Book a Demo!
CoCalc Logo Icon
StoreFeaturesDocsShareSupportNewsAboutPoliciesSign UpSign In
Download
7643 views
Tweet
Share
Share
1
#include "mupdf/pdf.h"

2


3
/* Load or synthesize ToUnicode map for fonts */

4


5
static void

6
pdf_remap_cmap_range(fz_context *ctx, pdf_cmap *ucs_from_gid,

7
	unsigned int cpt, unsigned int gid, unsigned int n, pdf_cmap *ucs_from_cpt)

8
{

9
	unsigned int k;

10
	int ucsbuf[8];

11
	int ucslen;

12


13
	for (k = 0; k <= n; ++k)

14
	{

15
		ucslen = pdf_lookup_cmap_full(ucs_from_cpt, cpt + k, ucsbuf);

16
		if (ucslen == 1)

17
			pdf_map_range_to_range(ctx, ucs_from_gid, gid + k, gid + k, ucsbuf[0]);

18
		else if (ucslen > 1)

19
			pdf_map_one_to_many(ctx, ucs_from_gid, gid + k, ucsbuf, ucslen);

20
	}

21
}

22


23
static pdf_cmap *

24
pdf_remap_cmap(fz_context *ctx, pdf_cmap *gid_from_cpt, pdf_cmap *ucs_from_cpt)

25
{

26
	pdf_cmap *ucs_from_gid;

27
	unsigned int a, b, x;

28
	int i;

29


30
	ucs_from_gid = pdf_new_cmap(ctx);

31


32
	if (gid_from_cpt->usecmap)

33
		ucs_from_gid->usecmap = pdf_remap_cmap(ctx, gid_from_cpt->usecmap, ucs_from_cpt);

34


35
	for (i = 0; i < gid_from_cpt->rlen; ++i)

36
	{

37
		a = gid_from_cpt->ranges[i].low;

38
		b = gid_from_cpt->ranges[i].high;

39
		x = gid_from_cpt->ranges[i].out;

40
		pdf_remap_cmap_range(ctx, ucs_from_gid, a, x, b - a, ucs_from_cpt);

41
	}

42


43
	for (i = 0; i < gid_from_cpt->xlen; ++i)

44
	{

45
		a = gid_from_cpt->xranges[i].low;

46
		b = gid_from_cpt->xranges[i].high;

47
		x = gid_from_cpt->xranges[i].out;

48
		pdf_remap_cmap_range(ctx, ucs_from_gid, a, x, b - a, ucs_from_cpt);

49
	}

50


51
	/* Font encoding CMaps don't have one-to-many mappings, so we can ignore the mranges. */

52


53
	pdf_sort_cmap(ctx, ucs_from_gid);

54


55
	return ucs_from_gid;

56
}

57


58
void

59
pdf_load_to_unicode(fz_context *ctx, pdf_document *doc, pdf_font_desc *font,

60
	char **strings, char *collection, pdf_obj *cmapstm)

61
{

62
	unsigned int cpt;

63


64
	if (pdf_is_stream(ctx, doc, pdf_to_num(ctx, cmapstm), pdf_to_gen(ctx, cmapstm)))

65
	{

66
		pdf_cmap *ucs_from_cpt = pdf_load_embedded_cmap(ctx, doc, cmapstm);

67
		font->to_unicode = pdf_remap_cmap(ctx, font->encoding, ucs_from_cpt);

68
		pdf_drop_cmap(ctx, ucs_from_cpt);

69
		font->size += pdf_cmap_size(ctx, font->to_unicode);

70
	}

71


72
	else if (collection)

73
	{

74
		if (!strcmp(collection, "Adobe-CNS1"))

75
			font->to_unicode = pdf_load_system_cmap(ctx, "Adobe-CNS1-UCS2");

76
		else if (!strcmp(collection, "Adobe-GB1"))

77
			font->to_unicode = pdf_load_system_cmap(ctx, "Adobe-GB1-UCS2");

78
		else if (!strcmp(collection, "Adobe-Japan1"))

79
			font->to_unicode = pdf_load_system_cmap(ctx, "Adobe-Japan1-UCS2");

80
		else if (!strcmp(collection, "Adobe-Korea1"))

81
			font->to_unicode = pdf_load_system_cmap(ctx, "Adobe-Korea1-UCS2");

82


83
		return;

84
	}

85


86
	if (strings)

87
	{

88
		/* TODO one-to-many mappings */

89


90
		font->cid_to_ucs_len = 256;

91
		font->cid_to_ucs = fz_malloc_array(ctx, 256, sizeof *font->cid_to_ucs);

92
		font->size += 256 * sizeof *font->cid_to_ucs;

93


94
		for (cpt = 0; cpt < 256; cpt++)

95
		{

96
			if (strings[cpt])

97
				font->cid_to_ucs[cpt] = pdf_lookup_agl(strings[cpt]);

98
			else

99
				font->cid_to_ucs[cpt] = '?';

100
		}

101
	}

102


103
	if (!font->to_unicode && !font->cid_to_ucs)

104
	{

105
		/* TODO: synthesize a ToUnicode if it's a freetype font with

106
		 * cmap and/or post tables or if it has glyph names. */

107
	}

108
}

109


110