Tidy up careful_string_filter and add some extra transliterations.
[dcpomatic.git] / src / lib / util.cc
index 8a039764d0a77f4bec576e579b4b801c2f211ba3..7d0d2bf606c698d3a2a4881f2336c4e156456899 100644 (file)
  *  @brief Some utility functions and classes.
  */
 
-#include "util.h"
-#include "exceptions.h"
-#include "dcp_content_type.h"
-#include "filter.h"
+
+#define UNICODE 1
+
+
+#include "audio_buffers.h"
+#include "audio_processor.h"
 #include "cinema_sound_processor.h"
+#include "compose.hpp"
 #include "config.h"
-#include "ratio.h"
-#include "job.h"
 #include "cross.h"
-#include "video_content.h"
-#include "rect.h"
-#include "digester.h"
-#include "audio_processor.h"
 #include "crypto.h"
-#include "compose.hpp"
-#include "audio_buffers.h"
-#include "string_text.h"
-#include "font.h"
-#include "render_text.h"
+#include "dcp_content_type.h"
+#include "digester.h"
+#include "exceptions.h"
 #include "ffmpeg_image_proxy.h"
+#include "filter.h"
+#include "font.h"
 #include "image.h"
-#include "text_decoder.h"
+#include "job.h"
 #include "job_manager.h"
-#include "warnings.h"
+#include "ratio.h"
+#include "rect.h"
+#include "render_text.h"
+#include "string_text.h"
+#include "text_decoder.h"
+#include "util.h"
+#include "video_content.h"
+#include <dcp/atmos_asset.h>
 #include <dcp/decrypted_kdm.h>
 #include <dcp/locale_convert.h>
-#include <dcp/util.h>
-#include <dcp/raw_convert.h>
 #include <dcp/picture_asset.h>
+#include <dcp/raw_convert.h>
 #include <dcp/sound_asset.h>
 #include <dcp/subtitle_asset.h>
-#include <dcp/atmos_asset.h>
+#include <dcp/util.h>
+#include <dcp/warnings.h>
+LIBDCP_DISABLE_WARNINGS
 extern "C" {
 #include <libavfilter/avfilter.h>
 #include <libavformat/avformat.h>
 #include <libavcodec/avcodec.h>
 }
+LIBDCP_ENABLE_WARNINGS
 #include <curl/curl.h>
 #include <glib.h>
 #include <pangomm/init.h>
@@ -69,9 +75,9 @@ extern "C" {
 #include <boost/range/algorithm/replace_if.hpp>
 #include <boost/thread.hpp>
 #include <boost/filesystem.hpp>
-DCPOMATIC_DISABLE_WARNINGS
+LIBDCP_DISABLE_WARNINGS
 #include <boost/locale.hpp>
-DCPOMATIC_ENABLE_WARNINGS
+LIBDCP_ENABLE_WARNINGS
 #ifdef DCPOMATIC_WINDOWS
 #include <boost/locale.hpp>
 #include <dbghelp.h>
@@ -89,25 +95,23 @@ DCPOMATIC_ENABLE_WARNINGS
 
 #include "i18n.h"
 
-using std::string;
-using std::wstring;
-using std::setfill;
-using std::ostream;
+
+using std::bad_alloc;
+using std::cout;
 using std::endl;
-using std::vector;
-using std::min;
-using std::max;
-using std::map;
-using std::list;
-using std::multimap;
 using std::istream;
+using std::list;
+using std::make_pair;
+using std::make_shared;
+using std::map;
+using std::min;
+using std::ostream;
 using std::pair;
-using std::cout;
-using std::bad_alloc;
 using std::set_terminate;
-using std::make_pair;
 using std::shared_ptr;
-using std::make_shared;
+using std::string;
+using std::vector;
+using std::wstring;
 using boost::thread;
 using boost::optional;
 using boost::lexical_cast;
@@ -118,6 +122,7 @@ using dcp::raw_convert;
 using dcp::locale_convert;
 using namespace dcpomatic;
 
+
 /** Path to our executable, required by the stacktrace stuff and filled
  *  in during App::onInit().
  */
@@ -237,7 +242,7 @@ addr2line (void const * const addr)
        return system(addr2line_cmd);
 }
 
