Add codepoint data and Unicode version

Signed-off-by: Mattias Andrée <m@maandree.se>

8 files changed, 422 insertions, 3 deletions

diff --git a/Makefile b/Makefile
index 7190f5f..cb1eb13 100644
--- a/Makefile
+++ b/Makefile
@@ -21,11 +21,14 @@ UNICODE_VERSION = 17.0.0
 
 
 OBJ =\
+	libcmap_unicode_version.o\
 	libcmap_block_list.o\
 	libcmap_script_list.o\
+	libcmap_codepoint_list.o\
 	libcmap_find_in_no_block.o\
 	libcmap_find_block.o\
 	libcmap_find_script.o\
+	libcmap_find_codepoint_data.o\
 	libcmap_sprint_range.o\
 	libcmap_snprint_range.o\
 	libcmap_fprint_range.o\
@@ -40,22 +43,30 @@ LOBJ = $(OBJ:.o=.lo)
 
 PARSERS =\
 	Scripts.parse\
-	Blocks.parse
+	Blocks.parse\
+	UnicodeData.parse
 
 
 all: libcmap.a libcmap.$(LIBEXT)
 $(OBJ): $(HDR)
 $(LOBJ): $(HDR)
 $(PARSERS:=.o): parse-common.c
+$(PARSERS):
 
 libcmap_script_list.o: Scripts.txt.c
 libcmap_block_list.o: Blocks.txt.c
+libcmap_codepoint_list.o: UnicodeData.txt.c
+
+libcmap_unicode_version.o: Makefile
 
 .c.o:
-	$(CC) -c -o $@ $< $(CFLAGS) $(CPPFLAGS)
+	$(CC) -c -o $@ $< $(CFLAGS) $(CPPFLAGS) -D'UNICODE_VERSION="$(UNICODE_VERSION)"'
 
 .c.lo:
-	$(CC) -fPIC -c -o $@ $< $(CFLAGS) $(CPPFLAGS)
+	$(CC) -fPIC -c -o $@ $< $(CFLAGS) $(CPPFLAGS) -D'UNICODE_VERSION="$(UNICODE_VERSION)"'
+
+.o:
+	$(CC) -o $@ $<
 
 libcmap.a: $(OBJ)
 	@rm -f -- $@
@@ -71,12 +82,18 @@ Scripts.$(UNICODE_VERSION).txt:
 Blocks.$(UNICODE_VERSION).txt:
 	$(DOWNLOAD) 'https://www.unicode.org/Public/$(UNICODE_VERSION)/ucd/Blocks.txt' > $@
 
+UnicodeData.$(UNICODE_VERSION).txt:
+	$(DOWNLOAD) 'https://www.unicode.org/Public/$(UNICODE_VERSION)/ucd/UnicodeData.txt' > $@
+
 Scripts.txt.c: Scripts.$(UNICODE_VERSION).txt Scripts.parse
 	./Scripts.parse < $(@:.txt.c=).$(UNICODE_VERSION).txt > $@
 
 Blocks.txt.c: Blocks.$(UNICODE_VERSION).txt Blocks.parse
 	./Blocks.parse < $(@:.txt.c=).$(UNICODE_VERSION).txt > $@
 
