Unified the chapter and tag XML output code just like the parser code for both was unified a long time ago. Also output binary content of XML elements in hex format.

src/common/chapter_writer.cpp

@@ -21,10 +21,10 @@
 
 #include <matroska/KaxChapters.h>
 
+#include "chapters.h"
 #include "commonebml.h"
 #include "mm_io.h"
-#include "chapters.h"
-#include "xml_element_mapping.h"
+#include "xml_element_writer.h"
 
 using namespace std;
 using namespace libmatroska;
@@ -203,16 +203,8 @@ write_chapters_simple(int &chapter_num,
 
 // {{{ XML chapter output
 
-typedef struct {
-  int level;
-  int parent_idx;
-  int elt_idx;
-  EbmlElement *e;
-  mm_io_c *out;
-} chapter_writer_cb_t;
-
 static void
-pt(chapter_writer_cb_t *cb,
+pt(xml_writer_cb_t *cb,
    const char *tag) {
   int i;
 
@@ -221,91 +213,6 @@ pt(chapter_writer_cb_t *cb,
   cb->out->printf("%s", tag);
 }
 
-static void
-write_xml_element_rec(int level,
-                      int parent_idx,
-                      EbmlElement *e,
-                      mm_io_c *out) {
-  EbmlMaster *m;
-  int elt_idx, i;
-  bool found;
-  string s;
-
-  elt_idx = parent_idx;
-  found = false;
-  while ((chapter_elements[elt_idx].name != NULL) &&
-         (chapter_elements[elt_idx].level >=
-          chapter_elements[parent_idx].level)) {
-    if (chapter_elements[elt_idx].id == e->Generic().GlobalId) {
-      found = true;
-      break;
-    }
-    elt_idx++;
-  }
-
-  for (i = 0; i < level; i++)
-    out->printf("  ");
-
-  if (!found) {
-    out->printf("<!-- Unknown element '%s' -->\n", e->Generic().DebugName);
-    return;
-  }
-
-  out->printf("<%s>", chapter_elements[elt_idx].name);
-  switch (chapter_elements[elt_idx].type) {
-    case EBMLT_MASTER:
-      out->printf("\n");
-      m = dynamic_cast<EbmlMaster *>(e);
-      assert(m != NULL);
-      for (i = 0; i < m->ListSize(); i++)
-        write_xml_element_rec(level + 1, elt_idx, (*m)[i], out);
-
-      if (chapter_elements[elt_idx].end_hook != NULL) {
-        chapter_writer_cb_t cb;
-
-        cb.level = level;
-        cb.parent_idx = parent_idx;
-        cb.elt_idx = elt_idx;
-        cb.e = e;
-        cb.out = out;
-
-        chapter_elements[elt_idx].end_hook(&cb);
-      }
-
-      for (i = 0; i < level; i++)
-        out->printf("  ");
-      out->printf("</%s>\n", chapter_elements[elt_idx].name);
-      break;
-
-    case EBMLT_UINT:
-    case EBMLT_BOOL:
-      out->printf("%llu</%s>\n", uint64(*dynamic_cast<EbmlUInteger *>(e)),
-                  chapter_elements[elt_idx].name);
-      break;
-
-    case EBMLT_STRING:
-      s = escape_xml(string(*dynamic_cast<EbmlString *>(e)));
-      out->printf("%s</%s>\n", s.c_str(), chapter_elements[elt_idx].name);
-      break;
-
-    case EBMLT_USTRING:
-      s = UTFstring_to_cstrutf8(UTFstring(*static_cast
-                                          <EbmlUnicodeString *>(e)).c_str());
-      s = escape_xml(s);
-      out->printf("%s</%s>\n", s.c_str(), chapter_elements[elt_idx].name);
-      break;
-
-    case EBMLT_TIME:
-      out->printf(FMT_TIMECODEN "</%s>\n", 
-                  ARG_TIMECODEN(uint64(*dynamic_cast<EbmlUInteger *>(e))),
-                  chapter_elements[elt_idx].name);
-      break;
-
-    default:
-      assert(false);
-  }
-}
-
 static int
 cet_index(const char *name) {
   int i;
@@ -321,9 +228,9 @@ cet_index(const char *name) {
 static void
 end_write_chapter_atom(void *data) {
   KaxChapterAtom *atom;
-  chapter_writer_cb_t *cb;
+  xml_writer_cb_t *cb;
 
