summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
Diffstat
-rw-r--r--include/fitz-world.h1
-rw-r--r--include/fitz/cmap.h26
-rw-r--r--include/fitz/tree.h4
-rw-r--r--include/mupdf/rsrc.h45
-rw-r--r--mupdf/pdf_build.c10
-rw-r--r--mupdf/pdf_cmap.c739
-rw-r--r--mupdf/pdf_font.c20
-rw-r--r--mupdf/pdf_interpret.c18
-rw-r--r--mupdf/pdf_store.c4
-rw-r--r--mupdf/pdf_type3.c4
-rw-r--r--mupdf/pdf_unicode.c25
-rw-r--r--world/Jamfile1
-rw-r--r--world/node_misc2.c15
-rw-r--r--world/node_toxml.c2
-rw-r--r--world/res_cmap.c467
15 files changed, 755 insertions, 626 deletions
diff --git a/include/fitz-world.h b/include/fitz-world.h
index 86c9e5f1..0f1855f5 100644
--- a/include/fitz-world.h
+++ b/include/fitz-world.h
@@ -11,7 +11,6 @@
#error "fitz-base.h must be included before fitz-world.h"
#endif
-#include "fitz/cmap.h"
#include "fitz/font.h"
#include "fitz/pixmap.h"
#include "fitz/colorspace.h"
diff --git a/include/fitz/cmap.h b/include/fitz/cmap.h
deleted file mode 100644
index c5a4ad6f..00000000
--- a/include/fitz/cmap.h
+++ /dev/null
@@ -1,26 +0,0 @@
-typedef struct fz_cmap_s fz_cmap;
-
-fz_error *fz_newcmap(fz_cmap **cmapp);
-void fz_debugcmap(fz_cmap *cmap);
-fz_cmap *fz_keepcmap(fz_cmap *cmap);
-void fz_dropcmap(fz_cmap *cmap);
-
-char *fz_getcmapname(fz_cmap *cmap);
-void fz_setcmapname(fz_cmap *cmap, char *name);
-char *fz_getusecmapname(fz_cmap *cmap);
-void fz_setusecmapname(fz_cmap *cmap, char *usecmap);
-void fz_setusecmap(fz_cmap *cmap, fz_cmap *usecmap);
-fz_cmap *fz_getusecmap(fz_cmap *cmap);
-void fz_setwmode(fz_cmap *cmap, int wmode);
-int fz_getwmode(fz_cmap *cmap);
-
-fz_error *fz_addcodespacerange(fz_cmap *cmap, unsigned lo, unsigned hi, int n);
-
-fz_error *fz_setcidlookup(fz_cmap *cmap, int map[256]);
-
-fz_error *fz_addcidrange(fz_cmap *cmap, int srclo, int srchi, int dstlo);
-fz_error *fz_endcidrange(fz_cmap *cmap);
-
-int fz_lookupcid(fz_cmap *cmap, int cpt);
-unsigned char *fz_decodecpt(fz_cmap *cmap, unsigned char *s, int *cpt);
-
diff --git a/include/fitz/tree.h b/include/fitz/tree.h
index 22665736..69143b1f 100644
--- a/include/fitz/tree.h
+++ b/include/fitz/tree.h
@@ -134,7 +134,7 @@ struct fz_linknode_s
struct fz_metanode_s
{
fz_node super;
- void *name;
+ char *name;
void *dict;
};
@@ -156,7 +156,7 @@ fz_rect fz_boundnode(fz_node *node, fz_matrix ctm);
void fz_dropnode(fz_node *node);
/* branch nodes */
-fz_error *fz_newmetanode(fz_node **nodep, void *name, void *dict);
+fz_error *fz_newmetanode(fz_node **nodep, char *name, void *dict);
fz_error *fz_newovernode(fz_node **nodep);
fz_error *fz_newmasknode(fz_node **nodep);
fz_error *fz_newblendnode(fz_node **nodep, fz_colorspace *cs, fz_blendkind b, int k, int i);
diff --git a/include/mupdf/rsrc.h b/include/mupdf/rsrc.h
index 71475452..9d1e4595 100644
--- a/include/mupdf/rsrc.h
+++ b/include/mupdf/rsrc.h
@@ -140,7 +140,38 @@ fz_error *pdf_loadimage(pdf_image **imgp, pdf_xref *xref, fz_obj *obj, fz_obj *r
fz_error *pdf_loadtile(fz_image *image, fz_pixmap *tile);
/*
- * CMap and Font
+ * CMap
+ */
+
+typedef struct pdf_cmap_s pdf_cmap;
+
+fz_error *pdf_newcmap(pdf_cmap **cmapp);
+pdf_cmap *pdf_keepcmap(pdf_cmap *cmap);
+void pdf_dropcmap(pdf_cmap *cmap);
+
+void pdf_debugcmap(pdf_cmap *cmap);
+int pdf_getwmode(pdf_cmap *cmap);
+pdf_cmap *fz_getusecmap(pdf_cmap *cmap);
+void fz_setwmode(pdf_cmap *cmap, int wmode);
+void fz_setusecmap(pdf_cmap *cmap, pdf_cmap *usecmap);
+
+fz_error *pdf_addcodespace(pdf_cmap *cmap, unsigned lo, unsigned hi, int n);
+
+fz_error *pdf_maprangetotable(pdf_cmap *cmap, int low, int *map, int len);
+fz_error *pdf_maprangetorange(pdf_cmap *cmap, int srclo, int srchi, int dstlo);
+fz_error *pdf_maponetomany(pdf_cmap *cmap, int one, int *many, int len);
+fz_error *pdf_sortcmap(pdf_cmap *cmap);
+
+int pdf_lookupcmap(pdf_cmap *cmap, int cpt);
+unsigned char *pdf_decodecmap(pdf_cmap *cmap, unsigned char *s, int *cpt);
+
+fz_error *pdf_parsecmap(pdf_cmap **cmapp, fz_stream *file);
+fz_error *pdf_loadembeddedcmap(pdf_cmap **cmapp, pdf_xref *xref, fz_obj *ref);
+fz_error *pdf_loadsystemcmap(pdf_cmap **cmapp, char *name);
+fz_error *pdf_newidentitycmap(pdf_cmap **cmapp, int wmode, int bytes);
+
+/*
+ * Font
*/
void pdf_loadencoding(char **estrings, char *encoding);
@@ -171,13 +202,13 @@ struct pdf_font_s
float missingwidth;
/* Encoding (CMap) */
- fz_cmap *encoding;
- fz_cmap *tottfcmap;
+ pdf_cmap *encoding;
+ pdf_cmap *tottfcmap;
int ncidtogid;
unsigned short *cidtogid;
/* ToUnicode */
- fz_cmap *tounicode;
+ pdf_cmap *tounicode;
int ncidtoucs;
unsigned short *cidtoucs;
@@ -192,12 +223,6 @@ struct pdf_font_s
fz_tree *charprocs[256];
};
-/* cmap.c */
-fz_error *pdf_parsecmap(fz_cmap **cmapp, fz_stream *file);
-fz_error *pdf_loadembeddedcmap(fz_cmap **cmapp, pdf_xref *xref, fz_obj *ref);
-fz_error *pdf_loadsystemcmap(fz_cmap **cmapp, char *name);
-fz_error *pdf_makeidentitycmap(fz_cmap **cmapp, int wmode, int bytes);
-
/* unicode.c */
fz_error *pdf_loadtounicode(pdf_font *font, pdf_xref *xref, char **strings, char *collection, fz_obj *cmapstm);
diff --git a/mupdf/pdf_build.c b/mupdf/pdf_build.c
index 736b2de1..40f71d0f 100644
--- a/mupdf/pdf_build.c
+++ b/mupdf/pdf_build.c
@@ -281,7 +281,6 @@ addpatternshape(pdf_gstate *gs, fz_node *shape,
fz_matrix inv;
fz_matrix ptm;
fz_rect bbox;
- fz_obj *name;
fz_obj *dict;
int x, y, x0, y0, x1, y1;
@@ -296,9 +295,6 @@ addpatternshape(pdf_gstate *gs, fz_node *shape,
error = fz_newtransformnode(&xform, ptm);
if (error) return error;
- error = fz_newname(&name, "Pattern");
- if (error) return error;
-
error = fz_packobj(&dict, "<< /Tree %p /XStep %f /YStep %f "
" /Matrix[%f %f %f %f %f %f] >>",
pat->tree, pat->xstep, pat->ystep,
@@ -307,7 +303,7 @@ addpatternshape(pdf_gstate *gs, fz_node *shape,
pat->matrix.e, pat->matrix.f);
if (error) return error;
- error = fz_newmetanode(&meta, name, dict);
+ error = fz_newmetanode(&meta, "Pattern", dict);
if (error) return error;
error = fz_newovernode(&over);
@@ -776,8 +772,8 @@ pdf_showtext(pdf_csi *csi, fz_obj *text)
while (buf < end)
{
- buf = fz_decodecpt(font->encoding, buf, &cpt);
- cid = fz_lookupcid(font->encoding, cpt);
+ buf = pdf_decodecmap(font->encoding, buf, &cpt);
+ cid = pdf_lookupcmap(font->encoding, cpt);
if (cid == -1)
cid = 0;
diff --git a/mupdf/pdf_cmap.c b/mupdf/pdf_cmap.c
index 45a0eed7..4729322f 100644
--- a/mupdf/pdf_cmap.c
+++ b/mupdf/pdf_cmap.c
@@ -1,5 +1,545 @@
-#include <fitz.h>
-#include <mupdf.h>
+/*
+ * The CMap data structure here is constructed on the fly by
+ * adding simple range-to-range mappings. Then the data structure
+ * is optimized to contain both range-to-range and range-to-table
+ * lookups.