+UnicodeData.txt.c: UnicodeData.$(UNICODE_VERSION).txt UnicodeData.parse
+	./UnicodeData.parse < $(@:.txt.c=).$(UNICODE_VERSION).txt > $@
+
 install: libcmap.a libcmap.$(LIBEXT)
 	mkdir -p -- "$(DESTDIR)$(PREFIX)/lib"
 	mkdir -p -- "$(DESTDIR)$(PREFIX)/include"
diff --git a/TODO b/TODO
new file mode 100644
index 0000000..6c085d3
--- /dev/null
+++ b/TODO
@@ -0,0 +1 @@
+Add man pages
diff --git a/UnicodeData.parse.c b/UnicodeData.parse.c
new file mode 100644
index 0000000..173fb6a
--- /dev/null
+++ b/UnicodeData.parse.c
@@ -0,0 +1,139 @@
+/* See LICENSE file for copyright and license details. */
+#include "parse-common.c"
+
+
+struct segment {
+	uint32_t first, last;
+};
+
+
+static struct segment *segments = NULL;
+static size_t nsegments = 0u;
+static int ret = 0;
+
+
+static char *
+split(char **text)
+{
+	char *field = *text;
+	if (field) {
+		*text = strchr(*text, ';');
+		if (*text)
+			*(*text)++ = '\0';
+	}
+	return field;
+}
+
+
+static int
+output_character_data(char *text)
+{
+	char *field;
+
+	if (!(field = split(&text))) return 0; /* codepoint */
+
+	if (!(field = split(&text))) return 0;
+	ret |= printf(".name = \"%s\"", field);
+
+	if (!(field = split(&text))) return 0;
+	ret |= printf(", .general_category = GENERAL_CATEGORY_%s", field);
+
+	if (!(field = split(&text))) return 0;
+	ret |= printf(", .canonical_combining_classes = %su", field);
+
+	if (!(field = split(&text))) return 0;
+	ret |= printf(", .bidirectional_category = BIDIRECTIONAL_CATEGORY_%s", field);
+
+	if (!(field = split(&text))) return 0; /* TODO Character Decomposition Mapping */
+	if (!(field = split(&text))) return 0; /* TODO Decimal digit value */
+	if (!(field = split(&text))) return 0; /* TODO Digit value */
+	if (!(field = split(&text))) return 0; /* TODO Numeric value */
+
+	if (!(field = split(&text))) return 0;
+	ret |= printf(", .bidirectional_mirrored = BIDIRECTIONAL_MIRRORED_%s", field);
+
+	if (!(field = split(&text))) return 0;
+	ret |= printf(", .unicode_1_name = \"%s\"", field);
+
+	if (!(field = split(&text))) return 0; /* ISO 10646 comment field -- always empty as of Unicode 5.2.0 */
+
+	if (!(field = split(&text))) return 0; /* TODO Uppercase Mapping */
+	if (!(field = split(&text))) return 0; /* TODO Lowercase Mapping */
+	if (!(field = split(&text))) return 0; /* TODO Titlecase Mapping */
+
+	return 1;
+}
+
+
+static void
+parse_line(char *text, size_t lineno)
+{
+	unsigned long int codepoint;
+	char *end;
+
+	if (!isxdigit(*text)) {
+	malformat:
+		fprintf(stderr, "%s: line %zu in is malformatted\n", argv0, lineno);
+		exit(1);
+	}
+
+	errno = 0;
+
+	codepoint = strtoul(text, &end, 16);
+	if (errno || codepoint > ULTIMATE_CODEPOINT || *end != ';')
+		goto malformat;
+
+	if (!nsegments) {
+	new_segment:
+		ret |= printf("static const struct libcmap_codepoint_data data_from_%04lx[] = {\n", codepoint);
+		segments = realloc(segments, ++nsegments * sizeof(*segments));
+		if (!segments) {
+			fprintf(stderr, "%s: realloc %zu: %s\n", argv0, nsegments * sizeof(*segments), strerror(errno));
+			exit(1);
+		}
+		segments[nsegments - 1u].first = (uint32_t)codepoint;
+		goto new_codepoint;
+	} else if (segments[nsegments - 1u].last + 1u == (uint32_t)codepoint) {
+		ret |= printf(",\n");
+	new_codepoint:
+		segments[nsegments - 1u].last = (uint32_t)codepoint;
+		ret |= printf("\t{");
+		if (!output_character_data(text))
+			goto malformat;
+		ret |= printf("}");
+	} else {
+		if ((uint32_t)codepoint < segments[nsegments - 1u].last) {
+			fprintf(stderr, "%s: UnicodeData is not sorted\n", argv0);
+			exit(1);
+		}
+		ret |= printf("\n};\n");
+		goto new_segment;
+	}
+}
+
+
+static int
+output(void)
+{
+	size_t i;
+
+	if (!nsegments) {
+		fprintf(stderr, "%s: character list is empty\n", argv0);
+		exit(1);
+	}
+
+	ret |= printf("\n};\n");
+
+	ret |= printf("static const struct libcmap_codepoint_segment list[] = {\n");
+	for (i = 0u; i < nsegments; i++) {
+		if (i)
+			ret |= printf(",\n");
+		ret |= printf("\t{.range = {0x%04"PRIX32", 0x%04"PRIX32"}, .data = data_from_%04"PRIx32"}",
+		              segments[i].first, segments[i].last, segments[i].first);
+	}
+	ret |= printf("\n};\n");
+
+	free(segments);
+
+	return ret;
+}
diff --git a/libcmap.h b/libcmap.h
index a62ddee..9c32c77 100644
--- a/libcmap.h
+++ b/libcmap.h
@@ -36,6 +36,88 @@
 #define LIBCMAP_RANGE_SIZE(RANGE) ((size_t)((RANGE)->last - (RANGE)->first + 1u))
 
