Add common page range parsing function for tools.

8 files changed, 92 insertions, 236 deletions

diff --git a/include/mupdf/fitz/string.h b/include/mupdf/fitz/string.h
index 8daa28fc..2af6ba93 100644
--- a/include/mupdf/fitz/string.h
+++ b/include/mupdf/fitz/string.h
@@ -136,4 +136,11 @@ float fz_strtof_no_exp(const char *string, char **tailptr);
 */
 int fz_grisu(float f, char *s, int *exp);
 
+/*
+	Check and parse string into page ranges:
+		( ','? ([0-9]+|'N') ( '-' ([0-9]+|N) )? )+
+*/
+int fz_is_page_range(fz_context *ctx, const char *s);
+const char *fz_parse_page_range(fz_context *ctx, const char *s, int *a, int *b, int n);
+
 #endif
diff --git a/source/fitz/string.c b/source/fitz/string.c
index 95b216fe..df171795 100644
--- a/source/fitz/string.c
+++ b/source/fitz/string.c
@@ -416,3 +416,50 @@ fz_off_t fz_atoo(const char *s)
 		return 0;
 	return fz_atoo_imp(s);
 }
+
+int fz_is_page_range(fz_context *ctx, const char *s)
+{
+	/* TODO: check the actual syntax... */
+	while (*s)
+	{
+		if ((*s < '0' || *s > '9') && *s != 'N' && *s != '-' && *s != ',')
+			return 0;
+		s++;
+	}
+	return 1;
+}
+
+const char *fz_parse_page_range(fz_context *ctx, const char *s, int *a, int *b, int n)
+{
+	if (!s || !s[0])
+		return NULL;
+
+	if (s[0] == ',')
+		s += 1;
+
+	if (s[0] == 'N')
+	{
+		*a = n;
+		s += 1;
+	}
+	else
+		*a = strtol(s, (char**)&s, 10);
+
+	if (s[0] == '-')
+	{
+		if (s[1] == 'N')
+		{
+			*b = n;
+			s += 2;
+		}
+		else
+			*b = strtol(s+1, (char**)&s, 10);
+	}
+	else
+		*b = *a;
+
+	*a = fz_clampi(*a, 1, n);
+	*b = fz_clampi(*b, 1, n);
+
+	return s;
+}
diff --git a/source/pdf/pdf-clean-file.c b/source/pdf/pdf-clean-file.c
index 2a80f0cc..e886dc5f 100644
--- a/source/pdf/pdf-clean-file.c
+++ b/source/pdf/pdf-clean-file.c
@@ -206,39 +206,18 @@ static void retainpages(fz_context *ctx, globals *glo, int argc, char **argv)
 	while (argc - argidx)
 	{
 		int page, spage, epage;
-		char *spec, *dash;
-		char *pagelist = argv[argidx];
+		const char *pagelist = argv[argidx];
 
 		pagecount = pdf_count_pages(ctx, doc);
-		spec = fz_strsep(&pagelist, ",");
-		while (spec)
-		{
-			dash = strchr(spec, '-');
-
-			if (dash == spec)
-				spage = epage = pagecount;
-			else
-				spage = epage = atoi(spec);
-
-			if (dash)
-			{
-				if (strlen(dash) > 1)
-					epage = atoi(dash + 1);
-				else
-					epage = pagecount;
-			}
-
-			spage = fz_clampi(spage, 1, pagecount);
-			epage = fz_clampi(epage, 1, pagecount);
 
