Fix Whisper (#27)

PulumiWebServer/Nix/flake.lock

@@ -156,11 +156,11 @@
         ]
       },
       "locked": {
-        "lastModified": 1705660020,
-        "narHash": "sha256-1tOuNh+UbiZlaC8RrpQzzypgnLBC67eRlBunfkE4sbQ=",
+        "lastModified": 1706473109,
+        "narHash": "sha256-iyuAvpKTsq2u23Cr07RcV5XlfKExrG8gRpF75hf1uVc=",
         "owner": "nix-community",
         "repo": "home-manager",
-        "rev": "2064348e555b6aa963da6372a8f14e6acb80a176",
+        "rev": "d634c3abafa454551f2083b054cd95c3f287be61",
         "type": "github"
       },
       "original": {
@@ -226,11 +226,11 @@
     },
     "nixpkgs": {
       "locked": {
-        "lastModified": 1705458851,
-        "narHash": "sha256-uQvEhiv33Zj/Pv364dTvnpPwFSptRZgVedDzoM+HqVg=",
+        "lastModified": 1706373441,
+        "narHash": "sha256-S1hbgNbVYhuY2L05OANWqmRzj4cElcbLuIkXTb69xkk=",
         "owner": "NixOS",
         "repo": "nixpkgs",
-        "rev": "8bf65f17d8070a0a490daf5f1c784b87ee73982c",
+        "rev": "56911ef3403a9318b7621ce745f5452fb9ef6867",
         "type": "github"
       },
       "original": {
@@ -242,11 +242,11 @@
     },
     "nixpkgs-stable": {
       "locked": {
-        "lastModified": 1705033721,
-        "narHash": "sha256-K5eJHmL1/kev6WuqyqqbS1cdNnSidIZ3jeqJ7GbrYnQ=",
+        "lastModified": 1705957679,
+        "narHash": "sha256-Q8LJaVZGJ9wo33wBafvZSzapYsjOaNjP/pOnSiKVGHY=",
         "owner": "NixOS",
         "repo": "nixpkgs",
-        "rev": "a1982c92d8980a0114372973cbdfe0a307f1bdea",
+        "rev": "9a333eaa80901efe01df07eade2c16d183761fa3",
         "type": "github"
       },
       "original": {
@@ -273,11 +273,11 @@
     },
     "nixpkgs_3": {
       "locked": {
-        "lastModified": 1704842529,
-        "narHash": "sha256-OTeQA+F8d/Evad33JMfuXC89VMetQbsU4qcaePchGr4=",
+        "lastModified": 1706173671,
+        "narHash": "sha256-lciR7kQUK2FCAYuszyd7zyRRmTaXVeoZsCyK6QFpGdk=",
         "owner": "NixOS",
         "repo": "nixpkgs",
-        "rev": "eabe8d3eface69f5bb16c18f8662a702f50c20d5",
+        "rev": "4fddc9be4eaf195d631333908f2a454b03628ee5",
         "type": "github"
       },
       "original": {
@@ -380,11 +380,11 @@
         "nixpkgs": "nixpkgs_2"
       },
       "locked": {
-        "lastModified": 1703937582,
-        "narHash": "sha256-M4y/xbrocPoLwG4qUUdCoBvOHumPAGlMoeo8SpWjn0M=",
+        "lastModified": 1706573865,
+        "narHash": "sha256-Gfk5mMz6gRLVwljXrPTfqAX7Bp8UQK4vKtXu9kYkzQM=",
         "ref": "refs/heads/main",
-        "rev": "cdbc73b07f3cac88e446fbe73c4b0c6616448319",
-        "revCount": 8,
+        "rev": "e96ae78665db13cbf4745c04d634a8a878549768",
+        "revCount": 10,
         "type": "git",
         "url": "https://gitea.patrickstevens.co.uk/patrick/puregym-unofficial-dotnet"
       },
@@ -469,11 +469,11 @@
         "nixpkgs-stable": "nixpkgs-stable"
       },
       "locked": {
-        "lastModified": 1705356877,
-        "narHash": "sha256-274jL1cH64DcXUXebVMZBRUsTs3FvFlPIPkCN/yhSnI=",
+        "lastModified": 1706410821,
+        "narHash": "sha256-iCfXspqUOPLwRobqQNAQeKzprEyVowLMn17QaRPQc+M=",
         "owner": "Mic92",
         "repo": "sops-nix",
-        "rev": "87755331580fdf23df7e39b46d63ac88236bf42c",
+        "rev": "73bf36912e31a6b21af6e0f39218e067283c67ef",
         "type": "github"
       },
       "original": {
@@ -571,11 +571,11 @@
         "scripts": "scripts_4"
       },
       "locked": {
-        "lastModified": 1705693021,
-        "narHash": "sha256-Ew7yxjYvwG9IE6YNPxcHNQUn2T8es1yWoK/f+x7UyCM=",
+        "lastModified": 1706487253,
+        "narHash": "sha256-t72k/g6topd9INA2PEjO+NIaC56EyTUIOtDar6V/nAM=",
         "owner": "Smaug123",
         "repo": "static-site-pipeline",
-        "rev": "da3a4d2e53e3068755fcd49b74bcfcf2800c197d",
+        "rev": "18314d94f67917c42202cc1201cbccde7ddb181a",
         "type": "github"
       },
       "original": {
@@ -591,11 +591,11 @@
         "nixpkgs": "nixpkgs_7"
       },
       "locked": {
-        "lastModified": 1704121990,
-        "narHash": "sha256-ss1gDu3C7anyqEo0ksDuOPgcsD3EOUqtCwAt8ei6rM4=",
+        "lastModified": 1706487294,
+        "narHash": "sha256-h5Ptbc6Kks4zZllR2nHnbVPJAJyYOAHEqGyEBFNMvzE=",
         "owner": "Smaug123",
         "repo": "whisper.cpp",
-        "rev": "2ff0257983b4a3fe9cece30b215ba2b5087f2613",
+        "rev": "96214a91134021ad6416f06cb7d35419a73b302c",
         "type": "github"
       },
       "original": {

