summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
Diffstat
-rw-r--r--mupdf/mupdf.h8
-rw-r--r--mupdf/pdf_build.c55
-rw-r--r--mupdf/pdf_interpret.c84
3 files changed, 86 insertions, 61 deletions
diff --git a/mupdf/mupdf.h b/mupdf/mupdf.h
index c5fa9ca9..da04d32e 100644
--- a/mupdf/mupdf.h
+++ b/mupdf/mupdf.h
@@ -599,13 +599,17 @@ struct pdf_csi_s
fz_device *dev;
pdf_xref *xref;
- fz_obj *obj, *array;
+ fz_obj *obj;
char name[64];
+ unsigned char string[256];
+ int stringlen;
float stack[32];
int istack[32];
int top;
int xbalance;
+ int intext;
+ int inarray;
/* path object state */
fz_path *path;
@@ -632,6 +636,8 @@ void pdf_setcolor(pdf_csi *csi, int what, float *v);
void pdf_setpattern(pdf_csi *csi, int what, pdf_pattern *pat, float *v);
void pdf_setshade(pdf_csi *csi, int what, fz_shade *shade);
void pdf_showpath(pdf_csi*, int close, int fill, int stroke, int evenodd);
+void pdf_showspace(pdf_csi *csi, float tadj);
+void pdf_showstring(pdf_csi *csi, unsigned char *buf, int len);
void pdf_showtext(pdf_csi*, fz_obj *text);
void pdf_flushtext(pdf_csi*);
void pdf_showimage(pdf_csi*, fz_pixmap *image);
diff --git a/mupdf/pdf_build.c b/mupdf/pdf_build.c
index 739ec976..3c9f21d1 100644
--- a/mupdf/pdf_build.c
+++ b/mupdf/pdf_build.c
@@ -629,11 +629,18 @@ pdf_showglyph(pdf_csi *csi, int cid)
}
}
-static void
+void
pdf_showspace(pdf_csi *csi, float tadj)
{
pdf_gstate *gstate = csi->gstate + csi->gtop;
pdf_fontdesc *fontdesc = gstate->font;
+
+ if (!fontdesc)
+ {
+ fz_warn("cannot draw text since font and size not set");
+ return;
+ }
+
if (fontdesc->wmode == 0)
csi->tm = fz_concat(fz_translate(tadj * gstate->scale, 0), csi->tm);
else
@@ -641,13 +648,11 @@ pdf_showspace(pdf_csi *csi, float tadj)
}
void
-pdf_showtext(pdf_csi *csi, fz_obj *text)
+pdf_showstring(pdf_csi *csi, unsigned char *buf, int len)
{
pdf_gstate *gstate = csi->gstate + csi->gtop;
pdf_fontdesc *fontdesc = gstate->font;
- unsigned char *buf;
- unsigned char *end;
- int i, len;
+ unsigned char *end = buf + len;
int cpt, cid;
if (!fontdesc)
@@ -656,34 +661,38 @@ pdf_showtext(pdf_csi *csi, fz_obj *text)
return;
}
+ while (buf < end)
+ {
+ buf = pdf_decodecmap(fontdesc->encoding, buf, &cpt);
+ cid = pdf_lookupcmap(fontdesc->encoding, cpt);
+ if (cid >= 0)
+ pdf_showglyph(csi, cid);
+ else
+ fz_warn("cannot encode character with code point %#x", cpt);
+ if (cpt == 32)
+ pdf_showspace(csi, gstate->wordspace);
+ }
+}
+
+void
+pdf_showtext(pdf_csi *csi, fz_obj *text)
+{
+ pdf_gstate *gstate = csi->gstate + csi->gtop;
+ int i;
+
if (fz_isarray(text))
{
for (i = 0; i < fz_arraylen(text); i++)
{
fz_obj *item = fz_arrayget(text, i);
if (fz_isstring(item))
- pdf_showtext(csi, item);
+ pdf_showstring(csi, (unsigned char *)fz_tostrbuf(item), fz_tostrlen(item));
else
pdf_showspace(csi, - fz_toreal(item) * gstate->size * 0.001f);
}
}
-
- if (fz_isstring(text))
+ else if (fz_isstring(text))
{
- buf = (unsigned char *)fz_tostrbuf(text);
- len = fz_tostrlen(text);
- end = buf + len;
-
- while (buf < end)
- {
- buf = pdf_decodecmap(fontdesc->encoding, buf, &cpt);
- cid = pdf_lookupcmap(fontdesc->encoding, cpt);
- if (cid >= 0)
- pdf_showglyph(csi, cid);
- else
- fz_warn("cannot encode character with code point %#x", cpt);
- if (cpt == 32)
- pdf_showspace(csi, gstate->wordspace);
- }
+ pdf_showstring(csi, (unsigned char *)fz_tostrbuf(text), fz_tostrlen(text));
}
}
diff --git a/mupdf/pdf_interpret.c b/mupdf/pdf_interpret.c
index 6c7e0307..55d58e79 100644
--- a/mupdf/pdf_interpret.c
+++ b/mupdf/pdf_interpret.c
@@ -12,12 +12,14 @@ pdf_newcsi(pdf_xref *xref, fz_device *dev, fz_matrix ctm)
csi->top = 0;
csi->obj = nil;
- csi->array = nil;
csi->name[0] = 0;
+ csi->stringlen = 0;
memset(csi->stack, 0, sizeof csi->stack);
memset(csi->istack, 0, sizeof csi->istack);
csi->xbalance = 0;
+ csi->intext = 0;
+ csi->inarray = 0;
