Support for reading <MultiComment> from XML files and writing those to the file. Untested. <Language> elements are now checked for a valid ISO639-2 code.

doc/matroska-tags.xml

@@ -19,7 +19,8 @@
     <General>
       <Subject>Subject</Subject>
       <Bibliography>Bibliography</Bibliography>
-      <Language>Language</Language>
+      <!-- All languages must be valid ISO639-2 codes. -->
+      <Language>ger</Language>
       <Rating>QUJDREVG</Rating>
       <Encoder>Encoder</Encoder>
       <EncodeSettings>EncodeSettings</EncodeSettings>
@@ -131,7 +132,7 @@
         <Address>Address</Address>
         <URL>URL</URL>
         <Email>Email</Email>
-        <Language>Language</Language>
+        <Language>eng</Language>
       </Title>
       <!-- Let's test whether a complete second Title entry works. -->
       <Title>
@@ -142,7 +143,7 @@
         <Address>Address2</Address>
         <URL>URL2</URL>
         <Email>Email2</Email>
-        <Language>Language2</Language>
+        <Language>fre</Language>
       </Title>
     </MultiTitle>
 
@@ -150,7 +151,7 @@
 <!--     <MultiComment> -->
 <!--       <Name>CommentName</Name> -->
 <!--       <Comments>Comments</Comments> -->
-<!--       <Language>Language</Language> -->
+<!--       <Language>ita</Language> -->
 <!--     </MultiComment> -->
 
   </Tag>

src/iso639.cpp

@@ -535,7 +535,7 @@ iso639_language_t iso639_languages[] =
    {"Zuni", "zun", NULL},
    {NULL, NULL, NULL}};
 
-char *get_iso639_english_name(char *iso639_2_code) {
+char *get_iso639_english_name(const char *iso639_2_code) {
   int i;
 
   i = 0;
@@ -548,7 +548,7 @@ char *get_iso639_english_name(char *iso639_2_code) {
   return NULL;
 }
 
-int is_valid_iso639_2_code(char *iso639_2_code) {
+int is_valid_iso639_2_code(const char *iso639_2_code) {
   int i;
 
   i = 0;

src/iso639.h

@@ -27,8 +27,8 @@ typedef struct {
 
 extern iso639_language_t iso639_language_list[];
 
-int is_valid_iso639_2_code(char *iso639_2_code);
+int is_valid_iso639_2_code(const char *iso639_2_code);
-char *get_iso639_english_name(char *iso639_2_code);
+char *get_iso639_english_name(const char *iso639_2_code);
 void list_iso639_languages();
 
 #endif // __ISO639_H

src/tagparser.h

@@ -115,8 +115,9 @@ using namespace libmatroska;
 #define E_MultiComment                   77
 #define E_CommentName                    78
 #define E_Comments                       79
+#define E_CommentLanguage                81
 
-// MAX: 80
+// MAX: 81
 
 typedef struct {
   XML_Parser parser;
@@ -190,20 +191,7 @@ void perror(parser_data_t *pdata, const char *fmt, ...);
                                     "allowed under <%s>.", name, \
                                     parent_name.c_str());
 
-void start_element(void *user_data, const char *name,
-                   const char **atts);
-void start_level1(parser_data_t *pdata, const char *name);
-void start_level2(parser_data_t *pdata, const char *name);
-void start_level3(parser_data_t *pdata, const char *name);
-void start_level4(parser_data_t *pdata, const char *name);
-void start_level5(parser_data_t *pdata, const char *name);
-
 void end_element(void *user_data, const char *name);
-void end_level1(parser_data_t *pdata, const char *name);
-void end_level2(parser_data_t *pdata, const char *name);
-void end_level3(parser_data_t *pdata, const char *name);
-void end_level4(parser_data_t *pdata, const char *name);
-void end_level5(parser_data_t *pdata, const char *name);
 
 void parse_xml_tags(const char *name, KaxTags *tags);
 

src/tagparser_end.cpp

@@ -34,6 +34,7 @@
 
 #include "base64.h"
 #include "common.h"
+#include "iso639.h"
 #include "mm_io.h"
 #include "tagparser.h"
 
