Merge RFC4 (#1865)

This commit is the result of the squashing of rfc4_dev branch in a single
commit. It implements mostly RFC 4 related work.

* Grid handling:
  - remove obsolete and presumably unfinished implementation of grid catalog functionality
  - all grid functionality is in grids.cpp/.hpp
  - vertical and horizontal grid shift: rework to no longer load whole grid into memory
  - remove hgrids and vgrids member from PJ structure, and store them in hgridshift/vgridshift/deformation structures
  - build systems: add optional libtiff dependency. Must be explicitly disabled if not desired
  - add support for horizontal and vertical grids in GeoTIFF, if libtiff is available
  - add GenericShiftGridSet and GenericShiftGrid classes, relying on TIFF grids, that can be used for generic purpose grid-based adjustment
  - add a +proj=xyzgridshift method to perform geocentric translation by grid. Used for French NTF to RGF93 transformation using gr3df97a.tif grid
  - deformation: add support for +grids= for GeoTIFF grids
  - horizontal grid shift: fix failures on points slightly outside a subgrid (fixes #209)

* File management:
  - add a filemanager.cpp/.hpp to deal with file related work
  - test for legacy proj_api.h fileapi
  - proj.h: add proj_context_set_fileapi() and proj_context_set_sqlite3_vfs_name() (fixes #866)
  - add capability to read resource files from the user writable directory

* Network access:
  - build systems: add optional curl dependency
  - add a curl-based default implementation for network related functionality
  - proj.h: add C API to control network functionality, and optionaly provide network callbacks
  - add data/proj.ini with default settings
  - add a SQLite3 local cache of downloaded chunks
  - add proj_is_download_needed() and proj_download_file()

* Use Win32 Unicode APIs and expect all strings to be UTF-8 (fixes #1765)
  For backward compatibility, if PROJ_LIB content is found to be not UTF-8 or
  pointing to a non existing directory, then an attempt at interpretating it
  in the ANSI page encoding is done.
  proj_context_set_search_paths() now assumes strings to be in UTF-8, and
  functions returning paths will also return values in UTF-8.

1 files changed, 2544 insertions, 0 deletions

diff --git a/src/networkfilemanager.cpp b/src/networkfilemanager.cpp
new file mode 100644
index 00000000..64969d0f
--- /dev/null
+++ b/src/networkfilemanager.cpp
@@ -0,0 +1,2544 @@
+/******************************************************************************
+ * Project:  PROJ
+ * Purpose:  Functionality related to network access and caching
+ * Author:   Even Rouault, <even.rouault at spatialys.com>
+ *
+ ******************************************************************************
+ * Copyright (c) 2019-2020, Even Rouault, <even.rouault at spatialys.com>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+ * DEALINGS IN THE SOFTWARE.
+ *****************************************************************************/
+
+#ifndef FROM_PROJ_CPP
+#define FROM_PROJ_CPP
+#endif
+#define LRU11_DO_NOT_DEFINE_OUT_OF_CLASS_METHODS
+
+#include <stdlib.h>
+
+#include <algorithm>
+#include <limits>
+#include <string>
+
+#include "filemanager.hpp"
+#include "proj.h"
+#include "proj/internal/internal.hpp"
+#include "proj/internal/lru_cache.hpp"
+#include "proj_internal.h"
+#include "sqlite3_utils.hpp"
+
+#ifdef __MINGW32__
+// mingw32-win32 doesn't implement std::mutex
+namespace {
+class MyMutex {
+  public:
+    // cppcheck-suppress functionStatic
+    void lock() { pj_acquire_lock(); }
+    // cppcheck-suppress functionStatic
+    void unlock() { pj_release_lock(); }
+};
+}
+#else
+#include <mutex>
+#define MyMutex std::mutex
+#endif
+
+#ifdef CURL_ENABLED
+#include <curl/curl.h>
+#include <sqlite3.h> // for sqlite3_snprintf
+#endif
+
+#include <sys/stat.h>
+
+#ifdef _WIN32
+#include <shlobj.h>
+#else
+#include <sys/types.h>
+#include <unistd.h>
+#endif
+
+#if defined(_WIN32)
+#include <windows.h>
+#elif defined(__MACH__) && defined(__APPLE__)
+#include <mach-o/dyld.h>
+#elif defined(__FreeBSD__)
+#include <sys/sysctl.h>
+#include <sys/types.h>
+#endif
+
+#include <time.h>
+
+//! @cond Doxygen_Suppress
+
+#define STR_HELPER(x) #x
+#define STR(x) STR_HELPER(x)
+
+using namespace NS_PROJ::internal;
+
+NS_PROJ_START
+
+// ---------------------------------------------------------------------------
+
+static void sleep_ms(int ms) {
+#ifdef _WIN32
+    Sleep(ms);
+#else
+    usleep(ms * 1000);
+#endif
+}
+
+// ---------------------------------------------------------------------------
+
+constexpr size_t DOWNLOAD_CHUNK_SIZE = 16 * 1024;
+constexpr int MAX_CHUNKS = 64;
+
+struct FileProperties {
+    unsigned long long size = 0;
+    time_t lastChecked = 0;
+    std::string lastModified{};
+    std::string etag{};
+};
+
+class NetworkChunkCache {
+  public:
+    void insert(PJ_CONTEXT *ctx, const std::string &url,
+                unsigned long long chunkIdx, std::vector<unsigned char> &&data);
+
+    std::shared_ptr<std::vector<unsigned char>>
+    get(PJ_CONTEXT *ctx, const std::string &url, unsigned long long chunkIdx);
+
+    std::shared_ptr<std::vector<unsigned char>> get(PJ_CONTEXT *ctx,
+                                                    const std::string &url,
+                                                    unsigned long long chunkIdx,
+                                                    FileProperties &props);
+
+    void clearMemoryCache();
+
+    static void clearDiskChunkCache(PJ_CONTEXT *ctx);
+
+  private:
+    struct Key {
+        std::string url;
+        unsigned long long chunkIdx;
+
+        Key(const std::string &urlIn, unsigned long long chunkIdxIn)
+            : url(urlIn), chunkIdx(chunkIdxIn) {}
+        bool operator==(const Key &other) const {
+            return url == other.url && chunkIdx == other.chunkIdx;
+        }
+    };
+
+    struct KeyHasher {
+        std::size_t operator()(const Key &k) const {
+            return std::hash<std::string>{}(k.url) ^
+                   (std::hash<unsigned long long>{}(k.chunkIdx) << 1);
+        }
+    };
+
+    lru11::Cache<
+        Key, std::shared_ptr<std::vector<unsigned char>>, MyMutex,
+        std::unordered_map<
+            Key,
+            typename std::list<lru11::KeyValuePair<
+                Key, std::shared_ptr<std::vector<unsigned char>>>>::iterator,
+            KeyHasher>>
+        cache_{MAX_CHUNKS};
+};
+
+// ---------------------------------------------------------------------------
+
+static NetworkChunkCache gNetworkChunkCache{};
+
+// ---------------------------------------------------------------------------
+
+class NetworkFilePropertiesCache {
+  public:
+    void insert(PJ_CONTEXT *ctx, const std::string &url, FileProperties &props);
+
+    bool tryGet(PJ_CONTEXT *ctx, const std::string &url, FileProperties &props);
+
+    void clearMemoryCache();
+
+  private:
+    lru11::Cache<std::string, FileProperties, MyMutex> cache_{};
+};
+
+// ---------------------------------------------------------------------------
+
+static NetworkFilePropertiesCache gNetworkFileProperties{};
+
+// ---------------------------------------------------------------------------
+
+class DiskChunkCache {
+    PJ_CONTEXT *ctx_ = nullptr;
+    std::string path_{};
+    sqlite3 *hDB_ = nullptr;
+    std::string thisNamePtr_{};
+    std::unique_ptr<SQLite3VFS> vfs_{};
+
+    explicit DiskChunkCache(PJ_CONTEXT *ctx, const std::string &path);
+ 
+    bool initialize();
+    void commitAndClose();
+
+    bool createDBStructure();
+    bool checkConsistency();
+    bool get_links(sqlite3_int64 chunk_id, sqlite3_int64 &link_id,
+                   sqlite3_int64 &prev, sqlite3_int64 &next,
+                   sqlite3_int64 &head, sqlite3_int64 &tail);
+    bool update_links_of_prev_and_next_links(sqlite3_int64 prev,
+                                             sqlite3_int64 next);
+    bool update_linked_chunks(sqlite3_int64 link_id, sqlite3_int64 prev,
+                              sqlite3_int64 next);
+    bool update_linked_chunks_head_tail(sqlite3_int64 head, sqlite3_int64 tail);
+
+    DiskChunkCache(const DiskChunkCache &) = delete;
+    DiskChunkCache &operator=(const DiskChunkCache &) = delete;
+
+  public:
+    static std::unique_ptr<DiskChunkCache> open(PJ_CONTEXT *ctx);
+    ~DiskChunkCache();
+
+    sqlite3 *handle() { return hDB_; }
+    std::unique_ptr<SQLiteStatement> prepare(const char *sql);
+    bool move_to_head(sqlite3_int64 chunk_id);
+    bool move_to_tail(sqlite3_int64 chunk_id);
+    void closeAndUnlink();
+};
+
+// ---------------------------------------------------------------------------
+
+static bool pj_context_get_grid_cache_is_enabled(PJ_CONTEXT *ctx) {
+    pj_load_ini(ctx);
+    return ctx->gridChunkCache.enabled;
+}
+
+// ---------------------------------------------------------------------------
+
+static long long pj_context_get_grid_cache_max_size(PJ_CONTEXT *ctx) {
+    pj_load_ini(ctx);
+    return ctx->gridChunkCache.max_size;
+}
+
+// ---------------------------------------------------------------------------
+
+static int pj_context_get_grid_cache_ttl(PJ_CONTEXT *ctx) {
+    pj_load_ini(ctx);
+    return ctx->gridChunkCache.ttl;
+}
+
+// ---------------------------------------------------------------------------
+
+std::unique_ptr<DiskChunkCache> DiskChunkCache::open(PJ_CONTEXT *ctx) {
+    if (!pj_context_get_grid_cache_is_enabled(ctx)) {
+        return nullptr;
+    }
+    const auto cachePath = pj_context_get_grid_cache_filename(ctx);
+    if (cachePath.empty()) {
+        return nullptr;
+    }
+
+    auto diskCache =
+        std::unique_ptr<DiskChunkCache>(new DiskChunkCache(ctx, cachePath));
+    if (!diskCache->initialize())
+        diskCache.reset();
+    return diskCache;
+}
+
+// ---------------------------------------------------------------------------
+
+DiskChunkCache::DiskChunkCache(PJ_CONTEXT *ctx, const std::string &path)
+    : ctx_(ctx), path_(path) {}
+
+// ---------------------------------------------------------------------------
+
+bool DiskChunkCache::initialize() {
+    std::string vfsName;
+    if (ctx_->custom_sqlite3_vfs_name.empty()) {
+        vfs_ = SQLite3VFS::create(true, false, false);
+        if (vfs_ == nullptr) {
+            return false;
+        }
+        vfsName = vfs_->name();
+    } else {
+        vfsName = ctx_->custom_sqlite3_vfs_name;
+    }
+    sqlite3_open_v2(path_.c_str(), &hDB_,
+                    SQLITE_OPEN_READWRITE | SQLITE_OPEN_CREATE,
+                    vfsName.c_str());
+    if (!hDB_) {
+        pj_log(ctx_, PJ_LOG_ERROR, "Cannot open %s", path_.c_str());
+        return false;
+    }
+
+    // Cannot run more than 30 times / a bit more than one second.
