Merge commit 'bd2126e954a85806526338c75c8084a52b50f7dd'

CHANGES.md

@@ -1,11 +1,16 @@
 0.3.0 (unreleased)
 =====
+
 - Add basic example.
 - Add optional custom parser argument to override the default parsing mechanism.
 - Add binary format for encoding frames.
+- Update default metadata mappings to follow musicbrainz's picard mapping.
+- Add `MIME` module to guess MIME type of files (#4).
+- Generic RIFF format parser, adds support for wave files (#6).
 
 0.2.0 (2023-07-01)
 =====
+
 - Add support for FLAC.
 - id3v2: use "bpm" instead of "tempo".
 - id3v2: convert "tlen" to "duration".
@@ -15,4 +20,5 @@
 
 0.1.0 (2023-02-08)
 =====
+
 - Initial release.

README.md

@@ -4,7 +4,7 @@ The metadata library
 A pure OCaml library to read metadata from various formats. For now, are
 supported:
 
-- audio formats: ID3v1 and ID3v2 (for mp3), ogg/vorbis, ogg/opus and flac
+- audio formats: ID3v1 and ID3v2 (for mp3), ogg/vorbis, ogg/opus, flac and wav
 - image formats: jpeg and png
 - video formats: mp4 and avi
 

examples/dune

@@ -13,6 +13,11 @@
  (modules dump)
  (libraries metadata))
 
+(executable
+ (name mimetype)
+ (modules mimetype)
+ (libraries metadata))
+
 (executable
  (name test)
  (modules test)
@@ -68,9 +73,7 @@
    -i
    "sine=frequency=440:sample_rate=44100:duration=10"
    -f
-   mp3
-   -write_id3v1
-   true
+   wav
    -metadata
    "title=The title"
    -metadata

examples/mimetype

@@ -0,0 +1,2 @@
+#!/bin/sh
+dune exec --no-print-directory ./mimetype.exe -- $@

examples/mimetype.ml

@@ -0,0 +1,4 @@
+let () =
+  let fname = Sys.argv.(1) in
+  let mime = try Metadata.MIME.of_file fname with Not_found -> "unknown" in
+  print_endline mime

examples/test.ml

@@ -9,7 +9,7 @@ let () =
   p "mp3v2" (Metadata.ID3v2.parse_file "test.mp3");
   p "mp3v1" (Metadata.ID3v1.parse_file "test.mp3");
   p "mp3" (Metadata.ID3.parse_file "test.mp3");
-  p "wav" (Metadata.ID3.parse_file "test.wav");
+  p "wav" (Metadata.WAV.parse_file "test.wav");
   p "png" (Metadata.PNG.parse_file "test.png");
   p "jpg" (Metadata.JPEG.parse_file "test.jpg");
   p "avi" (Metadata.AVI.parse_file "test.avi");

src/metadata.ml

@@ -1,5 +1,7 @@
 module CharEncoding = MetadataCharEncoding
 
+module MIME = MetadataMIME
+
 module Make (E : CharEncoding.T) = struct
   include MetadataBase
   module ID3v1 = MetadataID3v1
@@ -10,6 +12,8 @@ module Make (E : CharEncoding.T) = struct
   module PNG = MetadataPNG
   module AVI = MetadataAVI
   module MP4 = MetadataMP4
+  module WAV = MetadataWAV
+  module RIFF = MetadataRIFF
 
   (** Charset conversion function. *)
   let recode = E.convert
@@ -41,7 +45,7 @@ module Make (E : CharEncoding.T) = struct
       | [] -> raise Invalid
 
   module Audio = struct
-    let parsers = [ID3.parse; OGG.parse; FLAC.parse]
+    let parsers = [ID3.parse; OGG.parse; FLAC.parse; WAV.parse]
     let parse = first_valid parsers
 
     let parse_file ?custom_parser file =
@@ -65,7 +69,7 @@ module Make (E : CharEncoding.T) = struct
   end
 
   module Any = struct
-    let parsers = Audio.parsers @ Image.parsers @ Video.parsers
+    let parsers = Audio.parsers @ Image.parsers @ Video.parsers @ [RIFF.parse]
 
     (** Genering parsing of metadata. *)
     let parse = first_valid parsers

src/metadata.mli

@@ -3,6 +3,16 @@
 (** Functions for handling charset conversion. *)
 module CharEncoding = MetadataCharEncoding
 
+(** Guess the MIME type of a file. *)
+module MIME : sig
+  (** Guess the MIME type from file contents. Raises [Not_found] if none was
+      found. *)
+  val of_string : string -> string
+
+  (** Same as [of_string] but takes a file name as argument. *)
+  val of_file : string -> string
+end
+
 (** Generate metadata parsers given functions for converting charsets. *)
 module Make : functor (_ : CharEncoding.T) -> sig
   (** Raised when the metadata is not valid. *)
@@ -84,6 +94,12 @@ module Make : functor (_ : CharEncoding.T) -> sig
   (** MP4 metadata. *)
   module MP4 = MetadataMP4
 