@@ -41,8 +42,8 @@ using namespace std;
 using namespace libebml;
 using namespace libmatroska;
 
-void el_get_uint(parser_data_t *pdata, EbmlElement *el,
+static void el_get_uint(parser_data_t *pdata, EbmlElement *el,
-                 uint64_t min_value = 0) {
+                        uint64_t min_value = 0) {
   int64 value;
 
   strip(*pdata->bin);
@@ -56,8 +57,8 @@ void el_get_uint(parser_data_t *pdata, EbmlElement *el,
   *(static_cast<EbmlUInteger *>(el)) = value;
 }
 
-void el_get_sint(parser_data_t *pdata, EbmlElement *el,
+static void el_get_sint(parser_data_t *pdata, EbmlElement *el,
-                 int64_t min_value = -9223372036854775807LL-1) {
+                        int64_t min_value = -9223372036854775807LL-1) {
   int64 value;
 
   strip(*pdata->bin);
@@ -71,8 +72,8 @@ void el_get_sint(parser_data_t *pdata, EbmlElement *el,
   *(static_cast<EbmlSInteger *>(el)) = value;
 }
 
-void el_get_float(parser_data_t *pdata, EbmlElement *el,
+static void el_get_float(parser_data_t *pdata, EbmlElement *el,
-                  float min_value = (float)1.40129846432481707e-45) {
+                         float min_value = (float)1.40129846432481707e-45) {
   char *endptr;
   float value;
 
@@ -87,15 +88,21 @@ void el_get_float(parser_data_t *pdata, EbmlElement *el,
   *(static_cast<EbmlFloat *>(el)) = value;
 }
 
-void el_get_string(parser_data_t *pdata, EbmlElement *el) {
+static void el_get_string(parser_data_t *pdata, EbmlElement *el,
+                          bool check_language = false) {
   strip(*pdata->bin);
   if (pdata->bin->length() == 0)
     perror(pdata, "Expected a string but found only whitespaces.");
 
+  if (check_language && !is_valid_iso639_2_code(pdata->bin->c_str()))
+    perror(pdata, "'%s' is not a valid ISO639-2 language code. See the "
+           "output of 'mkvmerge --list-languages' for a list of all "
+           "valid language codes.", pdata->bin->c_str());
+
   *(static_cast<EbmlString *>(el)) = pdata->bin->c_str();
 }
 
-void el_get_utf8string(parser_data_t *pdata, EbmlElement *el) {
+static void el_get_utf8string(parser_data_t *pdata, EbmlElement *el) {
   strip(*pdata->bin);
   if (pdata->bin->length() == 0)
     perror(pdata, "Expected a string but found only whitespaces.");
@@ -104,7 +111,7 @@ void el_get_utf8string(parser_data_t *pdata, EbmlElement *el) {
     cstrutf8_to_UTFstring(pdata->bin->c_str());
 }
 
-void el_get_binary(parser_data_t *pdata, EbmlElement *el) {
+static void el_get_binary(parser_data_t *pdata, EbmlElement *el) {
   int64_t result;
   binary *buffer;
   mm_io_c *io;
@@ -146,7 +153,7 @@ void el_get_binary(parser_data_t *pdata, EbmlElement *el) {
 // ISO 8601 format: 2003-07-17T19:50:53+0200
 //                  012345678901234567890123
 //                            1         2
-void el_get_date(parser_data_t *pdata, EbmlElement *el) {
+static void el_get_date(parser_data_t *pdata, EbmlElement *el) {
   const char *errmsg = "Expected a date in ISO 8601 format but found '%s'. "
     "The ISO 8601 date format looks like this: YYYY-MM-DDTHH:MM:SS:-TZTZ, "
     "e.g. 2003-07-17T19:50:52+0200. The time zone (TZ) may also be negative.";
@@ -223,7 +230,27 @@ void el_get_date(parser_data_t *pdata, EbmlElement *el) {
   (static_cast<EbmlDate *>(el))->SetEpochDate(tme);
 }
 