+    for (int i = 0;; i++) {
+        int ret =
+            sqlite3_exec(hDB_, "BEGIN EXCLUSIVE", nullptr, nullptr, nullptr);
+        if (ret == SQLITE_OK) {
+            break;
+        }
+        if (ret != SQLITE_BUSY) {
+            pj_log(ctx_, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB_));
+            sqlite3_close(hDB_);
+            hDB_ = nullptr;
+            return false;
+        }
+        const char *max_iters = getenv("PROJ_LOCK_MAX_ITERS");
+        if (i >= (max_iters && max_iters[0] ? atoi(max_iters)
+                                            : 30)) { // A bit more than 1 second
+            pj_log(ctx_, PJ_LOG_ERROR, "Cannot take exclusive lock on %s",
+                   path_.c_str());
+            sqlite3_close(hDB_);
+            hDB_ = nullptr;
+            return false;
+        }
+        pj_log(ctx_, PJ_LOG_TRACE, "Lock taken on cache. Waiting a bit...");
+        // Retry every 5 ms for 50 ms, then every 10 ms for 100 ms, then
+        // every 100 ms
+        sleep_ms(i < 10 ? 5 : i < 20 ? 10 : 100);
+    }
+    char **pasResult = nullptr;
+    int nRows = 0;
+    int nCols = 0;
+    sqlite3_get_table(hDB_,
+                      "SELECT 1 FROM sqlite_master WHERE name = 'properties'",
+                      &pasResult, &nRows, &nCols, nullptr);
+    sqlite3_free_table(pasResult);
+    if (nRows == 0) {
+        if (!createDBStructure()) {
+            sqlite3_close(hDB_);
+            hDB_ = nullptr;
+            return false;
+        }
+    }
+
+    if (getenv("PROJ_CHECK_CACHE_CONSISTENCY")) {
+        checkConsistency();
+    }
+    return true;
+}
+
+// ---------------------------------------------------------------------------
+
+static const char *cache_db_structure_sql =
+    "CREATE TABLE properties("
+    " url          TEXT PRIMARY KEY NOT NULL,"
+    " lastChecked  TIMESTAMP NOT NULL,"
+    " fileSize     INTEGER NOT NULL,"
+    " lastModified TEXT,"
+    " etag         TEXT"
+    ");"
+    "CREATE TABLE downloaded_file_properties("
+    " url          TEXT PRIMARY KEY NOT NULL,"
+    " lastChecked  TIMESTAMP NOT NULL,"
+    " fileSize     INTEGER NOT NULL,"
+    " lastModified TEXT,"
+    " etag         TEXT"
+    ");"
+    "CREATE TABLE chunk_data("
+    " id        INTEGER PRIMARY KEY AUTOINCREMENT CHECK (id > 0),"
+    " data      BLOB NOT NULL"
+    ");"
+    "CREATE TABLE chunks("
+    " id        INTEGER PRIMARY KEY AUTOINCREMENT CHECK (id > 0),"
+    " url       TEXT NOT NULL,"
+    " offset    INTEGER NOT NULL,"
+    " data_id   INTEGER NOT NULL,"
+    " data_size INTEGER NOT NULL,"
+    " CONSTRAINT fk_chunks_url FOREIGN KEY (url) REFERENCES properties(url),"
+    " CONSTRAINT fk_chunks_data FOREIGN KEY (data_id) REFERENCES chunk_data(id)"
+    ");"
+    "CREATE INDEX idx_chunks ON chunks(url, offset);"
+    "CREATE TABLE linked_chunks("
+    " id        INTEGER PRIMARY KEY AUTOINCREMENT CHECK (id > 0),"
+    " chunk_id  INTEGER NOT NULL,"
+    " prev      INTEGER,"
+    " next      INTEGER,"
+    " CONSTRAINT fk_links_chunkid FOREIGN KEY (chunk_id) REFERENCES chunks(id),"
+    " CONSTRAINT fk_links_prev FOREIGN KEY (prev) REFERENCES linked_chunks(id),"
+    " CONSTRAINT fk_links_next FOREIGN KEY (next) REFERENCES linked_chunks(id)"
+    ");"
+    "CREATE INDEX idx_linked_chunks_chunk_id ON linked_chunks(chunk_id);"
+    "CREATE TABLE linked_chunks_head_tail("
+    "  head       INTEGER,"
+    "  tail       INTEGER,"
+    "  CONSTRAINT lht_head FOREIGN KEY (head) REFERENCES linked_chunks(id),"
+    "  CONSTRAINT lht_tail FOREIGN KEY (tail) REFERENCES linked_chunks(id)"
+    ");"
+    "INSERT INTO linked_chunks_head_tail VALUES (NULL, NULL);";
+
+bool DiskChunkCache::createDBStructure() {
+
+    pj_log(ctx_, PJ_LOG_TRACE, "Creating cache DB structure");
+    if (sqlite3_exec(hDB_, cache_db_structure_sql, nullptr, nullptr, nullptr) !=
+        SQLITE_OK) {
+        pj_log(ctx_, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB_));
+        return false;
+    }
+    return true;
+}
+
+// ---------------------------------------------------------------------------
+
+// Used by checkConsistency() and insert()
+#define INVALIDATED_SQL_LITERAL "'invalidated'"
+
+bool DiskChunkCache::checkConsistency() {
+
+    auto stmt = prepare("SELECT * FROM chunk_data WHERE id NOT IN (SELECT "
+                        "data_id FROM chunks)");
+    if (!stmt) {
+        return false;
+    }
+    if (stmt->execute() != SQLITE_DONE) {
+        fprintf(stderr, "Rows in chunk_data not referenced by chunks.\n");
+        return false;
+    }
+
+    stmt = prepare("SELECT * FROM chunks WHERE id NOT IN (SELECT chunk_id FROM "
+                   "linked_chunks)");
+    if (!stmt) {
+        return false;
+    }
+    if (stmt->execute() != SQLITE_DONE) {
+        fprintf(stderr, "Rows in chunks not referenced by linked_chunks.\n");
+        return false;
+    }
+
+    stmt = prepare("SELECT * FROM chunks WHERE url <> " INVALIDATED_SQL_LITERAL
+                   " AND url "
+                   "NOT IN (SELECT url FROM properties)");
+    if (!stmt) {
+        return false;
+    }
+    if (stmt->execute() != SQLITE_DONE) {
+        fprintf(stderr, "url values in chunks not referenced by properties.\n");
+        return false;
+    }
+
+    stmt = prepare("SELECT head, tail FROM linked_chunks_head_tail");
+    if (!stmt) {
+        return false;
+    }
+    if (stmt->execute() != SQLITE_ROW) {
+        fprintf(stderr, "linked_chunks_head_tail empty.\n");
+        return false;
+    }
+    const auto head = stmt->getInt64();
+    const auto tail = stmt->getInt64();
+    if (stmt->execute() != SQLITE_DONE) {
+        fprintf(stderr, "linked_chunks_head_tail has more than one row.\n");
+        return false;
+    }
+
+    stmt = prepare("SELECT COUNT(*) FROM linked_chunks");
+    if (!stmt) {
+        return false;
+    }
+    if (stmt->execute() != SQLITE_ROW) {
+        fprintf(stderr, "linked_chunks_head_tail empty.\n");
+        return false;
+    }
+    const auto count_linked_chunks = stmt->getInt64();
+
+    if (head) {
+        auto id = head;
+        std::set<sqlite3_int64> visitedIds;
+        stmt = prepare("SELECT next FROM linked_chunks WHERE id = ?");
+        if (!stmt) {
+            return false;
+        }
+        while (true) {
+            visitedIds.insert(id);
+            stmt->reset();
+            stmt->bindInt64(id);
+            if (stmt->execute() != SQLITE_ROW) {
+                fprintf(stderr, "cannot find linked_chunks.id = %d.\n",
+                        static_cast<int>(id));
+                return false;
+            }
+            auto next = stmt->getInt64();
+            if (next == 0) {
+                if (id != tail) {
+                    fprintf(stderr,
+                            "last item when following next is not tail.\n");
+                    return false;
+                }
+                break;
+            }
+            if (visitedIds.find(next) != visitedIds.end()) {
+                fprintf(stderr, "found cycle on linked_chunks.next = %d.\n",
+                        static_cast<int>(next));
+                return false;
+            }
+            id = next;
+        }
+        if (visitedIds.size() != static_cast<size_t>(count_linked_chunks)) {
+            fprintf(stderr,
+                    "ghost items in linked_chunks when following next.\n");
+            return false;
+        }
+    } else if (count_linked_chunks) {
+        fprintf(stderr, "linked_chunks_head_tail.head = NULL but linked_chunks "
+                        "not empty.\n");
+        return false;
+    }
+
+    if (tail) {
+        auto id = tail;
+        std::set<sqlite3_int64> visitedIds;
+        stmt = prepare("SELECT prev FROM linked_chunks WHERE id = ?");
+        if (!stmt) {
+            return false;
+        }
+        while (true) {
+            visitedIds.insert(id);
+            stmt->reset();
+            stmt->bindInt64(id);
+            if (stmt->execute() != SQLITE_ROW) {
+                fprintf(stderr, "cannot find linked_chunks.id = %d.\n",
+                        static_cast<int>(id));
+                return false;
+            }
+            auto prev = stmt->getInt64();
+            if (prev == 0) {
+                if (id != head) {
+                    fprintf(stderr,
+                            "last item when following prev is not head.\n");
+                    return false;
+                }
+                break;
+            }
+            if (visitedIds.find(prev) != visitedIds.end()) {
+                fprintf(stderr, "found cycle on linked_chunks.prev = %d.\n",
+                        static_cast<int>(prev));
+                return false;
+            }
+            id = prev;
+        }
+        if (visitedIds.size() != static_cast<size_t>(count_linked_chunks)) {
+            fprintf(stderr,
+                    "ghost items in linked_chunks when following prev.\n");
+            return false;
+        }
+    } else if (count_linked_chunks) {
+        fprintf(stderr, "linked_chunks_head_tail.tail = NULL but linked_chunks "
+                        "not empty.\n");
+        return false;
+    }
+
+    fprintf(stderr, "check ok\n");
+    return true;
+}
+
+// ---------------------------------------------------------------------------
+
+void DiskChunkCache::commitAndClose() {
+    if (hDB_) {
+        if( sqlite3_exec(hDB_, "COMMIT", nullptr, nullptr, nullptr) != SQLITE_OK ) {
+            pj_log(ctx_, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB_));
+        }
+        sqlite3_close(hDB_);
+        hDB_ = nullptr;
+    }
+}
+
+// ---------------------------------------------------------------------------
+
+DiskChunkCache::~DiskChunkCache() {
+    commitAndClose();
+}
+
+// ---------------------------------------------------------------------------
+
+void DiskChunkCache::closeAndUnlink() {
+    commitAndClose();
+    if (vfs_) {
+        vfs_->raw()->xDelete(vfs_->raw(), path_.c_str(), 0);
+    }
+}
+
+// ---------------------------------------------------------------------------
+
+std::unique_ptr<SQLiteStatement> DiskChunkCache::prepare(const char *sql) {
+    sqlite3_stmt *hStmt = nullptr;
+    sqlite3_prepare_v2(hDB_, sql, -1, &hStmt, nullptr);
+    if (!hStmt) {
+        pj_log(ctx_, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB_));
+        return nullptr;
+    }
+    return std::unique_ptr<SQLiteStatement>(new SQLiteStatement(hStmt));
+}
+
+// ---------------------------------------------------------------------------
+
+bool DiskChunkCache::get_links(sqlite3_int64 chunk_id, sqlite3_int64 &link_id,
+                               sqlite3_int64 &prev, sqlite3_int64 &next,
+                               sqlite3_int64 &head, sqlite3_int64 &tail) {
+    auto stmt =
+        prepare("SELECT id, prev, next FROM linked_chunks WHERE chunk_id = ?");
+    if (!stmt)
+        return false;
+    stmt->bindInt64(chunk_id);
+    {
+        const auto ret = stmt->execute();
+        if (ret != SQLITE_ROW) {
+            pj_log(ctx_, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB_));
+            return false;
+        }
+    }
+    link_id = stmt->getInt64();
+    prev = stmt->getInt64();
+    next = stmt->getInt64();
+
+    stmt = prepare("SELECT head, tail FROM linked_chunks_head_tail");
+    {
+        const auto ret = stmt->execute();
+        if (ret != SQLITE_ROW) {
+            pj_log(ctx_, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB_));
+            return false;
+        }
+    }
+    head = stmt->getInt64();
+    tail = stmt->getInt64();
+    return true;
+}
+
+// ---------------------------------------------------------------------------
+
+bool DiskChunkCache::update_links_of_prev_and_next_links(sqlite3_int64 prev,
+                                                         sqlite3_int64 next) {
+    if (prev) {
+        auto stmt = prepare("UPDATE linked_chunks SET next = ? WHERE id = ?");
+        if (!stmt)
+            return false;
+        if (next)
+            stmt->bindInt64(next);
+        else
+            stmt->bindNull();
+        stmt->bindInt64(prev);
+        const auto ret = stmt->execute();
+        if (ret != SQLITE_DONE) {
+            pj_log(ctx_, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB_));
+            return false;
+        }
+    }
+
+    if (next) {
+        auto stmt = prepare("UPDATE linked_chunks SET prev = ? WHERE id = ?");
+        if (!stmt)
+            return false;
+        if (prev)
+            stmt->bindInt64(prev);
+        else
