jpeginfo: decode basic TIFF tag names

tools/jpeginfo.c

@@ -23,14 +23,7 @@
 #include <stdbool.h>
 #include <stdarg.h>
 
-// --- TIFF --------------------------------------------------------------------
+// --- Utilities ---------------------------------------------------------------
-// https://www.adobe.io/content/dam/udp/en/open/standards/tiff/TIFF6.pdf
-// https://www.adobe.io/content/dam/udp/en/open/standards/tiff/TIFFPM6.pdf
-// https://www.cipa.jp/std/documents/e/DC-008-2012_E.pdf
-//
-// libtiff is a mess, and the format is not particularly complicated.
-// Also, we'd still want to duplicate its tag tables.
-// Exif libraries are senselessly copylefted.
 
 static uint32_t
 u32be(const uint8_t *p)
@@ -56,13 +49,20 @@ u16le(const uint8_t *p)
 	return (uint16_t) p[1] << 8 | p[0];
 }
 
+// --- TIFF --------------------------------------------------------------------
+// https://www.adobe.io/content/dam/udp/en/open/standards/tiff/TIFF6.pdf
+// https://www.adobe.io/content/dam/udp/en/open/standards/tiff/TIFFPM6.pdf
+// https://www.cipa.jp/std/documents/e/DC-008-2012_E.pdf
+//
+// libtiff is a mess, and the format is not particularly complicated.
+// Also, we'd still want to duplicate its tag tables.
+// Exif libraries are senselessly copylefted.
+
 static struct un {
 	uint32_t (*u32) (const uint8_t *);
 	uint16_t (*u16) (const uint8_t *);
 } unbe = {u32be, u16be}, unle = {u32le, u16le};
 
-// - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
-
 struct tiffer {
 	struct un *un;
 	const uint8_t *begin, *p, *end;
@@ -324,6 +324,120 @@ tiffer_next_entry(struct tiffer *self, struct tiffer_entry *entry)
 	return true;
 }
 
+// --- TIFF/Exif/MPF/* tags ----------------------------------------------------
+
+struct tiff_entry {
+	const char *name;
+	uint16_t tag;
+};
+
+static struct tiff_entry tiff_entries[] = {
+	{"NewSubfileType", 254},
+	{"SubfileType", 255},
+	{"ImageWidth", 256},
+	{"ImageLength", 257},
+	{"BitsPerSample", 258},
+	{"Compression", 259},
+	{"PhotometricInterpretation", 262},
+	{"Threshholding", 263},
+	{"CellWidth", 264},
+	{"CellLength", 265},
+	{"FillOrder", 266},
+	{"DocumentName", 269},
+	{"ImageDescription", 270},
+	{"Make", 271},
+	{"Model", 272},
+	{"StripOffsets", 273},
+	{"Orientation", 274},
+	{"SamplesPerPixel", 277},
+	{"RowsPerStrip", 278},
+	{"StripByteCounts", 279},
+	{"MinSampleValue", 280},
+	{"MaxSampleValue", 281},
+	{"XResolution", 282},
+	{"YResolution", 283},
+	{"PlanarConfiguration", 284},
+	{"PageName", 285},
+	{"XPosition", 286},
+	{"YPosition", 287},
+	{"FreeOffsets", 288},
+	{"FreeByteCounts", 289},
+	{"GrayResponseUnit", 290},
+	{"GrayResponseCurve", 291},
+	{"T4Options", 292},
+	{"T6Options", 293},
+	{"ResolutionUnit", 296},
+	{"PageNumber", 297},
+	{"TransferFunction", 301},
+	{"Software", 305},
+	{"DateTime", 306},
+	{"Artist", 315},
+	{"HostComputer", 316},
+	{"Predictor", 317},
+	{"WhitePoint", 318},
+	{"PrimaryChromaticities", 319},
+	{"ColorMap", 320},
+	{"HalftoneHints", 321},
+	{"TileWidth", 322},
+	{"TileLength", 323},
+	{"TileOffsets", 324},
+	{"TileByteCounts", 325},
+	{"InkSet", 332},
+	{"InkNames", 333},
+	{"NumberOfInks", 334},
+	{"DotRange", 336},
+	{"TargetPrinter", 337},
+	{"ExtraSamples", 338},
+	{"SampleFormat", 339},
+	{"SMinSampleValue", 340},
+	{"SMaxSampleValue", 341},
+	{"TransferRange", 342},
+	{"JPEGProc", 512},
+	{"JPEGInterchangeFormat", 513},
+	{"JPEGInterchangeFormatLngth", 514},
+	{"JPEGRestartInterval", 515},
+	{"JPEGLosslessPredictors", 517},
+	{"JPEGPointTransforms", 518},
+	{"JPEGQTables", 519},
+	{"JPEGDCTables", 520},
+	{"JPEGACTables", 521},
+	{"YCbCrCoefficients", 529},
+	{"YCbCrSubSampling", 530},
+	{"YCbCrPositioning", 531},
+	{"ReferenceBlackWhite", 532},
+	{"Copyright", 33432},
+	{}
+};
+
+// Compression
+static struct tiff_entry tiff_compression_values[] = {
+	{"Uncompressed", 1},
+	{"CCITT 1D", 2},
+	{"Group 3 Fax", 3},
+	{"Group 4 Fax", 4},
+	{"LZW", 5},
+	{"JPEG", 6},
+	{"PackBits", 32773},
+	{}
+};
+
+// PhotometricInterpretation
+static struct tiff_entry tiff_photometric_interpretation_values[] = {
+	{"WhiteIsZero", 0},
+	{"BlackIsZero", 1},
+	{"RGB", 2},
+	{"RGB Palette", 3},
+	{"Transparency mask", 4},
+	{"CMYK", 5},
+	{"YCbCr", 6},
+	{"CIELab", 8},
+	{}
+};
+
+// TODO(p): Insert tags and values from other documentation,
+// so far only Appendix A from TIFF 6.0 is present.
+// There are still quite a few missing constant names from there.
+
 // --- Analysis ----------------------------------------------------------------
 
 static jv
@@ -351,6 +465,17 @@ add_error(jv o, const char *message)
 
