Support for decoding textual log files

This is for use by the cbl-log and cblite tools.

- New class TextLogDecoder parses LiteCore text logs the same way as
  LogDecode does for binary logs.
- MultiLogDecoder recognizes when a log file is text, and opens it
  using a TextLogDecoder.
- Fixed a few minor bugs in MultiLogDecoder.
- Added a few more ANSI escape codes to Tool.hh

Author: snej

LiteCore/Logging/LogDecoder.hh

@@ -31,6 +31,9 @@ namespace litecore {
             unsigned microsecs;
         };
 
+        static constexpr Timestamp kMinTimestamp{0, 0};
+        static constexpr Timestamp kMaxTimestamp{std::numeric_limits<time_t>::max(), 999999};
+
         virtual ~LogIterator() = default;
 
         /** Decodes the entire log and writes it to the output stream, with timestamps.

LiteCore/Logging/MultiLogDecoder.hh

@@ -12,6 +12,7 @@
 
 #pragma once
 #include "LogDecoder.hh"
+#include "TextLogDecoder.hh"
 #include <algorithm>
 #include <climits>
 #include <fstream>
@@ -27,22 +28,20 @@ namespace litecore {
     class MultiLogDecoder : public LogIterator {
       public:
         MultiLogDecoder() {
-            _startTime = {UINT_MAX, 0};
-            for ( unsigned i = 0; i <= kMaxLevel; i++ ) _startTimeByLevel[i] = {UINT_MAX, 0};
+            _startTime = kMaxTimestamp;
+            for ( unsigned i = 0; i <= kMaxLevel; i++ ) _startTimeByLevel[i] = kMaxTimestamp;
         }
 
         /// Adds a log iterator. Must be called before calling \ref next().
         /// The iterator is assumed to be at its start, so its \ref next() will be called first.
         void add(LogIterator* log) {
             assert(!_current);
-            if ( !log->next() ) return;
-
-            _logs.push(log);
 
             auto startTime = log->startTime();
             _startTime     = std::min(_startTime, startTime);
-            auto level     = log->level();
-            if ( level >= 0 && level <= kMaxLevel )
+            if ( !log->next() ) return;
+            _logs.push(log);
+            if ( auto level = log->level(); level >= 0 && level <= kMaxLevel )
                 _startTimeByLevel[level] = std::min(_startTimeByLevel[level], startTime);
         }
 
@@ -52,19 +51,28 @@ namespace litecore {
             if ( !in ) return false;
             in.exceptions(std::ifstream::badbit);
             _inputs.push_back(std::move(in));
-            LogDecoder decoder(_inputs.back());
+            std::unique_ptr<LogIterator> decoder;
+            if ( TextLogDecoder::looksTextual(_inputs.back()) )
+                decoder = std::make_unique<TextLogDecoder>(_inputs.back());
+            else
+                decoder = std::make_unique<LogDecoder>(_inputs.back());
             _decoders.push_back(std::move(decoder));
-            add(&_decoders.back());
+            add(_decoders.back().get());
             return true;
         }
 
         /// Time when the earliest log began
         [[nodiscard]] Timestamp startTime() const override { return _startTime; }
 
+        /// Time that earliest logs at `level` begin, or kMaxTimestamp if none.
+        [[nodiscard]] Timestamp startTimeOfLevel(unsigned level) const { return _startTimeByLevel[level]; }
+
         /// First time when logs of all levels are available
         [[nodiscard]] Timestamp fullStartTime() const {
-            Timestamp fullStartTime = {0, 0};
-            for ( unsigned i = 0; i <= kMaxLevel; i++ ) fullStartTime = std::max(fullStartTime, _startTimeByLevel[i]);
+            Timestamp fullStartTime = kMinTimestamp;
+            for ( auto& ts : _startTimeByLevel ) {
+                if ( fullStartTime < ts && ts != kMaxTimestamp ) fullStartTime = ts;
+            }
             return fullStartTime;
         }
 
@@ -116,10 +124,10 @@ namespace litecore {
         std::priority_queue<LogIterator*, std::vector<LogIterator*>, logcmp> _logs;
         LogIterator*                                                         _current{nullptr};
         Timestamp                                                            _startTime{};
-        Timestamp                                                            _startTimeByLevel[kMaxLevel + 1]{};
+        std::array<Timestamp, kMaxLevel + 1>                                 _startTimeByLevel{};
 
-        std::deque<LogDecoder>    _decoders;
-        std::deque<std::ifstream> _inputs;
+        std::vector<std::unique_ptr<LogIterator>> _decoders;
+        std::deque<std::ifstream>                 _inputs;
     };
 
 

