(* COCANWIKI - a wiki written in Objective CAML.
* Written by Richard W.M. Jones <rich@merjis.com>.
* Copyright (C) 2004 Merjis Ltd.
- * $Id: page.ml,v 1.19 2004/09/23 15:16:21 rich Exp $
+ * $Id: page.ml,v 1.32 2004/10/10 19:19:58 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
open Printf
open ExtString
+open ExtList
open Cocanwiki
open Cocanwiki_template
open Cocanwiki_ok
open Cocanwiki_date
open Cocanwiki_server_settings
-
-(* Maximum level of redirection. *)
-let max_redirect = 4
+open Cocanwiki_links
type fp_status = FPOK of int * string * string * Dbi.datetime * bool
| FPRedirect of string
| FPNotFound
-let run r (q : cgi) (dbh : Dbi.connection) hostid {edit_anon=edit_anon} user =
+(* 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;
+ view_anon = view_anon } as host)
+ user =
let template_page = get_template dbh hostid "page.html" in
let template_404 = get_template dbh hostid "page_404.html" in
(* 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, navigation
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, navigation =
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; `Bool navigation ] ->
+ has_host_css, has_feedback_email, mailing_list, search_box,
+ navigation
| _ -> assert false in
(* Can the user edit? Manage users? etc. *)
- let can_edit = can_edit edit_anon user in
- let can_manage_users = can_manage_users user in
- let can_manage_contacts = can_manage_contacts user in
- let can_manage_site = can_manage_site user in
- let can_edit_global_css = can_edit_global_css user in
+ let can_edit = can_edit host user in
+ let can_manage_users = can_manage_users host user in
+ let can_manage_contacts = can_manage_contacts host user in
+ let can_manage_site = can_manage_site host user in
+ let can_edit_global_css = can_edit_global_css host 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
+ * <body> 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 <b>bold</b> text.<br/>", then
+ * after this step we will have the following list:
+ * [ "This is some "; "<b>"; "bold"; "</b>"; " text."; "<br/>" ]
+ *)
+ 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 <body> ... </body> 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") [] xhtml in
+ let body, tail =
+ list_split ((=) "</body>") [] 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 =
+ "<span class=\"" ^ span_class ^ "\">" ^ text ^ "</span>"
+ 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
+
+ (* Check the templates table for extensions. *)
+ let get_extension url =
+ let sth = dbh#prepare_cached "select extension from templates
+ where ? ~ url_regexp
+ order by ordering
+ limit 1" in
+ sth#execute [`String url];
+
+ try
+ let name = sth#fetch1string () in
+ Some (List.assoc name !extensions)
+ with
+ Not_found -> None
+ in
+
(* This code generates ordinary pages. *)
let make_page title description pageid last_modified_date has_page_css
- version page page' =
+ version page page' extension =
let t = template_page in
t#set "title" title;
- t#set "description" description;
- t#set "pageid" (string_of_int pageid);
- t#set "last_modified_date" (printable_date last_modified_date);
+ t#set "last_modified_date" last_modified_date;
+
+ (match description with
+ None -> t#conditional "has_description" false
+ | Some description ->
+ t#conditional "has_description" true;
+ t#set "description" description);
if page <> page' then (* redirection *) (
t#set "page" page';
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 "navigation" navigation;
t#conditional "can_edit" can_edit;
t#conditional "can_manage_users" can_manage_users;
t#conditional "has_stats" has_stats;
(* Pull out the sections in this page. *)
- let sth = dbh#prepare_cached
- "select ordering, sectionname, content, divname
- from contents
- where pageid = ?
- order by ordering" in
- sth#execute [`Int pageid];
-
let sections =
- sth#map
- (function [`Int ordering;
- (`Null | `String _) as sectionname;
- `String content;
- (`Null | `String _) as divname] ->
- let divname, has_divname =
- match divname with
- `Null -> "", false
- | `String divname -> divname, true in
- let sectionname, has_sectionname =
- match sectionname with
- `Null -> "", false
- | `String sectionname -> sectionname, true in
- let linkname = linkname_of_sectionname sectionname in
- [ "ordering", Template.VarString (string_of_int ordering);
- "has_sectionname", Template.VarConditional has_sectionname;
- "sectionname", Template.VarString sectionname;
- "linkname", Template.VarString linkname;
- "content",
- Template.VarString
- (Wikilib.xhtml_of_content dbh hostid content);
- "has_divname", Template.VarConditional has_divname;
- "divname", Template.VarString divname ]
- | _ -> assert false) in
+ match pageid with
+ None -> []
+ | Some pageid ->
+ let sth = dbh#prepare_cached
+ "select ordering, sectionname, content, divname
+ from contents where pageid = ? order by ordering" in
+ sth#execute [`Int pageid];
+
+ sth#map
+ (function [`Int ordering;
+ (`Null | `String _) as sectionname;
+ `String content;
+ (`Null | `String _) as divname] ->
+ let divname, has_divname =
+ match divname with
+ `Null -> "", false
+ | `String divname -> divname, true in
+ let sectionname, has_sectionname =
+ match sectionname with
+ `Null -> "", false
+ | `String sectionname -> sectionname, true in
+ let linkname = linkname_of_sectionname sectionname in
+ [ "ordering", Template.VarString (string_of_int ordering);
+ "has_sectionname",
+ Template.VarConditional has_sectionname;
+ "sectionname", Template.VarString sectionname;
+ "linkname", Template.VarString linkname;
+ "content",
+ Template.VarString
+ (Wikilib.xhtml_of_content dbh hostid content);
+ "has_divname", Template.VarConditional has_divname;
+ "divname", Template.VarString divname ]
+ | _ -> assert false) in
+
+ (* Call an extension to generate the first section in this page? *)
+ let sections =
+ match extension with
+ None -> sections
+ | Some extension ->
+ let content = extension dbh hostid page' in
+ let section = [
+ "ordering", Template.VarString "0";
+ "has_sectionname", Template.VarConditional false;
+ "linkname", Template.VarString "";
+ "content", Template.VarString content;
+ "has_divname", Template.VarConditional true;
+ "divname", Template.VarString "form_div";
+ ] in
+ section :: sections in
t#table "sections" sections;
t#conditional "user_logged_in" true;
t#set "username" username);
- q#template t
+ (* If logged in, we want to update the recently_visited table. *)
+ if pageid <> None then (
+ match user with
+ | User (userid, _, _) ->
+ let sth = dbh#prepare_cached "delete from recently_visited
+ where hostid = ? and userid = ?
