +in +let register uri = + (* Warning: this fail if uri is already registered *) + (map_of_uri uri)#add uri +in let return_all_foo_uris map doctype filter outchan = + (** return all URIs contained in 'map' which satisfy predicate 'filter'; URIs + are written in an XMLish format ('doctype' is the XML doctype) onto 'outchan' + *) Http_daemon.send_basic_headers ~code:200 outchan; Http_daemon.send_header "Content-Type" "text/xml" outchan; + Http_daemon.send_headers common_headers outchan; Http_daemon.send_CRLF outchan; output_string outchan @@ -113,14 +140,14 @@ let return_all_foo_uris map doctype filter outchan = output_string outchan (sprintf "\t\n" uri)); output_string outchan (sprintf "\n" doctype) in -let return_all_xml_uris = return_all_foo_uris xml_map "alluris" in +let return_all_xml_uris = return_all_foo_uris cic_map "alluris" in let return_all_rdf_uris = return_all_foo_uris rdf_map "allrdfuris" in let return_ls = let (++) (oldann, oldtypes, oldbody) (newann, newtypes, newbody) = ((if newann > oldann then newann else oldann), (if newtypes > oldtypes then newtypes else oldtypes), - (if newbody > oldbody then newbody else oldbody)) - in + (if newbody > oldbody then newbody else oldbody)) + in let basepart_RE = Pcre.regexp "^([^.]*\\.[^.]*)((\\.body)|(\\.types))?(\\.ann)?" in @@ -149,7 +176,7 @@ let return_ls = let oldflags = try Hashtbl.find objs basepart - with Not_found -> (false, No, No) (* no ann, no types no body *) + with Not_found -> (false, No, No) (* no ann, no types, no body *) in let newflags = match o with @@ -161,7 +188,7 @@ let return_ls = in Hashtbl.replace objs basepart (oldflags ++ newflags) in - xml_map#iter (* BLEARGH Dbm module lacks support for fold-like functions *) + cic_map#iter (* BLEARGH Dbm module lacks support for fold-like functions *) (fun key _ -> match key with | uri when Pcre.pmatch ~rex:dir_RE uri -> (* directory hit *) @@ -187,7 +214,8 @@ let return_ls = objs "") in Http_daemon.respond - ~headers:["Content-Type", "text/plain"] ~body outchan + ~headers:(("Content-Type", "text/plain") :: common_headers) + ~body outchan | Fmt_xml -> let body = sprintf @@ -220,23 +248,34 @@ let return_ls = objs "")) in Http_daemon.respond - ~headers:["Content-Type", "text/xml"] ~body outchan + ~headers:(("Content-Type", "text/xml") :: common_headers) + ~body outchan +in +let (index_line_sep_RE, index_sep_RE, trailing_types_RE, + heading_cic_RE, heading_theory_RE, heading_nuprl_RE, + heading_rdf_cic_RE, heading_rdf_theory_RE) = + (Pcre.regexp "[ \t]+", Pcre.regexp "\r\n|\r|\n", + Pcre.regexp "\\.types$", + Pcre.regexp "^cic:", Pcre.regexp "^theory:", Pcre.regexp "^nuprl:", + Pcre.regexp "^helm:rdf.*//cic:", Pcre.regexp "^helm:rdf.