src/paca.mjs (tokenizeRegexStep): Support tokenizing range exps {m,n}

1 files changed, 540 insertions, 0 deletions

diff --git a/tests/paca.mjs b/tests/paca.mjs
index 82102f1..4f085e4 100644
--- a/tests/paca.mjs
+++ b/tests/paca.mjs
@@ -353,6 +353,546 @@ const test_tokenizeRegexStep = t => {
 			);
 		}
 	});
+
+	t.testing("multichar range operator {m,n} is parsed", () => {
+		const table = [{
+			regex: "a{1,2}",
+			steps: [{
+				in: {
+					out:     [],
+					state:   ConcatStep.ACCEPTING,
+					context: null,
+				},
+				out: {
+					out:     ["a", { operator: "concat" }],
+					state:   ConcatStep.ACCEPTING,
+					context: null,
+				},
+			}, {
+				in: {
+					out:     [],
+					state:   ConcatStep.ACCEPTING,
+					context: null,
+				},
+				out: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [],
+						to:    [],
+						where: "from",
+					},
+				},
+			}, {
+				in: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [],
+						to:    [],
+						where: "from",
+					},
+				},
+				out: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [ "1" ],
+						to:    [],
+						where: "from",
+					},
+				},
+			}, {
+				in: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [ "1" ],
+						to:    [],
+						where: "from",
+					},
+				},
+				out: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [ "1" ],
+						to:    [],
+						where: "to",
+					},
+				},
+			}, {
+				in: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [ "1" ],
+						to:    [],
+						where: "to",
+					},
+				},
+				out: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [ "1" ],
+						to:    [ "2" ],
+						where: "to",
+					},
+				},
+			}, {
+				in: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [ "1" ],
+						to:    [ "2" ],
+						where: "to",
+					},
+				},
+				out: {
+					out:     [{
+						operator: "range",
+						from:     1,
+						to:       2,
+					}],
+					state:   ConcatStep.ACCEPTING,
+					context: null,
+				},
+			}],
+		}, {
+			regex: "a{,2}",
+			steps: [{
+				in: {
+					out:     [],
+					state:   ConcatStep.ACCEPTING,
+					context: null,
+				},
+				out: {
+					out:     ["a", { operator: "concat" }],
+					state:   ConcatStep.ACCEPTING,
+					context: null,
+				},
+			}, {
+				in: {
+					out:     [],
+					state:   ConcatStep.ACCEPTING,
+					context: null,
+				},
+				out: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [],
+						to:    [],
+						where: "from",
+					},
+				},
+			}, {
+				in: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [],
+						to:    [],
+						where: "from",
+					},
+				},
+				out: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [],
+						to:    [],
+						where: "to",
+					},
+				},
+			}, {
+				in: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [],
+						to:    [],
+						where: "to",
+					},
+				},
+				out: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [],
+						to:    [ "2" ],
+						where: "to",
+					},
+				},
+			}, {
+				in: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [],
+						to:    [ "2" ],
+						where: "to",
+					},
+				},
+				out: {
+					out:     [{
+						operator: "range",
+						from:     -1,
+						to:       2,
+					}],
+					state:   ConcatStep.ACCEPTING,
+					context: null,
+				},
+			}],
+		}, {
+			regex: "a{1,}",
+			steps: [{
+				in: {
+					out:     [],
+					state:   ConcatStep.ACCEPTING,
+					context: null,
+				},
+				out: {
+					out:     ["a", { operator: "concat" }],
+					state:   ConcatStep.ACCEPTING,
+					context: null,
+				},
+			}, {
+				in: {
+					out:     [],
+					state:   ConcatStep.ACCEPTING,
+					context: null,
+				},
+				out: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [],
+						to:    [],
+						where: "from",
+					},
+				},
+			}, {
+				in: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [],
+						to:    [],
+						where: "from",
+					},
+				},
+				out: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [ "1" ],
+						to:    [],
+						where: "from",
+					},
+				},
+			}, {
+				in: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [ "1" ],
+						to:    [],
+						where: "from",
+					},
+				},
+				out: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [ "1" ],
+						to:    [],
+						where: "to",
+					},
+				},
+			}, {
+				in: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [ "1" ],
+						to:    [],
+						where: "to",
