From 4318efab0b880c38237daf992b401de1584ca9f2 Mon Sep 17 00:00:00 2001 From: Stefano Zacchiroli Date: Tue, 21 Jan 2003 11:37:12 +0000 Subject: [PATCH] - disambiguation implemented! --- helm/searchEngine/searchEngine.ml | 249 +++++++++++++++++++++++++++--- 1 file changed, 227 insertions(+), 22 deletions(-) diff --git a/helm/searchEngine/searchEngine.ml b/helm/searchEngine/searchEngine.ml index 0accf62ab..7b5635704 100644 --- a/helm/searchEngine/searchEngine.ml +++ b/helm/searchEngine/searchEngine.ml @@ -24,9 +24,9 @@ * http://cs.unibo.it/helm/. *) -let debug = false;; +let debug = true;; let debug_print s = if debug then prerr_endline s;; -Http_common.debug := debug;; +Http_common.debug := true;; open Printf;; @@ -34,6 +34,40 @@ let daemon_name = "Search Engine";; let default_port = 58085;; let port_env_var = "SEARCH_ENGINE_PORT";; +let pages_dir = "html";; (* relative to searchEngine's document root *) +let interactive_user_uri_choice_TPL = pages_dir ^ "/templateambigpdq1.html";; +let interactive_interpretation_choice_TPL = pages_dir ^ "/templateambigpdq2.html";; +let final_results_TPL = pages_dir ^ "/templateambigpdq3.html";; + +exception Chat_unfinished + + (** chain application of Pcre substitutions *) +let rec apply_substs substs line = + match substs with + | [] -> line + | (rex, templ) :: rest -> apply_substs rest (Pcre.replace ~rex ~templ line) + (** fold like function on files *) +let fold_file f init fname = + let inchan = open_in fname in + let rec fold_lines' value = + try + let line = input_line inchan in + fold_lines' (f value line) + with End_of_file -> value + in + let res = (try fold_lines' init with e -> (close_in inchan; raise e)) in + close_in inchan; + res + (** iter like function on files *) +let iter_file f = fold_file (fun _ line -> f line) () + +let (title_tag_RE, choices_tag_RE, msg_tag_RE, id_to_uris_RE, id_RE, + interpretations_RE, interpretations_labels_RE, results_RE, new_aliases_RE) = + (Pcre.regexp "@TITLE@", Pcre.regexp "@CHOICES@", Pcre.regexp "@MSG@", + Pcre.regexp "@ID_TO_URIS@", Pcre.regexp "@ID@", + Pcre.regexp "@INTERPRETATIONS@", Pcre.regexp "@INTERPRETATIONS_LABELS@", + Pcre.regexp "@RESULTS@", Pcre.regexp "@NEW_ALIASES@") + let port = try int_of_string (Sys.getenv port_env_var) @@ -45,14 +79,24 @@ let port = in let pp_result result = let result_string = MQueryUtil.text_of_result result "\n" in - (sprintf "

%s

" result_string) + (sprintf "\n\n\n\n

%s

\n\n" + result_string) in let pp_error = sprintf "

Error: %s

" in let bad_request body outchan = Http_daemon.respond_error ~status:(`Client_error `Bad_request) ~body outchan in -let callback req outchan = +let contype = "Content-Type", "text/html" in + +(* SEARCH ENGINE functions *) + +let refine_constraints (x, y, z) = (x, y, z), (Some x, Some y, Some z) in + +(* HTTP DAEMON CALLBACK *) + +let callback (req: Http_types.request) outchan = try + debug_print (sprintf "Received request: %s" req#path); (match req#path with | "/execute" -> let query_string = req#param "query" in @@ -63,36 +107,196 @@ let callback req outchan = Http_daemon.respond ~body: (sprintf "

