cwebp: add PNG metadata extraction

the values of XMP/EXIF/ICC are currently unused.

Change-Id: I39d78b9a179f2d36c9c8ea12776bfdf6d8d18663

examples/cwebp.c

@@ -327,7 +327,7 @@ static int ReadPicture(const char* const filename, WebPPicture* const pic,
     // If no size specified, try to decode it as PNG/JPEG (as appropriate).
     const InputFileFormat format = GetImageType(in_file);
     if (format == PNG_) {
-      ok = ReadPNG(in_file, pic, keep_alpha);
+      ok = ReadPNG(in_file, pic, keep_alpha, metadata);
     } else if (format == JPEG_) {
       ok = ReadJPEG(in_file, pic);
     } else if (format == TIFF_) {

examples/pngdec.c

@@ -21,15 +21,173 @@
 #include <stdlib.h>
 
 #include "webp/encode.h"
+#include "./metadata.h"
 
 static void PNGAPI error_function(png_structp png, png_const_charp dummy) {
   (void)dummy;  // remove variable-unused warning
   longjmp(png_jmpbuf(png), 1);
 }
 
-int ReadPNG(FILE* in_file, WebPPicture* const pic, int keep_alpha) {
+// Converts the NULL terminated 'hexstring' which contains 2-byte character
+// representations of hex values to raw data.
+// 'hexstring' may contain values consisting of [A-F][a-f][0-9] in pairs,
+// e.g., 7af2..., separated by any number of newlines.
+// 'expected_length' is the anticipated processed size.
+// On success the raw buffer is returned with its length equivalent to
+// 'expected_length'. NULL is returned if the processed length is less than
+// 'expected_length' or any character aside from those above is encountered.
+// The returned buffer must be freed by the caller.
+static uint8_t* HexStringToBytes(const char* hexstring,
+                                 size_t expected_length) {
+  const char* src = hexstring;
+  size_t actual_length = 0;
+  uint8_t* const raw_data = (uint8_t*)malloc(expected_length);
+  uint8_t* dst;
+
+  if (raw_data == NULL) return NULL;
+
+  for (dst = raw_data; actual_length < expected_length && *src != '\0'; ++src) {
+    char* end;
+    char val[3];
+    if (*src == '\n') continue;
+    val[0] = *src++;
+    val[1] = *src;
+    val[2] = '\0';
+    *dst++ = (uint8_t)strtol(val, &end, 16);
+    if (end != val + 2) break;
+    ++actual_length;
+  }
+
+  if (actual_length != expected_length) {
+    free(raw_data);
+    return NULL;
+  }
+  return raw_data;
+}
+
+static int ProcessRawProfile(const char* profile, size_t profile_len,
+                             MetadataPayload* const payload) {
+  const char* src = profile;
+  char* end;
+  int expected_length;
+
+  if (profile == NULL || profile_len == 0) return 0;
+
+  // ImageMagick formats 'raw profiles' as
+  // '\n<name>\n<length>(%8lu)\n<hex payload>\n'.
+  if (*src != '\n') {
+    fprintf(stderr, "Malformed raw profile, expected '\\n' got '\\x%.2X'\n",
+            *src);
+    return 0;
+  }
+  ++src;
+  // skip the profile name and extract the length.
+  while (*src != '\0' && *src++ != '\n') {}
+  expected_length = (int)strtol(src, &end, 10);
+  if (*end != '\n') {
+    fprintf(stderr, "Malformed raw profile, expected '\\n' got '\\x%.2X'\n",
+            *end);
+    return 0;
+  }
+  ++end;
+
+  // 'end' now points to the profile payload.
+  payload->bytes = HexStringToBytes(end, expected_length);
+  if (payload->bytes == NULL) return 0;
+  payload->size = expected_length;
+  return 1;
+}
+
+static const struct {
+  const char* name;
+  int (*process)(const char* profile, size_t profile_len,
+                 MetadataPayload* const payload);
+  size_t storage_offset;
+} kPNGMetadataMap[] = {
+  // http://www.sno.phy.queensu.ca/~phil/exiftool/TagNames/PNG.html#TextualData
+  // See also: ExifTool on CPAN.
+  { "Raw profile type exif", ProcessRawProfile, METADATA_OFFSET(exif) },
+  { "Raw profile type xmp",  ProcessRawProfile, METADATA_OFFSET(xmp) },
+  // XMP Specification Part 3, Section 3 #PNG
+  { "XML:com.adobe.xmp",     MetadataCopy,      METADATA_OFFSET(xmp) },
+  { NULL, NULL, 0 },
+};
+
+// Looks for metadata at both the beginning and end of the PNG file, giving
+// preference to the head.
+// Returns true on success. The caller must use MetadataFree() on 'metadata' in
+// all cases.
+static int ExtractMetadataFromPNG(png_structp png,
+                                  png_infop const head_info,
+                                  png_infop const end_info,
+                                  Metadata* const metadata) {
+  int p;
+
+  for (p = 0; p < 2; ++p)  {
+    png_infop const info = (p == 0) ? head_info : end_info;
+    png_textp text = NULL;
+    const int num = png_get_text(png, info, &text, NULL);
+    int i;
+    // Look for EXIF / XMP metadata.
+    for (i = 0; i < num; ++i, ++text) {
+      int j;
+      for (j = 0; kPNGMetadataMap[j].name != NULL; ++j) {
+        if (!strcmp(text->key, kPNGMetadataMap[j].name)) {
+          MetadataPayload* const payload =
+              (MetadataPayload*)((uint8_t*)metadata +
+                                 kPNGMetadataMap[j].storage_offset);
+          png_size_t text_length;
+          switch (text->compression) {
+#ifdef PNG_iTXt_SUPPORTED
+            case PNG_ITXT_COMPRESSION_NONE:
+            case PNG_ITXT_COMPRESSION_zTXt:
+              text_length = text->itxt_length;
+              break;
+#endif
+            default:
+            case PNG_TEXT_COMPRESSION_NONE:
+            case PNG_TEXT_COMPRESSION_zTXt:
+              text_length = text->text_length;
+              break;
+          }
+          if (payload->bytes != NULL) {
+            fprintf(stderr, "Ignoring additional '%s'\n", text->key);
+          } else if (!kPNGMetadataMap[j].process(text->text, text_length,
+                                                 payload)) {
+            fprintf(stderr, "Failed to process: '%s'\n", text->key);
+            return 0;
+          }
+          break;
+        }
+      }
+    }
+    // Look for an ICC profile.
+    {
+      png_charp name;
+      int comp_type;
+#if ((PNG_LIBPNG_VER_MAJOR << 8) | PNG_LIBPNG_VER_MINOR << 0) < \
+    ((1 << 8) | (5 << 0))
+      png_charp profile;
+#else  // >= libpng 1.5.0
+      png_bytep profile;
+#endif
+      png_uint_32 len;
+
+      if (png_get_iCCP(png, info,
+                       &name, &comp_type, &profile, &len) == PNG_INFO_iCCP) {
+        if (!MetadataCopy((const char*)profile, len, &metadata->iccp)) return 0;
+      }
+    }
+  }
+
+  return 1;
+}
+
+int ReadPNG(FILE* in_file, WebPPicture* const pic, int keep_alpha,
+            Metadata* const metadata) {
   png_structp png;
-  png_infop info;
+  png_infop info = NULL;
+  png_infop end_info = NULL;
   int color_type, bit_depth, interlaced;
   int has_alpha;
   int num_passes;
@@ -47,13 +205,16 @@ int ReadPNG(FILE* in_file, WebPPicture* const pic, int keep_alpha) {
   png_set_error_fn(png, 0, error_function, NULL);
   if (setjmp(png_jmpbuf(png))) {
  Error:
-    png_destroy_read_struct(&png, NULL, NULL);
+    MetadataFree(metadata);
+    png_destroy_read_struct(&png, &info, &end_info);
     free(rgb);
     goto End;
   }
 