+ *
+ * Any one-to-many mappings are inserted as one-to-table
+ * lookups in the beginning, and are not affected by the optimization
+ * stage.
+ *
+ * There is a special function to add a 256-length range-to-table mapping.
+ * The ranges do not have to be added in order.
+ *
+ * This code can be a lot simpler if we don't care about wasting memory,
+ * or can trust the parser to give us optimal mappings.
+ */
+
+#include "fitz.h"
+#include "mupdf.h"
+
+typedef struct pdf_range_s pdf_range;
+
+enum { MAXCODESPACE = 10 };
+enum { SINGLE, RANGE, TABLE, MULTI };
+
+struct pdf_range_s
+{
+ int low;
+ int high;
+ int flag; /* what kind of lookup is this */
+ int offset; /* either range-delta or table-index */
+};
+
+static int
+cmprange(const void *va, const void *vb)
+{
+ return ((const pdf_range*)va)->low - ((const pdf_range*)vb)->low;
+}
+
+struct pdf_cmap_s
+{
+ int refs;
+ char cmapname[32];
+
+ char usecmapname[32];
+ pdf_cmap *usecmap;
+
+ int wmode;
+
+ int ncspace;
+ struct {
+ int n;
+ unsigned char lo[4];
+ unsigned char hi[4];
+ } cspace[MAXCODESPACE];
+
+ int rlen, rcap;
+ pdf_range *ranges;
+
+ int tlen, tcap;
+ int *table;
+};
+
+/*
+ * Allocate, destroy and simple parameters.
+ */
+
+fz_error *
+pdf_newcmap(pdf_cmap **cmapp)
+{
+ pdf_cmap *cmap;
+
+ cmap = *cmapp = fz_malloc(sizeof(pdf_cmap));
+ if (!cmap)
+ return fz_outofmem;
+
+ cmap->refs = 1;
+ strcpy(cmap->cmapname, "");
+
+ strcpy(cmap->usecmapname, "");
+ cmap->usecmap = nil;
+
+ cmap->wmode = 0;
+
+ cmap->ncspace = 0;
+
+ cmap->rlen = 0;
+ cmap->rcap = 0;
+ cmap->ranges = nil;
+
+ cmap->tlen = 0;
+ cmap->tcap = 0;
+ cmap->table = nil;
+
+ return nil;
+}
+
+pdf_cmap *
+pdf_keepcmap(pdf_cmap *cmap)
+{
+ cmap->refs ++;
+ return cmap;
+}
+
+void
+pdf_dropcmap(pdf_cmap *cmap)
+{
+ if (--cmap->refs == 0)
+ {
+ if (cmap->usecmap)
+ pdf_dropcmap(cmap->usecmap);
+ fz_free(cmap->ranges);
+ fz_free(cmap->table);
+ fz_free(cmap);
+ }
+}
+
+pdf_cmap *
+pdf_getusecmap(pdf_cmap *cmap)
+{
+ return cmap->usecmap;
+}
+
+void
+pdf_setusecmap(pdf_cmap *cmap, pdf_cmap *usecmap)
+{
+ int i;
+
+ if (cmap->usecmap)
+ pdf_dropcmap(cmap->usecmap);
+ cmap->usecmap = pdf_keepcmap(usecmap);
+
+ if (cmap->ncspace == 0)
+ {
+ cmap->ncspace = usecmap->ncspace;
+ for (i = 0; i < usecmap->ncspace; i++)
+ cmap->cspace[i] = usecmap->cspace[i];
+ }
+}
+
+int
+pdf_getwmode(pdf_cmap *cmap)
+{
+ return cmap->wmode;
+}
+
+void
+pdf_setwmode(pdf_cmap *cmap, int wmode)
+{
+ cmap->wmode = wmode;
+}
+
+void
+pdf_debugcmap(pdf_cmap *cmap)
+{
+ int i, k, n;
+
+ printf("cmap $%p /%s {\n", cmap, cmap->cmapname);
+
+ if (cmap->usecmapname[0])
+ printf(" usecmap /%s\n", cmap->usecmapname);
+ if (cmap->usecmap)
+ printf(" usecmap $%p\n", cmap->usecmap);
+
+ printf(" wmode %d\n", cmap->wmode);
+
+ printf(" codespaces {\n");
+ for (i = 0; i < cmap->ncspace; i++)
+ {
+ printf(" <");
+ for (k = 0; k < cmap->cspace[i].n; k++)
+ printf("%02x", cmap->cspace[i].lo[k]);
+ printf("> <");
+ for (k = 0; k < cmap->cspace[i].n; k++)
+ printf("%02x", cmap->cspace[i].hi[k]);
+ printf(">\n");
+ }
+ printf(" }\n");
+
+ printf(" ranges (%d,%d) {\n", cmap->rlen, cmap->tlen);
+ for (i = 0; i < cmap->rlen; i++)
+ {
+ pdf_range *r = &cmap->ranges[i];
+ printf(" <%04x> <%04x> ", r->low, r->high);
+ if (r->flag == TABLE)
+ {
+ printf("[ ");
+ for (k = 0; k < r->high - r->low + 1; k++)
+ printf("%d ", cmap->table[r->offset + k]);
+ printf("]\n");
+ }
+ else if (r->flag == MULTI)
+ {
+ printf("< ");
+ n = cmap->table[r->offset];
+ for (k = 0; k < n; k++)
+ printf("%04x ", cmap->table[r->offset + 1 + k]);
+ printf(">\n");
+ }
+ else
+ printf("%d\n", r->offset);
+ }
+ printf(" }\n}\n");
+}
+
+/*
+ * Add a codespacerange section.
+ * These ranges are used by pdf_decodecmap to decode
+ * multi-byte encoded strings.
