First commit1.0

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

1 files changed, 470 insertions, 0 deletions

diff --git a/libparsepsf.c b/libparsepsf.c
new file mode 100644
index 0000000..7c7b655
--- /dev/null
+++ b/libparsepsf.c
@@ -0,0 +1,470 @@
+/* See LICENSE file for copyright and license details. */
+#include "libparsepsf.h"
+
+#include <errno.h>
+#include <stdio.h>
+#include <stdint.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include <grapheme.h>
+
+
+struct psf1_header {
+	uint8_t magic_x36;
+	uint8_t magic_x04;
+	uint8_t mode;
+#define PSF1_MODE512    0x01
+#define PSF1_MODEHASTAB 0x02
+/* #define PSF1_MODEHASSEQ 0x04 // really used */
+	uint8_t height;
+};
+#define PSF1_SEPARATOR  0xFFFF
+#define PSF1_STARTSEQ   0xFFFE
+
+struct psf2_header {
+	uint8_t magic_x72;
+	uint8_t magic_xb5;
+	uint8_t magic_x4a;
+	uint8_t magic_x86;
+	uint32_t version;
+#define PSF2_MAXVERSION 0
+	uint32_t header_size;
+	uint32_t flags;
+#define PSF2_HAS_UNICODE_TABLE 0x01
+	uint32_t num_glyphs;
+	uint32_t charsize; /* = height * ((width + 7) / 8) */
+	uint32_t height;
+	uint32_t width;
+};
+#define PSF2_SEPARATOR  0xFF
+#define PSF2_STARTSEQ   0xFE
+
+
+static void
+free_map(struct libparsepsf_unimap *node)
+{
+	size_t i;
+	for (i = 0; i < sizeof(node->nonterminal) / sizeof(*node->nonterminal); i++)
+		if (node->nonterminal[i])
+			free_map(node->nonterminal[i]);
+	free(node);
+}
+
+void
+libparsepsf_destroy_font(struct libparsepsf_font *font)
+{
+	free(font->glyph_data);
+	font->glyph_data = NULL;
+	if (font->map) {
+		free_map(font->map);
+		font->map = NULL;
+	}
+}
+
+
+static uint16_t
+letoh16(const uint8_t *le)
+{
+	uint16_t b0 = (uint16_t)((uint16_t)le[0] << 0);
+	uint16_t b1 = (uint16_t)((uint16_t)le[1] << 8);
+	return (uint16_t)(b0 | b1);
+}
+
+
+static uint32_t
+letoh32(uint32_t le)
+{
+	union {
+		uint32_t v;
+		uint8_t b[4];
+	} u = {.v = le};
+	uint32_t b0 = (uint32_t)((uint32_t)u.b[0] << 0);
+	uint32_t b1 = (uint32_t)((uint32_t)u.b[1] << 8);
+	uint32_t b2 = (uint32_t)((uint32_t)u.b[2] << 16);
+	uint32_t b3 = (uint32_t)((uint32_t)u.b[3] << 24);
+	return (uint32_t)(b0 | b1 | b2 | b3);
+}
+
+
+static uint32_t
+desurrogate(uint16_t high, uint16_t low)
+{
+	/* high surrogate has lower value */
+	uint32_t h = UINT32_C(0xD800) ^ (uint32_t)high;
+	uint32_t l = UINT32_C(0xDC00) ^ (uint32_t)low;
+	h <<= 10;
+	return (uint32_t)(h | l);
+}
+
+static int
+put_map_incomplete(struct libparsepsf_font *font, const uint8_t *seq, size_t seqlen,
+                   uint8_t *savedp, struct libparsepsf_unimap **nodep)
+{
+	size_t i;
+	if (font->map == NULL) {
+		font->map = calloc(1, sizeof(*font->map));
+		if (!font->map)
+			goto enomem;
+	}
+	if (!seqlen)
+		goto ebfont;
+	*nodep = font->map;
+	*savedp = seq[--seqlen];
+	for (i = 0; i < seqlen; i++) {
+		if (!(*nodep)->nonterminal[seq[i]]) {
+			(*nodep)->nonterminal[seq[i]] = calloc(1, sizeof(*font->map));
+			if (!(*nodep)->nonterminal[seq[i]])
