Rework handling of PDF names for speed and memory.

Currently, every PDF name is allocated in a pdf_obj structure, and
comparisons are done using strcmp. Given that we can predict most
of the PDF names we'll use in a given file, this seems wasteful.

The pdf_obj type is opaque outside the pdf-object.c file, so we can
abuse it slightly without anyone outside knowing.

We collect a sorted list of names used in PDF (resources/pdf/names.txt),
and we add a utility (namedump) that preprocesses this into 2 header
files.

The first (include/mupdf/pdf/pdf-names-table.h, included as part of
include/mupdf/pdf/object.h), defines a set of "PDF_NAME_xxxx"
entries. These are pdf_obj *'s that callers can use to mean "A PDF
object that means literal name 'xxxx'"

The second (source/pdf/pdf-name-impl.h) is a C array of names.

We therefore update the code so that rather than passing "xxxx" to
functions (such as pdf_dict_gets(...)) we now pass PDF_NAME_xxxx (to
pdf_dict_get(...)). This is a fairly natural (if widespread) change.

The pdf_dict_getp (and sibling) functions that take a path (e.g.
"foo/bar/baz") are therefore supplemented with equivalents that
take a list (pdf_dict_getl(... , PDF_NAME_foo, PDF_NAME_bar,
PDF_NAME_baz, NULL)).

The actual implementation of this relies on the fact that small
pointer values are never valid values. For a given pdf_obj *p,
if NULL < (intptr_t)p < PDF_NAME__LIMIT then p is a literal
entry in the name table.

This enables us to do fast pointer compares and to skip expensive
strcmps.

Also, bring "null", "true" and "false" into the same style as PDF names.

Rather than using full pdf_obj structures for null/true/false, use
special pointer values just above the PDF_NAME_ table. This saves
memory and makes comparisons easier.

1 files changed, 60 insertions, 60 deletions

diff --git a/source/pdf/pdf-annot.c b/source/pdf/pdf-annot.c
index 79114211..50c637d2 100644
--- a/source/pdf/pdf-annot.c
+++ b/source/pdf/pdf-annot.c
@@ -24,7 +24,7 @@ resolve_dest_rec(fz_context *ctx, pdf_document *doc, pdf_obj *dest, fz_link_kind
 
 	else if (pdf_is_dict(ctx, dest))
 	{
-		dest = pdf_dict_gets(ctx, dest, "D");
+		dest = pdf_dict_get(ctx, dest, PDF_NAME_D);
 		return resolve_dest_rec(ctx, doc, dest, kind, depth+1);
 	}
 
@@ -101,27 +101,27 @@ pdf_parse_link_dest(fz_context *ctx, pdf_document *doc, fz_link_kind kind, pdf_o
 	if (!pdf_is_name(ctx, obj))
 		return ld;
 
-	if (!strcmp("XYZ", pdf_to_name(ctx, obj)))
+	if (pdf_name_eq(ctx, PDF_NAME_XYZ, obj))
 	{
 		l_from_2 = t_from_3 = z_from_4 = 1;
 		ld.ld.gotor.flags |= fz_link_flag_r_is_zoom;
 	}
-	else if ((!strcmp("Fit", pdf_to_name(ctx, obj))) || (!strcmp("FitB", pdf_to_name(ctx, obj))))
+	else if ((pdf_name_eq(ctx, PDF_NAME_Fit, obj)) || (pdf_name_eq(ctx, PDF_NAME_FitB, obj)))
 	{
 		ld.ld.gotor.flags |= fz_link_flag_fit_h;
 		ld.ld.gotor.flags |= fz_link_flag_fit_v;
 	}
-	else if ((!strcmp("FitH", pdf_to_name(ctx, obj))) || (!strcmp("FitBH", pdf_to_name(ctx, obj))))
+	else if ((pdf_name_eq(ctx, PDF_NAME_FitH, obj)) || (pdf_name_eq(ctx, PDF_NAME_FitBH, obj)))
 	{
 		t_from_2 = 1;
 		ld.ld.gotor.flags |= fz_link_flag_fit_h;
 	}
-	else if ((!strcmp("FitV", pdf_to_name(ctx, obj))) || (!strcmp("FitBV", pdf_to_name(ctx, obj))))
+	else if ((pdf_name_eq(ctx, PDF_NAME_FitV, obj)) || (pdf_name_eq(ctx, PDF_NAME_FitBV, obj)))
 	{
 		l_from_2 = 1;
 		ld.ld.gotor.flags |= fz_link_flag_fit_v;
 	}
-	else if (!strcmp("FitR", pdf_to_name(ctx, obj)))
+	else if (pdf_name_eq(ctx, PDF_NAME_FitR, obj))
 	{
 		l_from_2 = b_from_3 = r_from_4 = t_from_5 = 1;
 		ld.ld.gotor.flags |= fz_link_flag_fit_h;
@@ -228,12 +228,12 @@ pdf_parse_file_spec(fz_context *ctx, pdf_document *doc, pdf_obj *file_spec)
 
