1 (* COCANWIKI - a wiki written in Objective CAML.
2 * Written by Richard W.M. Jones <rich@merjis.com>.
3 * Copyright (C) 2004 Merjis Ltd.
4 * $Id: page.ml,v 1.54 2006/08/14 15:53:52 rich Exp $
6 * This program is free software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License as published by
8 * the Free Software Foundation; either version 2 of the License, or
9 * (at your option) any later version.
11 * This program is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 * GNU General Public License for more details.
16 * You should have received a copy of the GNU General Public License
17 * along with this program; see the file COPYING. If not, write to
18 * the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
19 * Boston, MA 02111-1307, USA.
31 open Cocanwiki_template
34 open Cocanwiki_server_settings
36 open Cocanwiki_extensions
37 open Cocanwiki_strings
40 | FPOK of int32 * string * string * string option * Calendar.t * bool
42 | FPInternalRedirect of string
43 | FPExternalRedirect of string
46 (* Referer strings which help us decide if the user came from
47 * a search engine and highlight terms in the page appropriately.
49 let search_engines = [
50 Pcre.regexp "^http://.*google\\.", [ "q"; "as_q"; "as_epq"; "as_oq" ];
51 Pcre.regexp "^http://.*yahoo\\.", [ "p" ];
52 Pcre.regexp "^http://.*msn\\.", [ "q"; "MT" ]
54 let split_words = Pcre.regexp "\\W+"
56 let split_qs_re = Pcre.regexp "\\?"
58 let xhtml_re = Pcre.regexp "<.*?>|[^<>]+"
60 let run r (q : cgi) dbh hostid
61 ({ edit_anon = edit_anon; view_anon = view_anon } as host)
63 let page = q#param "page" in
64 let page = if page = "" then "index" else page in
66 (* The main "page" template is split in two to improve the speed of
67 * delivery of the page. The very first part ("page_header.html")
68 * contains the page <head>, crucially including all the links to the
69 * stylesheets. We send this first and flush it out to the client so
70 * that the client can begin requesting stylesheets, background images
71 * and so on. After this we compose the main page ("page.html") and
75 let template_page_header =
76 get_template ~page dbh hostid "page_header.html" in
77 let template_page = get_template ~page dbh hostid "page.html" in
79 (* This is the simpler template for 404 pages. *)
80 let template_404_header = get_template dbh hostid "page_404_header.html" in
81 let template_404 = get_template dbh hostid "page_404.html" in
83 (* Host-specific fields. *)
86 "select feedback_email is not null, mailing_list, navigation
87 from hosts where id = $hostid" in
88 let has_feedback_email, mailing_list, navigation =
90 | [Some has_feedback_email, mailing_list, navigation] ->
91 has_feedback_email, mailing_list, navigation
92 | _ -> assert false in
94 (* User permissions. *)
95 let can_edit = can_edit host user in
96 let can_manage_users = can_manage_users host user in
98 (* Do we have a stats page set up? *)
99 let has_stats = server_settings_stats_page dbh <> None in
101 (* Given the referer string, return the list of search terms. If none
102 * can be found, then throws Not_found.
104 let search_terms_from_referer referer =
106 List.find (fun (rex, _) -> Pcre.pmatch ~rex referer) search_engines in
108 match Pcre.split ~rex:split_qs_re ~max:2 referer with
109 | [url] | [url;""] -> url, ""
110 | [url;qs] -> url, qs
111 | _ -> assert false in
112 let args = Cgi_args.parse qs in
114 List.find (fun argname -> List.mem_assoc argname args) argnames in
115 let search_string = List.assoc argname args in
116 Pcre.split ~rex:split_words search_string
119 (* Given a full page of XHTML, highlight search terms found in the
120 * <body> part of the page.
122 let highlight_search_terms xhtml search_terms span_class =
123 (* Split the original XHTML into strings and tags. For example if
124 * the original string is: "This is some <b>bold</b> text.<br/>", then
125 * after this step we will have the following list:
126 * [ "This is some "; "<b>"; "bold"; "</b>"; " text."; "<br/>" ]
128 let xhtml = Pcre.extract_all ~rex:xhtml_re xhtml in
129 let xhtml = Array.to_list xhtml in
130 let xhtml = List.map (fun matches -> matches.(0)) xhtml in
132 (* Find the <body> ... </body> tags. We only want to apply
133 * highlighting to tags within this area.
135 let rec list_split f acc = function
136 | [] -> List.rev acc, []
137 | ((x :: _) as xs) when f x -> List.rev acc, xs
139 let acc = x :: acc in
143 list_split (fun str -> String.starts_with str "<body") [] xhtml in
145 list_split ((=) "</body>") [] body in
146 (* NB: Hopefully, xhtml = head @ body @ tail. *)
148 (* The search terms are a list of simple words. Turn into a big
149 * regular expression, because we want to substitute for each. We
150 * end up with a regexp like '(word1|word2|word3)'.
