Keep copy of decoded utf8 text string in pdf_obj.

Removes the need to alloc/free text strings in the API, allowing
for simple functions like pdf_dict_get_text_string.

9 files changed, 91 insertions, 61 deletions

diff --git a/source/pdf/pdf-annot.c b/source/pdf/pdf-annot.c
index ed6856de..ba92b7ef 100644
--- a/source/pdf/pdf-annot.c
+++ b/source/pdf/pdf-annot.c
@@ -392,10 +392,10 @@ pdf_set_annot_rect(fz_context *ctx, pdf_annot *annot, const fz_rect *rect)
 	pdf_dirty_annot(ctx, annot);
 }
 
-char *
-pdf_copy_annot_contents(fz_context *ctx, pdf_annot *annot)
+const char *
+pdf_get_annot_contents(fz_context *ctx, pdf_annot *annot)
 {
-	return pdf_to_utf8(ctx, pdf_dict_get(ctx, annot->obj, PDF_NAME(Contents)));
+	return pdf_dict_get_text_string(ctx, annot->obj, PDF_NAME(Contents));
 }
 
 void
@@ -1397,11 +1397,11 @@ pdf_annot_has_author(fz_context *ctx, pdf_annot *annot)
 	return is_allowed_subtype(ctx, annot, PDF_NAME(T), markup_subtypes);
 }
 
-char *
-pdf_copy_annot_author(fz_context *ctx, pdf_annot *annot)
+const char *
+pdf_get_annot_author(fz_context *ctx, pdf_annot *annot)
 {
 	check_allowed_subtypes(ctx, annot, PDF_NAME(T), markup_subtypes);
-	return pdf_to_utf8(ctx, pdf_dict_get(ctx, annot->obj, PDF_NAME(T)));
+	return pdf_dict_get_text_string(ctx, annot->obj, PDF_NAME(T));
 }
 
 void
diff --git a/source/pdf/pdf-appearance.c b/source/pdf/pdf-appearance.c
index c3e94baa..9753be92 100644
--- a/source/pdf/pdf-appearance.c
+++ b/source/pdf/pdf-appearance.c
@@ -906,11 +906,12 @@ pdf_write_free_text_appearance(fz_context *ctx, pdf_annot *annot, fz_buffer *buf
 {
 	const char *font;
 	float size, color[3];
-	char *text;
+	const char *text;
 	float w, h, t, b;
 	int q, r;
 
 	/* /Rotate is an undocumented annotation property supported by Adobe */
+	text = pdf_get_annot_contents(ctx, annot);
 	r = pdf_dict_get_int(ctx, annot->obj, PDF_NAME(Rotate));
 	q = pdf_annot_quadding(ctx, annot);
 	pdf_annot_default_appearance(ctx, annot, &font, &size, color);
@@ -933,13 +934,7 @@ pdf_write_free_text_appearance(fz_context *ctx, pdf_annot *annot, fz_buffer *buf
 		fz_append_printf(ctx, buf, "%g %g %g %g re\nS\n", b/2, b/2, w-b, h-b);
 	}
 
-	text = pdf_copy_annot_contents(ctx, annot);
-	fz_try(ctx)
-		write_variable_text(ctx, annot, buf, res, text, font, size, color, q, w, h, b*2, 1);
-	fz_always(ctx)
-		fz_free(ctx, text);
-	fz_catch(ctx)
-		fz_rethrow(ctx);
+	write_variable_text(ctx, annot, buf, res, text, font, size, color, q, w, h, b*2, 1);
 }
 
 static void
diff --git a/source/pdf/pdf-form.c b/source/pdf/pdf-form.c
index 54cb1472..93b88e2c 100644
--- a/source/pdf/pdf-form.c
+++ b/source/pdf/pdf-form.c
@@ -1102,7 +1102,7 @@ int pdf_text_widget_set_text(fz_context *ctx, pdf_document *doc, pdf_widget *tw,
 }
 