LiteCore/Logging/TextLogDecoder.hh

@@ -0,0 +1,132 @@
+//
+// Created by Jens Alfke on 12/9/25.
+//
+
+#pragma once
+#include "LogDecoder.hh"
+#include "ParseDate.hh"
+#include <istream>
+#include <regex>
+#include <stdexcept>
+
+namespace litecore {
+
+    /** Parses LiteCore-generated textual log files. */
+    class TextLogDecoder : public LogIterator {
+      public:
+        /** Returns true if the stream `in` appears to contain textual log data. */
+        static bool looksTextual(std::istream& in) {
+            auto pos       = in.tellg();
+            char chars[27] = {};
+            bool result    = true;
+            in.read((char*)chars, sizeof(chars));
+            in.seekg(pos);
+            return looksLikeLogLine(std::string_view(chars, std::size(chars)));
+        }
+
+        /** Initializes decoder with a stream written by LiteCore's textual log encoder. */
+        explicit TextLogDecoder(std::istream& in) : _in(in) {
+            _in.exceptions(std::istream::badbit);
+            if ( next() && next() )  // Read header line to get the initial timestamp
+                _startTime = _curTimestamp;
+        }
+
+        bool next() override {
+            if ( _line.empty() ) {
+                // Read next line if there's not one in the buffer:
+                if ( !_in || _in.peek() < 0 ) return false;
+                std::getline(_in, _line);
+                if ( _line.empty() ) return false;
+            }
+
+            // Example: 2025-12-09T06:47:55.507699Z WS Verbose Obj=/JRepl@1175308903/…/ Received 58-byte message
+
+            std::string_view rest(_line);
+            auto             nextColumn = [&] {
+                auto             next   = rest.find(' ');
+                std::string_view column = rest.substr(0, next);
+                rest                    = rest.substr(next + 1);
+                return column;
+            };
+
+            auto timestamp = nextColumn();
+            auto micros    = std::stoul(std::string(timestamp.substr(timestamp.size() - 7, 6)));
+            auto millis    = fleece::ParseISO8601Date(timestamp);
+            if ( millis == kInvalidDate || millis < 0x19000000 || micros > 999999 )
+                throw std::runtime_error("Could not parse timestamp in log line:  " + _line);
+            _curTimestamp = {millis / 1000, unsigned(micros)};
+
+            _curDomain = nextColumn();
+
+            auto levelStr = nextColumn();
+            if ( auto i = std::ranges::find(kLevelNames, levelStr); i != std::end(kLevelNames) )
+                _curLevel = i - std::begin(kLevelNames);
+            else
+                _curLevel = 0;
+
+            _curObject.clear();
+            _curObjectID = 0;
+            if ( rest.starts_with("Obj=/") ) {
+                std::string_view obj = nextColumn();
+                if ( auto size = obj.size(); size >= 13 && obj.ends_with('/') ) {
+                    if ( auto pos = obj.rfind('#'); pos != std::string::npos ) {
+                        _curObject   = obj.substr(5, size - 6);  // trim 'Obj=/' and '/ '
+                        _curObjectID = std::stoul(std::string(obj.substr(pos + 1, size - 2 - pos)));
+                    }
+                }
+            }
+
+            _curMessage = rest;
+
+            // Add any following non-log-format lines to the message:
+            _line.clear();
+            while ( _in && _in.peek() >= 0 ) {
+                std::getline(_in, _line);
+                if ( _line.empty() ) break;
+                if ( looksLikeLogLine(_line) ) break;
+                _curMessage += '\n';
+                _curMessage += _line;
+                _line.clear();
+            }
+
+            return true;
+        }
+
+        Timestamp startTime() const override { return _startTime; }
+
+        Timestamp timestamp() const override { return _curTimestamp; }
+
+        int8_t level() const override { return _curLevel; }
+
+        const std::string& domain() const override { return _curDomain; }
+
+        uint64_t objectID() const override { return _curObjectID; }
+
+        const std::string* objectDescription() const override { return &_curObject; }
+
+        void decodeMessageTo(std::ostream& out) override { out << _curMessage; }
+
+      private:
+        static bool looksLikeLogLine(std::string_view line) {
+            if ( line.size() < 27 ) return false;
+            for ( uint8_t c : line.substr(0, 27) ) {
+                if ( !isdigit(c) && c != '-' && c != ':' && c != '.' && c != 'Z' && c != 'T' ) return false;
+            }
+            return true;
+        }
+
+        static constexpr std::string_view kLevelNames[] = {"Debug", "Verbose", "Info", "WARNING", "ERROR"};
+
+        std::istream& _in;
+        Timestamp     _startTime{};
+        std::string   _line;
+
+        Timestamp   _curTimestamp;
+        int8_t      _curLevel;
+        std::string _curDomain;
+        std::string _curObject;
+        uint64_t    _curObjectID;
+        std::string _curMessage;
+    };
+
+}  // namespace litecore

tool_support/Tool.hh

@@ -182,8 +182,12 @@ class Tool {
 
     std::string ansiItalic() { return ansi("3"); }
 
+    std::string ansiNoItalic() { return ansi("23"); }
+
     std::string ansiUnderline() { return ansi("4"); }
 
+    std::string ansiNoUnderline() { return ansi("24"); }
+
     std::string ansiRed() { return ansi("31"); }
 
     std::string ansiReset() { return ansi("0"); }