153 Pcre.regexp ~flags:[`CASELESS]
154 ("(" ^ String.concat "|" search_terms ^ ")") in
156 (* Do the substitution, but only on text, not elements! *)
159 "<span class=\"" ^ span_class ^ "\">" ^ text ^ "</span>"
162 if String.length str > 0 && str.[0] != '<' then
163 Pcre.substitute ~rex ~subst str
167 (* Join the XHTML fragments back together again. *)
168 String.concat "" (List.concat [ head ; body ; tail ])
171 (* Check the templates table for extensions. *)
172 let get_extension url =
176 PGSQL(dbh) "select extension from templates
177 where $url ~ url_regexp
181 Some (List.assoc name !extensions)
183 Not_found | ExtList.List.Empty_list | Failure "hd" -> None
186 (* This code generates ordinary pages. *)
187 let make_page title description keywords
188 pageid last_modified_date has_page_css noodp
189 version page page' extension =
190 let t = template_page in
191 let th = template_page_header in
193 th#set "title" title;
194 t#set "last_modified_date" last_modified_date;
196 (match description with
197 None -> th#conditional "has_description" false
198 | Some description ->
199 th#conditional "has_description" true;
200 th#set "description" description);
203 None -> th#conditional "has_keywords" false
205 th#conditional "has_keywords" true;
206 th#set "keywords" keywords);
208 if page <> page' then (* redirection *) (
211 t#set "original_page" page; (* XXX title - get it from database *)
212 t#conditional "redirected" true
216 t#conditional "redirected" false
219 th#conditional "has_page_css" has_page_css;
221 (* If the per-page noodp is not null, set the noodp flag here. Otherwise
222 * we will use the default (from hosts.global_noodp) which was set
223 * in Cocanwiki_template.
227 | Some b -> th#conditional "noodp" b);
229 (* Are we showing an old version of the page? If so, warn. *)
232 t#conditional "is_old_version" false;
233 th#conditional "is_old_version" false
235 t#conditional "is_old_version" true;
236 th#conditional "is_old_version" true;
237 t#set "old_version" (Int32.to_string pageid);
238 th#set "old_version" (Int32.to_string pageid));
240 (* Just before we show the header, call any registered pre-page
241 * handlers. They might want to send cookies.
243 List.iter (fun handler ->
244 handler r q dbh hostid page') !pre_page_handlers;
246 (* At this point, we can print out the header and flush it back to
247 * the user, allowing the browser to start fetching stylesheets
248 * and background images while we compose the page.
251 ignore (print_string r th#to_string);
252 ignore (Request.rflush r);
254 t#conditional "has_feedback_email" has_feedback_email;
255 t#conditional "mailing_list" mailing_list;
256 t#conditional "navigation" navigation;
258 t#conditional "can_edit" can_edit;
259 t#conditional "can_manage_users" can_manage_users;
260 t#conditional "has_stats" has_stats;
262 (* Pull out the sections in this page. *)
267 let rows = PGSQL(dbh)
268 "select ordering, sectionname, content, divname, jsgo
269 from contents where pageid = $pageid order by ordering" in
272 (fun (ordering, sectionname, content, divname, jsgo) ->
273 let divname, has_divname =
276 | Some divname -> divname, true in
280 | Some jsgo -> jsgo, true in
281 let sectionname, has_sectionname =
282 match sectionname with
284 | Some sectionname -> sectionname, true in
285 let linkname = linkname_of_sectionname sectionname in
286 [ "ordering", Template.VarString (Int32.to_string ordering);
288 Template.VarConditional has_sectionname;
289 "sectionname", Template.VarString sectionname;
290 "linkname", Template.VarString linkname;
293 (Wikilib.xhtml_of_content r dbh hostid content);
294 "has_divname", Template.VarConditional has_divname;
295 "divname", Template.VarString divname;
296 "has_jsgo", Template.VarConditional has_jsgo;
297 "jsgo", Template.VarString jsgo ]) rows in
299 (* Call an extension to generate the first section in this page? *)
304 let content = extension r dbh hostid page' in
306 "ordering", Template.VarString "0";
307 "has_sectionname", Template.VarConditional false;
308 "linkname", Template.VarString "";
309 "content", Template.VarString content;
310 "has_divname", Template.VarConditional true;
311 "divname", Template.VarString "form_div";
312 "has_jsgo", Template.VarConditional false;
313 "jsgo", Template.VarString "";
315 section :: sections in
317 t#table "sections" sections;
322 t#conditional "user_logged_in" false
323 | User (_, username, _, _) ->
324 t#conditional "user_logged_in" true;
325 t#set "username" username);
327 (* Can anonymous users create accounts? If not them we don't
328 * want to offer to create accounts for them.
