add support for text/binary distinction within multifile subfiles

panda/src/downloadertools/multify.cxx

@@ -48,11 +48,15 @@ Filename chdir_to;             // -C
 bool got_chdir_to = false;
 size_t scale_factor = 0;       // -F
 pset<string> dont_compress;    // -Z
+pset<string> text_ext;         // -X
 vector_string sign_params;     // -S
 
 // Default extensions not to compress.  May be overridden with -Z.
 string dont_compress_str = "jpg,png,mp3,ogg";
 
+// Default text extensions.  May be overridden with -X.
+string text_ext_str = "txt";
+
 bool got_record_timestamp_flag = false;
 bool record_timestamp_flag = true;
 
@@ -210,6 +214,12 @@ help() {
     "      files that are not to be compressed.  The default if this is omitted is\n"
     "      \"" << dont_compress_str << "\".  Specify -Z \"\" (be sure to include the space) to allow\n"
     "      all files to be compressed.\n\n"
+    "  -X <extension_list>\n"
+    "      Specify a comma-separated list of filename extensions that represent\n"
+    "      text files.  These files are opened and read in text mode, and added to\n"
+    "      the multifile with the text flag set.  The default if this is omitted is\n"
+    "      \"" << text_ext_str << "\".  Specify -X \"\" (be sure to include the space) to record\n"
+    "      all files in binary mode.\n\n"
 
     "  -T <flag>\n"
     "      Enable or disable the recording of file timestamps within the multifile.\n"
@@ -271,6 +281,20 @@ is_named(const string &subfile_name, const vector_string &params) {
   return false;
 }
 
+bool
+is_text(const Filename &subfile_name) {
+  // Returns true if this filename should be read as a text file,
+  // false otherwise.
+
+  string ext = subfile_name.get_extension();
+  if (text_ext.find(ext) != text_ext.end()) {
+    // This extension is listed on the -X parameter list; it's a text file.
+    return true;
+  }
+
+  return false;
+}
+
 int
 get_compression_level(const Filename &subfile_name) {
   // Returns the appropriate compression level for the named file.
@@ -317,7 +341,7 @@ do_add_files(Multifile *multifile, const pvector<Filename> &filenames) {
   bool okflag = true;
   pvector<Filename>::const_iterator fi;
   for (fi = filenames.begin(); fi != filenames.end(); ++fi) {
-    const Filename &subfile_name = (*fi);
+    Filename subfile_name = (*fi);
 
     if (subfile_name.is_directory()) {
       if (!do_add_directory(multifile, subfile_name)) {
@@ -329,6 +353,12 @@ do_add_files(Multifile *multifile, const pvector<Filename> &filenames) {
       okflag = false;
 
     } else {
+      if (is_text(subfile_name)) {
+        subfile_name.set_text();
+      } else {
+        subfile_name.set_binary();
+      }
+
       string new_subfile_name;
       if (update) {
         new_subfile_name = multifile->update_subfile
@@ -623,6 +653,10 @@ list_files(const vector_string &params) {
         if (multifile->is_subfile_encrypted(i)) {
           encrypted_symbol = 'e';
         }
+        char text_symbol = ' ';
+        if (multifile->is_subfile_text(i)) {
+          text_symbol = 't';
+        }
         if (multifile->is_subfile_compressed(i)) {
           size_t orig_length = multifile->get_subfile_length(i);
           size_t internal_length = multifile->get_subfile_internal_length(i);
@@ -631,25 +665,25 @@ list_files(const vector_string &params) {
             ratio = (double)internal_length / (double)orig_length;
           }
           if (ratio > 1.0) {
-            printf("%12d worse %c %s %s\n",
+            printf("%12d worse %c%c %s %s\n",
                    (int)multifile->get_subfile_length(i),
-                   encrypted_symbol, 
+                   encrypted_symbol, text_symbol,
                    format_timestamp(multifile->get_record_timestamp(),
                                     multifile->get_subfile_timestamp(i)),
                    subfile_name.c_str());
           } else {
-            printf("%12d  %3.0f%% %c %s %s\n",
+            printf("%12d  %3.0f%% %c%c %s %s\n",
                    (int)multifile->get_subfile_length(i),
                    100.0 - ratio * 100.0, 
-                   encrypted_symbol, 
+                   encrypted_symbol, text_symbol,
                    format_timestamp(multifile->get_record_timestamp(),
                                     multifile->get_subfile_timestamp(i)),
                    subfile_name.c_str());
           }
         } else {
-          printf("%12d       %c %s %s\n", 
+          printf("%12d       %c%c %s %s\n", 
                  (int)multifile->get_subfile_length(i),
-                 encrypted_symbol, 
+                 encrypted_symbol, text_symbol,
                  format_timestamp(multifile->get_record_timestamp(),
                                   multifile->get_subfile_timestamp(i)),
                  subfile_name.c_str());
@@ -736,7 +770,7 @@ main(int argc, char *argv[]) {
 
