X-Git-Url: http://matita.cs.unibo.it/gitweb/?a=blobdiff_plain;f=helm%2FDEVEL%2Fpxp%2Fnetstring%2Fnetmappings.mli;fp=helm%2FDEVEL%2Fpxp%2Fnetstring%2Fnetmappings.mli;h=1c52d0729921d114b1abbe94d870288d44f49b45;hb=c03d2c1fdab8d228cb88aaba5ca0f556318bebc5;hp=0000000000000000000000000000000000000000;hpb=758057e85325f94cd88583feb1fdf6b038e35055;p=helm.git

diff --git a/helm/DEVEL/pxp/netstring/netmappings.mli b/helm/DEVEL/pxp/netstring/netmappings.mli
new file mode 100644
index 000000000..1c52d0729
--- /dev/null
+++ b/helm/DEVEL/pxp/netstring/netmappings.mli
@@ -0,0 +1,115 @@
+(* $Id$
+ * ----------------------------------------------------------------------
+ *)
+
+type from_uni_list =
+    U_nil
+  | U_single of (int*int)
+  | U_list of (int*int) list
+;;
+  (* A representation of (int*int) list that is optimized for the case that
+   * lists with 0 and 1 elements are the most frequent cases.
+   *)
+
+
+val to_unicode   : (Netconversion.encoding, 
+		    int array Lazy.t)           Hashtbl.t;;
+
+val from_unicode : (Netconversion.encoding, 
+		    from_uni_list array Lazy.t) Hashtbl.t;;
+  (* These hashtables are used internally by the parser to store
+   * the conversion tables from 8 bit encodings to Unicode and vice versa.
+   * It is normally not necessary to access these tables; the 
+   * Netconversion module does it already for you.
+   *
+   * Specification of the conversion tables:
+   *
+   * to_unicode: maps an 8 bit code to Unicode, i.e.
+   *    let m = Hashtbl.find `Enc_isoXXX to_unicode in
+   *    let unicode = m.(isocode) 
+   *    - This may be (-1) to indicate that the code point is not defined.
+   *
+   * from_unicode: maps Unicode to an 8 bit code, i.e.
+   *    let m = Hashtbl.find `Enc_isoXXX from_unicode in
+   *    let l = m.(unicode land 255)
+   *    Now search in l the pair (unicode, isocode), and return isocode.
+   *
+   * Note: It is guaranteed that both arrays have always 256 elements.
+   *)
+
+val lock : unit -> unit
+  (* In multi-threaded applications: obtains a lock which is required to
+   * Lazy.force the values found in to_unicode and from_unicode.
+   * In single-threaded applications: a NO-OP
+   *)
+
+val unlock : unit -> unit
+  (* In multi-threaded applications: releases the lock which is required to
+   * Lazy.force the values found in to_unicode and from_unicode.
+   * In single-threaded applications: a NO-OP
+   *)
+
+
+val init_mt : (unit -> unit) -> (unit -> unit) -> unit
+  (* Internally used; see netstring_mt.ml *)
+
+
+(* ---------------------------------------- *)
+
+(* The following comment was written when the conversion module belonged
+ * to the PXP package (Polymorhic XML Parser).
+ *)
+
+(* HOW TO ADD A NEW 8 BIT CODE:
+ *
+ * It is relatively simple to add a new 8 bit code to the system. This
+ * means that the parser can read and write files with the new encoding;
+ * this does not mean that the parser can represent the XML tree internally
+ * by the new encoding.
+ *
+ * - Put a new unimap file into the "mappings" directory. The file format
+ *   is simple; please look at the already existing files. 
+ *   The name of the file determines the internal name of the code:
+ *   If the file is called <name>.unimap, the code will be called
+ *   `Enc_<name>.
+ *
+ * - Extend the type "encoding" in pxp_types.mli and pxp_types.ml
+ *
+ * - Extend the two functions encoding_of_string and string_of_encoding
+ *   in pxp_types.ml
+ *
+ * - Recompile the parser
+ *
+ * Every encoding consumes at least 3kB of memory, but this may be much more 
+ * if the code points are dispersed on the Unicode code space.
+ *
+ * Perhaps the addition of new codes will become even simpler in future
+ * versions of PXP; but it is currently more important to support 
+ * non-8-bit codes, too.
+ * Every contribution of new codes to PXP is welcome!
+ *)
+
+
+(* ======================================================================
+ * History:
+ * 
+ * $Log$
+ * Revision 1.1  2000/11/17 09:57:28  lpadovan
+ * Initial revision
+ *
+ * Revision 1.2  2000/08/29 00:47:24  gerd
+ * 	New type for the conversion Unicode to 8bit.
+ * 	Conversion tables are now lazy. Thus also mutexes are required.
+ *
+ * Revision 1.1  2000/08/13 00:02:57  gerd
+ * 	Initial revision.
+ *
+ *
+ * ======================================================================
+ * OLD LOGS FROM THE PXP PACKAGE (FILE NAME pxp_mappings.mli):
+ * 
+ * Revision 1.1  2000/07/27 00:40:02  gerd
+ * 	Initial revision.
+ *
+ * 
+ *)