*//theory:") in let update_from_server logmsg server_url = (* use global maps *) debug_print ("Updating information from " ^ server_url); let xml_url_of_uri = function (* TODO missing sanity checks on server_url, e.g. it can contains $1 *) - | uri when (Pcre.pmatch ~pat:"^cic:" uri) -> - Pcre.replace ~pat:"^cic:" ~templ:server_url uri - | uri when (Pcre.pmatch ~pat:"^theory:" uri) -> - Pcre.replace ~pat:"^theory:" ~templ:server_url uri + | uri when (Pcre.pmatch ~rex:heading_cic_RE uri) -> + Pcre.replace ~rex:heading_cic_RE ~templ:server_url uri + | uri when (Pcre.pmatch ~rex:heading_theory_RE uri) -> + Pcre.replace ~rex:heading_theory_RE ~templ:server_url uri + | uri when (Pcre.pmatch ~rex:heading_nuprl_RE uri) -> + Pcre.replace ~rex:heading_nuprl_RE ~templ:server_url uri | uri -> raise (Http_getter_invalid_URI uri) in let rdf_url_of_uri = function (* TODO as above *) - | uri when (Pcre.pmatch ~pat:"^helm:rdf.*//cic:" uri) -> - Pcre.replace ~pat:"^helm:rdf.*//cic:" ~templ:server_url uri - | uri when (Pcre.pmatch ~pat:"^helm:rdf.*//theory:" uri) -> - Pcre.replace ~pat:"^helm:rdf.*//theory:" ~templ:server_url uri + | uri when (Pcre.pmatch ~rex:heading_rdf_cic_RE uri) -> + Pcre.replace ~rex:heading_rdf_cic_RE ~templ:server_url uri + | uri when (Pcre.pmatch ~rex:heading_rdf_theory_RE uri) -> + Pcre.replace ~rex:heading_rdf_theory_RE ~templ:server_url uri | uri -> raise (Http_getter_invalid_URI uri) in let log = ref (logmsg ^ "Processing server: " ^ server_url ^ "
\n") in @@ -252,15 +291,24 @@ let update_from_server logmsg server_url = (* use global maps *) | Some xml_index -> (log := !log ^ "Updating XML db ...
\n"; List.iter - (fun l -> - try - (match Pcre.split ~pat:"[ \\t]+" l with - | [uri; "gz"] -> xml_map#add uri ((xml_url_of_uri uri) ^ ".xml.gz") - | [uri] -> xml_map#add uri ((xml_url_of_uri uri) ^ ".xml") - | _ -> log := !log ^ "Ignoring invalid line: " ^ l ^ "
\n") - with Http_getter_invalid_URI uri -> - log := !log ^ "Ignoring invalid XML URI: " ^ uri ^ "
\n") - (Pcre.split ~pat:"\n+" xml_index)) (* xml_index lines *) + (function + | l when is_blank_line l -> () (* skip blank and commented lines *) + | l -> + try + (match Pcre.split ~rex:index_line_sep_RE l with + | [uri; "gz"] -> + assert (is_cic_uri uri || is_nuprl_uri uri) ; + (map_of_uri uri)#replace + uri ((xml_url_of_uri uri) ^ ".xml.gz") + | [uri] -> + assert (is_cic_uri uri || is_nuprl_uri uri) ; + (map_of_uri uri)#replace + uri ((xml_url_of_uri uri) ^ ".xml") + | _ -> + log := !log ^ "Ignoring invalid line: '" ^ l ^ "'
\n") + with Http_getter_invalid_URI uri -> + log := !log ^ "Ignoring invalid XML URI: '" ^ uri ^ "'
\n") + (Pcre.split ~rex:index_sep_RE xml_index)) (* xml_index lines *) | None -> ()); (match rdf_index with | Some rdf_index -> @@ -268,81 +316,151 @@ let update_from_server logmsg server_url = (* use global maps *) List.iter (fun l -> try - (match Pcre.split ~pat:"[ \\t]+" l with - | [uri; "gz"] -> rdf_map#add uri ((rdf_url_of_uri uri) ^ ".xml.gz") - | [uri] -> rdf_map#add uri ((rdf_url_of_uri uri) ^ ".xml") + (match Pcre.split ~rex:index_line_sep_RE l with + | [uri; "gz"] -> + rdf_map#replace uri ((rdf_url_of_uri uri) ^ ".xml.gz") + | [uri] -> rdf_map#replace uri ((rdf_url_of_uri uri) ^ ".xml") | _ -> log := !log ^ "Ignoring invalid line: " ^ l ^ "
\n") with Http_getter_invalid_URI uri -> log := !log ^ "Ignoring invalid RDF URI: " ^ uri ^ "
\n") - (Pcre.split ~pat:"\n+" rdf_index)) (* rdf_index lines *) + (Pcre.split ~rex:index_sep_RE rdf_index)) (* rdf_index lines *) | None -> ()); (match xsl_index with | Some xsl_index -> (log := !log ^ "Updating XSLT db ...
