31 files changed, 620 insertions, 607 deletions

diff --git a/pdf/base_object.c b/pdf/base_object.c
index 94f1acfc..1dc07f1a 100644
--- a/pdf/base_object.c
+++ b/pdf/base_object.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 typedef enum pdf_objkind_e
 {
diff --git a/pdf/mupdf-internal.h b/pdf/mupdf-internal.h
new file mode 100644
index 00000000..faec16b5
--- /dev/null
+++ b/pdf/mupdf-internal.h
@@ -0,0 +1,535 @@
+#ifndef MUPDF_INTERNAL_H
+#define MUPDF_INTERNAL_H
+
+#include "mupdf.h"
+#include "fitz-internal.h"
+
+void pdf_set_str_len(pdf_obj *obj, int newlen);
+void *pdf_get_indirect_document(pdf_obj *obj);
+
+/*
+ * PDF Images
+ */
+
+typedef struct pdf_image_params_s pdf_image_params;
+
+struct pdf_image_params_s
+{
+	int type;
+	fz_colorspace *colorspace;
+	union
+	{
+		struct
+		{
+			int columns;
+			int rows;
+			int k;
+			int eol;
+			int eba;
+			int eob;
+			int bi1;
+		}
+		fax;
+		struct
+		{
+			int ct;
+		}
+		jpeg;
+		struct
+		{
+			int columns;
+			int colors;
+			int predictor;
+			int bpc;
+		}
+		flate;
+		struct
+		{
+			int columns;
+			int colors;
+			int predictor;
+			int bpc;
+			int ec;
+		}
+		lzw;
+	}
+	u;
+};
+
+
+typedef struct pdf_image_s pdf_image;
+
+struct pdf_image_s
+{
+	fz_image base;
+	fz_pixmap *tile;
+	int n, bpc;
+	pdf_image_params params;
+	fz_buffer *buffer;
+	int colorkey[FZ_MAX_COLORS * 2];
+	float decode[FZ_MAX_COLORS * 2];
+	int imagemask;
+	int interpolate;
+	int usecolorkey;
+};
+
+enum
+{
+	PDF_IMAGE_RAW,
+	PDF_IMAGE_FAX,
+	PDF_IMAGE_JPEG,
+	PDF_IMAGE_RLD,
+	PDF_IMAGE_FLATE,
+	PDF_IMAGE_LZW,
+	PDF_IMAGE_JPX
+};
+
+/*
+ * tokenizer and low-level object parser
+ */
+
+enum
+{
+	PDF_TOK_ERROR, PDF_TOK_EOF,
+	PDF_TOK_OPEN_ARRAY, PDF_TOK_CLOSE_ARRAY,
+	PDF_TOK_OPEN_DICT, PDF_TOK_CLOSE_DICT,
+	PDF_TOK_OPEN_BRACE, PDF_TOK_CLOSE_BRACE,
+	PDF_TOK_NAME, PDF_TOK_INT, PDF_TOK_REAL, PDF_TOK_STRING, PDF_TOK_KEYWORD,
+	PDF_TOK_R, PDF_TOK_TRUE, PDF_TOK_FALSE, PDF_TOK_NULL,
+	PDF_TOK_OBJ, PDF_TOK_ENDOBJ,
+	PDF_TOK_STREAM, PDF_TOK_ENDSTREAM,
+	PDF_TOK_XREF, PDF_TOK_TRAILER, PDF_TOK_STARTXREF,
+	PDF_NUM_TOKENS
+};
+
+enum
+{
+	PDF_LEXBUF_SMALL = 256,
+	PDF_LEXBUF_LARGE = 65536
+};
+
+typedef struct pdf_lexbuf_s pdf_lexbuf;
+typedef struct pdf_lexbuf_large_s pdf_lexbuf_large;
+
+struct pdf_lexbuf_s
+{
+	int size;
+	int len;
+	int i;
+	float f;
+	char scratch[PDF_LEXBUF_SMALL];
+};
+
+struct pdf_lexbuf_large_s
+{
+	pdf_lexbuf base;
+	char scratch[PDF_LEXBUF_LARGE - PDF_LEXBUF_SMALL];
+};
+
+int pdf_lex(fz_stream *f, pdf_lexbuf *lexbuf);
+
+pdf_obj *pdf_parse_array(pdf_document *doc, fz_stream *f, pdf_lexbuf *buf);
+pdf_obj *pdf_parse_dict(pdf_document *doc, fz_stream *f, pdf_lexbuf *buf);
+pdf_obj *pdf_parse_stm_obj(pdf_document *doc, fz_stream *f, pdf_lexbuf *buf);
+pdf_obj *pdf_parse_ind_obj(pdf_document *doc, fz_stream *f, pdf_lexbuf *buf, int *num, int *gen, int *stm_ofs);
+
+/*
+ * xref and object / stream api
+ */
+
+typedef struct pdf_xref_entry_s pdf_xref_entry;
+
+struct pdf_xref_entry_s
+{
+	int ofs;	/* file offset / objstm object number */
+	int gen;	/* generation / objstm index */
+	int stm_ofs;	/* on-disk stream */
+	pdf_obj *obj;	/* stored/cached object */
+	int type;	/* 0=unset (f)ree i(n)use (o)bjstm */
+};
+
+typedef struct pdf_crypt_s pdf_crypt;
+typedef struct pdf_ocg_descriptor_s pdf_ocg_descriptor;
+typedef struct pdf_ocg_entry_s pdf_ocg_entry;
+
+struct pdf_ocg_entry_s
+{
+	int num;
+	int gen;
+	int state;
+};
+
+struct pdf_ocg_descriptor_s
+{
+	int len;
+	pdf_ocg_entry *ocgs;
+	pdf_obj *intent;
+};
+
+struct pdf_document_s
+{
+	fz_document super;
+
+	fz_context *ctx;
+	fz_stream *file;
+
+	int version;
+	int startxref;
+	int file_size;
+	pdf_crypt *crypt;
+	pdf_obj *trailer;
+	pdf_ocg_descriptor *ocg;
+
+	int len;
+	pdf_xref_entry *table;
+
+	int page_len;
+	int page_cap;
+	pdf_obj **page_objs;
+	pdf_obj **page_refs;
+
+	pdf_lexbuf_large lexbuf;
+};
+
+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);
+fz_buffer *pdf_load_image_stream(pdf_document *doc, int num, int gen, pdf_image_params *params);
+fz_stream *pdf_open_image_stream(pdf_document *doc, int num, int gen, pdf_image_params *params);
+fz_stream *pdf_open_stream_with_offset(pdf_document *doc, int num, int gen, pdf_obj *dict, int stm_ofs);
+fz_stream *pdf_open_image_decomp_stream(fz_context *ctx, fz_buffer *, pdf_image_params *params, int *factor);
+
+void pdf_repair_xref(pdf_document *doc, pdf_lexbuf *buf);
+void pdf_repair_obj_stms(pdf_document *doc);
+void pdf_debug_xref(pdf_document *);
+void pdf_resize_xref(pdf_document *doc, int newcap);
+
+/*
+ * Encryption
+ */
+
+pdf_crypt *pdf_new_crypt(fz_context *ctx, pdf_obj *enc, pdf_obj *id);
+void pdf_free_crypt(fz_context *ctx, pdf_crypt *crypt);
+
