Split part of fz_document interface for pdf_document into separate file.

Make a separate constructor function that does not link in the
interpreter, so we can save space in the mubusy binary by not
including the font and cmap resources.

11 files changed, 163 insertions, 106 deletions

diff --git a/android/jni/Core.mk b/android/jni/Core.mk
index 9fe01db0..cea16e74 100644
--- a/android/jni/Core.mk
+++ b/android/jni/Core.mk
@@ -102,6 +102,7 @@ LOCAL_SRC_FILES := \
 	$(MY_ROOT)/pdf/pdf_unicode.c \
 	$(MY_ROOT)/pdf/pdf_xobject.c \
 	$(MY_ROOT)/pdf/pdf_xref.c \
+	$(MY_ROOT)/pdf/pdf_xref_aux.c \
 	$(MY_ROOT)/xps/xps_common.c \
 	$(MY_ROOT)/xps/xps_doc.c \
 	$(MY_ROOT)/xps/xps_glyphs.c \
diff --git a/apps/mupdfclean.c b/apps/mupdfclean.c
index a1064f57..a3711c72 100644
--- a/apps/mupdfclean.c
+++ b/apps/mupdfclean.c
@@ -200,7 +200,7 @@ int pdfclean_main(int argc, char **argv)
 		exit(1);
 	}
 
-	xref = pdf_open_document(ctx, infile);
+	xref = pdf_open_document_no_run(ctx, infile);
 	if (pdf_needs_password(xref))
 		if (!pdf_authenticate_password(xref, password))
 			fz_throw(ctx, "cannot authenticate password: %s", infile);
diff --git a/apps/mupdfextract.c b/apps/mupdfextract.c
index a46e0d88..d1131f67 100644
--- a/apps/mupdfextract.c
+++ b/apps/mupdfextract.c
@@ -3,6 +3,7 @@
  */
 
 #include "mupdf.h"
+#include "mupdf-internal.h"
 
 static pdf_document *doc = NULL;
 static fz_context *ctx = NULL;
@@ -170,7 +171,7 @@ int pdfextract_main(int argc, char **argv)
 		exit(1);
 	}
 
-	doc = pdf_open_document(ctx, infile);
+	doc = pdf_open_document_no_run(ctx, infile);
 	if (pdf_needs_password(doc))
 		if (!pdf_authenticate_password(doc, password))
 			fz_throw(ctx, "cannot authenticate password: %s", infile);
diff --git a/apps/mupdfinfo.c b/apps/mupdfinfo.c
index 52018a9c..78838ee2 100644
--- a/apps/mupdfinfo.c
+++ b/apps/mupdfinfo.c
@@ -996,7 +996,7 @@ int pdfinfo_main(int argc, char **argv)
 
 			filename = argv[fz_optind];
 			printf("%s:\n", filename);
-			xref = pdf_open_document(ctx, filename);
+			xref = pdf_open_document_no_run(ctx, filename);
 			if (pdf_needs_password(xref))
 				if (!pdf_authenticate_password(xref, password))
 					fz_throw(ctx, "cannot authenticate password: %s", filename);
diff --git a/apps/mupdfposter.c b/apps/mupdfposter.c
index 415e3175..19198982 100644
--- a/apps/mupdfposter.c
+++ b/apps/mupdfposter.c
@@ -162,7 +162,7 @@ int pdfposter_main(int argc, char **argv)
 		exit(1);
 	}
 
-	xref = pdf_open_document(ctx, infile);
+	xref = pdf_open_document_no_run(ctx, infile);
 	if (pdf_needs_password(xref))
 		if (!pdf_authenticate_password(xref, password))
 			fz_throw(ctx, "cannot authenticate password: %s", infile);
diff --git a/apps/mupdfshow.c b/apps/mupdfshow.c
index 1c198e8d..7114f249 100644
--- a/apps/mupdfshow.c
+++ b/apps/mupdfshow.c
@@ -203,7 +203,7 @@ int pdfshow_main(int argc, char **argv)
 	fz_var(doc);
 	fz_try(ctx)
 	{
-		doc = pdf_open_document(ctx, filename);
+		doc = pdf_open_document_no_run(ctx, filename);
 		if (pdf_needs_password(doc))
 			if (!pdf_authenticate_password(doc, password))
 				fz_warn(ctx, "cannot authenticate password: %s", filename);
diff --git a/pdf/mupdf-internal.h b/pdf/mupdf-internal.h
index e947cbbd..bde094a1 100644
--- a/pdf/mupdf-internal.h
+++ b/pdf/mupdf-internal.h
@@ -191,6 +191,9 @@ struct pdf_document_s
 	pdf_lexbuf_large lexbuf;
 };
 
