* add hack for utf16 to ascii

* remove debug statements
* add invalid tags feedback

.gitignore

@@ -58,3 +58,4 @@ dist/
 spec/
 compile_commands.json
 examples/bin
+.data/

src/include/id3v2/types.h

@@ -86,6 +86,7 @@ typedef struct {
     id3v2_frame_t *frame;
     id3v2_frame_encoding_t encoding;
     char *text;
+    size_t size;
 } id3v2_frame_text_t;
 
 typedef struct {

src/include/id3v2/unicode.h

@@ -0,0 +1,13 @@
+#ifndef ID3V2_UNICODE_H
+#define ID3V2_UNICODE_H
+
+#include <stdio.h>
+
+/**
+* FIXME: hack for utf16 strings
+*/
+void utf16_to_ascii(const char *buf, size_t buf_size, char **out, size_t *out_size);
+
+
+#endif // ID3V2_UNICODE_H
+

src/lib/CMakeLists.txt

@@ -10,9 +10,9 @@ pkg_search_module(LIBSODIUM OPTIONAL libsodium)
 
 
 set(SOURCES
-#  ${id3v2lib_SOURCE_DIR}/src/include/id3v2/types.h
   synch.c
   helpers.c
+  unicode.c
   id3v2.c
   )
 
@@ -53,6 +53,7 @@ message("-- CMAKE_SYSTEM_NAME:      ${CMAKE_SYSTEM_NAME}")
 message("-- CMAKE_SYSTEM_PROCESSOR: ${CMAKE_SYSTEM_PROCESSOR}")
 message("-- CMAKE_SYSTEM:           ${CMAKE_SYSTEM}")
 message("-- CMAKE_OS_NAME:          ${CMAKE_OS_NAME}")
+message("-- CMAKE_INSTALL_PREFIX:   ${CMAKE_INSTALL_PREFIX}")
 message("-- HAS_APT:                ${HAS_APT}")
 message("-- HAS_RPM:                ${HAS_RPM}")
 

src/lib/id3v2.c

@@ -1,6 +1,7 @@
 #include "id3v2.h"
 #include "id3v2/synch.h"
 #include "id3v2/types.h"
+#include "id3v2/unicode.h"
 
 static id3v2_frame_list_t *scan_frames(id3v2_tag_t *t);
 
@@ -22,13 +23,15 @@ id3v2_tag_t *id3v2_from_file(const char *file_name) {
 
   if (fread(&header, sizeof(char), ID3_HEADER, fp) != ID3_HEADER) {
     puts("unable to read id3 header");
-    exit(1);
+    // exit(1);
+    goto error;
   }
 
   // check id3 tag marker
   if ((header[0] != 'I') || (header[1] != 'D') || (header[2] != '3')) {
     puts("invalid id3 header");
-    exit(1);
+    // exit(1);
+    goto error;
   }
 
   // get major version_major byte
@@ -41,7 +44,7 @@ id3v2_tag_t *id3v2_from_file(const char *file_name) {
   // decode the tag size
   t->tag_size = synch_decode(int_decode(header, 4, 6));
 
-  printf("Tag size: %u \n", t->tag_size);
+  // fprintf(stderr, "Tag size: %u \n", t->tag_size);
 
   fseek(fp, 10, SEEK_SET);
 
@@ -62,6 +65,13 @@ id3v2_tag_t *id3v2_from_file(const char *file_name) {
   fclose(fp);
 
   return t;
+
+error:
+  if (t != NULL) {
+    id3v2_tag_free(t);
+  }
+  fclose(fp);
+  return NULL;
 }
 
 id3v2_tag_t *id3v2_from_buffer(const char *buf) {
@@ -116,7 +126,7 @@ id3v2_frame_t *parse_frame(id3v2_tag_t *t, size_t *cursor_pos) {
 
   // validate the tag and return NULL for invalid
   if (!_id3v2_validate_frame_tag(tag)) {
-    printf("Invalid frame tag: position: %lu Tag: %s \n", *cursor_pos, tag);
+    // fprintf(stderr, "Invalid frame tag: position: %lu Tag: %s \n", *cursor_pos, tag);
     return NULL;
   }
 
