1 files changed, 124 insertions, 0 deletions
diff --git a/scraper/s2-dump-db-pdf-urls.py b/scraper/s2-dump-db-pdf-urls.py
new file mode 100644
index 00000000..dbcb91d8
--- /dev/null
+++ b/scraper/s2-dump-db-pdf-urls.py
@@ -0,0 +1,124 @@
+import os
+import glob
+import simplejson as json
+import click
+from urllib.parse import urlparse
+import operator
+from util import *
+
+PAPER_JSON_DIR = 'datasets/s2/db_papers'
+
+@click.command()
+def s2_dump_pdf_urls():
+  # loop over all the papers in db_papers
+  # get all the PDF urls, pick the best one
+  # store it and the paper id
+  # another script will fetch the urls from this process
+  rows = []
+  pdf_count = 0
+  ieee_count = 0
+  url_count = 0
+  doi_count = 0
+  empty_count = 0
+  domains = {}
+  pdf = []
+  doi = []
+  for fn in glob.iglob('{}/**/paper.json'.format(PAPER_JSON_DIR), recursive=True):
+    # if 'db_paper' in fn:
+    row = process_db_paper(fn)
+    # elif 'raw_paper' in fn:
+    # row = process_raw_paper(fn)
+    if row is not None:
+      rows.append(row)
+      if row[1] is not None:
+        pdf.append([row[0], row[1]])
+        pdf_count += 1
+      elif row[2] is not None:
+        doi.append([row[0], row[2]])
+        ieee_count += 1
+      elif row[3] is not None:
+        doi.append([row[0], row[3]])
+        doi_count += 1
+      elif row[4] is not None:
+        if 'pdf' not in row[4]:
+          doi.append([row[0], row[4]])
+        url_count += 1
+        domain = urlparse(row[4]).netloc
+        if domain in domains:
+          domains[domain] += 1
+        else:
+          domains[domain] = 1
+      else:
+        empty_count += 1
+  print("Wrote {} rows".format(len(rows)))
+  print("pdf count: {}".format(pdf_count))
+  print("ieee count: {}".format(ieee_count))
+  print("doi count: {}".format(doi_count))
+  print("url count: {}".format(url_count))
+  for domain, count in sorted(domains.items(), key=operator.itemgetter(1)):
+     print(" -- {} - {}".format(domain, count))
+  print("empty count: {}".format(empty_count))
+  write_csv('db_paper_pdf_list.csv', keys=['Paper ID', 'PDF URL', 'IEEE URL', 'DOI URL', 'Extra URL'], rows=rows)
+  write_csv('db_paper_pdf.csv', keys=None, rows=pdf)
+  write_csv('db_paper_doi.csv', keys=None, rows=doi)
+  # write_csv('raw_paper_pdf_list.csv', keys=['Paper ID', 'PDF URL', 'IEEE URL', 'DOI URL', 'Extra URL'], rows=rows)
+  # write_csv('raw_paper_pdf.csv', keys=None, rows=pdf)
+  # write_csv('raw_paper_doi.csv', keys=None, rows=doi)
+
+def process_db_paper(fn):
+  # print(fn)
+  paper = read_json(fn)
+  if paper is None:
+    return None
+  paper_id = paper['id']
+  pdf_url = None
+  ieee_url = None
+  doi_url = None
+  extra_url = None
+  if paper['s2PdfUrl']:
+    pdf_url = paper['s2PdfUrl']
+  for url in paper['pdfUrls']:
+    if 'ieeexplore.ieee.org' in url:
+      ieee_url = url
+    elif 'doi.org' in url:
+      doi_url = url
+    elif pdf_url is None and 'pdf' in url:
+      pdf_url = url
+    else:
+      extra_url = url
+  return [paper_id, pdf_url, ieee_url, doi_url, extra_url]
+
+def process_raw_paper(fn):
+  print(fn)
+  data = read_json(fn)
+  if 'paper' not in data:
+    print(data)
+    return
+  paper = data['paper']
+  if paper is None:
+    return None
+  paper_id = paper['id']
+  pdf_url = None
+  ieee_url = None
+  doi_url = None
+  extra_url = None
+  if 'primaryPaperLink' in paper and 'url' in paper['primaryPaperLink']:
+    primary_url = paper['primaryPaperLink']['url']
+    if 'pdf' in primary_url:
+      pdf_url = primary_url
+    elif 'doi' in primary_url:
+      doi_url = primary_url
+  for link in paper['links']:
+    url = link['url']
+    if 'ieeexplore.ieee.org' in url:
+      ieee_url = url
+    elif 'doi.org' in url:
+      doi_url = url
+    elif pdf_url is None and 'pdf' in url:
+      pdf_url = url
+    else:
+      extra_url = url
+  return [paper_id, pdf_url, ieee_url, doi_url, extra_url]
+
+if __name__ == '__main__':
+  s2_dump_pdf_urls()