maybe rm empty txts

1 files changed, 7 insertions, 1 deletions

diff --git a/s2-pdf-report.py b/s2-pdf-report.py
index 6ef5c0f7..7c89381f 100644
--- a/s2-pdf-report.py
+++ b/s2-pdf-report.py
@@ -12,10 +12,11 @@ PDF_DIR = 'datasets/s2/pdf'
 def pdf_report_first_pages():
   rows = []
   for fn in glob.iglob('{}/**/*.txt'.format(PDF_DIR), recursive=True):
-    row = process_paper(fn)
+    row, institutions = process_paper(fn)
     print(row)
     rows.append(row)
   write_report('reports/first_pages.html', title='First pages', keys=None, rows=rows)
+  write_report('reports/institutions.html', title='Institutions', keys=None, rows=institutions)
   print("Wrote {} rows".format(len(rows)))
 
 def process_paper(fn):
@@ -27,6 +28,7 @@ def process_paper(fn):
   with open(fn, 'r') as f:
     lines = []
     emails = []
+    institutions = []
     authors = [ (a[0], a[1], a[1].lower(),) for a in paper.authors ]
     journal = paper.journal.lower()
     found_authors = []
@@ -55,6 +57,7 @@ def process_paper(fn):
         # lines.append(NameLine(line))
         continue
       if 'university' in l or 'universiteit' in l or 'research center' in l:
+        institutions.append(line)
         lines.append(BoldLine(line))
         continue
       lines.append(line)
@@ -63,6 +66,9 @@ def process_paper(fn):
       lines,
       found_authors,
       emails,
+    ], [
+      paper_id,
+      sorted(institutions),
     ]
 
 class NameLine(object):