+pdf_document *pdf_open_document_no_run(fz_context *ctx, const char *filename);
+pdf_document *pdf_open_document_no_run_with_stream(fz_stream *file);
+
 void pdf_cache_object(pdf_document *doc, int num, int gen);
 
 fz_stream *pdf_open_inline_stream(pdf_document *doc, pdf_obj *stmobj, int length, fz_stream *chain, pdf_image_params *params);
diff --git a/pdf/mupdf.h b/pdf/mupdf.h
index f02d1828..762d1bc5 100644
--- a/pdf/mupdf.h
+++ b/pdf/mupdf.h
@@ -217,4 +217,9 @@ void pdf_run_page(pdf_document *doc, pdf_page *page, fz_device *dev, fz_matrix c
 
 void pdf_run_page_with_usage(pdf_document *doc, pdf_page *page, fz_device *dev, fz_matrix ctm, char *event, fz_cookie *cookie);
 
+/*
+	Metadata interface.
+*/
+int pdf_meta(pdf_document *doc, int key, void *ptr, int size);
+
 #endif
diff --git a/pdf/pdf_xref.c b/pdf/pdf_xref.c
index 3af2901c..cbc0c1ec 100644
--- a/pdf/pdf_xref.c
+++ b/pdf/pdf_xref.c
@@ -672,29 +672,19 @@ pdf_free_ocg(fz_context *ctx, pdf_ocg_descriptor *desc)
  * If password is not null, try to decrypt.
  */
 
-static void pdf_init_document(pdf_document *xref);
-
-pdf_document *
-pdf_open_document_with_stream(fz_stream *file)
+static void
+pdf_init_document(pdf_document *xref)
 {
-	pdf_document *xref;
+	fz_context *ctx = xref->ctx;
 	pdf_obj *encrypt, *id;
 	pdf_obj *dict = NULL;
 	pdf_obj *obj;
 	pdf_obj *nobj = NULL;
 	int i, repaired = 0;
-	fz_context *ctx = file->ctx;
 
 	fz_var(dict);
 	fz_var(nobj);
 
-	xref = fz_malloc_struct(ctx, pdf_document);
-	pdf_init_document(xref);
-	xref->lexbuf.base.size = PDF_LEXBUF_LARGE;
-
-	xref->file = fz_keep_stream(file);
-	xref->ctx = ctx;
-
 	fz_try(ctx)
 	{
 		pdf_load_xref(xref, &xref->lexbuf.base);
@@ -797,8 +787,6 @@ pdf_open_document_with_stream(fz_stream *file)
 	{
 		fz_warn(ctx, "Ignoring Broken Optional Content");
 	}
-
-	return xref;
 }
 
 void
@@ -1083,7 +1071,8 @@ pdf_resolve_indirect(pdf_obj *ref)
 	return ref;
 }
 
-int pdf_count_objects(pdf_document *doc)
+int
+pdf_count_objects(pdf_document *doc)
 {
 	return doc->len;
 }
@@ -1110,88 +1099,9 @@ pdf_update_object(pdf_document *xref, int num, int gen, pdf_obj *newobj)
 	x->ofs = 0;
 }
 
