A few general utility functions added for the sake of the forms work

10 files changed, 405 insertions, 3 deletions

diff --git a/fitz/fitz-internal.h b/fitz/fitz-internal.h
index 05d7b52f..2f7f8041 100644
--- a/fitz/fitz-internal.h
+++ b/fitz/fitz-internal.h
@@ -419,6 +419,13 @@ void fz_write_buffer_bits(fz_context *ctx, fz_buffer *buf, int val, int bits);
 
 void fz_write_buffer_pad(fz_context *ctx, fz_buffer *buf);
 
+/*
+	fz_buffer_printf: print formatted to a buffer. The buffer will
+	grow, but the caller must ensure that no more than 256 bytes are
+	added to the buffer per call.
+*/
+void fz_buffer_printf(fz_context *ctx, fz_buffer *buffer, char *fmt, ...);
+
 struct fz_stream_s
 {
 	fz_context *ctx;
diff --git a/fitz/fitz.h b/fitz/fitz.h
index ee00b2ed..a66976af 100644
--- a/fitz/fitz.h
+++ b/fitz/fitz.h
@@ -403,7 +403,7 @@ void *fz_calloc(fz_context *ctx, unsigned int count, unsigned int size);
 	exception on failure to allocate.
 */
 #define fz_malloc_struct(CTX, STRUCT) \
-	Memento_label(fz_calloc(CTX,1,sizeof(STRUCT)), #STRUCT)
+	((STRUCT *)Memento_label(fz_calloc(CTX,1,sizeof(STRUCT)), #STRUCT))
 
 /*
 	fz_malloc_array: Allocate a block of (non zeroed) memory (with
diff --git a/fitz/stm_buffer.c b/fitz/stm_buffer.c
index 58539bf8..d14b738f 100644
--- a/fitz/stm_buffer.c
+++ b/fitz/stm_buffer.c
@@ -159,3 +159,18 @@ void fz_write_buffer_pad(fz_context *ctx, fz_buffer *buf)
 {
 	buf->unused_bits = 0;
 }
+
+void
+fz_buffer_printf(fz_context *ctx, fz_buffer *buffer, char *fmt, ...)
+{
+	va_list args;
+	va_start(args, fmt);
+
+	/* Caller guarantees not to generate more than 256 bytes per call */
+	while(buffer->cap - buffer->len < 256)
+		fz_grow_buffer(ctx, buffer);
+
+	buffer->len += vsprintf(buffer->data + buffer->len, fmt, args);
+
+	va_end(args);
+}
diff --git a/pdf/mupdf-internal.h b/pdf/mupdf-internal.h
index aa4334b7..b178494c 100644
--- a/pdf/mupdf-internal.h
+++ b/pdf/mupdf-internal.h
@@ -135,6 +135,11 @@ 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);
 
 /*
+	pdf_print_token: print a lexed token to a buffer, growing if necessary
+*/
+void pdf_print_token(fz_context *ctx, fz_buffer *buf, int tok, pdf_lexbuf *lex);
+
+/*
  * xref and object / stream api
  */
 
@@ -296,8 +301,11 @@ struct pdf_xobject_s
 };
 
 pdf_xobject *pdf_load_xobject(pdf_document *doc, pdf_obj *obj);
+pdf_obj *pdf_new_xobject(pdf_document *doc, fz_rect *bbox, fz_matrix *mat);
 pdf_xobject *pdf_keep_xobject(fz_context *ctx, pdf_xobject *xobj);
 void pdf_drop_xobject(fz_context *ctx, pdf_xobject *xobj);
+void pdf_xobject_set_contents(pdf_document *xref, pdf_xobject *from, fz_buffer *buffer);
+
 
 /*
  * CMap
diff --git a/pdf/mupdf.h b/pdf/mupdf.h
index bda0e2fa..10cc6f08 100644
--- a/pdf/mupdf.h
+++ b/pdf/mupdf.h
@@ -22,6 +22,8 @@ pdf_obj *pdf_new_string(fz_context *ctx, char *str, int len);
 pdf_obj *pdf_new_indirect(fz_context *ctx, int num, int gen, void *doc);
 pdf_obj *pdf_new_array(fz_context *ctx, int initialcap);
 pdf_obj *pdf_new_dict(fz_context *ctx, int initialcap);
+pdf_obj *pdf_new_rect(fz_context *ctx, fz_rect *rect);
+pdf_obj *pdf_new_matrix(fz_context *ctx, fz_matrix *mtx);
 pdf_obj *pdf_copy_array(fz_context *ctx, pdf_obj *array);
 pdf_obj *pdf_copy_dict(fz_context *ctx, pdf_obj *dict);
 
