X-Git-Url: http://git.annexia.org/?a=blobdiff_plain;f=scripts%2Fpage.ml;h=a98e688358c070b5f1001a8ae26ca0199a51beb0;hb=eeb304015c65ccf593a77058d5db5f5a3e3b45d5;hp=03c3537ab6e7e337eff0aea631393979836ca0cd;hpb=0d88075fa62c0a801ea1e9246d85389f9161a4e1;p=cocanwiki.git diff --git a/scripts/page.ml b/scripts/page.ml index 03c3537..a98e688 100644 --- a/scripts/page.ml +++ b/scripts/page.ml @@ -1,7 +1,7 @@ (* COCANWIKI - a wiki written in Objective CAML. * Written by Richard W.M. Jones . * Copyright (C) 2004 Merjis Ltd. - * $Id: page.ml,v 1.17 2004/09/22 10:19:26 rich Exp $ + * $Id: page.ml,v 1.24 2004/09/27 18:08:02 rich Exp $ * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -30,6 +30,7 @@ open Cocanwiki open Cocanwiki_template open Cocanwiki_ok open Cocanwiki_date +open Cocanwiki_server_settings (* Maximum level of redirection. *) let max_redirect = 4 @@ -38,6 +39,20 @@ type fp_status = FPOK of int * string * string * Dbi.datetime * bool | FPRedirect of string | FPNotFound +(* Referer strings which help us decide if the user came from + * a search engine and highlight terms in the page appropriately. + *) +let search_engines = [ + Pcre.regexp "^http://.*google\\.", [ "q"; "as_q"; "as_epq"; "as_oq" ]; + Pcre.regexp "^http://.*yahoo\\.", [ "p" ]; + Pcre.regexp "^http://.*msn\\.", [ "q"; "MT" ] +] +let split_words = Pcre.regexp "\\W+" + +let split_qs_re = Pcre.regexp "\\?" + +let xhtml_re = Pcre.regexp "<.*?>|[^<>]+" + let run r (q : cgi) (dbh : Dbi.connection) hostid {edit_anon=edit_anon} user = let template_page = get_template dbh hostid "page.html" in let template_404 = get_template dbh hostid "page_404.html" in @@ -47,13 +62,15 @@ let run r (q : cgi) (dbh : Dbi.connection) hostid {edit_anon=edit_anon} user = (* Host-specific fields. *) let sth = dbh#prepare_cached "select css is not null, - feedback_email is not null + feedback_email is not null, + mailing_list, search_box from hosts where id = ?" in sth#execute [`Int hostid]; - let has_host_css, has_feedback_email = + let has_host_css, has_feedback_email, mailing_list, search_box = match sth#fetch1 () with - | [ `Bool has_host_css; `Bool has_feedback_email ] -> - has_host_css, has_feedback_email + | [ `Bool has_host_css; `Bool has_feedback_email; `Bool mailing_list; + `Bool search_box ] -> + has_host_css, has_feedback_email, mailing_list, search_box | _ -> assert false in (* Can the user edit? Manage users? etc. *) @@ -63,6 +80,79 @@ let run r (q : cgi) (dbh : Dbi.connection) hostid {edit_anon=edit_anon} user = let can_manage_site = can_manage_site user in let can_edit_global_css = can_edit_global_css user in + (* Do we have a stats page set up? *) + let has_stats = server_settings_stats_page dbh <> None in + + (* Given the referer string, return the list of search terms. If none + * can be found, then throws Not_found. + *) + let search_terms_from_referer referer = + let _, argnames = + List.find (fun (rex, _) -> Pcre.pmatch ~rex referer) search_engines in + let url, qs = + match Pcre.split ~rex:split_qs_re ~max:2 referer with + | [url] | [url;""] -> url, "" + | [url;qs] -> url, qs + | _ -> assert false in + let args = Cgi_args.parse qs in + let argname = + List.find (fun argname -> List.mem_assoc argname args) argnames in + let search_string = List.assoc argname args in + Pcre.split ~rex:split_words search_string + in + + (* Given a full page of XHTML, highlight search terms found in the + * part of the page. + *) + let highlight_search_terms xhtml search_terms span_class = + (* Split the original XHTML into strings and tags. For example if + * the original string is: "This is some bold text.
