big update to ORM-database driven website, no more mlcrunch hack

.gitignore

@@ -8,5 +8,14 @@
 .*.swp
 *.annot
 ._d/
+._bcdi/
+._ncdi/
+bibtex/bib2db
+fs/fs2db
 src/server
 src/static_*.ml
+content/
+bibtex/bibtex_lexer.ml
+bibtex/bibtex_parser.ml
+bibtex/bibtex_parser.mli
+

Makefile

@@ -1,7 +1,21 @@
 all:
-	@cd crunch && $(MAKE) all
-	@cd src && $(MAKE) all
+	@cd fs && $(MAKE) nc
+	@cd bibtex && $(MAKE)
+	@cd src && $(MAKE) 
+
+dbs: tdb sdb bdb
+
+sdb:
+	rm -f content/db/static.db
+	./fs/fs2db -db content/db/static.db content/static
+tdb:
+	rm -f content/db/tmpl.db
+	./fs/fs2db -db content/db/tmpl.db content/templates
+bdb:
+	rm -f content/db/bib.db
+	./bibtex/bib2db ./content/biblio.bib content/db/bib.db
 
 clean:
-	@cd crunch && $(MAKE) clean
+	@cd fs && $(MAKE) clean
 	@cd src && $(MAKE) clean
+	rm -f content/db/bib.db content/db/static.db content/db/tmpl.db

bibtex/Makefile

@@ -0,0 +1,8 @@
+OCAMLMAKEFILE = ../OCamlMakefile
+ANNOTATE = yes
+SOURCES = bibtex.ml bib.ml bibtex_parser.mly bibtex_lexer.mll bib2db.ml
+RESULT = bib2db
+PREDS = camlp4o
+PACKS = orm.syntax json-tc.syntax
+
+-include $(OCAMLMAKEFILE)

bibtex/bib.ml

@@ -0,0 +1,65 @@
+open Printf
+
+type t =
+  |Book 
+  |In_proceedings 
+  |In_book 
+  |In_collection
+  |Article 
+  |Proceedings
+  |Webpage
+  |Tech_report 
+  |Phd_thesis
+  |Masters_thesis
+  |Unpublished 
+  |Misc
+and ent = {
+  ty: t;
+  key: string;
+  authors: string list;
+  year: int option;
+  title: string;
+  misc: (string * string) list
+} with orm
+
+let t_of_string = function
+  |"book" -> Book 
+  |"inproceedings" -> In_proceedings 
+  |"article" -> Article
+  |"proceedings" -> Proceedings 
+  |"webpage" -> Webpage 
+  |"techreport" -> Tech_report
+  |"phdthesis" -> Phd_thesis
+  |"mastersthesis" -> Masters_thesis
+  |"unpublished" -> Unpublished 
+  |"misc" -> Misc
+  |"inbook" -> In_book 
+  |"incollection" -> In_collection
+  |x -> failwith ("unknown bibtex type: " ^ x)
+
+let author_html a = a
+
+let ent_to_summary_html e =
+  let authors = match List.rev e.authors with
+   | [] -> "XXX author"
+   | [a] -> a
+   | a :: tl ->
+      String.concat ", " (List.map author_html (List.rev tl)) ^ " and " ^ (author_html a) in
+  let year = match e.year with 
+   | None -> "XXX year"
+   | Some y -> string_of_int y in
+  let key x = try List.assoc x e.misc with Not_found -> "XXX " ^ x in
+  match e.ty with
+  |In_proceedings ->
+     sprintf "<b>%s</b>, %s, <i>%s</i>" e.title authors (key "booktitle")
+  |Article 
+  |Book 
+  |In_book 
+  |In_collection
+  |Proceedings
+  |Webpage
+  |Tech_report 
+  |Phd_thesis
+  |Masters_thesis
+  |Unpublished 
+  |Misc -> sprintf "<b>%s</b>" e.title

bibtex/bib2db.ml

@@ -0,0 +1,70 @@
+open Printf
+
+let read_bibtex f =
+  let ch = open_in f in
+  let lb = Lexing.from_channel ch in
+  try
+    let el = Bibtex_parser.command_list Bibtex_lexer.token lb in
+    eprintf "ok (%d entries).\n%!" (Bibtex.size el);
+    el
+  with
+    Parsing.Parse_error | Failure "unterminated string" -> 
+       (eprintf "Parse error character %d, in or after entry '%s'.\n%!" (Lexing.lexeme_start lb) !Bibtex.current_key;
+       exit 1)
+
+open Bibtex
+
+let atom = function |Id i -> "id " ^ i |String s -> "str " ^ s
+let atoms a = String.concat ", " (List.map atom a)
+let output_bibtex c () =
+  match c with
+  |Comment c -> ()
+  |Preamble al -> printf "preamble: %s\n" (atoms al)
+  |Abbrev (t,al) -> printf "abbr: %s %s\n" t (atoms al)
+  |Entry (ety, key, el) ->
+     printf "e: %s[%s] -> %s \n" ety key
+      (String.concat "\n" (List.map (fun (k,v) -> sprintf "  %s=%s" k (atoms v)) el))
+
+let string_of_atom = function
+  | Id x -> x
+  | String x -> Pcre.replace ~rex:(Pcre.regexp "[{|}]") ~itempl:(Pcre.subst "") x
+
+let normalize_author x =
+  (* Blogg, Joe -> Joe Blogg *)
+  let x = String.concat " " (
+    List.rev (Pcre.split ~rex:(Pcre.regexp ", *") x)) in
+  (* strip edge spaces *)
+  let itempl = Pcre.subst "" in
+  let x = Pcre.replace ~rex:(Pcre.regexp "^ +") ~itempl x in
+  Pcre.replace ~rex:(Pcre.regexp " +$") ~itempl x
+
+let string_of_atoms x =
+  String.concat " " (List.map string_of_atom x)
+
+let bib_of_bibtex = function
+  | Comment _ | Preamble _ | Abbrev _ -> None
+  | Entry (ety, key, el) as bt ->
+      let t = Bib.t_of_string ety in
+      let try_assoc k = try string_of_atoms (List.assoc k el) with Not_found -> "" in
+      let authors = List.map normalize_author (Pcre.split ~pat:" and " (try_assoc "author")) in
+      let year = try Some (int_of_string (try_assoc "year")) with _ -> None in
+      let title = try_assoc "title" in
+      let misc = List.map (fun (k,v) -> (k, (string_of_atoms v))) (List.filter (fun (k,_) -> match k with "author" |"year" |"title" -> false |_ -> true) el) in
+      Some { Bib.ty=t; key=key; authors=authors; year=year; title=title; misc=misc }
+
+let output_bibtexs cl =
+  fold output_bibtex cl ()
+
+let save_bibtexs bs dbname =
+  let db = Bib.ent_init dbname in
+  List.iter (fun c -> printf ".%!"; Bib.ent_save db c) bs
+
+let _ =
+  match Sys.argv with 
+  | [| _; file; db |] ->
+      let bt = read_bibtex file in
+      let bso = List.map bib_of_bibtex bt in
+      let bs = List.fold_left (fun a -> function None -> a | Some b -> b :: a) [] bso in
+      save_bibtexs bs db
+  | _ -> 
+      eprintf "Usage: %s <bibtex> <db>\n%!" Sys.argv.(0)