+		while ((pagelist = fz_parse_page_range(ctx, pagelist, &spage, &epage, pagecount)))
+		{
 			if (spage < epage)
 				for (page = spage; page <= epage; ++page)
 					retainpage(ctx, doc, parent, kids, page);
 			else
 				for (page = spage; page >= epage; --page)
 					retainpage(ctx, doc, parent, kids, page);
-
-			spec = fz_strsep(&pagelist, ",");
 		}
 
 		argidx++;
diff --git a/source/tools/muconvert.c b/source/tools/muconvert.c
index 7c2d1f4f..6f354b01 100644
--- a/source/tools/muconvert.c
+++ b/source/tools/muconvert.c
@@ -40,7 +40,7 @@ static void usage(void)
 		"\t\tcbz, pdf\n"
 		"\t-O -\tcomma separated list of options for output format\n"
 		"\n"
-		"\tpages\tcomma separated list of page numbers and ranges\n"
+		"\tpages\tcomma separated list of page numbers and ranges, where N is the last page\n"
 		"\n"
 		);
 	fprintf(stderr, "%s\n", fz_cbz_write_options_usage);
@@ -48,17 +48,6 @@ static void usage(void)
 	exit(1);
 }
 
-static int isrange(const char *s)
-{
-	while (*s)
-	{
-		if ((*s < '0' || *s > '9') && *s != 'N' && *s != '-' && *s != ',')
-			return 0;
-		s++;
-	}
-	return 1;
-}
-
 static void runpage(int number)
 {
 	fz_matrix ctm;
@@ -74,44 +63,18 @@ static void runpage(int number)
 	fz_drop_page(ctx, page);
 }
 
-static void runrange(char *s)
+static void runrange(const char *range)
 {
 	int start, end, i;
-	while (s[0])
-	{
-		if (s[0] == 'N')
-		{
-			start = count;
-			s += 1;
-		}
-		else
-			start = strtol(s, &s, 10);
-
-		if (s[0] == '-')
-		{
-			if (s[1] == 'N')
-			{
-				end = count;
-				s += 2;
-			}
-			else
-				end = strtol(s+1, &s, 10);
-		}
-		else
-			end = start;
-
-		start = fz_clampi(start, 1, count);
-		end = fz_clampi(end, 1, count);
 
+	while ((range = fz_parse_page_range(ctx, range, &start, &end, count)))
+	{
 		if (start < end)
 			for (i = start; i <= end; ++i)
 				runpage(i);
 		else
 			for (i = start; i >= end; --i)
 				runpage(i);
-
-		if (s[0] != ',')
-			break;
 	}
 }
 
@@ -188,7 +151,7 @@ int muconvert_main(int argc, char **argv)
 		fz_layout_document(ctx, doc, layout_w, layout_h, layout_em);
 		count = fz_count_pages(ctx, doc);
 
-		if (isrange(argv[i+1]))
+		if (i+1 < argc && fz_is_page_range(ctx, argv[i+1]))
 			runrange(argv[++i]);
 		else
 			runrange("1-N");
diff --git a/source/tools/mudraw.c b/source/tools/mudraw.c
index 6523380b..a58912a9 100644
--- a/source/tools/mudraw.c
+++ b/source/tools/mudraw.c
@@ -455,17 +455,6 @@ static int gettime(void)
 	return (now.tv_sec - first.tv_sec) * 1000 + (now.tv_usec - first.tv_usec) / 1000;
 }
 
-static int isrange(char *s)
-{
-	while (*s)
-	{
-		if ((*s < '0' || *s > '9') && *s != '-' && *s != ',')
-			return 0;
-		s++;
-	}
-	return 1;
-}
-
 static int has_percent_d(char *s)
 {
 	/* find '%[0-9]*d' */
@@ -1187,41 +1176,20 @@ static void drawpage(fz_context *ctx, fz_document *doc, int pagenum)
 	}
 }
 
-static void drawrange(fz_context *ctx, fz_document *doc, char *range)
+static void drawrange(fz_context *ctx, fz_document *doc, const char *range)
 {
 	int page, spage, epage, pagecount;
-	char *spec, *dash;
 
 	pagecount = fz_count_pages(ctx, doc);
-	spec = fz_strsep(&range, ",");
-	while (spec)
-	{
-		dash = strchr(spec, '-');
-
-		if (dash == spec)
-			spage = epage = pagecount;
-		else
-			spage = epage = atoi(spec);
-
-		if (dash)
-		{
-			if (strlen(dash) > 1)
-				epage = atoi(dash + 1);
-			else
-				epage = pagecount;
-		}
-
-		spage = fz_clampi(spage, 1, pagecount);
-		epage = fz_clampi(epage, 1, pagecount);
 