+            stmt->bindNull();
+        stmt->bindInt64(next);
+        const auto ret = stmt->execute();
+        if (ret != SQLITE_DONE) {
+            pj_log(ctx_, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB_));
+            return false;
+        }
+    }
+    return true;
+}
+
+// ---------------------------------------------------------------------------
+
+bool DiskChunkCache::update_linked_chunks(sqlite3_int64 link_id,
+                                          sqlite3_int64 prev,
+                                          sqlite3_int64 next) {
+    auto stmt =
+        prepare("UPDATE linked_chunks SET prev = ?, next = ? WHERE id = ?");
+    if (!stmt)
+        return false;
+    if (prev)
+        stmt->bindInt64(prev);
+    else
+        stmt->bindNull();
+    if (next)
+        stmt->bindInt64(next);
+    else
+        stmt->bindNull();
+    stmt->bindInt64(link_id);
+    const auto ret = stmt->execute();
+    if (ret != SQLITE_DONE) {
+        pj_log(ctx_, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB_));
+        return false;
+    }
+    return true;
+}
+
+// ---------------------------------------------------------------------------
+
+bool DiskChunkCache::update_linked_chunks_head_tail(sqlite3_int64 head,
+                                                    sqlite3_int64 tail) {
+    auto stmt =
+        prepare("UPDATE linked_chunks_head_tail SET head = ?, tail = ?");
+    if (!stmt)
+        return false;
+    if (head)
+        stmt->bindInt64(head);
+    else
+        stmt->bindNull(); // shouldn't happen normally
+    if (tail)
+        stmt->bindInt64(tail);
+    else
+        stmt->bindNull(); // shouldn't happen normally
+    const auto ret = stmt->execute();
+    if (ret != SQLITE_DONE) {
+        pj_log(ctx_, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB_));
+        return false;
+    }
+    return true;
+}
+
+// ---------------------------------------------------------------------------
+
+bool DiskChunkCache::move_to_head(sqlite3_int64 chunk_id) {
+
+    sqlite3_int64 link_id = 0;
+    sqlite3_int64 prev = 0;
+    sqlite3_int64 next = 0;
+    sqlite3_int64 head = 0;
+    sqlite3_int64 tail = 0;
+    if (!get_links(chunk_id, link_id, prev, next, head, tail)) {
+        return false;
+    }
+
+    if (link_id == head) {
+        return true;
+    }
+
+    if (!update_links_of_prev_and_next_links(prev, next)) {
+        return false;
+    }
+
+    if (head) {
+        auto stmt = prepare("UPDATE linked_chunks SET prev = ? WHERE id = ?");
+        if (!stmt)
+            return false;
+        stmt->bindInt64(link_id);
+        stmt->bindInt64(head);
+        const auto ret = stmt->execute();
+        if (ret != SQLITE_DONE) {
+            pj_log(ctx_, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB_));
+            return false;
+        }
+    }
+
+    return update_linked_chunks(link_id, 0, head) &&
+           update_linked_chunks_head_tail(link_id,
+                                          (link_id == tail) ? prev : tail);
+}
+
+// ---------------------------------------------------------------------------
+
+bool DiskChunkCache::move_to_tail(sqlite3_int64 chunk_id) {
+    sqlite3_int64 link_id = 0;
+    sqlite3_int64 prev = 0;
+    sqlite3_int64 next = 0;
+    sqlite3_int64 head = 0;
+    sqlite3_int64 tail = 0;
+    if (!get_links(chunk_id, link_id, prev, next, head, tail)) {
+        return false;
+    }
+
+    if (link_id == tail) {
+        return true;
+    }
+
+    if (!update_links_of_prev_and_next_links(prev, next)) {
+        return false;
+    }
+
+    if (tail) {
+        auto stmt = prepare("UPDATE linked_chunks SET next = ? WHERE id = ?");
+        if (!stmt)
+            return false;
+        stmt->bindInt64(link_id);
+        stmt->bindInt64(tail);
+        const auto ret = stmt->execute();
+        if (ret != SQLITE_DONE) {
+            pj_log(ctx_, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB_));
+            return false;
+        }
+    }
+
+    return update_linked_chunks(link_id, tail, 0) &&
+           update_linked_chunks_head_tail((link_id == head) ? next : head,
+                                          link_id);
+}
+
+// ---------------------------------------------------------------------------
+
+void NetworkChunkCache::insert(PJ_CONTEXT *ctx, const std::string &url,
+                               unsigned long long chunkIdx,
+                               std::vector<unsigned char> &&data) {
+    auto dataPtr(std::make_shared<std::vector<unsigned char>>(std::move(data)));
+    cache_.insert(Key(url, chunkIdx), dataPtr);
+
+    auto diskCache = DiskChunkCache::open(ctx);
+    if (!diskCache)
+        return;
+    auto hDB = diskCache->handle();
+
+    // Always insert DOWNLOAD_CHUNK_SIZE bytes to avoid fragmentation
+    std::vector<unsigned char> blob(*dataPtr);
+    assert(blob.size() <= DOWNLOAD_CHUNK_SIZE);
+    blob.resize(DOWNLOAD_CHUNK_SIZE);
+
+    // Check if there is an existing entry for that URL and offset
+    auto stmt = diskCache->prepare(
+        "SELECT id, data_id FROM chunks WHERE url = ? AND offset = ?");
+    if (!stmt)
+        return;
+    stmt->bindText(url.c_str());
+    stmt->bindInt64(chunkIdx * DOWNLOAD_CHUNK_SIZE);
+
+    const auto mainRet = stmt->execute();
+    if (mainRet == SQLITE_ROW) {
+        const auto chunk_id = stmt->getInt64();
+        const auto data_id = stmt->getInt64();
+        stmt =
+            diskCache->prepare("UPDATE chunk_data SET data = ? WHERE id = ?");
+        if (!stmt)
+            return;
+        stmt->bindBlob(blob.data(), blob.size());
+        stmt->bindInt64(data_id);
+        {
+            const auto ret = stmt->execute();
+            if (ret != SQLITE_DONE) {
+                pj_log(ctx, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB));
+                return;
+            }
+        }
+
+        diskCache->move_to_head(chunk_id);
+
+        return;
+    } else if (mainRet != SQLITE_DONE) {
+        pj_log(ctx, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB));
+        return;
+    }
+
+    // Lambda to recycle an existing entry that was either invalidated, or
+    // least recently used.
+    const auto reuseExistingEntry = [ctx, &blob, &diskCache, hDB, &url,
+                                     chunkIdx, &dataPtr](
+        std::unique_ptr<SQLiteStatement> &stmtIn) {
+        const auto chunk_id = stmtIn->getInt64();
+        const auto data_id = stmtIn->getInt64();
+        if (data_id <= 0) {
+            pj_log(ctx, PJ_LOG_ERROR, "data_id <= 0");
+            return;
+        }
+
+        auto l_stmt =
+            diskCache->prepare("UPDATE chunk_data SET data = ? WHERE id = ?");
+        if (!l_stmt)
+            return;
+        l_stmt->bindBlob(blob.data(), blob.size());
+        l_stmt->bindInt64(data_id);
+        {
+            const auto ret2 = l_stmt->execute();
+            if (ret2 != SQLITE_DONE) {
+                pj_log(ctx, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB));
+                return;
+            }
+        }
+
+        l_stmt = diskCache->prepare("UPDATE chunks SET url = ?, "
+                                    "offset = ?, data_size = ?, data_id = ? "
+                                    "WHERE id = ?");
+        if (!l_stmt)
+            return;
+        l_stmt->bindText(url.c_str());
+        l_stmt->bindInt64(chunkIdx * DOWNLOAD_CHUNK_SIZE);
+        l_stmt->bindInt64(dataPtr->size());
+        l_stmt->bindInt64(data_id);
+        l_stmt->bindInt64(chunk_id);
+        {
+            const auto ret2 = l_stmt->execute();
+            if (ret2 != SQLITE_DONE) {
+                pj_log(ctx, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB));
+                return;
+            }
+        }
+
+        diskCache->move_to_head(chunk_id);
+    };
+
+    // Find if there is an invalidated chunk we can reuse
+    stmt = diskCache->prepare(
+        "SELECT id, data_id FROM chunks "
+        "WHERE id = (SELECT tail FROM linked_chunks_head_tail) AND "
+        "url = " INVALIDATED_SQL_LITERAL);
+    if (!stmt)
+        return;
+    {
+        const auto ret = stmt->execute();
+        if (ret == SQLITE_ROW) {
+            reuseExistingEntry(stmt);
+            return;
+        } else if (ret != SQLITE_DONE) {
+            pj_log(ctx, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB));
+            return;
+        }
+    }
+
+    // Check if we have not reached the max size of the cache
+    stmt = diskCache->prepare("SELECT COUNT(*) FROM chunks");
+    if (!stmt)
+        return;
+    {
+        const auto ret = stmt->execute();
+        if (ret != SQLITE_ROW) {
+            pj_log(ctx, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB));
+            return;
+        }
+    }
+
+    const auto max_size = pj_context_get_grid_cache_max_size(ctx);
+    if (max_size > 0 &&
+        static_cast<long long>(stmt->getInt64() * DOWNLOAD_CHUNK_SIZE) >=
+            max_size) {
+        stmt = diskCache->prepare(
+            "SELECT id, data_id FROM chunks "
+            "WHERE id = (SELECT tail FROM linked_chunks_head_tail)");
+        if (!stmt)
+            return;
+
+        const auto ret = stmt->execute();
+        if (ret != SQLITE_ROW) {
+            pj_log(ctx, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB));
+            return;
+        }
+        reuseExistingEntry(stmt);
+        return;
+    }
+
+    // Otherwise just append a new entry
+    stmt = diskCache->prepare("INSERT INTO chunk_data(data) VALUES (?)");
+    if (!stmt)
+        return;
+    stmt->bindBlob(blob.data(), blob.size());
+    {
+        const auto ret = stmt->execute();
+        if (ret != SQLITE_DONE) {
+            pj_log(ctx, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB));
+            return;
+        }
+    }
+
+    const auto chunk_data_id = sqlite3_last_insert_rowid(hDB);
+
+    stmt = diskCache->prepare("INSERT INTO chunks(url, offset, data_id, "
+                              "data_size) VALUES (?,?,?,?)");
+    if (!stmt)
+        return;
+    stmt->bindText(url.c_str());
+    stmt->bindInt64(chunkIdx * DOWNLOAD_CHUNK_SIZE);
+    stmt->bindInt64(chunk_data_id);
+    stmt->bindInt64(dataPtr->size());
+    {
+        const auto ret = stmt->execute();
+        if (ret != SQLITE_DONE) {
+            pj_log(ctx, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB));
+            return;
+        }
+    }
+
+    const auto chunk_id = sqlite3_last_insert_rowid(hDB);
+
+    stmt = diskCache->prepare(
+        "INSERT INTO linked_chunks(chunk_id, prev, next) VALUES (?,NULL,NULL)");
+    if (!stmt)
+        return;
+    stmt->bindInt64(chunk_id);
+    if (stmt->execute() != SQLITE_DONE) {
+        pj_log(ctx, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB));
+        return;
+    }
+
+    stmt = diskCache->prepare("SELECT head FROM linked_chunks_head_tail");
+    if (!stmt)
+        return;
+    if (stmt->execute() != SQLITE_ROW) {
+        pj_log(ctx, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB));
+        return;
+    }
+    if (stmt->getInt64() == 0) {
+        stmt = diskCache->prepare(
+            "UPDATE linked_chunks_head_tail SET head = ?, tail = ?");
+        if (!stmt)
+            return;
+        stmt->bindInt64(chunk_id);
+        stmt->bindInt64(chunk_id);
+        if (stmt->execute() != SQLITE_DONE) {
+            pj_log(ctx, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB));
+            return;
+        }
+    }
+
+    diskCache->move_to_head(chunk_id);
+}
+
+// ---------------------------------------------------------------------------
+
+std::shared_ptr<std::vector<unsigned char>>
+NetworkChunkCache::get(PJ_CONTEXT *ctx, const std::string &url,
+                       unsigned long long chunkIdx) {
+    std::shared_ptr<std::vector<unsigned char>> ret;
+    if (cache_.tryGet(Key(url, chunkIdx), ret)) {
+        return ret;
+    }
+
+    auto diskCache = DiskChunkCache::open(ctx);
+    if (!diskCache)
+        return ret;
+    auto hDB = diskCache->handle();
+
+    auto stmt = diskCache->prepare(
+        "SELECT chunks.id, chunks.data_size, chunk_data.data FROM chunks "
+        "JOIN chunk_data ON chunks.id = chunk_data.id "
+        "WHERE chunks.url = ? AND chunks.offset = ?");
+    if (!stmt)
+        return ret;
+
+    stmt->bindText(url.c_str());
+    stmt->bindInt64(chunkIdx * DOWNLOAD_CHUNK_SIZE);
+
+    const auto mainRet = stmt->execute();
+    if (mainRet == SQLITE_ROW) {
+        const auto chunk_id = stmt->getInt64();