 
+enum libcmap_general_supercategory {
+	LIBCMAP_GENERAL_SUPERCATEGORY_LETTER = 0x0100u,
+	LIBCMAP_GENERAL_SUPERCATEGORY_MARK = 0x0200u,
+	LIBCMAP_GENERAL_SUPERCATEGORY_NUMBER = 0x0300u,
+	LIBCMAP_GENERAL_SUPERCATEGORY_PUNCTUATION = 0x0400u,
+	LIBCMAP_GENERAL_SUPERCATEGORY_SYMBOL = 0x0500u,
+	LIBCMAP_GENERAL_SUPERCATEGORY_SEPARATOR = 0x0600u,
+	LIBCMAP_GENERAL_SUPERCATEGORY_OTHER = 0x0700u
+};
+
+enum libcmap_general_category {
+	LIBCMAP_GENERAL_CATEGORY_LETTER_UPPERCASE = LIBCMAP_GENERAL_SUPERCATEGORY_LETTER + 1u,
+	LIBCMAP_GENERAL_CATEGORY_LETTER_LOWERCASE,
+	LIBCMAP_GENERAL_CATEGORY_LETTER_TITLECASE,
+	LIBCMAP_GENERAL_CATEGORY_LETTER_MODIFIER,
+	LIBCMAP_GENERAL_CATEGORY_LETTER_OTHER,
+	LIBCMAP_GENERAL_CATEGORY_MARK_NONSPACING = LIBCMAP_GENERAL_SUPERCATEGORY_MARK + 1u,
+	LIBCMAP_GENERAL_CATEGORY_MARK_SPACING_COMBINING,
+	LIBCMAP_GENERAL_CATEGORY_MARK_ENCODING,
+	LIBCMAP_GENERAL_CATEGORY_NUMBER_DECIMAL_DIGIT = LIBCMAP_GENERAL_SUPERCATEGORY_NUMBER + 1u,
+	LIBCMAP_GENERAL_CATEGORY_NUMBER_LETTER,
+	LIBCMAP_GENERAL_CATEGORY_NUMBER_OTHER,
+	LIBCMAP_GENERAL_CATEGORY_PUNCTUATION_CONNECTOR = LIBCMAP_GENERAL_SUPERCATEGORY_PUNCTUATION + 1u,
+	LIBCMAP_GENERAL_CATEGORY_PUNCTUATION_DASH,
+	LIBCMAP_GENERAL_CATEGORY_PUNCTUATION_OPEN,
+	LIBCMAP_GENERAL_CATEGORY_PUNCTUATION_CLOSE,
+	LIBCMAP_GENERAL_CATEGORY_PUNCTUATION_INITIAL_QUOTE,
+	LIBCMAP_GENERAL_CATEGORY_PUNCTUATION_FINAL_QUOTE,
+	LIBCMAP_GENERAL_CATEGORY_PUNCTUATION_OTHER,
+	LIBCMAP_GENERAL_CATEGORY_SYMBOL_MATH = LIBCMAP_GENERAL_SUPERCATEGORY_SYMBOL + 1u,
+	LIBCMAP_GENERAL_CATEGORY_SYMBOL_CURRENCY,
+	LIBCMAP_GENERAL_CATEGORY_SYMBOL_MODIFIER,
+	LIBCMAP_GENERAL_CATEGORY_SYMBOL_OTHER,
+	LIBCMAP_GENERAL_CATEGORY_SEPARATOR_SPACE = LIBCMAP_GENERAL_SUPERCATEGORY_SEPARATOR + 1u,
+	LIBCMAP_GENERAL_CATEGORY_SEPARATOR_LINE,
+	LIBCMAP_GENERAL_CATEGORY_SEPARATOR_PARAGRAPH,
+	LIBCMAP_GENERAL_CATEGORY_OTHER_CONTROL = LIBCMAP_GENERAL_SUPERCATEGORY_OTHER + 1u,
+	LIBCMAP_GENERAL_CATEGORY_OTHER_FORMAT,
+	LIBCMAP_GENERAL_CATEGORY_OTHER_SURROGATE,
+	LIBCMAP_GENERAL_CATEGORY_OTHER_PRIVATE_USE,
