Add a new 'info' mode to read the header of compressed files

flanglet · flanglet · commit dd9b9d3ddd01 · 2025-09-06T16:08:20.000-04:00
diff --git a/src/app/BlockDecompressor.cpp b/src/app/BlockDecompressor.cpp
@@ -87,6 +87,24 @@ int BlockDecompressor::decompress(uint64& inputSize)
     transform(upperInputName.begin(), upperInputName.end(), upperInputName.begin(), ::toupper);
     bool isStdIn = upperInputName == "STDIN";
 
+    // In mode "info", we want to display the information in the stream header only.
+    // We can reuse the existing code but we need to:
+    //   create an InfoPrinter with a dedicated INFO type
+    //   disable logging outside of this printer (=> _verbosity=0)
+    //   decompress no block (=> _outputName = NONE and --from=1 and --to=1)
+    //   disable threading for proper display (=> _jobs=1)
+    bool isInfo = _ctx.getString("mode") == "y";
+    int vl = _verbosity;
+
+    if (isInfo) {
+        _verbosity = 0;
+        _outputName = "NONE";
+        _ctx.putString("outputName", _outputName);
+        _ctx.putInt("from", 1);
+        _ctx.putInt("to", 1);
+        _ctx.putInt("jobs", 1);
+    }
+
     if (isStdIn == false) {
         vector<string> errors;
         bool isRecursive = (_inputName.length() < 2) ||
@@ -137,9 +155,10 @@ int BlockDecompressor::decompress(uint64& inputSize)
         ss.str(string());
     }
 
-    InfoPrinter listener(_verbosity, InfoPrinter::DECODING, cout);
+    InfoPrinter::Type ipt = isInfo ? InfoPrinter::INFO : InfoPrinter::DECODING;
+    InfoPrinter listener(vl, ipt, cout);
 
-    if (_verbosity > 2)
+    if ((vl > 2) || ((isInfo == true) && (vl > 0)))
         addListener(listener);
 
     int res = 0;
diff --git a/src/app/InfoPrinter.cpp b/src/app/InfoPrinter.cpp
@@ -18,7 +18,7 @@ limitations under the License.
 #include <ios>
 #include <sstream>
 #include "InfoPrinter.hpp"
-#include "../util.hpp"
+#include "../util/strings.hpp"
 
 using namespace kanzi;
 using namespace std;
@@ -145,29 +145,55 @@ void InfoPrinter::processEvent(const Event& evt)
         _map[hash(currentBlockId)] = nullptr;
     }
     else if ((evt.getType() == Event::AFTER_HEADER_DECODING) && (_level >= 3)) {
-        stringstream ss(evt.toString());
-        string s = ss.str();
+        // Special CSV format
+        stringstream ss;
         vector<string> tokens;
+        string s = evt.toString();
         const int nbTokens = tokenizeCSV(s, tokens);
-        ss.str(string());
 
-        if (nbTokens > 1)
-            ss << "Bitstream version: " << tokens[1] << endl;
+        if (_level >= 5) {
+            // JSON text
+            if (nbTokens > 1)
+                ss << ", \"bsVersion\":" << tokens[1];
+
+            if (nbTokens > 2)
+                ss << ", \"checksize\":" << tokens[2];
+
+            if (nbTokens > 3)
+                ss << ", \"blocksize\":" << tokens[3];
+
+            if (nbTokens > 4)
+                ss << ", \"entropy\":" << (tokens[4] == "" ? "none" : "\"" + tokens[4] + "\"");
 
-        if (nbTokens > 2)
-            ss << "Block checksum: " << tokens[2] << (tokens[2] == "NONE" ? "" : " bits") << endl;
+            if (nbTokens > 5)
+                ss << ", \"transforms\":" << (tokens[5] == "" ? "none" : "\"" + tokens[5] + "\"");
 
-        if (nbTokens > 3)
-            ss << "Block size: " << tokens[3] << " bytes" << endl;
+            if (nbTokens > 6)
+                ss << ", \"compressed\":" << (tokens[6] == "" ? "N/A" : tokens[6]);
 
-        if (nbTokens > 4)
-            ss << "Using " << (tokens[4] == "" ? "no" : tokens[4]) << " entropy codec (stage 1)" << endl;
+            if (nbTokens > 7)
+                ss << ", \"original\":" << (tokens[7] == "" ? "N/A" : tokens[7]);
+        }
+        else {
+            // Raw text
+            if (nbTokens > 1)
+                ss << "Bitstream version: " << tokens[1] << endl;
+
+            if (nbTokens > 2)
+                ss << "Block checksum: " << (tokens[2] == "0" ? "NONE" : tokens[2] + " bits") << endl;
+
+            if (nbTokens > 3)
+                ss << "Block size: " << tokens[3] << " bytes" << endl;
 