+ */
+fz_error *
+pdf_addcodespace(pdf_cmap *cmap, unsigned lo, unsigned hi, int n)
+{
+ int i;
+
+ if (cmap->ncspace + 1 == MAXCODESPACE)
+ return fz_throw("rangelimit: too many code space ranges");
+
+ cmap->cspace[cmap->ncspace].n = n;
+
+ for (i = 0; i < n; i++)
+ {
+ int o = (n - i - 1) * 8;
+ cmap->cspace[cmap->ncspace].lo[i] = (lo >> o) & 0xFF;
+ cmap->cspace[cmap->ncspace].hi[i] = (hi >> o) & 0xFF;
+ }
+
+ cmap->ncspace ++;
+
+ return nil;
+}
+
+/*
+ * Add an integer to the table.
+ */
+static fz_error *
+addtable(pdf_cmap *cmap, int value)
+{
+ if (cmap->tlen + 1 > cmap->tcap)
+ {
+ int newcap = cmap->tcap == 0 ? 256 : cmap->tcap * 2;
+ int *newtable = fz_realloc(cmap->table, newcap * sizeof(int));
+ if (!newtable)
+ return fz_outofmem;
+ cmap->tcap = newcap;
+ cmap->table = newtable;
+ }
+
+ cmap->table[cmap->tlen++] = value;
+
+ return nil;
+}
+
+/*
+ * Add a range.
+ */
+static fz_error *
+addrange(pdf_cmap *cmap, int low, int high, int flag, int offset)
+{
+ if (cmap->rlen + 1 > cmap->rcap)
+ {
+ pdf_range *newranges;
+ int newcap = cmap->rcap == 0 ? 256 : cmap->rcap * 2;
+ newranges = fz_realloc(cmap->ranges, newcap * sizeof(pdf_range));
+ if (!newranges)
+ return fz_outofmem;
+ cmap->rcap = newcap;
+ cmap->ranges = newranges;
+ }
+
+ cmap->ranges[cmap->rlen].low = low;
+ cmap->ranges[cmap->rlen].high = high;
+ cmap->ranges[cmap->rlen].flag = flag;
+ cmap->ranges[cmap->rlen].offset = offset;
+ cmap->rlen ++;
+
+ return nil;
+}
+
+/*
+ * Add a range-to-table mapping.
+ */
+fz_error *
+pdf_maprangetotable(pdf_cmap *cmap, int low, int *table, int len)
+{
+ fz_error *error;
+ int offset;
+ int high;
+ int i;
+
+ high = low + len;
+ offset = cmap->tlen;
+
+ for (i = 0; i < len; i++)
+ {
+ error = addtable(cmap, table[i]);
+ if (error)
+ return error;
+ }
+
+ return addrange(cmap, low, high, TABLE, offset);
+}
+
+/*
+ * Add a range of contiguous one-to-one mappings (ie 1..5 maps to 21..25)
+ */
+fz_error *
+pdf_maprangetorange(pdf_cmap *cmap, int low, int high, int offset)
+{
+ return addrange(cmap, low, high, high - low == 0 ? SINGLE : RANGE, offset);
+}
+
+/*
+ * Add a single one-to-many mapping.
+ */
+fz_error *
+pdf_maponetomany(pdf_cmap *cmap, int low, int *values, int len)
+{
+ fz_error *error;
+ int offset;
+ int i;
+
+ if (len == 1)
+ return addrange(cmap, low, low, SINGLE, values[0]);
+
+ offset = cmap->tlen;
+
+ error = addtable(cmap, len);
+ if (error)
+ return error;
+
+ for (i = 0; i < len; i++)
+ {
+ addtable(cmap, values[i]);
+ if (error)
+ return error;
+ }
+
+ return addrange(cmap, low, low, MULTI, offset);
+}
+
+/*
+ * Sort the input ranges.
+ * Merge contiguous input ranges to range-to-range if the output is contiguos.
+ * Merge contiguous input ranges to range-to-table if the output is random.
+ */
+fz_error *
+pdf_sortcmap(pdf_cmap *cmap)
+{
+ fz_error *error;
+ pdf_range *newranges;
+ int *newtable;
+ pdf_range *a; /* last written range on output */
+ pdf_range *b; /* current range examined on input */
+
+ qsort(cmap->ranges, cmap->rlen, sizeof(pdf_range), cmprange);
+
+ a = cmap->ranges;
+ b = cmap->ranges + 1;
+
+ while (b < cmap->ranges + cmap->rlen)
+ {
+ /* ignore one-to-many mappings */
+ if (b->flag == MULTI)
+ {
+ *(++a) = *b;
+ }
+
+ /* input contiguous */
+ else if (a->high + 1 == b->low)
+ {
+ /* output contiguous */
+ if (a->high - a->low + a->offset + 1 == b->offset)
+ {
+ /* SR -> R and SS -> R and RR -> R and RS -> R */
+ if (a->flag == SINGLE || a->flag == RANGE)
+ {
+ a->flag = RANGE;
+ a->high = b->high;
+ }
+
+ /* LS -> L */
+ else if (a->flag == TABLE && b->flag == SINGLE)
+ {
+ a->high = b->high;
+ error = addtable(cmap, b->offset);
+ if (error)
+ return error;
+ }
+
+ /* LR -> LR */
+ else if (a->flag == TABLE && b->flag == RANGE)
+ {
+ *(++a) = *b;
+ }
+
+ /* XX -> XX */
+ else
+ {
+ *(++a) = *b;
+ }
+ }
+
+ /* output separated */
+ else
+ {
+ /* SS -> L */
+ if (a->flag == SINGLE && b->flag == SINGLE)
+ {
+ a->flag = TABLE;
+ a->high = b->high;
+
+ error = addtable(cmap, a->offset);
+ if (error)
+ return error;
+
+ error = addtable(cmap, b->offset);
+ if (error)
+ return error;
+
+ a->offset = cmap->tlen - 2;
+ }
+
+ /* LS -> L */
+ else if (a->flag == TABLE && b->flag == SINGLE)
+ {
+ a->high = b->high;
+ error = addtable(cmap, b->offset);
+ if (error)
+ return error;
+ }
+
+ /* XX -> XX */
+ else
+ {
+ *(++a) = *b;
+ }
+ }
+ }
+
+ /* input separated: XX -> XX */
+ else
+ {
+ *(++a) = *b;
+ }
+
+ b ++;
+ }
+
+ cmap->rlen = a - cmap->ranges + 1;
+
+ assert(cmap->rlen > 0);
+
+ newranges = fz_realloc(cmap->ranges, cmap->rlen * sizeof(pdf_range));
+ if (!newranges)
+ return fz_outofmem;
+ cmap->rcap = cmap->rlen;
+ cmap->ranges = newranges;
+
+ if (cmap->tlen)
+ {
+ newtable = fz_realloc(cmap->table, cmap->tlen * sizeof(int));
+ if (!newtable)
+ return fz_outofmem;
+ cmap->tcap = cmap->tlen;
+ cmap->table = newtable;
+ }
+
+ return nil;
+}
+
+/*
+ * Lookup the mapping of a codepoint.
+ */
+int
+pdf_lookupcmap(pdf_cmap *cmap, int cpt)
+{
+ int l = 0;
+ int r = cmap->rlen - 1;
+ int m;
+
+ while (l <= r)
+ {
+ m = (l + r) >> 1;
+ if (cpt < cmap->ranges[m].low)
+ r = m - 1;
+ else if (cpt > cmap->ranges[m].high)
+ l = m + 1;
+ else
+ {
+ int i = cpt - cmap->ranges[m].low + cmap->ranges[m].offset;
+ if (cmap->ranges[m].flag == TABLE)
+ return cmap->table[i];
+ if (cmap->ranges[m].flag == MULTI)
+ return -1;
+ return i;
+ }
+ }
+
+ if (cmap->usecmap)
+ return pdf_lookupcmap(cmap->usecmap, cpt);
+
+ return -1;
+}
+
+/*
+ * Use the codespace ranges to extract a codepoint from a
+ * multi-byte encoded string.
