Merge pull request #193 from chrox/master

update mupdf and k2pdfopt to latest version

Makefile

@@ -107,7 +107,6 @@ $(MUPDF_LIB): $(JPEG_LIB) $(FREETYPE_LIB)
 		$(CFLAGS) \
 		-Wl,-E -Wl,-rpath,'$$ORIGIN' \
 		-Wl,--whole-archive $(MUPDF_LIB_STATIC) \
-		-Wl,--whole-archive $(MUPDF_JS_LIB_STATIC) \
 		-Wl,--no-whole-archive $(MUPDF_THIRDPARTY_LIBS) \
 		-Wl,-soname=$(notdir $(MUPDF_LIB)) \
 		$(JPEG_LIB) $(FREETYPE_LIB) \

ffi/koptcontext_h.lua

@@ -229,7 +229,9 @@ PIX * pixBlendBackgroundToColor(PIX *pixd, PIX *pixs, BOX *box, l_uint32 color,
 /* K2pdfopt */
 void bmp_init(WILLUSBITMAP *bmap);
 void bmp_free(WILLUSBITMAP *bmap);
+int  bmp_alloc(WILLUSBITMAP *bmap);
 int  bmp_copy(WILLUSBITMAP *dest,WILLUSBITMAP *src);
+unsigned char *bmp_rowptr_from_top(WILLUSBITMAP *bmp,int row);
 void wrectmaps_init(WRECTMAPS *wrectmaps);
 void wrectmaps_free(WRECTMAPS *wrectmaps);
 int  wrectmap_inside(WRECTMAP *wrmap,int xc,int yc);
@@ -242,7 +244,6 @@ void k2pdfopt_tocr_single_word(WILLUSBITMAP *src,
         char *word, int max_length,
         char *datadir, char *lang, int ocr_type,
         int allow_spaces, int std_proc);
-int bmpmupdf_pdffile_to_bmp(WILLUSBITMAP *bmp,char *filename,int pageno,double dpi,int bpp);
 void k2pdfopt_reflow_bmp(KOPTContext *kctx);
 void k2pdfopt_tocr_end();
 void pageregions_init(PAGEREGIONS *regions);