+	LIBCMAP_GENERAL_CATEGORY_OTHER_NOT_ASSIGNED
+};
+
+#define LIBCMAP_GENERAL_SUPERCATEGORY_MASK 0x7F00u
+
+
+enum libcmap_bidirectional_supercategory {
+	LIBCMAP_BIDIRECTIONAL_SUPERCATEGORY_STRONG = 0x0100u,
+	LIBCMAP_BIDIRECTIONAL_SUPERCATEGORY_WEAK = 0x0200u,
+	LIBCMAP_BIDIRECTIONAL_SUPERCATEGORY_NEUTRAL = 0x0300u,
+	LIBCMAP_BIDIRECTIONAL_SUPERCATEGORY_EXPLICIT = 0x0400u
+};
+
+enum libcmap_bidirectional_category {
+	LIBCMAP_BIDIRECTIONAL_CATEGORY_LEFT_TO_RIGHT = LIBCMAP_BIDIRECTIONAL_SUPERCATEGORY_STRONG + 1u,
+	LIBCMAP_BIDIRECTIONAL_CATEGORY_RIGHT_TO_LEFT,
+	LIBCMAP_BIDIRECTIONAL_CATEGORY_RIGHT_TO_LEFT_ARABIC,
+	LIBCMAP_BIDIRECTIONAL_CATEGORY_EUROPEAN_NUMBER = LIBCMAP_BIDIRECTIONAL_SUPERCATEGORY_WEAK + 1u,
+	LIBCMAP_BIDIRECTIONAL_CATEGORY_EUROPEAN_NUMBER_SEPARATOR,
+	LIBCMAP_BIDIRECTIONAL_CATEGORY_EUROPEAN_NUMBER_TERMINATOR,
+	LIBCMAP_BIDIRECTIONAL_CATEGORY_ARABIC_NUMBER,
+	LIBCMAP_BIDIRECTIONAL_CATEGORY_COMMON_SEPARATOR,
+	LIBCMAP_BIDIRECTIONAL_CATEGORY_NONSPACING_MARK,
+	LIBCMAP_BIDIRECTIONAL_CATEGORY_BOUNDARY_NEUTRAL,
+	LIBCMAP_BIDIRECTIONAL_CATEGORY_PARAGRAPH_SEPARATOR = LIBCMAP_BIDIRECTIONAL_SUPERCATEGORY_NEUTRAL + 1u,
+	LIBCMAP_BIDIRECTIONAL_CATEGORY_SEGMENT_SEPARATOR,
+	LIBCMAP_BIDIRECTIONAL_CATEGORY_WHITESPACE,
+	LIBCMAP_BIDIRECTIONAL_CATEGORY_OTHER_NEUTRALS,
+	LIBCMAP_BIDIRECTIONAL_CATEGORY_LEFT_TO_RIGHT_EMBEDDING = LIBCMAP_BIDIRECTIONAL_SUPERCATEGORY_EXPLICIT + 1u,
+	LIBCMAP_BIDIRECTIONAL_CATEGORY_LEFT_TO_RIGHT_OVERRIDE,
+	LIBCMAP_BIDIRECTIONAL_CATEGORY_RIGHT_TO_LEFT_EMBEDDING,
+	LIBCMAP_BIDIRECTIONAL_CATEGORY_RIGHT_TO_LEFT_OVERRIDE,
+	LIBCMAP_BIDIRECTIONAL_CATEGORY_POP_DIRECTIONAL_FORMAT,
+	LIBCMAP_BIDIRECTIONAL_CATEGORY_LEFT_TO_RIGHT_ISOLATE,
+	LIBCMAP_BIDIRECTIONAL_CATEGORY_RIGHT_TO_LEFT_ISOLATE,
+	LIBCMAP_BIDIRECTIONAL_CATEGORY_FIRST_STRONG_ISOLATE,
+	LIBCMAP_BIDIRECTIONAL_CATEGORY_POP_DIRECTIONAL_ISOLATE
+};
+
+#define LIBCMAP_BIDIRECTIONAL_SUPERCATEGORY_MASK 0x7F00u
+
+
 /**
  * Contiguous range of Unicode codepoints
  *
@@ -107,6 +189,52 @@ struct libcmap_script {
 