+	while ((range = fz_parse_page_range(ctx, range, &spage, &epage, pagecount)))
+	{
 		if (spage < epage)
 			for (page = spage; page <= epage; page++)
 				drawpage(ctx, doc, page);
 		else
 			for (page = spage; page >= epage; page--)
 				drawpage(ctx, doc, page);
-
-		spec = fz_strsep(&range, ",");
 	}
 }
 
@@ -1648,9 +1616,9 @@ int mudraw_main(int argc, char **argv)
 				}
 				else
 				{
-					if (fz_optind == argc || !isrange(argv[fz_optind]))
-						drawrange(ctx, doc, "1-");
-					if (fz_optind < argc && isrange(argv[fz_optind]))
+					if (fz_optind == argc || !fz_is_page_range(ctx, argv[fz_optind]))
+						drawrange(ctx, doc, "1-N");
+					if (fz_optind < argc && fz_is_page_range(ctx, argv[fz_optind]))
 						drawrange(ctx, doc, argv[fz_optind++]);
 				}
 
diff --git a/source/tools/pdfinfo.c b/source/tools/pdfinfo.c
index 0d30b22a..fd52d12f 100644
--- a/source/tools/pdfinfo.c
+++ b/source/tools/pdfinfo.c
@@ -912,10 +912,9 @@ printinfo(fz_context *ctx, globals *glo, char *filename, int show, int page)
 }
 
 static void
-showinfo(fz_context *ctx, globals *glo, char *filename, int show, char *pagelist)
+showinfo(fz_context *ctx, globals *glo, char *filename, int show, const char *pagelist)
 {
 	int page, spage, epage;
-	char *spec, *dash;
 	int allpages;
 	int pagecount;
 	fz_output *out = glo->out;
@@ -923,33 +922,12 @@ showinfo(fz_context *ctx, globals *glo, char *filename, int show, char *pagelist
 	if (!glo->doc)
 		infousage();
 
-	allpages = !strcmp(pagelist, "1-");
+	allpages = !strcmp(pagelist, "1-N");
 
 	pagecount = pdf_count_pages(ctx, glo->doc);
-	spec = fz_strsep(&pagelist, ",");
-	while (spec && pagecount)
-	{
-		dash = strchr(spec, '-');
-
-		if (dash == spec)
-			spage = epage = pagecount;
-		else
-			spage = epage = atoi(spec);
-
-		if (dash)
-		{
-			if (strlen(dash) > 1)
-				epage = atoi(dash + 1);
-			else
-				epage = pagecount;
-		}
-
-		if (spage > epage)
-			page = spage, spage = epage, epage = page;
-
-		spage = fz_clampi(spage, 1, pagecount);
-		epage = fz_clampi(epage, 1, pagecount);
 
+	while ((pagelist = fz_parse_page_range(ctx, pagelist, &spage, &epage, pagecount)))
+	{
 		if (allpages)
 			fz_printf(ctx, out, "Retrieving info from pages %d-%d...\n", spage, epage);
 		for (page = spage; page <= epage; page++)
@@ -963,26 +941,12 @@ showinfo(fz_context *ctx, globals *glo, char *filename, int show, char *pagelist
 				clearinfo(ctx, glo);
 			}
 		}
-
-		spec = fz_strsep(&pagelist, ",");
 	}
 
 	if (allpages)
 		printinfo(ctx, glo, filename, show, -1);
 }
 
-static int arg_is_page_range(const char *arg)
-{
-	int c;
-
-	while ((c = *arg++) != 0)
-	{
-		if ((c < '0' || c > '9') && (c != '-') && (c != ','))
-			return 0;
-	}
-	return 1;
-}
-
 static void
 pdfinfo_info(fz_context *ctx, fz_output *out, char *filename, char *password, int show, char *argv[], int argc)
 {
@@ -996,11 +960,11 @@ pdfinfo_info(fz_context *ctx, fz_output *out, char *filename, char *password, in
 	state = NO_FILE_OPENED;
 	while (argidx < argc)
 	{
-		if (state == NO_FILE_OPENED || !arg_is_page_range(argv[argidx]))
+		if (state == NO_FILE_OPENED || !fz_is_page_range(ctx, argv[argidx]))
 		{
 			if (state == NO_INFO_GATHERED)
 			{
-				showinfo(ctx, &glo, filename, show, "1-");
+				showinfo(ctx, &glo, filename, show, "1-N");
 			}
 