 // --- Exif --------------------------------------------------------------------
 
+// TODO(p): Decode more and better.
+static jv
+process_exif_entry(jv o, const struct tiffer_entry *entry)
+{
+	for (const struct tiff_entry *p = tiff_entries; p->name; p++) {
+		if (p->tag == entry->tag)
+			return add_to_subarray(o, "TIFF", jv_string(p->name));
+	}
+	return add_to_subarray(o, "TIFF", jv_number(entry->tag));
+}
+
 static jv
 parse_exif(jv o, const uint8_t *p, size_t len)
 {
@@ -358,11 +483,11 @@ parse_exif(jv o, const uint8_t *p, size_t len)
 	if (!tiffer_init(&T, p, len))
 		return add_warning(o, "invalid Exif");
 
-	// TODO(p): Decode more and better.
+	// TODO(p): Turn this into an array of objects indexed by tag name.
 	struct tiffer_entry entry;
 	while (tiffer_next_ifd(&T)) {
 		while (tiffer_next_entry(&T, &entry)) {
-			o = add_to_subarray(o, "TIFF", jv_number(entry.tag));
+			o = process_exif_entry(o, &entry);
 		}
 	}
 	return o;
@@ -467,7 +592,7 @@ parse_icc(jv o, const uint8_t *profile, size_t profile_len)
 // https://www.adobe.com/devnet-apps/photoshop/fileformatashtml/
 
 static jv
-parse_psir_block(jv o, uint16_t resource_id, const char *name,
+process_psir(jv o, uint16_t resource_id, const char *name,
 	const uint8_t *data, size_t len)
 {
 	// TODO(p): These is more to extract here. The name is most often empty.
@@ -477,45 +602,51 @@ parse_psir_block(jv o, uint16_t resource_id, const char *name,
 	return add_to_subarray(o, "PSIR", jv_number(resource_id));
 }
 
+static jv
+parse_psir_block(jv o, const uint8_t *p, size_t len, size_t *advance)
+{
+	*advance = 0;
+	if (len < 8 || memcmp(p, "8BIM", 4))
+		return add_warning(o, "bad PSIR block header");
+
+	uint16_t resource_id = u16be(p + 4);
+	uint8_t name_len = p[6];
+	const uint8_t *name = &p[7];
+
+	// Add one byte for the Pascal-ish string length prefix,
+	// then another one for padding to make the length even.
+	size_t name_len_full = (name_len + 2) & ~1U;
+
+	size_t resource_len_offset = 6 + name_len_full,
+		header_len = resource_len_offset + 4;
+	if (len < header_len)
+		return add_warning(o, "bad PSIR block header");
+
+	uint32_t resource_len = u32be(p + resource_len_offset);
+	size_t resource_len_padded = (resource_len + 1) & ~1U;
+	if (resource_len_padded < resource_len ||
+		len < header_len + resource_len_padded)
+		return add_warning(o, "runaway PSIR block");
+
+	char *cname = calloc(1, name_len_full);
+	strncpy(cname, (const char *) name, name_len);
+	o = process_psir(o, resource_id, cname, p + header_len, resource_len);
+	free(cname);
+
+	*advance = header_len + resource_len_padded;
+	return o;
+}
+
 static jv
 parse_psir(jv o, const uint8_t *p, size_t len)
 {
 	if (len == 0)
 		return add_warning(o, "empty PSIR data");
 
-	while (len) {
+	size_t advance = 0;
-		if (len < 8 || memcmp(p, "8BIM", 4))
+	while (len && (o = parse_psir_block(o, p, len, &advance), advance)) {
-			return add_warning(o, "bad PSIR block header");
+		p += advance;
-
+		len -= advance;
-		uint16_t resource_id = u16be(p + 4);
-		uint8_t name_len = p[6];
-		const uint8_t *name = &p[7];
-
-		// Add one byte for the Pascal-ish string length prefix,
-		// then another one for padding to make the length even.
-		size_t name_len_full = (name_len + 2) & ~1U;
-
-		size_t resource_len_offset = 6 + name_len_full,
-			header_len = resource_len_offset + 4;
-		if (len < header_len)
-			return add_warning(o, "bad PSIR block header");
-
-		uint32_t resource_len = u32be(p + resource_len_offset);
-		size_t resource_len_padded = (resource_len + 1) & ~1U;
-		if (resource_len_padded < resource_len ||
-			len < header_len + resource_len_padded)
-			return add_warning(o, "runaway PSIR block");
-
-		p += header_len;
-		len -= header_len;
-
-		char *cname = calloc(1, name_len_full);
-		strncpy(cname, (const char *) name, name_len);
-		o = parse_psir_block(o, resource_id, cname, p, resource_len);
-		free(cname);
-
-		p += resource_len_padded;
-		len -= resource_len_padded;
 	}
 	return o;
 }