preliminary parsing facility of BibTeX files: for now strings remain …

…unstructured

rascal/src/io/bibtex/Parser.rsc

@@ -0,0 +1,24 @@
+@contributor{Vadim Zaytsev - vadim@grammarware.net - SWAT, CWI}
+module io::bibtex::Parser
+
+import io::bibtex::Syntax;
+import ParseTree;
+import String;
+import IO;
+
+alias BibLib = list[BibEntry];
+alias BibEntry = tuple[str kind, str name, map[str,str] attrs];
+
+public BibLib loc2bib(loc l) = str2bib(readFile(l));
+BibLib str2bib(str s) = library2list(parse(#BibLibrary,trim(replaceAll(s,"\uC2A0"," "))));
+
+BibLib library2list(BibLibrary b) = [mapEntry(e) | OneBibEntry e <- b.es];
+
+BibEntry mapEntry(OneBibEntry e) = <"<e.kind.name>","<e.name>",mapKVs(e.pairs)>;
+
+map[str,str] mapKVs({BibPair ","}+ ps) = ("<p.key>":"<p.val>" | BibPair p <- ps);
+
+public void main()
+{
+	loc2bib(|home:///workspace/zaytsev.bib|);
+} 

rascal/src/io/bibtex/Syntax.rsc

@@ -0,0 +1,25 @@
+@contributor{Vadim Zaytsev - vadim@grammarware.net - SWAT, CWI}
+module io::bibtex::Syntax
+
+layout L = WS;
+lexical WS = [\uC2A0 \ \n\r\t]* !>> [\uC2A0 \ \n\r\t]; // note the nonbreakable space (0xC2A0 in Unicode)
+
+syntax OneBibEntry = BibEntryType kind "{" BibEntryName name "," {BibPair ","}+ pairs ","? "}";
+lexical BibEntryType = "@" [a-zA-Z]+ name;
+lexical BibEntryName = ![,]+ >> [,];
+syntax BibPair = BibKey key "=" BibValue val;
+lexical BibKey = [a-z]+;
+
+lexical BibValue
+	= BibValueQ
+	| BibValueC
+	| [a-zA-Z0-9]+ !>> [a-zA-Z0-9]
+	;
+
+lexical BibValueQ = [\"] BQElement* [\"] ;
+lexical BQElement = ![\"\\{}] | [\\] [\"\'`&$%a-zA-Z] | BibValueC;
+
+lexical BibValueC = [{] BCElement* [}];
+lexical BCElement = BibValueC | ![{}];
+
+start syntax BibLibrary = OneBibEntry+ es;