-        if (nbTokens > 5)
-            ss << "Using " << (tokens[5] == "" ? "no" : tokens[5]) << " transform (stage 2)" << endl;
+            if (nbTokens > 4)
+                ss << "Using " << (tokens[4] == "" ? "no" : tokens[4]) << " entropy codec (stage 1)" << endl;
 
-        if (nbTokens > 7)
-            ss << "Original size: " << tokens[7] << " byte(s)" << endl;;
+            if (nbTokens > 5)
+                ss << "Using " << (tokens[5] == "" ? "no" : tokens[5]) << " transform (stage 2)" << endl;
+
+            if ((nbTokens > 7) && (tokens[7] != ""))
+                ss << "Original size: " << tokens[7] << " byte(s)" << endl;
+        }
 
         _os << ss.str() << endl;
     }
@@ -190,6 +216,7 @@ void InfoPrinter::processHeaderInfo(const Event& evt)
 
    if (_headerInfo++ == 0) {
       // Display header
+      ss << endl;
       ss << "|" << "     File Name      ";
       ss << "|" << "Ver";
       ss << "|" << "Check";
@@ -200,7 +227,7 @@ void InfoPrinter::processHeaderInfo(const Event& evt)
 
       if (_level >= 4) {
          ss << "|" << " Entropy";
-         ss << "|" << "      Transforms      ";
+         ss << "|" << "        Transforms        ";
       }
 
       ss << "|" << endl;
@@ -264,10 +291,10 @@ void InfoPrinter::processHeaderInfo(const Event& evt)
    if ((_level >= 4) && (nbTokens > 5)) {
        string t = tokens[5];
 
-       if (t.length() > 22)
-          t = t.substr(0, 20) + "..";
+       if (t.length() > 26)
+          t = t.substr(0, 24) + "..";
 
-       ss << setw(22) << (t == "" ? "NONE" : t) << "|"; // transforms
+       ss << setw(26) << (t == "" ? "NONE" : t) << "|"; // transforms
    }
 
    _os << ss.str() << endl;
diff --git a/src/app/Kanzi.cpp b/src/app/Kanzi.cpp
@@ -32,8 +32,8 @@ limitations under the License.
 using namespace kanzi;
 using namespace std;
 
-static const string CMD_LINE_ARGS[14] = {
-    "-c", "-d", "-i", "-o", "-b", "-t", "-e", "-j", "-v", "-l", "-s", "-x", "-f", "-h"
+static const string CMD_LINE_ARGS[15] = {
+    "-c", "-d", "-i", "-o", "-b", "-t", "-e", "-j", "-v", "-l", "-s", "-x", "-f", "-h", "-y"
 };
 
 //static const int ARG_IDX_COMPRESS = 0;
@@ -53,7 +53,7 @@ static const int ARG_IDX_LEVEL = 9;
 static const string KANZI_VERSION = "2.4.0";
 static const string APP_HEADER = "Kanzi " + KANZI_VERSION + " (c) Frederic Langlet";
 static const string APP_SUB_HEADER = "Fast lossless data compressor.";
-static const string APP_USAGE = "Usage: kanzi [-c|-d] [flags and files in any order]";
+static const string APP_USAGE = "Usage: kanzi [-c|-d|-y] [flags and files in any order]";
 
 
 #ifdef CONCURRENCY_ENABLED
@@ -77,14 +77,16 @@ void printHelp(Printer& log, const string& mode, bool showHeader)
    log.println("Options\n", true);
    log.println("   -h, --help", true);
 
-   if ((mode.compare(0, 1, "c") != 0) && (mode.compare(0, 1, "d") != 0)) {
+   if ((mode != "c") && (mode != "d") && (mode != "y")) {
        log.println("        Display this message.", true);
        log.println("        Use in conjunction with -c to print information for compression,", true);
        log.println("        or -d to print information for decompression.\n", true);
        log.println("   -c, --compress", true);
        log.println("        Compress mode\n", true);
        log.println("   -d, --decompress", true);
        log.println("        Decompress mode\n", true);
+       log.println("   -y, --info", true);
+       log.println("        Info mode: display information about compressed files\n", true);
    }
    else {
        log.println("        Display this message.\n", true);
@@ -101,23 +103,26 @@ void printHelp(Printer& log, const string& mode, bool showHeader)
    log.println(ss.str(), true);
    log.println("        If this option is not provided, kanzi reads data from stdin.\n", true);
    ss.str(string());
-   log.println("   -o, --output=<outputName>", true);
 