+void pdf_crypt_obj(fz_context *ctx, pdf_crypt *crypt, pdf_obj *obj, int num, int gen);
+fz_stream *pdf_open_crypt(fz_stream *chain, pdf_crypt *crypt, int num, int gen);
+fz_stream *pdf_open_crypt_with_filter(fz_stream *chain, pdf_crypt *crypt, char *name, int num, int gen);
+
+int pdf_get_crypt_revision(pdf_document *doc);
+char *pdf_get_crypt_method(pdf_document *doc);
+int pdf_get_crypt_length(pdf_document *doc);
+unsigned char *pdf_get_crypt_key(pdf_document *doc);
+
+void pdf_debug_crypt(pdf_crypt *crypt);
+
+/*
+ * Functions, Colorspaces, Shadings and Images
+ */
+
+typedef struct pdf_function_s pdf_function;
+
+pdf_function *pdf_load_function(pdf_document *doc, pdf_obj *ref);
+void pdf_eval_function(fz_context *ctx, pdf_function *func, float *in, int inlen, float *out, int outlen);
+pdf_function *pdf_keep_function(fz_context *ctx, pdf_function *func);
+void pdf_drop_function(fz_context *ctx, pdf_function *func);
+unsigned int pdf_function_size(pdf_function *func);
+
+fz_colorspace *pdf_load_colorspace(pdf_document *doc, pdf_obj *obj);
+fz_pixmap *pdf_expand_indexed_pixmap(fz_context *ctx, fz_pixmap *src);
+
+fz_shade *pdf_load_shading(pdf_document *doc, pdf_obj *obj);
+
+fz_image *pdf_load_inline_image(pdf_document *doc, pdf_obj *rdb, pdf_obj *dict, fz_stream *file);
+int pdf_is_jpx_image(fz_context *ctx, pdf_obj *dict);
+
+/*
+ * Pattern
+ */
+
+typedef struct pdf_pattern_s pdf_pattern;
+
+struct pdf_pattern_s
+{
+	fz_storable storable;
+	int ismask;
+	float xstep;
+	float ystep;
+	fz_matrix matrix;
+	fz_rect bbox;
+	pdf_obj *resources;
+	fz_buffer *contents;
+};
+
+pdf_pattern *pdf_load_pattern(pdf_document *doc, pdf_obj *obj);
+pdf_pattern *pdf_keep_pattern(fz_context *ctx, pdf_pattern *pat);
+void pdf_drop_pattern(fz_context *ctx, pdf_pattern *pat);
+
+/*
+ * XObject
+ */
+
+typedef struct pdf_xobject_s pdf_xobject;
+
+struct pdf_xobject_s
+{
+	fz_storable storable;
+	fz_matrix matrix;
+	fz_rect bbox;
+	int isolated;
+	int knockout;
+	int transparency;
+	fz_colorspace *colorspace;
+	pdf_obj *resources;
+	fz_buffer *contents;
+	pdf_obj *me;
+};
+
+pdf_xobject *pdf_load_xobject(pdf_document *doc, pdf_obj *obj);
+pdf_xobject *pdf_keep_xobject(fz_context *ctx, pdf_xobject *xobj);
+void pdf_drop_xobject(fz_context *ctx, pdf_xobject *xobj);
+
+/*
+ * CMap
+ */
+
+typedef struct pdf_cmap_s pdf_cmap;
+typedef struct pdf_range_s pdf_range;
+
+enum { PDF_CMAP_SINGLE, PDF_CMAP_RANGE, PDF_CMAP_TABLE, PDF_CMAP_MULTI };
+
+struct pdf_range_s
+{
+	unsigned short low;
+	/* Next, we pack 2 fields into the same unsigned short. Top 14 bits
+	 * are the extent, bottom 2 bits are flags: single, range, table,
+	 * multi */
+	unsigned short extent_flags;
+	unsigned short offset;	/* range-delta or table-index */
+};
+
+struct pdf_cmap_s
+{
+	fz_storable storable;
+	char cmap_name[32];
+
+	char usecmap_name[32];
+	pdf_cmap *usecmap;
+
+	int wmode;
+
+	int codespace_len;
+	struct
+	{
+		unsigned short n;
+		unsigned short low;
+		unsigned short high;
+	} codespace[40];
+
+	int rlen, rcap;
+	pdf_range *ranges;
+
+	int tlen, tcap;
+	unsigned short *table;
+};
+
+pdf_cmap *pdf_new_cmap(fz_context *ctx);
+pdf_cmap *pdf_keep_cmap(fz_context *ctx, pdf_cmap *cmap);
+void pdf_drop_cmap(fz_context *ctx, pdf_cmap *cmap);
+void pdf_free_cmap_imp(fz_context *ctx, fz_storable *cmap);
+unsigned int pdf_cmap_size(fz_context *ctx, pdf_cmap *cmap);
+
+void pdf_debug_cmap(fz_context *ctx, pdf_cmap *cmap);
+int pdf_get_wmode(fz_context *ctx, pdf_cmap *cmap);
+void pdf_set_wmode(fz_context *ctx, pdf_cmap *cmap, int wmode);
+void pdf_set_usecmap(fz_context *ctx, pdf_cmap *cmap, pdf_cmap *usecmap);
+
+void pdf_add_codespace(fz_context *ctx, pdf_cmap *cmap, int low, int high, int n);
+void pdf_map_range_to_table(fz_context *ctx, pdf_cmap *cmap, int low, int *map, int len);
+void pdf_map_range_to_range(fz_context *ctx, pdf_cmap *cmap, int srclo, int srchi, int dstlo);
+void pdf_map_one_to_many(fz_context *ctx, pdf_cmap *cmap, int one, int *many, int len);
+void pdf_sort_cmap(fz_context *ctx, pdf_cmap *cmap);
+
+int pdf_lookup_cmap(pdf_cmap *cmap, int cpt);
+int pdf_lookup_cmap_full(pdf_cmap *cmap, int cpt, int *out);
+int pdf_decode_cmap(pdf_cmap *cmap, unsigned char *s, int *cpt);
+
+pdf_cmap *pdf_new_identity_cmap(fz_context *ctx, int wmode, int bytes);
+pdf_cmap *pdf_load_cmap(fz_context *ctx, fz_stream *file);
+pdf_cmap *pdf_load_system_cmap(fz_context *ctx, char *name);
+pdf_cmap *pdf_load_builtin_cmap(fz_context *ctx, char *name);
+pdf_cmap *pdf_load_embedded_cmap(pdf_document *doc, pdf_obj *ref);
+
+/*
+ * Font
+ */
+
+enum
+{
+	PDF_FD_FIXED_PITCH = 1 << 0,
+	PDF_FD_SERIF = 1 << 1,
+	PDF_FD_SYMBOLIC = 1 << 2,
+	PDF_FD_SCRIPT = 1 << 3,
+	PDF_FD_NONSYMBOLIC = 1 << 5,
+	PDF_FD_ITALIC = 1 << 6,
+	PDF_FD_ALL_CAP = 1 << 16,