-void end_level1(parser_data_t *pdata, const char *) {
+static bool is_multicomment(parser_data_t *pdata, const char *name) {
+  int parent;
+
+  parent = (*pdata->parents)[pdata->parents->size() - 2];
+
+  if (parent != E_MultiComment)
+    return false;
+
+  if (!strcmp(name, "Name"))
+    el_get_string(pdata, &GetChild<KaxTagMultiCommentName>(*pdata->m_comment));
+  else if (!strcmp(name, "Comments"))
+    el_get_utf8string(pdata, &GetChild<KaxTagMultiCommentComments>
+                      (*pdata->m_comment));
+  else if (!strcmp(name, "Language"))
+    el_get_string(pdata, &GetChild<KaxTagMultiCommentLanguage>
+                  (*pdata->m_comment), true);
+
+  return true;
+}
+
+static void end_level1(parser_data_t *pdata, const char *) {
   // Can only be "Tag"
   pdata->targets = NULL;
   pdata->general = NULL;
@@ -239,7 +266,7 @@ void end_level1(parser_data_t *pdata, const char *) {
   pdata->m_comment = NULL;
 }
 
-void end_level2(parser_data_t *pdata, const char *name) {
+static void end_level2(parser_data_t *pdata, const char *name) {
   if (!strcmp(name, "Targets")) {
     pdata->track_uid = NULL;
     pdata->chapter_uid = NULL;
@@ -290,10 +317,13 @@ void end_level2(parser_data_t *pdata, const char *name) {
   }
 }
 
-void end_level3(parser_data_t *pdata, const char *name) {
+static void end_level3(parser_data_t *pdata, const char *name) {
   string parent_name;
   int parent;
 
+  if (is_multicomment(pdata, name))
+    return;
+
   parent_name = (*pdata->parent_names)[pdata->parent_names->size() - 2];
   parent = (*pdata->parents)[pdata->parents->size() - 2];
 
@@ -309,7 +339,7 @@ void end_level3(parser_data_t *pdata, const char *name) {
     else if (!strcmp(name, "Bibliography"))
       el_get_utf8string(pdata, &GetChild<KaxTagBibliography>(*pdata->general));
     else if (!strcmp(name, "Language"))
-      el_get_string(pdata, &GetChild<KaxTagLanguage>(*pdata->general));
+      el_get_string(pdata, &GetChild<KaxTagLanguage>(*pdata->general), true);
     else if (!strcmp(name, "Rating"))
       el_get_binary(pdata, &GetChild<KaxTagRating>(*pdata->general));
     else if (!strcmp(name, "Encoder"))
@@ -428,10 +458,13 @@ void end_level3(parser_data_t *pdata, const char *name) {
     die("Unknown parent: level 3, %d", parent);
 }
 
-void end_level4(parser_data_t *pdata, const char *name) {
+static void end_level4(parser_data_t *pdata, const char *name) {
   string parent_name;
   int parent;
 
+  if (is_multicomment(pdata, name))
+    return;
+
   parent_name = (*pdata->parent_names)[pdata->parent_names->size() - 2];
   parent = (*pdata->parents)[pdata->parents->size() - 2];
 
@@ -522,16 +555,19 @@ void end_level4(parser_data_t *pdata, const char *name) {
       el_get_string(pdata, pdata->t_email);
     else if (!strcmp(name, "Language"))
       el_get_string(pdata, &GetChild<KaxTagMultiTitleLanguage>
-                    (*pdata->title));
+                    (*pdata->title), true);
 
   } else
     die("Unknown parent: level 4, %d", parent);
 }
 
-void end_level5(parser_data_t *pdata, const char *name) {
+static void end_level5(parser_data_t *pdata, const char *name) {
   string parent_name;
   int parent;
 
+  if (is_multicomment(pdata, name))
+    return;
+
   parent_name = (*pdata->parent_names)[pdata->parent_names->size() - 2];
   parent = (*pdata->parents)[pdata->parents->size() - 2];
 
@@ -550,6 +586,13 @@ void end_level5(parser_data_t *pdata, const char *name) {
     die("Unknown parent: level 4, %d", parent);
 }
 
+static void end_level6(parser_data_t *pdata, const char *name) {
+  if (is_multicomment(pdata, name))
+    return;
+
+  die("tagparser_end: Unknown element. This should not have happened.");
+}
+
 void end_element(void *user_data, const char *name) {
   parser_data_t *pdata;
 
