def repl_func(m): """ Replaces <CIT ID=0 /> with <REF ID=cit0 REFID=ref0 /> """ cit_id=m.group(1) # this is the actual unique identifier of a citation in the document ref_id=doc.citation_by_id[cit_id]["ref_id"] # this is the id of the reference the cit cites return u'<REF ID="'+safe_unicode(cit_id)+u'" REFID="'+safe_unicode(ref_id)+'" />'
def write(self, doc, filename): """ """ self.already_rendered=[] lines=[] lines.append('<?xml version="1.0" encoding="UTF-8"?>') lines.append('<!DOCTYPE PAPER SYSTEM "paper-structure-annotation.dtd">') lines.append("<PAPER>") lines.extend(self.writeMetadata(doc)) lines.extend(self.writeAbstract(doc)) lines.extend(self.writeBody(doc)) lines.extend(self.writeReferences(doc)) lines.append("</PAPER>") lines2=[] for line in lines: lines2.append(safe_unicode(line)) lines=lines2 ## text="\n".join(lines) f=codecs.open(filename,"w", encoding="utf-8",errors="ignore") ## f.writelines([line+"\n" for line in lines]) f.writelines(lines) f.close()