Session statistics (no persistence) #46

New issue

Closed

opened 2026-02-15 04:14:27 +03:00 by NiXTheDev · 0 comments

NiXTheDev commented

2026-02-15 04:14:27 +03:00

(Migrated from github.com)

Description

Show real-time processing statistics for the current message chain without persisting any data.

Use Case

Users want visibility into how their substitutions are performing without storing data long-term.

Implementation Plan

In-memory tracking (per message only)
- Track substitution count in current chain
- Measure compilation and execution time
- Count cache hits/misses
- No persistence beyond message processing
Smart placement logic
Calculate dynamically if performance text fits:
```
available_space = 4096 - result_length - performance_length - 1 (for newline)
```
If fits: Append to result message with newline separator
```
Result text

Performed X substitutions in Yms
```
If doesn't fit: Send as separate message
Edit handling (complex logic)
When user edits the sed command:

a) Removing 'p' flag:
- Delete the performance message (if separate)
- Or edit result to remove performance line
b) Adding 'p' flag:
- Check if fits in result message
- If yes: Inline it
- If no: Send as separate message
c) Command changed (different result):
- Recalculate space availability
- If was separate and now fits: Delete separate message, inline in result
- If was inline and now doesn't fit: Remove from result, send as separate
- If still doesn't fit: Update both messages
- If still fits inline: Update result message only
Time unit logic
- < 1000ms: "Xms"
- < 60000ms: "Xs" (e.g., "1.2s")
- < 3600000ms: "Xm Ys"
- = 3600000ms: "Xh Ym"
Data to collect
- Number of successful substitutions
- Total processing time
- Cache hit/miss count (optional)
- Time per substitution (optional)
Storage for edits
- Store bot message IDs in memory:
  - Result message ID
  - Performance message ID (if separate)
- Track if performance is inlined or separate
- Clean up after edit window expires (48h)

In-Memory Tracking (No Persistence)

Store performance message metadata in memory only:

interface PerformanceMessageInfo {
	chatId: number;
	targetMessageId: number; // User's sed command message
	resultMessageId: number; // Bot's reply with substitution result
	performanceMessageId?: number; // Separate performance message (undefined if inlined)
	isInlined: boolean;
	timestamp: number;
}

// In-memory Map: key = `${chatId}:${targetMessageId}`
const performanceMessageTracker = new Map<string, PerformanceMessageInfo>();

// Cleanup old entries periodically (after 48h, matching Telegram edit window)

Acceptance Criteria

Calculates available space dynamically (includes newline)
Inlines performance when space available
Sends separate message when no space
Handles edit: removing 'p' flag
Handles edit: adding 'p' flag
Handles edit: result changes (inline ↔ separate transition)
Updates correct messages on edits
Time units scale appropriately
No persistent storage of statistics

Example Scenarios

Scenario 1: Short result, fits inline

User: s/foo/bar/gp
Bot: bar baz

      Performed 1 substitution in 8ms

Scenario 2: Long result, separate message

User: s/complex_pattern/replacement/gp (result is 4064+ chars)
Bot [msg 1]: replacement text (4064 chars)
Bot [msg 2]: Performed 1 substitution in 2.5s

Scenario 3: Edit removes 'p' flag

User edits: s/foo/bar/gp → s/foo/bar/
Bot: Edits message to remove "Performed..." line
[Or deletes separate performance message]

Scenario 4: Edit changes result size (now fits)

Original: Long result, performance separate
Edit: Shorter result
Bot: Deletes performance message
Bot: Edits result to inline performance

Scenario 5: Edit changes result size (no longer fits)

Original: Short result, performance inline
Edit: Longer result
Bot: Edits result to remove performance line
Bot: Sends new performance message

Part of Epic #38
Merged from #49 (enhanced performance flag with substitution count)

## Description Show real-time processing statistics for the current message chain without persisting any data. ## Use Case Users want visibility into how their substitutions are performing without storing data long-term. ## Implementation Plan 1. **In-memory tracking** (per message only) - Track substitution count in current chain - Measure compilation and execution time - Count cache hits/misses - No persistence beyond message processing 2. **Smart placement logic** Calculate dynamically if performance text fits: ``` available_space = 4096 - result_length - performance_length - 1 (for newline) ``` If fits: Append to result message with newline separator ``` Result text Performed X substitutions in Yms ``` If doesn't fit: Send as separate message 3. **Edit handling (complex logic)** When user edits the sed command: a) **Removing 'p' flag:** - Delete the performance message (if separate) - Or edit result to remove performance line b) **Adding 'p' flag:** - Check if fits in result message - If yes: Inline it - If no: Send as separate message c) **Command changed (different result):** - Recalculate space availability - If was separate and now fits: Delete separate message, inline in result - If was inline and now doesn't fit: Remove from result, send as separate - If still doesn't fit: Update both messages - If still fits inline: Update result message only 4. **Time unit logic** - < 1000ms: "Xms" - < 60000ms: "Xs" (e.g., "1.2s") - < 3600000ms: "Xm Ys" - > = 3600000ms: "Xh Ym" 5. **Data to collect** - Number of successful substitutions - Total processing time - Cache hit/miss count (optional) - Time per substitution (optional) 6. **Storage for edits** - Store bot message IDs in memory: - Result message ID - Performance message ID (if separate) - Track if performance is inlined or separate - Clean up after edit window expires (48h) ## In-Memory Tracking (No Persistence) Store performance message metadata in memory only: ```typescript interface PerformanceMessageInfo { chatId: number; targetMessageId: number; // User's sed command message resultMessageId: number; // Bot's reply with substitution result performanceMessageId?: number; // Separate performance message (undefined if inlined) isInlined: boolean; timestamp: number; } // In-memory Map: key = `${chatId}:${targetMessageId}` const performanceMessageTracker = new Map<string, PerformanceMessageInfo>(); // Cleanup old entries periodically (after 48h, matching Telegram edit window) ``` ## Acceptance Criteria - [ ] Calculates available space dynamically (includes newline) - [ ] Inlines performance when space available - [ ] Sends separate message when no space - [ ] Handles edit: removing 'p' flag - [ ] Handles edit: adding 'p' flag - [ ] Handles edit: result changes (inline ↔ separate transition) - [ ] Updates correct messages on edits - [ ] Time units scale appropriately - [ ] No persistent storage of statistics ## Example Scenarios **Scenario 1: Short result, fits inline** ``` User: s/foo/bar/gp Bot: bar baz Performed 1 substitution in 8ms ``` **Scenario 2: Long result, separate message** ``` User: s/complex_pattern/replacement/gp (result is 4064+ chars) Bot [msg 1]: replacement text (4064 chars) Bot [msg 2]: Performed 1 substitution in 2.5s ``` **Scenario 3: Edit removes 'p' flag** ``` User edits: s/foo/bar/gp → s/foo/bar/ Bot: Edits message to remove "Performed..." line [Or deletes separate performance message] ``` **Scenario 4: Edit changes result size (now fits)** ``` Original: Long result, performance separate Edit: Shorter result Bot: Deletes performance message Bot: Edits result to inline performance ``` **Scenario 5: Edit changes result size (no longer fits)** ``` Original: Short result, performance inline Edit: Longer result Bot: Edits result to remove performance line Bot: Sends new performance message ``` ## Related Part of Epic #38 Merged from #49 (enhanced performance flag with substitution count)