From 4cb48dd0c0b5cb1b6b86dd48a203a466232d9f6c Mon Sep 17 00:00:00 2001
From: Klaus Thoden <kthoden@mpiwg-berlin.mpg.de>
Date: Mon, 22 Jan 2018 18:59:01 +0100
Subject: [PATCH] A library function for formatting the bibliography

---
 libeoaconvert.py | 41 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 41 insertions(+)

diff --git a/libeoaconvert.py b/libeoaconvert.py
index c321d1e..7341718 100644
--- a/libeoaconvert.py
+++ b/libeoaconvert.py
@@ -7,6 +7,8 @@
 import shlex
 import logging
 import configparser
+from lxml.html import soupparser
+
 
 ##################################
 # Reading the configuration file #
@@ -126,3 +128,42 @@ def deb_var(obj):
     name = [name for name in globals() if globals()[name] is obj][0]
     print("DEBUG: %s: %s" % (name, obj))
 # def deb_var ends here
+
+# next function adapted from TEI2EOADjango
+def format_citations(used_citekeys, bibdata):
+    """Return a formatted xmlstring of the used citations"""
+
+    md_file_header = "---\nlang: en\ntitle: Citations\n...\n\n"
+
+    with open("tmp_files/used_citations.md", "w") as citation_formatter:
+        citation_formatter.write(md_file_header)
+        citation_formatter.write("# Full parentheses\n")
+        for entry in used_citekeys:
+            citation_formatter.write("[@%s]\n" % entry)
+        citation_formatter.write("\n# Year parentheses\n")
+        for entry in used_citekeys:
+            citation_formatter.write("@%s\n" % entry)
+        citation_formatter.write("\n# References\n")
+
+    command = "pandoc -o tmp_files/formatted_citations.html -t html --filter=pandoc-citeproc --bibliography=%s --csl=/Users/kthoden/EOAKram/dev/eoa-csl/eoa.csl tmp_files/used_citations.md" % bibdata
+    arguments = shlex.split(command)
+    logging.info("Using external command pandoc with command %s" % command)
+    subprocess.call(arguments)
+
+    with open("tmp_files/formatted_citations.html", "r") as ding:
+        dd = soupparser.fromstring(ding, features="html.parser")
+
+    references = dd.xpath("//div[@class='references']")
+    return references
+
+    # full_paren_cites = dd.select("#full-parentheses ~ p > span")
+    # year_paren_cites = dd.select("#year-parentheses ~ p > span")
+    # citation_dict = {}
+    # for entry in used_citekeys:
+    #     title = (bibdata[entry]['title'])
+    #     full_paren = dd.select("#full-parentheses ~ p > span[data-cites='%s']" % entry)[0].text
+    #     year_paren = dd.select("#year-parentheses ~ p > span[data-cites='%s']" % entry)[0].text
+    #     citation_dict[entry] = (full_paren, year_paren, title)
+    # return citation_dict
+
+# def format_citations ends here