PulumiWebServer/Nix/whisper/whisper.py

@@ -11,7 +11,7 @@ app = Flask(__name__)
 youtube_regex = re.compile(
     r"^(?:https?://)?(?:www\.)?(?:youtu\.be/|youtube\.com/(?:embed/|v/|watch\?v=|watch\?.+&v=))((\w|-){11})(?:\S+)?$")
 
-alnum_regex = re.compile(r"^[a-zA-Z0-9]+$")
+acceptable_regex = re.compile(r"^[a-zA-Z0-9_]+$")
 
 
 def generate_output(wav_file):
@@ -70,8 +70,8 @@ def transcribe_file():
         file = request.args.get('file')
     except KeyError:
         return Response("must have a file as obtained from /upload, in the format ?file=...", status=400)
-    if alnum_regex.match(file) is None:
-        return Response(f"filename '{file}' was not alphanumeric", status=400)
+    if acceptable_regex.match(file) is None:
+        return Response(f"filename '{file}' did not match acceptable regex", status=400)
     return Response(generate_output(file), mimetype="text/event-stream")
 
 
@@ -86,6 +86,11 @@ def upload():
         return 'No "file" part in request', 400
     file = request.files['file']
 
+    try:
+        os.mkdir("/tmp/whisper")
+    except FileExistsError:
+        pass
+
     # Create temp file for this upload
     handle, temp_file = tempfile.mkstemp(text=False)
     try:
@@ -111,8 +116,8 @@ def download():
     except KeyError:
         return Response("must have a file parameter", status=400)
 
-    if alnum_regex.match(file) is None:
-        return Response(f"file '{file}' was not alphanumeric, bad format", status=400)
+    if acceptable_regex.match(file) is None:
+        return Response(f"file '{file}' did not match acceptable regex, bad format", status=400)
 
     return Response(open(f"/tmp/whisper/{file}.wav", 'rb').read(), mimetype="audio/wav")