", then + * after this step we will have the following list: + * [ "This is some "; ""; "bold"; ""; " text."; "
" ] + *) + let xhtml = Pcre.extract_all ~rex:xhtml_re xhtml in + let xhtml = Array.to_list xhtml in + let xhtml = List.map (fun matches -> matches.(0)) xhtml in + + (* Find the ... tags. We only want to apply + * highlighting to tags within this area. + *) + let rec list_split f acc = function + | [] -> List.rev acc, [] + | ((x :: _) as xs) when f x -> List.rev acc, xs + | x :: xs -> + let acc = x :: acc in + list_split f acc xs + in + let head, body = + list_split (fun str -> String.starts_with str "") [] body in + (* NB: Hopefully, xhtml = head @ body @ tail. *) + + (* The search terms are a list of simple words. Turn into a big + * regular expression, because we want to substitute for each. We + * end up with a regexp like '(word1|word2|word3)'. + *) + let rex = + Pcre.regexp ~flags:[`CASELESS] + ("(" ^ String.concat "|" search_terms ^ ")") in + + (* Do the substitution, but only on text, not elements! *) + let body = + let subst text = + "" ^ text ^ "" + in + List.map (fun str -> + if String.length str > 0 && str.[0] != '<' then + Pcre.substitute ~rex ~subst str + else + str) body in + + (* Join the XHTML fragments back together again. *) + String.concat "" (List.concat [ head ; body ; tail ]) + in + (* This code generates ordinary pages. *) let make_page title description pageid last_modified_date has_page_css version page page' = @@ -85,6 +175,8 @@ let run r (q : cgi) (dbh : Dbi.connection) hostid {edit_anon=edit_anon} user = t#conditional "has_page_css" has_page_css; t#conditional "has_feedback_email" has_feedback_email; + t#conditional "mailing_list" mailing_list; + t#conditional "search_box" search_box; t#conditional "can_edit" can_edit; t#conditional "can_manage_users" can_manage_users; @@ -92,6 +184,8 @@ let run r (q : cgi) (dbh : Dbi.connection) hostid {edit_anon=edit_anon} user = t#conditional "can_manage_site" can_manage_site; t#conditional "can_edit_global_css" can_edit_global_css; + t#conditional "has_stats" has_stats; + (* Pull out the sections in this page. *) let sth = dbh#prepare_cached "select ordering, sectionname, content, divname @@ -144,7 +238,24 @@ let run r (q : cgi) (dbh : Dbi.connection) hostid {edit_anon=edit_anon} user = t#conditional "user_logged_in" true; t#set "username" username); - q#template t + (* If we are coming from a search engine then we want to highlight + * search terms throughout the whole page ... + *) + try + let referer = Table.get (Request.headers_in r) "Referer" in + let search_terms = search_terms_from_referer referer in + + (* Highlight the search terms. *) + let xhtml = t#to_string in + let xhtml = highlight_search_terms xhtml search_terms "search_term" in + + (* Deliver the page. *) + q#header (); + print_string r xhtml + with + Not_found -> + (* No referer / no search terms / not a search engine referer. *) + q#template t in (* This code generates 404 pages. *) @@ -170,6 +281,8 @@ let run r (q : cgi) (dbh : Dbi.connection) hostid {edit_anon=edit_anon} user = t#conditional "can_manage_site" can_manage_site; t#conditional "can_edit_global_css" can_edit_global_css; + t#conditional "has_stats" has_stats; + q#template t in @@ -254,7 +367,7 @@ let run r (q : cgi) (dbh : Dbi.connection) hostid {edit_anon=edit_anon} user = error ~title:"Too many redirections" ~back_button:true q ("Too many redirects between pages. This may happen because " ^ "of a cycle of redirections."); - raise CgiExit + return () ) else match fetch_page page' version allow_redirect with | FPOK (pageid, title, description, last_modified_date, has_page_css)->