@@ -70,9 +72,11 @@ pdf_obj *pdf_dict_get_key(pdf_obj *dict, int idx);
 pdf_obj *pdf_dict_get_val(pdf_obj *dict, int idx);
 pdf_obj *pdf_dict_get(pdf_obj *dict, pdf_obj *key);
 pdf_obj *pdf_dict_gets(pdf_obj *dict, char *key);
+pdf_obj *pdf_dict_getp(pdf_obj *dict, char *key);
 pdf_obj *pdf_dict_getsa(pdf_obj *dict, char *key, char *abbrev);
 void pdf_dict_put(pdf_obj *dict, pdf_obj *key, pdf_obj *val);
 void pdf_dict_puts(pdf_obj *dict, char *key, pdf_obj *val);
+void pdf_dict_putp(pdf_obj *dict, char *key, pdf_obj *val);
 void pdf_dict_del(pdf_obj *dict, pdf_obj *key);
 void pdf_dict_dels(pdf_obj *dict, char *key);
 void pdf_sort_dict(pdf_obj *dict);
@@ -118,6 +122,11 @@ void pdf_delete_object(pdf_document *xref, int num);
 void pdf_update_object(pdf_document *xref, int num, pdf_obj *obj);
 
 /*
+	pdf_get_stream: Return the contents for object in xref table
+*/
+fz_buffer *pdf_get_stream(pdf_document *xref, int num);
+
+/*
 	pdf_update_stream: Replace stream contents for object in xref table with the passed in buffer.
 
 	The buffer contents must match the /Filter setting.
diff --git a/pdf/pdf_lex.c b/pdf/pdf_lex.c
index 6774167a..c6ab6604 100644
--- a/pdf/pdf_lex.c
+++ b/pdf/pdf_lex.c
@@ -461,3 +461,55 @@ pdf_lex(fz_stream *f, pdf_lexbuf *buf)
 		}
 	}
 }
+
+void pdf_print_token(fz_context *ctx, fz_buffer *fzbuf, int tok, pdf_lexbuf *buf)
+{
+	switch(tok)
+	{
+	case PDF_TOK_NAME:
+		fz_buffer_printf(ctx, fzbuf, "/%s", buf->scratch);
+		break;
+	case PDF_TOK_STRING:
+		{
+			int i;
+			fz_buffer_printf(ctx, fzbuf, "<");
+			for (i = 0; i < buf->len; i++)
+				fz_buffer_printf(ctx, fzbuf, "%02X", buf->scratch[i]);
+			fz_buffer_printf(ctx, fzbuf, ">");
+		}
+		break;
+	case PDF_TOK_OPEN_DICT:
+		fz_buffer_printf(ctx, fzbuf, "<<");
+		break;
+	case PDF_TOK_CLOSE_DICT:
+		fz_buffer_printf(ctx, fzbuf, ">>");
+		break;
+	case PDF_TOK_OPEN_ARRAY:
+		fz_buffer_printf(ctx, fzbuf, "[");
+		break;
+	case PDF_TOK_CLOSE_ARRAY:
+		fz_buffer_printf(ctx, fzbuf, "]");
+		break;
+	case PDF_TOK_OPEN_BRACE:
+		fz_buffer_printf(ctx, fzbuf, "{");
+		break;
+	case PDF_TOK_CLOSE_BRACE:
+		fz_buffer_printf(ctx, fzbuf, "}");
+		break;
+	case PDF_TOK_INT:
+		fz_buffer_printf(ctx, fzbuf, "%d", buf->i);
+		break;
+	case PDF_TOK_REAL:
+		{
+			char sbuf[256];
+			sprintf(sbuf, "%g", buf->f);
+			if (strchr(sbuf, 'e')) /* bad news! */
+				sprintf(sbuf, fabsf(buf->f) > 1 ? "%1.1f" : "%1.8f", buf->f);
+			fz_buffer_printf(ctx, fzbuf, "%s", sbuf);
+		}
+		break;
+	default:
+		fz_buffer_printf(ctx, fzbuf, "%s", buf->scratch);
+		break;
+	}
+}
diff --git a/pdf/pdf_object.c b/pdf/pdf_object.c
index cc5cc662..60f10a54 100644
--- a/pdf/pdf_object.c
+++ b/pdf/pdf_object.c
@@ -567,6 +567,98 @@ pdf_array_contains(pdf_obj *arr, pdf_obj *obj)
 	return 0;
 }
 