+	PDF_FD_SMALL_CAP = 1 << 17,
+	PDF_FD_FORCE_BOLD = 1 << 18
+};
+
+enum { PDF_ROS_CNS, PDF_ROS_GB, PDF_ROS_JAPAN, PDF_ROS_KOREA };
+
+void pdf_load_encoding(char **estrings, char *encoding);
+int pdf_lookup_agl(char *name);
+const char **pdf_lookup_agl_duplicates(int ucs);
+
+extern const unsigned short pdf_doc_encoding[256];
+extern const char * const pdf_mac_roman[256];
+extern const char * const pdf_mac_expert[256];
+extern const char * const pdf_win_ansi[256];
+extern const char * const pdf_standard[256];
+
+typedef struct pdf_font_desc_s pdf_font_desc;
+typedef struct pdf_hmtx_s pdf_hmtx;
+typedef struct pdf_vmtx_s pdf_vmtx;
+
+struct pdf_hmtx_s
+{
+	unsigned short lo;
+	unsigned short hi;
+	int w;	/* type3 fonts can be big! */
+};
+
+struct pdf_vmtx_s
+{
+	unsigned short lo;
+	unsigned short hi;
+	short x;
+	short y;
+	short w;
+};
+
+struct pdf_font_desc_s
+{
+	fz_storable storable;
+	unsigned int size;
+
+	fz_font *font;
+
+	/* FontDescriptor */
+	int flags;
+	float italic_angle;
+	float ascent;
+	float descent;
+	float cap_height;
+	float x_height;
+	float missing_width;
+
+	/* Encoding (CMap) */
+	pdf_cmap *encoding;
+	pdf_cmap *to_ttf_cmap;
+	int cid_to_gid_len;
+	unsigned short *cid_to_gid;
+
+	/* ToUnicode */
+	pdf_cmap *to_unicode;
+	int cid_to_ucs_len;
+	unsigned short *cid_to_ucs;
+
+	/* Metrics (given in the PDF file) */
+	int wmode;
+
+	int hmtx_len, hmtx_cap;
+	pdf_hmtx dhmtx;
+	pdf_hmtx *hmtx;
+
+	int vmtx_len, vmtx_cap;
+	pdf_vmtx dvmtx;
+	pdf_vmtx *vmtx;
+
+	int is_embedded;
+};
+
+void pdf_set_font_wmode(fz_context *ctx, pdf_font_desc *font, int wmode);
+void pdf_set_default_hmtx(fz_context *ctx, pdf_font_desc *font, int w);
+void pdf_set_default_vmtx(fz_context *ctx, pdf_font_desc *font, int y, int w);
+void pdf_add_hmtx(fz_context *ctx, pdf_font_desc *font, int lo, int hi, int w);
+void pdf_add_vmtx(fz_context *ctx, pdf_font_desc *font, int lo, int hi, int x, int y, int w);
+void pdf_end_hmtx(fz_context *ctx, pdf_font_desc *font);
+void pdf_end_vmtx(fz_context *ctx, pdf_font_desc *font);
+pdf_hmtx pdf_get_hmtx(fz_context *ctx, pdf_font_desc *font, int cid);
+pdf_vmtx pdf_get_vmtx(fz_context *ctx, pdf_font_desc *font, int cid);
+
+void pdf_load_to_unicode(pdf_document *doc, pdf_font_desc *font, char **strings, char *collection, pdf_obj *cmapstm);
+
+int pdf_font_cid_to_gid(fz_context *ctx, pdf_font_desc *fontdesc, int cid);
+
+unsigned char *pdf_find_builtin_font(char *name, unsigned int *len);
+unsigned char *pdf_find_substitute_font(int mono, int serif, int bold, int italic, unsigned int *len);
+unsigned char *pdf_find_substitute_cjk_font(int ros, int serif, unsigned int *len);
+
+pdf_font_desc *pdf_load_type3_font(pdf_document *doc, pdf_obj *rdb, pdf_obj *obj);
+pdf_font_desc *pdf_load_font(pdf_document *doc, pdf_obj *rdb, pdf_obj *obj);
+
+pdf_font_desc *pdf_new_font_desc(fz_context *ctx);
+pdf_font_desc *pdf_keep_font(fz_context *ctx, pdf_font_desc *fontdesc);
+void pdf_drop_font(fz_context *ctx, pdf_font_desc *font);
+
+void pdf_debug_font(fz_context *ctx, pdf_font_desc *fontdesc);
+
+/*
+ * Interactive features
+ */
+
+typedef struct pdf_annot_s pdf_annot;
+
+struct pdf_annot_s
+{
+	pdf_obj *obj;
+	fz_rect rect;
+	pdf_xobject *ap;
+	fz_matrix matrix;
+	pdf_annot *next;
+};
+
+fz_link_dest pdf_parse_link_dest(pdf_document *doc, pdf_obj *dest);
+fz_link_dest pdf_parse_action(pdf_document *doc, pdf_obj *action);
+pdf_obj *pdf_lookup_dest(pdf_document *doc, pdf_obj *needle);
+pdf_obj *pdf_lookup_name(pdf_document *doc, char *which, pdf_obj *needle);
+pdf_obj *pdf_load_name_tree(pdf_document *doc, char *which);
+
+fz_link *pdf_load_link_annots(pdf_document *, pdf_obj *annots, fz_matrix page_ctm);
+
+pdf_annot *pdf_load_annots(pdf_document *, pdf_obj *annots);
+void pdf_free_annot(fz_context *ctx, pdf_annot *link);
+
+/*
+ * Page tree, pages and related objects
+ */
+
+struct pdf_page_s
+{
+	fz_matrix ctm; /* calculated from mediabox and rotate */
+	fz_rect mediabox;
+	int rotate;
+	int transparency;
+	pdf_obj *resources;
+	fz_buffer *contents;
+	fz_link *links;
+	pdf_annot *annots;
+};
+
+/*
+ * Content stream parsing
+ */
+
+void pdf_run_glyph(pdf_document *doc, pdf_obj *resources, fz_buffer *contents, fz_device *dev, fz_matrix ctm, void *gstate);
+
+/*
+ * PDF interface to store
+ */
+
+void pdf_store_item(fz_context *ctx, pdf_obj *key, void *val, unsigned int itemsize);
+void *pdf_find_item(fz_context *ctx, fz_store_free_fn *free, pdf_obj *key);
+void pdf_remove_item(fz_context *ctx, fz_store_free_fn *free, pdf_obj *key);
+
+#endif
diff --git a/pdf/mupdf.h b/pdf/mupdf.h
index 5d775044..7566bc36 100644
--- a/pdf/mupdf.h
+++ b/pdf/mupdf.h
@@ -1,9 +1,9 @@
-#ifndef _MUPDF_H_
-#define _MUPDF_H_
+#ifndef MUPDF_H
+#define MUPDF_H
 