-DCPOMATIC_DISABLE_WARNINGS
+LIBDCP_DISABLE_WARNINGS
 /** This is called when C signals occur on Windows (e.g. SIGSEGV)
  *  (NOT C++ exceptions!).  We write a backtrace to backtrace_file by dark means.
  *  Adapted from code here: http://spin.atomicobject.com/2013/01/13/exceptions-stack-traces-c/
@@ -245,9 +250,11 @@ DCPOMATIC_DISABLE_WARNINGS
 LONG WINAPI
 exception_handler(struct _EXCEPTION_POINTERS * info)
 {
-       auto f = fopen_boost (backtrace_file, "w");
-       fprintf (f, "C-style exception %d\n", info->ExceptionRecord->ExceptionCode);
-       fclose(f);
+       dcp::File f(backtrace_file, "w");
+       if (f) {
+               fprintf(f.get(), "C-style exception %d\n", info->ExceptionRecord->ExceptionCode);
+               f.close();
+       }
 
        if (info->ExceptionRecord->ExceptionCode != EXCEPTION_STACK_OVERFLOW) {
                CONTEXT* context = info->ContextRecord;
@@ -294,7 +301,7 @@ exception_handler(struct _EXCEPTION_POINTERS * info)
 
        return EXCEPTION_CONTINUE_SEARCH;
 }
-DCPOMATIC_ENABLE_WARNINGS
+LIBDCP_ENABLE_WARNINGS
 #endif
 
 void
@@ -365,10 +372,12 @@ dcpomatic_setup ()
        SetUnhandledExceptionFilter(exception_handler);
 #endif
 
-DCPOMATIC_DISABLE_WARNINGS
+#ifdef DCPOMATIC_HAVE_AVREGISTER
+LIBDCP_DISABLE_WARNINGS
        av_register_all ();
        avfilter_register_all ();
-DCPOMATIC_ENABLE_WARNINGS
+LIBDCP_ENABLE_WARNINGS
+#endif
 
 #ifdef DCPOMATIC_OSX
        /* Add our library directory to the libltdl search path so that
@@ -392,14 +401,14 @@ DCPOMATIC_ENABLE_WARNINGS
 #endif
 
        Pango::init ();
-       dcp::init (tags_path());
+       dcp::init (libdcp_resources_path());
 
 #if defined(DCPOMATIC_WINDOWS) || defined(DCPOMATIC_OSX)
        /* Render something to fontconfig to create its cache */
        list<StringText> subs;
        dcp::SubtitleString ss(
                optional<string>(), false, false, false, dcp::Colour(), 42, 1, dcp::Time(), dcp::Time(), 0, dcp::HAlign::CENTER, 0, dcp::VAlign::CENTER, dcp::Direction::LTR,
-               "Hello dolly", dcp::Effect::NONE, dcp::Colour(), dcp::Time(), dcp::Time()
+               "Hello dolly", dcp::Effect::NONE, dcp::Colour(), dcp::Time(), dcp::Time(), 0
                );
        subs.push_back (StringText(ss, 0));
        render_text (subs, list<shared_ptr<Font>>(), dcp::Size(640, 480), DCPTime(), 24);
@@ -483,16 +492,15 @@ digest_head_tail (vector<boost::filesystem::path> files, boost::uintmax_t size)
        char* p = buffer.get ();
        int i = 0;
        while (i < int64_t (files.size()) && to_do > 0) {
-               auto f = fopen_boost (files[i], "rb");
+               dcp::File f(files[i], "rb");
                if (!f) {
                        throw OpenFileError (files[i].string(), errno, OpenFileError::READ);
                }
 
                boost::uintmax_t this_time = min (to_do, boost::filesystem::file_size (files[i]));
-               checked_fread (p, this_time, f, files[i]);
+               f.checked_read(p, this_time);
                p += this_time;
                to_do -= this_time;
-               fclose (f);
 