 
 /**
+ * Information about a specific codepoint
+ */
+struct libcmap_codepoint_data {
+	/**
+	 * The name as published in the Unicode Standard code charts
+	 *
+	 * It uniquely identifies the character unless it starts with '<'
+	 */
+	const char *name;
+
+	/* TODO documentation is needed */
+	enum libcmap_general_category general_category;
+	enum libcmap_bidirectional_category bidirectional_category;
+	uint16_t canonical_combining_classes;
+	uint16_t bidirectional_mirrored;
+	const char *unicode_1_name;
+};
+
+
+/**
+ * Contiguous block of codepoints with
+ * data for each codepoint
+ */
+struct libcmap_codepoint_segment {
+	/**
+	 * The range of codepoints the segment covers
+	 */
+	struct libcmap_range range;
+
+	/**
+	 * data for each codepoint in the segment,
+	 *
+	 * `.data[codepoint - .range.first]` gives the entry for
+	 * a specific codepoint without [`.range.first, `.range.last`]
+	 */
+	const struct libcmap_codepoint_data *data;
+};
+
+
+/**
+ * The version of the Unicode Standard that the library is
+ * compiled with data for
+ */
+extern const char *const libcmap_unicode_version;
+
+/**
  * Listing of all Unicode codepoints not assigned any block
  * 
  * This listing can potentially be empty
@@ -163,6 +291,23 @@ extern const struct libcmap_script *const libcmap_script_list;
  */
 extern const size_t libcmap_script_list_size;
 