-#ifndef _FITZ_H_
-#error "fitz.h must be included before mupdf.h"
-#endif
+#include "fitz.h"
+
+typedef struct pdf_document_s pdf_document;
 
 /*
  * Dynamic objects.
@@ -39,6 +39,7 @@ int pdf_is_string(pdf_obj *obj);
 int pdf_is_array(pdf_obj *obj);
 int pdf_is_dict(pdf_obj *obj);
 int pdf_is_indirect(pdf_obj *obj);
+int pdf_is_stream(pdf_document *doc, int num, int gen);
 
 int pdf_objcmp(pdf_obj *a, pdf_obj *b);
 
@@ -81,220 +82,27 @@ int pdf_fprint_obj(FILE *fp, pdf_obj *obj, int tight);
 void pdf_debug_obj(pdf_obj *obj);
 void pdf_debug_ref(pdf_obj *obj);
 
-void pdf_set_str_len(pdf_obj *obj, int newlen); /* private */
-void *pdf_get_indirect_document(pdf_obj *obj); /* private */
-
-/*
- * PDF Images
- */
-
-typedef struct pdf_image_params_s pdf_image_params;
-
-struct pdf_image_params_s
-{
-	int type;
-	fz_colorspace *colorspace;
-	union
-	{
-		struct
-		{
-			int columns;
-			int rows;
-			int k;
-			int eol;
-			int eba;
-			int eob;
-			int bi1;
-		}
-		fax;
-		struct
-		{
-			int ct;
-		}
-		jpeg;
-		struct
-		{
-			int columns;
-			int colors;
-			int predictor;
-			int bpc;
-		}
-		flate;
-		struct
-		{
-			int columns;
-			int colors;
-			int predictor;
-			int bpc;
-			int ec;
-		}
-		lzw;
-	}
-	u;
-};
-
-
-typedef struct pdf_image_s pdf_image;
-
-struct pdf_image_s
-{
-	fz_image base;
-	fz_pixmap *tile;
-	int n, bpc;
-	pdf_image_params params;
-	fz_buffer *buffer;
-	int colorkey[FZ_MAX_COLORS * 2];
-	float decode[FZ_MAX_COLORS * 2];
-	int imagemask;
-	int interpolate;
-	int usecolorkey;
-};
-
-enum
-{
-	PDF_IMAGE_RAW,
-	PDF_IMAGE_FAX,
-	PDF_IMAGE_JPEG,
-	PDF_IMAGE_RLD,
-	PDF_IMAGE_FLATE,
-	PDF_IMAGE_LZW,
-	PDF_IMAGE_JPX
-};
-
-typedef struct pdf_document_s pdf_document;
-
-/*
- * tokenizer and low-level object parser
- */
-
-enum
-{
-	PDF_TOK_ERROR, PDF_TOK_EOF,
-	PDF_TOK_OPEN_ARRAY, PDF_TOK_CLOSE_ARRAY,
-	PDF_TOK_OPEN_DICT, PDF_TOK_CLOSE_DICT,
-	PDF_TOK_OPEN_BRACE, PDF_TOK_CLOSE_BRACE,
-	PDF_TOK_NAME, PDF_TOK_INT, PDF_TOK_REAL, PDF_TOK_STRING, PDF_TOK_KEYWORD,
-	PDF_TOK_R, PDF_TOK_TRUE, PDF_TOK_FALSE, PDF_TOK_NULL,
-	PDF_TOK_OBJ, PDF_TOK_ENDOBJ,
-	PDF_TOK_STREAM, PDF_TOK_ENDSTREAM,
-	PDF_TOK_XREF, PDF_TOK_TRAILER, PDF_TOK_STARTXREF,
-	PDF_NUM_TOKENS
-};
-
-enum
-{
-	PDF_LEXBUF_SMALL = 256,
-	PDF_LEXBUF_LARGE = 65536
-};
-
-    
-
-typedef struct pdf_lexbuf_s pdf_lexbuf;
-typedef struct pdf_lexbuf_large_s pdf_lexbuf_large;
-
-struct pdf_lexbuf_s
-{
-	int size;
-	int len;
-	int i;
-	float f;
-	char scratch[PDF_LEXBUF_SMALL];
-};
-
-struct pdf_lexbuf_large_s
-{
-	pdf_lexbuf base;
-	char scratch[PDF_LEXBUF_LARGE - PDF_LEXBUF_SMALL];
-};
-
-    
-
-int pdf_lex(fz_stream *f, pdf_lexbuf *lexbuf);
-
-pdf_obj *pdf_parse_array(pdf_document *doc, fz_stream *f, pdf_lexbuf *buf);
-pdf_obj *pdf_parse_dict(pdf_document *doc, fz_stream *f, pdf_lexbuf *buf);
-pdf_obj *pdf_parse_stm_obj(pdf_document *doc, fz_stream *f, pdf_lexbuf *buf);
-pdf_obj *pdf_parse_ind_obj(pdf_document *doc, fz_stream *f, pdf_lexbuf *buf, int *num, int *gen, int *stm_ofs);
-
-fz_rect pdf_to_rect(fz_context *ctx, pdf_obj *array);
-fz_matrix pdf_to_matrix(fz_context *ctx, pdf_obj *array);
 char *pdf_to_utf8(fz_context *ctx, pdf_obj *src);
-unsigned short *pdf_to_ucs2(fz_context *ctx, pdf_obj *src);
+unsigned short *pdf_to_ucs2(fz_context *ctx, pdf_obj *src); /* sumatrapdf */
 pdf_obj *pdf_to_utf8_name(fz_context *ctx, pdf_obj *src);
 char *pdf_from_ucs2(fz_context *ctx, unsigned short *str);
 
-/*
- * xref and object / stream api
- */
-
-typedef struct pdf_xref_entry_s pdf_xref_entry;
-typedef struct pdf_crypt_s pdf_crypt;
-typedef struct pdf_ocg_descriptor_s pdf_ocg_descriptor;
-typedef struct pdf_ocg_entry_s pdf_ocg_entry;
-
-struct pdf_xref_entry_s
-{
-	int ofs;	/* file offset / objstm object number */
-	int gen;	/* generation / objstm index */
-	int stm_ofs;	/* on-disk stream */
-	pdf_obj *obj;	/* stored/cached object */
-	int type;	/* 0=unset (f)ree i(n)use (o)bjstm */
-};
-
-struct pdf_ocg_entry_s
-{
-	int num;
-	int gen;
-	int state;
-};
-
-struct pdf_ocg_descriptor_s
-{
-	int len;
-	pdf_ocg_entry *ocgs;
-	pdf_obj *intent;
-};
-
-struct pdf_document_s
-{
-	fz_document super;
-
-	fz_context *ctx;
-	fz_stream *file;
-
-	int version;
-	int startxref;
-	int file_size;
-	pdf_crypt *crypt;
-	pdf_obj *trailer;
-	pdf_ocg_descriptor *ocg;
-
-	int len;
-	pdf_xref_entry *table;
-
-	int page_len;
-	int page_cap;
-	pdf_obj **page_objs;
-	pdf_obj **page_refs;
-
-	pdf_lexbuf_large lexbuf;
-};
+fz_rect pdf_to_rect(fz_context *ctx, pdf_obj *array);
+fz_matrix pdf_to_matrix(fz_context *ctx, pdf_obj *array);
 
