Moves images to improved documentation structure

Moves various image files demonstrating libomath’s usage and featuring artwork from showcases to the improved `docs/images` structure to reorganize the project's documentation in a logical and maintainable fashion.

This change ensures consistency and simplifies updating documentation assets.
This commit is contained in:
2025-10-31 16:47:28 +03:00
parent 9212b1f51f
commit d12a2611b8
17 changed files with 304 additions and 68 deletions

View File

@@ -45,7 +45,7 @@ The pattern string is parsed into a sequence of byte **tokens**:
* **Wildcard byte**: `?` or `??` matches **any single byte**.
* **Separators**: any ASCII whitespace (space, tab, newline) is **ignored** and may be used to group tokens.
> ✔️ Valid: `"48 8B ?? 05 00"`, `"90 90 90"`, `"??"`, `"00??FF"` (no spaces)
> ✔️ Valid: `"48 8B ?? 05 00"`, `"90 90 90"`, `"??"`
> ❌ Invalid: odd number of hex digits in a token, non-hex characters (besides `?` and whitespace)
If the string cannot be parsed into a clean sequence of tokens, `parse_pattern()` returns `std::unexpected(PatternScanError::INVALID_PATTERN_STRING)`, and the public scan function returns **end**.

View File

@@ -0,0 +1,155 @@
# `omath::PePatternScanner` — Scan PE images for byte patterns
> Header: your projects `pe_pattern_scanner.hpp`
> Namespace: `omath`
> Platform: **Windows / PE (Portable Executable) images**
> Depends on: `<filesystem>`, `<optional>`, `<cstdint>`
> Companion: works well with `omath::PatternScanner` (same pattern grammar)
`PePatternScanner` searches **Portable Executable (PE)** binaries for a hex pattern (with wildcards). You can scan:
* a **loaded module** in the current process, or
* a **PE file on disk** (by section name; defaults to **`.text`**).
---
## Pattern string grammar (same as `PatternScanner`)
* **Hex byte**: two hex digits → one byte (`90`, `4F`, `00`, `ff`).
* **Wildcard byte**: `?` or `??` matches **any byte**.
* **Whitespace**: ignored (use to group tokens).
✔️ `"48 8B ?? ?? 89"`, `"55 8B EC"`, `"??"`
❌ odd digit counts, non-hex characters (besides `?` and whitespace)
---
## API
```cpp
namespace omath {
struct PeSectionScanResult {
std::uint64_t virtual_base_addr; // RVA base of the scanned section (ImageBase + SectionRVA)
std::uint64_t raw_base_addr; // file offset (start of section in the file)
std::ptrdiff_t target_offset; // offset from section base to the first matched byte
};
class PePatternScanner final {
public:
// Scan a module already loaded in *this* process.
// module_base_address: HMODULE / ImageBase (e.g., from GetModuleHandle)
// Returns absolute address (process VA) of the first match, or nullopt.
static std::optional<std::uintptr_t>
scan_for_pattern_in_loaded_module(const void* module_base_address,
const std::string_view& pattern);
// Scan a PE file on disk, by section name (default ".text").
// Returns section bases (virtual + raw) and match offset within the section, or nullopt.
static std::optional<PeSectionScanResult>
scan_for_pattern_in_file(const std::filesystem::path& path_to_file,
const std::string_view& pattern,
const std::string_view& target_section_name = ".text");
};
} // namespace omath
```
---
## Return values
* **Loaded module**: `std::optional<std::uintptr_t>`
* `value()` = **process virtual address** (ImageBase + SectionRVA + match offset).
* `nullopt` = no match or parse/PE error.
* **File scan**: `std::optional<PeSectionScanResult>`
* `virtual_base_addr` = **ImageBase + SectionRVA** of the scanned section (as if mapped).
* `raw_base_addr` = **file offset** of section start.
* `target_offset` = offset from the section base to the **first matched byte**.
* To get addresses:
* **Virtual (RVA)** of hit = `virtual_base_addr + target_offset`
* **Raw file offset** of hit = `raw_base_addr + target_offset`
---
## Usage examples
### Scan a loaded module (current process)
```cpp
#include <windows.h>
#include "pe_pattern_scanner.hpp"
using omath::PePatternScanner;
auto hMod = ::GetModuleHandleW(L"kernel32.dll");
if (hMod) {
auto addr = PePatternScanner::scan_for_pattern_in_loaded_module(
hMod, "48 8B ?? ?? 89" // hex + wildcards
);
if (addr) {
// Use the absolute process VA:
std::uintptr_t hit_va = *addr;
// ...
}
}
```
### Scan a PE file on disk (default section “.text”)
```cpp
#include "pe_pattern_scanner.hpp"
using omath::PePatternScanner;
auto res = PePatternScanner::scan_for_pattern_in_file(
R"(C:\Windows\System32\kernel32.dll)", "55 8B EC"
);
if (res) {
auto rva_hit = res->virtual_base_addr + res->target_offset;
auto raw_hit = res->raw_base_addr + res->target_offset;
// rva_hit: where it would be in memory; raw_hit: file offset
}
```
### Scan another section (e.g., “.rdata”)
```cpp
auto res = PePatternScanner::scan_for_pattern_in_file(
"foo.dll", "48 8D 0D ?? ?? ?? ??", ".rdata"
);
```
---
## Notes & edge cases
* **PE only**: these functions assume a valid **PE/COFF** layout. Non-PE files or corrupted headers yield `nullopt`.
* **Section name**: `scan_for_pattern_in_file` defaults to **`.text`**; pass a different name to target other sections.
* **Alignment & RVAs**: `virtual_base_addr` is computed from section headers (RVA aligned per section alignment). The returned “virtual” base is suitable for RVA math; the **process VA** returned by the “loaded module” API already includes the image base.
* **Architecture**: works for 32-bit and 64-bit PEs; `std::uintptr_t` size matches the build architecture.
* **Performance**: Pattern matching is **O(N × M)** (sliding window with wildcards). For large images, prefer scanning only necessary sections.
* **Wildcards**: Each `?` matches **one byte** (no nibble masks). If you need `A?`-style nibble wildcards, extend the parser (see `PatternScanner`).
* **Safety**: For loaded modules, you must have access to the memory; scanning read-only sections is fine, but never write. For file scans, ensure the file path is accessible.
---
## Troubleshooting
* **`nullopt`**: Verify the pattern (valid tokens), correct **section**, and that youre scanning the intended module/file (check bitness and version).
* **“Loaded module” address math**: If you need an **offset from the module base**, compute `offset = hit_va - reinterpret_cast<std::uintptr_t>(module_base_address)`.
* **Multiple matches**: Only the **first** match is returned. If you need all matches, extend the implementation to continue scanning from `target_offset + 1`.
---
## See also
* `omath::PatternScanner` — raw buffer/iterator scanning with the same pattern grammar.
* `omath::Triangle`, `omath::Vector3` — math types used elsewhere in the library.
---
*Last updated: 31 Oct 2025*