DataZooDE
diff --git a/‎CMakeLists.txt‎
Lines changed: 31 additions & 0 deletions b/‎CMakeLists.txt‎
Lines changed: 31 additions & 0 deletions
diff --git a/‎src/bundle_locator.cpp‎
Lines changed: 113 additions & 62 deletions b/‎src/bundle_locator.cpp‎
Lines changed: 113 additions & 62 deletions
diff --git a/‎src/include/bundle_locator.hpp‎
Lines changed: 14 additions & 2 deletions b/‎src/include/bundle_locator.hpp‎
Lines changed: 14 additions & 2 deletions
diff --git a/‎src/include/macho_bundle.hpp‎
Lines changed: 72 additions & 0 deletions b/‎src/include/macho_bundle.hpp‎
Lines changed: 72 additions & 0 deletions
@@ -250,6 +250,7 @@ add_library(flapi-lib STATIC
     src/endpoint_config_parser.cpp
     src/extended_yaml_parser.cpp
     src/heartbeat_worker.cpp
+    src/macho_bundle.cpp
     src/open_api_doc_generator.cpp
     src/pack.cpp
     src/password_hasher.cpp
@@ -344,6 +345,36 @@ target_link_libraries(flapi PRIVATE flapi-lib)
 target_compile_definitions(flapi PRIVATE FLAPI_VERSION="${CMAKE_PROJECT_VERSION}")
 set_target_properties(flapi PROPERTIES ENABLE_EXPORTS TRUE)
 
+# macOS reserved-segment for self-packaging (#48). Allocates a
+# placeholder __FLAPI/__bundle section at link time that `flapi pack`
+# overwrites with the user's config tree, then re-signs. Required for
+# notarised distribution -- appending after __LINKEDIT invalidates
+# the signature.
+if(APPLE)
+    set(FLAPI_RESERVED_BUNDLE_MIB "16" CACHE STRING
+        "Size in MiB of the reserved __FLAPI/__bundle Mach-O segment.")
+    set(FLAPI_BUNDLE_PLACEHOLDER
+        "${CMAKE_BINARY_DIR}/_flapi_bundle_placeholder.bin")
+    # Use dd directly via execve (no shell). The previous `bash -c
+    # "... >/dev/null 2>&1"` wrapping broke under ninja-on-macos because
+    # the redirect operators were parsed by the outer shell rather than
+    # the bash -c subshell, producing "/dev/null 2: Permission denied".
+    # dd prints a 2-3 line summary on stderr; that's fine in CI logs.
+    add_custom_command(
+        OUTPUT "${FLAPI_BUNDLE_PLACEHOLDER}"
+        COMMAND dd
+                if=/dev/zero
+                "of=${FLAPI_BUNDLE_PLACEHOLDER}"
+                bs=1m
+                count=${FLAPI_RESERVED_BUNDLE_MIB}
+        COMMENT "flapi: generating ${FLAPI_RESERVED_BUNDLE_MIB}-MiB __FLAPI/__bundle placeholder")
+    add_custom_target(flapi_bundle_placeholder
+        DEPENDS "${FLAPI_BUNDLE_PLACEHOLDER}")
+    target_link_options(flapi PRIVATE
+        "-Wl,-sectcreate,__FLAPI,__bundle,${FLAPI_BUNDLE_PLACEHOLDER}")
+    add_dependencies(flapi flapi_bundle_placeholder)
+endif()
+
 # Add Windows-specific libraries
 if(WIN32)
     target_link_libraries(flapi PRIVATE Dbghelp.lib)
 
@@ -1,4 +1,5 @@
 #include "bundle_locator.hpp"
+#include "macho_bundle.hpp"
 #include "selfpath.hpp"
 
 #include <algorithm>
@@ -14,8 +15,8 @@ namespace {
 constexpr std::size_t kEocdRecordSize = 22;
 constexpr std::size_t kMaxCommentLen = 0xffffu;
 
-// We accept padding well in excess of the 10 KiB spike default; the
-// total tail buffer is EOCD + max comment + 64 KiB pad budget.
+// Total tail buffer = EOCD + max comment + 64 KiB pad budget. Generous
+// vs. the spike's 10 KiB libarchive tar-block default; cheap to read.
 constexpr std::size_t kPaddingBudget = 65536;
 constexpr std::size_t kScanBudget = kEocdRecordSize + kMaxCommentLen + kPaddingBudget;
 
@@ -32,75 +33,50 @@ std::uint32_t ReadU32(const std::uint8_t* p) {
          | (static_cast<std::uint32_t>(p[3]) << 24);
 }
 