+int pdf_count_objects(pdf_document *doc);
 pdf_obj *pdf_resolve_indirect(pdf_obj *ref);
-void pdf_cache_object(pdf_document *doc, int num, int gen);
 pdf_obj *pdf_load_object(pdf_document *doc, int num, int gen);
 void pdf_update_object(pdf_document *doc, int num, int gen, pdf_obj *newobj);
 
-int pdf_is_stream(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);
 fz_buffer *pdf_load_raw_stream(pdf_document *doc, int num, int gen);
 fz_buffer *pdf_load_stream(pdf_document *doc, int num, int gen);
-fz_buffer *pdf_load_image_stream(pdf_document *doc, int num, int gen, pdf_image_params *params);
 fz_stream *pdf_open_raw_stream(pdf_document *doc, int num, int gen);
-fz_stream *pdf_open_image_stream(pdf_document *doc, int num, int gen, pdf_image_params *params);
 fz_stream *pdf_open_stream(pdf_document *doc, int num, int gen);
-fz_stream *pdf_open_stream_with_offset(pdf_document *doc, int num, int gen, pdf_obj *dict, int stm_ofs);
-fz_stream *pdf_open_image_decomp_stream(fz_context *ctx, fz_buffer *, pdf_image_params *params, int *factor);
+
+fz_image *pdf_load_image(pdf_document *doc, pdf_obj *obj);
+
+fz_outline *pdf_load_outline(pdf_document *doc);
 
 /*
 	pdf_open_document: Open a PDF document.
@@ -336,15 +144,8 @@ pdf_document *pdf_open_document_with_stream(fz_stream *file);
 */
 void pdf_close_document(pdf_document *doc);
 
-/* private */
-void pdf_repair_xref(pdf_document *doc, pdf_lexbuf *buf);
-void pdf_repair_obj_stms(pdf_document *doc);
-void pdf_debug_xref(pdf_document *);
-void pdf_resize_xref(pdf_document *doc, int newcap);
-
-/*
- * Encryption
- */
+int pdf_needs_password(pdf_document *doc);
+int pdf_authenticate_password(pdf_document *doc, char *pw);
 
 enum
 {
@@ -359,325 +160,10 @@ enum
 	PDF_DEFAULT_PERM_FLAGS = 0xfffc
 };
 
-pdf_crypt *pdf_new_crypt(fz_context *ctx, pdf_obj *enc, pdf_obj *id);
-void pdf_free_crypt(fz_context *ctx, pdf_crypt *crypt);
-
-void pdf_crypt_obj(fz_context *ctx, pdf_crypt *crypt, pdf_obj *obj, int num, int gen);
-fz_stream *pdf_open_crypt(fz_stream *chain, pdf_crypt *crypt, int num, int gen);
-fz_stream *pdf_open_crypt_with_filter(fz_stream *chain, pdf_crypt *crypt, char *name, int num, int gen);
-
-int pdf_needs_password(pdf_document *doc);
-int pdf_authenticate_password(pdf_document *doc, char *pw);
 int pdf_has_permission(pdf_document *doc, int p);
 