+        const auto data_size = stmt->getInt64();
+        int blob_size = 0;
+        const void *blob = stmt->getBlob(blob_size);
+        if (blob_size < data_size) {
+            pj_log(ctx, PJ_LOG_ERROR,
+                   "blob_size=%d < data_size for chunk_id=%d", blob_size,
+                   static_cast<int>(chunk_id));
+            return ret;
+        }
+        if (data_size > static_cast<sqlite3_int64>(DOWNLOAD_CHUNK_SIZE)) {
+            pj_log(ctx, PJ_LOG_ERROR, "data_size > DOWNLOAD_CHUNK_SIZE");
+            return ret;
+        }
+        ret.reset(new std::vector<unsigned char>());
+        ret->assign(reinterpret_cast<const unsigned char *>(blob),
+                    reinterpret_cast<const unsigned char *>(blob) +
+                        static_cast<size_t>(data_size));
+        cache_.insert(Key(url, chunkIdx), ret);
+
+        if (!diskCache->move_to_head(chunk_id))
+            return ret;
+    } else if (mainRet != SQLITE_DONE) {
+        pj_log(ctx, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB));
+    }
+
+    return ret;
+}
+
+// ---------------------------------------------------------------------------
+
+std::shared_ptr<std::vector<unsigned char>>
+NetworkChunkCache::get(PJ_CONTEXT *ctx, const std::string &url,
+                       unsigned long long chunkIdx, FileProperties &props) {
+    if (!gNetworkFileProperties.tryGet(ctx, url, props)) {
+        return nullptr;
+    }
+
+    return get(ctx, url, chunkIdx);
+}
+
+// ---------------------------------------------------------------------------
+
+void NetworkChunkCache::clearMemoryCache() { cache_.clear(); }
+
+// ---------------------------------------------------------------------------
+
+void NetworkChunkCache::clearDiskChunkCache(PJ_CONTEXT *ctx) {
+    auto diskCache = DiskChunkCache::open(ctx);
+    if (!diskCache)
+        return;
+    diskCache->closeAndUnlink();
+}
+
+// ---------------------------------------------------------------------------
+
+void NetworkFilePropertiesCache::insert(PJ_CONTEXT *ctx, const std::string &url,
+                                        FileProperties &props) {
+    time(&props.lastChecked);
+    cache_.insert(url, props);
+
+    auto diskCache = DiskChunkCache::open(ctx);
+    if (!diskCache)
+        return;
+    auto hDB = diskCache->handle();
+    auto stmt = diskCache->prepare("SELECT fileSize, lastModified, etag "
+                                   "FROM properties WHERE url = ?");
+    if (!stmt)
+        return;
+    stmt->bindText(url.c_str());
+    if (stmt->execute() == SQLITE_ROW) {
+        FileProperties cachedProps;
+        cachedProps.size = stmt->getInt64();
+        const char *lastModified = stmt->getText();
+        cachedProps.lastModified = lastModified ? lastModified : std::string();
+        const char *etag = stmt->getText();
+        cachedProps.etag = etag ? etag : std::string();
+        if (props.size != cachedProps.size ||
+            props.lastModified != cachedProps.lastModified ||
+            props.etag != cachedProps.etag) {
+
+            // If cached properties don't match recent fresh ones, invalidate
+            // cached chunks
+            stmt = diskCache->prepare("SELECT id FROM chunks WHERE url = ?");
+            if (!stmt)
+                return;
+            stmt->bindText(url.c_str());
+            std::vector<sqlite3_int64> ids;
+            while (stmt->execute() == SQLITE_ROW) {
+                ids.emplace_back(stmt->getInt64());
+                stmt->resetResIndex();
+            }
+
+            for (const auto id : ids) {
+                diskCache->move_to_tail(id);
+            }
+
+            stmt = diskCache->prepare(
+                "UPDATE chunks SET url = " INVALIDATED_SQL_LITERAL ", "
+                "offset = -1, data_size = 0 WHERE url = ?");
+            if (!stmt)
+                return;
+            stmt->bindText(url.c_str());
+            if (stmt->execute() != SQLITE_DONE) {
+                pj_log(ctx, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB));
+                return;
+            }
+        }
+
+        stmt = diskCache->prepare("UPDATE properties SET lastChecked = ?, "
+                                  "fileSize = ?, lastModified = ?, etag = ? "
+                                  "WHERE url = ?");
+        if (!stmt)
+            return;
+        stmt->bindInt64(props.lastChecked);
+        stmt->bindInt64(props.size);
+        if (props.lastModified.empty())
+            stmt->bindNull();
+        else
+            stmt->bindText(props.lastModified.c_str());
+        if (props.etag.empty())
+            stmt->bindNull();
+        else
+            stmt->bindText(props.etag.c_str());
+        stmt->bindText(url.c_str());
+        if (stmt->execute() != SQLITE_DONE) {
+            pj_log(ctx, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB));
+            return;
+        }
+    } else {
+        stmt = diskCache->prepare("INSERT INTO properties (url, lastChecked, "
+                                  "fileSize, lastModified, etag) VALUES "
+                                  "(?,?,?,?,?)");
+        if (!stmt)
+            return;
+        stmt->bindText(url.c_str());
+        stmt->bindInt64(props.lastChecked);
+        stmt->bindInt64(props.size);
+        if (props.lastModified.empty())
+            stmt->bindNull();
+        else
+            stmt->bindText(props.lastModified.c_str());
+        if (props.etag.empty())
+            stmt->bindNull();
+        else
+            stmt->bindText(props.etag.c_str());
+        if (stmt->execute() != SQLITE_DONE) {
+            pj_log(ctx, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB));
+            return;
+        }
+    }
+}
+
+// ---------------------------------------------------------------------------
+
+bool NetworkFilePropertiesCache::tryGet(PJ_CONTEXT *ctx, const std::string &url,
+                                        FileProperties &props) {
+    if (cache_.tryGet(url, props)) {
+        return true;
+    }
+
+    auto diskCache = DiskChunkCache::open(ctx);
+    if (!diskCache)
+        return false;
+    auto stmt =
+        diskCache->prepare("SELECT lastChecked, fileSize, lastModified, etag "
+                           "FROM properties WHERE url = ?");
+    if (!stmt)
+        return false;
+    stmt->bindText(url.c_str());
+    if (stmt->execute() != SQLITE_ROW) {
+        return false;
+    }
+    props.lastChecked = stmt->getInt64();
+    props.size = stmt->getInt64();
+    const char *lastModified = stmt->getText();
+    props.lastModified = lastModified ? lastModified : std::string();
+    const char *etag = stmt->getText();
+    props.etag = etag ? etag : std::string();
+
+    const auto ttl = pj_context_get_grid_cache_ttl(ctx);
+    if (ttl > 0) {
+        time_t curTime;
+        time(&curTime);
+        if (curTime > props.lastChecked + ttl) {
+            props = FileProperties();
+            return false;
+        }
+    }
+    cache_.insert(url, props);
+    return true;
+}
+
+// ---------------------------------------------------------------------------
+
+void NetworkFilePropertiesCache::clearMemoryCache() { cache_.clear(); }
+
+// ---------------------------------------------------------------------------
+
+class NetworkFile : public File {
+    PJ_CONTEXT *m_ctx;
+    std::string m_url;
+    PROJ_NETWORK_HANDLE *m_handle;
+    unsigned long long m_pos = 0;
+    size_t m_nBlocksToDownload = 1;
+    unsigned long long m_lastDownloadedOffset;
+    FileProperties m_props;
+    proj_network_close_cbk_type m_closeCbk;
+    bool m_hasChanged = false;
+
+    NetworkFile(const NetworkFile &) = delete;
+    NetworkFile &operator=(const NetworkFile &) = delete;
+
+  protected:
+    NetworkFile(PJ_CONTEXT *ctx, const std::string &url,
+                PROJ_NETWORK_HANDLE *handle,
+                unsigned long long lastDownloadOffset,
+                const FileProperties &props)
+        : File(url), m_ctx(ctx), m_url(url), m_handle(handle),
+          m_lastDownloadedOffset(lastDownloadOffset), m_props(props),
+          m_closeCbk(ctx->networking.close) {}
+
+  public:
+    ~NetworkFile() override;
+
+    size_t read(void *buffer, size_t sizeBytes) override;
+    size_t write(const void *, size_t) override { return 0; }
+    bool seek(unsigned long long offset, int whence) override;
+    unsigned long long tell() override;
+    void reassign_context(PJ_CONTEXT *ctx) override;
+    bool hasChanged() const override { return m_hasChanged; }
+
+    static std::unique_ptr<File> open(PJ_CONTEXT *ctx, const char *filename);
+
+    static bool get_props_from_headers(PJ_CONTEXT *ctx,
+                                       PROJ_NETWORK_HANDLE *handle,
+                                       FileProperties &props);
+};
+
+// ---------------------------------------------------------------------------
+
+bool NetworkFile::get_props_from_headers(PJ_CONTEXT *ctx,
+                                         PROJ_NETWORK_HANDLE *handle,
+                                         FileProperties &props) {
+    const char *contentRange = ctx->networking.get_header_value(
+        ctx, handle, "Content-Range", ctx->networking.user_data);
+    if (contentRange) {
+        const char *slash = strchr(contentRange, '/');
+        if (slash) {
+            props.size = std::stoull(slash + 1);
+
+            const char *lastModified = ctx->networking.get_header_value(
+                ctx, handle, "Last-Modified", ctx->networking.user_data);
+            if (lastModified)
+                props.lastModified = lastModified;
+
+            const char *etag = ctx->networking.get_header_value(
+                ctx, handle, "ETag", ctx->networking.user_data);
+            if (etag)
+                props.etag = etag;
+
+            return true;
+        }
+    }
+    return false;
+}
+
+// ---------------------------------------------------------------------------
+
+std::unique_ptr<File> NetworkFile::open(PJ_CONTEXT *ctx, const char *filename) {
+    FileProperties props;
+    if (gNetworkChunkCache.get(ctx, filename, 0, props)) {
+        return std::unique_ptr<File>(new NetworkFile(
+            ctx, filename, nullptr,
+            std::numeric_limits<unsigned long long>::max(), props));
+    } else {
+        std::vector<unsigned char> buffer(DOWNLOAD_CHUNK_SIZE);
+        size_t size_read = 0;
+        std::string errorBuffer;
+        errorBuffer.resize(1024);
+
+        auto handle = ctx->networking.open(
+            ctx, filename, 0, buffer.size(), &buffer[0], &size_read,
+            errorBuffer.size(), &errorBuffer[0], ctx->networking.user_data);
+        buffer.resize(size_read);
+        if (!handle) {
+            errorBuffer.resize(strlen(errorBuffer.data()));
+            pj_log(ctx, PJ_LOG_ERROR, "Cannot open %s: %s", filename,
+                   errorBuffer.c_str());
+            pj_ctx_set_errno(ctx, PJD_ERR_NETWORK_ERROR);
+        }
+
+        bool ok = false;
+        if (handle) {
+            if (get_props_from_headers(ctx, handle, props)) {
+                ok = true;
+                gNetworkFileProperties.insert(ctx, filename, props);
+                gNetworkChunkCache.insert(ctx, filename, 0, std::move(buffer));
+            }
+        }
+
+        return std::unique_ptr<File>(
+            ok ? new NetworkFile(ctx, filename, handle, size_read, props)
+               : nullptr);
+    }
+}
+
+// ---------------------------------------------------------------------------
+
+std::unique_ptr<File> pj_network_file_open(PJ_CONTEXT* ctx, const char* filename) {
+    return NetworkFile::open(ctx, filename);
+}
+
+// ---------------------------------------------------------------------------
+
+size_t NetworkFile::read(void *buffer, size_t sizeBytes) {
+
+    if (sizeBytes == 0)
+        return 0;
+
+    auto iterOffset = m_pos;
+    while (sizeBytes) {
+        const auto chunkIdxToDownload = iterOffset / DOWNLOAD_CHUNK_SIZE;
+        const auto offsetToDownload = chunkIdxToDownload * DOWNLOAD_CHUNK_SIZE;
+        std::vector<unsigned char> region;
+        auto pChunk = gNetworkChunkCache.get(m_ctx, m_url, chunkIdxToDownload);
+        if (pChunk != nullptr) {
+            region = *pChunk;
+        } else {
+            if (offsetToDownload == m_lastDownloadedOffset) {
+                // In case of consecutive reads (of small size), we use a
+                // heuristic that we will read the file sequentially, so
+                // we double the requested size to decrease the number of
+                // client/server roundtrips.