@@ -130,23 +140,34 @@ id3v2_frame_t *parse_frame(id3v2_tag_t *t, size_t *cursor_pos) {
   *cursor_pos += 4;
 
   f->size = int_decode((unsigned char *)t->tag_buffer, 4, *cursor_pos);
+  // fprintf(stderr, "frame size: %d \n", f->size);
   if (t->version_major == 4) {
     // version_major 4 gets the synch safe size
     f->size = synch_decode(f->size);
+    fprintf(stderr, "sync safe frame: %d \n", f->size);
   }
 
   // move the cursor forward
   *cursor_pos += 4;
 
   // get the flags
   memcpy(&f->flags, t->tag_buffer + *cursor_pos, 2);
-
   *cursor_pos += 2;
 
-  // allocate and copy the frame buffer
-  f->buffer = calloc(f->size, sizeof(char));
-  memcpy(f->buffer, t->tag_buffer + *cursor_pos, f->size);
+  // if (strcmp(tag, "TIT2") == 0) {
+  //   fprintf(stderr, "Found title \n");
+  //   fprintf(stderr, "flags: %x %x\n", f->flags[0],  f->flags[1]);
+  //   fprintf(stderr, "first bytes: %x %x\n", (t->tag_buffer + *cursor_pos)[0], (t->tag_buffer + *cursor_pos)[1]);
+  //   for(size_t ii=0; ii < f->size; ii++) {
+  //     fprintf(stderr, "0x%.2x\n", (t->tag_buffer + *cursor_pos)[ii]);
+  //   }
+  // }
 
+  // allocate and copy the frame buffer
+  // f->buffer = calloc(f->size + 1, sizeof(char));
+  f->buffer = malloc(sizeof(char) * f->size);
+  memcpy(f->buffer, t->tag_buffer + (*cursor_pos), (size_t)f->size);
+
   *cursor_pos += f->size;
   //  printf("func cursor pos: %lu \n", *cursor_pos);
 
@@ -205,39 +226,51 @@ id3v2_frame_text_t *id3v2_frame_text(id3v2_frame_t *f) {
 
   // buffer position
   size_t cursor = 0;
-  // size of the null terminator to apply
-  size_t null_term_size = 1;
+
+  if (f->size <= 2) {
+    t->text = NULL;
+    t->encoding = LATIN1;
+    return t;
+  }
 
   // determine encoding
   // TODO: utf byte order marker?
+  // FIXME: using a hack for utf16, we should make this 
+  // more robust or use a library
   switch (f->buffer[cursor]) {
   case 0x00:
     t->encoding = LATIN1;
     break;
   case 0x01:
     t->encoding = UTF16;
-    null_term_size = 2;
     break;
   case 0x02:
     t->encoding = UTF16BE;
-    null_term_size = 2;
     break;
   case 0x03:
     t->encoding = UTF8;
-    null_term_size = 2;
     break;
   default:
-    // printf("Illegal encoding for frame: %s \n", f->tag);
     t->encoding = LATIN1;
   }
   cursor++;
 
-  if (f->size > 2) {
-    printf("text frame size: %u \n", f->size);
-    t->text = calloc(f->size, sizeof(char));
+  switch(t->encoding) {
+  case UTF16:{
+      char *out = NULL;
+      size_t out_size = 0;
+      utf16_to_ascii(f->buffer, f->size, &out, &out_size); 
+      t->encoding = LATIN1;
+      t->text = out;
+      t->size = out_size;
+      // fprintf(stderr, "Output: %s\n", t->text);
+      break;
+  }
+  default: // LATIN1
+    t->text = malloc(sizeof(char) * f->size);
+    t->size = f->size;
     memcpy(t->text, f->buffer + cursor, f->size);
-  } else {
-    t->text = NULL;
+    break;
   }
 
   return t;
@@ -435,6 +468,7 @@ static void write_frame_header(id3v2_tag_t *t, id3v2_frame_t *f,
 void id3v2_tag_write_to_buffer(id3v2_tag_t *t, unsigned char **buffer,
                                size_t *size) {
 