ffi/mupdf_h.lua

@@ -1,8 +1,30 @@
 local ffi = require("ffi")
 ffi.cdef[[
 typedef struct fz_alloc_context_s fz_alloc_context;
+typedef struct fz_locks_context_s fz_locks_context;
 typedef struct fz_colorspace_s fz_colorspace;
+typedef struct fz_document_s fz_document;
 typedef struct fz_context_s fz_context;
+typedef struct fz_matrix_s fz_matrix;
+typedef struct fz_cookie_s fz_cookie;
+typedef struct fz_pixmap_s fz_pixmap;
+typedef struct fz_device_s fz_device;
+typedef struct fz_irect_s fz_irect;
+typedef struct fz_rect_s fz_rect;
+typedef struct fz_page_s fz_page;
+
+struct fz_matrix_s {
+  float a, b, c, d, e, f;
+};
+struct fz_rect_s {
+  float x0, y0;
+  float x1, y1;
+};
+struct fz_irect_s {
+  int x0, y0;
+  int x1, y1;
+};
+
 typedef void fz_store_free_fn(struct fz_context_s *, struct fz_storable_s *);
 struct fz_storable_s {
   int refs;
@@ -23,12 +45,32 @@ struct fz_pixmap_s {
   unsigned char *samples;
   int free_samples;
 };
-typedef struct fz_pixmap_s fz_pixmap;
-struct fz_context_s *fz_new_context_imp(struct fz_alloc_context_s *, struct fz_locks_context_s *, unsigned int, const char *);
-struct fz_pixmap_s *fz_new_pixmap(struct fz_context_s *, struct fz_colorspace_s *, int, int);
-void fz_convert_pixmap(struct fz_context_s *, struct fz_pixmap_s *, struct fz_pixmap_s *);
-void fz_drop_pixmap(struct fz_context_s *, struct fz_pixmap_s *);
-struct fz_colorspace_s *fz_device_gray(struct fz_context_s *);
-void fz_free_context(struct fz_context_s *);
-struct fz_pixmap_s *fz_load_png(struct fz_context_s *, unsigned char *, int);
+
+fz_context *fz_new_context_imp(fz_alloc_context*, fz_locks_context*, unsigned int, const char *);
+void fz_register_document_handlers(fz_context *ctx);
+fz_colorspace *fz_device_gray(fz_context*);
+fz_colorspace *fz_device_rgb(fz_context *ctx);
+fz_document *fz_open_document(fz_context *ctx, const char *filename);
+int fz_count_pages(fz_document *doc);
+fz_page *fz_load_page(fz_document *doc, int number);
+fz_rect *fz_bound_page(fz_document *doc, fz_page *page, fz_rect *rect);
+fz_matrix *fz_scale(fz_matrix *m, float sx, float sy);
+fz_rect *fz_transform_rect(fz_rect *restrict rect, const fz_matrix *restrict transform);
+fz_irect *fz_round_rect(fz_irect *restrict bbox, const fz_rect *restrict rect);
+void fz_clear_pixmap_with_value(fz_context *ctx, fz_pixmap *pix, int value);
+void fz_run_page(fz_document *doc, fz_page *page, fz_device *dev, const fz_matrix *transform, fz_cookie *cookie);
+fz_pixmap *fz_new_pixmap_with_bbox(fz_context *ctx, fz_colorspace *colorspace, const fz_irect *bbox);
+fz_device *fz_new_draw_device(fz_context *ctx, fz_pixmap *dest);
+void fz_free_device(fz_device *dev);
+fz_pixmap *fz_new_pixmap(fz_context*, fz_colorspace *, int, int);
+void fz_convert_pixmap(fz_context*, fz_pixmap*, fz_pixmap*);
+int fz_pixmap_width(fz_context *ctx, fz_pixmap *pix);
+int fz_pixmap_height(fz_context *ctx, fz_pixmap *pix);
+int fz_pixmap_components(fz_context *ctx, fz_pixmap *pix);
+unsigned char *fz_pixmap_samples(fz_context *ctx, fz_pixmap *pix);
+void fz_drop_pixmap(fz_context*, fz_pixmap*);
+void fz_free_page(fz_document *doc, fz_page *page);
+void fz_free_context(fz_context*);
+void fz_close_document(fz_document *doc);
+fz_pixmap *fz_load_png(fz_context*, unsigned char *, int);
 ]]