%s

" result_string) - outchan + ~headers:[contype] outchan | "/locate" -> let id = req#param "id" in let result = MQueryGenerator.locate id in - Http_daemon.respond ~body:(pp_result result) outchan + Http_daemon.respond ~headers:[contype] ~body:(pp_result result) outchan + | "/getpage" -> + (let is_permitted _ = true in + let remove_fragment uri = Pcre.replace ~pat:"#.*" uri in + match req#param "url" with + | page when is_permitted (remove_fragment page) -> + Http_daemon.respond_file + ~fname:(sprintf "%s/%s" pages_dir (remove_fragment page)) outchan + | page -> Http_daemon.respond_forbidden ~url:page outchan) | "/searchPattern" -> let term_string = req#param "term" in - let precision = int_of_string (req#param "precision") in let lexbuf = Lexing.from_string term_string in - let (dom, mkterm) = - CicTextualParser.main CicTextualLexer.token lexbuf + let (context, metasenv) = ([], []) in + let (dom, mk_metasenv_and_expr) = + CicTextualParserContext.main + ~context ~metasenv CicTextualLexer.token lexbuf + in + let id_to_uris_raw = req#param "aliases" in + let tokens = Pcre.split ~pat:"\\s" id_to_uris_raw in + let rec parse_tokens keys lookup = function (* TODO spostarla fuori *) + | [] -> keys, lookup + | "alias" :: key :: value :: rest -> + parse_tokens + (key::keys) + (fun id -> + if id = key then + Some (Disambiguate.cic_textual_parser_uri_of_string value) + else lookup id) + rest + | _ -> failwith "Can't parse aliases" + in + let parse_choices choices_raw = + let choices = Pcre.split ~pat:";" choices_raw in + List.fold_left + (fun f x -> + match Pcre.split ~pat:"\\s" x with + | ""::id::tail + | id::tail when id<>"" -> + (fun id' -> + if id = id' then + Some (List.map (fun u -> Netencoding.Url.decode u) tail) + else + f id') + | _ -> failwith "Can't parse choices") + (fun _ -> None) + choices in - (match dom with - | [] -> (* no free variables *) - let term = mkterm (fun _ -> None) in - prerr_endline (CicPp.ppterm term); - let result = MQueryGenerator.searchPattern [] [] term precision in - Http_daemon.respond ~body:(pp_result result) outchan - | _ -> + let id_to_uris = parse_tokens [] (fun _ -> None) tokens in + let id_to_choices = + try + let choices_raw = req#param "choices" in + parse_choices choices_raw + with Http_types.Param_not_found _ -> (fun _ -> None) + in + let module Chat: Disambiguate.Callbacks = + struct + + let output_html = prerr_endline + + let interactive_user_uri_choice + ~selection_mode ?ok + ?enable_button_for_non_vars ~(title: string) ~(msg: string) + ~(id: string) (choices: string list) + = + (match id_to_choices id with + | Some choices -> choices + | None -> + let msg = Pcre.replace ~pat:"\"" ~templ:"\\\"" msg in + (match selection_mode with + | `SINGLE -> assert false + | `EXTENDED -> + iter_file + (fun line -> + let formatted_choices = + String.concat "," + (List.map (fun uri -> sprintf "\"%s\"" uri) choices) + in + let processed_line = + apply_substs + [title_tag_RE, title; + choices_tag_RE, formatted_choices; + msg_tag_RE, msg; + id_to_uris_RE, id_to_uris_raw; + id_RE, id] + line + in + output_string outchan processed_line) + interactive_user_uri_choice_TPL; + raise Chat_unfinished)) + + let interactive_interpretation_choice interpretations = + let html_interpretations_labels = + String.concat ", " + (List.map + (fun l -> + "\"" ^ + (String.concat "
" + (List.map + (fun (id, value) -> + (sprintf "alias %s %s" id value)) + l)) ^ + "\"") + interpretations) + in + let html_interpretations = + String.concat ", " + (List.map + (fun l -> + "\"" ^ + (String.concat " " + (List.map + (fun (id, value) -> + (sprintf "alias %s %s" + id + (Disambiguate.wrong_xpointer_format_from_wrong_xpointer_format' + value))) + l)) ^ + "\"") + interpretations) + in + iter_file + (fun line -> + let processed_line = + apply_substs + [interpretations_RE, html_interpretations; + interpretations_labels_RE, html_interpretations_labels] + line + in + output_string outchan processed_line) + interactive_interpretation_choice_TPL; + raise Chat_unfinished + + let input_or_locate_uri ~title = + UriManager.uri_of_string "cic:/Coq/Init/DataTypes/nat_ind.con" + + end + in + let module Disambiguate' = Disambiguate.Make (Chat) in + let (id_to_uris', metasenv', term') = + Disambiguate'.disambiguate_input + context metasenv dom mk_metasenv_and_expr id_to_uris + in + (match metasenv' with + | [] -> + let must = MQueryLevels2.get_constraints term' in + let must',only = refine_constraints must in + let results = MQueryGenerator.searchPattern must' only in + debug_print "FASE 3"; + iter_file + (fun line -> + let new_aliases = + match id_to_uris' with + | (domain, f) -> + String.concat ", " + (List.map + (fun name -> + sprintf "\"alias %s cic:%s\"" + name + (match f name with + | None -> assert false + | Some t -> + Disambiguate.string_of_cic_textual_parser_uri + t)) + domain) + in + let processed_line = + apply_substs + [results_RE, MQueryUtil.text_of_result results "\n"; + new_aliases_RE, new_aliases] + line + in + output_string outchan processed_line) + final_results_TPL + | _ -> (* unable to instantiate some implicit variable *) Http_daemon.respond - ~body:(pp_error - "identifiers resolution in the environment not yet implemented") + ~headers:[contype] + ~body:"some implicit variables are still unistantiated :-(" 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 (sprintf "%s done!" req#path) with - | Http_request.Param_not_found attr_name -> + | Chat_unfinished -> prerr_endline "Chat unfinished, Try again!" + | Http_types.Param_not_found attr_name -> bad_request (sprintf "Parameter '%s' is missing" attr_name) outchan - | Failure "int_of_string" -> - bad_request "Invalid 'precision' value, must be an integer" outchan | exc -> Http_daemon.respond ~body:(pp_error ("Uncaught exception: " ^ (Printexc.to_string exc))) @@ -101,6 +305,7 @@ in printf "%s started and listening on port %d\n" daemon_name port; printf "current directory is %s\n" (Sys.getcwd ()); flush stdout; +Unix.putenv "http_proxy" ""; Mqint.set_database Mqint.postgres_db; Mqint.init "host=mowgli.cs.unibo.it dbname=helm_mowgli_new_schema user=helm"; Http_daemon.start' ~port callback; -- 2.39.2