+/**
+ * Data about codepoints
+ *
+ * There are gaps, hence the list is segmented
+ *
+ * The list is sorted in ascending order
+ *
+ * @seealso  libcmap_codepoint_list_size
+ * @seealso  libcmap_find_codepoint_data
+ */
+extern const struct libcmap_codepoint_segment *const libcmap_codepoint_list;
+
+/**
+ * The number of elements in `libcmap_codepoint_list`
+ */
+extern const size_t libcmap_codepoint_list_size;
+
 
 /**
  * Find a codepoint within `libcmap_no_block`
@@ -237,6 +382,23 @@ const struct libcmap_script *libcmap_find_script(uint32_t codepoint, size_t *off
 
 
 /**
+ * Find data about a specified codepoint
+ *
+ * @param   codepoint  The codepoint
+ * @return             Information from UnicodeData about the codepoint,
+ *                     `NULL` if not found
+ *
+ * `NULL`-return doesn't necessary mean it is unassigned,
+ * but rather just that it doesn't appear in UnicodeData,
+ * which is common for ideographs
+ * 
+ * @seealso  libcmap_codepoint_list
+ */
+LIBCMAP_PURE_
+const struct libcmap_codepoint_data *libcmap_find_codepoint_data(uint32_t codepoint);
+
+
+/**
  * Print a string representing a Unicode codepoint range
  * 
  * @param   buf     Output buffer to print to, must be sufficiently large;
diff --git a/libcmap_codepoint_list.c b/libcmap_codepoint_list.c
new file mode 100644
index 0000000..d706441
--- /dev/null
+++ b/libcmap_codepoint_list.c
@@ -0,0 +1,68 @@
+/* See LICENSE file for copyright and license details. */
+#include "common.h"
+
+
+#define GENERAL_CATEGORY_Lu LIBCMAP_GENERAL_CATEGORY_LETTER_UPPERCASE
+#define GENERAL_CATEGORY_Ll LIBCMAP_GENERAL_CATEGORY_LETTER_LOWERCASE
+#define GENERAL_CATEGORY_Lt LIBCMAP_GENERAL_CATEGORY_LETTER_TITLECASE
+#define GENERAL_CATEGORY_Lm LIBCMAP_GENERAL_CATEGORY_LETTER_MODIFIER
+#define GENERAL_CATEGORY_Lo LIBCMAP_GENERAL_CATEGORY_LETTER_OTHER
+#define GENERAL_CATEGORY_Mn LIBCMAP_GENERAL_CATEGORY_MARK_NONSPACING
+#define GENERAL_CATEGORY_Mc LIBCMAP_GENERAL_CATEGORY_MARK_SPACING_COMBINING
+#define GENERAL_CATEGORY_Me LIBCMAP_GENERAL_CATEGORY_MARK_ENCODING
+#define GENERAL_CATEGORY_Nd LIBCMAP_GENERAL_CATEGORY_NUMBER_DECIMAL_DIGIT
+#define GENERAL_CATEGORY_Nl LIBCMAP_GENERAL_CATEGORY_NUMBER_LETTER
+#define GENERAL_CATEGORY_No LIBCMAP_GENERAL_CATEGORY_NUMBER_OTHER
+#define GENERAL_CATEGORY_Pc LIBCMAP_GENERAL_CATEGORY_PUNCTUATION_CONNECTOR
+#define GENERAL_CATEGORY_Pd LIBCMAP_GENERAL_CATEGORY_PUNCTUATION_DASH
+#define GENERAL_CATEGORY_Ps LIBCMAP_GENERAL_CATEGORY_PUNCTUATION_OPEN
+#define GENERAL_CATEGORY_Pe LIBCMAP_GENERAL_CATEGORY_PUNCTUATION_CLOSE
+#define GENERAL_CATEGORY_Pi LIBCMAP_GENERAL_CATEGORY_PUNCTUATION_INITIAL_QUOTE
+#define GENERAL_CATEGORY_Pf LIBCMAP_GENERAL_CATEGORY_PUNCTUATION_FINAL_QUOTE
+#define GENERAL_CATEGORY_Po LIBCMAP_GENERAL_CATEGORY_PUNCTUATION_OTHER
+#define GENERAL_CATEGORY_Sm LIBCMAP_GENERAL_CATEGORY_SYMBOL_MATH
+#define GENERAL_CATEGORY_Sc LIBCMAP_GENERAL_CATEGORY_SYMBOL_CURRENCY
+#define GENERAL_CATEGORY_Sk LIBCMAP_GENERAL_CATEGORY_SYMBOL_MODIFIER
+#define GENERAL_CATEGORY_So LIBCMAP_GENERAL_CATEGORY_SYMBOL_OTHER
+#define GENERAL_CATEGORY_Zs LIBCMAP_GENERAL_CATEGORY_SEPARATOR_SPACE
+#define GENERAL_CATEGORY_Zl LIBCMAP_GENERAL_CATEGORY_SEPARATOR_LINE
+#define GENERAL_CATEGORY_Zp LIBCMAP_GENERAL_CATEGORY_SEPARATOR_PARAGRAPH
+#define GENERAL_CATEGORY_Cc LIBCMAP_GENERAL_CATEGORY_OTHER_CONTROL
+#define GENERAL_CATEGORY_Cf LIBCMAP_GENERAL_CATEGORY_OTHER_FORMAT
+#define GENERAL_CATEGORY_Cs LIBCMAP_GENERAL_CATEGORY_OTHER_SURROGATE
+#define GENERAL_CATEGORY_Co LIBCMAP_GENERAL_CATEGORY_OTHER_PRIVATE_USE
+/* #define GENERAL_CATEGORY_Cn LIBCMAP_GENERAL_CATEGORY_OTHER_NOT_ASSIGNED */
+
+#define BIDIRECTIONAL_CATEGORY_L   LIBCMAP_BIDIRECTIONAL_CATEGORY_LEFT_TO_RIGHT