 /* Get either the listed value or the export value. */
-int pdf_choice_widget_options(fz_context *ctx, pdf_document *doc, pdf_widget *tw, int exportval, char *opts[])
+int pdf_choice_widget_options(fz_context *ctx, pdf_document *doc, pdf_widget *tw, int exportval, const char *opts[])
 {
 	pdf_annot *annot = (pdf_annot *)tw;
 	pdf_obj *optarr;
@@ -1122,11 +1122,11 @@ int pdf_choice_widget_options(fz_context *ctx, pdf_document *doc, pdf_widget *tw
 			/* If it is a two element array, the second item is the one that we want if we want the listing value. */
 			if (m == 2)
 				if (exportval)
-					opts[i] = pdf_to_utf8(ctx, pdf_array_get(ctx, pdf_array_get(ctx, optarr, i), 0));
+					opts[i] = pdf_array_get_text_string(ctx, pdf_array_get(ctx, optarr, i), 0);
 				else
-					opts[i] = pdf_to_utf8(ctx, pdf_array_get(ctx, pdf_array_get(ctx, optarr, i), 1));
+					opts[i] = pdf_array_get_text_string(ctx, pdf_array_get(ctx, optarr, i), 1);
 			else
-				opts[i] = pdf_to_utf8(ctx, pdf_array_get(ctx, optarr, i));
+				opts[i] = pdf_array_get_text_string(ctx, optarr, i);
 		}
 	}
 	return n;
@@ -1148,7 +1148,7 @@ int pdf_choice_widget_is_multiselect(fz_context *ctx, pdf_document *doc, pdf_wid
 	}
 }
 
-int pdf_choice_widget_value(fz_context *ctx, pdf_document *doc, pdf_widget *tw, char *opts[])
+int pdf_choice_widget_value(fz_context *ctx, pdf_document *doc, pdf_widget *tw, const char *opts[])
 {
 	pdf_annot *annot = (pdf_annot *)tw;
 	pdf_obj *optarr;
@@ -1162,32 +1162,27 @@ int pdf_choice_widget_value(fz_context *ctx, pdf_document *doc, pdf_widget *tw,
 	if (pdf_is_string(ctx, optarr))
 	{
 		if (opts)
-			opts[0] = pdf_to_utf8(ctx, optarr);
-
+			opts[0] = pdf_to_text_string(ctx, optarr);
 		return 1;
 	}
 	else
 	{
 		n = pdf_array_len(ctx, optarr);
-
 		if (opts)
 		{
 			for (i = 0; i < n; i++)
 			{
 				pdf_obj *elem = pdf_array_get(ctx, optarr, i);
-
 				if (pdf_is_array(ctx, elem))
 					elem = pdf_array_get(ctx, elem, 1);
-
-				opts[i] = pdf_to_utf8(ctx, elem);
+				opts[i] = pdf_to_text_string(ctx, elem);
 			}
 		}
-
 		return n;
 	}
 }
 
-void pdf_choice_widget_set_value(fz_context *ctx, pdf_document *doc, pdf_widget *tw, int n, char *opts[])
+void pdf_choice_widget_set_value(fz_context *ctx, pdf_document *doc, pdf_widget *tw, int n, const char *opts[])
 {
 	pdf_annot *annot = (pdf_annot *)tw;
 	pdf_obj *optarr = NULL, *opt;
diff --git a/source/pdf/pdf-link.c b/source/pdf/pdf-link.c
index 56c11973..e9f1c36f 100644
--- a/source/pdf/pdf-link.c
+++ b/source/pdf/pdf-link.c
@@ -134,7 +134,7 @@ pdf_parse_file_spec(fz_context *ctx, pdf_document *doc, pdf_obj *file_spec, pdf_
 		return NULL;
 	}
 
-	path = pdf_to_utf8(ctx, filename);
+	path = fz_strdup(ctx, pdf_to_text_string(ctx, filename));
 #ifdef _WIN32
 	if (!pdf_name_eq(ctx, pdf_dict_get(ctx, file_spec, PDF_NAME(FS)), PDF_NAME(URL)))
 	{
diff --git a/source/pdf/pdf-nametree.c b/source/pdf/pdf-nametree.c
index 30dd5974..4682024f 100644
--- a/source/pdf/pdf-nametree.c
+++ b/source/pdf/pdf-nametree.c
@@ -140,9 +140,13 @@ pdf_load_name_tree_imp(fz_context *ctx, pdf_obj *dict, pdf_document *doc, pdf_ob
 			pdf_obj *val = pdf_array_get(ctx, names, i + 1);
 			if (pdf_is_string(ctx, key))
 			{
-				key = pdf_to_utf8_name(ctx, key);
-				pdf_dict_put(ctx, dict, key, val);
-				pdf_drop_obj(ctx, key);
+				key = pdf_new_name(ctx, pdf_to_text_string(ctx, key));
+				fz_try(ctx)
+					pdf_dict_put(ctx, dict, key, val);
+				fz_always(ctx)
+					pdf_drop_obj(ctx, key);
+				fz_catch(ctx)
+					fz_rethrow(ctx);
 			}
 			else if (pdf_is_name(ctx, key))
 			{
diff --git a/source/pdf/pdf-object.c b/source/pdf/pdf-object.c
index ec66376a..bc856644 100644
--- a/source/pdf/pdf-object.c
+++ b/source/pdf/pdf-object.c
@@ -59,6 +59,7 @@ typedef struct pdf_obj_num_s
 typedef struct pdf_obj_string_s
 {
 	pdf_obj super;
+	char *text; /* utf8 encoded text string */
 	unsigned int len;
 	char buf[1];
 } pdf_obj_string;
@@ -141,6 +142,7 @@ pdf_new_string(fz_context *ctx, const char *str, size_t len)
 	obj->super.refs = 1;
 	obj->super.kind = PDF_STRING;
 	obj->super.flags = 0;
+	obj->text = NULL;
 	obj->len = l;
 	memcpy(obj->buf, str, len);
 	obj->buf[len] = '\0';
@@ -336,6 +338,32 @@ int pdf_to_str_len(fz_context *ctx, pdf_obj *obj)
 	return 0;
 }
 