ffi/mupdfbmp.lua

@@ -0,0 +1,72 @@
+--[[
+converting pdf page to willus bmp
+--]]
+
+local ffi = require("ffi")
+local dummy = require("ffi/mupdf_h")
+local dummy = require("ffi/koptcontext_h")
+local mupdf = ffi.load("libs/libmupdf.so")
+local k2pdfopt = ffi.load("libs/libk2pdfopt.so.2")
+
+local Bitmap = {}
+
+function Bitmap:pixmap_to_bmp(bmp, ctx, pixmap)
+    bmp.width = mupdf.fz_pixmap_width(ctx, pixmap)
+    bmp.height = mupdf.fz_pixmap_height(ctx, pixmap)
+    local ncomp = mupdf.fz_pixmap_components(ctx, pixmap)
+    if ncomp ~= 2 and ncomp ~= 4 then return -1 end
+    bmp.bpp = ncomp == 2 and 8 or 24
+    k2pdfopt.bmp_alloc(bmp)
+    if ncomp == 2 then
+        for i=0, 255 do
+            bmp.red[i] = i
+            bmp.green[i] = i
+            bmp.blue[i] = i
+        end
+    end
+    local p = mupdf.fz_pixmap_samples(ctx, pixmap)
+    if ncomp == 2 then
+        for row=0, bmp.height - 1 do
+            local dest = k2pdfopt.bmp_rowptr_from_top(bmp, row)
+            for col=0, bmp.width - 1 do
+                dest[0] = p[0]
+                dest = dest + 1
+                p = p + 2
+            end
+        end
+    end
+    return 0
+end
+
+function Bitmap:pdffile_to_bmp(bmp, filename, pageno, dpi, bpp)
+	local ctx = mupdf.fz_new_context_imp(nil, nil, bit.lshift(8, 20), "1.4")
+    if ctx == nil then return -1 end
+    mupdf.fz_register_document_handlers(ctx)
+    local colorspace = bpp == 8 and mupdf.fz_device_gray(ctx) or mupdf.fz_device_rgb(ctx)
+    local doc = mupdf.fz_open_document(ctx, filename)
+    if mupdf.fz_count_pages(doc) < pageno or pageno < 1 then
+        return -1
+    end
+    local page = mupdf.fz_load_page(doc, pageno-1)
+    local ctm = ffi.new("fz_matrix[1]")[0]
+    local bbox = ffi.new("fz_irect[1]")[0]
+    local bounds = ffi.new("fz_rect[1]")[0]
+    mupdf.fz_bound_page(doc, page, bounds)
+    mupdf.fz_scale(ctm, dpi/72, dpi/72)
+    mupdf.fz_transform_rect(bounds, ctm)
+    mupdf.fz_round_rect(bbox, bounds)
+    local pix = mupdf.fz_new_pixmap_with_bbox(ctx, colorspace, bbox)
+    mupdf.fz_clear_pixmap_with_value(ctx, pix, 0xFF)
+    local dev = mupdf.fz_new_draw_device(ctx, pix)
+    mupdf.fz_run_page(doc, page, dev, ctm, nil)
+    mupdf.fz_free_device(dev)
+    if self:pixmap_to_bmp(bmp, ctx, pix) ~= 0 then return -1 end
+    mupdf.fz_drop_pixmap(ctx, pix)
+    mupdf.fz_free_page(doc, page)
+    mupdf.fz_close_document(doc)
+    mupdf.fz_free_context(ctx)
+
+    return 0
+end
+
+return Bitmap

ffi/mupdfimg.lua

@@ -11,7 +11,7 @@ require("ffi/mupdf_h")
 local Image = {}
 
 function Image:initContext(cache_size)
-	self.context = mupdf.fz_new_context_imp(nil, nil, cache_size or bit.lshift(8, 20), "1.3")
+	self.context = mupdf.fz_new_context_imp(nil, nil, cache_size or bit.lshift(8, 20), "1.4")
 end
 
 function Image:_getFileData(filename)
@@ -29,7 +29,7 @@ function Image:toBlitBuffer()
 	local pixmap = ffi.new("fz_pixmap[1]")
 	pixmap = self.pixmap
 	if self.pixmap.n ~= 2 then
-		self.pixmap = mupdf.fz_new_pixmap(self.context, mupdf.fz_device_gray(self.context), 
+		self.pixmap = mupdf.fz_new_pixmap(self.context, mupdf.fz_device_gray(self.context),
 										  pixmap.w, pixmap.h);
 		mupdf.fz_convert_pixmap(self.context, self.pixmap, pixmap);
 		mupdf.fz_drop_pixmap(self.context, pixmap);

mupdf.patch

@@ -1,48 +1,48 @@
 diff --git a/source/pdf/pdf-font.c b/source/pdf/pdf-font.c
-index f79a076..4ecb159 100644
+index 8454e92..0d3bfea 100644
 --- a/source/pdf/pdf-font.c
 +++ b/source/pdf/pdf-font.c
-@@ -185,7 +185,12 @@ pdf_load_builtin_font(fz_context *ctx, pdf_font_desc *fontdesc, char *fontname)
- 	if (!data)
- 		fz_throw(ctx, FZ_ERROR_GENERIC, "cannot find builtin font: '%s'", fontname);
+@@ -188,7 +188,12 @@ pdf_load_builtin_font(fz_context *ctx, pdf_font_desc *fontdesc, char *fontname,
+ 		if (!data)
+ 			fz_throw(ctx, FZ_ERROR_GENERIC, "cannot find builtin font: '%s'", fontname);
 