-int pdf_get_crypt_revision(pdf_document *doc);
-char *pdf_get_crypt_method(pdf_document *doc);
-int pdf_get_crypt_length(pdf_document *doc);
-unsigned char *pdf_get_crypt_key(pdf_document *doc);
-
-void pdf_debug_crypt(pdf_crypt *crypt);
-
-/*
- * Functions, Colorspaces, Shadings and Images
- */
-
-typedef struct pdf_function_s pdf_function;
-
-pdf_function *pdf_load_function(pdf_document *doc, pdf_obj *ref);
-void pdf_eval_function(fz_context *ctx, pdf_function *func, float *in, int inlen, float *out, int outlen);
-pdf_function *pdf_keep_function(fz_context *ctx, pdf_function *func);
-void pdf_drop_function(fz_context *ctx, pdf_function *func);
-unsigned int pdf_function_size(pdf_function *func);
-
-fz_colorspace *pdf_load_colorspace(pdf_document *doc, pdf_obj *obj);
-fz_pixmap *pdf_expand_indexed_pixmap(fz_context *ctx, fz_pixmap *src);
-
-fz_shade *pdf_load_shading(pdf_document *doc, pdf_obj *obj);
-
-fz_image *pdf_load_inline_image(pdf_document *doc, pdf_obj *rdb, pdf_obj *dict, fz_stream *file);
-fz_image *pdf_load_image(pdf_document *doc, pdf_obj *obj);
-int pdf_is_jpx_image(fz_context *ctx, pdf_obj *dict);
-
-/*
- * Pattern
- */
-
-typedef struct pdf_pattern_s pdf_pattern;
-
-struct pdf_pattern_s
-{
-	fz_storable storable;
-	int ismask;
-	float xstep;
-	float ystep;
-	fz_matrix matrix;
-	fz_rect bbox;
-	pdf_obj *resources;
-	fz_buffer *contents;
-};
-
-pdf_pattern *pdf_load_pattern(pdf_document *doc, pdf_obj *obj);
-pdf_pattern *pdf_keep_pattern(fz_context *ctx, pdf_pattern *pat);
-void pdf_drop_pattern(fz_context *ctx, pdf_pattern *pat);
-
-/*
- * XObject
- */
-
-typedef struct pdf_xobject_s pdf_xobject;
-
-struct pdf_xobject_s
-{
-	fz_storable storable;
-	fz_matrix matrix;
-	fz_rect bbox;
-	int isolated;
-	int knockout;
-	int transparency;
-	fz_colorspace *colorspace;
-	pdf_obj *resources;
-	fz_buffer *contents;
-	pdf_obj *me;
-};
-
-pdf_xobject *pdf_load_xobject(pdf_document *doc, pdf_obj *obj);
-pdf_xobject *pdf_keep_xobject(fz_context *ctx, pdf_xobject *xobj);
-void pdf_drop_xobject(fz_context *ctx, pdf_xobject *xobj);
-
-/*
- * CMap
- */
-
-typedef struct pdf_cmap_s pdf_cmap;
-typedef struct pdf_range_s pdf_range;
-
-enum { PDF_CMAP_SINGLE, PDF_CMAP_RANGE, PDF_CMAP_TABLE, PDF_CMAP_MULTI };
-
-struct pdf_range_s
-{
-	unsigned short low;
-	/* Next, we pack 2 fields into the same unsigned short. Top 14 bits
-	 * are the extent, bottom 2 bits are flags: single, range, table,
-	 * multi */
-	unsigned short extent_flags;
-	unsigned short offset;	/* range-delta or table-index */
-};
-
-struct pdf_cmap_s
-{
-	fz_storable storable;
-	char cmap_name[32];
-
-	char usecmap_name[32];
-	pdf_cmap *usecmap;
-
-	int wmode;
-
-	int codespace_len;
-	struct
-	{
-		unsigned short n;
-		unsigned short low;
-		unsigned short high;
-	} codespace[40];
-
-	int rlen, rcap;
-	pdf_range *ranges;
-
-	int tlen, tcap;
-	unsigned short *table;
-};
-
-pdf_cmap *pdf_new_cmap(fz_context *ctx);
-pdf_cmap *pdf_keep_cmap(fz_context *ctx, pdf_cmap *cmap);
-void pdf_drop_cmap(fz_context *ctx, pdf_cmap *cmap);
-void pdf_free_cmap_imp(fz_context *ctx, fz_storable *cmap);
-unsigned int pdf_cmap_size(fz_context *ctx, pdf_cmap *cmap);
-
-void pdf_debug_cmap(fz_context *ctx, pdf_cmap *cmap);
-int pdf_get_wmode(fz_context *ctx, pdf_cmap *cmap);
-void pdf_set_wmode(fz_context *ctx, pdf_cmap *cmap, int wmode);
-void pdf_set_usecmap(fz_context *ctx, pdf_cmap *cmap, pdf_cmap *usecmap);
-
-void pdf_add_codespace(fz_context *ctx, pdf_cmap *cmap, int low, int high, int n);
-void pdf_map_range_to_table(fz_context *ctx, pdf_cmap *cmap, int low, int *map, int len);
-void pdf_map_range_to_range(fz_context *ctx, pdf_cmap *cmap, int srclo, int srchi, int dstlo);
-void pdf_map_one_to_many(fz_context *ctx, pdf_cmap *cmap, int one, int *many, int len);
-void pdf_sort_cmap(fz_context *ctx, pdf_cmap *cmap);
-
-int pdf_lookup_cmap(pdf_cmap *cmap, int cpt);
-int pdf_lookup_cmap_full(pdf_cmap *cmap, int cpt, int *out);
-int pdf_decode_cmap(pdf_cmap *cmap, unsigned char *s, int *cpt);
-
-pdf_cmap *pdf_new_identity_cmap(fz_context *ctx, int wmode, int bytes);
-pdf_cmap *pdf_load_cmap(fz_context *ctx, fz_stream *file);
-pdf_cmap *pdf_load_system_cmap(fz_context *ctx, char *name);
-pdf_cmap *pdf_load_builtin_cmap(fz_context *ctx, char *name);
-pdf_cmap *pdf_load_embedded_cmap(pdf_document *doc, pdf_obj *ref);
-
-/*
- * Font
- */
-
-enum
-{
-	PDF_FD_FIXED_PITCH = 1 << 0,
-	PDF_FD_SERIF = 1 << 1,
-	PDF_FD_SYMBOLIC = 1 << 2,
-	PDF_FD_SCRIPT = 1 << 3,
-	PDF_FD_NONSYMBOLIC = 1 << 5,
-	PDF_FD_ITALIC = 1 << 6,
-	PDF_FD_ALL_CAP = 1 << 16,
-	PDF_FD_SMALL_CAP = 1 << 17,
-	PDF_FD_FORCE_BOLD = 1 << 18
-};
-
-enum { PDF_ROS_CNS, PDF_ROS_GB, PDF_ROS_JAPAN, PDF_ROS_KOREA };
-
-void pdf_load_encoding(char **estrings, char *encoding);
-int pdf_lookup_agl(char *name);
-const char **pdf_lookup_agl_duplicates(int ucs);
-
-extern const unsigned short pdf_doc_encoding[256];
-extern const char * const pdf_mac_roman[256];
-extern const char * const pdf_mac_expert[256];
-extern const char * const pdf_win_ansi[256];
-extern const char * const pdf_standard[256];
-
-typedef struct pdf_font_desc_s pdf_font_desc;
-typedef struct pdf_hmtx_s pdf_hmtx;
-typedef struct pdf_vmtx_s pdf_vmtx;
-
-struct pdf_hmtx_s
-{
-	unsigned short lo;
-	unsigned short hi;
-	int w;	/* type3 fonts can be big! */
-};
-
-struct pdf_vmtx_s
-{
-	unsigned short lo;
-	unsigned short hi;
-	short x;
-	short y;
-	short w;
-};
-
-struct pdf_font_desc_s
-{
-	fz_storable storable;
-	unsigned int size;
-
-	fz_font *font;
-
-	/* FontDescriptor */
-	int flags;
-	float italic_angle;
-	float ascent;
-	float descent;
-	float cap_height;
-	float x_height;
-	float missing_width;
-
-	/* Encoding (CMap) */
-	pdf_cmap *encoding;
-	pdf_cmap *to_ttf_cmap;
-	int cid_to_gid_len;
-	unsigned short *cid_to_gid;
-
-	/* ToUnicode */
-	pdf_cmap *to_unicode;
-	int cid_to_ucs_len;
-	unsigned short *cid_to_ucs;
-
-	/* Metrics (given in the PDF file) */
-	int wmode;
-
-	int hmtx_len, hmtx_cap;
-	pdf_hmtx dhmtx;
-	pdf_hmtx *hmtx;
-
-	int vmtx_len, vmtx_cap;
-	pdf_vmtx dvmtx;
-	pdf_vmtx *vmtx;
-
-	int is_embedded;
-};
-
-void pdf_set_font_wmode(fz_context *ctx, pdf_font_desc *font, int wmode);
-void pdf_set_default_hmtx(fz_context *ctx, pdf_font_desc *font, int w);
-void pdf_set_default_vmtx(fz_context *ctx, pdf_font_desc *font, int y, int w);
-void pdf_add_hmtx(fz_context *ctx, pdf_font_desc *font, int lo, int hi, int w);
-void pdf_add_vmtx(fz_context *ctx, pdf_font_desc *font, int lo, int hi, int x, int y, int w);
-void pdf_end_hmtx(fz_context *ctx, pdf_font_desc *font);
-void pdf_end_vmtx(fz_context *ctx, pdf_font_desc *font);
-pdf_hmtx pdf_get_hmtx(fz_context *ctx, pdf_font_desc *font, int cid);
-pdf_vmtx pdf_get_vmtx(fz_context *ctx, pdf_font_desc *font, int cid);
-
-void pdf_load_to_unicode(pdf_document *doc, pdf_font_desc *font, char **strings, char *collection, pdf_obj *cmapstm);
-
-int pdf_font_cid_to_gid(fz_context *ctx, pdf_font_desc *fontdesc, int cid);
-
-unsigned char *pdf_find_builtin_font(char *name, unsigned int *len);
-unsigned char *pdf_find_substitute_font(int mono, int serif, int bold, int italic, unsigned int *len);
-unsigned char *pdf_find_substitute_cjk_font(int ros, int serif, unsigned int *len);
-
-pdf_font_desc *pdf_load_type3_font(pdf_document *doc, pdf_obj *rdb, pdf_obj *obj);
-pdf_font_desc *pdf_load_font(pdf_document *doc, pdf_obj *rdb, pdf_obj *obj);
-
-pdf_font_desc *pdf_new_font_desc(fz_context *ctx);
-pdf_font_desc *pdf_keep_font(fz_context *ctx, pdf_font_desc *fontdesc);
-void pdf_drop_font(fz_context *ctx, pdf_font_desc *font);
-
-void pdf_debug_font(fz_context *ctx, pdf_font_desc *fontdesc);
-
-/*
- * Interactive features
- */
-
-typedef struct pdf_annot_s pdf_annot;
-
-struct pdf_annot_s
-{
-	pdf_obj *obj;
-	fz_rect rect;
-	pdf_xobject *ap;
-	fz_matrix matrix;
-	pdf_annot *next;
-};
-
-fz_link_dest pdf_parse_link_dest(pdf_document *doc, pdf_obj *dest);
-fz_link_dest pdf_parse_action(pdf_document *doc, pdf_obj *action);
-pdf_obj *pdf_lookup_dest(pdf_document *doc, pdf_obj *needle);
-pdf_obj *pdf_lookup_name(pdf_document *doc, char *which, pdf_obj *needle);
-pdf_obj *pdf_load_name_tree(pdf_document *doc, char *which);
-
-fz_outline *pdf_load_outline(pdf_document *doc);
-
-fz_link *pdf_load_link_annots(pdf_document *, pdf_obj *annots, fz_matrix page_ctm);
-
-pdf_annot *pdf_load_annots(pdf_document *, pdf_obj *annots);
-void pdf_free_annot(fz_context *ctx, pdf_annot *link);
-
-/*
- * Page tree, pages and related objects
- */
-
 typedef struct pdf_page_s pdf_page;
 
