Generate file_formats from a namedtuple spec and correct seven mimetypes

CHANGELOG.md

@@ -1,6 +1,20 @@
 Release History
 ===============
 
+Unreleased
+----------
+  - `file_formats` mimetype `vtt`: `.vtt` → `text/vtt`
+  - `file_formats` mimetype `mp3`: `.mp3` → `audio/mpeg`
+  - `file_formats` mimetype `zip`: `.zip` → `application/zip`
+  - `file_formats` mimetype `svg`: `image/svg` → `image/svg+xml`
+  - `file_formats` mimetype `h5p`: `.h5p` → `application/h5p+zip`
+  - `file_formats` mimetype `zim`: `.zim` → `application/x-openzim`
+  - `file_formats` mimetype `graphie`: `.graphie` → `application/x-graphie`
+  - `file_formats.getformat(id, default)` returns `default` for an unknown `id`
+  - Removed `resources/formatlookup.json`; use `file_formats.FORMATLIST`
+  - Added `kolibri-constants/FileFormats`
+
+
 0.2.17 Apr 13, 2026
 -------------------
   - Added library enum constants (KOLIBRI, COMMUNITY) for identifying channel library membership

js/FileFormats.js

@@ -0,0 +1,80 @@
+// -*- coding: utf-8 -*-
+// Generated by scripts/generate_from_specs.py
+// FileFormats
+
+export default {
+    MP4: "mp4",
+    MP4_MIMETYPE: "video/mp4",
+    WEBM: "webm",
+    WEBM_MIMETYPE: "video/webm",
+    VTT: "vtt",
+    VTT_MIMETYPE: "text/vtt",
+    PDF: "pdf",
+    PDF_MIMETYPE: "application/pdf",
+    EPUB: "epub",
+    EPUB_MIMETYPE: "application/epub+zip",
+    MP3: "mp3",
+    MP3_MIMETYPE: "audio/mpeg",
+    JPG: "jpg",
+    JPG_MIMETYPE: "image/jpeg",
+    JPEG: "jpeg",
+    PNG: "png",
+    PNG_MIMETYPE: "image/png",
+    GIF: "gif",
+    GIF_MIMETYPE: "image/gif",
+    JSON: "json",
+    JSON_MIMETYPE: "application/json",
+    SVG: "svg",
+    SVG_MIMETYPE: "image/svg+xml",
+    GRAPHIE: "graphie",
+    GRAPHIE_MIMETYPE: "application/x-graphie",
+    PERSEUS: "perseus",
+    PERSEUS_MIMETYPE: "application/perseus+zip",
+    H5P: "h5p",
+    H5P_MIMETYPE: "application/h5p+zip",
+    ZIM: "zim",
+    ZIM_MIMETYPE: "application/x-openzim",
+    HTML5: "zip",
+    HTML5_MIMETYPE: "application/zip",
+    BLOOMPUB: "bloompub",
+    BLOOMPUB_MIMETYPE: "application/bloompub+zip",
+    BLOOMD: "bloomd",
+    HTML5_ARTICLE: "kpub",
+    HTML5_ARTICLE_MIMETYPE: "application/kpub+zip",
+    AVI: "avi",
+    MOV: "mov",
+    MPG: "mpg",
+    WMV: "wmv",
+    MKV: "mkv",
+    FLV: "flv",
+    OGV: "ogv",
+    M4V: "m4v",
+    SRT: "srt",
+    TTML: "ttml",
+    SAMI: "sami",
+    SCC: "scc",
+    DFXP: "dfxp",
+};
+
+export const FORMATLIST = [
+    { id: "mp4", mimetype: "video/mp4" },
+    { id: "webm", mimetype: "video/webm" },
+    { id: "vtt", mimetype: "text/vtt" },
+    { id: "pdf", mimetype: "application/pdf" },
+    { id: "epub", mimetype: "application/epub+zip" },
+    { id: "mp3", mimetype: "audio/mpeg" },
+    { id: "jpg", mimetype: "image/jpeg" },
+    { id: "jpeg", mimetype: "image/jpeg" },
+    { id: "png", mimetype: "image/png" },
+    { id: "gif", mimetype: "image/gif" },
+    { id: "json", mimetype: "application/json" },
+    { id: "svg", mimetype: "image/svg+xml" },
+    { id: "graphie", mimetype: "application/x-graphie" },
+    { id: "perseus", mimetype: "application/perseus+zip" },
+    { id: "h5p", mimetype: "application/h5p+zip" },
+    { id: "zim", mimetype: "application/x-openzim" },
+    { id: "zip", mimetype: "application/zip" },
+    { id: "bloompub", mimetype: "application/bloompub+zip" },
+    { id: "bloomd", mimetype: "application/bloompub+zip" },
+    { id: "kpub", mimetype: "application/kpub+zip" },
+];