-}  // namespace
-
-std::optional<BundleLocation> LocateBundle(const std::filesystem::path& path) {
-    std::error_code ec;
-    const auto file_size = std::filesystem::file_size(path, ec);
-    if (ec || file_size < kEocdRecordSize) {
-        return std::nullopt;
-    }
-
-    std::ifstream in(path, std::ios::binary);
-    if (!in.is_open()) {
-        return std::nullopt;
-    }
-
-    const std::size_t tail_bytes =
-        static_cast<std::size_t>(std::min<std::uint64_t>(file_size, kScanBudget));
-    const std::uint64_t tail_start = file_size - tail_bytes;
-
-    std::vector<std::uint8_t> tail(tail_bytes);
-    in.seekg(static_cast<std::streamoff>(tail_start), std::ios::beg);
-    in.read(reinterpret_cast<char*>(tail.data()),
-            static_cast<std::streamsize>(tail_bytes));
-    if (!in) {
-        return std::nullopt;
-    }
-    if (tail.size() < kEocdRecordSize) {
+// Scan a buffer for the most-recent valid EOCD record. `buf_start_in_file`
+// is the absolute file offset of buf[0]; used to translate the result.
+// `logical_eof_in_buf` is the buf-relative position to treat as EOF for
+// padding-tolerance purposes (== buf.size() in the common case).
+std::optional<BundleLocation> ScanBufferForEocd(
+        const std::vector<std::uint8_t>& buf,
+        std::uint64_t buf_start_in_file,
+        std::size_t logical_eof_in_buf) {
+    if (logical_eof_in_buf > buf.size() || logical_eof_in_buf < kEocdRecordSize) {
         return std::nullopt;
     }
-
-    // Reverse-scan from the latest valid signature position. The latest
-    // (largest-offset) EOCD wins, since any earlier signature byte
-    // sequence in random leading data is a false positive.
-    const std::size_t max_start = tail.size() - kEocdRecordSize;
+    const std::size_t max_start = logical_eof_in_buf - kEocdRecordSize;
     for (std::size_t i = max_start + 1; i-- > 0; ) {
-        if (tail[i]     != 0x50 ||
-            tail[i + 1] != 0x4b ||
-            tail[i + 2] != 0x05 ||
-            tail[i + 3] != 0x06) {
+        if (buf[i]     != 0x50 ||
+            buf[i + 1] != 0x4b ||
+            buf[i + 2] != 0x05 ||
+            buf[i + 3] != 0x06) {
             continue;
         }
+        const std::uint8_t* p = buf.data() + i;
+        const std::uint16_t this_disk     = ReadU16(p + 4);
+        const std::uint16_t cd_start_disk = ReadU16(p + 6);
+        const std::uint16_t entries_this  = ReadU16(p + 8);
+        const std::uint16_t entries_total = ReadU16(p + 10);
+        const std::uint32_t cd_size       = ReadU32(p + 12);
+        const std::uint32_t cd_offset_arch = ReadU32(p + 16);
+        const std::uint16_t comment_len   = ReadU16(p + 20);
 
-        const std::uint8_t* p = tail.data() + i;
-        const std::uint16_t this_disk        = ReadU16(p + 4);
-        const std::uint16_t cd_start_disk    = ReadU16(p + 6);
-        const std::uint16_t entries_this     = ReadU16(p + 8);
-        const std::uint16_t entries_total    = ReadU16(p + 10);
-        const std::uint32_t cd_size          = ReadU32(p + 12);
-        const std::uint32_t cd_offset_arch   = ReadU32(p + 16);
-        const std::uint16_t comment_len      = ReadU16(p + 20);
-
-        // Multi-disk archives are not supported.
         if (this_disk != 0 || cd_start_disk != 0) {
             continue;
         }
         if (entries_this != entries_total) {
             continue;
         }
 
-        // The comment must fit in the tail.
         const std::size_t comment_end = i + kEocdRecordSize + comment_len;
-        if (comment_end > tail.size()) {
+        if (comment_end > logical_eof_in_buf) {
             continue;
         }
 
-        // Anything after the comment up to file-EOF must be zero
-        // padding -- the libarchive tar-block rounding tolerance.
+        // Anything after the comment up to logical EOF must be zero.
         bool padding_ok = true;
-        for (std::size_t j = comment_end; j < tail.size(); ++j) {
-            if (tail[j] != 0) {
+        for (std::size_t j = comment_end; j < logical_eof_in_buf; ++j) {
+            if (buf[j] != 0) {
                 padding_ok = false;
                 break;
             }
@@ -109,36 +85,111 @@ std::optional<BundleLocation> LocateBundle(const std::filesystem::path& path) {
             continue;
         }
 