+                if (m_nBlocksToDownload < 100)
+                    m_nBlocksToDownload *= 2;
+            } else {
+                // Random reads. Cancel the above heuristics.
+                m_nBlocksToDownload = 1;
+            }
+
+            // Ensure that we will request at least the number of blocks
+            // to satisfy the remaining buffer size to read.
+            const auto endOffsetToDownload =
+                ((iterOffset + sizeBytes + DOWNLOAD_CHUNK_SIZE - 1) /
+                 DOWNLOAD_CHUNK_SIZE) *
+                DOWNLOAD_CHUNK_SIZE;
+            const auto nMinBlocksToDownload = static_cast<size_t>(
+                (endOffsetToDownload - offsetToDownload) / DOWNLOAD_CHUNK_SIZE);
+            if (m_nBlocksToDownload < nMinBlocksToDownload)
+                m_nBlocksToDownload = nMinBlocksToDownload;
+
+            // Avoid reading already cached data.
+            // Note: this might get evicted if concurrent reads are done, but
+            // this should not cause bugs. Just missed optimization.
+            for (size_t i = 1; i < m_nBlocksToDownload; i++) {
+                if (gNetworkChunkCache.get(m_ctx, m_url,
+                                           chunkIdxToDownload + i) != nullptr) {
+                    m_nBlocksToDownload = i;
+                    break;
+                }
+            }
+
+            if (m_nBlocksToDownload > MAX_CHUNKS)
+                m_nBlocksToDownload = MAX_CHUNKS;
+
+            region.resize(m_nBlocksToDownload * DOWNLOAD_CHUNK_SIZE);
+            size_t nRead = 0;
+            std::string errorBuffer;
+            errorBuffer.resize(1024);
+            if (!m_handle) {
+                m_handle = m_ctx->networking.open(
+                    m_ctx, m_url.c_str(), offsetToDownload,
+                    m_nBlocksToDownload * DOWNLOAD_CHUNK_SIZE, &region[0],
+                    &nRead, errorBuffer.size(), &errorBuffer[0],
+                    m_ctx->networking.user_data);
+                if (!m_handle) {
+                    pj_ctx_set_errno(m_ctx, PJD_ERR_NETWORK_ERROR);
+                    return 0;
+                }
+            } else {
+                nRead = m_ctx->networking.read_range(
+                    m_ctx, m_handle, offsetToDownload,
+                    m_nBlocksToDownload * DOWNLOAD_CHUNK_SIZE, &region[0],
+                    errorBuffer.size(), &errorBuffer[0],
+                    m_ctx->networking.user_data);
+            }
+            if (nRead == 0) {
+                errorBuffer.resize(strlen(errorBuffer.data()));
+                if (!errorBuffer.empty()) {
+                    pj_log(m_ctx, PJ_LOG_ERROR, "Cannot read in %s: %s",
+                           m_url.c_str(), errorBuffer.c_str());
+                }
+                pj_ctx_set_errno(m_ctx, PJD_ERR_NETWORK_ERROR);
+                return 0;
+            }
+
+            if (!m_hasChanged) {
+                FileProperties props;
+                if (get_props_from_headers(m_ctx, m_handle, props)) {
+                    if (props.size != m_props.size ||
+                        props.lastModified != m_props.lastModified ||
+                        props.etag != m_props.etag) {
+                        gNetworkFileProperties.insert(m_ctx, m_url, props);
+                        gNetworkChunkCache.clearMemoryCache();
+                        m_hasChanged = true;
+                    }
+                }
+            }
+
+            region.resize(nRead);
+            m_lastDownloadedOffset = offsetToDownload + nRead;
+
+            const auto nChunks =
+                (region.size() + DOWNLOAD_CHUNK_SIZE - 1) / DOWNLOAD_CHUNK_SIZE;
+            for (size_t i = 0; i < nChunks; i++) {
+                std::vector<unsigned char> chunk(
+                    region.data() + i * DOWNLOAD_CHUNK_SIZE,
+                    region.data() +
+                        std::min((i + 1) * DOWNLOAD_CHUNK_SIZE, region.size()));
+                gNetworkChunkCache.insert(m_ctx, m_url, chunkIdxToDownload + i,
+                                          std::move(chunk));
+            }
+        }
+        const size_t nToCopy = static_cast<size_t>(
+            std::min(static_cast<unsigned long long>(sizeBytes),
+                     region.size() - (iterOffset - offsetToDownload)));
+        memcpy(buffer, region.data() + iterOffset - offsetToDownload, nToCopy);
+        buffer = static_cast<char *>(buffer) + nToCopy;
+        iterOffset += nToCopy;
+        sizeBytes -= nToCopy;
+        if (region.size() < static_cast<size_t>(DOWNLOAD_CHUNK_SIZE) &&
+            sizeBytes != 0) {
+            break;
+        }
+    }
+
+    size_t nRead = static_cast<size_t>(iterOffset - m_pos);
+    m_pos = iterOffset;
+    return nRead;
+}
+
+// ---------------------------------------------------------------------------
+
+bool NetworkFile::seek(unsigned long long offset, int whence) {
+    if (whence == SEEK_SET) {
+        m_pos = offset;
+    } else if (whence == SEEK_CUR) {
+        m_pos += offset;
+    } else {
+        if (offset != 0)
+            return false;
+        m_pos = m_props.size;
+    }
+    return true;
+}
+
+// ---------------------------------------------------------------------------
+
+unsigned long long NetworkFile::tell() { return m_pos; }
+
+// ---------------------------------------------------------------------------
+
+NetworkFile::~NetworkFile() {
+    if (m_handle) {
+        m_ctx->networking.close(m_ctx, m_handle, m_ctx->networking.user_data);
+    }
+}
+
+// ---------------------------------------------------------------------------
+
+void NetworkFile::reassign_context(PJ_CONTEXT *ctx) {
+    m_ctx = ctx;
+    if (m_closeCbk != m_ctx->networking.close) {
+        pj_log(m_ctx, PJ_LOG_ERROR,
+               "Networking close callback has changed following context "
+               "reassignment ! This is highly suspicious");
+    }
+}
+
+// ---------------------------------------------------------------------------
+
+#ifdef CURL_ENABLED
+
+struct CurlFileHandle {
+    std::string m_url;
+    CURL *m_handle;
+    std::string m_headers{};
+    std::string m_lastval{};
+    std::string m_useragent{};
+    char m_szCurlErrBuf[CURL_ERROR_SIZE + 1] = {};
+
+    CurlFileHandle(const CurlFileHandle &) = delete;
+    CurlFileHandle &operator=(const CurlFileHandle &) = delete;
+
+    explicit CurlFileHandle(const char *url, CURL *handle);
+    ~CurlFileHandle();
+
+    static PROJ_NETWORK_HANDLE *
+    open(PJ_CONTEXT *, const char *url, unsigned long long offset,
+         size_t size_to_read, void *buffer, size_t *out_size_read,
+         size_t error_string_max_size, char *out_error_string, void *);
+};
+
+// ---------------------------------------------------------------------------
+
+static std::string GetExecutableName() {
+#if defined(__linux)
+    std::string path;
+    path.resize(1024);
+    const auto ret = readlink("/proc/self/exe", &path[0], path.size());
+    if (ret > 0) {
+        path.resize(ret);
+        const auto pos = path.rfind('/');
+        if (pos != std::string::npos) {
+            path = path.substr(pos + 1);
+        }
+        return path;
+    }
+#elif defined(_WIN32)
+    std::string path;
+    path.resize(1024);
+    if (GetModuleFileNameA(nullptr, &path[0],
+                           static_cast<DWORD>(path.size()))) {
+        path.resize(strlen(path.c_str()));
+        const auto pos = path.rfind('\\');
+        if (pos != std::string::npos) {
+            path = path.substr(pos + 1);
+        }
+        return path;
+    }
+#elif defined(__MACH__) && defined(__APPLE__)
+    std::string path;
+    path.resize(1024);
+    uint32_t size = static_cast<uint32_t>(path.size());
+    if (_NSGetExecutablePath(&path[0], &size) == 0) {
+        path.resize(strlen(path.c_str()));
+        const auto pos = path.rfind('/');
+        if (pos != std::string::npos) {
+            path = path.substr(pos + 1);
+        }
+        return path;
+    }
+#elif defined(__FreeBSD__)
+    int mib[4];
+    mib[0] = CTL_KERN;
+    mib[1] = KERN_PROC;
+    mib[2] = KERN_PROC_PATHNAME;
+    mib[3] = -1;
+    std::string path;
+    path.resize(1024);
+    size_t size = path.size();
+    if (sysctl(mib, 4, &path[0], &size, nullptr, 0) == 0) {
+        path.resize(strlen(path.c_str()));
+        const auto pos = path.rfind('/');
+        if (pos != std::string::npos) {
+            path = path.substr(pos + 1);
+        }
+        return path;
+    }
+#endif
+
+    return std::string();
+}
+
+// ---------------------------------------------------------------------------
+
+CurlFileHandle::CurlFileHandle(const char *url, CURL *handle)
+    : m_url(url), m_handle(handle) {
+    curl_easy_setopt(handle, CURLOPT_URL, m_url.c_str());
+
+    if (getenv("PROJ_CURL_VERBOSE"))
+        curl_easy_setopt(handle, CURLOPT_VERBOSE, 1);
+
+// CURLOPT_SUPPRESS_CONNECT_HEADERS is defined in curl 7.54.0 or newer.
+#if LIBCURL_VERSION_NUM >= 0x073600
+    curl_easy_setopt(handle, CURLOPT_SUPPRESS_CONNECT_HEADERS, 1L);
+#endif
+
+    // Enable following redirections.  Requires libcurl 7.10.1 at least.
