Add CMap processing scripts, and turn cmapdump into mutool.

A dumping script written in python.
A flattening script written in python (for easier editing).
A subsetting script written in shell to minimize CMaps by reusing subsets.

Use 'mutool cmapdump' to bootstrap or verify cmap dumps.

1 files changed, 7 insertions, 134 deletions

diff --git a/scripts/cmapcleanz.c b/scripts/cmapcleanz.c
index 07373f61..dee04044 100644
--- a/scripts/cmapcleanz.c
+++ b/scripts/cmapcleanz.c
@@ -1,36 +1,10 @@
 /* cmapclean.c -- parse a CMap file and write it back out */
 
-/* We never want to build memento versions of the cmapdump util */
-#undef MEMENTO
-
-/* We never want large file access here */
-#undef FZ_LARGEFILE
-
 #include <stdio.h>
 #include <string.h>
 
 #include "mupdf/pdf.h"
 
-#include "../source/fitz/context.c"
-#include "../source/fitz/error.c"
-#include "../source/fitz/memory.c"
-#include "../source/fitz/output.c"
-#include "../source/fitz/string.c"
-#include "../source/fitz/buffer.c"
-#include "../source/fitz/stream-open.c"
-#include "../source/fitz/stream-read.c"
-#include "../source/fitz/strtod.c"
-#include "../source/fitz/strtof.c"
-#include "../source/fitz/ftoa.c"
-#include "../source/fitz/printf.c"
-#ifdef _WIN32
-#include "../source/fitz/time.c"
-#endif
-
-#include "../source/pdf/pdf-lex.c"
-#include "../source/pdf/pdf-cmap.c"
-#include "../source/pdf/pdf-cmap-parse.c"
-
 void pc(unsigned int c)
 {
 	if (c <= 0xff) printf("<%02x>", c);
@@ -63,7 +37,7 @@ main(int argc, char **argv)
 
 	fi = fz_open_file(ctx, argv[1]);
 	cmap = pdf_load_cmap(ctx, fi);
-	fz_close(fi);
+	fz_drop_stream(ctx, fi);
 
 	printf("begincmap\n");
 	printf("/CMapName /%s def\n", cmap->cmap_name);
@@ -105,7 +79,7 @@ main(int argc, char **argv)
 		for (k = 0; k < cmap->rlen; k++) {
 			if (cmap->ranges[k].high - cmap->ranges[k].low == 0) {
 				pc(cmap->ranges[k].low);
-				printf("%u\n", cmap->ranges[k].out);
+				printf(" %u\n", cmap->ranges[k].out);
 			}
 		}
 		printf("endcidchar\n");
@@ -117,8 +91,9 @@ main(int argc, char **argv)
 		for (k = 0; k < cmap->rlen; k++) {
 			if (cmap->ranges[k].high - cmap->ranges[k].low > 0) {
 				pc(cmap->ranges[k].low);
+				putchar(' ');
 				pc(cmap->ranges[k].high);
-				printf("%u\n", cmap->ranges[k].out);
+				printf(" %u\n", cmap->ranges[k].out);
 			}
 		}
 		printf("endcidrange\n");
@@ -160,6 +135,7 @@ main(int argc, char **argv)
 
 	/* 1-to-many */
 
+#if 0
 	if (cmap->mlen > 0)
 	{
 		printf("beginbfchar\n");
@@ -173,113 +149,10 @@ main(int argc, char **argv)
 		}
 		printf("endbfchar\n");
 	}
+#endif
 
 	printf("endcmap\n");
 
-	fz_free_context(ctx);
-	return 0;
-}
-
-void fz_new_font_context(fz_context *ctx)
-{
-}
-
-void fz_drop_font_context(fz_context *ctx)
-{
-}
-
-fz_font_context *fz_keep_font_context(fz_context *ctx)
-{
-	return NULL;
-}
-
-void fz_new_colorspace_context(fz_context *ctx)
-{
-}
-
-void fz_drop_colorspace_context(fz_context *ctx)
-{
-}
-
-fz_colorspace_context *fz_keep_colorspace_context(fz_context *ctx)
-{
-	return NULL;
-}
-
-void fz_new_aa_context(fz_context *ctx)
-{
-}
-
-void fz_drop_aa_context(fz_context *ctx)
-{
-}
-
-void fz_copy_aa_context(fz_context *dst, fz_context *src)
-{
-}
-
-void *fz_keep_storable(fz_context *ctx, const fz_storable *sc)
-{
-	fz_storable *s = (fz_storable *)sc;
-	return fz_keep_imp(ctx, s, &s->refs);
-}
-
-void fz_drop_storable(fz_context *ctx, const fz_storable *sc)
-{
-	fz_storable *s = (fz_storable *)sc;
-	if (fz_drop_imp(ctx, s, &s->refs))
-		s->drop(ctx, s);
-}
-
-void fz_new_store_context(fz_context *ctx, size_t max)
-{
-}
-
-void fz_drop_store_context(fz_context *ctx)
-{
-}
-
-fz_store *fz_keep_store_context(fz_context *ctx)
-{
-	return NULL;
-}
-
-int fz_store_scavenge(fz_context *ctx, size_t size, int *phase)
-{
-	return 0;
-}
-
-void fz_new_glyph_cache_context(fz_context *ctx)
-{
-}
-
-void fz_drop_glyph_cache_context(fz_context *ctx)
-{
-}
-
-fz_glyph_cache *fz_keep_glyph_cache(fz_context *ctx)
-{
-	return NULL;
-}
-
-void fz_new_document_handler_context(fz_context *ctx)
-{
-}
-
-void fz_drop_document_handler_context(fz_context *ctx)
-{
-}
-
-fz_document_handler_context *fz_keep_document_handler_context(fz_context *ctx)
-{
-	return NULL;
-}
-
-void fz_default_image_decode(void *arg, int w, int h, int l2factor, fz_irect *irect)
-{
-}
-
-int fz_default_image_scale(void *arg, int w, int h, int src_w, int src_h)
-{
+	fz_drop_context(ctx);
 	return 0;
 }