-  cb = (chapter_writer_cb_t *)data;
+  cb = (xml_writer_cb_t *)data;
   atom = dynamic_cast<KaxChapterAtom *>(cb->e);
   assert(atom != NULL);
   if (FINDFIRST(atom, KaxChapterTimeStart) == NULL)
@@ -333,9 +240,9 @@ end_write_chapter_atom(void *data) {
 static void
 end_write_chapter_display(void *data) {
   KaxChapterDisplay *display;
-  chapter_writer_cb_t *cb;
+  xml_writer_cb_t *cb;
 
-  cb = (chapter_writer_cb_t *)data;
+  cb = (xml_writer_cb_t *)data;
   display = dynamic_cast<KaxChapterDisplay *>(cb->e);
   assert(display != NULL);
   if (FINDFIRST(display, KaxChapterString) == NULL)
@@ -359,7 +266,7 @@ write_chapters_xml(KaxChapters *chapters,
     end_write_chapter_display;
 
   for (i = 0; i < chapters->ListSize(); i++)
-    write_xml_element_rec(1, 0, (*chapters)[i], out);
+    write_xml_element_rec(1, 0, (*chapters)[i], out, chapter_elements);
 }
 
 // }}}

src/common/tagwriter.cpp

@@ -14,163 +14,13 @@
 
 #include "os.h"
 
-#include <stdarg.h>
-#include <stdio.h>
-#include <time.h>
-
-#include <matroska/KaxTag.h>
 #include <matroska/KaxTags.h>
-#include <matroska/KaxTagMulti.h>
+
+#include "xml_element_writer.h"
 
 using namespace libmatroska;
 using namespace std;
 
-#include "base64.h"
-#include "common.h"
-#include "commonebml.h"
-#include "matroska.h"
-#include "mm_io.h"
-#include "xml_element_mapping.h"
-
-static void
-print_binary(int level,
-             const char *name,
-             EbmlElement *e,
-             mm_io_c *out) {
-  EbmlBinary *b;
-  string s;
-  int i, idx, old_idx;
-
-  b = (EbmlBinary *)e;
-  s = base64_encode((const unsigned char *)b->GetBuffer(), b->GetSize(), true,
-                    72 - level - 2);
-  if (s[s.length() - 1] == '\n')
-    s.erase(s.length() - 1);
-
-  if ((level * 2 + 2 * strlen(name) + 2 + 3 + s.length()) <= 78) {
-    out->printf("%s</%s>\n", s.c_str(), name);
-    return;
-  }
-
-  out->printf("\n");
-
-  for (i = 0; i < (level + 2); i++)
-    out->printf("  ");
-
-  old_idx = 0;
-  for (idx = 0; idx < s.length(); idx++)
-    if (s[idx] == '\n') {
-      out->printf("%s\n", s.substr(old_idx, idx - old_idx).c_str());
-      for (i = 0; i < (level + 2); i++)
-        out->printf("  ");
-      old_idx = idx + 1;
-    }
-
-  if (old_idx < s.length())
-    out->printf("%s", s.substr(old_idx).c_str());
-
-  out->printf("\n");
-
-  for (i = 0; i < level; i++)
-    out->printf("  ");
-  out->printf("</%s>\n", name);
-}
-
-typedef struct {
-  int level;
-  int parent_idx;
-  int elt_idx;
-  EbmlElement *e;
-} tag_writer_cb_t;
-
-static void
-write_xml_element_rec(int level,
-                      int parent_idx,
-                      EbmlElement *e,
-                      mm_io_c *out) {
-  EbmlMaster *m;
-  int elt_idx, i;
-  bool found;
-  string s;
-
-  elt_idx = parent_idx;
-  found = false;
-  while ((tag_elements[elt_idx].name != NULL) &&
-         (tag_elements[elt_idx].level >=
-          tag_elements[parent_idx].level)) {
-    if (tag_elements[elt_idx].id == e->Generic().GlobalId) {
-      found = true;
-      break;
-    }
-    elt_idx++;
-  }
-
-  for (i = 0; i < level; i++)
-    out->printf("  ");
-
-  if (!found) {
-    out->printf("<!-- Unknown element '%s' -->\n", e->Generic().DebugName);
-    return;
-  }
-
-  out->printf("<%s>", tag_elements[elt_idx].name);
-  switch (tag_elements[elt_idx].type) {
-    case EBMLT_MASTER:
-      out->printf("\n");
-      m = dynamic_cast<EbmlMaster *>(e);
-      assert(m != NULL);
-      for (i = 0; i < m->ListSize(); i++)
-        write_xml_element_rec(level + 1, elt_idx, (*m)[i], out);
-
-      if (tag_elements[elt_idx].end_hook != NULL) {
-        tag_writer_cb_t cb;
-
-        cb.level = level;
-        cb.parent_idx = parent_idx;
-        cb.elt_idx = elt_idx;
-        cb.e = e;
-
-        tag_elements[elt_idx].end_hook(&cb);
-      }
-
-      for (i = 0; i < level; i++)
-        out->printf("  ");
-      out->printf("</%s>\n", tag_elements[elt_idx].name);
-      break;
-
-    case EBMLT_UINT:
-    case EBMLT_BOOL:
-      out->printf("%llu</%s>\n", uint64(*dynamic_cast<EbmlUInteger *>(e)),
-                  tag_elements[elt_idx].name);
-      break;
-
-    case EBMLT_STRING:
-      s = escape_xml(string(*dynamic_cast<EbmlString *>(e)));
-      out->printf("%s</%s>\n", s.c_str(), tag_elements[elt_idx].name);
-      break;
-
-    case EBMLT_USTRING:
-      s = UTFstring_to_cstrutf8(UTFstring(*static_cast
-                                          <EbmlUnicodeString *>(e)).c_str());
-      s = escape_xml(s);
-      out->printf("%s</%s>\n", s.c_str(), tag_elements[elt_idx].name);
-      break;
-
-    case EBMLT_TIME:
-      out->printf(FMT_TIMECODEN "</%s>\n", 
-                  ARG_TIMECODEN(uint64(*dynamic_cast<EbmlUInteger *>(e))),
-                  tag_elements[elt_idx].name);
-      break;
-
-    case EBMLT_BINARY:
-      print_binary(level, tag_elements[elt_idx].name, e, out);
-      break;
-
-    default:
-      assert(false);
-  }
-}
-
 void
 write_tags_xml(KaxTags &tags,
                mm_io_c *out) {
@@ -182,5 +32,5 @@ write_tags_xml(KaxTags &tags,
   }
 