+pdf_obj *pdf_new_rect(fz_context *ctx, fz_rect *rect)
+{
+	pdf_obj *arr = NULL;
+	pdf_obj *item = NULL;
+
+	fz_var(arr);
+	fz_var(item);
+	fz_try(ctx)
+	{
+		arr = pdf_new_array(ctx, 4);
+
+		item = pdf_new_real(ctx, rect->x0);
+		pdf_array_put(arr, 0, item);
+		pdf_drop_obj(item);
+		item = NULL;
+
+		item = pdf_new_real(ctx, rect->y0);
+		pdf_array_put(arr, 1, item);
+		pdf_drop_obj(item);
+		item = NULL;
+
+		item = pdf_new_real(ctx, rect->x1 - rect->x0);
+		pdf_array_put(arr, 2, item);
+		pdf_drop_obj(item);
+		item = NULL;
+
+		item = pdf_new_real(ctx, rect->y1 - rect->y0);
+		pdf_array_put(arr, 3, item);
+		pdf_drop_obj(item);
+		item = NULL;
+	}
+	fz_catch(ctx)
+	{
+		pdf_drop_obj(item);
+		pdf_drop_obj(arr);
+		fz_rethrow(ctx);
+	}
+
+	return arr;
+}
+
+pdf_obj *pdf_new_matrix(fz_context *ctx, fz_matrix *mtx)
+{
+	pdf_obj *arr = NULL;
+	pdf_obj *item = NULL;
+
+	fz_var(arr);
+	fz_var(item);
+	fz_try(ctx)
+	{
+		arr = pdf_new_array(ctx, 6);
+
+		item = pdf_new_real(ctx, mtx->a);
+		pdf_array_put(arr, 0, item);
+		pdf_drop_obj(item);
+		item = NULL;
+
+		item = pdf_new_real(ctx, mtx->b);
+		pdf_array_put(arr, 1, item);
+		pdf_drop_obj(item);
+		item = NULL;
+
+		item = pdf_new_real(ctx, mtx->c);
+		pdf_array_put(arr, 2, item);
+		pdf_drop_obj(item);
+		item = NULL;
+
+		item = pdf_new_real(ctx, mtx->d);
+		pdf_array_put(arr, 3, item);
+		pdf_drop_obj(item);
+		item = NULL;
+
+		item = pdf_new_real(ctx, mtx->e);
+		pdf_array_put(arr, 4, item);
+		pdf_drop_obj(item);
+		item = NULL;
+
+		item = pdf_new_real(ctx, mtx->f);
+		pdf_array_put(arr, 5, item);
+		pdf_drop_obj(item);
+		item = NULL;
+	}
+	fz_catch(ctx)
+	{
+		pdf_drop_obj(item);
+		pdf_drop_obj(arr);
+		fz_rethrow(ctx);
+	}
+
+	return arr;
+}
+
 /* dicts may only have names as keys! */
 
 static int keyvalcmp(const void *ap, const void *bp)
@@ -743,6 +835,36 @@ pdf_dict_gets(pdf_obj *obj, char *key)
 }
 
 pdf_obj *
+pdf_dict_getp(pdf_obj *obj, char *keys)
+{
+	char buf[256];
+	char *k, *e;
+
+	if (strlen(keys)+1 > 256)
+		fz_throw(obj->ctx, "buffer overflow in pdf_dict_getp");
+
+	strcpy(buf, keys);
+
+	e = buf;
+	while (*e && obj)
+	{
+		k = e;
+		while (*e != '/' && *e != '\0')
+			e++;
+
+		if (*e == '/')
+		{
+			*e = '\0';
+			e++;
+		}
+
+		obj = pdf_dict_gets(obj, k);
+	}
+
+	return obj;
+}
+
+pdf_obj *
 pdf_dict_get(pdf_obj *obj, pdf_obj *key)
 {
 	if (!key || key->kind != PDF_NAME)
@@ -829,6 +951,63 @@ pdf_dict_puts(pdf_obj *obj, char *key, pdf_obj *val)
 }
 