 	if (pdf_is_dict(ctx, file_spec)) {
 #if defined(_WIN32) || defined(_WIN64)
-		filename = pdf_dict_gets(ctx, file_spec, "DOS");
+		filename = pdf_dict_get(ctx, file_spec, PDF_NAME_DOS);
 #else
-		filename = pdf_dict_gets(ctx, file_spec, "Unix");
+		filename = pdf_dict_get(ctx, file_spec, PDF_NAME_Unix);
 #endif
 		if (!filename)
-			filename = pdf_dict_getsa(ctx, file_spec, "UF", "F");
+			filename = pdf_dict_geta(ctx, file_spec, PDF_NAME_UF, PDF_NAME_F);
 	}
 
 	if (!pdf_is_string(ctx, filename))
@@ -274,38 +274,38 @@ pdf_parse_action(fz_context *ctx, pdf_document *doc, pdf_obj *action)
 	if (!action)
 		return ld;
 
-	obj = pdf_dict_gets(ctx, action, "S");
-	if (!strcmp(pdf_to_name(ctx, obj), "GoTo"))
+	obj = pdf_dict_get(ctx, action, PDF_NAME_S);
+	if (pdf_name_eq(ctx, PDF_NAME_GoTo, obj))
 	{
-		dest = pdf_dict_gets(ctx, action, "D");
+		dest = pdf_dict_get(ctx, action, PDF_NAME_D);
 		ld = pdf_parse_link_dest(ctx, doc, FZ_LINK_GOTO, dest);
 	}
-	else if (!strcmp(pdf_to_name(ctx, obj), "URI"))
+	else if (pdf_name_eq(ctx, PDF_NAME_URI, obj))
 	{
 		ld.kind = FZ_LINK_URI;
-		ld.ld.uri.is_map = pdf_to_bool(ctx, pdf_dict_gets(ctx, action, "IsMap"));
-		ld.ld.uri.uri = pdf_to_utf8(ctx, doc, pdf_dict_gets(ctx, action, "URI"));
+		ld.ld.uri.is_map = pdf_to_bool(ctx, pdf_dict_get(ctx, action, PDF_NAME_IsMap));
+		ld.ld.uri.uri = pdf_to_utf8(ctx, doc, pdf_dict_get(ctx, action, PDF_NAME_URI));
 	}
-	else if (!strcmp(pdf_to_name(ctx, obj), "Launch"))
+	else if (pdf_name_eq(ctx, PDF_NAME_Launch, obj))
 	{
 		ld.kind = FZ_LINK_LAUNCH;
-		file_spec = pdf_dict_gets(ctx, action, "F");
+		file_spec = pdf_dict_get(ctx, action, PDF_NAME_F);
 		ld.ld.launch.file_spec = pdf_parse_file_spec(ctx, doc, file_spec);
-		ld.ld.launch.new_window = pdf_to_int(ctx, pdf_dict_gets(ctx, action, "NewWindow"));
-		ld.ld.launch.is_uri = !strcmp(pdf_to_name(ctx, pdf_dict_gets(ctx, file_spec, "FS")), "URL");
+		ld.ld.launch.new_window = pdf_to_int(ctx, pdf_dict_get(ctx, action, PDF_NAME_NewWindow));
+		ld.ld.launch.is_uri = pdf_name_eq(ctx, PDF_NAME_URL, pdf_dict_get(ctx, file_spec, PDF_NAME_FS));
 	}
-	else if (!strcmp(pdf_to_name(ctx, obj), "Named"))
+	else if (pdf_name_eq(ctx, PDF_NAME_Named, obj))
 	{
 		ld.kind = FZ_LINK_NAMED;
-		ld.ld.named.named = fz_strdup(ctx, pdf_to_name(ctx, pdf_dict_gets(ctx, action, "N")));
+		ld.ld.named.named = fz_strdup(ctx, pdf_to_name(ctx, pdf_dict_get(ctx, action, PDF_NAME_N)));
 	}
-	else if (!strcmp(pdf_to_name(ctx, obj), "GoToR"))
+	else if (pdf_name_eq(ctx, PDF_NAME_GoToR, obj))
 	{
-		dest = pdf_dict_gets(ctx, action, "D");
-		file_spec = pdf_dict_gets(ctx, action, "F");
+		dest = pdf_dict_get(ctx, action, PDF_NAME_D);
+		file_spec = pdf_dict_get(ctx, action, PDF_NAME_F);
 		ld = pdf_parse_link_dest(ctx, doc, FZ_LINK_GOTOR, dest);
 		ld.ld.gotor.file_spec = pdf_parse_file_spec(ctx, doc, file_spec);
-		ld.ld.gotor.new_window = pdf_to_int(ctx, pdf_dict_gets(ctx, action, "NewWindow"));
+		ld.ld.gotor.new_window = pdf_to_int(ctx, pdf_dict_get(ctx, action, PDF_NAME_NewWindow));
 	}
 	return ld;
 }
@@ -318,7 +318,7 @@ pdf_load_link(fz_context *ctx, pdf_document *doc, pdf_obj *dict, const fz_matrix
 	fz_rect bbox;
 	fz_link_dest ld;
 