   for (i = 0; i < tags.ListSize(); i++)
-    write_xml_element_rec(1, 0, tags[i], out);
+    write_xml_element_rec(1, 0, tags[i], out, tag_elements);
 }

src/common/tagwriter.h

@@ -15,8 +15,6 @@
 #ifndef __TAGWRITER_H
 #define __TAGWRITER_H
 
-#include <stdio.h>
-
 #include <matroska/KaxTags.h>
 #include <matroska/KaxTag.h>
 

src/common/xml_element_writer.cpp

@@ -0,0 +1,153 @@
+/*
+   mkvmerge -- utility for splicing together matroska files
+   from component media subtypes
+  
+   Distributed under the GPL
+   see the file COPYING for details
+   or visit http://www.gnu.org/copyleft/gpl.html
+  
+   $Id$
+  
+   XML chapter writer functions
+  
+   Written by Moritz Bunkus <moritz@bunkus.org>.
+*/
+
+#include <ebml/EbmlBinary.h>
+#include <ebml/EbmlMaster.h>
+#include <ebml/EbmlSInteger.h>
+#include <ebml/EbmlString.h>
+#include <ebml/EbmlUInteger.h>
+#include <ebml/EbmlUnicodeString.h>
+#include <ebml/EbmlDate.h>
+
+#include "base64.h"
+#include "common.h"
+#include "commonebml.h"
+#include "mm_io.h"
+#include "xml_element_writer.h"
+
+using namespace libebml;
+
+static void
+print_binary(int level,
+             const char *name,
+             EbmlElement *e,
+             mm_io_c *out) {
+  EbmlBinary *b;
+  const unsigned char *p;
+  string s;
+  int i, size;
+
+  b = (EbmlBinary *)e;
+  p = (const unsigned char *)b->GetBuffer();
+  size = b->GetSize();
+
+  for (i = 0; i < size; i++) {
+    if ((i % 16) != 0)
+      s += " ";
+    s += mxsprintf("%02x", *p);
+    ++p;
+    if ((((i + 1) % 16) == 0) && ((i + 1) < size))
+      s += mxsprintf("\n%*s", (level + 1) * 2, "");
+  }
+
+  if ((level * 2 + 2 * strlen(name) + 2 + 3 + s.length()) <= 78)
+    out->printf("%s</%s>\n", s.c_str(), name);
+  else
+    out->printf("\n%*s%s\n%*s</%s>\n", (level + 1) * 2, "", s.c_str(),
+                level * 2, "", name);
+}
+
+void
+write_xml_element_rec(int level,
+                      int parent_idx,
+                      EbmlElement *e,
+                      mm_io_c *out,
+                      const parser_element_t *element_map) {
+  EbmlMaster *m;
+  int elt_idx, i;
+  bool found;
+  string s;
+
+  elt_idx = parent_idx;
+  found = false;
+  while ((element_map[elt_idx].name != NULL) &&
+         (element_map[elt_idx].level >=
+          element_map[parent_idx].level)) {
+    if (element_map[elt_idx].id == e->Generic().GlobalId) {
+      found = true;
+      break;
+    }
+    elt_idx++;
+  }
+
+  for (i = 0; i < level; i++)
+    out->printf("  ");
+
+  if (!found) {
+    out->printf("<!-- Unknown element '%s' -->\n", e->Generic().DebugName);
+    return;
+  }
+
+  out->printf("<%s%s>", element_map[elt_idx].name,