 void
+pdf_dict_putp(pdf_obj *obj, char *keys, pdf_obj *val)
+{
+	fz_context *ctx = obj->ctx;
+	char buf[256];
+	char *k, *e;
+	pdf_obj *cobj = NULL;
+
+	if (strlen(keys)+1 > 256)
+		fz_throw(obj->ctx, "buffer overflow in pdf_dict_getp");
+
+	strcpy(buf, keys);
+
+	e = buf;
+	while (*e)
+	{
+		k = e;
+		while (*e != '/' && *e != '\0')
+			e++;
+
+		if (*e == '/')
+		{
+			*e = '\0';
+			e++;
+		}
+
+		if (*e)
+		{
+			/* Not the last key in the key path. Create subdict if not already there. */
+			cobj = pdf_dict_gets(obj, k);
+			if (cobj == NULL)
+			{
+				cobj = pdf_new_dict(ctx, 1);
+				fz_try(ctx)
+				{
+					pdf_dict_puts(obj, k, cobj);
+				}
+				fz_always(ctx)
+				{
+					pdf_drop_obj(cobj);
+				}
+				fz_catch(ctx)
+				{
+					fz_rethrow(ctx);
+				}
+			}
+			/* Move to subdict */
+			obj = cobj;
+		}
+		else
+		{
+			/* Last key. Use it to store the value */
+			pdf_dict_puts(obj, k, val);
+		}
+	}
+}
+
+void
 pdf_dict_dels(pdf_obj *obj, char *key)
 {
 	RESOLVE(obj);
diff --git a/pdf/pdf_stream.c b/pdf/pdf_stream.c
index fcf9acdc..829210ea 100644
--- a/pdf/pdf_stream.c
+++ b/pdf/pdf_stream.c
@@ -358,7 +358,7 @@ pdf_open_image_stream(pdf_document *xref, int num, int gen, int orig_num, int or
 	pdf_cache_object(xref, num, gen);
 	/* RJW: "cannot load stream object (%d %d R)", num, gen */
 
-	if (x->stm_ofs == 0)
+	if (x->stm_ofs == 0 && x->stm_buf == NULL)
 		fz_throw(xref->ctx, "object is not a stream");
 
 	return pdf_open_filter(xref->file, xref, x->obj, orig_num, orig_gen, x->stm_ofs, params);
diff --git a/pdf/pdf_xobject.c b/pdf/pdf_xobject.c
index 1de0b20e..88ab5a01 100644
--- a/pdf/pdf_xobject.c
+++ b/pdf/pdf_xobject.c
@@ -23,7 +23,6 @@ pdf_free_xobject_imp(fz_context *ctx, fz_storable *xobj_)
 	if (xobj->resources)
 		pdf_drop_obj(xobj->resources);
 	if (xobj->contents)
-		//fz_drop_buffer(ctx, xobj->contents);
 		pdf_drop_obj(xobj->contents);
 	pdf_drop_obj(xobj->me);
 	fz_free(ctx, xobj);
@@ -111,3 +110,122 @@ pdf_load_xobject(pdf_document *xref, pdf_obj *dict)
 