+  (** WAV metadata. *)
+  module WAV = MetadataWAV
+
+  (** RIFF metdata. *)
+  module RIFF = MetadataRIFF
+
   (** Convert the charset encoding of a string. *)
   val recode :
     ?source:[ `ISO_8859_1 | `UTF_16 | `UTF_16BE | `UTF_16LE | `UTF_8 ] ->

src/metadataAVI.ml

@@ -1,61 +1,3 @@
-open MetadataBase
-module R = Reader
+let parse = MetadataRIFF.parse ~format:"AVI "
 
-(* Tag normalization. *)
-let tagn =
-  [
-    ("IART", "artist");
-    ("ICMT", "comment");
-    ("ICOP", "copyright");
-    ("ICRD", "date");
-    ("ICRD", "date");
-    ("IGNR", "genre");
-    ("INAM", "title");
-    ("IPRD", "album");
-    ("IPRT", "track");
-    ("ISFT", "encoder");
-  ]
-
-let parse f : metadata =
-  if R.read f 4 <> "RIFF" then raise Invalid;
-  let _ (* file size *) = R.int32_le f in
-  if R.read f 4 <> "AVI " then raise Invalid;
-  let ans = ref [] in
-  let chunk () =
-    let tag = R.read f 4 in
-    let size = R.int32_le f in
-    if tag <> "LIST" then R.drop f size
-    else (
-      let subtag = R.read f 4 in
-      match subtag with
-        | "INFO" ->
-            let remaining = ref (size - 4) in
-            while !remaining > 0 do
-              let tag = R.read f 4 in
-              let size = R.int32_le f in
-              match R.read_tag ~length:(size - 1) ~label:tag f with
-                | None -> ()
-                | Some s ->
-                    R.drop f 1;
-                    (* null-terminated *)
-                    let padding = size mod 2 in
-                    R.drop f padding;
-                    remaining := !remaining - (8 + size + padding);
-                    let tag =
-                      match List.assoc_opt tag tagn with
-                        | Some tag -> tag
-                        | None -> tag
-                    in
-                    ans := (tag, s) :: !ans
-            done
-        | "movi" -> raise Exit (* stop parsing there *)
-        | _ -> R.drop f (size - 4))
-  in
-  try
-    while true do
-      chunk ()
-    done;
-    assert false
-  with _ -> List.rev !ans
-
-let parse_file ?custom_parser file = R.with_file ?custom_parser parse file
+let parse_file = MetadataRIFF.parse_file ~format:"AVI "

src/metadataID3v2.ml

@@ -58,23 +58,38 @@ let normalize_id = function
   | "TALB" -> "album"
   | "TBPM" -> "bpm"
   | "TCOM" -> "composer"
-  | "TCON" -> "content"
+  | "TCON" -> "genre"
   | "TCOP" -> "copyright"
   | "TDAT" -> "date"
   | "TDOR" -> "original release time"
   | "TDRC" -> "recording time"
-  | "TENC" -> "encoder"
+  | "TENC" -> "encodedby"
+  | "TEXT" -> "lyricist"
+  | "TIT1" -> "grouping"
   | "TIT2" -> "title"
+  | "TIT3" -> "subtitle"
+  | "TKEY" -> "key"
   | "TLAN" -> "language"
   | "TLEN" -> "length"
-  | "TMED" -> "media type"
-  | "TOPE" -> "performer"
+  | "TMED" -> "media"
+  | "TOAL" -> "originalalbum"
+  | "TOFN" -> "originalfilename"
+  | "TOPE" -> "originalartist"
+  | "TPOS" -> "discnumber"
   | "TPE1" -> "artist"
-  | "TPE2" -> "band"
-  | "TPUB" -> "publisher"
+  | "TPE2" -> "albumartist"
+  | "TPE3" -> "conductor"
+  | "TPE4" -> "remixer"
+  | "TPUB" -> "label"
   | "TRCK" -> "tracknumber"
+  | "TSOA" -> "albumsort"
+  | "TSO2" -> "albumartistsort"
+  | "TSOT" -> "titlesort"
+  | "TSRC" -> "isrc"
   | "TSSE" -> "encoder"
+  | "TSST" -> "discsubtitle"
   | "TYER" -> "year"
+  | "WOAR" -> "website"
   | "WXXX" -> "url"
   | id -> id
 