 +#ifndef NOBUILTINFONT
- 	fontdesc->font = fz_new_font_from_memory(ctx, fontname, data, len, 0, 1);
+ 		fontdesc->font = fz_new_font_from_memory(ctx, fontname, data, len, 0, 1);
 +#else
-+	fontdesc->font = fz_new_font_from_file(ctx, fontname, data, 0, 1);
-+	free(data);
++		fontdesc->font = fz_new_font_from_file(ctx, fontname, data, 0, 1);
++		free(data);
 +#endif
+ 	}
 
- 	if (!strcmp(fontname, "Symbol") || !strcmp(fontname, "ZapfDingbats"))
- 		fontdesc->flags |= PDF_FD_SYMBOLIC;
-@@ -205,7 +210,12 @@ pdf_load_substitute_font(fz_context *ctx, pdf_font_desc *fontdesc, char *fontnam
- 	if (!data)
- 		fz_throw(ctx, FZ_ERROR_GENERIC, "cannot find substitute font");
+ 	if (!strcmp(clean_name, "Symbol") || !strcmp(clean_name, "ZapfDingbats"))
+@@ -212,7 +217,12 @@ pdf_load_substitute_font(fz_context *ctx, pdf_font_desc *fontdesc, char *fontnam
+ 		if (!data)
+ 			fz_throw(ctx, FZ_ERROR_GENERIC, "cannot find substitute font");
 
 +#ifndef NOBUILTINFONT
- 	fontdesc->font = fz_new_font_from_memory(ctx, fontname, data, len, 0, 1);
+ 		fontdesc->font = fz_new_font_from_memory(ctx, fontname, data, len, 0, 1);
 +#else
-+	fontdesc->font = fz_new_font_from_file(ctx, fontname, data, 0, 1);
-+	free(data);
++		fontdesc->font = fz_new_font_from_file(ctx, fontname, data, 0, 1);
++		free(data);
 +#endif
+ 		fontdesc->font->ft_bold = bold && !ft_is_bold(fontdesc->font->ft_face);
+ 		fontdesc->font->ft_italic = italic && !ft_is_italic(fontdesc->font->ft_face);
+ 	}
+@@ -234,7 +244,12 @@ pdf_load_substitute_cjk_font(fz_context *ctx, pdf_font_desc *fontdesc, char *fon
+ 			fz_throw(ctx, FZ_ERROR_GENERIC, "cannot find builtin CJK font");
 
- 	fontdesc->font->ft_substitute = 1;
- 	fontdesc->font->ft_bold = bold && !ft_is_bold(fontdesc->font->ft_face);
-@@ -223,7 +233,12 @@ pdf_load_substitute_cjk_font(fz_context *ctx, pdf_font_desc *fontdesc, char *fon
- 		fz_throw(ctx, FZ_ERROR_GENERIC, "cannot find builtin CJK font");
-
- 	/* a glyph bbox cache is too big for droid sans fallback (51k glyphs!) */
+ 		/* a glyph bbox cache is too big for droid sans fallback (51k glyphs!) */
 +#ifndef NOBUILTINFONT
- 	fontdesc->font = fz_new_font_from_memory(ctx, fontname, data, len, 0, 0);
+ 		fontdesc->font = fz_new_font_from_memory(ctx, fontname, data, len, 0, 0);
 +#else
-+	fontdesc->font = fz_new_font_from_file(ctx, fontname, data, 0, 1);
-+	free(data);
++		fontdesc->font = fz_new_font_from_file(ctx, fontname, data, 0, 1);
++		free(data);
 +#endif
+ 	}
 