+				goto enomem;
+		}
+		*nodep = (*nodep)->nonterminal[seq[i]];
+	}
+
+	return 0;
+
+ebfont:
+	errno = EBFONT;
+	return -1;
+enomem:
+	errno = ENOMEM;
+	return -1;
+}
+
+static int
+put_map_finalise(size_t index, uint8_t saved, struct libparsepsf_unimap *node)
+{
+	/* unfortunately this actually happens in the real world */
+#if 0
+	if (node->terminal[saved]) {
+		return 0;
+# if 0
+		errno = EBFONT;
+		return -1;
+# endif
+	}
+#endif
+
+	node->terminal[saved] = index + 1;
+	return 0;
+}
+
+static int
+put_map(struct libparsepsf_font *font, size_t index, const uint8_t *seq, size_t seqlen)
+{
+	uint8_t saved = 0xFF;
+	struct libparsepsf_unimap *node = NULL;
+	if (put_map_incomplete(font, seq, seqlen, &saved, &node))
+		return -1;
+	return put_map_finalise(index, saved, node);
+}
+
+static int
+decode_utf8(const uint8_t *data, size_t size, size_t *np, uint32_t *cpp)
+{
+	uint8_t head;
+	uint32_t cp;
+
+	head = *data;
+	*np = 1;
+	if (!(head & 0x80)) {
+		if (cpp)
+			*cpp = (uint32_t)head;
+		return 0;
+	} else if (!(head & 0x40)) {
+		return -1;
+	}
+	size--;
+	cp = (uint32_t)head;
+	head <<= 1;
+
+	while (head & 0x80) {
+		head <<= 1;
+		if ((data[*np] & 0xC0) != 0x80)
+			return -1;
+		cp <<= 6;
+		cp |= (uint32_t)(data[*np] ^ 0x80);
+		*np += 1;
+	}
+	if (*np > 4)
+		return -1;
+
+	cp &= (UINT32_C(1) << (*np * 5 + 1)) - 1;
+	if ((cp & UINT32_C(0xFFF800)) == UINT32_C(0xD800) ||
+	    cp > UINT32_C(0x10FFFF) ||
+	    cp < UINT32_C(1) << (*np == 2 ? 7 : *np * 5 - 4))
+		return -1;
+
+	if (cpp)
+		*cpp = cp;
+	return 0;
+}
+
+int
+libparsepsf_parse_font(const void *data, size_t size, struct libparsepsf_font *fontp, uint32_t *unrecognised_versionp)
+{
+	union {
+		struct psf1_header psf1;
+		struct psf2_header psf2;
+	} header;
+	const uint8_t *udata = data;
+	size_t glyphs_offset;
+	size_t charsize;
+	size_t off;
+	size_t i;
+	size_t n;
+	uint32_t u32;
+	uint16_t u16, u16b;
+	uint8_t u8, utf8[4], utf8_saved;
+	struct libparsepsf_unimap *utf8_node;
+
+	*unrecognised_versionp = 0;
+	fontp->glyph_data = NULL;
+	fontp->map = NULL;
+
+	if (size < 4)
+		goto ebfont;
+
+	if (udata[0] == 0x36) { /* TODO untested */
+		if (size < sizeof(header.psf1))
+			goto ebfont;
+		memcpy(&header.psf1, udata, sizeof(header.psf1));
+		if (header.psf1.magic_x36 != 0x36 ||
+		    header.psf1.magic_x04 != 0x04)
+			goto ebfont;
+		fontp->num_glyphs = (header.psf1.mode & PSF1_MODE512) ? 512 : 256;
+		fontp->height     = (size_t)header.psf1.height;
+		fontp->width      = 8;
+		glyphs_offset = sizeof(header.psf1);
+		charsize = fontp->height;
+		if (glyphs_offset > size ||
+		    !fontp->num_glyphs ||
+		    charsize > (size - glyphs_offset) / fontp->num_glyphs)
+			goto ebfont;
+		if (header.psf1.mode & PSF1_MODEHASTAB) {
+			off = glyphs_offset + fontp->num_glyphs * charsize;
+			for (i = 0; i < fontp->num_glyphs; i++) {
+				for (;;) {
+					if (off + 2 > size)
+						goto ebfont;
+					u16 = letoh16(&udata[off]);
+					off += 2;