+const char *pdf_to_string(fz_context *ctx, pdf_obj *obj, size_t *sizep)
+{
+	RESOLVE(obj);
+	if (OBJ_IS_STRING(obj))
+	{
+		if (sizep)
+			*sizep = STRING(obj)->len;
+		return STRING(obj)->buf;
+	}
+	if (sizep)
+		*sizep = 0;
+	return "";
+}
+
+const char *pdf_to_text_string(fz_context *ctx, pdf_obj *obj)
+{
+	RESOLVE(obj);
+	if (OBJ_IS_STRING(obj))
+	{
+		if (!STRING(obj)->text)
+			STRING(obj)->text = pdf_new_utf8_from_pdf_string(ctx, STRING(obj)->buf, STRING(obj)->len);
+		return STRING(obj)->text;
+	}
+	return "";
+}
+
 void pdf_set_int(fz_context *ctx, pdf_obj *obj, int64_t i)
 {
 	if (OBJ_IS_INT(obj))
@@ -1669,6 +1697,11 @@ pdf_drop_obj(fz_context *ctx, pdf_obj *obj)
 				pdf_drop_array(ctx, obj);
 			else if (obj->kind == PDF_DICT)
 				pdf_drop_dict(ctx, obj);
+			else if (obj->kind == PDF_STRING)
+			{
+				fz_free(ctx, STRING(obj)->text);
+				fz_free(ctx, obj);
+			}
 			else
 				fz_free(ctx, obj);
 		}
@@ -2254,10 +2287,12 @@ const char *pdf_dict_get_name(fz_context *ctx, pdf_obj *dict, pdf_obj *key)
 
 const char *pdf_dict_get_string(fz_context *ctx, pdf_obj *dict, pdf_obj *key, size_t *sizep)
 {
-	pdf_obj *val = pdf_dict_get(ctx, dict, key);
-	if (sizep)
-		*sizep = pdf_to_str_len(ctx, val);
-	return pdf_to_str_buf(ctx, val);
+	return pdf_to_string(ctx, pdf_dict_get(ctx, dict, key), sizep);
+}
+
+const char *pdf_dict_get_text_string(fz_context *ctx, pdf_obj *dict, pdf_obj *key)
+{
+	return pdf_to_text_string(ctx, pdf_dict_get(ctx, dict, key));
 }
 
 int pdf_array_get_bool(fz_context *ctx, pdf_obj *array, int index)
@@ -2274,3 +2309,13 @@ float pdf_array_get_real(fz_context *ctx, pdf_obj *array, int index)
 {
 	return pdf_to_real(ctx, pdf_array_get(ctx, array, index));
 }
+
+const char *pdf_array_get_string(fz_context *ctx, pdf_obj *array, int index, size_t *sizep)
+{
+	return pdf_to_string(ctx, pdf_array_get(ctx, array, index), sizep);
+}
+
+const char *pdf_array_get_text_string(fz_context *ctx, pdf_obj *array, int index)
+{
+	return pdf_to_text_string(ctx, pdf_array_get(ctx, array, index));
+}
diff --git a/source/pdf/pdf-outline.c b/source/pdf/pdf-outline.c
index ea8ecbeb..b091cf4a 100644
--- a/source/pdf/pdf-outline.c
+++ b/source/pdf/pdf-outline.c
@@ -24,7 +24,7 @@ pdf_load_outline_imp(fz_context *ctx, pdf_document *doc, pdf_obj *dict)
 
 			obj = pdf_dict_get(ctx, dict, PDF_NAME(Title));
 			if (obj)
-				node->title = pdf_to_utf8(ctx, obj);
+				node->title = fz_strdup(ctx, pdf_to_text_string(ctx, obj));
 
 			if ((obj = pdf_dict_get(ctx, dict, PDF_NAME(Dest))) != NULL)
 				node->uri = pdf_parse_link_dest(ctx, doc, obj);
diff --git a/source/pdf/pdf-parse.c b/source/pdf/pdf-parse.c
index f4075c5a..fb1cc595 100644
--- a/source/pdf/pdf-parse.c
+++ b/source/pdf/pdf-parse.c
@@ -80,9 +80,11 @@ skip_language_code_utf8(const unsigned char *s, size_t n, size_t i)
 	return 0;
 }
 