+ */
+unsigned char *
+pdf_decodecmap(pdf_cmap *cmap, unsigned char *buf, int *cpt)
+{
+ int i, k;
+
+ for (k = 0; k < cmap->ncspace; k++)
+ {
+ unsigned char *lo = cmap->cspace[k].lo;
+ unsigned char *hi = cmap->cspace[k].hi;
+ int n = cmap->cspace[k].n;
+ int c = 0;
+
+ for (i = 0; i < n; i++)
+ {
+ if (lo[i] <= buf[i] && buf[i] <= hi[i])
+ c = (c << 8) | buf[i];
+ else
+ break;
+ }
+
+ if (i == n) {
+ *cpt = c;
+ return buf + n;
+ }
+ }
+
+ *cpt = 0;
+ return buf + 1;
+}
+
+/*
+ * CMap parser
+ */
enum
{
@@ -48,7 +588,7 @@ static int mylex(fz_stream *file, char *buf, int n, int *sl)
return token;
}
-static fz_error *parsecmapname(fz_cmap *cmap, fz_stream *file)
+static fz_error *parsecmapname(pdf_cmap *cmap, fz_stream *file)
{
char buf[256];
int token;
@@ -56,14 +596,14 @@ static fz_error *parsecmapname(fz_cmap *cmap, fz_stream *file)
token = mylex(file, buf, sizeof buf, &len);
if (token == PDF_TNAME) {
- fz_setcmapname(cmap, buf);
+ strlcpy(cmap->cmapname, buf, sizeof(cmap->cmapname));
return nil;
}
return fz_throw("syntaxerror in CMap after /CMapName");
}
-static fz_error *parsewmode(fz_cmap *cmap, fz_stream *file)
+static fz_error *parsewmode(pdf_cmap *cmap, fz_stream *file)
{
char buf[256];
int token;
@@ -71,14 +611,14 @@ static fz_error *parsewmode(fz_cmap *cmap, fz_stream *file)
token = mylex(file, buf, sizeof buf, &len);
if (token == PDF_TINT) {
- fz_setwmode(cmap, atoi(buf));
+ pdf_setwmode(cmap, atoi(buf));
return nil;
}
return fz_throw("syntaxerror in CMap after /WMode");
}
-static fz_error *parsecodespacerange(fz_cmap *cmap, fz_stream *file)
+static fz_error *parsecodespacerange(pdf_cmap *cmap, fz_stream *file)
{
char buf[256];
int token;
@@ -100,7 +640,7 @@ static fz_error *parsecodespacerange(fz_cmap *cmap, fz_stream *file)
if (token == PDF_TSTRING)
{
hi = codefromstring(buf, len);
- error = fz_addcodespacerange(cmap, lo, hi, len);
+ error = pdf_addcodespace(cmap, lo, hi, len);
if (error)
return error;
}
@@ -113,7 +653,7 @@ static fz_error *parsecodespacerange(fz_cmap *cmap, fz_stream *file)
return fz_throw("syntaxerror in CMap codespacerange section");
}
-static fz_error *parsecidrange(fz_cmap *cmap, fz_stream *file)
+static fz_error *parsecidrange(pdf_cmap *cmap, fz_stream *file)
{
char buf[256];
int token;
@@ -145,7 +685,7 @@ static fz_error *parsecidrange(fz_cmap *cmap, fz_stream *file)
dst = atoi(buf);
- error = fz_addcidrange(cmap, lo, hi, dst);
+ error = pdf_maprangetorange(cmap, lo, hi, dst);
if (error)
return error;
}
@@ -154,7 +694,7 @@ cleanup:
return fz_throw("syntaxerror in CMap cidrange section");
}
-static fz_error *parsecidchar(fz_cmap *cmap, fz_stream *file)
+static fz_error *parsecidchar(pdf_cmap *cmap, fz_stream *file)
{
char buf[256];
int token;
@@ -180,7 +720,7 @@ static fz_error *parsecidchar(fz_cmap *cmap, fz_stream *file)
dst = atoi(buf);
- error = fz_addcidrange(cmap, src, src, dst);
+ error = pdf_maprangetorange(cmap, src, src, dst);
if (error)
return error;
}
@@ -189,7 +729,41 @@ cleanup:
return fz_throw("syntaxerror in CMap cidchar section");
}
-static fz_error *parsebfrange(fz_cmap *cmap, fz_stream *file)
+static fz_error *parsebfrangearray(pdf_cmap *cmap, fz_stream *file, int lo, int hi)
+{
+ char buf[256];
+ int token;
+ int len;
+ fz_error *error;
+ int dst[256];
+ int i;
+
+ while (1)
+ {
+ token = mylex(file, buf, sizeof buf, &len);
+ /* Note: does not handle [ /Name /Name ... ] */
+
+ if (token == PDF_TCARRAY)
+ return nil;
+
+ else if (token != PDF_TSTRING)
+ return fz_throw("syntaxerror in CMap bfrange array section");
+
+ if (len / 2)
+ {
+ for (i = 0; i < len / 2; i++)
+ dst[i] = codefromstring(buf + i * 2, 2);
+
+ error = pdf_maponetomany(cmap, lo, dst, len / 2);
+ if (error)
+ return error;
+ }
+
+ lo ++;
+ }
+}
+
+static fz_error *parsebfrange(pdf_cmap *cmap, fz_stream *file)
{
char buf[256];
int token;
@@ -216,28 +790,64 @@ static fz_error *parsebfrange(fz_cmap *cmap, fz_stream *file)
hi = codefromstring(buf, len);
token = mylex(file, buf, sizeof buf, &len);
- /* Note: does not handle [ /Name /Name /Name ... ] */
- if (token != PDF_TSTRING)
- goto cleanup;
- dst = codefromstring(buf, len);
+ if (token == PDF_TSTRING)
+ {
+ if (len == 2)
+ {
+ dst = codefromstring(buf, len);
+ error = pdf_maprangetorange(cmap, lo, hi, dst);
+ if (error)
+ return error;
+ }
+ else
+ {
+ int dststr[256];
+ int i;
+
+ if (len / 2)
+ {
+ for (i = 0; i < len / 2; i++)
+ dststr[i] = codefromstring(buf + i * 2, 2);
+
+ while (lo <= hi)
+ {
+ dststr[i-1] ++;
+ error = pdf_maponetomany(cmap, lo, dststr, i);
+ if (error)
+ return error;
+ lo ++;
+ }
+ }
+ }
+ }
- error = fz_addcidrange(cmap, lo, hi, dst);
- if (error)
- return error;
+ else if (token == PDF_TOARRAY)
+ {
+ error = parsebfrangearray(cmap, file, lo, hi);
+ if (error)
+ return error;
+ }
+
+ else
+ {
+ goto cleanup;
+ }
}
cleanup:
return fz_throw("syntaxerror in CMap bfrange section");
}
-static fz_error *parsebfchar(fz_cmap *cmap, fz_stream *file)
+static fz_error *parsebfchar(pdf_cmap *cmap, fz_stream *file)
{
char buf[256];
int token;
int len;
fz_error *error;
- int src, dst;
+ int dst[256];
+ int src;
+ int i;
while (1)
{
@@ -256,11 +866,15 @@ static fz_error *parsebfchar(fz_cmap *cmap, fz_stream *file)
if (token != PDF_TSTRING)
goto cleanup;
- dst = codefromstring(buf, len);
+ if (len / 2)
+ {
+ for (i = 0; i < len / 2; i++)
+ dst[i] = codefromstring(buf + i * 2, 2);