                ++i;
        }
@@ -503,17 +511,16 @@ digest_head_tail (vector<boost::filesystem::path> files, boost::uintmax_t size)
        p = buffer.get ();
        i = files.size() - 1;
        while (i >= 0 && to_do > 0) {
-               auto f = fopen_boost (files[i], "rb");
+               dcp::File f(files[i], "rb");
                if (!f) {
                        throw OpenFileError (files[i].string(), errno, OpenFileError::READ);
                }
 
                boost::uintmax_t this_time = min (to_do, boost::filesystem::file_size (files[i]));
-               dcpomatic_fseek (f, -this_time, SEEK_END);
-               checked_fread (p, this_time, f, files[i]);
+               f.seek(-this_time, SEEK_END);
+               f.checked_read(p, this_time);
                p += this_time;
                to_do -= this_time;
-               fclose (f);
 
                --i;
        }
@@ -522,19 +529,14 @@ digest_head_tail (vector<boost::filesystem::path> files, boost::uintmax_t size)
        return digester.get ();
 }
 
-/** Round a number up to the nearest multiple of another number.
- *  @param c Index.
- *  @param stride Array of numbers to round, indexed by c.
- *  @param t Multiple to round to.
- *  @return Rounded number.
- */
-int
-stride_round_up (int c, int const * stride, int t)
+
+string
+simple_digest (vector<boost::filesystem::path> paths)
 {
-       int const a = stride[c] + (t - 1);
-       return a - (a % t);
+       return digest_head_tail(paths, 1000000) + raw_convert<string>(boost::filesystem::file_size(paths.front()));
 }
 
+
 /** Trip an assert if the caller is not in the UI thread */
 void
 ensure_ui_thread ()
@@ -595,7 +597,7 @@ short_audio_channel_name (int c)
                _("BsR"),
                _("DBP"),
                _("DBS"),
-               "",
+               _("Sign"),
                ""
        };
 
@@ -657,16 +659,6 @@ fit_ratio_within (float ratio, dcp::Size full_frame)
        return dcp::Size (full_frame.width, lrintf (full_frame.width / ratio));
 }
 
-void *
-wrapped_av_malloc (size_t s)
-{
-       auto p = av_malloc (s);
-       if (!p) {
-               throw bad_alloc ();
-       }
-       return p;
-}
-
 map<string, string>
 split_get_request (string url)
 {
@@ -775,19 +767,32 @@ careful_string_filter (string s)
        */
 
        /* First transliterate using libicu to try to remove accents in a "nice" way */
-       auto icu_utf16 = icu::UnicodeString::fromUTF8(icu::StringPiece(s));
+       auto transliterated = icu::UnicodeString::fromUTF8(icu::StringPiece(s));
        auto status = U_ZERO_ERROR;
        auto transliterator = icu::Transliterator::createInstance("NFD; [:M:] Remove; NFC", UTRANS_FORWARD, status);
-       transliterator->transliterate(icu_utf16);
-       s.clear ();
-       icu_utf16.toUTF8String(s);
+       transliterator->transliterate(transliterated);
+
+       /* Some things are missed by ICU's transliterator */
+       std::map<wchar_t, wchar_t> replacements = {
+               { L'ł',         L'l' },
+               { L'Ł',         L'L' }
+       };
+
+       icu::UnicodeString transliterated_more;
+       for (int i = 0; i < transliterated.length(); ++i) {
+               auto replacement = replacements.find(transliterated[i]);
+               if (replacement != replacements.end()) {
+                       transliterated_more += replacement->second;
+               } else {
+                       transliterated_more += transliterated[i];
+               }
+       }
 
        /* Then remove anything that's not in a very limited character set */
-       wstring ws = boost::locale::conv::utf_to_utf<wchar_t>(s);
-       string out;
-       string const allowed = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz-_%.+";
-       for (size_t i = 0; i < ws.size(); ++i) {
-               wchar_t c = ws[i];
+       wstring out;
+       wstring const allowed = L"0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz-_%.+";
+       for (size_t i = 0; i < transliterated_more.length(); ++i) {
+               wchar_t c = transliterated_more[i];
                if (allowed.find(c) != string::npos) {
                        out += c;
                }
@@ -872,35 +877,6 @@ increment_eyes (Eyes e)
        return Eyes::LEFT;
 }
 
