version 0.5.5

Avoid crashes when passing unknown dicts to the -u flag

Fix #87

.github/workflows/main.yml

@@ -20,12 +20,13 @@ jobs:
       fail-fast: true
 
       matrix:
-        os: [ubuntu-latest]
+        os: [ubuntu-20.04, ubuntu-latest]
     steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v3
         with:
           submodules: 'recursive'
       - uses: jwlawson/actions-setup-cmake@v1.4
+        if: matrix.os != 'ubuntu-latest'
         with:
           cmake-version: '3.5.1'
           github-api-token: ${{ secrets.GITHUB_TOKEN }}

CMakeLists.txt

@@ -95,7 +95,7 @@ set(CPACK_PACKAGE_VENDOR "Evgeniy Dushistov <dushistov@mail.ru>")
 set(CPACK_PACKAGE_DESCRIPTION_FILE "${CMAKE_CURRENT_SOURCE_DIR}/README.org")
 set(CPACK_PACKAGE_VERSION_MAJOR "0")
 set(CPACK_PACKAGE_VERSION_MINOR "5")
-set(CPACK_PACKAGE_VERSION_PATCH "4")
+set(CPACK_PACKAGE_VERSION_PATCH "5")
 
 set(sdcv_VERSION
 	"${CPACK_PACKAGE_VERSION_MAJOR}.${CPACK_PACKAGE_VERSION_MINOR}.${CPACK_PACKAGE_VERSION_PATCH}")

NEWS

@@ -1,3 +1,13 @@
+Version 0.5.5
+    - Avoid crashes when passing unknown dicts to the -u flag (by NiLuJe)
+    - Use off_t for stuff mainly assigned to a stat.st_size value
+Version 0.5.4
+    - Use binary search for synonyms
+    - Various improvments in work with synonyms
+    - Added --json (same as --json-output) to match man
+    - Show all matched result
+    - More robust parsing of ifo file
+    - Prevent crash if file size of files not matched expecting one for .oft files
 Version 0.5.3
   - Use single quotes around JSON data to reduce need for escaping
   - Store integer magic in cache file
@@ -36,7 +46,7 @@ Version 0.4.2
 * Russian translation update
 
 Version 0.4.1
-* Recreate cache if idx file was modified 
+* Recreate cache if idx file was modified
 * Abbility to use pager(SDCV_PAGER)
 * Add Chinese (traditional)  translation
 * Add  Ukrainian translation

README.org

@@ -1,6 +1,9 @@
 #+OPTIONS: ^:nil
 [[https://github.com/Dushistov/sdcv/actions?query=workflow%3ACI+branch%3Amaster][https://github.com/Dushistov/sdcv/workflows/CI/badge.svg]]
 [[https://github.com/Dushistov/sdcv/blob/master/LICENSE][https://img.shields.io/badge/license-GPL%202-brightgreen.svg]]
+
+* sdcv
+*sdcv* is a simple, cross-platform, text-based utility for working with dictionaries in [[http://stardict-4.sourceforge.net/][StarDict]] format.
 * How to compile and install
 #+BEGIN_SRC sh
 mkdir /tmp/build-sdcv

src/dictziplib.hpp

@@ -27,7 +27,7 @@ public:
 private:
     const char *start; /* start of mmap'd area */
     const char *end; /* end of mmap'd area */
-    unsigned long size; /* size of mmap */
+    off_t size; /* size of mmap */
 
     int type;
     z_stream zStream;
@@ -47,7 +47,7 @@ private:
     std::string origFilename;
     std::string comment;
     unsigned long crc;
-    unsigned long length;
+    off_t length;
     unsigned long compressedLength;
     DictCache cache[DICT_CACHE_SIZE];
     MapFile mapfile;

src/mapfile.hpp

@@ -22,13 +22,13 @@ public:
     ~MapFile();
     MapFile(const MapFile &) = delete;
     MapFile &operator=(const MapFile &) = delete;
-    bool open(const char *file_name, unsigned long file_size);
+    bool open(const char *file_name, off_t file_size);
     gchar *begin() { return data; }
 
 private:
     char *data = nullptr;
-    unsigned long size = 0ul;
 #ifdef HAVE_MMAP
+    size_t size = 0u;
     int mmap_fd = -1;
 #elif defined(_WIN32)
     HANDLE hFile = 0;
@@ -36,9 +36,8 @@ private:
 #endif
 };
 
-inline bool MapFile::open(const char *file_name, unsigned long file_size)
+inline bool MapFile::open(const char *file_name, off_t file_size)
 {
-    size = file_size;
 #ifdef HAVE_MMAP
     if ((mmap_fd = ::open(file_name, O_RDONLY)) < 0) {
         // g_print("Open file %s failed!\n",fullfilename);
@@ -46,14 +45,16 @@ inline bool MapFile::open(const char *file_name, unsigned long file_size)
     }
     struct stat st;
     if (fstat(mmap_fd, &st) == -1 || st.st_size < 0 || (st.st_size == 0 && S_ISREG(st.st_mode))
-        || sizeof(st.st_size) > sizeof(file_size) || static_cast<unsigned long>(st.st_size) != file_size) {
+        || st.st_size != file_size) {
         close(mmap_fd);
         return false;
     }
 
