1 files changed, 123 insertions, 0 deletions
diff --git a/include/wireframe/pcapng/reader.hpp b/include/wireframe/pcapng/reader.hpp
new file mode 100644
index 0000000..d01b431
--- /dev/null
+++ b/include/wireframe/pcapng/reader.hpp
@@ -0,0 +1,123 @@
+#pragma once
+
+#include <array>
+#include <cstdint>
+#include <cstdio>
+#include <optional>
+#include <span>
+#include <vector>
+
+// Minimal pcapng reader, paired with writer.hpp: reads Enhanced Packet
+// Blocks sequentially, skipping the Section Header Block, Interface
+// Description Block, and any other block type transparently.
+//
+// Assumes little-endian block encoding (checked against the Section
+// Header Block's byte-order magic, not just assumed) since that's what
+// writer.hpp emits and what pcapng writers on this class of hardware
+// (tcpdump, dumpcap) produce. A big-endian file is out of scope - this
+// pairs with our own writer, not general pcapng interop.
+namespace wireframe::pcapng {
+
+struct PacketRecord {
+    std::uint32_t interface_id;
+    std::uint64_t timestamp_us;
+    std::uint32_t original_len;
+    std::vector<unsigned char> data;
+};
+
+class Reader {
+public:
+    explicit Reader(std::FILE* file) : file_(file) {}
+
+    // Returns the next packet, or nullopt once the file is exhausted or
+    // a malformed/unsupported block is hit - treated as end of stream
+    // rather than a hard error, to keep this reader small.
+    std::optional<PacketRecord> next_packet() {
+        for (;;) {
+            std::array<std::uint8_t, 4> field{};
+            if (std::fread(field.data(), 1, 4, file_) != 4) return std::nullopt;
+            std::uint32_t type = get_u32(field);
+
+            if (std::fread(field.data(), 1, 4, file_) != 4) return std::nullopt;
+            std::uint32_t total_len = get_u32(field);
+            if (total_len < 12) return std::nullopt;
+
+            std::size_t body_len = total_len - 12;
+            // total_len is an untrusted 32-bit value straight from the
+            // file; without a cap, a corrupted/hostile file can claim
+            // a multi-gigabyte block and OOM the process on the
+            // allocation below before a single byte is even read to
+            // check whether the file actually contains that much data
+            // (found by fuzzing fuzz_pcapng_reader.cpp - real crash,
+            // not theoretical). Bounded well above any block our own
+            // writer produces (packets capped at a 65535 snaplen; this
+            // reader is explicitly scoped to pair with that writer,
+            // not arbitrary pcapng interop).
+            if (body_len > kMaxBlockBodyLen) return std::nullopt;
+            std::vector<std::uint8_t> body(body_len);
+            if (body_len > 0 && std::fread(body.data(), 1, body_len, file_) != body_len) {
+                return std::nullopt;
+            }
+
+            if (std::fread(field.data(), 1, 4, file_) != 4) return std::nullopt;
+            if (get_u32(field) != total_len) return std::nullopt;  // corrupt trailer
+
+            if (type == kBlockTypeShb) {
+                if (body_len < 4 || get_u32({body.data(), 4}) != kByteOrderMagic) {
+                    return std::nullopt;  // not little-endian, or malformed
+                }
+                continue;
+            }
+            if (type == kBlockTypeIdb) {
+                // LinkType is the first 2 bytes of the IDB body (see
+                // writer.hpp's write_interface_description). Only the
+                // first IDB is captured - correct for a file our own
+                // writer produced, which only ever writes one
+                // interface, matching this reader's documented scope.
+                if (!link_type_ && body_len >= 2) {
+                    link_type_ = static_cast<std::uint16_t>(body[0] | (body[1] << 8));
+                }
+                continue;
+            }
+            if (type != kBlockTypeEpb) continue;  // anything else: skip
+
+            if (body_len < 20) return std::nullopt;
+
+            PacketRecord record;
+            record.interface_id = get_u32({body.data() + 0, 4});
+            std::uint32_t ts_high = get_u32({body.data() + 4, 4});
+            std::uint32_t ts_low = get_u32({body.data() + 8, 4});
+            record.timestamp_us = (static_cast<std::uint64_t>(ts_high) << 32) | ts_low;
+            std::uint32_t caplen = get_u32({body.data() + 12, 4});
+            record.original_len = get_u32({body.data() + 16, 4});
+
+            if (body_len < 20 + caplen) return std::nullopt;
+            record.data.assign(body.begin() + 20, body.begin() + 20 + caplen);
+            return record;
+        }
+    }
+
+    // The interface's link type, learned from the Interface
+    // Description Block once next_packet() has read past it (which
+    // happens before it ever returns the first EPB, so this is
+    // populated by the time the first successful next_packet() call
+    // returns). nullopt if no IDB has been seen yet.
+    std::optional<std::uint16_t> link_type() const { return link_type_; }
+
+private:
+    static std::uint32_t get_u32(std::span<const std::uint8_t> b) {
+        return static_cast<std::uint32_t>(b[0]) | (static_cast<std::uint32_t>(b[1]) << 8) |
+               (static_cast<std::uint32_t>(b[2]) << 16) | (static_cast<std::uint32_t>(b[3]) << 24);
+    }
+
+    static constexpr std::uint32_t kBlockTypeShb = 0x0A0D0D0A;
+    static constexpr std::uint32_t kBlockTypeIdb = 0x00000001;
+    static constexpr std::uint32_t kBlockTypeEpb = 0x00000006;
+    static constexpr std::uint32_t kByteOrderMagic = 0x1A2B3C4D;
+    static constexpr std::size_t kMaxBlockBodyLen = 1 << 20;  // 1 MiB
+
+    std::FILE* file_;
+    std::optional<std::uint16_t> link_type_;
+};
+
+}  // namespace wireframe::pcapng