  	fontdesc->font->ft_substitute = 1;
- }
 diff --git a/source/pdf/pdf-fontfile.c b/source/pdf/pdf-fontfile.c
-index c9990da..9c091a7 100644
+index 2e33768..a5ab5e0 100644
 --- a/source/pdf/pdf-fontfile.c
 +++ b/source/pdf/pdf-fontfile.c
 @@ -14,6 +14,8 @@
@@ -81,50 +81,50 @@ index c9990da..9c091a7 100644
 +}
 +
 +unsigned char *
-+pdf_lookup_builtin_font(char *name, unsigned int *len)
++pdf_lookup_builtin_font(const char *name, unsigned int *len)
 +{
 +	*len = 0;
 +	if (!strcmp("Courier", name)) {
-+		return get_font_file("NimbusMonL-Regu.cff");
++		return get_font_file("urw/NimbusMon-Reg.cff");
 +	}
 +	if (!strcmp("Courier-Bold", name)) {
-+		return get_font_file("NimbusMonL-Bold.cff");
++		return get_font_file("urw/NimbusMon-Bol.cff");
 +	}
 +	if (!strcmp("Courier-Oblique", name)) {
-+		return get_font_file("NimbusMonL-ReguObli.cff");
++		return get_font_file("urw/NimbusMon-Obl.cff");
 +	}
 +	if (!strcmp("Courier-BoldOblique", name)) {
-+		return get_font_file("NimbusMonL-BoldObli.cff");
++		return get_font_file("urw/NimbusMon-BolObl.cff");
 +	}
 +	if (!strcmp("Helvetica", name)) {
-+		return get_font_file("NimbusSanL-Regu.cff");
++		return get_font_file("urw/NimbusSan-Reg.cff");
 +	}
 +	if (!strcmp("Helvetica-Bold", name)) {
-+		return get_font_file("NimbusSanL-Bold.cff");
++		return get_font_file("urw/NimbusSan-Bol.cff");
 +	}
 +	if (!strcmp("Helvetica-Oblique", name)) {
-+		return get_font_file("NimbusSanL-ReguItal.cff");
++		return get_font_file("urw/NimbusSan-Ita.cff");
 +	}
 +	if (!strcmp("Helvetica-BoldOblique", name)) {
-+		return get_font_file("NimbusSanL-BoldItal.cff");
++		return get_font_file("urw/NimbusSan-BolIta.cff");
 +	}
 +	if (!strcmp("Times-Roman", name)) {
-+		return get_font_file("NimbusRomNo9L-Regu.cff");
++		return get_font_file("urw/NimbusRom-Reg.cff");
 +	}
 +	if (!strcmp("Times-Bold", name)) {
-+		return get_font_file("NimbusRomNo9L-Medi.cff");
++		return get_font_file("urw/NimbusRom-Med.cff");
 +	}
 +	if (!strcmp("Times-Italic", name)) {
-+		return get_font_file("NimbusRomNo9L-ReguItal.cff");
++		return get_font_file("urw/NimbusRom-Ita.cff");
 +	}
 +	if (!strcmp("Times-BoldItalic", name)) {
-+		return get_font_file("NimbusRomNo9L-MediItal.cff");
++		return get_font_file("urw/NimbusRom-MedIta.cff");
 +	}
 +	if (!strcmp("Symbol", name)) {
-+		return get_font_file("StandardSymL.cff");
++		return get_font_file("urw/StandardSymL.cff");
 +	}
 +	if (!strcmp("ZapfDingbats", name)) {
-+		return get_font_file("Dingbats.cff");
++		return get_font_file("urw/Dingbats.cff");
 +	}
 +	return NULL;
 +}

pdf.c

@@ -184,6 +184,8 @@ static int openDocument(lua_State *L) {
 
 	doc->context = fz_new_context(&my_alloc_default, NULL, cache_size);
 
+	fz_register_document_handlers(doc->context);
+
 	fz_try(doc->context) {
 		doc->xref = fz_open_document(doc->context, filename);
 	}