+    curl_easy_setopt(handle, CURLOPT_FOLLOWLOCATION, 1);
+    curl_easy_setopt(handle, CURLOPT_MAXREDIRS, 10);
+
+    if (getenv("PROJ_UNSAFE_SSL")) {
+        curl_easy_setopt(handle, CURLOPT_SSL_VERIFYPEER, 0L);
+        curl_easy_setopt(handle, CURLOPT_SSL_VERIFYHOST, 0L);
+    }
+
+    curl_easy_setopt(handle, CURLOPT_ERRORBUFFER, m_szCurlErrBuf);
+
+    if (getenv("PROJ_NO_USERAGENT") == nullptr) {
+        m_useragent = "PROJ " STR(PROJ_VERSION_MAJOR) "." STR(
+            PROJ_VERSION_MINOR) "." STR(PROJ_VERSION_PATCH);
+        const auto exeName = GetExecutableName();
+        if (!exeName.empty()) {
+            m_useragent = exeName + " using " + m_useragent;
+        }
+        curl_easy_setopt(handle, CURLOPT_USERAGENT, m_useragent.data());
+    }
+}
+
+// ---------------------------------------------------------------------------
+
+CurlFileHandle::~CurlFileHandle() { curl_easy_cleanup(m_handle); }
+
+// ---------------------------------------------------------------------------
+
+static size_t pj_curl_write_func(void *buffer, size_t count, size_t nmemb,
+                                 void *req) {
+    const size_t nSize = count * nmemb;
+    auto pStr = static_cast<std::string *>(req);
+    if (pStr->size() + nSize > pStr->capacity()) {
+        // to avoid servers not honouring Range to cause excessive memory
+        // allocation
+        return 0;
+    }
+    pStr->append(static_cast<const char *>(buffer), nSize);
+    return nmemb;
+}
+
+// ---------------------------------------------------------------------------
+
+static double GetNewRetryDelay(int response_code, double dfOldDelay,
+                               const char *pszErrBuf,
+                               const char *pszCurlError) {
+    if (response_code == 429 || response_code == 500 ||
+        (response_code >= 502 && response_code <= 504) ||
+        // S3 sends some client timeout errors as 400 Client Error
+        (response_code == 400 && pszErrBuf &&
+         strstr(pszErrBuf, "RequestTimeout")) ||
+        (pszCurlError && strstr(pszCurlError, "Connection timed out"))) {
+        // Use an exponential backoff factor of 2 plus some random jitter
+        // We don't care about cryptographic quality randomness, hence:
+        // coverity[dont_call]
+        return dfOldDelay * (2 + rand() * 0.5 / RAND_MAX);
+    } else {
+        return 0;
+    }
+}
+
+// ---------------------------------------------------------------------------
+
+constexpr double MIN_RETRY_DELAY_MS = 500;
+constexpr double MAX_RETRY_DELAY_MS = 60000;
+
+PROJ_NETWORK_HANDLE *CurlFileHandle::open(PJ_CONTEXT *ctx, const char *url,
+                                          unsigned long long offset,
+                                          size_t size_to_read, void *buffer,
+                                          size_t *out_size_read,
+                                          size_t error_string_max_size,
+                                          char *out_error_string, void *) {
+    CURL *hCurlHandle = curl_easy_init();
+    if (!hCurlHandle)
+        return nullptr;
+
+    auto file =
+        std::unique_ptr<CurlFileHandle>(new CurlFileHandle(url, hCurlHandle));
+
+    double oldDelay = MIN_RETRY_DELAY_MS;
+    std::string headers;
+    std::string body;
+
+    char szBuffer[128];
+    sqlite3_snprintf(sizeof(szBuffer), szBuffer, "%llu-%llu", offset,
+                     offset + size_to_read - 1);
+
+    while (true) {
+        curl_easy_setopt(hCurlHandle, CURLOPT_RANGE, szBuffer);
+
+        headers.clear();
+        headers.reserve(16 * 1024);
+        curl_easy_setopt(hCurlHandle, CURLOPT_HEADERDATA, &headers);
+        curl_easy_setopt(hCurlHandle, CURLOPT_HEADERFUNCTION,
+                         pj_curl_write_func);
+
+        body.clear();
+        body.reserve(size_to_read);
+        curl_easy_setopt(hCurlHandle, CURLOPT_WRITEDATA, &body);
+        curl_easy_setopt(hCurlHandle, CURLOPT_WRITEFUNCTION,
+                         pj_curl_write_func);
+
+        file->m_szCurlErrBuf[0] = '\0';
+
+        curl_easy_perform(hCurlHandle);
+
+        long response_code = 0;
+        curl_easy_getinfo(hCurlHandle, CURLINFO_HTTP_CODE, &response_code);
+
+        curl_easy_setopt(hCurlHandle, CURLOPT_HEADERDATA, nullptr);
+        curl_easy_setopt(hCurlHandle, CURLOPT_HEADERFUNCTION, nullptr);
+
+        curl_easy_setopt(hCurlHandle, CURLOPT_WRITEDATA, nullptr);
+        curl_easy_setopt(hCurlHandle, CURLOPT_WRITEFUNCTION, nullptr);
+
+        if (response_code == 0 || response_code >= 300) {
+            const double delay =
+                GetNewRetryDelay(static_cast<int>(response_code), oldDelay,
+                                 body.c_str(), file->m_szCurlErrBuf);
+            if (delay != 0 && delay < MAX_RETRY_DELAY_MS) {
+                pj_log(ctx, PJ_LOG_TRACE,
+                       "Got a HTTP %ld error. Retrying in %d ms", response_code,
+                       static_cast<int>(delay));
+                sleep_ms(static_cast<int>(delay));
+                oldDelay = delay;
+            } else {
+                if (out_error_string) {
+                    if (file->m_szCurlErrBuf[0]) {
+                        snprintf(out_error_string, error_string_max_size, "%s",
+                                 file->m_szCurlErrBuf);
+                    } else {
+                        snprintf(out_error_string, error_string_max_size,
+                                 "HTTP error %ld: %s", response_code,
+                                 body.c_str());
+                    }
+                }
+                return nullptr;
+            }
+        } else {
+            break;
+        }
+    }
+
+    if (out_error_string && error_string_max_size) {
+        out_error_string[0] = '\0';
+    }
+
+    if (!body.empty()) {
+        memcpy(buffer, body.data(), std::min(size_to_read, body.size()));
+    }
+    *out_size_read = std::min(size_to_read, body.size());
+
+    file->m_headers = std::move(headers);
+    return reinterpret_cast<PROJ_NETWORK_HANDLE *>(file.release());
+}
+
+// ---------------------------------------------------------------------------
+
+static void pj_curl_close(PJ_CONTEXT *, PROJ_NETWORK_HANDLE *handle,
+                          void * /*user_data*/) {
+    delete reinterpret_cast<CurlFileHandle *>(handle);
+}
+
+// ---------------------------------------------------------------------------
+
+static size_t pj_curl_read_range(PJ_CONTEXT *ctx,
+                                 PROJ_NETWORK_HANDLE *raw_handle,
+                                 unsigned long long offset, size_t size_to_read,
+                                 void *buffer, size_t error_string_max_size,
+                                 char *out_error_string, void *) {
+    auto handle = reinterpret_cast<CurlFileHandle *>(raw_handle);
+    auto hCurlHandle = handle->m_handle;
+
+    double oldDelay = MIN_RETRY_DELAY_MS;
+    std::string headers;
+    std::string body;
+
+    char szBuffer[128];
+    sqlite3_snprintf(sizeof(szBuffer), szBuffer, "%llu-%llu", offset,
+                     offset + size_to_read - 1);
+
+    while (true) {
+        curl_easy_setopt(hCurlHandle, CURLOPT_RANGE, szBuffer);
+
+        headers.clear();
+        headers.reserve(16 * 1024);
+        curl_easy_setopt(hCurlHandle, CURLOPT_HEADERDATA, &headers);
+        curl_easy_setopt(hCurlHandle, CURLOPT_HEADERFUNCTION,
+                         pj_curl_write_func);
+
+        body.clear();
+        body.reserve(size_to_read);
+        curl_easy_setopt(hCurlHandle, CURLOPT_WRITEDATA, &body);
+        curl_easy_setopt(hCurlHandle, CURLOPT_WRITEFUNCTION,
+                         pj_curl_write_func);
+
+        handle->m_szCurlErrBuf[0] = '\0';
+
+        curl_easy_perform(hCurlHandle);
+
+        long response_code = 0;
+        curl_easy_getinfo(hCurlHandle, CURLINFO_HTTP_CODE, &response_code);
+
+        curl_easy_setopt(hCurlHandle, CURLOPT_WRITEDATA, nullptr);
+        curl_easy_setopt(hCurlHandle, CURLOPT_WRITEFUNCTION, nullptr);
+
+        if (response_code == 0 || response_code >= 300) {
+            const double delay =
+                GetNewRetryDelay(static_cast<int>(response_code), oldDelay,
+                                 body.c_str(), handle->m_szCurlErrBuf);
+            if (delay != 0 && delay < MAX_RETRY_DELAY_MS) {
+                pj_log(ctx, PJ_LOG_TRACE,
+                       "Got a HTTP %ld error. Retrying in %d ms", response_code,
+                       static_cast<int>(delay));
+                sleep_ms(static_cast<int>(delay));
+                oldDelay = delay;
+            } else {
+                if (out_error_string) {
+                    if (handle->m_szCurlErrBuf[0]) {
+                        snprintf(out_error_string, error_string_max_size, "%s",
+                                 handle->m_szCurlErrBuf);
+                    } else {
+                        snprintf(out_error_string, error_string_max_size,
+                                 "HTTP error %ld: %s", response_code,
+                                 body.c_str());
+                    }
+                }
+                return 0;
+            }
+        } else {
+            break;
+        }
+    }
+    if (out_error_string && error_string_max_size) {
+        out_error_string[0] = '\0';
+    }
+
+    if (!body.empty()) {
+        memcpy(buffer, body.data(), std::min(size_to_read, body.size()));
+    }
+    handle->m_headers = std::move(headers);
+
+    return std::min(size_to_read, body.size());
+}
+
+// ---------------------------------------------------------------------------
+
+static const char *pj_curl_get_header_value(PJ_CONTEXT *,
+                                            PROJ_NETWORK_HANDLE *raw_handle,
+                                            const char *header_name, void *) {
+    auto handle = reinterpret_cast<CurlFileHandle *>(raw_handle);
+    auto pos = ci_find(handle->m_headers, header_name);
+    if (pos == std::string::npos)
+        return nullptr;
+    pos += strlen(header_name);
+    const char *c_str = handle->m_headers.c_str();
+    if (c_str[pos] == ':')
+        pos++;
+    while (c_str[pos] == ' ')
+        pos++;
+    auto posEnd = pos;
+    while (c_str[posEnd] != '\r' && c_str[posEnd] != '\n' &&
+           c_str[posEnd] != '\0')
+        posEnd++;
+    handle->m_lastval = handle->m_headers.substr(pos, posEnd - pos);
+    return handle->m_lastval.c_str();
+}
+
+#else
+
+// ---------------------------------------------------------------------------
+
+static PROJ_NETWORK_HANDLE *
+no_op_network_open(PJ_CONTEXT *, const char * /* url */,
+                   unsigned long long, /* offset */
+                   size_t,             /* size to read */
+                   void *,             /* buffer to update with bytes read*/
+                   size_t *,           /* output: size actually read */
+                   size_t error_string_max_size, char *out_error_string,
+                   void * /*user_data*/) {
+    if (out_error_string) {
+        snprintf(out_error_string, error_string_max_size, "%s",
+                 "Network functionality not available");
+    }
+    return nullptr;
+}
+
+// ---------------------------------------------------------------------------
+
+static void no_op_network_close(PJ_CONTEXT *, PROJ_NETWORK_HANDLE *,
+                                void * /*user_data*/) {}
+
+#endif
+
+// ---------------------------------------------------------------------------
+
+void FileManager::fillDefaultNetworkInterface(PJ_CONTEXT *ctx) {
+#ifdef CURL_ENABLED
+    ctx->networking.open = CurlFileHandle::open;
+    ctx->networking.close = pj_curl_close;
+    ctx->networking.read_range = pj_curl_read_range;
+    ctx->networking.get_header_value = pj_curl_get_header_value;
+#else
+    ctx->networking.open = no_op_network_open;
+    ctx->networking.close = no_op_network_close;
+#endif
+}
+
+// ---------------------------------------------------------------------------
+
+void FileManager::clearMemoryCache() {
+    gNetworkChunkCache.clearMemoryCache();
+    gNetworkFileProperties.clearMemoryCache();
+}
+
+NS_PROJ_END
+
+//! @endcond
+
+// ---------------------------------------------------------------------------
+
+#ifdef WIN32
+static const char dir_chars[] = "/\\";
+#else
+static const char dir_chars[] = "/";
+#endif
+
+static bool is_tilde_slash(const char *name) {
+    return *name == '~' && strchr(dir_chars, name[1]);
+}
+
+static bool is_rel_or_absolute_filename(const char *name) {
+    return strchr(dir_chars, *name) ||
+           (*name == '.' && strchr(dir_chars, name[1])) ||
+           (!strncmp(name, "..", 2) && strchr(dir_chars, name[2])) ||
+           (name[0] != '\0' && name[1] == ':' && strchr(dir_chars, name[2]));
+}
+
+static std::string build_url(PJ_CONTEXT *ctx, const char *name) {
+    if (!is_tilde_slash(name) && !is_rel_or_absolute_filename(name) &&
+        !starts_with(name, "http://") && !starts_with(name, "https://")) {
+        std::string remote_file(pj_context_get_url_endpoint(ctx));
+        if (!remote_file.empty()) {
+            if (remote_file.back() != '/') {
+                remote_file += '/';
+            }
+            remote_file += name;
+            auto pos = remote_file.rfind('.');
+            if (pos + 4 == remote_file.size()) {
+                remote_file = remote_file.substr(0, pos) + ".tif";
+            } else {
+                // For example for resource files like 'alaska'
+                remote_file += ".tif";
+            }
+        }
+        return remote_file;
+    }
+    return name;
+}
+
+// ---------------------------------------------------------------------------
+
+/** Define a custom set of callbacks for network access.