-   if (mode.compare(0, 1, "c") == 0) {
-       log.println("        Optional name of the output file or directory (defaults to", true);
-       log.println("        <inputName.knz> if input is <inputName> or 'stdout' if input is 'stdin').", true);
-       log.println("        or 'none' or 'stdout'.\n", true);
-   }
-   else if (mode.compare(0, 1, "d") == 0) {
-       log.println("        Optional name of the output file or directory (defaults to", true);
-       log.println("        <inputName> if input is <inputName.knz> or 'stdout' if input is 'stdin').", true);
-       log.println("        or 'none' or 'stdout'.\n", true);
-   }
-   else {
-       log.println("        Optional name of the output file or 'none' or 'stdout'.\n", true);
+   if (mode != "y") {
+       log.println("   -o, --output=<outputName>", true);
+   
+       if (mode == "c") {
+           log.println("        Optional name of the output file or directory (defaults to", true);
+           log.println("        <inputName.knz> if input is <inputName> or 'stdout' if input is 'stdin').", true);
+           log.println("        or 'none' or 'stdout'.\n", true);
+       }
+       else if (mode == "d") {
+           log.println("        Optional name of the output file or directory (defaults to", true);
+           log.println("        <inputName> if input is <inputName.knz> or 'stdout' if input is 'stdin').", true);
+           log.println("        or 'none' or 'stdout'.\n", true);
+       }
+       else {
+           log.println("        Optional name of the output file or 'none' or 'stdout'.\n", true);
+       }
    }
 
-   if (mode.compare(0, 1, "c") == 0) {
+   if (mode == "c") {
        log.println("   -b, --block=<size>", true);
        log.println("        Size of blocks (default 4|8|16|32 MiB based on level, max 1 GiB, min 1 KiB).", true);
        log.println("        'auto' means that the compressor derives the best value", true);
@@ -162,24 +167,29 @@ void printHelp(Printer& log, const string& mode, bool showHeader)
    log.println("        4=display block size and timings, 5=display extra information", true);
    log.println("        Verbosity is reduced to 1 when files are processed concurrently", true);
    log.println("        Verbosity is reduced to 0 when the output is 'stdout'\n", true);
-   log.println("   -f, --force", true);
-   log.println("        Overwrite the output file if it already exists\n", true);
-   log.println("   --rm", true);
 
-   if (mode.compare(0, 1, "c") == 0) {
+   if (mode != "y") {
+       log.println("   -f, --force", true);
+       log.println("        Overwrite the output file if it already exists\n", true);
+   }
+
+   if (mode == "c") {
+       log.println("   --rm", true);
        log.println("        Remove the input file after successful compression.", true);
+       log.println("        If the input is a folder, all processed files under the folder are removed.\n", true);
    }
-   else {
+   else if (mode == "d") {
+       log.println("   --rm", true);
        log.println("        Remove the input file after successful decompression.", true);
+       log.println("        If the input is a folder, all processed files under the folder are removed.\n", true);
    }
 
-   log.println("        If the input is a folder, all processed files under the folder are removed.\n", true);
    log.println("   --no-link", true);
    log.println("        Skip links\n", true);
    log.println("   --no-dot-file", true);
    log.println("        Skip dot files\n", true);
 
-   if (mode.compare(0, 1, "d") == 0) {
+   if (mode == "d") {
        log.println("   --from=blockId", true);
        log.println("        Decompress starting at the provided block (included).", true);
        log.println("        The first block ID is 1.\n", true);
@@ -191,7 +201,7 @@ void printHelp(Printer& log, const string& mode, bool showHeader)
        log.println("  kanzi --decompress --input=foo.knz --force --verbose=2 --jobs=2\n", true);
    }
 
-   if (mode.compare(0, 1, "c") == 0) {
+   if (mode == "c") {
        log.println("", true);
        log.println("Transforms\n", true);
        log.println("  BWT: Burrows Wheeler Transform is a transform that reorders symbols", true);
@@ -399,8 +409,8 @@ int processCommandLine(int argc, const char* argv[], Context& map, Printer& log)
 
         // Extract verbosity, output and mode first
         if ((arg == "-c") || (arg.compare(0, 10, "--compress") == 0)) {
-            if (mode == "d") {
-                cerr << "Both compression and decompression options were provided." << endl;
+            if (mode != "") {
+                cerr << "Only one mode can be provided (already got '" << mode << "'" << endl;
                 return Error::ERR_INVALID_PARAM;
             }
 