@@ -558,7 +601,9 @@ void end_element(void *user_data, const char *name) {
   if (pdata->data_allowed && (pdata->bin == NULL))
     perror(pdata, "Element <%s> does not contain any data.", name);
 
-  if (pdata->depth == 2)
+  if (pdata->depth == 1)
+    ;                           // Nothing to do here!
+  else if (pdata->depth == 2)
     end_level1(pdata, name);
   else if (pdata->depth == 3)
     end_level2(pdata, name);
@@ -568,6 +613,10 @@ void end_element(void *user_data, const char *name) {
     end_level4(pdata, name);
   else if (pdata->depth == 6)
     end_level5(pdata, name);
+  else if (pdata->depth == 7)
+    end_level6(pdata, name);
+  else
+    die("tagparser_end: depth > 7: %d", pdata->depth);
 
   if (pdata->bin != NULL) {
     delete pdata->bin;

src/tagparser_start.cpp

@@ -77,7 +77,53 @@ template <typename Type>Type &GetNextEmptyChild(EbmlMaster &master,
 	return *(static_cast<Type *>(e));
 }
 
-void start_level1(parser_data_t *pdata, const char *name) {
+static bool is_multicomment(parser_data_t *pdata, const char *name,
+                            EbmlElement *parent_elt) {
+  string parent_name;
+  int parent;
+
+  parent_name = (*pdata->parent_names)[pdata->parent_names->size() - 2];
+  parent = (*pdata->parents)[pdata->parents->size() - 1];
+
+  if (((parent == E_MultiComment) || (parent == E_CommentName) ||
+       (parent == E_Comments) || (parent == E_CommentLanguage)) &&
+      !strcmp(name, "MultiComment"))
+    perror_nochild();
+
+  if ((parent != E_MultiComment) && !strcmp(name, "MultiComment")) {
+    if (pdata->m_comment == NULL)
+      pdata->m_comment =
+        &GetEmptyChild<KaxTagMultiComment>(*((EbmlMaster *)parent_elt));
+    else
+      pdata->m_comment =
+        &GetNextEmptyChild<KaxTagMultiComment>(*((EbmlMaster *)parent_elt),
+                                               *pdata->m_comment);
+    pdata->parents->push_back(E_MultiComment);
+
+    return true;
+
+  } else if (parent == E_MultiComment) {
+    pdata->data_allowed = true;
+
+    if (!strcmp(name, "Name")) {
+      check_instances(pdata->m_comment, KaxTagMultiCommentName);
+      pdata->parents->push_back(E_CommentName);
+    } else if (!strcmp(name, "Comments")) {
+      check_instances(pdata->m_comment, KaxTagMultiCommentComments);
+      pdata->parents->push_back(E_Comments);
+    } else if (!strcmp(name, "Language")) {
+      check_instances(pdata->m_comment, KaxTagMultiCommentLanguage);
+      pdata->parents->push_back(E_CommentLanguage);
+    } else
+      perror_nochild();
+
+    return true;
+  }
+
+  return false;
+}
+
+static void start_level1(parser_data_t *pdata, const char *name) {
   string parent_name;
 
   parent_name = (*pdata->parent_names)[pdata->parent_names->size() - 2];
@@ -94,9 +140,12 @@ void start_level1(parser_data_t *pdata, const char *name) {
   pdata->parents->push_back(E_Tag);
 }
 
-void start_level2(parser_data_t *pdata, const char *name) {
+static void start_level2(parser_data_t *pdata, const char *name) {
   string parent_name;
 
+  if (is_multicomment(pdata, name, pdata->tag))
+    return;
+
   parent_name = (*pdata->parent_names)[pdata->parent_names->size() - 2];
 
   if (!strcmp(name, "Targets")) {
@@ -188,7 +237,7 @@ void start_level2(parser_data_t *pdata, const char *name) {
     perror_nochild();
 }
 
-void start_level3(parser_data_t *pdata, const char *name) {
+static void start_level3(parser_data_t *pdata, const char *name) {
   string parent_name;
   int parent;
 
