python script to read RePEc .rdf files and create _data/wp.yml

_config.yml

@@ -167,6 +167,7 @@ exclude:
   - /assets/images/logo/*.pdf
   - /assets/images/logo/*.tex
   - /assets/images/logo/*.sh
+  - /assets/RePEc/rdf2yml.py
 keep_files:
   - .git
   - /assets/images/*

assets/RePEc/rdf2yml.py

+import os
+import re
+import sys
+
+# Parse .rdf files
+wp_first_year = 2011
+wps = dict()
+wp_year = wp_first_year
+while True:
+    filename = 'dynare-' + str(wp_year) + '.rdf'
+    if not os.path.isfile(filename):
+        break
+
+    fh = open(filename, 'r')
+    wpn = -1
+    wp = dict()
+    authors = []
+    urls = []
+    url_function = []
+    for line in fh:
+        spl = re.split('[:]', line)
+        key = spl[0]
+
+        if key == 'Template-Type' or key == '# Local Variables':
+            if wpn >= 1:
+                wp['urls'] = urls
+                wp['url_function'] = url_function
+                wp['authors'] = authors
+                wps[wpn] = wp
+
+            wpn = -1
+            wp = dict()
+            authors = []
+            urls = []
+            url_function = []
+        elif key == 'Number':
+            wpn = int(spl[1])
+        elif key == 'Title':
+            wp['title'] = ":".join(spl[1:]).strip()
+        elif key == 'File-URL':
+            urls.append(':'.join((spl[1:])).strip())
+        elif key == 'File-Function':
+            url_function.append(spl[1].strip())
+        elif key == 'Creation-Date':
+            spl = re.split('[-]', spl[1])
+            wp['year'] = int(spl[0])
+        elif key == 'Author-Name':
+            authors.append(spl[1].strip())
+
+    fh.close()
+    wp_year += 1
+
+# Write .yml file
+yml = open('wp.yml', 'w')
+for key in sorted(wps, reverse=True):
+    yml.write('- number: {:d}\n'.format(key))
+    yml.write('  authors: ')
+    firstauthor = True
+    for author in wps[key]['authors']:
+        name = author.split(',')
+        last = name[0]
+        first = " ".join(name[1:]).strip()
+        if firstauthor:
+            firstauthor = False
+        else:
+            yml.write(', ')
+        yml.write('{:s} {:s}'.format(first, last))
+    yml.write('\n')
+    yml.write('  title: "{:s}"\n'.format(wps[key]['title']))
+    yml.write('  year: {:d}\n'.format(wps[key]['year']))
+
+    # Have to do this because some entries are mal-formed (do not contain File-Function info)
+    if len(wps[key]['urls']) == 1:
+        yml.write('  url: {:s}\n'.format(wps[key]['urls'][0]))
+    else:
+        for idx, url in enumerate(wps[key]['urls']):
+            if wps[key]['url_function'][idx] == 'Main text':
+                yml.write('  url: {:s}\n'.format(wps[key]['urls'][idx]))
+            elif wps[key]['url_function'][idx] == 'Data used in the paper':
+                yml.write('  data_url: {:s}\n'.format(wps[key]['urls'][idx]))
+            elif wps[key]['url_function'][idx] == 'Source code of programs used in the paper' or \
+                 wps[key]['url_function'][idx] == 'Dynare source code used in the paper':
+                yml.write('  source_url: {:s}\n'.format(wps[key]['urls'][idx]))
+            else:
+                print(wps[key])
+                print('Unhandled url case: {:s}'.format(wps[key]['url_function'][idx]))
+                sys.exit()
+    yml.write('\n')
+
+yml.close()
+print('Done writing wp.yml')
+

wp/index.md

@@ -12,7 +12,7 @@ title: Working Papers
 <ol reversed start="{{ latestwp }}">
 {% for wp in site.data.wp limit:5 %}
 <li>
-  {{ wp.authors }} ({{ wp.year }}), "<a href="http://www.dynare.org/wp-repo/dynarewp0{{ wp.number }}.pdf">{{ wp.title }}</a>"
+  {{ wp.authors }} ({{ wp.year }}), "<a href="{{ wp.url }}">{{ wp.title }}</a>"
   {% if wp.source_url %}(<a href="{{ wp.source_url }}">source</a>){% endif %}
   {% if wp.data_url %}(<a href="{{ wp.data_url }}">data</a>){% endif %}
 </li>

wp/list.md

@@ -12,7 +12,7 @@ title: Working Papers
 <ol reversed start="{{ latestwp }}">
 {% for wp in site.data.wp %}
 <li>
-  {{ wp.authors }} ({{ wp.year }}), "<a href="http://www.dynare.org/wp-repo/dynarewp0{{ wp.number }}.pdf">{{ wp.title }}</a>"
+  {{ wp.authors }} ({{ wp.year }}), "<a href="{{ wp.url }}">{{ wp.title }}</a>"
   {% if wp.source_url %}(<a href="{{ wp.source_url }}">source</a>){% endif %}
   {% if wp.data_url %}(<a href="{{ wp.data_url }}">data</a>){% endif %}
 </li>