csi->path = fz_newpath();
csi->clip = 0;
@@ -46,6 +48,7 @@ pdf_clearstack(pdf_csi *csi)
csi->obj = nil;
csi->name[0] = 0;
+ csi->stringlen = 0;
for (i = 0; i < csi->top; i++)
{
@@ -150,7 +153,6 @@ pdf_freecsi(pdf_csi *csi)
if (csi->path) fz_freepath(csi->path);
if (csi->text) fz_freetext(csi->text);
- if (csi->array) fz_dropobj(csi->array);
pdf_clearstack(csi);
@@ -458,6 +460,7 @@ static void pdf_run_BMC(pdf_csi *csi)
static void pdf_run_BT(pdf_csi *csi)
{
+ csi->intext = 1;
csi->tm = fz_identity;
csi->tlm = fz_identity;
}
@@ -598,6 +601,7 @@ static void pdf_run_ET(pdf_csi *csi)
{
pdf_flushtext(csi);
csi->accumulate = 1;
+ csi->intext = 0;
}
static void pdf_run_EX(pdf_csi *csi)
@@ -863,12 +867,18 @@ static void pdf_run_Tstar(pdf_csi *csi)
static void pdf_run_Tj(pdf_csi *csi)
{
- pdf_showtext(csi, csi->obj);
+ if (csi->stringlen)
+ pdf_showstring(csi, csi->string, csi->stringlen);
+ else
+ pdf_showtext(csi, csi->obj);
}
static void pdf_run_TJ(pdf_csi *csi)
{
- pdf_showtext(csi, csi->obj);
+ if (csi->stringlen)
+ pdf_showstring(csi, csi->string, csi->stringlen);
+ else
+ pdf_showtext(csi, csi->obj);
}
static void pdf_run_W(pdf_csi *csi)
@@ -1134,7 +1144,10 @@ static void pdf_run_squote(pdf_csi *csi)
csi->tlm = fz_concat(m, csi->tlm);
csi->tm = csi->tlm;
- pdf_showtext(csi, csi->obj);
+ if (csi->stringlen)
+ pdf_showstring(csi, csi->string, csi->stringlen);
+ else
+ pdf_showtext(csi, csi->obj);
}
static void pdf_run_dquote(pdf_csi *csi)
@@ -1149,7 +1162,10 @@ static void pdf_run_dquote(pdf_csi *csi)
csi->tlm = fz_concat(m, csi->tlm);
csi->tm = csi->tlm;
- pdf_showtext(csi, csi->obj);
+ if (csi->stringlen)
+ pdf_showstring(csi, csi->string, csi->stringlen);
+ else
+ pdf_showtext(csi, csi->obj);
}
static fz_error
@@ -1762,7 +1778,6 @@ pdf_runcsifile(pdf_csi *csi, fz_obj *rdb, fz_stream *file, char *buf, int buflen
fz_error error;
int tok;
int len;
- fz_obj *obj;
pdf_clearstack(csi);
@@ -1775,25 +1790,21 @@ pdf_runcsifile(pdf_csi *csi, fz_obj *rdb, fz_stream *file, char *buf, int buflen
if (error)
return fz_rethrow(error, "lexical error in content stream");
- if (csi->array)
+ if (csi->inarray)
{
if (tok == PDF_TCARRAY)
{
- csi->obj = csi->array;
- csi->array = nil;
+ csi->inarray = 0;
csi->top ++;
}
else if (tok == PDF_TINT || tok == PDF_TREAL)
{
- obj = fz_newreal(atof(buf));
- fz_arraypush(csi->array, obj);
- fz_dropobj(obj);
+ pdf_gstate *gstate = csi->gstate + csi->gtop;
+ pdf_showspace(csi, -atof(buf) * gstate->size * 0.001f);
}
else if (tok == PDF_TSTRING)
{
- obj = fz_newstring(buf, len);
- fz_arraypush(csi->array, obj);
- fz_dropobj(obj);
+ pdf_showstring(csi, (unsigned char *)buf, len);
}
else if (tok == PDF_TEOF)
{
@@ -1812,9 +1823,18 @@ pdf_runcsifile(pdf_csi *csi, fz_obj *rdb, fz_stream *file, char *buf, int buflen
case PDF_TEOF:
return fz_okay;
- /* TODO: optimize text-object array parsing */
case PDF_TOARRAY:
- csi->array = fz_newarray(8);
+ if (!csi->intext)
+ {
+ error = pdf_parsearray(&csi->obj, csi->xref, file, buf, buflen);
+ if (error)
+ return fz_rethrow(error, "cannot parse array");
+ csi->top ++;
+ }
+ else
+ {
+ csi->inarray = 1;
+ }
break;
case PDF_TODICT:
@@ -1842,25 +1862,15 @@ pdf_runcsifile(pdf_csi *csi, fz_obj *rdb, fz_stream *file, char *buf, int buflen
break;
case PDF_TSTRING:
- csi->obj = fz_newstring(buf, len);
- csi->top ++;
- break;
-
- case PDF_TTRUE:
- csi->istack[csi->top] = 1;
- csi->stack[csi->top] = 1;
- csi->top ++;
- break;
-
- case PDF_TFALSE:
- csi->istack[csi->top] = 0;
- csi->stack[csi->top] = 0;
- csi->top ++;
- break;
-
- case PDF_TNULL:
- csi->istack[csi->top] = 0;
- csi->stack[csi->top] = 0;
+ if (len <= sizeof(csi->string))
+ {
+ memcpy(csi->string, buf, len);
+ csi->stringlen = len;
+ }
+ else
+ {
+ csi->obj = fz_newstring(buf, len);
+ }
csi->top ++;
break;
generated by cgit v1.2.3 (git 2.25.1) at 2025-04-01 13:18:59 +0000