@@ -409,15 +419,25 @@ int processCommandLine(int argc, const char* argv[], Context& map, Printer& log)
         }
 
         if ((arg == "-d") || (arg.compare(0, 12, "--decompress") == 0)) {
-            if (mode == "c") {
-                cerr << "Both compression and decompression options were provided." << endl;
+            if (mode != "") {
+                cerr << "Only one mode can be provided (already got '" << mode << "'" << endl;
                 return Error::ERR_INVALID_PARAM;
             }
 
             mode = "d";
             continue;
         }
 
+        if ((arg == "-y") || (arg.compare(0, 10, "--info") == 0)) {
+            if (mode != "") {
+                cerr << "Only one mode can be provided (already got '" << mode << "')" << endl;
+                return Error::ERR_INVALID_PARAM;
+            }
+
+            mode = "y";
+            continue;
+        }
+
         if ((ctx == ARG_IDX_VERBOSE) || (arg.compare(0, 10, "--verbose=") == 0)) {
            if (verboseFlag == true) {
                 WARNING_OPT_DUPLICATE("verbosity level", arg);
@@ -467,12 +487,17 @@ int processCommandLine(int argc, const char* argv[], Context& map, Printer& log)
         }
     }
     else {
-        string str = outputName;
-        transform(str.begin(), str.end(), str.begin(), ::toupper);
+        if (mode == "y") {
+           log.println("Warning: ignoring option [" + outputName + "]. Only applicable in compression or decompression mode.", verbose > 0);
+        }
+        else {
+           string str = outputName;
+           transform(str.begin(), str.end(), str.begin(), ::toupper);
 
-        if (str == "STDOUT") {
-            verbose = 0;
-            verboseFlag = true;
+           if (str == "STDOUT") {
+               verbose = 0;
+               verboseFlag = true;
+           }
         }
     }
 
@@ -494,7 +519,8 @@ int processCommandLine(int argc, const char* argv[], Context& map, Printer& log)
            arg = arg.substr(k);
         }
 
-        if ((arg == "-c") || (arg == "-d") || (arg == "--compress") || (arg == "--decompress")) {
+        if ((arg == "-c") || (arg == "-d") || (arg == "-y") || (arg == "--compress") || (arg == "--decompress") ||
+            (arg == "--info")) {
             if (ctx != -1) {
                 WARNING_OPT_NOVALUE(CMD_LINE_ARGS[ctx]);
             }
@@ -511,6 +537,11 @@ int processCommandLine(int argc, const char* argv[], Context& map, Printer& log)
                 WARNING_OPT_DUPLICATE(arg, "true");
             }
 
+            if (mode == "y") {
+                WARNING_OPT_COMP_ONLY(arg);
+                continue;
+            }
+
             overwrite = 1;
             ctx = -1;
             continue;
@@ -559,6 +590,10 @@ int processCommandLine(int argc, const char* argv[], Context& map, Printer& log)
                 WARNING_OPT_DUPLICATE(arg, "true");
             }
 
+            if (mode == "y") {
+                WARNING_OPT_COMP_ONLY(arg);
+                continue;
+            }
             remove = 1;
             ctx = -1;
             continue;
@@ -630,6 +665,11 @@ int processCommandLine(int argc, const char* argv[], Context& map, Printer& log)
                 WARNING_OPT_DUPLICATE(msg, arg);
             }
             else {
+                if (mode == "y") {
+                    WARNING_OPT_COMP_ONLY(arg);
+                    continue;
+                }
+
                 if ((arg.length() >= 2) && (arg[0] == '.') && (arg[1] == PATH_SEPARATOR)) {
                    arg = (arg.length() == 2) ? arg.substr(0, 1) : arg.substr(2);
                 }
@@ -1072,7 +1112,7 @@ int main(int argc, const char* argv[])
             }
         }
 
-        if (mode == "d") {
+        if ((mode == "d") || (mode == "y")) {
             try {
                 BlockDecompressor bd(ctx);
                 uint64 read = 0;
diff --git a/src/io/CompressedInputStream.cpp b/src/io/CompressedInputStream.cpp
@@ -379,7 +379,7 @@ void CompressedInputStream::readHeader()
         string inputName = _ctx.getString("inputName", "");
         ss << inputName << ",";
         ss << bsVersion << ",";
-        string ckSize = "NONE";
+        string ckSize = "0";
 
         if (_hasher32 != nullptr)
             ckSize = "32";
