From acc966b488cdf14cf17f4da703f459000771ef5c Mon Sep 17 00:00:00 2001
From: Blallo <blallo@autistici.org>
Date: Wed, 15 Sep 2021 17:53:39 +0200
Subject: [PATCH] Do not singletonize the retriever

---
 techrec/forge.py          |  8 +++--
 techrec/http_retriever.py | 69 +++++++++++++--------------------------
 2 files changed, 27 insertions(+), 50 deletions(-)

diff --git a/techrec/forge.py b/techrec/forge.py
index 138601c..e006fc6 100644
--- a/techrec/forge.py
+++ b/techrec/forge.py
@@ -8,7 +8,7 @@ from time import sleep
 from typing import Callable, Optional
 
 from .config_manager import get_config
-from .http_retriever import RETRIEVER
+from .http_retriever import download
 
 logger = logging.getLogger("forge")
 Validator = Callable[[datetime, datetime, str], bool]
@@ -23,8 +23,10 @@ async def get_timefile_exact(time) -> str:
     remote_path = os.path.join(
         remote_repo, time.strftime(get_config()["AUDIO_INPUT_FORMAT"])
     )
-    if remote.startswith("http"):
-        local = await RETRIEVER.get(remote_path)
+    if remote_path.startswith("http://") or remote_path.startswith("https://"):
+        logger.info(f"downloading {remote_path}")
+        print(f"DOWNLOADING -> {remote_path}")
+        local = await download(remote_path)
         return local
     return local_path
 
diff --git a/techrec/http_retriever.py b/techrec/http_retriever.py
index 1a3bd67..68d8934 100644
--- a/techrec/http_retriever.py
+++ b/techrec/http_retriever.py
@@ -1,6 +1,8 @@
 # -*- encoding: utf-8 -*-
 import asyncio
 import os
+from typing import Optional
+from tempfile import mkdtemp
 
 import aiohttp  # type: ignore
 
@@ -9,52 +11,25 @@ from .config_manager import get_config
 CHUNK_SIZE = 2 ** 12
 
 
-class HTTPRetriever(object):
+async def download(remote: str, staging: Optional[str] = None) -> str:
     """
-    This class offers the `get` method to retrieve the file from the local staging path
-    or, if missing, from the given remote.
+    This will download to AUDIO_STAGING the remote file and return the local
+    path of the downloaded file
     """
-    _instance = None
-
-    def __new__(cls):
-        if self._instance is None:
-            self._instance = super().__new__(cls)
-        return self._instance
-
-    def __init__(self):
-        self.repo_path = get_config()["AUDIO_STAGING"]
-        self.repo = dict(
-            path=os.path.join(self.repo_path, path) for i in os.listdir(self.repo_path)
-        )
-
-    async def get(remote: str) -> str:
-        """
-        This will look in the local staging path (ideally on a tmpfs or something
-        similar), and return the file from there if present. Otherwise, it will download
-        it from the remote.
-        """
-        if remote in self.repo:
-            return self.repo[remote]
-        file = await self._download_from_remote(remote)
-        self.repo[] = file
-        return file
-
-    async def _download(remote: str) -> str:
-        """
-        This will download to AUDIO_STAGING the remote file and return the local path
-        of the downloaded file
-        """
-        _, filename = os.path.split(remote)
-        local = os.path.join(get_config()["AUDIO_STAGING"], filename)
-        async with aiohttp.ClientSession() as session:
-            async with session.get(remote) as resp:
-                with open(local) as f:
-                    while True:
-                        chunk = await resp.content.read(CHUNK_SIZE)
-                        if not chunk:
-                            break
-                        f.write(chunk)
-        return local
-
-
-RETRIEVER = HTTPRetriever()
+    _, filename = os.path.split(remote)
+    if staging:
+        base = staging
+    else:
+        # if no staging is specified, and you want to clean the storage
+        # used by techrec: rm -rf /tmp/techrec*
+        base = mkdtemp(prefix="techrec-", dir="/tmp")
+    local = os.path.join(base, filename)
+    async with aiohttp.ClientSession() as session:
+        async with session.get(remote) as resp:
+            with open(local, "wb") as f:
+                while True:
+                    chunk = await resp.content.read(CHUNK_SIZE)
+                    if not chunk:
+                        break
+                    f.write(chunk)
+    return local