shithub: neatpost

--- a/Makefile

+++ b/Makefile

@@ -4,7 +4,7 @@

 CC = cc

 CFLAGS = -Wall -O2 "-DTROFFFDIR=\"$(FDIR)\""

 LDFLAGS =

-OBJS = post.o ps.o font.o dev.o clr.o dict.o iset.o

+OBJS = post.o ps.o font.o dev.o clr.o dict.o iset.o sbuf.o

 OBJSPDF = post.o pdf.o pdfext.o font.o dev.o clr.o dict.o iset.o sbuf.o

 all: post pdf

--- a/pdf.c

+++ b/pdf.c

@@ -733,7 +733,7 @@

 	if (lnk[0] == '#') {	/* internal links */

 		pdfout("  /A << /S /GoTo /D (%s) >>\n", lnk + 1);

 	} else {		/* external links */

-		pdfout("  /A << /S /URI /URI (%s) >>\n", lnk);

+		pdfout("  /A << /S /URI /URI %s >>\n", pdftext_static(lnk));

 	pdfout(">>\n");

 	obj_end();

@@ -786,7 +786,7 @@

 				cnt++;

 		obj_beg(objs[i]);

 		pdfout("<<\n");

-		pdfout("  /Title (%s)\n", desc[i]);

+		pdfout("  /Title %s\n", pdftext_static(desc[i]));

 		/* the parent field */

 		for (j = i - 1; j >= 0 && level[j] >= level[i]; j--)

@@ -996,9 +996,9 @@

 	info_id = obj_beg(0);

 	pdfout("<<\n");

 	if (pdf_title[0])

-		pdfout("  /Title (%s)\n", pdf_title);

+		pdfout("  /Title %s\n", pdftext_static(pdf_title));

 	if (pdf_author[0])

-		pdfout("  /Author (%s)\n", pdf_author);

+		pdfout("  /Author %s\n", pdftext_static(pdf_author));

 	pdfout("  /Creator (Neatroff)\n");

 	pdfout("  /Producer (Neatpost)\n");

 	pdfout(">>\n");

--- a/post.c

+++ b/post.c

@@ -54,18 +54,16 @@

 static int utf8len(int c)

-	if (c <= 0x7f)

+	if (~c & 0x80)		/* ASCII */

+		return c > 0;

+	if (~c & 0x40)		/* invalid UTF-8 */

 		return 1;

-	if (c >= 0xfc)

-		return 6;

-	if (c >= 0xf8)

-		return 5;

-	if (c >= 0xf0)

-		return 4;

-	if (c >= 0xe0)

-		return 3;

-	if (c >= 0xc0)

+	if (~c & 0x20)

 		return 2;

+	if (~c & 0x10)

+		return 3;

+	if (~c & 0x08)

+		return 4;

 	return 1;

@@ -546,6 +544,69 @@

 	memset(new + oldsz * memsz, 0, (newsz - oldsz) * memsz);

 	free(old);

 	return new;

+}

+/* the unicode codepoint of the given utf-8 character */

+static int utf8code(char *s)

+{

+	int c = (unsigned char) s[0];

+	if (!(c & 0x80))

+		return c;

+	if (!(c & 0x20))

+		return ((c & 0x1f) << 6) | (s[1] & 0x3f);

+	if (!(c & 0x10))

+		return ((c & 0x0f) << 12) | ((s[1] & 0x3f) << 6) | (s[2] & 0x3f);

+	if (!(c & 0x08))

+		return ((c & 0x07) << 18) | ((s[1] & 0x3f) << 12) | ((s[2] & 0x3f) << 6) | (s[3] & 0x3f);

+	return c;

+}

+static int pdftext_ascii(char *s)

+{

+	for (; *s; s++)

+		if (((unsigned char) *s) & 0x80 || *s == '(' || *s == ')')

+			return 0;

+	return 1;

+}

+/* encode s as pdf text string */

+static char *pdftext(char *s)

+{

+	struct sbuf *sb = sbuf_make();

+	if (pdftext_ascii(s)) {

+		sbuf_chr(sb, '(');

+		sbuf_str(sb, s);

+		sbuf_chr(sb, ')');

+		return sbuf_done(sb);

+	}

+	/* read utf-8 and write utf-16 */

+	sbuf_str(sb, "<FEFF");		/* unicode byte order marker */

+	while (*s) {

+		int l = utf8len((unsigned char) *s);

+		int c = utf8code(s);

+		if ((c >= 0 && c <= 0xd7ff) || (c >= 0xe000 && c <= 0xffff)) {

+			sbuf_printf(sb, "%02X%02X", c >> 8, c & 0xff);

+		}

+		if (c >= 0x010000 && c <= 0x10ffff) {

+			int c1 = 0xd800 + ((c - 0x10000) >> 10);

+			int c2 = 0xdc00 + ((c - 0x10000) & 0x3ff);

+			sbuf_printf(sb, "%02X%02X", c1 >> 8, c1 & 0xff);

+			sbuf_printf(sb, "%02X%02X", c2 >> 8, c2 & 0xff);

+		}

+		s += l;

+	}

+	sbuf_chr(sb, '>');

+	return sbuf_done(sb);

+}

+/* encode s as pdf text string; returns a static buffer */

+char *pdftext_static(char *s)

+{

+	static char buf[1024];

+	char *r = pdftext(s);

+	snprintf(buf, sizeof(buf), "%s", r);

+	free(r);

+	return buf;

 static char *usage =

--- a/post.h

+++ b/post.h

@@ -110,6 +110,8 @@

 /* memory allocation */

 void *mextend(void *old, long oldsz, long newsz, int memsz);

+/* helper functions */

+char *pdftext_static(char *s);

 /* string buffers */

 struct sbuf *sbuf_make(void);

--- a/ps.c

+++ b/ps.c

@@ -317,9 +317,9 @@

 			lnk[0] == '#' ? lnk + 1 : lnk);

 	} else {

 		outf("[ /Rect [ %d %d t %d %d t ] "

-			"/Action << /Subtype /URI /URI (%s) >> /Open true "

+			"/Action << /Subtype /URI /URI %s >> /Open true "

 			"/Subtype /Link /LNK pdfmark\n",

-			o_h, o_v, o_h + hwid, o_v + vwid, lnk);

+			o_h, o_v, o_h + hwid, o_v + vwid, pdftext_static(lnk));

@@ -346,7 +346,7 @@

 		for (j = i + 1; j < n && level[j] > level[i]; j++)

 			if (level[j] == level[i] + 1)

 				cnt++;

-		outf("[ /Title (%s)", desc[i]);

+		outf("[ /Title %s", pdftext_static(desc[i]));

 		if (page[i] > 0)

 			outf(" /Page %d", page[i]);

 		if (cnt > 0)

@@ -385,9 +385,9 @@

 	out("[");

 	if (ps_title[0])

-		out(" /Title (%s)", ps_title);

+		out(" /Title %s", pdftext_static(ps_title));

 	if (ps_author[0])

-		out(" /Author (%s)", ps_author);

+		out(" /Author %s", pdftext_static(ps_author));

 	out(" /Creator (Neatroff) /DOCINFO pdfmark\n");

 	out("%%%%Trailer\n");

 	out("done\n");