 			closexref(ctx, &glo);
@@ -1026,7 +990,7 @@ pdfinfo_info(fz_context *ctx, fz_output *out, char *filename, char *password, in
 	}
 
 	if (state == NO_INFO_GATHERED)
-		showinfo(ctx, &glo, filename, show, "1-");
+		showinfo(ctx, &glo, filename, show, "1-N");
 
 	closexref(ctx, &glo);
 }
diff --git a/source/tools/pdfmerge.c b/source/tools/pdfmerge.c
index 597bd4ce..fc097994 100644
--- a/source/tools/pdfmerge.c
+++ b/source/tools/pdfmerge.c
@@ -13,7 +13,7 @@ static void usage(void)
 		"\t-o\tname of PDF file to create\n"
 		"\t-O\tcomma separated list of output options\n"
 		"\tinput.pdf\tname of input file from which to copy pages\n"
-		"\tpages\tcomma separated list of page ranges to copy (for example: 1-5,6,10-)\n\n"
+		"\tpages\tcomma separated list of page numbers and ranges\n"
 		);
 	fprintf(stderr, "%s\n", fz_pdf_write_options_usage);
 	exit(1);
@@ -23,19 +23,6 @@ static fz_context *ctx = NULL;
 static pdf_document *doc_des = NULL;
 static pdf_document *doc_src = NULL;
 
-/* This isrange is a duplicate with mudraw.c Not sure how we want to organize or if
- * we are fine with the small amount of code duplication */
-static int isrange(char *s)
-{
-	while (*s)
-	{
-		if ((*s < '0' || *s > '9') && *s != '-' && *s != ',')
-			return 0;
-		s++;
-	}
-	return 1;
-}
-
 static void page_merge(int page_from, int page_to, pdf_graft_map *graft_map)
 {
 	pdf_obj *pageref = NULL;
@@ -91,46 +78,24 @@ static void page_merge(int page_from, int page_to, pdf_graft_map *graft_map)
 	}
 }
 
-static void merge_range(char *range)
+static void merge_range(const char *range)
 {
-	int page, spage, epage, src_pagecount, des_pagecount;
-	char *spec, *dash;
+	int start, end, i, count;
 	pdf_graft_map *graft_map;
 
-	src_pagecount = fz_count_pages(ctx, (fz_document*) doc_src);
-	des_pagecount = fz_count_pages(ctx, (fz_document*) doc_des);
-	spec = fz_strsep(&range, ",");
+	count = pdf_count_pages(ctx, doc_src);
 	graft_map = pdf_new_graft_map(ctx, doc_src);
 