-void
-checked_fwrite (void const * ptr, size_t size, FILE* stream, boost::filesystem::path path)
-{
-       size_t N = fwrite (ptr, 1, size, stream);
-       if (N != size) {
-               if (ferror(stream)) {
-                       fclose (stream);
-                       throw FileError (String::compose("fwrite error %1", errno), path);
-               } else {
-                       fclose (stream);
-                       throw FileError ("Unexpected short write", path);
-               }
-       }
-}
-
-void
-checked_fread (void* ptr, size_t size, FILE* stream, boost::filesystem::path path)
-{
-       size_t N = fread (ptr, 1, size, stream);
-       if (N != size) {
-               if (ferror(stream)) {
-                       fclose (stream);
-                       throw FileError (String::compose("fread error %1", errno), path);
-               } else {
-                       fclose (stream);
-                       throw FileError ("Unexpected short read", path);
-               }
-       }
-}
 
 size_t
 utf8_strlen (string s)
@@ -949,8 +925,8 @@ void
 emit_subtitle_image (ContentTimePeriod period, dcp::SubtitleImage sub, dcp::Size size, shared_ptr<TextDecoder> decoder)
 {
        /* XXX: this is rather inefficient; decoding the image just to get its size */
-       FFmpegImageProxy proxy (sub.png_image(), VideoRange::FULL);
-       auto image = proxy.image().image;
+       FFmpegImageProxy proxy (sub.png_image());
+       auto image = proxy.image(Image::Alignment::PADDED).image;
        /* set up rect with height and width */
        dcpomatic::Rect<double> rect(0, 0, image->size().width / double(size.width), image->size().height / double(size.height));
 
@@ -1040,71 +1016,45 @@ show_jobs_on_console (bool progress)
 
 /** XXX: could use mmap? */
 void
-copy_in_bits (boost::filesystem::path from, boost::filesystem::path to, boost::function<void (float)> progress)
+copy_in_bits (boost::filesystem::path from, boost::filesystem::path to, std::function<void (float)> progress)
 {
-       auto f = fopen_boost (from, "rb");
+       dcp::File f(from, "rb");
        if (!f) {
                throw OpenFileError (from, errno, OpenFileError::READ);
        }
-       auto t = fopen_boost (to, "wb");
+       dcp::File t(to, "wb");
        if (!t) {
-               fclose (f);
                throw OpenFileError (to, errno, OpenFileError::WRITE);
        }
 
        /* on the order of a second's worth of copying */
        boost::uintmax_t const chunk = 20 * 1024 * 1024;
 
-       auto buffer = static_cast<uint8_t*> (malloc(chunk));
-       if (!buffer) {
-               throw std::bad_alloc ();
-       }
+       std::vector<uint8_t> buffer(chunk);
 
        boost::uintmax_t const total = boost::filesystem::file_size (from);
        boost::uintmax_t remaining = total;
 
        while (remaining) {
                boost::uintmax_t this_time = min (chunk, remaining);
-               size_t N = fread (buffer, 1, chunk, f);
+               size_t N = f.read(buffer.data(), 1, chunk);
                if (N < this_time) {
-                       fclose (f);
-                       fclose (t);
-                       free (buffer);
                        throw ReadFileError (from, errno);
                }
 
-               N = fwrite (buffer, 1, this_time, t);
+               N = t.write(buffer.data(), 1, this_time);
                if (N < this_time) {
-                       fclose (f);
-                       fclose (t);
-                       free (buffer);
                        throw WriteFileError (to, errno);
                }
 
                progress (1 - float(remaining) / total);
                remaining -= this_time;
        }
-
-       fclose (f);
-       fclose (t);
-       free (buffer);
-}
-
-double
-db_to_linear (double db)
-{
-       return pow(10, db / 20);
-}
-
-double
-linear_to_db (double linear)
-{
-       return 20 * log10(linear);
 }
 
 
 dcp::Size
-scale_for_display (dcp::Size s, dcp::Size display_container, dcp::Size film_container)
+scale_for_display (dcp::Size s, dcp::Size display_container, dcp::Size film_container, PixelQuanta quanta)
 {
        /* Now scale it down if the display container is smaller than the film container */
        if (display_container != film_container) {
@@ -1115,6 +1065,7 @@ scale_for_display (dcp::Size s, dcp::Size display_container, dcp::Size film_cont
 
                s.width = lrintf (s.width * scale);
                s.height = lrintf (s.height * scale);
+               s = quanta.round (s);
        }
 
        return s;