\n"; List.iter - (fun l -> xsl_map#add l (server_url ^ "/" ^ l)) - (Pcre.split ~pat:"\n+" xsl_index); + (fun l -> xsl_map#replace l (server_url ^ "/" ^ l)) + (Pcre.split ~rex:index_sep_RE xsl_index); log := !log ^ "All done!
\n") | None -> ()); + debug_print "done with this server"; !log in +let update_from_all_servers () = (* use global maps *) + cic_map#clear; nuprl_map#clear; rdf_map#clear; xsl_map#clear; + let log = + List.fold_left + update_from_server + "" (* initial logmsg: empty *) + (* reverse order: 1st server is the most important one *) + (List.rev !Http_getter_env.servers) + in + cic_map#sync; nuprl_map#sync; rdf_map#sync; xsl_map#sync; + log +in (* thread action *) + let callback (req: Http_types.request) outchan = try debug_print ("Connection from " ^ req#clientAddr); debug_print ("Received request: " ^ req#path); (match req#path with - | "/help" -> return_html_msg Http_getter_const.usage_string outchan + | "/help" -> + return_html_raw + (Http_getter_const.usage_string (Http_getter_env.env_to_string ())) + outchan | "/getxml" | "/getxslt" | "/getdtd" | "/resolve" | "/register" -> (let uri = req#param "uri" in (* common parameter *) match req#path with | "/getxml" -> let enc = parse_enc req in - let patch_dtd = - try parse_patch_dtd req with Http_types.Param_not_found _ -> true - in + let patch = parse_patch req in Http_getter_cache.respond_xml - ~url:(resolve uri) ~uri ~enc ~patch_dtd outchan + ~url:(resolve uri) ~uri ~enc ~patch outchan | "/getxslt" -> - let patch_dtd = - try parse_patch_dtd req with Http_types.Param_not_found _ -> true - in - Http_getter_cache.respond_xsl ~url:(resolve uri) ~patch_dtd outchan + let patch = parse_patch req in + Http_getter_cache.respond_xsl ~url:(resolve uri) ~patch outchan | "/getdtd" -> - let patch_dtd = - try parse_patch_dtd req with Http_types.Param_not_found _ -> true - in + let patch = parse_patch req in Http_getter_cache.respond_dtd - ~patch_dtd ~url:(Http_getter_env.dtd_dir ^ "/" ^ uri) outchan + ~patch ~url:(Http_getter_env.dtd_dir ^ "/" ^ uri) outchan | "/resolve" -> (try - return_xml_msg + return_xml_raw (sprintf "\n" (resolve uri)) outchan with Http_getter_unresolvable_URI uri -> - return_xml_msg "\n" outchan) + return_xml_raw "\n" outchan) | "/register" -> let url = req#param "url" in register uri url; return_html_msg "Register done" outchan | _ -> assert false) | "/update" -> - (xml_map#clear; rdf_map#clear; xsl_map#clear; - let log = - List.fold_left - update_from_server - "" (* initial logmsg: empty *) - (* reverse order: 1st server is the most important one *) - (List.rev Http_getter_env.servers) + Http_getter_env.reload (); (* reload servers list from servers file *) + let log = update_from_all_servers () in + return_html_msg log outchan + | "/list_servers" -> + return_html_raw + (sprintf "\n%s\n
" + (String.concat "\n" + (List.map + (let i = ref ~-1 in + fun s -> incr i; sprintf "%d%s" !i s) + !Http_getter_env.servers))) + outchan + | "/add_server" -> + let name = req#param "url" in + (try + let position = + try + let res = int_of_string (req#param "position") in + if res < 0 then + raise (Failure "int_of_string"); + res + with Failure "int_of_string" -> + raise (Http_getter_bad_request + (sprintf "position must be a non negative integer (%s given)" + (req#param "position"))) + in + if position = 0 then (* fallback to default value *) + raise (Http_types.Param_not_found "foo") + else if position > 0 then begin (* add server and update all *) + Http_getter_env.add_server ~position name; + let log = update_from_all_servers () in + return_html_msg + (sprintf "Added server %s in position %d)