+					},
+				},
+				out: {
+					out:     [{
+						operator: "range",
+						from:     1,
+						to:       -1,
+					}],
+					state:   ConcatStep.ACCEPTING,
+					context: null,
+				},
+			}],
+		}, {
+			regex: "a{,}",
+			steps: [{
+				in: {
+					out:     [],
+					state:   ConcatStep.ACCEPTING,
+					context: null,
+				},
+				out: {
+					out:     ["a", { operator: "concat" }],
+					state:   ConcatStep.ACCEPTING,
+					context: null,
+				},
+			}, {
+				in: {
+					out:     [],
+					state:   ConcatStep.ACCEPTING,
+					context: null,
+				},
+				out: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [],
+						to:    [],
+						where: "from",
+					},
+				},
+			}, {
+				in: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [],
+						to:    [],
+						where: "from",
+					},
+				},
+				out: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [],
+						to:    [],
+						where: "to",
+					},
+				},
+			}, {
+				in: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [],
+						to:    [],
+						where: "to",
+					},
+				},
+				out: {
+					out:     [{
+						operator: "range",
+						from:     -1,
+						to:       -1,
+					}],
+					state:   ConcatStep.ACCEPTING,
+					context: null,
+				},
+			}],
+		}, {
+			regex: "a{123,456}",
+			steps: [{
+				in: {
+					out:     [],
+					state:   ConcatStep.ACCEPTING,
+					context: null,
+				},
+				out: {
+					out:     ["a", { operator: "concat" }],
+					state:   ConcatStep.ACCEPTING,
+					context: null,
+				},
+			}, {
+				in: {
+					out:     [],
+					state:   ConcatStep.ACCEPTING,
+					context: null,
+				},
+				out: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [],
+						to:    [],
+						where: "from",
+					},
+				},
+			}, {
+				in: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [],
+						to:    [],
+						where: "from",
+					},
+				},
+				out: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [ "1" ],
+						to:    [],
+						where: "from",
+					},
+				},
+			}, {
+				in: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [ "1" ],
+						to:    [],
+						where: "from",
+					},
+				},
+				out: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [ "1", "2" ],
+						to:    [],
+						where: "from",
+					},
+				},
+			}, {
+				in: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [ "1", "2" ],
+						to:    [],
+						where: "from",
+					},
+				},
+				out: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [ "1", "2", "3" ],
+						to:    [],
+						where: "from",
+					},
+				},
+			}, {
+				in: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [ "1", "2", "3" ],
+						to:    [],
+						where: "from",
+					},
+				},
+				out: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [ "1", "2", "3" ],
+						to:    [],
+						where: "to",
+					},
+				},
+			}, {
+				in: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [ "1", "2", "3" ],
+						to:    [],
+						where: "to",
+					},
+				},
+				out: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [ "1", "2", "3" ],
+						to:    [ "4" ],
+						where: "to",
+					},
+				},
+			}, {
+				in: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [ "1", "2", "3" ],
+						to:    [ "4" ],
+						where: "to",
+					},
+				},
+				out: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [ "1", "2", "3" ],
+						to:    [ "4", "5" ],
+						where: "to",
+					},
+				},
+			}, {
+				in: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [ "1", "2", "3" ],
+						to:    [ "4", "5" ],
+						where: "to",
+					},
+				},
+				out: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [ "1", "2", "3" ],
+						to:    [ "4", "5", "6" ],
+						where: "to",
+					},
+				},
+			}, {
+				in: {
+					out:     [],
+					state:   ConcatStep.RANGE,
+					context: {
+						from:  [ "1", "2", "3" ],
+						to:    [ "4", "5", "6" ],
+						where: "to",
+					},
+				},
+				out: {
+					out:     [{
+						operator: "range",
+						from:     123,
+						to:       456,
+					}],
+					state:   ConcatStep.ACCEPTING,
+					context: null,
+				},
+			}],
+		}];
+		for (const case_ of table) {
+			const stepFn = tokenizeRegexStep(case_.regex);
+			for (const i in case_.regex) {
+				const step = case_.steps[i];
+				const char = case_.regex[i];
+				t.assertEq(
+					stepFn(step.in, char, Number(i)),
+					step.out,
+				);
+			}
+		}
+	});
 };
 
 const test_tokenizeRegexFn = t => {