+ and url = ?" in
+ sth#execute [`Int hostid; `Int userid; `String page'];
+ let sth = dbh#prepare_cached
+ "insert into recently_visited (hostid, userid, url)
+ values (?, ?, ?)" in
+ sth#execute [`Int hostid; `Int userid; `String page'];
+ dbh#commit ()
+ | _ -> ()
+ );
+
+ (* Navigation links. *)
+ if navigation then (
+ let max_links = 18 in (* Show no more links than this. *)
+
+ (* What links here. *)
+ let wlh = what_links_here dbh hostid page' in
+ let wlh = List.take max_links wlh in
+ let wlh_urls = List.map fst wlh in (* Just the URLs ... *)
+
+ let rv =
+ match user with
+ | User (userid, _, _) ->
+ (* Recently visited URLs, but don't repeat any from the 'what
+ * links here' section, and don't link to self.
+ *)
+ let not_urls = page' :: wlh_urls in
+ let limit = max_links - List.length wlh_urls in
+ let qs = Dbi.placeholders (List.length not_urls) in
+ let sth =
+ dbh#prepare_cached
+ ("select rv.url, p.title, rv.visit_time
+ from recently_visited rv, pages p
+ where rv.hostid = ? and rv.userid = ?
+ and rv.url not in " ^ qs ^ "
+ and rv.hostid = p.hostid and rv.url = p.url
+ order by 3 desc
+ limit ?") in
+ let args = List.map (fun s -> `String s) not_urls in
+ sth#execute
+ ([`Int hostid; `Int userid] @ args @ [`Int limit]);
+ sth#map
+ (function [`String url; `String title; _] ->
+ url, title
+ | _ -> assert false)
+ | _ -> [] in
+
+ (* Links to page. *)
+ let f (page, title) = [ "page", Template.VarString page;
+ "title", Template.VarString title ] in
+ let table = List.map f wlh in
+ t#table "what_links_here" table;
+ t#conditional "has_what_links_here" (wlh <> []);
+
+ let table = List.map f rv in
+ t#table "recently_visited" table;
+ t#conditional "has_recently_visited" (rv <> []);
+
+ (* If both lists are empty (ie. an empty navigation box would
+ * appear), then disable navigation altogether.
+ *)
+ if wlh = [] && rv = [] then t#conditional "navigation" false
+ );
+
+ (* 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. *)
(* Fetch a page by name. This function can give three answers:
* (1) Page fetched OK (fetches some details of the page).
* (2) Page is a redirect (fetches the name of the redirect page).
- * (3) Page not found in database, ie. 404 error.
+ * (3) Page not found in database, could be template or 404 error.
*)
(* XXX Should do a case-insensitive matching of URLs, and if the URL differs
* in case only should redirect to the lowercase version.
) else
match fetch_page page' version allow_redirect with
| FPOK (pageid, title, description, last_modified_date, has_page_css)->
- make_page title description pageid last_modified_date has_page_css
- version page page'
+ (* Check if the page is also a template. *)
+ let extension = get_extension page' in
+ make_page title (Some description) (Some pageid)
+ (printable_date last_modified_date) has_page_css
+ version page page' extension
| FPRedirect page' ->
loop page' (i+1)
| FPNotFound ->
- make_404 ()
+ (* Might be a templated page with no content in it. *)
+ let extension = get_extension page' in
+ (match extension with
+ | (Some _) as extension ->
+ let title = page' in
+ make_page title None None
+ "Now" false None page page'
+ extension
+ | None ->
+ make_404 ())
in
loop page 0
let () =
- register_script run
+ register_script ~restrict:[CanView] run