-/*
- * Convenience function to open a file then call pdf_open_document_with_stream.
- */
-
-pdf_document *
-pdf_open_document(fz_context *ctx, const char *filename)
-{
-	fz_stream *file = NULL;
-	pdf_document *xref;
-
-	fz_var(file);
-	fz_try(ctx)
-	{
-		file = fz_open_file(ctx, filename);
-		xref = pdf_open_document_with_stream(file);
-	}
-	fz_catch(ctx)
-	{
-		fz_close(file);
-		fz_throw(ctx, "cannot load document '%s'", filename);
-	}
-
-	fz_close(file);
-	return xref;
-}
-
-/* Document interface wrappers */
-
-static void pdf_close_document_shim(fz_document *doc)
-{
-	pdf_close_document((pdf_document*)doc);
-}
-
-static int pdf_needs_password_shim(fz_document *doc)
-{
-	return pdf_needs_password((pdf_document*)doc);
-}
-
-static int pdf_authenticate_password_shim(fz_document *doc, char *password)
-{
-	return pdf_authenticate_password((pdf_document*)doc, password);
-}
-
-static fz_outline *pdf_load_outline_shim(fz_document *doc)
-{
-	return pdf_load_outline((pdf_document*)doc);
-}
-
-static int pdf_count_pages_shim(fz_document *doc)
-{
-	return pdf_count_pages((pdf_document*)doc);
-}
-
-static fz_page *pdf_load_page_shim(fz_document *doc, int number)
-{
-	return (fz_page*) pdf_load_page((pdf_document*)doc, number);
-}
-
-static fz_link *pdf_load_links_shim(fz_document *doc, fz_page *page)
-{
-	return pdf_load_links((pdf_document*)doc, (pdf_page*)page);
-}
-
-static fz_rect pdf_bound_page_shim(fz_document *doc, fz_page *page)
-{
-	return pdf_bound_page((pdf_document*)doc, (pdf_page*)page);
-}
-
-static void pdf_run_page_shim(fz_document *doc, fz_page *page, fz_device *dev, fz_matrix transform, fz_cookie *cookie)
-{
-	pdf_run_page((pdf_document*)doc, (pdf_page*)page, dev, transform, cookie);
-}
-
-static void pdf_free_page_shim(fz_document *doc, fz_page *page)
-{
-	pdf_free_page((pdf_document*)doc, (pdf_page*)page);
-}
-
-static int pdf_meta(fz_document *doc_, int key, void *ptr, int size)
+int
+pdf_meta(pdf_document *doc, int key, void *ptr, int size)
 {
-	pdf_document *doc = (pdf_document *)doc_;
-
 	switch(key)
 	{
 	/*
@@ -1263,9 +1173,72 @@ static int pdf_meta(fz_document *doc_, int key, void *ptr, int size)
 	}
 }
 