+ *
+ * All callbacks should be provided (non NULL pointers).
+ *
+ * @param ctx PROJ context, or NULL
+ * @param open_cbk Callback to open a remote file given its URL
+ * @param close_cbk Callback to close a remote file.
+ * @param get_header_value_cbk Callback to get HTTP headers
+ * @param read_range_cbk Callback to read a range of bytes inside a remote file.
+ * @param user_data Arbitrary pointer provided by the user, and passed to the
+ * above callbacks. May be NULL.
+ * @return TRUE in case of success.
+ * @since 7.0
+ */
+int proj_context_set_network_callbacks(
+    PJ_CONTEXT *ctx, proj_network_open_cbk_type open_cbk,
+    proj_network_close_cbk_type close_cbk,
+    proj_network_get_header_value_cbk_type get_header_value_cbk,
+    proj_network_read_range_type read_range_cbk, void *user_data) {
+    if (ctx == nullptr) {
+        ctx = pj_get_default_ctx();
+    }
+    if (!open_cbk || !close_cbk || !get_header_value_cbk || !read_range_cbk) {
+        return false;
+    }
+    ctx->networking.open = open_cbk;
+    ctx->networking.close = close_cbk;
+    ctx->networking.get_header_value = get_header_value_cbk;
+    ctx->networking.read_range = read_range_cbk;
+    ctx->networking.user_data = user_data;
+    return true;
+}
+
+// ---------------------------------------------------------------------------
+
+/** Enable or disable network access.
+*
+* This overrides the default endpoint in the PROJ configuration file or with
+* the PROJ_NETWORK environment variable.
+*
+* @param ctx PROJ context, or NULL
+* @param enable TRUE if network access is allowed.
+* @return TRUE if network access is possible. That is either libcurl is
+*         available, or an alternate interface has been set.
+* @since 7.0
+*/
+int proj_context_set_enable_network(PJ_CONTEXT *ctx, int enable) {
+    if (ctx == nullptr) {
+        ctx = pj_get_default_ctx();
+    }
+    // Load ini file, now so as to override its network settings
+    pj_load_ini(ctx);
+    ctx->networking.enabled_env_variable_checked = true;
+    ctx->networking.enabled = enable != FALSE;
+#ifdef CURL_ENABLED
+    return ctx->networking.enabled;
+#else
+    return ctx->networking.enabled &&
+           ctx->networking.open != NS_PROJ::no_op_network_open;
+#endif
+}
+
+//! @endcond
+
+// ---------------------------------------------------------------------------
+
+/** Define the URL endpoint to query for remote grids.
+*
+* This overrides the default endpoint in the PROJ configuration file or with
+* the PROJ_NETWORK_ENDPOINT environment variable.
+*
+* @param ctx PROJ context, or NULL
+* @param url Endpoint URL. Must NOT be NULL.
+* @since 7.0
+*/
+void proj_context_set_url_endpoint(PJ_CONTEXT *ctx, const char *url) {
+    if (ctx == nullptr) {
+        ctx = pj_get_default_ctx();
+    }
+    // Load ini file, now so as to override its network settings
+    pj_load_ini(ctx);
+    ctx->endpoint = url;
+}
+
+// ---------------------------------------------------------------------------
+
+/** Enable or disable the local cache of grid chunks
+*
+* This overrides the setting in the PROJ configuration file.
+*
+* @param ctx PROJ context, or NULL
+* @param enabled TRUE if the cache is enabled.
+* @since 7.0
+*/
+void proj_grid_cache_set_enable(PJ_CONTEXT *ctx, int enabled) {
+    if (ctx == nullptr) {
+        ctx = pj_get_default_ctx();
+    }
+    // Load ini file, now so as to override its settings
+    pj_load_ini(ctx);
+    ctx->gridChunkCache.enabled = enabled != FALSE;
+}
+
+// ---------------------------------------------------------------------------
+
+/** Override, for the considered context, the path and file of the local
+* cache of grid chunks.
+*
+* @param ctx PROJ context, or NULL
+* @param fullname Full name to the cache (encoded in UTF-8). If set to NULL,
+*                 caching will be disabled.
+* @since 7.0
+*/
+void proj_grid_cache_set_filename(PJ_CONTEXT *ctx, const char *fullname) {
+    if (ctx == nullptr) {
+        ctx = pj_get_default_ctx();
+    }
+    // Load ini file, now so as to override its settings
+    pj_load_ini(ctx);
+    ctx->gridChunkCache.filename = fullname ? fullname : std::string();
+}
+
+// ---------------------------------------------------------------------------
+
+/** Override, for the considered context, the maximum size of the local
+* cache of grid chunks.
+*
+* @param ctx PROJ context, or NULL
+* @param max_size_MB Maximum size, in mega-bytes (1024*1024 bytes), or
+*                    negative value to set unlimited size.
+* @since 7.0
+*/
+void proj_grid_cache_set_max_size(PJ_CONTEXT *ctx, int max_size_MB) {
+    if (ctx == nullptr) {
+        ctx = pj_get_default_ctx();
+    }
+    // Load ini file, now so as to override its settings
+    pj_load_ini(ctx);
+    ctx->gridChunkCache.max_size =
+        max_size_MB < 0 ? -1
+                        : static_cast<long long>(max_size_MB) * 1024 * 1024;
+    if (max_size_MB == 0) {
+        // For debug purposes only
+        const char *env_var = getenv("PROJ_GRID_CACHE_MAX_SIZE_BYTES");
+        if (env_var && env_var[0] != '\0') {
+            ctx->gridChunkCache.max_size = atoi(env_var);
+        }
+    }
+}
+
+// ---------------------------------------------------------------------------
+
+/** Override, for the considered context, the time-to-live delay for
+* re-checking if the cached properties of files are still up-to-date.
+*
+* @param ctx PROJ context, or NULL
+* @param ttl_seconds Delay in seconds. Use negative value for no expiration.
+* @since 7.0
+*/
+void proj_grid_cache_set_ttl(PJ_CONTEXT *ctx, int ttl_seconds) {
+    if (ctx == nullptr) {
+        ctx = pj_get_default_ctx();
+    }
+    // Load ini file, now so as to override its settings
+    pj_load_ini(ctx);
+    ctx->gridChunkCache.ttl = ttl_seconds;
+}
+
+// ---------------------------------------------------------------------------
+
+/** Clear the local cache of grid chunks.
+*
+* @param ctx PROJ context, or NULL
+* @since 7.0
+*/
+void proj_grid_cache_clear(PJ_CONTEXT *ctx) {
+    if (ctx == nullptr) {
+        ctx = pj_get_default_ctx();
+    }
+    NS_PROJ::gNetworkChunkCache.clearDiskChunkCache(ctx);
+}
+
+// ---------------------------------------------------------------------------
+
+/** Return if a file must be downloaded or is already available in the
+ * PROJ user-writable directory.
+ *
+ * The file will be determinted to have to be downloaded if it does not exist
+ * yet in the user-writable directory, or if it is determined that a more recent
+ * version exists. To determine if a more recent version exists, PROJ will
+ * use the "downloaded_file_properties" table of its grid cache database.
+ * Consequently files manually placed in the user-writable
+ * directory without using this function would be considered as
+ * non-existing/obsolete and would be unconditionnaly downloaded again.
+ *
+ * This function can only be used if networking is enabled, and either
+ * the default curl network API or a custom one have been installed.
+ *
+ * @param ctx PROJ context, or NULL
+ * @param url_or_filename URL or filename (without directory component)
+ * @param ignore_ttl_setting If set to FALSE, PROJ will only check the
+ *                           recentness of an already downloaded file, if
+ *                           the delay between the last time it has been
+ *                           verified and the current time exceeds the TTL
+ *                           setting. This can save network accesses.
+ *                           If set to TRUE, PROJ will unconditionnally
+ *                           check from the server the recentness of the file.
+ * @return TRUE if the file must be downloaded with proj_download_file()
+ * @since 7.0
+ */
+
+int proj_is_download_needed(PJ_CONTEXT *ctx, const char *url_or_filename,
+                            int ignore_ttl_setting) {
+    if (ctx == nullptr) {
+        ctx = pj_get_default_ctx();
+    }
+    if (!pj_context_is_network_enabled(ctx)) {
+        pj_log(ctx, PJ_LOG_ERROR, "Networking capabilities are not enabled");
+        return false;
+    }
+
+    const auto url(build_url(ctx, url_or_filename));
+    const char *filename = strrchr(url.c_str(), '/');
+    if (filename == nullptr)
+        return false;
+    const auto localFilename(
+        pj_context_get_user_writable_directory(ctx, false) + filename);
+
+    auto f = NS_PROJ::FileManager::open(ctx, localFilename.c_str(),
+                                        NS_PROJ::FileAccess::READ_ONLY);
+    if (!f) {
+        return true;
+    }
+    f.reset();
+
+    auto diskCache = NS_PROJ::DiskChunkCache::open(ctx);
+    if (!diskCache)
+        return false;
+    auto stmt =
+        diskCache->prepare("SELECT lastChecked, fileSize, lastModified, etag "
+                           "FROM downloaded_file_properties WHERE url = ?");
+    if (!stmt)
+        return true;
+    stmt->bindText(url.c_str());
+    if (stmt->execute() != SQLITE_ROW) {
+        return true;
+    }
+
+    NS_PROJ::FileProperties cachedProps;
+    cachedProps.lastChecked = stmt->getInt64();
+    cachedProps.size = stmt->getInt64();
+    const char *lastModified = stmt->getText();
+    cachedProps.lastModified = lastModified ? lastModified : std::string();
+    const char *etag = stmt->getText();
+    cachedProps.etag = etag ? etag : std::string();
+
+    if (!ignore_ttl_setting) {
+        const auto ttl = NS_PROJ::pj_context_get_grid_cache_ttl(ctx);
+        if (ttl > 0) {
+            time_t curTime;
+            time(&curTime);
+            if (curTime > cachedProps.lastChecked + ttl) {
+
+                unsigned char dummy;
+                size_t size_read = 0;
+                std::string errorBuffer;
+                errorBuffer.resize(1024);
+                auto handle = ctx->networking.open(
+                    ctx, url.c_str(), 0, 1, &dummy, &size_read,
+                    errorBuffer.size(), &errorBuffer[0],
+                    ctx->networking.user_data);
+                if (!handle) {
+                    errorBuffer.resize(strlen(errorBuffer.data()));
+                    pj_log(ctx, PJ_LOG_ERROR, "Cannot open %s: %s", url.c_str(),
+                           errorBuffer.c_str());
+                    return false;
+                }
+                NS_PROJ::FileProperties props;
+                if (!NS_PROJ::NetworkFile::get_props_from_headers(ctx, handle,
+                                                                  props)) {
+                    ctx->networking.close(ctx, handle,
+                                          ctx->networking.user_data);
+                    return false;
+                }
+                ctx->networking.close(ctx, handle, ctx->networking.user_data);
+
+                if (props.size != cachedProps.size ||
+                    props.lastModified != cachedProps.lastModified ||
+                    props.etag != cachedProps.etag) {
+                    return true;
+                }
+
+                stmt = diskCache->prepare(
+                    "UPDATE downloaded_file_properties SET lastChecked = ? "
+                    "WHERE url = ?");
+                if (!stmt)
+                    return false;
+                stmt->bindInt64(curTime);
+                stmt->bindText(url.c_str());
+                if (stmt->execute() != SQLITE_DONE) {
+                    auto hDB = diskCache->handle();
+                    pj_log(ctx, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB));
+                    return false;
+                }
+            }
+        }
+    }
+
+    return false;
+}
+
+// ---------------------------------------------------------------------------
+
+/** Download a file in the PROJ user-writable directory.