@@ -196,6 +245,9 @@ void start_level3(parser_data_t *pdata, const char *name) {
   parent = (*pdata->parents)[pdata->parents->size() - 1];
 
   if (parent == E_Targets) {
+    if (is_multicomment(pdata, name, pdata->targets))
+      return;
+
     pdata->data_allowed = true;
 
     if (!strcmp(name, "TrackUID")) {
@@ -219,6 +271,9 @@ void start_level3(parser_data_t *pdata, const char *name) {
       perror_nochild();
 
   } else if (parent == E_General) {
+    if (is_multicomment(pdata, name, pdata->general))
+      return;
+
     pdata->data_allowed = true;
 
     if (!strcmp(name, "Subject")) {
@@ -287,6 +342,9 @@ void start_level3(parser_data_t *pdata, const char *name) {
       perror_nochild();
 
   } else if (parent == E_Genres) {
+    if (is_multicomment(pdata, name, pdata->genres))
+      return;
+
     pdata->data_allowed = true;
 
     if (!strcmp(name, "AudioGenre")) {
@@ -312,6 +370,9 @@ void start_level3(parser_data_t *pdata, const char *name) {
       perror_nochild();
 
   } else if (parent == E_AudioSpecific) {
+    if (is_multicomment(pdata, name, pdata->audio_specific))
+      return;
+
     pdata->data_allowed = true;
 
     if (!strcmp(name, "AudioEncryption")) {
@@ -348,6 +409,9 @@ void start_level3(parser_data_t *pdata, const char *name) {
       perror_nochild();
 
   } else if (parent == E_ImageSpecific) {
+    if (is_multicomment(pdata, name, pdata->image_specific))
+      return;
+
     pdata->data_allowed = true;
 
     if (!strcmp(name, "CaptureDPI")) {
@@ -372,6 +436,9 @@ void start_level3(parser_data_t *pdata, const char *name) {
       perror_nochild();
 
   } else if (parent == E_MultiCommercial) {
+    if (is_multicomment(pdata, name, pdata->m_commercial))
+      return;
+
     if (!strcmp(name, "Commercial")) {
       if (pdata->commercial == NULL)
         pdata->commercial =
@@ -385,6 +452,9 @@ void start_level3(parser_data_t *pdata, const char *name) {
       perror_nochild();
 
   } else if (parent == E_MultiDate) {
+    if (is_multicomment(pdata, name, pdata->m_date))
+      return;
+
     if (!strcmp(name, "Date")) {
       if (pdata->date == NULL)
         pdata->date = &GetEmptyChild<KaxTagDate>(*pdata->m_date);
@@ -396,6 +466,9 @@ void start_level3(parser_data_t *pdata, const char *name) {
       perror_nochild();
 
   } else if (parent == E_MultiEntity) {
+    if (is_multicomment(pdata, name, pdata->m_entity))
+      return;
+
     if (!strcmp(name, "Entity")) {
       if (pdata->entity == NULL)
         pdata->entity = &GetEmptyChild<KaxTagEntity>(*pdata->m_entity);
@@ -407,6 +480,9 @@ void start_level3(parser_data_t *pdata, const char *name) {
       perror_nochild();
 
   } else if (parent == E_MultiIdentifier) {
+    if (is_multicomment(pdata, name, pdata->m_identifier))
+      return;
+
     if (!strcmp(name, "Identifier")) {
       if (pdata->identifier == NULL)
         pdata->identifier =
@@ -420,6 +496,9 @@ void start_level3(parser_data_t *pdata, const char *name) {
       perror_nochild();
 
   } else if (parent == E_MultiLegal) {
+    if (is_multicomment(pdata, name, pdata->m_legal))
+      return;
+
     if (!strcmp(name, "Legal")) {
       if (pdata->legal == NULL)
         pdata->legal = &GetEmptyChild<KaxTagLegal>(*pdata->m_legal);
@@ -431,6 +510,9 @@ void start_level3(parser_data_t *pdata, const char *name) {
       perror_nochild();
 