+					if (u16 == PSF1_STARTSEQ) {
+						break;
+					} else if (u16 == PSF1_SEPARATOR) {
+						goto next_char_psf1;
+					} else if ((u16 & UINT32_C(0xF800)) == UINT32_C(0xD800)) {
+						if (off + 2 > size)
+							goto ebfont;
+						u16b = letoh16(&udata[off]);
+						off += 2;
+						if (((u16 ^ u16b) & 0xDC00) != 0x0400)
+							goto ebfont;
+						u32 = desurrogate(u16 < u16b ? u16 : u16b,
+						                  u16 < u16b ? u16b : u16);
+					} else {
+						u32 = (uint32_t)u16;
+					}
+					n = grapheme_cp_encode(u32, utf8, sizeof(utf8));
+					if (n > sizeof(utf8))
+						abort();
+					if (put_map(fontp, i, utf8, n))
+						goto fail;
+				}
+				utf8_saved = 0xFF;
+				utf8_node = NULL;
+				for (;;) {
+					if (off + 2 > size)
+						goto ebfont;
+					u16 = letoh16(&udata[off]);
+					off += 2;
+					if (u16 == PSF1_STARTSEQ || u16 == PSF1_SEPARATOR) {
+						if (put_map_finalise(i, utf8_saved, utf8_node))
+							goto fail;
+						if (u16 == PSF1_SEPARATOR)
+							goto next_char_psf1;
+						utf8_saved = 0xFF;
+						utf8_node = NULL;
+						continue;
+					} else if ((u16 & UINT32_C(0xF800)) == UINT32_C(0xD800)) {
+						if (off + 2 > size)
+							goto ebfont;
+						u16b = letoh16(&udata[off]);
+						off += 2;
+						if (((u16 ^ u16b) & 0xDC00) != 0x0400)
+							goto ebfont;
+						u32 = desurrogate(u16 > u16b ? u16 : u16b,
+						                  u16 > u16b ? u16b : u16);
+					} else {
+						u32 = (uint32_t)u16;
+					}
+					n = grapheme_cp_encode(u32, utf8, sizeof(utf8));
+					if (n > sizeof(utf8))
+						abort();
+					if (put_map_incomplete(fontp, utf8, n, &utf8_saved, &utf8_node))
+						goto fail;
+				}
+			next_char_psf1:;
+			}
+		}
+
+	} else {
+		if (size < sizeof(header.psf2))
+			goto ebfont;
+		memcpy(&header.psf2, udata, sizeof(header.psf2));
+		if (header.psf2.magic_x72 != 0x72 ||
+		    header.psf2.magic_xb5 != 0xb5 ||
+		    header.psf2.magic_x4a != 0x4a ||
+		    header.psf2.magic_x86 != 0x86)
+			goto ebfont;
+		header.psf2.version     = letoh32(header.psf2.version);
+		header.psf2.header_size = letoh32(header.psf2.header_size);
+		header.psf2.flags       = letoh32(header.psf2.flags);
+		header.psf2.num_glyphs  = letoh32(header.psf2.num_glyphs);
+		header.psf2.charsize    = letoh32(header.psf2.charsize);
+		header.psf2.height      = letoh32(header.psf2.height);
+		header.psf2.width       = letoh32(header.psf2.width);
+		if (header.psf2.height * ((header.psf2.width + 7) / 8) != header.psf2.charsize)
+			goto ebfont;
+		if (header.psf2.version > PSF2_MAXVERSION)
+			*unrecognised_versionp = 0;
+		fontp->num_glyphs = (size_t)header.psf2.num_glyphs;
+		fontp->height     = (size_t)header.psf2.height;
+		fontp->width      = (size_t)header.psf2.width;
+		glyphs_offset     = (size_t)header.psf2.header_size;
+		charsize = (size_t)header.psf2.charsize;
+		if (glyphs_offset > size ||
+		    !fontp->num_glyphs ||
+		    charsize > (size - glyphs_offset) / fontp->num_glyphs)