- error = fz_addcidrange(cmap, src, src, dst);
- if (error)
- return error;
+ error = pdf_maponetomany(cmap, src, dst, i);
+ if (error)
+ return error;
+ }
}
cleanup:
@@ -268,16 +882,16 @@ cleanup:
}
fz_error *
-pdf_parsecmap(fz_cmap **cmapp, fz_stream *file)
+pdf_parsecmap(pdf_cmap **cmapp, fz_stream *file)
{
fz_error *error;
- fz_cmap *cmap;
+ pdf_cmap *cmap;
char key[64];
char buf[256];
int token;
int len;
- error = fz_newcmap(&cmap);
+ error = pdf_newcmap(&cmap);
if (error)
return error;
@@ -316,7 +930,7 @@ pdf_parsecmap(fz_cmap **cmapp, fz_stream *file)
else if (token == TUSECMAP)
{
- fz_setusecmapname(cmap, key);
+ strlcpy(cmap->usecmapname, key, sizeof(cmap->usecmapname));
}
else if (token == TBEGINCODESPACERANGE)
@@ -357,7 +971,7 @@ pdf_parsecmap(fz_cmap **cmapp, fz_stream *file)
/* ignore everything else */
}
- error = fz_endcidrange(cmap);
+ error = pdf_sortcmap(cmap);
if (error)
goto cleanup;
@@ -365,7 +979,7 @@ pdf_parsecmap(fz_cmap **cmapp, fz_stream *file)
return nil;
cleanup:
- fz_dropcmap(cmap);
+ pdf_dropcmap(cmap);
return error;
}
@@ -373,19 +987,19 @@ cleanup:
* Load CMap stream in PDF file
*/
fz_error *
-pdf_loadembeddedcmap(fz_cmap **cmapp, pdf_xref *xref, fz_obj *stmref)
+pdf_loadembeddedcmap(pdf_cmap **cmapp, pdf_xref *xref, fz_obj *stmref)
{
fz_obj *stmobj = stmref;
fz_error *error = nil;
fz_stream *file;
- fz_cmap *cmap = nil;
- fz_cmap *usecmap;
+ pdf_cmap *cmap = nil;
+ pdf_cmap *usecmap;
fz_obj *wmode;
fz_obj *obj;
if ((*cmapp = pdf_finditem(xref->store, PDF_KCMAP, stmref)))
{
- fz_keepcmap(*cmapp);
+ pdf_keepcmap(*cmapp);
return nil;
}
@@ -409,7 +1023,7 @@ pdf_loadembeddedcmap(fz_cmap **cmapp, pdf_xref *xref, fz_obj *stmref)
if (fz_isint(wmode))
{
pdf_logfont("wmode %d\n", wmode);
- fz_setwmode(cmap, fz_toint(wmode));
+ pdf_setwmode(cmap, fz_toint(wmode));
}
obj = fz_dictgets(stmobj, "UseCMap");
@@ -419,8 +1033,8 @@ pdf_loadembeddedcmap(fz_cmap **cmapp, pdf_xref *xref, fz_obj *stmref)
error = pdf_loadsystemcmap(&usecmap, fz_toname(obj));
if (error)
goto cleanup;
- fz_setusecmap(cmap, usecmap);
- fz_dropcmap(usecmap);
+ pdf_setusecmap(cmap, usecmap);
+ pdf_dropcmap(usecmap);
}
else if (fz_isindirect(obj))
{
@@ -428,8 +1042,8 @@ pdf_loadembeddedcmap(fz_cmap **cmapp, pdf_xref *xref, fz_obj *stmref)
error = pdf_loadembeddedcmap(&usecmap, xref, obj);
if (error)
goto cleanup;
- fz_setusecmap(cmap, usecmap);
- fz_dropcmap(usecmap);
+ pdf_setusecmap(cmap, usecmap);
+ pdf_dropcmap(usecmap);
}
pdf_logfont("}\n");
@@ -445,7 +1059,7 @@ pdf_loadembeddedcmap(fz_cmap **cmapp, pdf_xref *xref, fz_obj *stmref)
cleanup:
if (cmap)
- fz_dropcmap(cmap);
+ pdf_dropcmap(cmap);
fz_dropobj(stmobj);
return error;
}
@@ -454,16 +1068,19 @@ cleanup:
* Load predefined CMap from system
*/
fz_error *
-pdf_loadsystemcmap(fz_cmap **cmapp, char *name)
+pdf_loadsystemcmap(pdf_cmap **cmapp, char *name)
{
fz_error *error = nil;
fz_stream *file;
char *cmapdir;
char *usecmapname;
- fz_cmap *usecmap;
- fz_cmap *cmap;
+ pdf_cmap *usecmap;
+ pdf_cmap *cmap;
char path[1024];
+ cmap = nil;
+ file = nil;
+
pdf_logfont("load system cmap %s {\n", name);
cmapdir = getenv("CMAPDIR");
@@ -484,15 +1101,15 @@ pdf_loadsystemcmap(fz_cmap **cmapp, char *name)
fz_dropstream(file);
- usecmapname = fz_getusecmapname(cmap);
- if (usecmapname)
+ usecmapname = cmap->usecmapname;
+ if (usecmapname[0])
{
pdf_logfont("usecmap %s\n", usecmapname);
error = pdf_loadsystemcmap(&usecmap, usecmapname);
if (error)
goto cleanup;
- fz_setusecmap(cmap, usecmap);
- fz_dropcmap(usecmap);
+ pdf_setusecmap(cmap, usecmap);
+ pdf_dropcmap(usecmap);
}
pdf_logfont("}\n");
@@ -502,7 +1119,7 @@ pdf_loadsystemcmap(fz_cmap **cmapp, char *name)
cleanup:
if (cmap)
- fz_dropcmap(cmap);
+ pdf_dropcmap(cmap);
if (file)
fz_dropstream(file);
return error;
@@ -512,34 +1129,36 @@ cleanup:
* Create an Identity-* CMap (for both 1 and 2-byte encodings)
*/
fz_error *
-pdf_makeidentitycmap(fz_cmap **cmapp, int wmode, int bytes)
+pdf_newidentitycmap(pdf_cmap **cmapp, int wmode, int bytes)
{
fz_error *error;
- fz_cmap *cmap;
+ pdf_cmap *cmap;
- error = fz_newcmap(&cmap);
+ error = pdf_newcmap(&cmap);
if (error)
return error;
- error = fz_addcodespacerange(cmap, 0x0000, 0xffff, bytes);
+ sprintf(cmap->cmapname, "Identity-%c", wmode ? 'V' : 'H');
+
+ error = pdf_addcodespace(cmap, 0x0000, 0xffff, bytes);
if (error) {
- fz_dropcmap(cmap);
+ pdf_dropcmap(cmap);
return error;
}
- error = fz_addcidrange(cmap, 0x0000, 0xffff, 0);
+ error = pdf_maprangetorange(cmap, 0x0000, 0xffff, 0);
if (error) {
- fz_dropcmap(cmap);
+ pdf_dropcmap(cmap);
return error;
}
- error = fz_endcidrange(cmap);
+ error = pdf_sortcmap(cmap);
if (error) {
- fz_dropcmap(cmap);
+ pdf_dropcmap(cmap);
return error;
}
- fz_setwmode(cmap, wmode);
+ pdf_setwmode(cmap, wmode);
*cmapp = cmap;
return nil;
diff --git a/mupdf/pdf_font.c b/mupdf/pdf_font.c
index 064988ca..0c465b8f 100644
--- a/mupdf/pdf_font.c
+++ b/mupdf/pdf_font.c
@@ -62,7 +62,7 @@ static inline int ftcidtogid(pdf_font *font, int cid)
{
if (font->tottfcmap)
{
- cid = fz_lookupcid(font->tottfcmap, cid);
+ cid = pdf_lookupcmap(font->tottfcmap, cid);
return FT_Get_Char_Index(font->ftface, cid);
}
@@ -223,11 +223,11 @@ static void ftdropfont(fz_font *font)
{
pdf_font *pfont = (pdf_font*)font;
if (pfont->encoding)
- fz_dropcmap(pfont->encoding);
+ pdf_dropcmap(pfont->encoding);
if (pfont->tottfcmap)
