refactor: homogenize usage of auth-header

.github/workflows/test_tip.yml

@@ -2,6 +2,9 @@
 name: Source
 on: [push, release]
 
+env:
+  PAPERSCRAPER_EMAIL: jannis.born@gmx.de
+
 jobs:
   test-source-install:
     runs-on: ubuntu-latest

paperscraper/get_dumps/utils/chemrxiv/crossref_api.py

@@ -10,6 +10,7 @@
 """
 
 import logging
+import os
 import sys
 from time import sleep
 from typing import Dict, Generator, List, Optional
@@ -58,7 +59,9 @@ def __init__(
         self.end_date = end_date
         self.page_size = min(max(1, page_size), 1000)
         self.max_retries = max_retries
-        self.mailto = mailto
+        self.mailto = mailto or os.getenv(
+            "PAPERSCRAPER_EMAIL", "your_email@example.com"
+        )
         self.request_delay_seconds = max(0.0, request_delay_seconds)
 
     def iter_items(self) -> Generator[Dict, None, None]:
@@ -140,7 +143,7 @@ def _request(self, params: Dict) -> Dict:
 
         headers = {
             "Accept": "application/json",
-            "User-Agent": "paperscraper (Crossref fallback)",
+            "User-Agent": "paperscraper",
         }
 
         for attempt in range(self.max_retries):

paperscraper/pdf/fallbacks.py

@@ -4,6 +4,7 @@
 import datetime
 import io
 import logging
+import os
 import re
 import sys
 import threading
@@ -123,7 +124,7 @@ def fallback_bioc_pmc(
         bool: True if the XML file was successfully downloaded, False otherwise.
     """
     ncbi_tool = "paperscraper"
-    ncbi_email = "your_email@example.com"
+    ncbi_email = os.getenv("PAPERSCRAPER_EMAIL", "your_email@example.com")
 
     converter_url = "https://www.ncbi.nlm.nih.gov/pmc/utils/idconv/v1.0/"
     params = {

paperscraper/pdf/pdf.py

@@ -151,7 +151,7 @@ def save_pdf(
         api_keys = load_api_keys(api_keys)
     doi = paper_metadata["doi"]
     url = f"https://doi.org/{doi}"
-    user_agent = {"User-Agent": "paperscraper/1.0 (+https)"}
+    user_agent = {"User-Agent": "paperscraper"}
     success = False
     metadata_written = False
 

paperscraper/pubmed/pubmed.py

@@ -12,7 +12,10 @@
 logger = logging.getLogger(__name__)
 logger.setLevel(logging.INFO)
 
-PUBMED = PubMed(tool=os.getenv("NCBI_TOOL", "paperscraper"), email="abc@def.gh")
+PUBMED = PubMed(
+    tool="paperscraper",
+    email=os.getenv("PAPERSCRAPER_EMAIL", "your_email@example.com"),
+)
 
 pubmed_field_mapper = {"publication_date": "date"}