Commit 85db1540 authored by jan.koester's avatar jan.koester
Browse files

zlib support

parent b4d53888

CMakeLists.txt

+1 −0
Changes for CMakeLists.txt: 1 added line, 0 removed lines.
Original line number Diff line number Diff line
@@ -24,6 +24,7 @@ find_package(smtpclient REQUIRED) 
find_package(libuuidp REQUIRED)

find_package(tinyxml2 REQUIRED)

find_package(paritypp REQUIRED)

find_package(ZLIB REQUIRED)



if(APPLE OR ${CMAKE_HOST_SYSTEM_NAME} MATCHES "FreeBSD")

    find_package(date REQUIRED)

editor/CMakeLists.txt

+3 −0
Changes for editor/CMakeLists.txt: 3 added lines, 0 removed lines.
Original line number Diff line number Diff line
@@ -25,6 +25,7 @@ if(NOT TARGET blogidev) 
    find_package(smtpclient REQUIRED)

    find_package(Brotli REQUIRED)

    find_package(paritypp REQUIRED)

    find_package(ZLIB REQUIRED)

    add_definitions(-D${CMAKE_HOST_SYSTEM_NAME} -DWIN32_LEAN_AND_MEAN)

endif()
@@ -57,6 +58,7 @@ if(TARGET blogidev) 
        blogidev

        json-c::json-c

        tinyxml2::tinyxml2

        ZLIB::ZLIB

    )

    target_sources(blogi-editor PRIVATE

        ${CMAKE_CURRENT_SOURCE_DIR}/../src/plugin.cpp
@@ -75,6 +77,7 @@ else() 
        uuidp::uuidp

        authdb::client

        Brotli::Brotli

        ZLIB::ZLIB

        dl

    )

endif()

editor/src/htmlimport.cpp

+50 −0
Changes for editor/src/htmlimport.cpp: 50 added lines, 0 removed lines.
Original line number Diff line number Diff line
@@ -41,6 +41,8 @@ 
#include <netinet/in.h>

#include <arpa/inet.h>



#include <zlib.h>



using namespace blogi::htmlimport;



namespace {
@@ -123,6 +125,50 @@ bool isSafeExternalHost(const std::string &host, std::string &error) 
    return safe;

}



// Best-effort gzip decompression for fetchExternalResource: plenty of real-

// world servers/CDNs return a gzip-compressed body regardless of what (or

// whether) the client's Accept-Encoding says -- confirmed against a live

// site that ignored an explicit "Accept-Encoding: identity" -- and

// libhttppp's HttpClient doesn't decompress bodies itself or expose the

// Content-Encoding header, so without this every gzipped page/stylesheet/

// script would be handed to the HTML/CSS parser as opaque binary. Detected

// purely by the gzip magic (1f 8b) for that reason. A no-op (returns true,

// @p data untouched) when the magic isn't present. Bounded by @p maxBytes

// on the DECOMPRESSED size to defend against a decompression-bomb response

// (small on the wire, huge once inflated) -- returns false, leaving @p data

// untouched, on any zlib error or if that cap is hit.

bool gunzipIfNeeded(std::vector<std::uint8_t> &data, size_t maxBytes)

{

    if (data.size() < 2 || data[0] != 0x1f || data[1] != 0x8b) return true;



    z_stream strm{};

    // 16 + MAX_WBITS: gzip-header format specifically (plain MAX_WBITS

    // expects a zlib-wrapped stream instead, which the gzip magic above

    // rules out).

    if (inflateInit2(&strm, 16 + MAX_WBITS) != Z_OK) return false;



    std::vector<std::uint8_t> out;

    std::vector<std::uint8_t> chunk(65536);

    strm.next_in = data.data();

    strm.avail_in = static_cast<uInt>(data.size());



    int rc;

    do {

        strm.next_out = chunk.data();

        strm.avail_out = static_cast<uInt>(chunk.size());

        rc = inflate(&strm, Z_NO_FLUSH);

        if (rc != Z_OK && rc != Z_STREAM_END) { inflateEnd(&strm); return false; }



        size_t produced = chunk.size() - strm.avail_out;

        if (out.size() + produced > maxBytes) { inflateEnd(&strm); return false; }

        out.insert(out.end(), chunk.begin(), chunk.begin() + produced);

    } while (rc != Z_STREAM_END);



    inflateEnd(&strm);

    data.swap(out);

    return true;

}



} // namespace



bool blogi::htmlimport::fetchExternalResource(const std::string &url,
@@ -184,6 +230,10 @@ bool blogi::htmlimport::fetchExternalResource(const std::string &url, 
        }



        out.assign(resp.begin(), resp.end());

        if (!gunzipIfNeeded(out, maxBytes)) {

            error = "failed to decompress gzip response";

            return false;

        }

        return true;

    } catch (const std::exception &e) {

        error = e.what();

editor/src/htmlimport.h

+4 −0
Changes for editor/src/htmlimport.h: 4 added lines, 0 removed lines.
Original line number Diff line number Diff line
@@ -50,6 +50,10 @@ namespace htmlimport { 
     *  streaming more than @p maxBytes before the check runs).

     *  @p contentTypePrefix, if non-empty, rejects responses whose

     *  Content-Type doesn't start with it (a missing header is tolerated).

     *  A gzip-magic-prefixed body is transparently decompressed (many real

     *  servers/CDNs send one regardless of Accept-Encoding, and the result

     *  is re-checked against @p maxBytes to bound decompression-bomb

     *  responses) -- see gunzipIfNeeded in htmlimport.cpp.

     *  Returns false and sets @p error on any failure; never throws. */

    bool fetchExternalResource(const std::string &url,

                               size_t maxBytes,