   extern char *optarg;
   extern int optind;
-  static const char *optflags = "crutxkvz123456789Z:T:S:f:OC:ep:P:F:h";
+  static const char *optflags = "crutxkvz123456789Z:T:X:S:f:OC:ep:P:F:h";
   int flag = getopt(argc, argv, optflags);
   Filename rel_path;
   while (flag != EOF) {
@@ -804,6 +838,9 @@ main(int argc, char *argv[]) {
     case 'Z':
       dont_compress_str = optarg;
       break;
+    case 'X':
+      text_ext_str = optarg;
+      break;
     case 'S':
       sign_params.push_back(optarg);
       break;
@@ -890,6 +927,9 @@ main(int argc, char *argv[]) {
   // Split out the extensions named by -Z into different words.
   tokenize_extensions(dont_compress_str, dont_compress);
 
+  // Ditto for -X.
+  tokenize_extensions(text_ext_str, text_ext);
+
   // Build a list of remaining parameters.
   vector_string params;
   params.reserve(argc - 1);

panda/src/express/config_express.cxx

@@ -61,6 +61,15 @@ ConfigVariableBool keep_temporary_files
           "default) to delete these.  Mainly useful for debugging "
           "when the process goes wrong."));
 
+ConfigVariableBool multifile_always_binary
+("multifile-always-binary", false,
+ PRC_DESC("This is a temporary transition variable.  Set this true "
+          "to enable the old behavior for multifiles: all subfiles are "
+          "always added to and extracted from the multifile in binary mode.  "
+          "Set it false to enable the new behavior: subfiles may be added "
+          "or extracted in either binary or text mode, according to the "
+          "set_binary() or set_text() flag on the Filename."));
+
 ConfigVariableBool collect_tcp
 ("collect-tcp", false,
  PRC_DESC("Set this true to enable accumulation of several small consecutive "

panda/src/express/config_express.h

@@ -50,6 +50,7 @@ extern ConfigVariableInt patchfile_buffer_size;
 extern ConfigVariableInt patchfile_zone_size;
 
 extern ConfigVariableBool keep_temporary_files;
+extern ConfigVariableBool multifile_always_binary;
 
 extern EXPCL_PANDAEXPRESS ConfigVariableBool collect_tcp;
 extern EXPCL_PANDAEXPRESS ConfigVariableDouble collect_tcp_interval;

panda/src/express/multifile.cxx

@@ -467,6 +467,11 @@ set_scale_factor(size_t scale_factor) {
 //               is replaced without examining its contents (but see
 //               also update_subfile).
 //
+//               Filename::set_binary() or set_text() must have been
+//               called previously to specify the nature of the source
+//               file.  If set_text() was called, the text flag will
+//               be set on the subfile.
+//
 //               Returns the subfile name on success (it might have
 //               been modified slightly), or empty string on failure.
 ////////////////////////////////////////////////////////////////////
@@ -475,15 +480,24 @@ add_subfile(const string &subfile_name, const Filename &filename,
             int compression_level) {
   nassertr(is_write_valid(), string());
 
