validate video urls

yt_fts/summarize.py

@@ -0,0 +1,71 @@
+import sys
+from urllib.parse import urlparse, parse_qs
+
+from rich.console import Console
+
+# determine if input_video is url or video id 
+# if it's a url get the video id 
+# check if the video id is in database
+# if video id is in database get the full transcript
+# if the video id is not in the database download the transcript
+# feed the transcript to an llm and print the summary
+
+
+class SummarizeHandler:
+    def __init__(self, openai_client, input_video):
+
+        self.console = Console()
+        self.openai_client = openai_client
+        self.input_video = input_video
+
+        if "https" in input_video:
+            self.video_id = self.get_video_id_from_url(input_video)
+        else:
+            self.video_id = input_video
+
+
+
+    def summarize_video(self):
+        console = self.console
+        input_video = self.input_video
+
+
+
+    def get_video_id_from_url(self, video_url):
+        # https://www.youtube.com/watch?v=Xjk6d5fPs_k
+        # https://youtu.be/Xjk6d5fPs_k?si=BBb2URutUT2gG4th
+        # https://youtu.be/Xjk6d5fPs_k
+        # https://www.youtube.com/watch?v=Xjk6d5fPs_k&si=BBb2URutUT2gG4th
+
+        console = self.console
+        video_url = video_url.strip('/')
+        parsed = urlparse(video_url)
+        domain = parsed.netloc
+        path = parsed.path.split('/')
+        query = parse_qs(parsed.query)
+
+        valid_domains = ["youtube.com", "youtu.be", "www.youtube.com"]
+
+        if domain not in valid_domains:
+            console.print("[red]Error:[/red] "
+                          f"Invalid URL, domain \"{domain}\" not supported.")
+            sys.exit(1)
+
+
+        if domain in ["youtube.com", "www.youtube.com"] and "watch" in path:
+            video_id = query.get('v', [None])[0]
+        elif domain == "youtu.be":
+            video_id = path[-1]
+        else:
+            console.print("[red]Error:[/red] "
+                          "Invalid URL, please provide a valid YouTube video URL.")
+            sys.exit(1)
+
+        if video_id:
+            return video_id
+
+        console.print("[red]Error:[/red] "
+                      "Invalid URL, please provide a valid YouTube video URL.")
+        sys.exit(1)
+
+

yt_fts/yt_fts.py

@@ -2,9 +2,14 @@
 import sys
 import click
 
+from openai import OpenAI
 from rich.console import Console
 
 from .download import DownloadHandler
+from .export import ExportHandler 
+from .search import SearchHandler
+from .summarize import SummarizeHandler
+
 from .list import list_channels
 from .utils import show_message
 from .config import (
@@ -164,7 +169,6 @@ def delete(channel):
 @click.option("-f", "--format", default="txt",
               help="The format to export transcripts to. Supported formats: txt, vtt")
 def export(channel, format):
-    from .export import ExportHandler 
 
     export_handler = ExportHandler(
         scope = "channel",
@@ -188,7 +192,6 @@ def export(channel, format):
 @click.option("-l", "--limit", default=10, type=int, help="Number of results to return")
 @click.option("-e", "--export", is_flag=True, help="Export search results to a CSV file.")
 def search(text, channel, video_id, export, limit):
-    from yt_fts.search import SearchHandler
 
     if len(text) > 40:
         show_message("search_too_long")
@@ -229,8 +232,6 @@ def search(text, channel, video_id, export, limit):
               help="OpenAI API key. If not provided, the script will attempt to read it from the OPENAI_API_KEY "
                    "environment variable.")
 def vsearch(text, channel, video_id, limit, export, openai_api_key):
-    from openai import OpenAI
-    from .search import SearchHandler
 
     if openai_api_key is None:
         openai_api_key = os.environ.get("OPENAI_API_KEY")
@@ -343,6 +344,32 @@ def llm(prompt, channel, openai_api_key=None):
     sys.exit(0)
 
 
+@cli.command(
+    name="summarize",
+    help="summarize a youtube video"
+)
+@click.argument("video", required=True)
+@click.option("--model", "-m", default="gpt-4o",
+              help="Model to use in summary")
+@click.option("--openai-api-key", default=None,
+              help="OpenAI API key. If not provided, the script will attempt to read it from"
+                   " the OPENAI_API_KEY environment variable.")
+def summarize(video, model, openai_api_key):
+    if openai_api_key is None:
+        openai_api_key = os.environ.get("OPENAI_API_KEY")
+
+    if openai_api_key is None:
+        console.print("[red]Error:[/red] OPENAI_API_KEY environment variable not set\n"
+                      "To set the key run: export \"OPENAI_API_KEY=<your_key>\" or pass "
+                      "one in with --openai-api-key")
+        sys.exit(1)
+
+    openai_client = OpenAI(api_key=openai_api_key)
+
+    summarize_handler = SummarizeHandler(openai_client, video)
+    summarize_handler.summarize_video()
+
+
 @cli.command(
     help="""
     Show config settings