Progress indicator

[paperdoorknob] / fetch.py

diff --git a/fetch.py b/fetch.py
index 1d2070183a44dded893850630f68d6c2b9c06bb9..e998394445196fcf677766a78cd3805d6b241bd1 100644 (file)
--- a/fetch.py
+++ b/fetch.py
@@ -13,6 +13,12 @@ from typing import IO, Iterator
 import requests
 import requests_cache
 
 import requests
 import requests_cache
 

+from version import paperdoorknob_version
+
+
+_headers = {'User-Agent': f'paperdoorknob/{paperdoorknob_version} ' +
+            '(https://git.scottworley.com/paperdoorknob/)'}
+

 
 class Fetcher(ABC):
     @abstractmethod
 
 class Fetcher(ABC):
     @abstractmethod


@@ -27,7 +33,7 @@ class _SessionFetcher(Fetcher):
         self._timeout = timeout
 
     def fetch(self, url: str) -> bytes:
         self._timeout = timeout
 
     def fetch(self, url: str) -> bytes:

-        with self._session.get(url, timeout=self._timeout) as r:
+        with self._session.get(url, timeout=self._timeout, headers=_headers) as r:

             r.raise_for_status()
             return r.content
 
             r.raise_for_status()
             return r.content
 


@@ -44,7 +50,7 @@ class _CachingFetcher(Fetcher):
         self._cache_hit_count = 0
 
     def fetch(self, url: str) -> bytes:
         self._cache_hit_count = 0
 
     def fetch(self, url: str) -> bytes:

-        with self._session.get(url, timeout=self._timeout) as r:
+        with self._session.get(url, timeout=self._timeout, headers=_headers) as r:

             r.raise_for_status()
             self._request_count += 1
             self._cache_hit_count += int(r.from_cache)
             r.raise_for_status()
             self._request_count += 1
             self._cache_hit_count += int(r.from_cache)


@@ -89,7 +95,7 @@ class FakeFetcher(Fetcher):
     def fetch(self, url: str) -> bytes:
         self._fetch_count += 1
         if url not in self._resources:
     def fetch(self, url: str) -> bytes:
         self._fetch_count += 1
         if url not in self._resources:

-            raise requests.HTTPError("URL not found")
+            raise requests.HTTPError("URL not found", url)

         return self._resources[url]
 
     def request_count(self) -> int:
         return self._resources[url]
 
     def request_count(self) -> int: