|
LogViewer
A modern log viewer with AI-assisted analysis (Qt 6)
|
Version: 1.11.0
Application: LogViewer
LogViewer is a desktop application for inspecting, filtering, and analysing log files. It supports multiple log formats — from plain XML/CSV to binary formats such as AUTOSAR DLT and POSIX Evlog — and provides a rich set of analysis panels, AI-assisted insights, and a side-by-side comparison view.
Key capabilities at a glance:
When you first launch LogViewer you see the main window with:
All panels can be resized, floated, or hidden. Drag a panel's title bar to move it; double-click the title bar to float it as a separate window.
| Extension | Format | Notes |
|---|---|---|
.xml | Generic XML | Default when extension is unrecognised |
.csv | Comma-Separated Values | |
.json | JSON array or NDJSON | Array [{…},…] or one object per line |
.jsonl | Newline-delimited JSON | Alias for NDJSON; same parser as .json |
.asc | CAN/Vector CANalyzer | Optional DBC for signal decoding |
.dlt | AUTOSAR DLT binary | Storage header + verbose/non-verbose payloads |
.evl | POSIX 1003.25 Evlog binary | Optional template directory for BINARY payloads |
LogViewer accepts two JSON layouts:
[{"ts":"…","level":"INFO","msg":"…"}, …]Nested objects are flattened with dot notation: {"ctx":{"host":"db"}} becomes the key ctx.host. Arrays are flattened with index suffixes: {"tags":["a","b"]} becomes tags.0 and tags.1.
If you open a file with an unknown or missing extension, LogViewer prompts you to pick the format before parsing begins.
File > Open (Ctrl+O) opens a file-picker dialog. The dialog filters by all supported extensions. After you select a file, the application parses it in the background and displays a progress indicator in the status bar. The UI remains interactive during loading.
Merging multiple files into one view is possible by opening them one after another — subsequent opens append events to the current dataset. Use File > Clear to start fresh.
File > Open Results Folder… (Ctrl+Alt+O), or dropping a folder on the window, opens the Run Explorer for a test-run results folder (one safeAPI *__merged_logs.txt per test, with Robot Framework's xunit.xml / output.xml). It lists every test in run order with its verdict, duration, entry count and failure message: the step failure that failed the test, or — when the log records no failing step, e.g. a failed suite setup — the message from xunit.xml (else output.xml). Filter by verdict, sort by any column, and tick Group failures by message to collapse failures that differ only in numbers. Double-click a test to load its log; Open with Previous Test (button or right-click) loads it merged with the test that ran just before it, to check whether that test's teardown left the system in a bad state. The folder is scanned in the background; Cancel stops the scan.
File > Follow File (Tail) (Ctrl+T) watches the currently loaded log file and automatically appends new events as they are written to disk. This is useful for monitoring a running service without reopening the file manually.
Supported formats: NDJSON (.json / .jsonl), CAN/ASC (.asc), AUTOSAR DLT (.dlt).
Not supported: XML (.xml), CSV (.csv) — these formats require full-file parsing.
Drag a log file from a file manager onto the main window. The application detects the extension and begins parsing immediately. If the extension is unrecognised, the format-picker dialog appears.
File > Recent Files lists the last ten files you opened. Click any entry to re-open it. Paths are stored per session; they are cleared if a file is deleted or moved.
Before or after opening an .asc file, choose File > Load DBC… and select a .dbc CAN database. Once loaded:
SIG:<name> fields.If you load an .asc file when a DBC is already loaded, decoding happens automatically. The active DBC path is saved in the session and reused on next launch.
Evlog BINARY payloads can be decoded into structured fields using template files. Choose File > Load Evlog Templates… and select a directory containing .t, .tmpl, or .template files. Templates are keyed by (facility, event_type) and declare typed field layouts with an optional format string.
.evl files are parsed with template decoding enabled.info field.Template file syntax example:
The events table shows one row per log event. Default columns:
| Column | Description |
|---|---|
| # | Sequential row number |
| Timestamp | Event time (float seconds for ASC/DLT/Evlog, ISO date for XML/CSV) |
| Level / Type | Severity level or event type (colour-coded based on configurable field) |
| Source / Actor | Originating component or actor (format-dependent) |
| Message / Info | The main event text |
Column visibility, order, and width are user-configurable in Tools > Settings > Columns. Column widths are persisted across sessions.
Rows are colour-coded based on the type filter field, which is auto-detected when a file is loaded. Colours are mapped in Tools > Settings > Colors (any column can have colour rules — the first matching column wins).
| Action | Result |
|---|---|
| Ctrl+C | Copies selected rows as tab-separated plain text |
| Right-click > Copy as JSON | Copies selected events as a JSON array |
| Right-click > Copy as CSV | Copies selected events as CSV with a header row |
Multiple rows can be selected with Shift+Click or Ctrl+Click before copying.
Press Ctrl+G (or Edit > Jump to Timestamp) to open the Jump to Timestamp dialog. Enter a timestamp value in the format used by the current log file (e.g. 1234567.890 for ASC float seconds, or 2024-03-15 14:22:03.456 for XML/CSV ISO format). The table scrolls to and selects the event with the nearest matching timestamp.
All filter controls are in the Filters panel on the left. Changes take effect immediately (debounced at 150 ms to avoid excessive recomputation on large datasets).
The search bar at the top of the window (Ctrl+F to focus) supports four match modes selectable from the dropdown next to it:
| Mode | Description |
|---|---|
| Plain | Case-insensitive substring search |
| Regex | Full regular expression (RE2 syntax) |
| Fuzzy | Approximate matching — tolerates minor typos |
| Wildcard | Shell-style * and ? patterns |
Press Enter or click Search to run. Matching rows are highlighted; use the up/down arrows next to the bar to navigate between results.
The Type Filter panel shows all distinct values of the type filter field. Check or uncheck values to show or hide event categories. Select All / Deselect All buttons are available.
Auto-detection: When a file is loaded, LogViewer automatically detects the type field by scanning the first 100 events for common names: level, type, severity, priority, category, and others. If none are found, a prompt asks for the field name — the answer is saved and used for future loads.
Manual override: Set a fixed field name in Tools > Settings > General ("Type Filter Field"). Leave the field empty to re-enable auto-detection. Setting it explicitly is useful when your logs use a non-standard field name or you have multiple file types with different naming conventions.
The Time Range panel lets you restrict visible events to a time window:
The range is applied as an additional filter on top of any active text/type filters. Clear the fields and click Apply to remove the time constraint.
The Actors panel in the left dock shows a tree of actor definitions. Enable or disable individual actors to include or exclude their events. Actors are defined in the Actor Definitions panel (content tab).
Filter profiles let you save and quickly switch between named filter states.
Switch between analysis views using the tab bar in the content area. Panels refresh lazily — computation only runs when the tab is visible, keeping the UI fast on large datasets.
The Statistics tab displays aggregate metrics for the loaded data:
SIG:* signalThe Signals tab plots numeric SIG:* field values over time as line series.
Each series is automatically downsampled to 2 000 points when the dataset is larger, keeping the chart responsive. Zoom by scrolling; pan by clicking and dragging the axis.
The Timeline tab shows a bar histogram of event volume over time. Use it to identify bursts, quiet periods, and temporal clusters at a glance.
The Patterns tab groups structurally similar log lines using template clustering. Messages that share the same structure but differ only in numeric or variable parts are collapsed into a single template with a match count.
The Trace Viewer tab groups visible events by a user-selected correlation field (e.g. requestId, sessionId, traceId) and displays them in a tree. Each top-level row represents one unique value and shows aggregate statistics.
Select the field to group by from the Correlation field dropdown. Click Rescan Fields if you change the loaded data and need to update the available field names.
| Column | Meaning |
|---|---|
| Trace / Span | Unique value of the correlation field |
| Events | Total number of matching events |
| Errors | Events whose level/severity is ERROR, CRITICAL, or FATAL |
| Duration | Time between first and last event in the group |
| First Seen | Timestamp of the earliest event |
| Last Seen | Timestamp of the most recent event |
Rows with at least one error are shown in red.
Expand a top-level row to see sub-breakdowns:
type, action, level) within the group.spanId / parentSpanId fields (OpenTelemetry-style distributed tracing), child spans are nested under their parent, forming a call tree.Type in the Filter traces box to instantly show only rows whose Trace ID contains the search text (case-insensitive).
Double-click any top-level row to filter the Events table to the events in that trace group. Use the Clear Filter button to restore the full view.
The Bookmarks tab lets you annotate important events:
Bookmarks are saved in the session file.
The Scenarios tab lets you define multi-step event patterns and search for them in the loaded log:
The Actors tab displays an event attribution tree. Events are assigned to actors based on the actor definitions you configure in the Actor Definitions panel:
The Side-by-Side panel loads two log files independently and displays them in a split view, making it easy to compare what happened in two logs at the same point in time.
Each side has its own Open button and a filename label. Click the button to open a file-picker dialog. Both sides support all log formats (XML, CSV, JSON, ASC, DLT, Evlog). File parsing runs on a background thread; the split view remains interactive while loading.
You can also drag a file from the file manager directly onto the Side by Side panel:
This is the same drag-and-drop gesture as the main window (which merges a file into the main events table), just targeted to one side of the comparison view.
Select a mode from the Sync dropdown in the toolbar:
When you click a row on either side, the application computes the timestamp of that event and scrolls the opposite side to the event with the nearest timestamp (binary search, O(log n)). This works best when both logs share a common time base (e.g. two DLT recordings from the same system clock).
If the selected event has no parseable timestamp the panel falls back to row-mirroring (same row index on the other side).
Use Manual sync when the two files have different time bases — for example, comparing a DLT log (epoch-based absolute seconds) with an ASC log (elapsed seconds from start of recording).
offset = rightRefTimestamp − leftRefTimestamp. All subsequent sync lookups apply this offset: targetRight = leftTimestamp + offset.The sync status label below the toolbar shows whether reference points have been set and the computed offset value.
Both views scroll completely independently. Useful when you only need to see two files simultaneously without needing them to stay aligned.
The Signal Browser is a left-dock panel (second tab, alongside Filters). It shows the CAN frame and signal tree from a loaded DBC file.
When no DBC is loaded, the panel shows a placeholder message. Load a DBC via File > Load DBC… to populate it.
Open Tools > Settings > AI and configure:
| Setting | Description |
|---|---|
| Provider | ollama, lmstudio, openai, anthropic, google, xai |
| Base URL | API endpoint (pre-filled with defaults; edit for local providers) |
| Model | Model name (e.g. llama3.2, gpt-4o, claude-3-5-sonnet) |
| API Key | Required for cloud providers; stored encrypted on disk |
| Timeout | Request timeout in seconds (30–3600) |
Local providers (recommended for privacy):
ollama serve, pull a model (ollama pull llama3.2), and set the URL to http://localhost:11434.http://localhost:1234.The analysis runs on a background thread. Results appear in the response area when complete.
Filter-aware: AI analysis respects the active filters. Only the events currently visible in the table are sent for analysis. On large datasets, a smart sampler caps at 5 000 events with even distribution across the timeline to stay within the model's context window.
A layout captures the current dock geometry — which panels are visible, their sizes, and their positions. Layouts let you switch between different working configurations in one click.