-  if (!filename.exists()) {
+  Filename fname = filename;
+  if (multifile_always_binary) {
+    fname.set_binary();
+  }
+
+  nassertr(fname.is_binary_or_text(), string());
+
+  if (!fname.exists()) {
     return string();
   }
   string name = standardize_subfile_name(subfile_name);
   if (!name.empty()) {
     Subfile *subfile = new Subfile;
     subfile->_name = name;
-    subfile->_source_filename = filename;
-    subfile->_source_filename.set_binary();
+    subfile->_source_filename = fname;
+    if (fname.is_text()) {
+      subfile->_flags |= SF_text;
+    }
     
     add_new_subfile(subfile, compression_level);
   }
@@ -500,6 +514,7 @@ add_subfile(const string &subfile_name, const Filename &filename,
 //  Description: Adds a file from a stream as a subfile to the Multifile.
 //               The indicated istream will be read and its contents
 //               added to the Multifile at the next call to flush().
+//               The file will be added as a binary subfile.
 //
 //               Note that the istream must remain untouched and
 //               unused by any other code until flush() is called.  At
@@ -538,13 +553,25 @@ add_subfile(const string &subfile_name, istream *subfile_data,
 //               compared byte-for-byte to the disk file, and it is
 //               replaced only if it is different; otherwise, the
 //               multifile is left unchanged.
+//
+//               Filename::set_binary() or set_text() must have been
+//               called previously to specify the nature of the source
+//               file.  If set_text() was called, the text flag will
+//               be set on the subfile.
 ////////////////////////////////////////////////////////////////////
 string Multifile::
 update_subfile(const string &subfile_name, const Filename &filename,
                int compression_level) {
   nassertr(is_write_valid(), string());
 
-  if (!filename.exists()) {
+  Filename fname = filename;
+  if (multifile_always_binary) {
+    fname.set_binary();
+  }
+
+  nassertr(fname.is_binary_or_text(), string());
+
+  if (!fname.exists()) {
     return string();
   }
   string name = standardize_subfile_name(subfile_name);
@@ -552,7 +579,7 @@ update_subfile(const string &subfile_name, const Filename &filename,
     int index = find_subfile(name);
     if (index >= 0) {
       // The subfile already exists; compare it to the source file.
-      if (compare_subfile(index, filename)) {
+      if (compare_subfile(index, fname)) {
         // The files are identical; do nothing.
         return name;
       }
@@ -562,8 +589,10 @@ update_subfile(const string &subfile_name, const Filename &filename,
     // source file.  Add the new source file.
     Subfile *subfile = new Subfile;
     subfile->_name = name;
-    subfile->_source_filename = filename;
-    subfile->_source_filename.set_binary();
+    subfile->_source_filename = fname;
+    if (fname.is_text()) {
+      subfile->_flags |= SF_text;
+    }
 
     add_new_subfile(subfile, compression_level);
   }
@@ -1644,6 +1673,23 @@ is_subfile_encrypted(int index) const {
   return (_subfiles[index]->_flags & SF_encrypted) != 0;
 }
 
+////////////////////////////////////////////////////////////////////
+//     Function: Multifile::is_subfile_text
+//       Access: Published
+//  Description: Returns true if the indicated subfile represents text
+//               data, or false if it represents binary data.  If the
+//               file is text data, it may have been processed by
+//               end-of-line conversion when it was added.  (But the
+//               actual bits in the multifile will represent the
+//               standard Unix end-of-line convention, e.g. \n instead
+//               of \r\n.)
+////////////////////////////////////////////////////////////////////
+bool Multifile::
+is_subfile_text(int index) const {
+  nassertr(index >= 0 && index < (int)_subfiles.size(), false);
+  return (_subfiles[index]->_flags & SF_text) != 0;
+}
+
 ////////////////////////////////////////////////////////////////////
 //     Function: Multifile::get_index_end
 //       Access: Published
@@ -1776,7 +1822,20 @@ extract_subfile(int index, const Filename &filename) {
   nassertr(index >= 0 && index < (int)_subfiles.size(), false);
   