   } else if (parent == E_MultiTitle) {
+    if (is_multicomment(pdata, name, pdata->m_title))
+      return;
+
     if (!strcmp(name, "Title")) {
       if (pdata->title == NULL)
         pdata->title = &GetEmptyChild<KaxTagTitle>(*pdata->m_title);
@@ -445,7 +527,7 @@ void start_level3(parser_data_t *pdata, const char *name) {
     die("Unknown parent: level 3, %d", parent);
 }
 
-void start_level4(parser_data_t *pdata, const char *name) {
+static void start_level4(parser_data_t *pdata, const char *name) {
   string parent_name;
   int parent;
 
@@ -453,6 +535,9 @@ void start_level4(parser_data_t *pdata, const char *name) {
   parent = (*pdata->parents)[pdata->parents->size() - 1];
 
   if (parent == E_Commercial) {
+    if (is_multicomment(pdata, name, pdata->commercial))
+      return;
+
     pdata->data_allowed = true;
 
     if (!strcmp(name, "Type")) {
@@ -492,6 +577,9 @@ void start_level4(parser_data_t *pdata, const char *name) {
       perror_nochild();
 
   } else if (parent == E_Date) {
+    if (is_multicomment(pdata, name, pdata->date))
+      return;
+
     pdata->data_allowed = true;
 
     if (!strcmp(name, "Type")) {
@@ -507,6 +595,9 @@ void start_level4(parser_data_t *pdata, const char *name) {
       perror_nochild();
 
   } else if (parent == E_Entity) {
+    if (is_multicomment(pdata, name, pdata->entity))
+      return;
+
     pdata->data_allowed = true;
 
     if (!strcmp(name, "Type")) {
@@ -539,6 +630,9 @@ void start_level4(parser_data_t *pdata, const char *name) {
       perror_nochild();
 
   } else if (parent == E_Identifier) {
+    if (is_multicomment(pdata, name, pdata->identifier))
+      return;
+
     pdata->data_allowed = true;
 
     if (!strcmp(name, "Type")) {
@@ -554,6 +648,9 @@ void start_level4(parser_data_t *pdata, const char *name) {
       perror_nochild();
 
   } else if (parent == E_Legal) {
+    if (is_multicomment(pdata, name, pdata->legal))
+      return;
+
     pdata->data_allowed = true;
 
     if (!strcmp(name, "Type")) {
@@ -577,6 +674,9 @@ void start_level4(parser_data_t *pdata, const char *name) {
       perror_nochild();
 
   } else if (parent == E_Title) {
+    if (is_multicomment(pdata, name, pdata->title))
+      return;
+
     pdata->data_allowed = true;
 
     if (!strcmp(name, "Type")) {
@@ -620,7 +720,7 @@ void start_level4(parser_data_t *pdata, const char *name) {
     die("Unknown parent: level 4, %d", parent);
 }
 
-void start_level5(parser_data_t *pdata, const char *name) {
+static void start_level5(parser_data_t *pdata, const char *name) {
   string parent_name;
   int parent;
 
@@ -628,6 +728,9 @@ void start_level5(parser_data_t *pdata, const char *name) {
   parent = (*pdata->parents)[pdata->parents->size() - 1];
 
   if (parent == E_MultiPrice) {
+    if (is_multicomment(pdata, name, pdata->m_price))
+      return;
+
     pdata->data_allowed = true;
 
     if (!strcmp(name, "Currency")) {
@@ -643,7 +746,12 @@ void start_level5(parser_data_t *pdata, const char *name) {
       perror_nochild();
 
   } else
-    die("Unknown parent: level 4, %d", parent);
+    die("Unknown parent: level 5, %d", parent);
+}
+
+static void start_level6(parser_data_t *pdata, const char *name) {
+  if (!is_multicomment(pdata, name, NULL))
+    die("tagparser_start: Unknown element - should not have happened.");
 }
 
 void start_element(void *user_data, const char *name,
@@ -681,8 +789,10 @@ void start_element(void *user_data, const char *name,
     start_level4(pdata, name);
   else if (pdata->depth == 5)
     start_level5(pdata, name);
+  else if (pdata->depth == 6)
+    start_level6(pdata, name);
   else
-    die("Depth > 5: %d", pdata->depth);
+    die("Depth > 6: %d", pdata->depth);
 
   if (pdata->parent_names->size() != pdata->parents->size())
     perror_unknown();