+#define BIDIRECTIONAL_CATEGORY_R   LIBCMAP_BIDIRECTIONAL_CATEGORY_RIGHT_TO_LEFT
+#define BIDIRECTIONAL_CATEGORY_AL  LIBCMAP_BIDIRECTIONAL_CATEGORY_RIGHT_TO_LEFT_ARABIC
+#define BIDIRECTIONAL_CATEGORY_EN  LIBCMAP_BIDIRECTIONAL_CATEGORY_EUROPEAN_NUMBER
+#define BIDIRECTIONAL_CATEGORY_ES  LIBCMAP_BIDIRECTIONAL_CATEGORY_EUROPEAN_NUMBER_SEPARATOR
+#define BIDIRECTIONAL_CATEGORY_ET  LIBCMAP_BIDIRECTIONAL_CATEGORY_EUROPEAN_NUMBER_TERMINATOR
+#define BIDIRECTIONAL_CATEGORY_AN  LIBCMAP_BIDIRECTIONAL_CATEGORY_ARABIC_NUMBER
+#define BIDIRECTIONAL_CATEGORY_CS  LIBCMAP_BIDIRECTIONAL_CATEGORY_COMMON_SEPARATOR
+#define BIDIRECTIONAL_CATEGORY_NSM LIBCMAP_BIDIRECTIONAL_CATEGORY_NONSPACING_MARK
+#define BIDIRECTIONAL_CATEGORY_BN  LIBCMAP_BIDIRECTIONAL_CATEGORY_BOUNDARY_NEUTRAL
+#define BIDIRECTIONAL_CATEGORY_B   LIBCMAP_BIDIRECTIONAL_CATEGORY_PARAGRAPH_SEPARATOR
+#define BIDIRECTIONAL_CATEGORY_S   LIBCMAP_BIDIRECTIONAL_CATEGORY_SEGMENT_SEPARATOR
+#define BIDIRECTIONAL_CATEGORY_WS  LIBCMAP_BIDIRECTIONAL_CATEGORY_WHITESPACE
+#define BIDIRECTIONAL_CATEGORY_ON  LIBCMAP_BIDIRECTIONAL_CATEGORY_OTHER_NEUTRALS
+#define BIDIRECTIONAL_CATEGORY_LRE LIBCMAP_BIDIRECTIONAL_CATEGORY_LEFT_TO_RIGHT_EMBEDDING
+#define BIDIRECTIONAL_CATEGORY_LRO LIBCMAP_BIDIRECTIONAL_CATEGORY_LEFT_TO_RIGHT_OVERRIDE
+#define BIDIRECTIONAL_CATEGORY_RLE LIBCMAP_BIDIRECTIONAL_CATEGORY_RIGHT_TO_LEFT_EMBEDDING
+#define BIDIRECTIONAL_CATEGORY_RLO LIBCMAP_BIDIRECTIONAL_CATEGORY_RIGHT_TO_LEFT_OVERRIDE
+#define BIDIRECTIONAL_CATEGORY_PDF LIBCMAP_BIDIRECTIONAL_CATEGORY_POP_DIRECTIONAL_FORMAT
+#define BIDIRECTIONAL_CATEGORY_LRI LIBCMAP_BIDIRECTIONAL_CATEGORY_LEFT_TO_RIGHT_ISOLATE
+#define BIDIRECTIONAL_CATEGORY_RLI LIBCMAP_BIDIRECTIONAL_CATEGORY_RIGHT_TO_LEFT_ISOLATE
+#define BIDIRECTIONAL_CATEGORY_FSI LIBCMAP_BIDIRECTIONAL_CATEGORY_FIRST_STRONG_ISOLATE
+#define BIDIRECTIONAL_CATEGORY_PDI LIBCMAP_BIDIRECTIONAL_CATEGORY_POP_DIRECTIONAL_ISOLATE
+
+#define BIDIRECTIONAL_MIRRORED_Y 1u
+#define BIDIRECTIONAL_MIRRORED_N 0u
+
+
+
+#include "UnicodeData.txt.c"
+
+const struct libcmap_codepoint_segment *const libcmap_codepoint_list = list;
+const size_t libcmap_codepoint_list_size = sizeof(list) / sizeof(*list);
diff --git a/libcmap_find_codepoint_data.c b/libcmap_find_codepoint_data.c
new file mode 100644
index 0000000..cd2e52f
--- /dev/null
+++ b/libcmap_find_codepoint_data.c
@@ -0,0 +1,25 @@
+/* See LICENSE file for copyright and license details. */
+#include "common.h"
+
+
+const struct libcmap_codepoint_data *
+libcmap_find_codepoint_data(uint32_t codepoint)
+{
+	size_t low = 0u, high = libcmap_codepoint_list_size - 1u;
+	size_t i;
+
+	while (low <= high) {
+		i = low + (high - low) / 2u;
+		if (libcmap_codepoint_list[i].range.last < codepoint) {
+			low = i + 1u;
+		} else if (libcmap_codepoint_list[i].range.first > codepoint) {
+			if (!high)
+				break;
+			high = i - 1u;
+		} else {
+			return &libcmap_codepoint_list[i].data[codepoint - libcmap_codepoint_list[i].range.first];
+		}
+	}
+
+	return NULL;	
+}
diff --git a/libcmap_unicode_version.c b/libcmap_unicode_version.c
new file mode 100644
index 0000000..5184701
--- /dev/null
+++ b/libcmap_unicode_version.c
@@ -0,0 +1,5 @@
+/* See LICENSE file for copyright and license details. */
+#include "common.h"
+
+
+const char *const libcmap_unicode_version = UNICODE_VERSION;
diff --git a/parse-common.c b/parse-common.c
index 899debb..aeac342 100644
--- a/parse-common.c
+++ b/parse-common.c
@@ -1,6 +1,8 @@
 /* See LICENSE file for copyright and license details. */
 #include <ctype.h>
 #include <errno.h>
+#include <inttypes.h>
+#include <stdint.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>