-	obj = pdf_dict_gets(ctx, dict, "Rect");
+	obj = pdf_dict_get(ctx, dict, PDF_NAME_Rect);
 	if (obj)
 		pdf_to_rect(ctx, obj, &bbox);
 	else
@@ -326,15 +326,15 @@ pdf_load_link(fz_context *ctx, pdf_document *doc, pdf_obj *dict, const fz_matrix
 
 	fz_transform_rect(&bbox, page_ctm);
 
-	obj = pdf_dict_gets(ctx, dict, "Dest");
+	obj = pdf_dict_get(ctx, dict, PDF_NAME_Dest);
 	if (obj)
 		ld = pdf_parse_link_dest(ctx, doc, FZ_LINK_GOTO, obj);
 	else
 	{
-		action = pdf_dict_gets(ctx, dict, "A");
+		action = pdf_dict_get(ctx, dict, PDF_NAME_A);
 		/* fall back to additional action button's down/up action */
 		if (!action)
-			action = pdf_dict_getsa(ctx, pdf_dict_gets(ctx, dict, "AA"), "U", "D");
+			action = pdf_dict_geta(ctx, pdf_dict_get(ctx, dict, PDF_NAME_AA), PDF_NAME_U, PDF_NAME_D);
 
 		ld = pdf_parse_action(ctx, doc, action);
 	}
@@ -423,56 +423,56 @@ pdf_transform_annot(fz_context *ctx, pdf_annot *annot)
 
 fz_annot_type pdf_annot_obj_type(fz_context *ctx, pdf_obj *obj)
 {
-	char *subtype = pdf_to_name(ctx, pdf_dict_gets(ctx, obj, "Subtype"));
-	if (!strcmp(subtype, "Text"))
+	pdf_obj *subtype = pdf_dict_get(ctx, obj, PDF_NAME_Subtype);
+	if (pdf_name_eq(ctx, PDF_NAME_Text, subtype))
 		return FZ_ANNOT_TEXT;
-	else if (!strcmp(subtype, "Link"))
+	else if (pdf_name_eq(ctx, PDF_NAME_Link, subtype))
 		return FZ_ANNOT_LINK;
-	else if (!strcmp(subtype, "FreeText"))
+	else if (pdf_name_eq(ctx, PDF_NAME_FreeText, subtype))
 		return FZ_ANNOT_FREETEXT;
-	else if (!strcmp(subtype, "Line"))
+	else if (pdf_name_eq(ctx, PDF_NAME_Line, subtype))
 		return FZ_ANNOT_LINE;
-	else if (!strcmp(subtype, "Square"))
+	else if (pdf_name_eq(ctx, PDF_NAME_Square, subtype))
 		return FZ_ANNOT_SQUARE;
-	else if (!strcmp(subtype, "Circle"))
+	else if (pdf_name_eq(ctx, PDF_NAME_Circle, subtype))
 		return FZ_ANNOT_CIRCLE;
-	else if (!strcmp(subtype, "Polygon"))
+	else if (pdf_name_eq(ctx, PDF_NAME_Polygon, subtype))
 		return FZ_ANNOT_POLYGON;
-	else if (!strcmp(subtype, "PolyLine"))
+	else if (pdf_name_eq(ctx, PDF_NAME_PolyLine, subtype))
 		return FZ_ANNOT_POLYLINE;
-	else if (!strcmp(subtype, "Highlight"))
+	else if (pdf_name_eq(ctx, PDF_NAME_Highlight, subtype))
 		return FZ_ANNOT_HIGHLIGHT;
-	else if (!strcmp(subtype, "Underline"))
+	else if (pdf_name_eq(ctx, PDF_NAME_Underline, subtype))
 		return FZ_ANNOT_UNDERLINE;
-	else if (!strcmp(subtype, "Squiggly"))
+	else if (pdf_name_eq(ctx, PDF_NAME_Squiggly, subtype))
 		return FZ_ANNOT_SQUIGGLY;
-	else if (!strcmp(subtype, "StrikeOut"))
+	else if (pdf_name_eq(ctx, PDF_NAME_StrikeOut, subtype))
 		return FZ_ANNOT_STRIKEOUT;
-	else if (!strcmp(subtype, "Stamp"))
+	else if (pdf_name_eq(ctx, PDF_NAME_Stamp, subtype))
 		return FZ_ANNOT_STAMP;
-	else if (!strcmp(subtype, "Caret"))
+	else if (pdf_name_eq(ctx, PDF_NAME_Caret, subtype))
 		return FZ_ANNOT_CARET;
-	else if (!strcmp(subtype, "Ink"))
+	else if (pdf_name_eq(ctx, PDF_NAME_Ink, subtype))
 		return FZ_ANNOT_INK;
-	else if (!strcmp(subtype, "Popup"))
+	else if (pdf_name_eq(ctx, PDF_NAME_Popup, subtype))
 		return FZ_ANNOT_POPUP;
-	else if (!strcmp(subtype, "FileAttachment"))
+	else if (pdf_name_eq(ctx, PDF_NAME_FileAttachment, subtype))
 		return FZ_ANNOT_FILEATTACHMENT;
-	else if (!strcmp(subtype, "Sound"))
+	else if (pdf_name_eq(ctx, PDF_NAME_Sound, subtype))
 		return FZ_ANNOT_SOUND;
-	else if (!strcmp(subtype, "Movie"))
+	else if (pdf_name_eq(ctx, PDF_NAME_Movie, subtype))
 		return FZ_ANNOT_MOVIE;
-	else if (!strcmp(subtype, "Widget"))
+	else if (pdf_name_eq(ctx, PDF_NAME_Widget, subtype))
 		return FZ_ANNOT_WIDGET;
-	else if (!strcmp(subtype, "Screen"))
+	else if (pdf_name_eq(ctx, PDF_NAME_Screen, subtype))
 		return FZ_ANNOT_SCREEN;
-	else if (!strcmp(subtype, "PrinterMark"))
+	else if (pdf_name_eq(ctx, PDF_NAME_PrinterMark, subtype))
 		return FZ_ANNOT_PRINTERMARK;
-	else if (!strcmp(subtype, "TrapNet"))
+	else if (pdf_name_eq(ctx, PDF_NAME_TrapNet, subtype))
 		return FZ_ANNOT_TRAPNET;
-	else if (!strcmp(subtype, "Watermark"))
+	else if (pdf_name_eq(ctx, PDF_NAME_Watermark, subtype))
 		return FZ_ANNOT_WATERMARK;
-	else if (!strcmp(subtype, "3D"))
+	else if (pdf_name_eq(ctx, PDF_NAME_3D, subtype))
 		return FZ_ANNOT_3D;
 	else
 		return -1;
@@ -538,9 +538,9 @@ pdf_load_annots(fz_context *ctx, pdf_document *doc, pdf_page *page, pdf_obj *ann
 				doc->update_appearance(ctx, doc, annot);
 
 			obj = annot->obj;
-			rect = pdf_dict_gets(ctx, obj, "Rect");
-			ap = pdf_dict_gets(ctx, obj, "AP");
-			as = pdf_dict_gets(ctx, obj, "AS");
+			rect = pdf_dict_get(ctx, obj, PDF_NAME_Rect);
+			ap = pdf_dict_get(ctx, obj, PDF_NAME_AP);
+			as = pdf_dict_get(ctx, obj, PDF_NAME_AS);
 
 			/* We only collect annotations with an appearance
 			 * stream into this list, so remove any that don't
@@ -553,11 +553,11 @@ pdf_load_annots(fz_context *ctx, pdf_document *doc, pdf_page *page, pdf_obj *ann
 				&& hp->gen == pdf_to_gen(ctx, obj)
 				&& (hp->state & HOTSPOT_POINTER_DOWN))
 			{
-				n = pdf_dict_gets(ctx, ap, "D"); /* down state */
+				n = pdf_dict_get(ctx, ap, PDF_NAME_D); /* down state */
 			}
 
 			if (n == NULL)
-				n = pdf_dict_gets(ctx, ap, "N"); /* normal state */
+				n = pdf_dict_get(ctx, ap, PDF_NAME_N); /* normal state */
 
 			/* lookup current state in sub-dictionary */
 			if (!pdf_is_stream(ctx, doc, pdf_to_num(ctx, n), pdf_to_gen(ctx, n)))