   Filename fname = filename;
-  fname.set_binary();
+  if (multifile_always_binary) {
+    fname.set_binary();
+  }
+
+  nassertr(fname.is_binary_or_text(), false);
+  if (!fname.is_binary_or_text()) {
+    // If we haven't specified binary or text, infer it from the type
+    // of the subfile.
+    if ((_subfiles[index]->_flags & SF_text) != 0) {
+      fname.set_text();
+    } else {
+      fname.set_binary();
+    }
+  }
   fname.make_dir();
   pofstream out;
   if (!fname.open_write(out, true)) {
@@ -1828,6 +1887,13 @@ extract_subfile_to(int index, ostream &out) {
 //               file on disk with the nth subfile.  Returns true if
 //               the files are equivalent, or false if they are
 //               different (or the file is missing).
+//
+//               If Filename::set_binary() or set_text() has already
+//               been called, it specifies the nature of the source
+//               file.  If this is different from the text flag of the
+//               subfile, the comparison will always return false.
+//               If this has not been specified, it will be set from
+//               the text flag of the subfile.
 ////////////////////////////////////////////////////////////////////
 bool Multifile::
 compare_subfile(int index, const Filename &filename) {
@@ -1840,27 +1906,62 @@ compare_subfile(int index, const Filename &filename) {
     return false;
   }
 
+  Filename fname = filename;
+  if (fname.is_binary()) {
+    // If we've specified a binary file, it had better be a binary
+    // subfile.
+    if ((_subfiles[index]->_flags & SF_text) != 0) {
+      if (express_cat.is_debug()) {
+        express_cat.debug()
+          << "File is not binary: " << filename << "\n";
+      }
+      return false;
+    }
+
+  } else if (fname.is_text()) {
+    // If we've specified a text file, it had better be a text
+    // subfile.
+    if ((_subfiles[index]->_flags & SF_text) == 0) {
+      if (express_cat.is_debug()) {
+        express_cat.debug()
+          << "File is not text: " << filename << "\n";
+      }
+      return false;
+    }
+
+  } else {
+    // If we haven't specified binary or text, infer it from the type
+    // of the subfile.
+    if ((_subfiles[index]->_flags & SF_text) != 0) {
+      fname.set_text();
+    } else {
+      fname.set_binary();
+    }
+  }
+
   istream *in1 = open_read_subfile(index);
   if (in1 == (istream *)NULL) {
     return false;
   }
 
   pifstream in2;
-  Filename bin_filename = Filename::binary_filename(filename);
-  if (!bin_filename.open_read(in2)) {
+
+  if (!fname.open_read(in2)) {
     express_cat.info()
       << "Cannot read " << filename << "\n";
     return false;
   }
 
-  // Check the file size.
-  in2.seekg(0, ios::end);
-  streampos file_size = in2.tellg();
-
-  if (file_size != (streampos)get_subfile_length(index)) {
-    // The files have different sizes.
-    close_read_subfile(in1);
-    return false;
+  if (fname.is_binary()) {
+    // Check the file size.
+    in2.seekg(0, ios::end);
+    streampos file_size = in2.tellg();
+    
+    if (file_size != (streampos)get_subfile_length(index)) {
+      // The files have different sizes.
+      close_read_subfile(in1);
+      return false;
+    }
   }
 
   // Check the file data, byte-for-byte.

panda/src/express/multifile.h

@@ -129,6 +129,7 @@ PUBLISHED:
   time_t get_subfile_timestamp(int index) const;
   bool is_subfile_compressed(int index) const;
   bool is_subfile_encrypted(int index) const;
+  bool is_subfile_text(int index) const;
 
   streampos get_index_end() const;
   streampos get_subfile_internal_start(int index) const;
@@ -161,6 +162,7 @@ private:
     SF_compressed     = 0x0008,
     SF_encrypted      = 0x0010,
     SF_signature      = 0x0020,
+    SF_text           = 0x0040,
   };
 
   class Subfile {