+ *
+ * The file will only be downloaded if it does not exist yet in the
+ * user-writable directory, or if it is determined that a more recent
+ * version exists. To determine if a more recent version exists, PROJ will
+ * use the "downloaded_file_properties" table of its grid cache database.
+ * Consequently files manually placed in the user-writable
+ * directory without using this function would be considered as
+ * non-existing/obsolete and would be unconditionnaly downloaded again.
+ *
+ * This function can only be used if networking is enabled, and either
+ * the default curl network API or a custom one have been installed.
+ *
+ * @param ctx PROJ context, or NULL
+ * @param url_or_filename URL or filename (without directory component)
+ * @param ignore_ttl_setting If set to FALSE, PROJ will only check the
+ *                           recentness of an already downloaded file, if
+ *                           the delay between the last time it has been
+ *                           verified and the current time exceeds the TTL
+ *                           setting. This can save network accesses.
+ *                           If set to TRUE, PROJ will unconditionnally
+ *                           check from the server the recentness of the file.
+ * @param progress_cbk Progress callback, or NULL.
+ *                     The passed percentage is in the [0, 1] range.
+ *                     The progress callback must return TRUE
+ *                     if download must be continued.
+ * @param user_data User data to provide to the progress callback, or NULL
+ * @return TRUE if the download was successful (or not needed)
+ * @since 7.0
+ */
+
+int proj_download_file(PJ_CONTEXT *ctx, const char *url_or_filename,
+                       int ignore_ttl_setting,
+                       int (*progress_cbk)(PJ_CONTEXT *, double pct,
+                                           void *user_data),
+                       void *user_data) {
+    if (ctx == nullptr) {
+        ctx = pj_get_default_ctx();
+    }
+    if (!pj_context_is_network_enabled(ctx)) {
+        pj_log(ctx, PJ_LOG_ERROR, "Networking capabilities are not enabled");
+        return false;
+    }
+    if (!proj_is_download_needed(ctx, url_or_filename, ignore_ttl_setting)) {
+        return true;
+    }
+
+    const auto url(build_url(ctx, url_or_filename));
+    const char *filename = strrchr(url.c_str(), '/');
+    if (filename == nullptr)
+        return false;
+    const auto localFilename(pj_context_get_user_writable_directory(ctx, true) +
+                             filename);
+
+#ifdef _WIN32
+    const int nPID = GetCurrentProcessId();
+#else
+    const int nPID = getpid();
+#endif
+    char szUniqueSuffix[128];
+    snprintf(szUniqueSuffix, sizeof(szUniqueSuffix), "%d_%p", nPID, &url);
+    const auto localFilenameTmp(localFilename + szUniqueSuffix);
+    auto f = NS_PROJ::FileManager::open(ctx, localFilenameTmp.c_str(),
+                                        NS_PROJ::FileAccess::CREATE);
+    if (!f) {
+        pj_log(ctx, PJ_LOG_ERROR, "Cannot create %s", localFilenameTmp.c_str());
+        return false;
+    }
+
+    constexpr size_t FULL_FILE_CHUNK_SIZE = 1024 * 1024;
+    std::vector<unsigned char> buffer(FULL_FILE_CHUNK_SIZE);
+    // For testing purposes only
+    const char *env_var_PROJ_FULL_FILE_CHUNK_SIZE =
+        getenv("PROJ_FULL_FILE_CHUNK_SIZE");
+    if (env_var_PROJ_FULL_FILE_CHUNK_SIZE &&
+        env_var_PROJ_FULL_FILE_CHUNK_SIZE[0] != '\0') {
+        buffer.resize(atoi(env_var_PROJ_FULL_FILE_CHUNK_SIZE));
+    }
+    size_t size_read = 0;
+    std::string errorBuffer;
+    errorBuffer.resize(1024);
+    auto handle = ctx->networking.open(
+        ctx, url.c_str(), 0, buffer.size(), &buffer[0], &size_read,
+        errorBuffer.size(), &errorBuffer[0], ctx->networking.user_data);
+    if (!handle) {
+        errorBuffer.resize(strlen(errorBuffer.data()));
+        pj_log(ctx, PJ_LOG_ERROR, "Cannot open %s: %s", url.c_str(),
+               errorBuffer.c_str());
+        f.reset();
+        NS_PROJ::FileManager::unlink(ctx, localFilenameTmp.c_str());
+        return false;
+    }
+
+    time_t curTime;
+    time(&curTime);
+    NS_PROJ::FileProperties props;
+    if (!NS_PROJ::NetworkFile::get_props_from_headers(ctx, handle, props)) {
+        ctx->networking.close(ctx, handle, ctx->networking.user_data);
+        f.reset();
+        NS_PROJ::FileManager::unlink(ctx, localFilenameTmp.c_str());
+        return false;
+    }
+
+    if (size_read <
+        std::min(static_cast<unsigned long long>(buffer.size()), props.size)) {
+        pj_log(ctx, PJ_LOG_ERROR, "Did not get as many bytes as expected");
+        ctx->networking.close(ctx, handle, ctx->networking.user_data);
+        f.reset();
+        NS_PROJ::FileManager::unlink(ctx, localFilenameTmp.c_str());
+        return false;
+    }
+    if (f->write(buffer.data(), size_read) != size_read) {
+        pj_log(ctx, PJ_LOG_ERROR, "Write error");
+        ctx->networking.close(ctx, handle, ctx->networking.user_data);
+        f.reset();
+        NS_PROJ::FileManager::unlink(ctx, localFilenameTmp.c_str());
+        return false;
+    }
+
+    unsigned long long totalDownloaded = size_read;
+    while (totalDownloaded < props.size) {
+        if (totalDownloaded + buffer.size() > props.size) {
+            buffer.resize(static_cast<size_t>(props.size - totalDownloaded));
+        }
+        errorBuffer.resize(1024);
+        size_read = ctx->networking.read_range(
+            ctx, handle, totalDownloaded, buffer.size(), &buffer[0],
+            errorBuffer.size(), &errorBuffer[0], ctx->networking.user_data);
+
+        if (size_read < buffer.size()) {
+            pj_log(ctx, PJ_LOG_ERROR, "Did not get as many bytes as expected");
+            ctx->networking.close(ctx, handle, ctx->networking.user_data);
+            f.reset();
+            NS_PROJ::FileManager::unlink(ctx, localFilenameTmp.c_str());
+            return false;
+        }
+        if (f->write(buffer.data(), size_read) != size_read) {
+            pj_log(ctx, PJ_LOG_ERROR, "Write error");
+            ctx->networking.close(ctx, handle, ctx->networking.user_data);
+            f.reset();
+            NS_PROJ::FileManager::unlink(ctx, localFilenameTmp.c_str());
+            return false;
+        }
+
+        totalDownloaded += size_read;
+        if (progress_cbk &&
+            !progress_cbk(ctx, double(totalDownloaded) / props.size,
+                          user_data)) {
+            ctx->networking.close(ctx, handle, ctx->networking.user_data);
+            f.reset();
+            NS_PROJ::FileManager::unlink(ctx, localFilenameTmp.c_str());
+            return false;
+        }
+    }
+
+    ctx->networking.close(ctx, handle, ctx->networking.user_data);
+    f.reset();
+    NS_PROJ::FileManager::unlink(ctx, localFilename.c_str());
+    if (!NS_PROJ::FileManager::rename(ctx, localFilenameTmp.c_str(),
+                                      localFilename.c_str())) {
+        pj_log(ctx, PJ_LOG_ERROR, "Cannot rename %s to %s",
+               localFilenameTmp.c_str(), localFilename.c_str());
+        return false;
+    }
+
+    auto diskCache = NS_PROJ::DiskChunkCache::open(ctx);
+    if (!diskCache)
+        return false;
+    auto stmt =
+        diskCache->prepare("SELECT lastChecked, fileSize, lastModified, etag "
+                           "FROM downloaded_file_properties WHERE url = ?");
+    if (!stmt)
+        return false;
+    stmt->bindText(url.c_str());
+
+    props.lastChecked = curTime;
+    auto hDB = diskCache->handle();
+
+    if (stmt->execute() == SQLITE_ROW) {
+        stmt = diskCache->prepare(
+            "UPDATE downloaded_file_properties SET lastChecked = ?, "
+            "fileSize = ?, lastModified = ?, etag = ? "
+            "WHERE url = ?");
+        if (!stmt)
+            return false;
+        stmt->bindInt64(props.lastChecked);
+        stmt->bindInt64(props.size);
+        if (props.lastModified.empty())
+            stmt->bindNull();
+        else
+            stmt->bindText(props.lastModified.c_str());
+        if (props.etag.empty())
+            stmt->bindNull();
+        else
+            stmt->bindText(props.etag.c_str());
+        stmt->bindText(url.c_str());
+        if (stmt->execute() != SQLITE_DONE) {
+            pj_log(ctx, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB));
+            return false;
+        }
+    } else {
+        stmt = diskCache->prepare(
+            "INSERT INTO downloaded_file_properties (url, lastChecked, "
+            "fileSize, lastModified, etag) VALUES "
+            "(?,?,?,?,?)");
+        if (!stmt)
+            return false;
+        stmt->bindText(url.c_str());
+        stmt->bindInt64(props.lastChecked);
+        stmt->bindInt64(props.size);
+        if (props.lastModified.empty())
+            stmt->bindNull();
+        else
+            stmt->bindText(props.lastModified.c_str());
+        if (props.etag.empty())
+            stmt->bindNull();
+        else
+            stmt->bindText(props.etag.c_str());
+        if (stmt->execute() != SQLITE_DONE) {
+            pj_log(ctx, PJ_LOG_ERROR, "%s", sqlite3_errmsg(hDB));
+            return false;
+        }
+    }
+    return true;
+}
+
+// ---------------------------------------------------------------------------
+
+//! @cond Doxygen_Suppress
+
+bool pj_context_is_network_enabled(PJ_CONTEXT *ctx) {
+    if (ctx == nullptr) {
+        ctx = pj_get_default_ctx();
+    }
+    if (ctx->networking.enabled_env_variable_checked) {
+        return ctx->networking.enabled;
+    }
+    const char *enabled = getenv("PROJ_NETWORK");
+    if (enabled && enabled[0] != '\0') {
+        ctx->networking.enabled = ci_equal(enabled, "ON") ||
+                                  ci_equal(enabled, "YES") ||
+                                  ci_equal(enabled, "TRUE");
+    }
+    pj_load_ini(ctx);
+    ctx->networking.enabled_env_variable_checked = true;
+    return ctx->networking.enabled;
+}
+
+// ---------------------------------------------------------------------------
+
+std::string pj_context_get_grid_cache_filename(PJ_CONTEXT *ctx) {
+    pj_load_ini(ctx);
+    if (!ctx->gridChunkCache.filename.empty()) {
+        return ctx->gridChunkCache.filename;
+    }
+    const std::string path(pj_context_get_user_writable_directory(ctx, true));
+    ctx->gridChunkCache.filename = path + "/cache.db";
+    return ctx->gridChunkCache.filename;
+}
+
+//! @endcond