+  // start with the known header size
   size_t total = 10;
   total += id3v2_tag_total_frame_size(t);
 

src/lib/unicode.c

@@ -0,0 +1,26 @@
+#include "id3v2/unicode.h"
+#include <stdlib.h>
+#include <string.h>
+
+
+void utf16_to_ascii(const char *buf, size_t buf_size, char **out, size_t *out_size) {
+  // create output
+  *out_size = (buf_size / 2);
+  char *o = (char *)malloc(sizeof(char) * (*out_size));
+  memset(o, 0, *out_size);
+
+  // output byte index
+  size_t cidx = 0;
+
+  for(size_t i=3; i < buf_size; i++) {
+    if (buf[i] == 0x00) {
+      continue;
+    }
+    o[cidx] = buf[i];
+    cidx++;
+  }
+
+  // swap address with given buffer
+  *out = o;
+
+}

src/test/CMakeLists.txt

@@ -16,3 +16,6 @@ include_directories(../lib)
 add_executable(${PROJECT_NAME} ${SOURCES})
 
 target_link_libraries(${PROJECT_NAME} m id3v2)
+
+add_executable(${PROJECT_NAME}_unicode unicode.c)
+target_link_libraries(${PROJECT_NAME}_unicode m id3v2)

src/test/main.c

@@ -83,7 +83,7 @@ void write_id3_buffer(id3v2_tag_t *t, const char *save_to,
   fclose(fp);
 }
 
-void utf16_to_ascii(const unsigned char *utf16_buf, size_t utf16_size,
+void ___utf16_to_ascii(const unsigned char *utf16_buf, size_t utf16_size,
                     unsigned char **buf, size_t *size) {
 
   // check if there is a BOM, if so we skip
@@ -160,12 +160,12 @@ int main(int argc, char **argv) {
 
   id3v2_frame_text_t *tt;
 
-  id3v2_tag_write_artist(tag, "Super Sublime");
-  tt = id3v2_tag_artist(tag);
-  if (tt != NULL) {
-    printf("Artist: %s \n", tt->text);
-    id3v2_tag_free_text_frame(tt);
-  }
+  // id3v2_tag_write_artist(tag, "Super Sublime");
+  // tt = id3v2_tag_artist(tag);
+  // if (tt != NULL) {
+  //   printf("Artist: %s \n", tt->text);
+  //   id3v2_tag_free_text_frame(tt);
+  // }
 
   tt = id3v2_tag_album(tag);
   if (tt != NULL) {
@@ -179,6 +179,12 @@ int main(int argc, char **argv) {
     id3v2_tag_free_text_frame(tt);
   }
 
+  tt = id3v2_tag_title(tag);
+  if (tt != NULL) {
+    printf("Title: %s \n", tt->text);
+    id3v2_tag_free_text_frame(tt);
+  }
+
   // update title
   id3v2_tag_write_title(tag, "an updated title");
   tt = id3v2_tag_title(tag);

src/test/unicode.c

@@ -0,0 +1,26 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include "../include/id3v2/unicode.h"
+/*
+  test a utf16 hack that only removes 0 byte indexes
+*/
+
+static char tb[] = {
+    0x01, 0xff, 0xfe, 0x4c, 0x00, 0x6f, 0x00, 0x6f, 0x00, 0x6b, 0x00, 0x20,
+    0x00, 0x41, 0x00, 0x74, 0x00, 0x20, 0x00, 0x4d, 0x00, 0x65, 0x00, 0x20,
+    0x00, 0x4e, 0x00, 0x6f, 0x00, 0x77, 0x00, 0x20, 0x00, 0x28, 0x00, 0x66,
+    0x00, 0x65, 0x00, 0x61, 0x00, 0x74, 0x00, 0x75, 0x00, 0x72, 0x00, 0x69,
+    0x00, 0x6e, 0x00, 0x67, 0x00, 0x20, 0x00, 0x4c, 0x00, 0x69, 0x00, 0x6c,
+    0x00, 0x20, 0x00, 0x57, 0x00, 0x61, 0x00, 0x79, 0x00, 0x6e, 0x00, 0x65,
+    0x00, 0x20, 0x00, 0x26, 0x00, 0x20, 0x00, 0x42, 0x00, 0x75, 0x00, 0x73,
+    0x00, 0x74, 0x00, 0x61, 0x00, 0x20, 0x00, 0x52, 0x00, 0x68, 0x00, 0x79,
+    0x00, 0x6d, 0x00, 0x65, 0x00, 0x73, 0x00, 0x29, 0x00};
+
+int main(int argc, char **argv) {
+  puts("unicode main");
+  char *out = NULL;
+  size_t out_size = 0;
+  utf16_to_ascii(tb, 105, &out, &out_size);
+  fprintf(stderr, "Size: %zu, Text: %s\n", out_size, out);
+  return EXIT_SUCCESS;
+}