-    data = (gchar *)mmap(nullptr, file_size, PROT_READ, MAP_SHARED, mmap_fd, 0);
+    size = static_cast<size_t>(st.st_size);
+    data = (gchar *)mmap(nullptr, size, PROT_READ, MAP_SHARED, mmap_fd, 0);
     if ((void *)data == (void *)(-1)) {
         // g_print("mmap file %s failed!\n",idxfilename);
+        size = 0u;
         data = nullptr;
         return false;
     }

src/sdcv.cpp

@@ -186,10 +186,13 @@ try {
         }
 
         // add bookname to list
-        gchar **p = get_impl(use_dict_list);
-        while (*p) {
-            order_list.push_back(bookname_to_ifo.at(*p));
-            ++p;
+        for (gchar **p = get_impl(use_dict_list); *p != nullptr; ++p) {
+            auto it = bookname_to_ifo.find(*p);
+            if (it != bookname_to_ifo.end()) {
+                order_list.push_back(it->second);
+            } else {
+                fprintf(stderr, _("Unknown dictionary: %s\n"), *p);
+            }
         }
     } else {
         std::string ordering_cfg_file = std::string(g_get_user_config_dir()) + G_DIR_SEPARATOR_S "sdcv_ordering";
@@ -201,7 +204,12 @@ try {
         if (ordering_file != nullptr) {
             std::string line;
             while (stdio_getline(ordering_file, line)) {
-                order_list.push_back(bookname_to_ifo.at(line));
+                auto it = bookname_to_ifo.find(line);
+                if (it != bookname_to_ifo.end()) {
+                    order_list.push_back(it->second);
+                } else {
+                    fprintf(stderr, _("Unknown dictionary: %s\n"), line.c_str());
+                }
             }
             fclose(ordering_file);
         }

src/stardict_lib.cpp

@@ -429,7 +429,7 @@ public:
         if (idxfile)
             fclose(idxfile);
     }
-    bool load(const std::string &url, gulong wc, gulong fsize, bool verbose) override;
+    bool load(const std::string &url, gulong wc, off_t fsize, bool verbose) override;
     const gchar *get_key(glong idx) override;
     void get_data(glong idx) override { get_key(idx); }
     const gchar *get_key_and_data(glong idx) override
@@ -489,7 +489,7 @@ public:
     {
     }
     ~WordListIndex() { g_free(idxdatabuf); }
-    bool load(const std::string &url, gulong wc, gulong fsize, bool verbose) override;
+    bool load(const std::string &url, gulong wc, off_t fsize, bool verbose) override;
     const gchar *get_key(glong idx) override { return wordlist[idx]; }
     void get_data(glong idx) override;
     const gchar *get_key_and_data(glong idx) override
@@ -615,7 +615,7 @@ bool OffsetIndex::save_cache(const std::string &url, bool verbose)
     return false;
 }
 
-bool OffsetIndex::load(const std::string &url, gulong wc, gulong fsize, bool verbose)
+bool OffsetIndex::load(const std::string &url, gulong wc, off_t fsize, bool verbose)
 {
     wordcount = wc;
     gulong npages = (wc - 1) / ENTR_PER_PAGE + 2;
@@ -758,7 +758,7 @@ bool OffsetIndex::lookup(const char *str, std::set<glong> &idxs, glong &next_idx
     return bFound;
 }
 
-bool WordListIndex::load(const std::string &url, gulong wc, gulong fsize, bool)
+bool WordListIndex::load(const std::string &url, gulong wc, off_t fsize, bool)
 {
     gzFile in = gzopen(url.c_str(), "rb");
     if (in == nullptr)
@@ -771,7 +771,7 @@ bool WordListIndex::load(const std::string &url, gulong wc, gulong fsize, bool)
     if (len < 0)
         return false;
 
-    if (gulong(len) != fsize)
+    if (static_cast<off_t>(len) != fsize)
         return false;
 
     wordlist.resize(wc + 1);
@@ -920,7 +920,7 @@ bool Dict::Lookup(const char *str, std::set<glong> &idxs, glong &next_idx)
 
 bool Dict::load(const std::string &ifofilename, bool verbose)
 {
-    gulong idxfilesize;
+    off_t idxfilesize;
     if (!load_ifofile(ifofilename, idxfilesize))
         return false;
 
@@ -964,7 +964,7 @@ bool Dict::load(const std::string &ifofilename, bool verbose)
     return true;
 }
 
-bool Dict::load_ifofile(const std::string &ifofilename, gulong &idxfilesize)
+bool Dict::load_ifofile(const std::string &ifofilename, off_t &idxfilesize)
 {
     DictInfo dict_info;
     if (!dict_info.load_from_ifo_file(ifofilename, false))

src/stardict_lib.hpp

@@ -77,8 +77,8 @@ struct DictInfo {
     std::string website;
     std::string date;
     std::string description;
-    guint32 index_file_size;
-    guint32 syn_file_size;
+    off_t index_file_size;
+    off_t syn_file_size;
     std::string sametypesequence;
 
     bool load_from_ifo_file(const std::string &ifofilename, bool istreedict);
@@ -91,7 +91,7 @@ public:
     guint32 wordentry_size;
 
     virtual ~IIndexFile() {}
-    virtual bool load(const std::string &url, gulong wc, gulong fsize, bool verbose) = 0;
+    virtual bool load(const std::string &url, gulong wc, off_t fsize, bool verbose) = 0;
     virtual const gchar *get_key(glong idx) = 0;
     virtual void get_data(glong idx) = 0;
     virtual const gchar *get_key_and_data(glong idx) = 0;
@@ -160,7 +160,7 @@ private:
     std::unique_ptr<IIndexFile> idx_file;
     std::unique_ptr<SynFile> syn_file;
 
-    bool load_ifofile(const std::string &ifofilename, gulong &idxfilesize);
+    bool load_ifofile(const std::string &ifofilename, off_t &idxfilesize);
 };
 
 class Libs