- fz_dropcmap(pfont->tottfcmap);
+ pdf_dropcmap(pfont->tottfcmap);
if (pfont->tounicode)
- fz_dropcmap(pfont->tounicode);
+ pdf_dropcmap(pfont->tounicode);
fz_free(pfont->cidtogid);
fz_free(pfont->cidtoucs);
if (pfont->ftface)
@@ -509,7 +509,7 @@ loadsimplefont(pdf_font **fontp, pdf_xref *xref, fz_obj *dict, fz_obj *ref)
}
}
- error = pdf_makeidentitycmap(&font->encoding, 0, 1);
+ error = pdf_newidentitycmap(&font->encoding, 0, 1);
if (error)
goto cleanup;
@@ -676,9 +676,9 @@ loadcidfont(pdf_font **fontp, pdf_xref *xref, fz_obj *dict, fz_obj *ref, fz_obj
{
pdf_logfont("encoding /%s\n", fz_toname(encoding));
if (!strcmp(fz_toname(encoding), "Identity-H"))
- error = pdf_makeidentitycmap(&font->encoding, 0, 2);
+ error = pdf_newidentitycmap(&font->encoding, 0, 2);
else if (!strcmp(fz_toname(encoding), "Identity-V"))
- error = pdf_makeidentitycmap(&font->encoding, 1, 2);
+ error = pdf_newidentitycmap(&font->encoding, 1, 2);
else
error = pdf_loadsystemcmap(&font->encoding, fz_toname(encoding));
}
@@ -694,8 +694,8 @@ loadcidfont(pdf_font **fontp, pdf_xref *xref, fz_obj *dict, fz_obj *ref, fz_obj
if (error)
goto cleanup;
- fz_setfontwmode((fz_font*)font, fz_getwmode(font->encoding));
- pdf_logfont("wmode %d\n", fz_getwmode(font->encoding));
+ fz_setfontwmode((fz_font*)font, pdf_getwmode(font->encoding));
+ pdf_logfont("wmode %d\n", pdf_getwmode(font->encoding));
if (kind == TRUETYPE)
{
@@ -820,7 +820,7 @@ loadcidfont(pdf_font **fontp, pdf_xref *xref, fz_obj *dict, fz_obj *ref, fz_obj
* Vertical
*/
- if (fz_getwmode(font->encoding) == 1)
+ if (pdf_getwmode(font->encoding) == 1)
{
fz_obj *obj;
int dw2y = 880;
diff --git a/mupdf/pdf_interpret.c b/mupdf/pdf_interpret.c
index e5ecf8ee..057ac6d1 100644
--- a/mupdf/pdf_interpret.c
+++ b/mupdf/pdf_interpret.c
@@ -291,46 +291,32 @@ runkeyword(pdf_csi *csi, pdf_xref *xref, fz_obj *rdb, char *buf)
else if (!strcmp(buf, "MP"))
{
- fz_node *meta;
if (csi->top != 1)
goto syntaxerror;
- error = fz_newmetanode(&meta, csi->stack[0], nil);
- if (error) return error;
- fz_insertnodelast(gstate->head, meta);
}
else if (!strcmp(buf, "DP"))
{
- fz_node *meta;
if (csi->top != 2)
goto syntaxerror;
- error = fz_newmetanode(&meta, csi->stack[0], csi->stack[1]);
- if (error) return error;
- fz_insertnodelast(gstate->head, meta);
}
else if (!strcmp(buf, "BMC"))
{
- fz_node *meta;
if (csi->top != 1)
goto syntaxerror;
- error = fz_newmetanode(&meta, csi->stack[0], nil);
- if (error) return error;
- fz_insertnodelast(gstate->head, meta);
}
else if (!strcmp(buf, "BDC"))
{
- fz_node *meta;
if (csi->top != 2)
goto syntaxerror;
- error = fz_newmetanode(&meta, csi->stack[0], csi->stack[1]);
- if (error) return error;
- fz_insertnodelast(gstate->head, meta);
}
else if (!strcmp(buf, "EMC"))
{
+ if (csi->top != 0)
+ goto syntaxerror;
}
else if (!strcmp(buf, "cm"))
diff --git a/mupdf/pdf_store.c b/mupdf/pdf_store.c
index d46cd628..f632657f 100644
--- a/mupdf/pdf_store.c
+++ b/mupdf/pdf_store.c
@@ -57,7 +57,7 @@ static void dropitem(pdf_itemkind kind, void *val)
case PDF_KIMAGE: fz_dropimage(val); break;
case PDF_KPATTERN: pdf_droppattern(val); break;
case PDF_KSHADE: fz_dropshade(val); break;
- case PDF_KCMAP: fz_dropcmap(val); break;
+ case PDF_KCMAP: pdf_dropcmap(val); break;
case PDF_KFONT: fz_dropfont(val); break;
}
}
@@ -112,7 +112,7 @@ pdf_storeitem(pdf_store *store, pdf_itemkind kind, fz_obj *key, void *val)
case PDF_KIMAGE: fz_keepimage(val); break;
case PDF_KPATTERN: pdf_keeppattern(val); break;
case PDF_KSHADE: fz_keepshade(val); break;
- case PDF_KCMAP: fz_keepcmap(val); break;
+ case PDF_KCMAP: pdf_keepcmap(val); break;
case PDF_KFONT: fz_keepfont(val); break;
}
diff --git a/mupdf/pdf_type3.c b/mupdf/pdf_type3.c
index 092db4de..b19a7308 100644
--- a/mupdf/pdf_type3.c
+++ b/mupdf/pdf_type3.c
@@ -11,7 +11,7 @@ t3dropfont(fz_font *font)
int i;
pdf_font *pfont = (pdf_font*)font;
if (pfont->encoding)
- fz_dropcmap(pfont->encoding);
+ pdf_dropcmap(pfont->encoding);
for (i = 0; i < 256; i++)
if (pfont->charprocs[i])
fz_droptree(pfont->charprocs[i]);
@@ -188,7 +188,7 @@ pdf_loadtype3font(pdf_font **fontp, pdf_xref *xref, fz_obj *dict, fz_obj *ref)
fz_dropobj(encoding);
- error = pdf_makeidentitycmap(&font->encoding, 0, 1);
+ error = pdf_newidentitycmap(&font->encoding, 0, 1);
if (error)
goto cleanup;
diff --git a/mupdf/pdf_unicode.c b/mupdf/pdf_unicode.c
index 28100b78..7affd1b9 100644
--- a/mupdf/pdf_unicode.c
+++ b/mupdf/pdf_unicode.c
@@ -10,7 +10,7 @@ pdf_loadtounicode(pdf_font *font, pdf_xref *xref,
char **strings, char *collection, fz_obj *cmapstm)
{
fz_error *error;
- fz_cmap *cmap;
+ pdf_cmap *cmap;
int cid;
int ucs;
int i;
@@ -23,28 +23,31 @@ pdf_loadtounicode(pdf_font *font, pdf_xref *xref,
if (error)
return error;
- error = fz_newcmap(&font->tounicode);
+ error = pdf_newcmap(&font->tounicode);
if (error)
goto cleanup;
for (i = 0; i < (strings ? 256 : 65536); i++)
{
- cid = fz_lookupcid(font->encoding, i);
+ cid = pdf_lookupcmap(font->encoding, i);
if (cid > 0)
{
- ucs = fz_lookupcid(cmap, i);
- error = fz_addcidrange(font->tounicode, cid, cid, ucs);
- if (error)
- goto cleanup;
+ ucs = pdf_lookupcmap(cmap, i);
+ if (ucs > 0)
+ {
+ error = pdf_maprangetorange(font->tounicode, cid, cid, ucs);
+ if (error)
+ goto cleanup;
+ }
}
}
- error = fz_endcidrange(font->tounicode);
+ error = pdf_sortcmap(font->tounicode);
if (error)
goto cleanup;
