…tures

- Replace custom ip_address concept with net::ip_address in template
  declarations across TCP/UDP proxy and filter classes
- Fix bind() calls to use sizeof(sockaddr_in) / sizeof(sockaddr_in6)
  instead of sizeof(sockaddr), preventing WSAEFAULT on IPv6 sockets
- Update SOCKS5 CONNECT request to set ATYP = 1 (IPv4) or 4 (IPv6)
  according to RFC 1928
- Add explicit enum underlying types (uint8_t) for action_t and
  direction_t
- Improve SOCKS5 negotiation logging and error handling

docs: add firewall troubleshooting tip to README (#83)

- Documented potential Windows Firewall interference with ProxiFyre
- Added instructions to create an inbound firewall rule for ProxiFyre.exe
- Improves onboarding for newcomers who may face connectivity issues
- Closes #83

Revised NukaColaM's code (#46).

Fixes for PR #84 from https://github.com/PoneyClairDeLune

- Fix case-insensitive matching in socks_local_router exclusion logic
  * Update exclusion checks to follow same path/name selection logic as pattern matching
  * Remove redundant to_upper() calls since network_process fields are already uppercase
  * Ensure consistent behavior between exclusion and association logic

- Add comprehensive documentation for match_app_name function
  * Document two-stage matching process (exclusion then pattern matching)
  * Clarify intelligent field selection based on path separators
  * Add detailed parameter descriptions with examples

- Update configuration documentation and examples
  * Add missing excludedList parameter documentation in ProxiFyreSettings constructor
  * Include "excludes" array in JSON configuration example with various exclusion patterns
  * Demonstrate support for process names, full paths, and partial path matching

- Enhance code consistency across C++ and C# components
  * Align exclusion logic implementation with documented behavior
  * Improve documentation completeness for better maintainability

These changes ensure robust process filtering with consistent case-insensitive
matching and comprehensive documentation for configuration and usage.

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>

- Implement full result caching for match_app_name function
  * Cache complete matching results (not just exclusions) using process ID + app pattern key
  * Add composite cache key with custom hash function for optimal performance
  * Implement time-based expiration (30 seconds TTL) and size-limited cache (2000 entries)
  * Add automatic cleanup for expired and oldest entries to prevent memory bloat

- Enhance cache management and invalidation
  * Clear cache when exclusion list or proxy mappings change via clear_match_cache()
  * Update associate_process_name_to_proxy and exclude_process_name to invalidate cache
  * Add comprehensive cache cleanup strategy with timestamp-based sorting

- Optimize packet processing hot path performance
  * Eliminate repeated string operations for same process/app combinations
  * Convert O(n) linear searches to O(1) hash map lookups for cached results
  * Maintain thread safety with mutex protection for concurrent access

- Update documentation to reflect caching behavior
  * Document comprehensive result caching in match_app_name function
  * Add detailed comments for cache data structures and management
  * Explain performance benefits and cache invalidation strategy

This optimization significantly improves performance for high-traffic scenarios where
the same processes generate many packets, eliminating redundant exclusion checks and
pattern matching operations through intelligent caching.

- Remove mutex acquisition from cleanup_match_cache() function
- Function now assumes caller holds match_cache_mutex_ (which is correct)
- Eliminates unsafe lock unlock/relock pattern in match_app_name()
- Simplifies code flow with single lock ownership per operation
- Called only from match_app_name() where lock is already held

This change improves thread safety by removing complex mutex manipulation
and clarifies lock ownership responsibilities. The cache cleanup is now
a simple utility function that operates under the caller's lock protection.

- Make excludedList parameter optional with default null value in ProxiFyreSettings constructor
- Add NullValueHandling.Ignore to JsonProperty attribute for ExcludedList
- Ensure ExcludedList is always initialized as empty list when field is missing from JSON
- Remove duplicate class summary comment for ProxiFyreSettings

This change prevents deserialization errors when app-config.json doesn't contain
the "excludes" field, making the exclusion configuration truly optional while
maintaining backward compatibility with existing configurations.

…reSettings class

fix: enhance process exclusion logic and improve configuration documentation (PR #84)