+/* Convert Unicode/PdfDocEncoding string into utf-8 */
 char *
-pdf_to_utf8_imp(fz_context *ctx, const unsigned char *srcptr, size_t srclen)
+pdf_new_utf8_from_pdf_string(fz_context *ctx, const char *ssrcptr, size_t srclen)
 {
+	const unsigned char *srcptr = (const unsigned char*)ssrcptr;
 	char *dstptr, *dst;
 	size_t dstlen = 0;
 	int ucs;
@@ -168,30 +170,29 @@ pdf_to_utf8_imp(fz_context *ctx, const unsigned char *srcptr, size_t srclen)
 	return dst;
 }
 
-/* Convert Unicode/PdfDocEncoding string into utf-8 */
+/* Convert text string object to UTF-8 */
 char *
-pdf_to_utf8(fz_context *ctx, pdf_obj *src)
+pdf_new_utf8_from_pdf_string_obj(fz_context *ctx, pdf_obj *src)
 {
-	unsigned char *srcptr;
+	const char *srcptr;
 	size_t srclen;
-	srcptr = (unsigned char *) pdf_to_str_buf(ctx, src);
-	srclen = pdf_to_str_len(ctx, src);
-	return pdf_to_utf8_imp(ctx, srcptr, srclen);
+	srcptr = pdf_to_string(ctx, src, &srclen);
+	return pdf_new_utf8_from_pdf_string(ctx, srcptr, srclen);
 }
 
 /* Load text stream and convert to UTF-8 */
 char *
-pdf_load_stream_as_utf8(fz_context *ctx, pdf_obj *src)
+pdf_new_utf8_from_pdf_stream_obj(fz_context *ctx, pdf_obj *src)
 {
 	fz_buffer *stmbuf;
-	unsigned char *srcptr;
+	char *srcptr;
 	size_t srclen;
 	char *dst = NULL;
 
 	stmbuf = pdf_load_stream(ctx, src);
-	srclen = fz_buffer_storage(ctx, stmbuf, &srcptr);
+	srclen = fz_buffer_storage(ctx, stmbuf, (unsigned char **)&srcptr);
 	fz_try(ctx)
-		dst = pdf_to_utf8_imp(ctx, srcptr, srclen);
+		dst = pdf_new_utf8_from_pdf_string(ctx, srcptr, srclen);
 	fz_always(ctx)
 		fz_drop_buffer(ctx, stmbuf);
 	fz_catch(ctx)
@@ -204,17 +205,8 @@ char *
 pdf_load_stream_or_string_as_utf8(fz_context *ctx, pdf_obj *src)
 {
 	if (pdf_is_stream(ctx, src))
-		return pdf_load_stream_as_utf8(ctx, src);
-	return pdf_to_utf8(ctx, src);
-}
-
-pdf_obj *
-pdf_to_utf8_name(fz_context *ctx, pdf_obj *src)
-{
-	char *buf = pdf_to_utf8(ctx, src);
-	pdf_obj *dst = pdf_new_name(ctx, buf);
-	fz_free(ctx, buf);
-	return dst;
+		return pdf_new_utf8_from_pdf_stream_obj(ctx, src);
+	return pdf_new_utf8_from_pdf_string_obj(ctx, src);
 }
 
 static pdf_obj *
diff --git a/source/pdf/pdf-xref.c b/source/pdf/pdf-xref.c
index 9d32ea5e..d60eb089 100644
--- a/source/pdf/pdf-xref.c
+++ b/source/pdf/pdf-xref.c
@@ -2220,7 +2220,7 @@ pdf_lookup_metadata(fz_context *ctx, pdf_document *doc, const char *key, char *b
 	if (strstr(key, "info:") == key)
 	{
 		pdf_obj *info;
-		char *s;
+		const char *s;
 		int n;
 
 		info = pdf_dict_get(ctx, pdf_trailer(ctx, doc), PDF_NAME(Info));
@@ -2231,9 +2231,8 @@ pdf_lookup_metadata(fz_context *ctx, pdf_document *doc, const char *key, char *b
 		if (!info)
 			return -1;
 
-		s = pdf_to_utf8(ctx, info);
+		s = pdf_to_text_string(ctx, info);
 		n = (int)fz_strlcpy(buf, s, size);
-		fz_free(ctx, s);
 		return n;
 	}