feat(parse): add/update/rename parser presets

thi-ng · Apr 19, 2020 · 12f2499 · 12f2499
1 parent e16426b
commit 12f2499
Show file tree

Hide file tree

Showing 7 changed files with 92 additions and 59 deletions.
diff --git a/packages/parse/src/presets/alpha.ts b/packages/parse/src/presets/alpha.ts
@@ -1,23 +1,23 @@
+import { ALPHA as _ALPHA, ALPHA_NUM as _ALPHA_NUM } from "@thi.ng/strings";
+import { oneOf } from "../prims/one-of";
 import { range } from "../prims/range";
-import { alt } from "../combinators/alt";
-import { DIGIT } from "./digits";
 
 /**
  * Matches single char in `a` - `z` range.
  */
-export const LOWER_CASE = range<string>("a", "z");
+export const LOWER_CASE = range("a", "z");
 
 /**
  * Matches single char in `A` - `Z` range.
  */
-export const UPPER_CASE = range<string>("A", "Z");
+export const UPPER_CASE = range("A", "Z");
 
 /**
  * Matches single in {@link LOWER_CASE} or {@link UPPER_CASE}.
  */
-export const ALPHA = alt([LOWER_CASE, UPPER_CASE]);
+export const ALPHA = oneOf(_ALPHA);
 
 /**
  * Matches single in {@link ALPHA} or {@link DIGIT}.
  */
-export const ALPHA_NUM = alt([ALPHA, DIGIT]);
+export const ALPHA_NUM = oneOf(_ALPHA_NUM);
diff --git a/packages/parse/src/presets/bits.ts b/packages/parse/src/presets/bits.ts
@@ -0,0 +1,8 @@
+import { oneOf } from "../prims/one-of";
+import { xfInt } from "../xform/number";
+import { repeat } from "../combinators/repeat";
+import { xform } from "../combinators/xform";
+
+export const BIT = oneOf("01");
+
+export const BINARY_UINT = xform(repeat(BIT, 1, 32, "uint"), xfInt(2));
diff --git a/packages/parse/src/presets/digits.ts b/packages/parse/src/presets/digits.ts
@@ -1,32 +1,12 @@
-import { alt } from "../combinators/alt";
-import { oneOrMore, zeroOrMore } from "../combinators/repeat";
+import { oneOrMore } from "../combinators/repeat";
 import { range } from "../prims/range";
 
 /**
  * Matches single decimal digit.
  */
-export const DIGIT = range<string>("0", "9", "digit");
-
-/**
- * Matches single hex digit (case insensitive).
- */
-export const HEX_DIGIT = alt([
-    DIGIT,
-    range<string>("a", "f"),
-    range<string>("A", "F"),
-]);
-
-/**
- * Matches zero or more {@link DIGIT}s.
- */
-export const DIGITS_0 = zeroOrMore(DIGIT);
+export const DIGIT = range("0", "9", "digit");
 
 /**
  * Matches one or more {@link DIGIT}s.
  */
-export const DIGITS_1 = oneOrMore(DIGIT);
-
-/**
- * Matches one or more {@link HEX_DIGIT}s.
- */
-export const HEX_DIGITS_1 = oneOrMore(HEX_DIGIT);
+export const DIGITS = oneOrMore(DIGIT);
diff --git a/packages/parse/src/presets/escape.ts b/packages/parse/src/presets/escape.ts
@@ -0,0 +1,22 @@
+import { always } from "../prims/always";
+import { litD } from "../prims/lit";
+import { seq } from "../combinators/seq";
+import { hoist } from "../xform/hoist";
+import { xform } from "../combinators/xform";
+import { repeat } from "../combinators/repeat";
+import { HEX_DIGIT } from "./hex";
+import { stringD } from "../prims/string";
+import { xfInt } from "../xform/number";
+
+export const ESC = hoist(seq([litD("\\"), always()], "esc"));
+
+/**
+ * Matches a single `\uNNNN` escaped unicode hex literal and transforms
+ * it into it actual character via `String.fromCharCode()`.
+ */
+export const UNICODE = xform(
+    seq([stringD("\\u"), repeat(HEX_DIGIT, 4, 4)], "unicode"),
+    ($, ctx) => (
+        ($!.result = String.fromCharCode(xfInt(16)($, ctx)!.result)), $
+    )
+);
diff --git a/packages/parse/src/presets/hex.ts b/packages/parse/src/presets/hex.ts
@@ -0,0 +1,21 @@
+import { HEX } from "@thi.ng/strings";
+import { oneOrMore, repeat } from "../combinators/repeat";
+import { xform } from "../combinators/xform";
+import { oneOf } from "../prims/one-of";
+import { xfInt } from "../xform/number";
+
+/**
+ * Matches single hex digit (case insensitive).
+ */
+export const HEX_DIGIT = oneOf(HEX);
+
+/**
+ * Matches one or more {@link HEX_DIGIT}s.
+ */
+export const HEX_DIGITS = oneOrMore(HEX_DIGIT);
+
+/**
+ * Matches 1-8 successive {@link HEX_DIGIT} and transforms with
+ * {@link xfInt} to JS number.
+ */
+export const HEX_UINT = xform(repeat(HEX_DIGIT, 1, 8, "uint"), xfInt(16));
diff --git a/packages/parse/src/presets/numbers.ts b/packages/parse/src/presets/numbers.ts
@@ -1,47 +1,49 @@
 import { alt } from "../combinators/alt";
 import { maybe } from "../combinators/maybe";
+import { zeroOrMore } from "../combinators/repeat";
 import { seq } from "../combinators/seq";
 import { xform } from "../combinators/xform";
 import { lit } from "../prims/lit";
 import { oneOf } from "../prims/one-of";
+import { comp } from "../xform/comp";
+import { join } from "../xform/join";
 import { xfFloat, xfInt } from "../xform/number";
-import { DIGITS_0, DIGITS_1, HEX_DIGITS_1 } from "./digits";
+import { xfID } from "../xform/with-id";
+import { DIGIT, DIGITS } from "./digits";
 
 /**
  * Matches single `+` or `-` char.
  */
 export const SIGN = maybe(oneOf("-+"));
 
 /**
- * Matches optionally signed {@link DIGITS_1} and result transformed w/
+ * Matches optionally signed {@link DIGITS} and result transformed w/
  * {@link xfInt} to JS number.
  */
-export const INT = xform(seq([SIGN, DIGITS_1], "int"), xfInt());
+export const INT = xform(seq([SIGN, DIGITS], "int"), xfInt());
 
 /**
- * Matches same as {@link DIGITS_1}, but result transformed w/
+ * Matches same as {@link DIGITS} but result transformed w/
  * {@link xfInt} to JS number.
  */
-export const UINT = xform(DIGITS_1, xfInt());
+export const UINT = xform(DIGITS, comp(xfID("uint"), xfInt()));
 
-/**
- * Matches same as {@link HEX_DIGITS_1}, but result transformed w/
- * {@link xfInt} to JS number.
- */
-export const HEX_UINT = xform(HEX_DIGITS_1, xfInt(16));
-
-const EXP = maybe(seq([maybe(oneOf("eE")), SIGN, DIGITS_1]));
+const EXP = maybe(seq([maybe(oneOf("eE")), SIGN, DIGITS]));
 
 const DOT = lit(".");
 
-const FRACT0 = maybe(seq([DOT, DIGITS_0]));
-const FRACT1 = seq([DOT, DIGITS_1]);
+const FRACT0 = maybe(seq([DOT, zeroOrMore(DIGIT)]));
+const FRACT1 = seq([DOT, DIGITS]);
+
+const _REAL = seq([SIGN, alt([FRACT1, seq([DIGITS, FRACT0])]), EXP], "real");
+
+/**
+ * Matches IEEE754 floating point number.
+ */
+export const REAL = join(_REAL);
 
 /**
- * Matches IEEE754 floating point number and transforms result w/
- * {@link xfFloat}.
+ * Like {@link REAL} but transforms result w/ {@link xfFloat} to JS
+ * number.
  */
-export const FLOAT = xform(
-    seq([SIGN, alt([FRACT1, seq([DIGITS_1, FRACT0])]), EXP], "float"),
-    xfFloat
-);
+export const FLOAT = xform(_REAL, xfFloat);
diff --git a/packages/parse/src/presets/whitespace.ts b/packages/parse/src/presets/whitespace.ts
@@ -1,16 +1,16 @@
 import { discard } from "../combinators/discard";
 import { oneOrMore, zeroOrMore } from "../combinators/repeat";
-import { oneOf } from "../prims/one-of";
+import { oneOf, oneOfD } from "../prims/one-of";
 
 /**
- * Matches single whitespace char.
+ * Matches & discards single whitespace char: ` \t\n\r`.
  */
-export const WS = oneOf(" \t\n\r", "ws");
+export const WS = oneOfD(" \t\n\r");
 
 /**
- * Matches single space or tab char.
+ * Matches & discards single space or tab char.
  */
-export const SPACE = oneOf(" \t", "space");
+export const SPACE = oneOfD(" \t");
 
 /**
  * Matches single `\n` or `\r` char.
@@ -20,24 +20,24 @@ export const NL = oneOf("\n\r");
 /**
  * Matches & discards single `\n` or `\r` char.
  */
-export const DNL = discard(NL);
+export const DNL = oneOfD("\n\r");
 
 /**
  * Zero or more {@link WS}. Result will be discarded.
  */
-export const WS_0 = discard(zeroOrMore(WS));
+export const WS0 = discard(zeroOrMore(WS));
 
 /**
  * One or more {@link WS}. Result will be discarded.
  */
-export const WS_1 = discard(oneOrMore(WS));
+export const WS1 = discard(oneOrMore(WS));
 
 /**
  * Zero or more {@link SPACE}. Result will be discarded.
  */
-export const SPACE_0 = discard(zeroOrMore(SPACE));
+export const SPACES0 = discard(zeroOrMore(SPACE));
 
 /**
  * One or more {@link SPACE}. Result will be discarded.
  */
-export const SPACE_1 = discard(oneOrMore(SPACE));
+export const SPACES = discard(oneOrMore(SPACE));