-struct pdf_page_s
-{
-	fz_matrix ctm; /* calculated from mediabox and rotate */
-	fz_rect mediabox;
-	int rotate;
-	int transparency;
-	pdf_obj *resources;
-	fz_buffer *contents;
-	fz_link *links;
-	pdf_annot *annots;
-};
-
 int pdf_find_page_number(pdf_document *doc, pdf_obj *pageobj);
 int pdf_count_pages(pdf_document *doc);
 
@@ -715,10 +201,6 @@ fz_rect pdf_bound_page(pdf_document *doc, pdf_page *page);
 void pdf_free_page(pdf_document *doc, pdf_page *page);
 
 /*
- * Content stream parsing
- */
-
-/*
 	pdf_run_page: Interpret a loaded page and render it on a device.
 
 	page: A page loaded by pdf_load_page.
@@ -731,14 +213,5 @@ void pdf_free_page(pdf_document *doc, pdf_page *page);
 void pdf_run_page(pdf_document *doc, pdf_page *page, fz_device *dev, fz_matrix ctm, fz_cookie *cookie);
 
 void pdf_run_page_with_usage(pdf_document *doc, pdf_page *page, fz_device *dev, fz_matrix ctm, char *event, fz_cookie *cookie);
-void pdf_run_glyph(pdf_document *doc, pdf_obj *resources, fz_buffer *contents, fz_device *dev, fz_matrix ctm, void *gstate);
-
-/*
- * PDF interface to store
- */
-
-void pdf_store_item(fz_context *ctx, pdf_obj *key, void *val, unsigned int itemsize);
-void *pdf_find_item(fz_context *ctx, fz_store_free_fn *free, pdf_obj *key);
-void pdf_remove_item(fz_context *ctx, fz_store_free_fn *free, pdf_obj *key);
 
 #endif
diff --git a/pdf/pdf_annot.c b/pdf/pdf_annot.c
index 81e5d887..bd0ad808 100644
--- a/pdf/pdf_annot.c
+++ b/pdf/pdf_annot.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 static pdf_obj *
 resolve_dest_rec(pdf_document *xref, pdf_obj *dest, int depth)
diff --git a/pdf/pdf_cmap.c b/pdf/pdf_cmap.c
index 987da592..712ebfe7 100644
--- a/pdf/pdf_cmap.c
+++ b/pdf/pdf_cmap.c
@@ -15,8 +15,8 @@
  * or can trust the parser to give us optimal mappings.
  */
 
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 /* Macros for accessing the combined extent_flags field */
 #define pdf_range_high(r) ((r)->low + ((r)->extent_flags >> 2))
diff --git a/pdf/pdf_cmap_load.c b/pdf/pdf_cmap_load.c
index d4e161c8..6867d845 100644
--- a/pdf/pdf_cmap_load.c
+++ b/pdf/pdf_cmap_load.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 unsigned int
 pdf_cmap_size(fz_context *ctx, pdf_cmap *cmap)
diff --git a/pdf/pdf_cmap_parse.c b/pdf/pdf_cmap_parse.c
index 4933dea5..3396e3ae 100644
--- a/pdf/pdf_cmap_parse.c
+++ b/pdf/pdf_cmap_parse.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 /*
  * CMap parser
diff --git a/pdf/pdf_cmap_table.c b/pdf/pdf_cmap_table.c
index fbe528e0..21dfe5fa 100644
--- a/pdf/pdf_cmap_table.c
+++ b/pdf/pdf_cmap_table.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 #ifndef NOCJK
 #include "../generated/cmap_cns.h"
diff --git a/pdf/pdf_colorspace.c b/pdf/pdf_colorspace.c
index 05d0c135..5bf84535 100644
--- a/pdf/pdf_colorspace.c
+++ b/pdf/pdf_colorspace.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 /* ICCBased */
 
diff --git a/pdf/pdf_crypt.c b/pdf/pdf_crypt.c
index 510acc7b..2927527e 100644
--- a/pdf/pdf_crypt.c
+++ b/pdf/pdf_crypt.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 enum
 {
diff --git a/pdf/pdf_encoding.c b/pdf/pdf_encoding.c
index 9a42a1c1..eb36ed96 100644
--- a/pdf/pdf_encoding.c
+++ b/pdf/pdf_encoding.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 #include "data_encodings.h"
 #include "data_glyphlist.h"
diff --git a/pdf/pdf_font.c b/pdf/pdf_font.c
index 5e54e0bb..11c97d37 100644
--- a/pdf/pdf_font.c
+++ b/pdf/pdf_font.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 #include <ft2build.h>
 #include FT_FREETYPE_H
diff --git a/pdf/pdf_fontfile.c b/pdf/pdf_fontfile.c
index 543ce763..4dbbe91c 100644
--- a/pdf/pdf_fontfile.c
+++ b/pdf/pdf_fontfile.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 #ifdef NOCJK
 #define NOCJKFONT
diff --git a/pdf/pdf_function.c b/pdf/pdf_function.c
index 681c0157..e1c53beb 100644
--- a/pdf/pdf_function.c
+++ b/pdf/pdf_function.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 enum
 {
diff --git a/pdf/pdf_image.c b/pdf/pdf_image.c
index 4cdd374a..17996a74 100644
--- a/pdf/pdf_image.c
+++ b/pdf/pdf_image.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 typedef struct pdf_image_key_s pdf_image_key;
 
diff --git a/pdf/pdf_interpret.c b/pdf/pdf_interpret.c
index ca957d09..1c2a11f3 100644
--- a/pdf/pdf_interpret.c
+++ b/pdf/pdf_interpret.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 #define TILE
 