\n%s" + name position log) + outchan + end else (* position < 0 *) (* error! *) + assert false (* already checked above *) + with Http_types.Param_not_found _ -> (* add as 1st server by default *) + Http_getter_env.add_server ~position:0 name; + let log = update_from_server (* quick update (new server only) *) + (sprintf "Added server %s in head position
\n" name) name + in + return_html_msg log outchan) + | "/remove_server" -> + let position = + try + let res = int_of_string (req#param "position") in + if res < 0 then + raise (Failure "int_of_string"); + res + with Failure "int_of_string" -> + raise (Http_getter_bad_request + (sprintf "position must be a non negative integer (%s given)" + (req#param "position"))) in - xml_map#sync; rdf_map#sync; xsl_map#sync; - return_html_msg log outchan) + let server_name = + try + List.nth !Http_getter_env.servers position + with Failure "nth" -> + raise (Http_getter_bad_request + (sprintf "no server with position %d" position)) + in + Http_getter_env.remove_server position; + let log = update_from_all_servers () in + return_html_msg + (sprintf "Removed server %s (position %d)
\n%s" + server_name position log) + outchan | "/getalluris" -> return_all_xml_uris (fun uri -> - (Pcre.pmatch ~pat:"^cic:" uri) && - not (Pcre.pmatch ~pat:"\\.types$" uri)) + (Pcre.pmatch ~rex:heading_cic_RE uri) && + not (Pcre.pmatch ~rex:trailing_types_RE uri)) outchan | "/getallrdfuris" -> (let classs = req#param "class" in @@ -361,22 +479,41 @@ let callback (req: Http_types.request) outchan = | "/getempty" -> Http_daemon.respond ~body:Http_getter_const.empty_xml outchan | invalid_request -> - Http_daemon.respond_error ~status:(`Client_error `Bad_request) outchan) + Http_daemon.respond_error ~status:(`Client_error `Bad_request) outchan); + debug_print "Done!\n" with | Http_types.Param_not_found attr_name -> return_400 (sprintf "Parameter '%s' is missing" attr_name) outchan | Http_getter_bad_request msg -> return_html_error msg outchan | Http_getter_internal_error msg -> return_html_internal_error msg outchan + | Shell.Subprocess_error l -> + return_html_internal_error + (String.concat "
\n" + (List.map + (fun (cmd, code) -> + sprintf "Command '%s' returned %s" + cmd (string_of_proc_status code)) + l)) + outchan | exc -> return_html_error ("Uncaught exception: " ^ (Printexc.to_string exc)) outchan in - (* daemon initialization *) -Http_getter_env.dump_env (); -flush stdout; -Unix.putenv "http_proxy" ""; -Http_daemon.start' - ~timeout:None ~port:Http_getter_env.port ~mode:`Thread callback + (* daemon initialization *) + +let main () = + print_string (Http_getter_env.env_to_string ()); + flush stdout; + Unix.putenv "http_proxy" ""; + at_exit save_maps; + Sys.catch_break true; + try + Http_daemon.start' + ~timeout:(Some 600) ~port:Http_getter_env.port ~mode:`Thread callback + with Sys.Break -> () (* 'save_maps' already registered with 'at_exit' *) +in + +main ()