Saving a layout:
Restoring a layout:
Deleting a layout:
Two built-in layouts are always available:
| Layout | Optimised for |
|---|---|
| XML / CSV Log | Generic log inspection: events table + filters + details |
| CAN Analysis | Signal Browser, Signal Plot, Statistics, and Timeline visible |
Named layouts are persisted in the application settings and survive restarts.
File > Export sub-menu:
| Action | Output |
|---|---|
| Export as CSV… | Comma-separated file — all visible (filtered) events |
| Export as JSON… | JSON array of event objects |
| Export as XML… | XML file mirroring the input XML structure |
All export actions apply the current filter, so only visible rows are written. A Save As dialog lets you choose the output path; the last-used directory is remembered per export type.
For quick clipboard transfer without saving a file, use right-click > Copy as JSON or Copy as CSV in the events table.
Three colour themes are available under View > Theme:
| Theme | Description |
|---|---|
| Light | White background, dark text |
| Dark | Dark background, light text; easier on the eyes in low-light environments |
| System | Follows the OS preference (light/dark mode toggle on macOS and Windows) |
The selected theme is applied instantly and persisted across sessions.
Settings are stored per-platform:
| Platform | Location |
|---|---|
| macOS | ~/Library/Application Support/LogViewer/config.json |
| Linux | ~/.config/LogViewer/config.json |
| Windows | APPDATA%\LogViewer\config.json |
Open the settings editor at Tools > Settings. The dialog has tabs:
level=ERROR → red background)You can also edit config.json directly; changes take effect on the next application start.
Open Tools > Manage Plugins… to browse, install, configure, and uninstall plugins without restarting the application. This is the single place for all plugin management — the Settings dialog no longer has a separate Plugins tab.
Each row shows:
| Column | Meaning |
|---|---|
| Name | Display name from the plugin's metadata |
| ID | Unique plugin identifier |
| Version | Plugin version string |
| Type | Functional category (Parser, Analyzer, AI Provider, …) |
| Status | Active / Initialized (green), Loaded (black), Disabled / Not loaded (grey), Error (red) |
| Auto-load | Whether the plugin loads automatically on startup |
| Button | What it does |
|---|---|
| Install… | File picker → accepts .zip plugin archives. The archive is extracted to the plugins directory and loaded immediately. |
| Uninstall | Removes the selected plugin's files from the plugins directory. |
| Enable | Loads a registered-but-not-running plugin into the current session. |
| Disable | Unloads a running plugin without removing its files. |
| Refresh | Rescans the plugins directory and updates the table. |
Selecting a row in the table populates the Plugin Details pane below with:
Auto-load on startup — tick the checkbox to have the plugin load automatically on every application start. The preference is saved immediately.
License key — if the selected plugin requires a license, enter the key in the field and click Set License. The key is validated by the plugin; a success or rejection message is shown.
The plugins directory is shown at the top of the dialog. Use Tools > Reload Plugins (Ctrl+Shift+P) to reload all plugins at once without opening the dialog.
| Shortcut | Action |
|---|---|
| Ctrl+O | Open log file |
| Ctrl+Alt+O | Open results folder (Run Explorer) |
| Ctrl+F | Focus search bar |
| Enter (in search bar) | Run search |
| Ctrl+G | Jump to timestamp |
| Ctrl+C | Copy selected events (plain text) |
| Ctrl+Shift+S | Switch to Side by Side tab |
| Ctrl+Z | Undo last filter change |
| Ctrl+S | Save session |
| Ctrl+Q | Quit application |
| F5 | Refresh / re-apply all filters |
Context-menu actions (right-click on the events table):
| Action | Description |
|---|---|
| Copy as JSON | Selected events → JSON array to clipboard |
| Copy as CSV | Selected events → CSV with header to clipboard |
| Add Bookmark | Annotate the selected event |
| Jump to Event | Scroll to this event (when invoked from search results) |
For build instructions and developer documentation see ARCHITECTURE.md, DEVELOPMENT.md, and SDK_GETTING_STARTED.md.