diff --git a/pdf/pdf_lex.c b/pdf/pdf_lex.c
index 322d945c..6774167a 100644
--- a/pdf/pdf_lex.c
+++ b/pdf/pdf_lex.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 #define IS_NUMBER \
 	'+':case'-':case'.':case'0':case'1':case'2':case'3':\
diff --git a/pdf/pdf_metrics.c b/pdf/pdf_metrics.c
index fc888257..ae4405ed 100644
--- a/pdf/pdf_metrics.c
+++ b/pdf/pdf_metrics.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 void
 pdf_set_font_wmode(fz_context *ctx, pdf_font_desc *font, int wmode)
diff --git a/pdf/pdf_nametree.c b/pdf/pdf_nametree.c
index 98a10241..b24d2c24 100644
--- a/pdf/pdf_nametree.c
+++ b/pdf/pdf_nametree.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 static pdf_obj *
 pdf_lookup_name_imp(fz_context *ctx, pdf_obj *node, pdf_obj *needle)
diff --git a/pdf/pdf_outline.c b/pdf/pdf_outline.c
index 793b5f77..d4bea75a 100644
--- a/pdf/pdf_outline.c
+++ b/pdf/pdf_outline.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 static fz_outline *
 pdf_load_outline_imp(pdf_document *xref, pdf_obj *dict)
diff --git a/pdf/pdf_page.c b/pdf/pdf_page.c
index ca38aacf..5b9106d1 100644
--- a/pdf/pdf_page.c
+++ b/pdf/pdf_page.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 struct info
 {
diff --git a/pdf/pdf_parse.c b/pdf/pdf_parse.c
index 57ba1d6f..fe9db368 100644
--- a/pdf/pdf_parse.c
+++ b/pdf/pdf_parse.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 fz_rect
 pdf_to_rect(fz_context *ctx, pdf_obj *array)
@@ -45,7 +45,7 @@ pdf_to_utf8(fz_context *ctx, pdf_obj *src)
 		for (i = 2; i + 1 < srclen; i += 2)
 		{
 			ucs = srcptr[i] << 8 | srcptr[i+1];
-			dstlen += runelen(ucs);
+			dstlen += fz_runelen(ucs);
 		}
 
 		dstptr = dst = fz_malloc(ctx, dstlen + 1);
@@ -53,7 +53,7 @@ pdf_to_utf8(fz_context *ctx, pdf_obj *src)
 		for (i = 2; i + 1 < srclen; i += 2)
 		{
 			ucs = srcptr[i] << 8 | srcptr[i+1];
-			dstptr += runetochar(dstptr, &ucs);
+			dstptr += fz_runetochar(dstptr, ucs);
 		}
 	}
 	else if (srclen >= 2 && srcptr[0] == 255 && srcptr[1] == 254)
@@ -61,7 +61,7 @@ pdf_to_utf8(fz_context *ctx, pdf_obj *src)
 		for (i = 2; i + 1 < srclen; i += 2)
 		{
 			ucs = srcptr[i] | srcptr[i+1] << 8;
-			dstlen += runelen(ucs);
+			dstlen += fz_runelen(ucs);
 		}
 
 		dstptr = dst = fz_malloc(ctx, dstlen + 1);
@@ -69,20 +69,20 @@ pdf_to_utf8(fz_context *ctx, pdf_obj *src)
 		for (i = 2; i + 1 < srclen; i += 2)
 		{
 			ucs = srcptr[i] | srcptr[i+1] << 8;
-			dstptr += runetochar(dstptr, &ucs);
+			dstptr += fz_runetochar(dstptr, ucs);
 		}
 	}
 	else
 	{
 		for (i = 0; i < srclen; i++)
-			dstlen += runelen(pdf_doc_encoding[srcptr[i]]);
+			dstlen += fz_runelen(pdf_doc_encoding[srcptr[i]]);
 
 		dstptr = dst = fz_malloc(ctx, dstlen + 1);
 
 		for (i = 0; i < srclen; i++)
 		{
 			ucs = pdf_doc_encoding[srcptr[i]];
-			dstptr += runetochar(dstptr, &ucs);
+			dstptr += fz_runetochar(dstptr, ucs);
 		}
 	}
 
diff --git a/pdf/pdf_pattern.c b/pdf/pdf_pattern.c
index 00e4b8fb..14175670 100644
--- a/pdf/pdf_pattern.c
+++ b/pdf/pdf_pattern.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 pdf_pattern *
 pdf_keep_pattern(fz_context *ctx, pdf_pattern *pat)
diff --git a/pdf/pdf_repair.c b/pdf/pdf_repair.c
index f42198eb..fda1e6b5 100644
--- a/pdf/pdf_repair.c
+++ b/pdf/pdf_repair.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 /* Scan file for objects and reconstruct xref table */
 
diff --git a/pdf/pdf_shade.c b/pdf/pdf_shade.c
index c3bddcbc..589b7613 100644
--- a/pdf/pdf_shade.c
+++ b/pdf/pdf_shade.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 #define HUGENUM 32000 /* how far to extend axial/radial shadings */
 #define FUNSEGS 32 /* size of sampled mesh for function-based shadings */
diff --git a/pdf/pdf_store.c b/pdf/pdf_store.c
index 38abe10b..ae8eac9a 100644
--- a/pdf/pdf_store.c
+++ b/pdf/pdf_store.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 static int
 pdf_make_hash_key(fz_store_hash *hash, void *key_)
diff --git a/pdf/pdf_stream.c b/pdf/pdf_stream.c
index 6afb2cdb..84f966ec 100644
--- a/pdf/pdf_stream.c
+++ b/pdf/pdf_stream.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 /*
  * Check if an object is a stream or not.
diff --git a/pdf/pdf_type3.c b/pdf/pdf_type3.c
index dc931e22..6603fc8b 100644
--- a/pdf/pdf_type3.c
+++ b/pdf/pdf_type3.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 static void
 pdf_run_glyph_func(void *doc, void *rdb_, fz_buffer *contents, fz_device *dev, fz_matrix ctm, void *gstate)
diff --git a/pdf/pdf_unicode.c b/pdf/pdf_unicode.c
index edfda708..c94a6683 100644
--- a/pdf/pdf_unicode.c
+++ b/pdf/pdf_unicode.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 /* Load or synthesize ToUnicode map for fonts */
 
diff --git a/pdf/pdf_xobject.c b/pdf/pdf_xobject.c
index 36c5c483..ffa86184 100644
--- a/pdf/pdf_xobject.c
+++ b/pdf/pdf_xobject.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 pdf_xobject *
 pdf_keep_xobject(fz_context *ctx, pdf_xobject *xobj)
diff --git a/pdf/pdf_xref.c b/pdf/pdf_xref.c
index f0dbde96..9e60955c 100644
--- a/pdf/pdf_xref.c
+++ b/pdf/pdf_xref.c
@@ -1,5 +1,5 @@
-#include "fitz.h"
-#include "mupdf.h"
+#include "fitz-internal.h"
+#include "mupdf-internal.h"
 
 static inline int iswhite(int ch)
 {
@@ -1074,6 +1074,11 @@ pdf_resolve_indirect(pdf_obj *ref)
 	return ref;
 }
 
+int pdf_count_objects(pdf_document *doc)
+{
+	return doc->len;
+}
+
 /* Replace numbered object -- for use by pdfclean and similar tools */
 void
 pdf_update_object(pdf_document *xref, int num, int gen, pdf_obj *newobj)