src/metadataMIME.ml

@@ -0,0 +1,64 @@
+(** Guess the mime-type of a file. *)
+
+module String = struct
+  include String
+
+  let contains_at offset ~substring s =
+    let n = String.length substring in
+    if String.length s < offset + n then false
+    else String.sub s offset n = substring
+end
+
+let prefixes =
+  [
+    "ID3", "audio/mpeg";
+    "OggS", "audio/ogg";
+    "%PDF-", "application/pdf";
+    "\137PNG\013\010\026\010", "image/png";
+  ]
+
+let advanced =
+  let wav s =
+    String.starts_with ~prefix:"RIFF" s &&
+    String.contains_at 8 ~substring:"WAVEfmt " s
+  in
+  let avi s =
+    String.starts_with ~prefix:"RIFF" s &&
+    String.contains_at 8 ~substring:"AVI " s
+  in
+  [
+    wav, "audio/wav";
+    avi, "video/x-msvideo"
+  ]
+
+let of_string s =
+  let ans = ref "" in
+  try
+    List.iter
+      (fun (f, mime) ->
+         if f s then
+           (
+             ans := mime;
+             raise Exit
+           )
+      ) advanced;
+    List.iter
+      (fun (prefix, mime) ->
+         if String.starts_with ~prefix s then
+           (
+             ans := mime;
+             raise Exit
+           )
+      ) prefixes;
+    raise Not_found
+  with
+  | Exit -> !ans
+
+let of_file fname =
+  let len = 16 in
+  let buf = Bytes.create len in
+  let ic = open_in fname in
+  let n = input ic buf 0 len in
+  let buf = if n = len then buf else Bytes.sub buf 0 n in
+  let s = Bytes.unsafe_to_string buf in
+  of_string s

src/metadataMP4.ml

@@ -5,9 +5,12 @@ let tagn =
   [
     ("\xa9nam", "title");
     ("\xa9ART", "artist");
+    ("cprt", "copyright");
     ("\xa9too", "encoder");
     ("\xa9day", "date");
     ("\xa9cpy", "copyright");
+    ("\xa9gen", "genre");
+    ("\xa9wrt", "composer");
     ("\xa9alb", "album");
     ("\xa9des", "description");
     ("\xa9cmt", "comment");

src/metadataRIFF.ml

@@ -0,0 +1,62 @@
+open MetadataBase
+module R = Reader
+
+(* Tag normalization. *)
+let tagn =
+  [
+    ("IART", "artist");
+    ("ICMT", "comment");
+    ("ICOP", "copyright");
+    ("ICRD", "date");
+    ("ICRD", "date");
+    ("IGNR", "genre");
+    ("INAM", "title");
+    ("IPRD", "album");
+    ("IPRT", "track");
+    ("ISFT", "encoder");
+    ("ITRK", "track");
+  ]
+
+let parse ?format f : metadata =
+  if R.read f 4 <> "RIFF" then raise Invalid;
+  let _ (* file size *) = R.int32_le f in
+  if format <> None && Some (R.read f 4) <> format then raise Invalid;
+  let ans = ref [] in
+  let chunk () =
+    let tag = R.read f 4 in
+    let size = R.int32_le f in
+    if tag <> "LIST" then R.drop f size
+    else (
+      let subtag = R.read f 4 in
+      match subtag with
+        | "INFO" ->
+            let remaining = ref (size - 4) in
+            while !remaining > 0 do
+              let tag = R.read f 4 in
+              let size = R.int32_le f in
+              match R.read_tag ~length:(size - 1) ~label:tag f with
+                | None -> ()
+                | Some s ->
+                    R.drop f 1;
+                    (* null-terminated *)
+                    let padding = size mod 2 in
+                    R.drop f padding;
+                    remaining := !remaining - (8 + size + padding);
+                    let tag =
+                      match List.assoc_opt tag tagn with
+                        | Some tag -> tag
+                        | None -> tag
+                    in
+                    ans := (tag, s) :: !ans
+            done
+        | "movi" -> raise Exit (* stop parsing there *)
+        | _ -> R.drop f (size - 4))
+  in
+  try
+    while true do
+      chunk ()
+    done;
+    assert false
+  with _ -> List.rev !ans
+
+let parse_file ?format ?custom_parser file = R.with_file ?custom_parser (parse ?format) file

src/metadataRIFF.mli

@@ -0,0 +1,3 @@
+val parse : ?format:string -> MetadataBase.Reader.t -> MetadataBase.metadata
+
+val parse_file : ?format:string -> ?custom_parser:MetadataBase.custom_parser -> string -> MetadataBase.metadata

src/metadataWAV.ml

@@ -0,0 +1,3 @@
+let parse = MetadataRIFF.parse ~format:"WAVE"
+
+let parse_file = MetadataRIFF.parse_file ~format:"WAVE"