import sys import re syntax_description = """ A "span" is an integer (the "index") followed by an optional description of one or more sequence subsets. A subset is defined by a series of subset descriptions, separated by a comma. A subset description is either an integer, two integers separated by a hypen to indicate a range, or an integer followed only by a hyphen, which will include all the following elements of the span to the end. No spaces are allowed in a span. All indices are zero-origin. Span examples for an index of "3": 3 3(1) 3(0-4) 3(1-4,6-9) 3(5-) Note that some sequence subsets must include two numbers, for example, border lines in a table. """.strip() class Span: def __init__(self, count, spec): def parse_range(match): start, hyphen, end = match.groups() if hyphen is None and end is None: end = start elif end is None: end = count - 1 return [int(start), int(end)] self.all = True subset_pat = "[-\\d,]+" span_rgx = re.compile(f"(\d+)(\({subset_pat}\))*") match = span_rgx.match(spec) self.spec = spec if match and match.group(0) == spec: range_rgx = re.compile("(\d+)(-)?(\d+)?") self.index = int(match.group(1)) if (match.group(2)): self.subsets = match.group(2).strip("()").split(",") matches = [range_rgx.match(e) for e in self.subsets] self.ranges = [parse_range(e) if e else None for e in matches] self.all = False else: self.ranges = [[0, count-1]] else: print(f'The span specification "{spec}" is incorrect.\n\n{syntax_description}\n') sys.exit(1) def __str__(self): subsets = "all" if self.all else ",".join([f"{e[0]}-{e[1]}" for e in self.ranges]) return f"{self.index}[{subsets}]" def __repr__(self): return self.__str__() class Spanset: def __init__(self, count, span_specs): self.count = count self.spans = {} for specs in span_specs.split(): for spec in self.parse_spec(specs): print("Spanset spec:", spec) self.spans[spec.index] = spec def parse_spec(self, spec): named_spec = {"top" : ["0"], "bottom" : [str(self.count-1)], "head" : ["1"], "outer" : ["0", str(self.count-1)], "inner" : [str(e) for e in range(1,self.count-1)], "all" : [str(e) for e in range(0,self.count+1)] }.get(spec) if named_spec is None: return [Span(self.count, spec)] else: return [Span(self.count, e) for e in named_spec] if __name__ == "__main__": import pprint for s in [ Span(10, "0"), Span(10, "1"), Span(10, "2(2)"), Span(10, "3(2-)"), Span(10, "4(2-4)"), Span(10, "5(2-4)"), Span(10, "6(2-4,6)"), Span(10, "7(2,6-8)"), Span(10, "8(2,7-8,9-11,13-14)")]: print(s.spec, "->", s) print("Spanset") S = Spanset(10, "1") for name in "top bottom head outer inner all 1(2-3)".split(): print(name, "->", S.parse_spec(name)) print("Instantiate:") s = Spanset(10, "3(1-2,4-5)") print(s.tex_hline(3))