le_utils/constants/file_formats.py

@@ -1,93 +1,71 @@
-import json
-import pkgutil
-from collections import namedtuple
+# -*- coding: utf-8 -*-
+# Generated by scripts/generate_from_specs.py
+from __future__ import unicode_literals
 
-""" File Format Constants """
+from collections import namedtuple
 
-# IMPORTANT: Keep these constants in sync with formatlookup.json and presetlookup.json
+# FileFormats
 
-# constants for Video format
 MP4 = "mp4"
 MP4_MIMETYPE = "video/mp4"
 WEBM = "webm"
 WEBM_MIMETYPE = "video/webm"
-# constants for video formats converitble to mp4
-AVI = "avi"
-MOV = "mov"
-MPG = "mpg"
-WMV = "wmv"
-MKV = "mkv"
-FLV = "flv"
-OGV = "ogv"
-M4V = "m4v"
-
-# constants for Subtitle format
 VTT = "vtt"
-VTT_MIMETYPE = ".vtt"
-# constants for formats convertible to VTT
-SRT = "srt"
-TTML = "ttml"
-SAMI = "sami"
-SCC = "scc"
-DFXP = "dfxp"
-
-# constants for Audio format
-MP3 = "mp3"
-MP3_MIMETYPE = ".mp3"
-
-# constants for Document format
+VTT_MIMETYPE = "text/vtt"
 PDF = "pdf"
 PDF_MIMETYPE = "application/pdf"
-
-# constants for Thumbnail format
+EPUB = "epub"
+EPUB_MIMETYPE = "application/epub+zip"
+MP3 = "mp3"
+MP3_MIMETYPE = "audio/mpeg"
 JPG = "jpg"
 JPG_MIMETYPE = "image/jpeg"
 JPEG = "jpeg"
 PNG = "png"
 PNG_MIMETYPE = "image/png"
 GIF = "gif"
 GIF_MIMETYPE = "image/gif"
-
-
 JSON = "json"
 JSON_MIMETYPE = "application/json"
 SVG = "svg"
-SVG_MIMETYPE = "image/svg"
+SVG_MIMETYPE = "image/svg+xml"
 GRAPHIE = "graphie"
-GRAPHIE_MIMETYPE = ".graphie"
-
-# constants for Exercise format
+GRAPHIE_MIMETYPE = "application/x-graphie"
 PERSEUS = "perseus"
 PERSEUS_MIMETYPE = "application/perseus+zip"
-
-# constants for HTML5 zip format
-HTML5 = "zip"
-HTML5_MIMETYPE = ".zip"
-
-# constants for Kolibri article zip format
-HTML5_ARTICLE = "kpub"
-HTML5_ARTICLE_MIMETYPE = "application/kpub+zip"
-
-# constants for H5P format
 H5P = "h5p"
-H5P_MIMETYPE = ".h5p"
-
-# constants for Zim format
+H5P_MIMETYPE = "application/h5p+zip"
 ZIM = "zim"
-ZIM_MIMETYPE = ".zim"
-
-# constants for ePub format
-EPUB = "epub"
-EPUB_MIMETYPE = "application/epub+zip"
-
-# constants for bloomPub format
+ZIM_MIMETYPE = "application/x-openzim"
+HTML5 = "zip"
+HTML5_MIMETYPE = "application/zip"
 BLOOMPUB = "bloompub"
-BLOOMD = "bloomd"
 BLOOMPUB_MIMETYPE = "application/bloompub+zip"
+BLOOMD = "bloomd"
+HTML5_ARTICLE = "kpub"
+HTML5_ARTICLE_MIMETYPE = "application/kpub+zip"
+AVI = "avi"
+MOV = "mov"
+MPG = "mpg"
+WMV = "wmv"
+MKV = "mkv"
+FLV = "flv"
+OGV = "ogv"
+M4V = "m4v"
+SRT = "srt"
+TTML = "ttml"
+SAMI = "sami"
+SCC = "scc"
+DFXP = "dfxp"
 