330 t#conditional "create_account_anon" host.create_account_anon;
332 (* If logged in, we want to update the recently_visited table. *)
333 if pageid <> None then (
335 | User (userid, _, _, _) ->
338 "delete from recently_visited
339 where hostid = $hostid and userid = $userid
342 "insert into recently_visited (hostid, userid, url)
343 values ($hostid, $userid, $page')";
347 (* Exceptions here are non-fatal. Just print them. *)
348 prerr_endline "exception updating recently_visited:";
349 prerr_endline (Printexc.to_string exn);
350 PGOCaml.rollback dbh;
352 PGOCaml.begin_work dbh;
356 (* Navigation links. *)
358 let max_links = 18 in (* Show no more links than this. *)
360 (* What links here. *)
361 let wlh = what_links_here dbh hostid page' in
362 let wlh = List.take max_links wlh in
363 let wlh_urls = List.map fst wlh in (* Just the URLs ... *)
367 | User (userid, _, _, _) ->
368 (* Recently visited URLs, but don't repeat any from the 'what
369 * links here' section, and don't link to self.
371 let not_urls = page' :: wlh_urls in
372 let limit = Int32.of_int (max_links - List.length wlh_urls) in
375 "select rv.url, p.title, rv.visit_time
376 from recently_visited rv, pages p
377 where rv.hostid = $hostid and rv.userid = $userid
378 and rv.url not in $@not_urls
379 and rv.hostid = p.hostid and rv.url = p.url
383 fun (url, title, _) -> url, title
388 let f (page, title) = [ "page", Template.VarString page;
389 "title", Template.VarString title ] in
390 let table = List.map f wlh in
391 t#table "what_links_here" table;
392 t#conditional "has_what_links_here" (wlh <> []);
394 let table = List.map f rv in
395 t#table "recently_visited" table;
396 t#conditional "has_recently_visited" (rv <> []);
398 (* If both lists are empty (ie. an empty navigation box would
399 * appear), then disable navigation altogether.
401 if wlh = [] && rv = [] then t#conditional "navigation" false
404 (* If we are coming from a search engine then we want to highlight
405 * search terms throughout the whole page ...
408 let referer = Table.get (Request.headers_in r) "Referer" in
409 let search_terms = search_terms_from_referer referer in
411 (* Highlight the search terms. *)
412 let xhtml = t#to_string in
413 let xhtml = highlight_search_terms xhtml search_terms "search_term" in
415 (* Deliver the page. *)
416 ignore (print_string r xhtml)
419 (* No referer / no search terms / not a search engine referer. *)
420 ignore (print_string r t#to_string)
423 (* This code generates 404 pages. *)
425 Request.set_status r 404; (* Return a 404 error code. *)
427 let th = template_404_header in
433 ('a'..'z' | 'A'..'Z' | '0'..'9') as c -> c
436 th#set "search_terms" search_terms;
438 (* Flush out the header while we start the search. *)
440 ignore (print_string r th#to_string);
441 ignore (Request.rflush r);
443 let t = template_404 in
444 t#set "query" search_terms;
445 t#set "canonical_hostname" host.canonical_hostname;
447 (* This is a simplified version of the code in search.ml. *)
449 (* Get the keywords from the query string. *)
450 let keywords = Pcre.split ~rex:split_words search_terms in
452 List.filter (fun s -> not (string_is_whitespace s)) keywords in
453 let keywords = List.map String.lowercase keywords in
455 (* Turn the keywords into a tsearch2 ts_query string. *)
456 let tsquery = String.concat "&" keywords in
458 (* Search the titles first. *)
461 "select url, title, last_modified_date,
462 (lower (title) = lower ($search_terms)) as exact
464 where hostid = $hostid
467 and title_description_fti @@ to_tsquery ('default', $tsquery)
468 order by exact desc, last_modified_date desc, title" in
472 | (Some url, title, last_modified, _) ->
473 url, title, last_modified
474 | _ -> assert false) rows in
476 let have_titles = titles <> [] in
477 t#conditional "have_titles" have_titles;
479 (* Search the contents. *)
482 "select c.id, p.url, p.title, p.last_modified_date
483 from contents c, pages p
484 where c.pageid = p.id
485 and p.hostid = $hostid
487 and p.redirect is null
488 and c.content_fti @@ to_tsquery ('default', $tsquery)
489 order by p.last_modified_date desc, p.title
494 | (contentid, Some url, title, last_modified) ->
495 contentid, url, title, last_modified
496 | _ -> assert false) rows in
498 let have_contents = contents <> [] in
499 t#conditional "have_contents" have_contents;
501 (* Pull out the actual text which matched so we can generate a summary.
502 * XXX tsearch2 can actually do better than this by emboldening
503 * the text which maps.