cleanup:
- fz_dropcmap(cmap);
+ pdf_dropcmap(cmap);
return error;
}
@@ -68,6 +71,8 @@ pdf_loadtounicode(pdf_font *font, pdf_xref *xref,
{
pdf_logfont("tounicode strings\n");
+ /* TODO use tounicode cmap here ... for one-to-many mappings */
+
font->ncidtoucs = 256;
font->cidtoucs = fz_malloc(256 * sizeof(unsigned short));
if (!font->cidtoucs)
@@ -235,7 +240,7 @@ extracttext(pdf_textline **line, fz_node *node, fz_matrix ctm)
box.y1 = MAX(0, MAX(vx.y, vy.y)) + y;
if (font->tounicode)
- c = fz_lookupcid(font->tounicode, g);
+ c = pdf_lookupcmap(font->tounicode, g);
else if (g < font->ncidtoucs)
c = font->cidtoucs[g];
else
diff --git a/world/Jamfile b/world/Jamfile
index b587c805..9f53ef0a 100644
--- a/world/Jamfile
+++ b/world/Jamfile
@@ -10,7 +10,6 @@ Library libworld :
node_text.c
node_tree.c
- res_cmap.c
res_colorspace.c
res_font.c
res_image.c
diff --git a/world/node_misc2.c b/world/node_misc2.c
index 8abcd4a5..6446d52f 100644
--- a/world/node_misc2.c
+++ b/world/node_misc2.c
@@ -162,7 +162,7 @@ fz_boundtransformnode(fz_transformnode *node, fz_matrix ctm)
*/
fz_error *
-fz_newmetanode(fz_node **nodep, void *name, void *dict)
+fz_newmetanode(fz_node **nodep, char *name, void *dict)
{
fz_metanode *node;
@@ -172,13 +172,8 @@ fz_newmetanode(fz_node **nodep, void *name, void *dict)
*nodep = (fz_node*)node;
fz_initnode((fz_node*)node, FZ_NMETA);
- node->name = nil;
- node->dict = nil;
-
- if (name)
- node->name = name;
- if (dict)
- node->dict = dict;
+ node->name = name;
+ node->dict = dict;
return nil;
}
@@ -186,10 +181,8 @@ fz_newmetanode(fz_node **nodep, void *name, void *dict)
void
fz_dropmetanode(fz_metanode *node)
{
- if (node->name)
- fz_warn("leaking meta node name");
if (node->dict)
- fz_warn("leaking meta node dict");
+ fz_warn("leaking meta node '%s'", node->name);
}
fz_rect
diff --git a/world/node_toxml.c b/world/node_toxml.c
index 2099c6a8..b7cd6c15 100644
--- a/world/node_toxml.c
+++ b/world/node_toxml.c
@@ -14,7 +14,7 @@ static void xmlmeta(fz_metanode *node, int level)
fz_node *child;
indent(level);
- printf("<meta>\n");
+ printf("<meta name=\"%s\">\n", node->name);
for (child = node->super.first; child; child = child->next)
xmlnode(child, level + 1);
diff --git a/world/res_cmap.c b/world/res_cmap.c
deleted file mode 100644
index bc2b4249..00000000
--- a/world/res_cmap.c
+++ /dev/null
@@ -1,467 +0,0 @@
-#include "fitz-base.h"
-#include "fitz-world.h"
-
-typedef struct fz_range_s fz_range;
-
-enum { MAXCODESPACE = 10 };
-enum { SINGLE, RANGE, LOOKUP };
-
-struct fz_range_s
-{
- int low;
- int high;
- int flag;
- int offset;
-};
-
-struct fz_cmap_s
-{
- int refs;
- char cmapname[32];
-
- char usecmapname[32];
- fz_cmap *usecmap;
-
- int wmode;
-
- int ncspace;
- struct {
- int n;
- unsigned char lo[4];
- unsigned char hi[4];
- } cspace[MAXCODESPACE];
-
- int rlen, rcap;
- fz_range *ranges;
-
- int tlen, tcap;
- int *lookup;
-};
-
-fz_error *
-fz_newcmap(fz_cmap **cmapp)
-{
- fz_cmap *cmap;
-
- cmap = *cmapp = fz_malloc(sizeof(fz_cmap));
- if (!cmap)
- return fz_outofmem;
-
- cmap->refs = 1;
- strcpy(cmap->cmapname, "");
-
- strcpy(cmap->usecmapname, "");
- cmap->usecmap = nil;
-
- cmap->wmode = 0;
-
- cmap->ncspace = 0;
-
- cmap->rlen = 0;
- cmap->rcap = 0;
- cmap->ranges = nil;
-
- cmap->tlen = 0;
- cmap->tcap = 0;
- cmap->lookup = nil;
-
- return nil;
-}
-
-fz_cmap *
-fz_keepcmap(fz_cmap *cmap)
-{
- cmap->refs ++;
- return cmap;
-}
-
-void
-fz_dropcmap(fz_cmap *cmap)
-{
- if (--cmap->refs == 0)
- {
- if (cmap->usecmap)
- fz_dropcmap(cmap->usecmap);
- fz_free(cmap->ranges);
- fz_free(cmap->lookup);
- fz_free(cmap);
- }
-}
-
-char *
-fz_getcmapname(fz_cmap *cmap)
-{
- if (cmap->cmapname[0])
- return cmap->cmapname;
- return nil;
-}
-
-void
-fz_setcmapname(fz_cmap *cmap, char *cmapname)
-{
- strlcpy(cmap->cmapname, cmapname, sizeof cmap->cmapname);
-}
-
-char *
-fz_getusecmapname(fz_cmap *cmap)
-{
- if (cmap->usecmapname[0])
- return cmap->usecmapname;
- return nil;
-}
-
-void
-fz_setusecmapname(fz_cmap *cmap, char *usecmap)
-{
- strlcpy(cmap->usecmapname, usecmap, sizeof cmap->usecmapname);
-}
-
-fz_cmap *
-fz_getusecmap(fz_cmap *cmap)
-{
- return cmap->usecmap;
-}
-
-void
-fz_setusecmap(fz_cmap *cmap, fz_cmap *usecmap)
-{
- int i;
-
- if (cmap->usecmap)
- fz_dropcmap(cmap->usecmap);
- cmap->usecmap = fz_keepcmap(usecmap);
-
- if (cmap->ncspace == 0)
- {
- cmap->ncspace = usecmap->ncspace;
- for (i = 0; i < usecmap->ncspace; i++)
- cmap->cspace[i] = usecmap->cspace[i];
- }
-}
-
-void
-fz_setwmode(fz_cmap *cmap, int wmode)
-{
- cmap->wmode = wmode;
-}
-
-int
-fz_getwmode(fz_cmap *cmap)
-{
- return cmap->wmode;
-}
-
-fz_error *
-fz_addcodespacerange(fz_cmap *cmap, unsigned lo, unsigned hi, int n)
-{
- int i;
-
- if (cmap->ncspace + 1 == MAXCODESPACE)
- return fz_throw("rangelimit: too many code space ranges");
-
- cmap->cspace[cmap->ncspace].n = n;
-
- for (i = 0; i < n; i++)
- {
- int o = (n - i - 1) * 8;
- cmap->cspace[cmap->ncspace].lo[i] = (lo >> o) & 0xFF;
- cmap->cspace[cmap->ncspace].hi[i] = (hi >> o) & 0xFF;
- }
-
- cmap->ncspace ++;
-
- return nil;
-}
-
-fz_error *
-fz_addcidrange(fz_cmap *cmap, int low, int high, int offset)
-{
- if (cmap->rlen + 1 > cmap->rcap)
- {
- fz_range *newranges;
- int newcap = cmap->rcap == 0 ? 256 : cmap->rcap * 2;
- newranges = fz_realloc(cmap->ranges, newcap * sizeof(fz_range));
- if (!newranges)