-# Zip-based formats identified by le_utils.archive.contents_sha256.
-# FROZEN: ricecooker and Studio's upload Cloud Function pin this; removing a format orphans its index entries.
-ARCHIVE_FORMATS = (HTML5, H5P, EPUB, HTML5_ARTICLE, BLOOMPUB, BLOOMD)
+ARCHIVE_FORMATS = (
+    HTML5,
+    H5P,
+    EPUB,
+    HTML5_ARTICLE,
+    BLOOMPUB,
+    BLOOMD,
+)
 
 choices = (
     (MP4, "MP4 Video"),
@@ -118,24 +96,35 @@ class Format(namedtuple("Format", ["id", "mimetype"])):
 
 
 def generate_list(constantlist):
-    for id, format in constantlist.items():
-        yield Format(id=id, **format)
-
-
-def _initialize_format_list():
-    constantlist = json.loads(pkgutil.get_data("le_utils", "resources/formatlookup.json").decode("utf-8"))
-    return generate_list(constantlist)
-
-
-FORMATLIST = list(_initialize_format_list())
-
-_FORMATLOOKUP = {f.id: f for f in FORMATLIST}
+    for id, record in constantlist.items():
+        yield Format(id=id, **record)
+
+
+FORMATLIST = [
+    Format(id="mp4", mimetype="video/mp4"),
+    Format(id="webm", mimetype="video/webm"),
+    Format(id="vtt", mimetype="text/vtt"),
+    Format(id="pdf", mimetype="application/pdf"),
+    Format(id="epub", mimetype="application/epub+zip"),
+    Format(id="mp3", mimetype="audio/mpeg"),
+    Format(id="jpg", mimetype="image/jpeg"),
+    Format(id="jpeg", mimetype="image/jpeg"),
+    Format(id="png", mimetype="image/png"),
+    Format(id="gif", mimetype="image/gif"),
+    Format(id="json", mimetype="application/json"),
+    Format(id="svg", mimetype="image/svg+xml"),
+    Format(id="graphie", mimetype="application/x-graphie"),
+    Format(id="perseus", mimetype="application/perseus+zip"),
+    Format(id="h5p", mimetype="application/h5p+zip"),
+    Format(id="zim", mimetype="application/x-openzim"),
+    Format(id="zip", mimetype="application/zip"),
+    Format(id="bloompub", mimetype="application/bloompub+zip"),
+    Format(id="bloomd", mimetype="application/bloompub+zip"),
+    Format(id="kpub", mimetype="application/kpub+zip"),
+]
+
+_FORMATLOOKUP = {r.id: r for r in FORMATLIST}
 
 
 def getformat(id, default=None):
-    """
-    Try to lookup a file format object for its `id` in internal representation defined
-    in resources/formatlookup.json.
-    Returns None if lookup by internal representation fails.
-    """
-    return _FORMATLOOKUP.get(id) or None
+    return _FORMATLOOKUP.get(id, default)