-static void
-pdf_init_document(pdf_document *doc)
+/*
+	Wrappers to implement the fz_document interface for pdf_document.
+
+	The functions are split across two files to allow calls to a
+	version of the constructor that does not link in the interpreter.
+	The interpreter references the built-in font and cmap resources
+	which are quite big. Not linking those into the mubusy binary
+	saves roughly 6MB of space.
+*/
+
+static void pdf_close_document_shim(fz_document *doc)
+{
+	pdf_close_document((pdf_document*)doc);
+}
+
+static int pdf_needs_password_shim(fz_document *doc)
+{
+	return pdf_needs_password((pdf_document*)doc);
+}
+
+static int pdf_authenticate_password_shim(fz_document *doc, char *password)
+{
+	return pdf_authenticate_password((pdf_document*)doc, password);
+}
+
+static fz_outline *pdf_load_outline_shim(fz_document *doc)
+{
+	return pdf_load_outline((pdf_document*)doc);
+}
+
+static int pdf_count_pages_shim(fz_document *doc)
+{
+	return pdf_count_pages((pdf_document*)doc);
+}
+
+static fz_page *pdf_load_page_shim(fz_document *doc, int number)
+{
+	return (fz_page*) pdf_load_page((pdf_document*)doc, number);
+}
+
+static fz_link *pdf_load_links_shim(fz_document *doc, fz_page *page)
+{
+	return pdf_load_links((pdf_document*)doc, (pdf_page*)page);
+}
+
+static fz_rect pdf_bound_page_shim(fz_document *doc, fz_page *page)
 {
+	return pdf_bound_page((pdf_document*)doc, (pdf_page*)page);
+}
+
+static void pdf_free_page_shim(fz_document *doc, fz_page *page)
+{
+	pdf_free_page((pdf_document*)doc, (pdf_page*)page);
+}
+
+static int pdf_meta_shim(fz_document *doc, int key, void *ptr, int size)
+{
+	return pdf_meta((pdf_document*)doc, key, ptr, size);
+}
+
+static pdf_document *
+pdf_new_document(fz_stream *file)
+{
+	fz_context *ctx = file->ctx;
+	pdf_document *doc = fz_malloc_struct(ctx, pdf_document);
+
 	doc->super.close = pdf_close_document_shim;
 	doc->super.needs_password = pdf_needs_password_shim;
 	doc->super.authenticate_password = pdf_authenticate_password_shim;
@@ -1274,7 +1247,46 @@ pdf_init_document(pdf_document *doc)
 	doc->super.load_page = pdf_load_page_shim;
 	doc->super.load_links = pdf_load_links_shim;
 	doc->super.bound_page = pdf_bound_page_shim;
-	doc->super.run_page = pdf_run_page_shim;
+	doc->super.run_page = NULL; /* see pdf_xref_aux.c */
 	doc->super.free_page = pdf_free_page_shim;
-	doc->super.meta = pdf_meta;
+	doc->super.meta = pdf_meta_shim;
+
+	doc->lexbuf.base.size = PDF_LEXBUF_LARGE;
+	doc->file = fz_keep_stream(file);
+	doc->ctx = ctx;
+
+	return doc;
+}
+
+pdf_document *
+pdf_open_document_no_run_with_stream(fz_stream *file)
+{
+	pdf_document *doc = pdf_new_document(file);
+	pdf_init_document(doc);
+	return doc;
+}
+
+pdf_document *
+pdf_open_document_no_run(fz_context *ctx, const char *filename)
+{
+	fz_stream *file = NULL;
+	pdf_document *doc;
+
+	fz_var(file);
+
+	fz_try(ctx)
+	{
+		file = fz_open_file(ctx, filename);
+		doc = pdf_new_document(file);
+		pdf_init_document(doc);
+	}
+	fz_always(ctx)
+	{
+		fz_close(file);
+	}
+	fz_catch(ctx)
+	{
+		fz_throw(ctx, "cannot load document '%s'", filename);
+	}
+	return doc;
 }
diff --git a/pdf/pdf_xref_aux.c b/pdf/pdf_xref_aux.c
new file mode 100644
index 00000000..2d760334
--- /dev/null
+++ b/pdf/pdf_xref_aux.c
@@ -0,0 +1,31 @@
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
+
+/*
+	These functions have been split out of pdf_xref.c to allow tools
+	to be linked without pulling in the interpreter.  The interpreter
+	references the built-in font and cmap resources which are quite
+	big. Not linking those into the tools saves roughly 6MB in the
+	resulting executables.
+*/
+
+static void pdf_run_page_shim(fz_document *doc, fz_page *page, fz_device *dev, fz_matrix transform, fz_cookie *cookie)
+{
+	pdf_run_page((pdf_document*)doc, (pdf_page*)page, dev, transform, cookie);
+}
+
+pdf_document *
+pdf_open_document_with_stream(fz_stream *file)
+{
+	pdf_document *doc = pdf_open_document_no_run_with_stream(file);
+	doc->super.run_page = pdf_run_page_shim;
+	return doc;
+}
+
+pdf_document *
+pdf_open_document(fz_context *ctx, const char *filename)
+{
+	pdf_document *doc = pdf_open_document_no_run(ctx, filename);
+	doc->super.run_page = pdf_run_page_shim;
+	return doc;
+}
diff --git a/win32/libmupdf.vcproj b/win32/libmupdf.vcproj
index 03b1bbd1..8706176c 100644
--- a/win32/libmupdf.vcproj
+++ b/win32/libmupdf.vcproj
@@ -345,6 +345,10 @@
 				RelativePath="..\pdf\pdf_xref.c"
 				>
 			</File>
+			<File
+				RelativePath="..\pdf\pdf_xref_aux.c"
+				>
+			</File>
 		</Filter>
 		<Filter
 			Name="fitz"