- return fz_outofmem;
- cmap->rcap = newcap;
- cmap->ranges = newranges;
- }
-
- cmap->ranges[cmap->rlen].low = low;
- cmap->ranges[cmap->rlen].high = high;
- cmap->ranges[cmap->rlen].flag = high - low == 0 ? SINGLE : RANGE;
- cmap->ranges[cmap->rlen].offset = offset;
- cmap->rlen ++;
-
- return nil;
-}
-
-static fz_error *
-addlookup(fz_cmap *cmap, int value)
-{
- if (cmap->tlen + 1 > cmap->tcap)
- {
- int newcap = cmap->tcap == 0 ? 256 : cmap->tcap * 2;
- int *newlookup = fz_realloc(cmap->lookup, newcap * sizeof(int));
- if (!newlookup)
- return fz_outofmem;
- cmap->tcap = newcap;
- cmap->lookup = newlookup;
- }
-
- cmap->lookup[cmap->tlen++] = value;
-
- return nil;
-}
-
-static int compare(const void *va, const void *vb)
-{
- return ((const fz_range*)va)->low - ((const fz_range*)vb)->low;
-}
-
-fz_error *
-fz_endcidrange(fz_cmap *cmap)
-{
- fz_error *error;
- fz_range *newranges;
- int *newlookup;
- fz_range *a; /* last written range on output */
- fz_range *b; /* current range examined on input */
-
- qsort(cmap->ranges, cmap->rlen, sizeof(fz_range), compare);
-
- a = cmap->ranges;
- b = cmap->ranges + 1;
-
- while (b < cmap->ranges + cmap->rlen)
- {
- /* input contiguous */
- if (a->high + 1 == b->low)
- {
- /* output contiguous */
- if (a->high - a->low + a->offset + 1 == b->offset)
- {
- /* SR -> R and SS -> R and RR -> R and RS -> R */
- if (a->flag == SINGLE || a->flag == RANGE)
- {
- a->flag = RANGE;
- a->high = b->high;
- }
-
- /* LS -> L */
- else if (a->flag == LOOKUP && b->flag == SINGLE)
- {
- a->high = b->high;
- error = addlookup(cmap, b->offset);
- if (error)
- return error;
- }
-
- /* LR -> LR */
- else if (a->flag == LOOKUP && b->flag == RANGE)
- {
- *(++a) = *b;
- }
- }
-
- /* output separated */
- else
- {
- /* SS -> L */
- if (a->flag == SINGLE && b->flag == SINGLE)
- {
- a->flag = LOOKUP;
- a->high = b->high;
-
- error = addlookup(cmap, a->offset);
- if (error)
- return error;
-
- error = addlookup(cmap, b->offset);
- if (error)
- return error;
-
- a->offset = cmap->tlen - 2;
- }
-
- /* LS -> L */
- else if (a->flag == LOOKUP && b->flag == SINGLE)
- {
- a->high = b->high;
- error = addlookup(cmap, b->offset);
- if (error)
- return error;
- }
-
- /* XX -> XX */
- else
- {
- *(++a) = *b;
- }
- }
- }
-
- /* input separated: XX -> XX */
- else
- {
- *(++a) = *b;
- }
-
- b ++;
- }
-
- cmap->rlen = a - cmap->ranges + 1;
-
- assert(cmap->rlen > 0);
-
- newranges = fz_realloc(cmap->ranges, cmap->rlen * sizeof(fz_range));
- if (!newranges)
- return fz_outofmem;
- cmap->rcap = cmap->rlen;
- cmap->ranges = newranges;
-
- if (cmap->tlen)
- {
- newlookup = fz_realloc(cmap->lookup, cmap->tlen * sizeof(int));
- if (!newlookup)
- return fz_outofmem;
- cmap->tcap = cmap->tlen;
- cmap->lookup = newlookup;
- }
-
- return nil;
-}
-
-fz_error *
-fz_setcidlookup(fz_cmap *cmap, int map[256])
-{
- int i;
-
- cmap->rlen = cmap->rcap = 1;
- cmap->ranges = fz_malloc(sizeof (fz_range));
- if (!cmap->ranges) {
- return fz_outofmem;
- }
-
- cmap->tlen = cmap->tcap = 256;
- cmap->lookup = fz_malloc(sizeof (int) * 256);
- if (!cmap->lookup) {
- fz_free(cmap->ranges);
- return fz_outofmem;
- }
-
- cmap->ranges[0].low = 0;
- cmap->ranges[0].high = 255;
- cmap->ranges[0].flag = LOOKUP;
- cmap->ranges[0].offset = 0;
-
- for (i = 0; i < 256; i++)
- cmap->lookup[i] = map[i];
-
- return nil;
-}
-
-int
-fz_lookupcid(fz_cmap *cmap, int cpt)
-{
- int l = 0;
- int r = cmap->rlen - 1;
- int m;
-
- while (l <= r)
- {
- m = (l + r) >> 1;
- if (cpt < cmap->ranges[m].low)
- r = m - 1;
- else if (cpt > cmap->ranges[m].high)
- l = m + 1;
- else
- {
- int i = cpt - cmap->ranges[m].low + cmap->ranges[m].offset;
- if (cmap->ranges[m].flag == LOOKUP)
- return cmap->lookup[i];
- return i;
- }
- }
-
- if (cmap->usecmap)
- return fz_lookupcid(cmap->usecmap, cpt);
-
- return -1;
-}
-
-unsigned char *
-fz_decodecpt(fz_cmap *cmap, unsigned char *buf, int *cpt)
-{
- int i, k;
-
- for (k = 0; k < cmap->ncspace; k++)
- {
- unsigned char *lo = cmap->cspace[k].lo;
- unsigned char *hi = cmap->cspace[k].hi;
- int n = cmap->cspace[k].n;
- int c = 0;
-
- for (i = 0; i < n; i++)
- {
- if (lo[i] <= buf[i] && buf[i] <= hi[i])
- c = (c << 8) | buf[i];
- else
- break;
- }
-
- if (i == n) {
- *cpt = c;
- return buf + n;
- }
- }
-
- *cpt = 0;
- return buf + 1;
-}
-
-void
-fz_debugcmap(fz_cmap *cmap)
-{
- int i, k;
-
- printf("cmap $%p /%s {\n", cmap, cmap->cmapname);
-
- if (cmap->usecmapname[0])
- printf(" usecmap /%s\n", cmap->usecmapname);
- if (cmap->usecmap)
- printf(" usecmap $%p\n", cmap->usecmap);
-
- printf(" wmode %d\n", cmap->wmode);
-
- printf(" codespaces {\n");
- for (i = 0; i < cmap->ncspace; i++)
- {
- printf(" <");
- for (k = 0; k < cmap->cspace[i].n; k++)
- printf("%02x", cmap->cspace[i].lo[k]);
- printf("> <");
- for (k = 0; k < cmap->cspace[i].n; k++)
- printf("%02x", cmap->cspace[i].hi[k]);
- printf(">\n");
- }
- printf(" }\n");
-
- printf(" ranges (%d,%d) {\n", cmap->rlen, cmap->tlen);
- for (i = 0; i < cmap->rlen; i++)
- {
- fz_range *r = &cmap->ranges[i];
- printf(" <%04x> <%04x> ", r->low, r->high);
- if (r->flag == LOOKUP)
- {
- printf("[ ");
- for (k = 0; k < r->high - r->low + 1; k++)
- printf("%d ", cmap->lookup[r->offset + k]);
- printf("]\n");
- }
- else
- printf("%d\n", r->offset);
- }
- printf(" }\n}\n");
-}
-
generated by cgit v1.2.3 (git 2.25.1) at 2025-03-24 03:19:33 +0000