+			goto ebfont;
+		if (header.psf2.flags & PSF2_HAS_UNICODE_TABLE) {
+			off = glyphs_offset + fontp->num_glyphs * charsize;
+			for (i = 0; i < fontp->num_glyphs; i++) {
+				for (;;) {
+					if (off == size)
+						goto ebfont;
+					u8 = udata[off];
+					if (u8 == PSF2_STARTSEQ) {
+						off += 1;
+						break;
+					} else if (u8 == PSF2_SEPARATOR) {
+						off += 1;
+						goto next_char_psf2;
+					}
+					if (decode_utf8(&udata[off], size - off, &n, NULL))
+						goto ebfont;
+					if (put_map(fontp, i, &udata[off], n))
+						goto fail;
+					off += n;
+				}
+				utf8_saved = 0xFF;
+				utf8_node = NULL;
+				for (;;) {
+					if (off == size)
+						goto ebfont;
+					u8 = udata[off];
+					if (u8 == 0xFE || u8 == 0xFF) {
+						if (put_map_finalise(i, utf8_saved, utf8_node))
+							goto fail;
+						off += 1;
+						if (u8 == 0xFF)
+							goto next_char_psf2;
+						utf8_saved = 0xFF;
+						utf8_node = NULL;
+					} else {
+						if (decode_utf8(&udata[off], size - off, &n, NULL))
+							goto ebfont;
+						if (put_map_incomplete(fontp, &udata[off], n, &utf8_saved, &utf8_node))
+							goto fail;
+						off += n;
+					}
+				}
+			next_char_psf2:;
+			}
+		}
+	}
+
+	if (charsize) {
+		fontp->glyph_data = malloc(fontp->num_glyphs * charsize);
+		if (!fontp->glyph_data)
+			goto enomem;
+	}
+	memcpy(fontp->glyph_data, &udata[glyphs_offset], fontp->num_glyphs * charsize);
+
+	return 0;
+
+enomem:
+	errno = ENOMEM;
+	goto fail;
+ebfont:
+	errno = EBFONT;
+fail:
+	libparsepsf_destroy_font(fontp);
+	return -1;
+}
+
+
+size_t
+libparsepsf_get_glyph(const struct libparsepsf_font *font, const char *c, size_t *remp, const char **next_cp)
+{
+	size_t glyph = 0, rem = 0, n;
+	uint32_t cp;
+	struct libparsepsf_unimap *node = font->map;
+
+	if (!node) {
+		if (!remp && !*c)
+			return 0;
+		if (decode_utf8((const uint8_t *)c, remp ? *remp : SIZE_MAX, &n, &cp)) {
+			errno = EILSEQ;
+			return 0;
+		}
+		if (remp)
+			rem = *remp - n;
+		c = &c[n];
+		glyph = (size_t)cp;
+		if (glyph >= font->num_glyphs)
+			return 0;
+		glyph -= 1;
+		goto out;
+
+	} else if (remp) {
+		rem = *remp;
+		if (!rem)
+			return 0;
+		for (; rem > 1; c = &c[1], rem -= 1) {
+			if (node->terminal[*(const uint8_t *)c]) {
+				glyph = node->terminal[*(const uint8_t *)c];
+				if (next_cp)
+					*next_cp = &c[1];
+				if (remp)
+					*remp = rem - 1;
+			}
+			node = node->nonterminal[*(const uint8_t *)c];
+			if (!node)
+				return glyph;
+		}
+
+	} else {
+		if (!c[0])
+			return 0;
+		for (; c[1]; c = &c[1]) {
+			if (node->terminal[*(const uint8_t *)c]) {
+				glyph = node->terminal[*(const uint8_t *)c];
+				if (next_cp)
+					*next_cp = &c[1];
+			}
+			node = node->nonterminal[*(const uint8_t *)c];
+			if (!node)
+				return glyph;
+		}
+	}
+
+	glyph = node->terminal[*(const uint8_t *)c];
+out:
+	if (glyph) {
+		if (next_cp)
+			*next_cp = &c[1];
+		if (remp)
+			*remp = rem - 1;
+	}
+	return glyph;
+}