 	return form;
 }
+
+pdf_obj *
+pdf_new_xobject(pdf_document *xref, fz_rect *bbox, fz_matrix *mat)
+{
+	int idict_num;
+	pdf_obj *idict = NULL;
+	pdf_obj *dict = NULL;
+	pdf_xobject *form = NULL;
+	pdf_obj *obj = NULL;
+	pdf_obj *res = NULL;
+	pdf_obj *procset = NULL;
+	fz_context *ctx = xref->ctx;
+
+	fz_var(idict);
+	fz_var(dict);
+	fz_var(form);
+	fz_var(obj);
+	fz_var(res);
+	fz_var(procset);
+	fz_try(ctx)
+	{
+		dict = pdf_new_dict(ctx, 0);
+
+		obj = pdf_new_rect(ctx, bbox);
+		pdf_dict_puts(dict, "BBox", obj);
+		pdf_drop_obj(obj);
+		obj = NULL;
+
+		obj = pdf_new_int(ctx, 1);
+		pdf_dict_puts(dict, "FormType", obj);
+		pdf_drop_obj(obj);
+		obj = NULL;
+
+		obj = pdf_new_int(ctx, 0);
+		pdf_dict_puts(dict, "Length", obj);
+		pdf_drop_obj(obj);
+		obj = NULL;
+
+		obj = pdf_new_matrix(ctx, mat);
+		pdf_dict_puts(dict, "Matrix", obj);
+		pdf_drop_obj(obj);
+		obj = NULL;
+
+		res = pdf_new_dict(ctx, 0);
+		procset = pdf_new_array(ctx, 2);
+		obj = fz_new_name(ctx, "PDF");
+		pdf_array_put(procset, 0, obj);
+		pdf_drop_obj(obj);
+		obj = NULL;
+		obj = fz_new_name(ctx, "Text");
+		pdf_array_put(procset, 1, obj);
+		pdf_drop_obj(obj);
+		obj = NULL;
+		pdf_dict_puts(res, "ProcSet", procset);
+		pdf_drop_obj(procset);
+		procset = NULL;
+		pdf_dict_puts(dict, "Resources", res);
+
+		obj = fz_new_name(ctx, "Form");
+		pdf_dict_puts(dict, "Subtype", obj);
+		pdf_drop_obj(obj);
+		obj = NULL;
+
+		obj = fz_new_name(ctx, "XObject");
+		pdf_dict_puts(dict, "Type", obj);
+		pdf_drop_obj(obj);
+		obj = NULL;
+
+		form = fz_malloc_struct(ctx, pdf_xobject);
+		FZ_INIT_STORABLE(form, 1, pdf_free_xobject_imp);
+		form->resources = NULL;
+		form->contents = NULL;
+		form->colorspace = NULL;
+		form->me = NULL;
+
+		form->bbox = *bbox;
+
+		form->matrix = fz_identity;
+
+		form->isolated = 0;
+		form->knockout = 0;
+		form->transparency = 0;
+
+		form->resources = res;
+		res = NULL;
+
+		idict_num = pdf_create_object(xref);
+		pdf_update_object(xref, idict_num, dict);
+		idict = pdf_new_indirect(ctx, idict_num, 0, xref);
+		pdf_drop_obj(dict);
+		dict = NULL;
+
+		pdf_store_item(ctx, idict, form, pdf_xobject_size(form));
+
+		form->contents = pdf_keep_obj(idict);
+		form->me = pdf_keep_obj(idict);
+
+		pdf_drop_xobject(ctx, form);
+		form = NULL;
+	}
+	fz_catch(ctx)
+	{
+		pdf_drop_obj(procset);
+		pdf_drop_obj(res);
+		pdf_drop_obj(obj);
+		pdf_drop_obj(dict);
+		pdf_drop_obj(idict);
+		pdf_drop_xobject(ctx, form);
+		fz_throw(ctx, "failed to create xobject)");
+	}
+
+	return idict;
+}
+
+void pdf_xobject_set_contents(pdf_document *xref, pdf_xobject *form, fz_buffer *buffer)
+{
+	pdf_dict_dels(form->contents, "Filter");
+	pdf_update_stream(xref, pdf_to_num(form->contents), buffer);
+}
diff --git a/pdf/pdf_xref.c b/pdf/pdf_xref.c
index 9b8a3ee8..af72bc09 100644
--- a/pdf/pdf_xref.c
+++ b/pdf/pdf_xref.c
@@ -1135,6 +1135,20 @@ pdf_update_object(pdf_document *xref, int num, pdf_obj *newobj)
 	x->ofs = 0;
 	x->obj = pdf_keep_obj(newobj);
 }
+
+fz_buffer *
+pdf_get_stream(pdf_document *xref, int num)
+{
+	pdf_xref_entry *x;
+
+	if (num < 0 || num >= xref->len)
+		fz_throw(xref->ctx, "object out of range (%d 0 R); xref size %d", num, xref->len);
+
+	x = &xref->table[num];
+
+	return x->stm_buf;
+}
+
 void
 pdf_update_stream(pdf_document *xref, int num, fz_buffer *newbuf)
 {