+              element_map[elt_idx].type == EBMLT_BINARY ?
+              " format=\"hex\"" : "");
+  switch (element_map[elt_idx].type) {
+    case EBMLT_MASTER:
+      out->printf("\n");
+      m = dynamic_cast<EbmlMaster *>(e);
+      assert(m != NULL);
+      for (i = 0; i < m->ListSize(); i++)
+        write_xml_element_rec(level + 1, elt_idx, (*m)[i], out, element_map);
+
+      if (element_map[elt_idx].end_hook != NULL) {
+        xml_writer_cb_t cb;
+
+        cb.level = level;
+        cb.parent_idx = parent_idx;
+        cb.elt_idx = elt_idx;
+        cb.e = e;
+        cb.out = out;
+
+        element_map[elt_idx].end_hook(&cb);
+      }
+
+      for (i = 0; i < level; i++)
+        out->printf("  ");
+      out->printf("</%s>\n", element_map[elt_idx].name);
+      break;
+
+    case EBMLT_UINT:
+    case EBMLT_BOOL:
+      out->printf("%llu</%s>\n", uint64(*dynamic_cast<EbmlUInteger *>(e)),
+                  element_map[elt_idx].name);
+      break;
+
+    case EBMLT_STRING:
+      s = escape_xml(string(*dynamic_cast<EbmlString *>(e)));
+      out->printf("%s</%s>\n", s.c_str(), element_map[elt_idx].name);
+      break;
+
+    case EBMLT_USTRING:
+      s = UTFstring_to_cstrutf8(UTFstring(*static_cast
+                                          <EbmlUnicodeString *>(e)).c_str());
+      s = escape_xml(s);
+      out->printf("%s</%s>\n", s.c_str(), element_map[elt_idx].name);
+      break;
+
+    case EBMLT_TIME:
+      out->printf(FMT_TIMECODEN "</%s>\n", 
+                  ARG_TIMECODEN(uint64(*dynamic_cast<EbmlUInteger *>(e))),
+                  element_map[elt_idx].name);
+      break;
+
+    case EBMLT_BINARY:
+      print_binary(level, element_map[elt_idx].name, e, out);
+      break;
+
+    default:
+      assert(false);
+  }
+}
+

src/common/xml_element_writer.h

@@ -0,0 +1,42 @@
+/*
+   mkvmerge -- utility for splicing together matroska files
+   from component media subtypes
+  
+   Distributed under the GPL
+   see the file COPYING for details
+   or visit http://www.gnu.org/copyleft/gpl.html
+  
+   $Id$
+  
+   XML chapter writer functions
+  
+   Written by Moritz Bunkus <moritz@bunkus.org>.
+*/
+
+#ifndef __XML_ELEMENT_WRITER_H
+#define __XML_ELEMENT_WRITER_H
+
+#include "xml_element_parser.h"
+
+namespace libebml {
+  class EbmlElement;
+};
+
+using namespace libebml;
+
+class mm_io_c;
+
+struct xml_writer_cb_t {
+  int level;
+  int parent_idx;
+  int elt_idx;
+  EbmlElement *e;
+  mm_io_c *out;
+};
+
+void MTX_DLL_API
+write_xml_element_rec(int level, int parent_idx,
+                      EbmlElement *e, mm_io_c *out,
+                      const parser_element_t *element_map);
+
+#endif // __XML_ELEMENT_WRITER_H