   info = png_create_info_struct(png);
   if (info == NULL) goto Error;
+  end_info = png_create_info_struct(png);
+  if (end_info == NULL) goto Error;
 
   png_init_io(png, in_file);
   png_read_info(png, info);
@@ -94,8 +255,15 @@ int ReadPNG(FILE* in_file, WebPPicture* const pic, int keep_alpha) {
       png_read_rows(png, &row, NULL, 1);
     }
   }
-  png_read_end(png, info);
+  png_read_end(png, end_info);
-  png_destroy_read_struct(&png, &info, NULL);
+
+  if (metadata != NULL &&
+      !ExtractMetadataFromPNG(png, info, end_info, metadata)) {
+    fprintf(stderr, "Error extracting PNG metadata!\n");
+    goto Error;
+  }
+
+  png_destroy_read_struct(&png, &info, &end_info);
 
   pic->width = width;
   pic->height = height;
@@ -103,7 +271,10 @@ int ReadPNG(FILE* in_file, WebPPicture* const pic, int keep_alpha) {
                  : WebPPictureImportRGB(pic, rgb, stride);
   free(rgb);
 
-  if (ok && has_alpha && keep_alpha == 2) {
+  if (!ok) {
+    goto Error;
+  }
+  if (has_alpha && keep_alpha == 2) {
     WebPCleanupTransparentArea(pic);
   }
 
@@ -111,10 +282,12 @@ int ReadPNG(FILE* in_file, WebPPicture* const pic, int keep_alpha) {
   return ok;
 }
 #else  // !WEBP_HAVE_PNG
-int ReadPNG(FILE* in_file, struct WebPPicture* const pic, int keep_alpha) {
+int ReadPNG(FILE* in_file, struct WebPPicture* const pic, int keep_alpha,
+            struct Metadata* const metadata) {
   (void)in_file;
   (void)pic;
   (void)keep_alpha;
+  (void)metadata;
   fprintf(stderr, "PNG support not compiled. Please install the libpng "
           "development package before building.\n");
   return 0;

examples/pngdec.h

@@ -16,13 +16,15 @@
 extern "C" {
 #endif
 
+struct Metadata;
 struct WebPPicture;
 
 // Reads a PNG from 'in_file', returning the decoded output in 'pic'.
 // If 'keep_alpha' is true and the PNG has an alpha channel, the output is RGBA
 // otherwise it will be RGB.
 // Returns true on success.
-int ReadPNG(FILE* in_file, struct WebPPicture* const pic, int keep_alpha);
+int ReadPNG(FILE* in_file, struct WebPPicture* const pic, int keep_alpha,
+            struct Metadata* const metadata);
 
 #if defined(__cplusplus) || defined(c_plusplus)
 }    // extern "C"