 	fz_try(ctx)
 	{
-		while (spec)
+		while ((range = fz_parse_page_range(ctx, range, &start, &end, count)))
 		{
-			dash = strchr(spec, '-');
-
-			if (dash == spec)
-				spage = epage = src_pagecount;
-			else
-				spage = epage = atoi(spec);
-
-			if (dash)
-			{
-				if (strlen(dash) > 1)
-					epage = atoi(dash + 1);
-				else
-					epage = src_pagecount;
-			}
-
-			spage = fz_clampi(spage, 1, src_pagecount);
-			epage = fz_clampi(epage, 1, src_pagecount);
-
-			if (spage < epage)
-				for (page = spage; page <= epage; page++, des_pagecount++)
-					page_merge(page, des_pagecount + 1, graft_map);
+			if (start < end)
+				for (i = start; i <= end; ++i)
+					page_merge(i, -1, graft_map);
 			else
-				for (page = spage; page >= epage; page--, des_pagecount++)
-					page_merge(page, des_pagecount + 1, graft_map);
-			spec = fz_strsep(&range, ",");
+				for (i = start; i >= end; --i)
+					page_merge(i, -1, graft_map);
 		}
 	}
 	fz_always(ctx)
@@ -191,8 +156,8 @@ int pdfmerge_main(int argc, char **argv)
 		{
 			pdf_drop_document(ctx, doc_src);
 			doc_src = pdf_open_document(ctx, input);
-			if (fz_optind == argc || !isrange(argv[fz_optind]))
-				merge_range("1-");
+			if (fz_optind == argc || !fz_is_page_range(ctx, argv[fz_optind]))
+				merge_range("1-N");
 			else
 				merge_range(argv[fz_optind++]);
 		}
diff --git a/source/tools/pdfpages.c b/source/tools/pdfpages.c
index 7f3577da..62aa4eeb 100644
--- a/source/tools/pdfpages.c
+++ b/source/tools/pdfpages.c
@@ -102,10 +102,9 @@ showpage(fz_context *ctx, pdf_document *doc, fz_output *out, int page)
 }
 
 static int
-showpages(fz_context *ctx, pdf_document *doc, fz_output *out, char *pagelist)
+showpages(fz_context *ctx, pdf_document *doc, fz_output *out, const char *pagelist)
 {
 	int page, spage, epage;
-	char *spec, *dash;
 	int pagecount;
 	int ret = 0;
 
@@ -113,53 +112,17 @@ showpages(fz_context *ctx, pdf_document *doc, fz_output *out, char *pagelist)
 		infousage();
 
 	pagecount = pdf_count_pages(ctx, doc);
-	spec = fz_strsep(&pagelist, ",");
-	while (spec && pagecount)
+	while ((pagelist = fz_parse_page_range(ctx, pagelist, &spage, &epage, pagecount)))
 	{
-		dash = strchr(spec, '-');
-
-		if (dash == spec)
-			spage = epage = pagecount;
-		else
-			spage = epage = atoi(spec);
-
-		if (dash)
-		{
-			if (strlen(dash) > 1)
-				epage = atoi(dash + 1);
-			else
-				epage = pagecount;
-		}
-
 		if (spage > epage)
 			page = spage, spage = epage, epage = page;
-
-		spage = fz_clampi(spage, 1, pagecount);
-		epage = fz_clampi(epage, 1, pagecount);
-
 		for (page = spage; page <= epage; page++)
-		{
 			ret |= showpage(ctx, doc, out, page);
-		}
-
-		spec = fz_strsep(&pagelist, ",");
 	}
 
 	return ret;
 }
 
-static int arg_is_page_range(const char *arg)
-{
-	int c;
-
-	while ((c = *arg++) != 0)
-	{
-		if ((c < '0' || c > '9') && (c != '-') && (c != ','))
-			return 0;
-	}
-	return 1;
-}
-
 static int
 pdfpages_pages(fz_context *ctx, fz_output *out, char *filename, char *password, char *argv[], int argc)
 {
@@ -171,11 +134,11 @@ pdfpages_pages(fz_context *ctx, fz_output *out, char *filename, char *password,
 	state = NO_FILE_OPENED;
 	while (argidx < argc)
 	{
-		if (state == NO_FILE_OPENED || !arg_is_page_range(argv[argidx]))
+		if (state == NO_FILE_OPENED || !fz_is_page_range(ctx, argv[argidx]))
 		{
 			if (state == NO_INFO_GATHERED)
 			{
-				showpages(ctx, doc, out, "1-");
+				showpages(ctx, doc, out, "1-N");
 			}
 
 			pdf_drop_document(ctx, doc);
@@ -199,7 +162,7 @@ pdfpages_pages(fz_context *ctx, fz_output *out, char *filename, char *password,
 	}
 
 	if (state == NO_INFO_GATHERED)
-		showpages(ctx, doc, out, "1-");
+		showpages(ctx, doc, out, "1-N");
 
 	pdf_drop_document(ctx, doc);