506 if contents = [] then []
510 List.map (fun (contentid, _,_,_) -> contentid) contents in
512 "select id, sectionname, content from contents
513 where id in $@contentids" in
514 List.map (fun (id, sectionname, content) ->
515 id, (sectionname, content)) rows
518 (* Generate the final tables. *)
520 List.map (fun (url, title, last_modified) ->
521 let last_modified = printable_date last_modified in
522 [ "url", Template.VarString url;
523 "title", Template.VarString title;
524 "last_modified", Template.VarString last_modified ]
526 t#table "titles" table;
530 (fun (contentid, url, title, last_modified) ->
531 let sectionname, content = List.assoc contentid content_map in
532 let have_sectionname, sectionname =
533 match sectionname with
535 | Some sectionname -> true, sectionname in
538 (Wikilib.text_of_xhtml
539 (Wikilib.xhtml_of_content r dbh hostid content)) in
540 let linkname = linkname_of_sectionname sectionname in
541 let last_modified = printable_date last_modified in
542 [ "url", Template.VarString url;
543 "title", Template.VarString title;
544 "have_sectionname", Template.VarConditional have_sectionname;
545 "sectionname", Template.VarString sectionname;
546 "linkname", Template.VarString linkname;
547 "content", Template.VarString content;
548 "last_modified", Template.VarString last_modified ]
550 t#table "contents" table;
552 (* Do we have any results? *)
553 let have_results = have_titles || have_contents in
555 t#conditional "have_results" have_results;
557 (* Deliver the rest of the page. *)
558 ignore (print_string r t#to_string)
561 (* Fetch a page by name. This function can give three answers:
562 * (1) Page fetched OK (fetches some details of the page).
563 * (2) Page is a redirect (fetches the name of the redirect page).
564 * (3) Page not found in database, could be template or 404 error.
566 let fetch_page page version allow_redirect =
569 if allow_redirect then (
570 let rows = PGSQL(dbh)
571 "select url, redirect, id, title, description, keywords,
572 last_modified_date, css is not null, noodp
574 where hostid = $hostid and lower (url) = lower ($page)" in
576 | [Some page', _, _, _, _, _, _, _, _]
577 when page <> page' -> (* different case *)
578 FPExternalRedirect page'
579 | [ _, None, id, title, description, keywords,
580 last_modified_date, has_page_css, noodp ] ->
581 let has_page_css = Option.get has_page_css in
582 FPOK (id, title, description, keywords, last_modified_date,
584 | [_, Some redirect, _, _, _, _, _, _, _] ->
585 FPInternalRedirect redirect
588 ) else (* redirects not allowed ... *) (
589 let rows = PGSQL(dbh)
590 "select id, title, description, keywords, last_modified_date,
591 css is not null, noodp
592 from pages where hostid = $hostid and url = $page" in
594 | [ id, title, description, keywords,
595 last_modified_date, has_page_css, noodp ] ->
596 let has_page_css = Option.get has_page_css in
597 FPOK (id, title, description, keywords, last_modified_date,
603 let rows = PGSQL(dbh)
604 "select id, title, description, keywords, last_modified_date,
605 css is not null, noodp
607 where hostid = $hostid and id = $version and
608 (url = $page or url_deleted = $page)" in
610 | [ id, title, description, keywords,
611 last_modified_date, has_page_css, noodp ] ->
612 let has_page_css = Option.get has_page_css in
613 FPOK (id, title, description, keywords, last_modified_date,
619 (* Here we deal with the complex business of redirects and versions. *)
620 (* Only allow the no_redirect and version syntax for editors. *)
621 let allow_redirect, version =
623 not (q#param_true "no_redirect"),
624 try Some (Int32.of_string (q#param "version")) with Not_found -> None
628 let rec loop page' i =
629 if i > max_redirect then (
630 error ~title:"Too many redirections" ~back_button:true
632 ("Too many redirects between pages. This may happen because " ^
633 "of a cycle of redirections.");
636 match fetch_page page' version allow_redirect with
637 | FPOK (pageid, title, description, keywords,
638 last_modified_date, has_page_css, noodp)->
639 (* Check if the page is also a template. *)
640 let extension = get_extension page' in
641 make_page title (Some description) keywords (Some pageid)
642 (printable_date last_modified_date) has_page_css noodp
643 version page page' extension
644 | FPInternalRedirect page' ->
646 | FPExternalRedirect page' ->
647 (* This normally happens when a user has requested an uppercase
648 * page name. We redirect to the true (lowercase) version.
650 q#redirect ("http://" ^ host.hostname ^ "/" ^ page')
652 (* Might be a templated page with no content in it. *)
653 let extension = get_extension page' in
654 (match extension with
655 | (Some _) as extension ->
657 make_page title None None None
658 "Now" false None None page page'
666 register_script ~restrict:[CanView] run