-        const std::uint64_t eocd_file_offset = tail_start + i;
-
-        // The central directory sits immediately before the EOCD.
+        const std::uint64_t eocd_file_offset = buf_start_in_file + i;
         if (cd_size > eocd_file_offset) {
             continue;
         }
         const std::uint64_t cd_file_offset = eocd_file_offset - cd_size;
-
         if (cd_offset_arch > cd_file_offset) {
             continue;
         }
         const std::uint64_t bundle_start = cd_file_offset - cd_offset_arch;
-
-        const std::uint64_t bundle_end = eocd_file_offset + kEocdRecordSize + comment_len;
+        const std::uint64_t bundle_end =
+            eocd_file_offset + kEocdRecordSize + comment_len;
         if (bundle_end < bundle_start) {
-            continue;  // overflow paranoia
+            continue;
         }
 
         BundleLocation loc;
         loc.offset = bundle_start;
         loc.size = bundle_end - bundle_start;
         return loc;
     }
-
     return std::nullopt;
 }
 
+}  // namespace
+
+std::optional<BundleLocation> LocateBundle(const std::filesystem::path& path) {
+    std::error_code ec;
+    const auto file_size = std::filesystem::file_size(path, ec);
+    if (ec || file_size < kEocdRecordSize) {
+        return std::nullopt;
+    }
+
+    std::ifstream in(path, std::ios::binary);
+    if (!in.is_open()) {
+        return std::nullopt;
+    }
+
+    const std::size_t tail_bytes =
+        static_cast<std::size_t>(std::min<std::uint64_t>(file_size, kScanBudget));
+    const std::uint64_t tail_start = file_size - tail_bytes;
+
+    std::vector<std::uint8_t> tail(tail_bytes);
+    in.seekg(static_cast<std::streamoff>(tail_start), std::ios::beg);
+    in.read(reinterpret_cast<char*>(tail.data()),
+            static_cast<std::streamsize>(tail_bytes));
+    if (!in) {
+        return std::nullopt;
+    }
+    return ScanBufferForEocd(tail, tail_start, tail.size());
+}
+
+std::optional<BundleLocation> LocateBundleInRange(
+        const std::filesystem::path& path,
+        std::uint64_t range_offset,
+        std::uint64_t range_size) {
+    if (range_size < kEocdRecordSize) {
+        return std::nullopt;
+    }
+    std::error_code ec;
+    const auto file_size = std::filesystem::file_size(path, ec);
+    if (ec || range_offset + range_size > file_size) {
+        return std::nullopt;
+    }
+    // Cap range_size at something sensible -- a runaway section_size
+    // value from a malformed Mach-O could otherwise allocate gigabytes.
+    constexpr std::uint64_t kMaxRangeBytes = 64ull * 1024ull * 1024ull;
+    const std::size_t to_read =
+        static_cast<std::size_t>(std::min<std::uint64_t>(range_size, kMaxRangeBytes));
+
+    std::ifstream in(path, std::ios::binary);
+    if (!in.is_open()) {
+        return std::nullopt;
+    }
+    std::vector<std::uint8_t> buf(to_read);
+    in.seekg(static_cast<std::streamoff>(range_offset), std::ios::beg);
+    in.read(reinterpret_cast<char*>(buf.data()),
+            static_cast<std::streamsize>(to_read));
+    if (!in) {
+        return std::nullopt;
+    }
+    return ScanBufferForEocd(buf, range_offset, buf.size());
+}
+
 std::optional<BundleLocation> LocateBundleInSelf() {
+    std::filesystem::path self_path;
+    try {
+        self_path = GetSelfPath();
+    } catch (...) {
+        return std::nullopt;
+    }
+
+    // Prefer the reserved Mach-O section if present (#48). Linux/Windows
+    // binaries lack the section, so this returns nullopt and we fall
+    // through to the EOF-tail scan unchanged.
+    if (auto sect = LocateFlapiSection(self_path); sect.has_value()) {
+        if (auto loc = LocateBundleInRange(self_path, sect->file_offset, sect->size);
+            loc.has_value()) {
+            return loc;
+        }
+        // Section is present but empty / unpopulated (e.g., un-packed
+        // build): fall through to the EOF-tail scan.
+    }
+
     try {
-        return LocateBundle(GetSelfPath());
+        return LocateBundle(self_path);
     } catch (...) {
         return std::nullopt;
     }
 
