Generalization of the context-aware scanning architecture for all character-level context-free languagesLeber, Žiga (Avtor) Kosar, Tomaž (Mentor) context-aware scanningpseudo-scannerless parsingscannerless parsinggeneralized LRright-nulled GLRMultiple representation editors utilize multiple editor panes to display a program in different notations. The internal representations of editor panes for the visual and textual notation differ significantly. In the former case the internal representation is a tree, where there are no limitations placed on the syntax of textual labels inside the nodes, in the later case it is text where the syntax of the corresponding lexemes is severely restricted. The limitations follow from the use of the traditional parsing architecture. We present a novel scanner-based architecture, which lifts these limitations. It can directly parse all possible specifications and has a comparable power as a scannerless- architecture. The architecture is based on right-nulled generalized LR (RNGLR) parsing and is a generalization of the context-aware scanning architecture. Our architecture degenerates to context-aware scanning for deterministic character-level context free languages, therefore, the performance is comparable in that case. Our implementation is currently outperformed by context-aware scanning in combination with RNGLR by at most 72%, tested on JSON, ANSI C, ISO 7185:1990 Pascal, Fortran 90 and VS COBOL II with files up to 1MB in size. For nearly deterministic languages the performance degrades gracefully and is predominantly linear.[Ž. Leber]20262025-11-14 17:57:14Doktorsko delo/naloga95962UDK: 004.4'13:004.4'232(043.3)COBISS_ID: 276907267sl