Introducing pandoc-citeproc for bibliographic rendering

eoaconvert.py

@@ -1,6 +1,6 @@
 #!/usr/bin/env python3
 # -*- coding: utf-8; mode: python -*-
-# Time-stamp: <2018-01-22 13:22:15 (kthoden)>
+# Time-stamp: <2018-01-22 16:45:16 (kthoden)>
 
 # license?
 __version__= "1.0"
@@ -21,6 +21,7 @@
 import re
 import string
 import shlex
+import json
 import subprocess
 import sys
 import shutil
@@ -261,7 +262,9 @@ def createBibEntryAuthorYear(bibEntry, boolSameAuthor):
     if bibEntry.entrytype() == "newspaper":
         strBibEntry = strAuthor + " (" + bibEntry.labelyear() + bibEntry.labelyearsuffix() + "). <i>" + bibEntry.title() + "</i>"
 
-    return(sanitize_bibentry(strBibEntry))
+    print(strBibEntry)
+
+    return sanitize_bibentry(strBibEntry)
 # def createBibEntryAuthorYear ends here
 
 def createBibEntryNumeric(bibEntry):
@@ -293,7 +296,7 @@ def createBibEntryNumeric(bibEntry):
         strBibEntry = strAuthor + " (" + bibEntry.labelyear() + bibEntry.labelyearsuffix() + ") <i>" + bibEntry.title() + "</i>."
 
     return strBibEntry
-# def createBibEntryAuthorYear ends here
+# def createBibEntryNumeric ends here
 
 def pdf_burst(input_file, tmpDir):
     """Split PDF file into single pages"""
@@ -893,24 +896,28 @@ def cleanup():
 print("-----------------------------------------------------")
 print("Working on Page Numbers for References")
 listAuxFiles = glob.glob(os.getcwd() + "/*.aux")
-for strFile in listAuxFiles:
-    tmpFile = open(strFile, "r")
-    lines = tmpFile.readlines()
-    tmpFile.close()
-    for line in lines:
-        # hyperref makes the lines much much longer
-        # \newlabel{BL}{{1.1}{4}{Forschungsüberblick zur Literatur über Alvarus Thomas}{section.1.1}{}}
-        # \newlabel{BL}{{1.1}{4}}
-        matched_label = re.match(r'\\newlabel\{(.*?)\}\{\{(.*?)\}\{(.*?)\}', line)
-        # matchObjectLabel = re.match(r'\newlabel\{(.*?)\}', line)
-        if matched_label:
-            # matchObjectPage = re.match(r'(.*?)\}\{(\d{1,})\}\}$', line)
-            # if matchObjectPage:
-            dictPagelabels[matched_label.group(1)] = matched_label.group(3)
-        # parsing out information on cite works
-        matched_citation = re.match(r'\\abx@aux@cite{(.*?)}', line)
-        if matched_citation is not None:
-            set_citations.add(matched_citation.group(1))
+if len(listAuxFiles) == 0:
+    print("No aux file found. Exiting")
+    sys.exit(1)
+else:
+    for strFile in listAuxFiles:
+        tmpFile = open(strFile, "r")
+        lines = tmpFile.readlines()
+        tmpFile.close()
+        for line in lines:
+            # hyperref makes the lines much much longer
+            # \newlabel{BL}{{1.1}{4}{Forschungsüberblick zur Literatur über Alvarus Thomas}{section.1.1}{}}
+            # \newlabel{BL}{{1.1}{4}}
+            matched_label = re.match(r'\\newlabel\{(.*?)\}\{\{(.*?)\}\{(.*?)\}', line)
+            # matchObjectLabel = re.match(r'\newlabel\{(.*?)\}', line)
+            if matched_label:
+                # matchObjectPage = re.match(r'(.*?)\}\{(\d{1,})\}\}$', line)
+                # if matchObjectPage:
+                dictPagelabels[matched_label.group(1)] = matched_label.group(3)
+            # parsing out information on cite works
+            matched_citation = re.match(r'\\abx@aux@cite{(.*?)}', line)
+            if matched_citation is not None:
+                set_citations.add(matched_citation.group(1))
 
 print(dictPagelabels)
 print(set_citations)
@@ -949,6 +956,20 @@ def cleanup():
     HAS_BIBLIOGRAPHY = False
     input()
 
+# the new solution: pandoc-citeproc
+interim_bib_json_file = (options.filename) + "-bib.json"
+citeproc_command = "pandoc-citeproc --bib2json  %s" % bib_database + ".bib"
+logging.debug(citeproc_command)
+citeproc_arguments = shlex.split(citeproc_command)
+citeproc_process = subprocess.Popen(citeproc_arguments, stdout=subprocess.PIPE)
+citeproc_json = citeproc_process.stdout.read()
+
+jsonObj = json.loads(citeproc_json)
+
+for x in jsonObj:
+    print(x["title"])
+
+# the old solution
 # Copy interim .bbl-File to interim bib.tex file
 interim_bibtex_file = (options.filename) + "bib.tex"
 try: