Resolve bidirectional text as UAX #9 does, passing all of Unicode's bidi tests

1 files changed, 37 insertions, 0 deletions

diff --git a/text/text.h b/text/text.h
new file mode 100644
index 0000000..e4e50ac
--- /dev/null
+++ b/text/text.h
@@ -0,0 +1,37 @@
+#ifndef VEGA_TEXT_H
+#define VEGA_TEXT_H
+#include <stdbool.h>
+#include <stddef.h>
+#include <stdint.h>
+
+/* Text as people write it (docs/design/004-text.md): the Unicode properties
+ * it needs (tables generated from the UCD by text/unicode.py) and the
+ * algorithms over them, checked against Unicode's conformance tests. */
+
+/* Bidi_Class (UAX #9), in the order of text/unicode.py. */
+typedef enum {
+    VEGA_BIDI_L, VEGA_BIDI_R, VEGA_BIDI_AL, VEGA_BIDI_EN, VEGA_BIDI_ES, VEGA_BIDI_ET, VEGA_BIDI_AN, VEGA_BIDI_CS,
+    VEGA_BIDI_NSM, VEGA_BIDI_BN, VEGA_BIDI_B, VEGA_BIDI_S, VEGA_BIDI_WS, VEGA_BIDI_ON, VEGA_BIDI_LRE, VEGA_BIDI_LRO,
+    VEGA_BIDI_RLE, VEGA_BIDI_RLO, VEGA_BIDI_PDF, VEGA_BIDI_LRI, VEGA_BIDI_RLI, VEGA_BIDI_FSI, VEGA_BIDI_PDI,
+    VEGA_BIDI_CLASSES
+} VegaBidiClass;
+VegaBidiClass vega_bidi_class(uint32_t code);
+/* A paired bracket: 1 opening, 2 closing (its pair in *pair), 0 none. */
+int vega_bracket(uint32_t code, uint32_t *pair);
+/* The character that mirrors code (Bidi_Mirroring_Glyph), or 0. */
+uint32_t vega_mirror(uint32_t code);
+
+/* The embedding levels of one paragraph by the Unicode Bidirectional
+ * Algorithm (UAX #9, rules P2 to I2), then L1 as for one line: classes are
+ * its characters' Bidi_Class; codes, when given, are the characters, so
+ * that paired brackets resolve (rule N0); direction is VEGA_BIDI_AUTO (from
+ * its first strong character, P2-P3), VEGA_BIDI_LTR or VEGA_BIDI_RTL. Each
+ * character's level goes to levels, VEGA_BIDI_REMOVED for those rule X9
+ * removes (embedding controls and boundary neutrals). Returns the
+ * paragraph's level, or -1 without memory. */
+enum { VEGA_BIDI_AUTO = 0, VEGA_BIDI_LTR = 1, VEGA_BIDI_RTL = 2, VEGA_BIDI_REMOVED = 0xff };
+int vega_bidi_levels(const VegaBidiClass *classes, const uint32_t *codes, size_t count, int direction, uint8_t *levels);
+/* The visual order of one line (rule L2), left to right: the positions of
+ * its characters, those removed left out; returns how many. */
+size_t vega_bidi_order(const uint8_t *levels, size_t count, uint32_t *order);
+#endif