Refactor code based on Wiley input file changes

* Add smart_open library so input files can be streamed directly from S3 based on a URI
* Remove unncessary dependencies
* Update test fixture from .xlsx to .csv file

.gitignore

@@ -132,4 +132,4 @@ dmypy.json
 *.pdf
 *.xlsx
 .DS_Store
-!/fixtures/test.xlsx
+!/fixtures/test.csv

Pipfile

@@ -4,10 +4,9 @@ verify_ssl = true
 name = "pypi"
 
 [packages]
-pandas = "*"
-openpyxl = "*"
 click = "*"
 coveralls = "*"
+smart-open = "*"
 
 [dev-packages]
 black = "==21.7b0"

awd/crossref.py

@@ -1,14 +1,12 @@
-import pandas
 import requests
+import smart_open
 
 
 def get_dois_from_spreadsheet(file):
-    """Retriev DOIs from the Wiley-provided spreadsheet."""
-    excel_data_df = pandas.read_excel(
-        file, sheet_name="MIT Article List", skiprows=range(0, 4)
-    )
-    for doi in excel_data_df["DOI"].tolist():
-        yield doi
+    """Retriev DOIs from the Wiley-provided CSV file."""
+    with smart_open.open(file, encoding="utf-8-sig") as csvfile:
+        for doi in csvfile.read().splitlines():
+            yield doi
 
 
 def get_crossref_work_from_doi(api_url, doi):

fixtures/test.csv

@@ -0,0 +1 @@
+10.1002/term.3131

tests/test_crossref.py

@@ -22,7 +22,7 @@ def work_record():
 
 
 def test_get_dois_from_spreadsheet():
-    dois = crossref.get_dois_from_spreadsheet("fixtures/test.xlsx")
+    dois = crossref.get_dois_from_spreadsheet("fixtures/test.csv")
     for doi in dois:
         assert doi == "10.1002/term.3131"