@@ -27,8 +27,20 @@ struct BundleLocation {
 // rounding pushing the EOCD off file-EOF.
 std::optional<BundleLocation> LocateBundle(const std::filesystem::path& path);
 
-// Convenience: scan the currently running executable. Returns nullopt
-// if either the self-path lookup or the EOCD scan fails.
+// Scans a specific byte range of `path` for a ZIP EOCD. Used by the
+// macOS section-mode locator (#48) where the bundle lives inside a
+// reserved Mach-O segment, not at file EOF. The returned
+// BundleLocation.offset is the absolute file offset; padding-tolerance
+// runs against `range_size` (treat range end as logical EOF).
+std::optional<BundleLocation> LocateBundleInRange(
+    const std::filesystem::path& path,
+    std::uint64_t range_offset,
+    std::uint64_t range_size);
+
+// Convenience: scan the currently running executable. On macOS, first
+// looks at the reserved __FLAPI/__bundle Mach-O section; if absent or
+// unpopulated, falls back to a reverse-EOF scan. Returns nullopt if
+// either fails.
 std::optional<BundleLocation> LocateBundleInSelf();
 
 }  // namespace flapi
@@ -0,0 +1,72 @@
+#pragma once
+
+#include <cstdint>
+#include <filesystem>
+#include <optional>
+#include <string>
+#include <vector>
+
+namespace flapi {
+
+// The reserved Mach-O segment+section we use to host the bundled
+// ZIP on macOS. Allocated at link time with a placeholder of
+// FLAPI_RESERVED_BUNDLE_MIB MiB and overwritten by `flapi pack`
+// at packaging time so the binary signature stays valid after a
+// re-`codesign`.
+constexpr const char* kFlapiSegName = "__FLAPI";
+constexpr const char* kFlapiSectName = "__bundle";
+
+struct MachOSection {
+    // File offset of the section's first byte, suitable for seek().
+    std::uint64_t file_offset = 0;
+    // Allocated section size in bytes (== reserved capacity).
+    std::uint64_t size = 0;
+};
+
+// Returns true if the bytes at `magic_bytes` (read from the head of
+// a file) are a Mach-O magic value we recognise. Cheap pre-check.
+bool IsMachOMagic(const std::uint8_t magic_bytes[4]);
+
+// Locate the __FLAPI/__bundle section in a Mach-O file on disk.
+// Returns nullopt if:
+//   - the file isn't a thin (non-fat) Mach-O,
+//   - the file is malformed,
+//   - the section doesn't exist (e.g., on a non-macOS build).
+//
+// Fat / universal binaries are currently not supported -- a follow-up
+// can iterate slices. macOS releases produced by this repo are thin
+// per-architecture, so the gap is acceptable for now.
+std::optional<MachOSection> LocateFlapiSection(const std::filesystem::path& path);
+
+// Overload that scans a buffer instead of opening a file. Used by
+// unit tests against synthetic Mach-O fixtures.
+std::optional<MachOSection> LocateFlapiSectionInBuffer(
+    const std::vector<std::uint8_t>& buffer);
+
+// Overwrite the reserved section at `binary` with `payload`. The
+// trailing capacity beyond payload.size() is zero-padded so that the
+// EOCD of the embedded ZIP still reverse-scans cleanly from segment
+// EOF. Throws ArchiveIOError when the payload is bigger than the
+// reserved capacity (see PackOptions::host_binary_override for the
+// override knob).
+//
+// Caller is responsible for re-signing the binary afterwards
+// (CodesignBinary), since the signature covers the section bytes.
+void OverwriteFlapiSection(const std::filesystem::path& binary,
+                           const MachOSection& section,
+                           const std::vector<std::uint8_t>& payload);
+
+// Result of a codesign attempt.
+struct CodesignResult {
+    int exit_code = 0;
+    std::string identity;     // "-" for ad-hoc; otherwise the CODESIGN_IDENTITY value
+    std::string stderr_tail;  // last few KB of codesign stderr, for diagnostics
+};
+
+// Invoke `codesign --force --sign <identity> <binary>` where identity
+// is taken from the CODESIGN_IDENTITY env var, defaulting to "-"
+// (ad-hoc). On non-Darwin builds this is a no-op that returns
+// exit_code = 0 -- the caller doesn't need a platform guard.
+CodesignResult CodesignBinary(const std::filesystem::path& binary);
+
+}  // namespace flapi