Add check for empty <LabelText> in <ContentVersion>
[libdcp.git] / src / verify.cc
index c42a10d4b961522596b31c75a65e0b412c001502..42ee192119b85280ada83512167a01aaa2a4805d 100644 (file)
@@ -386,6 +386,12 @@ enum class VerifyAssetResult {
 static VerifyAssetResult
 verify_asset (shared_ptr<const DCP> dcp, shared_ptr<const ReelFileAsset> reel_file_asset, function<void (float)> progress)
 {
+       /* When reading the DCP the hash will have been set to the one from the PKL/CPL.
+        * We want to calculate the hash of the actual file contents here, so that we
+        * can check it.  unset_hash() means that this calculation will happen on the
+        * call to hash().
+        */
+       reel_file_asset->asset_ref()->unset_hash();
        auto const actual_hash = reel_file_asset->asset_ref()->hash(progress);
 
        auto pkls = dcp->pkls();
@@ -849,16 +855,18 @@ verify_closed_caption_asset (
 }
 
 
-/** Check the timing of the individual subtitles and make sure there are no empty <Text> nodes */
+/** Check the timing of the individual subtitles and make sure there are no empty <Text> nodes etc. */
 static
 void
 verify_text_details (
+       dcp::Standard standard,
        vector<shared_ptr<Reel>> reels,
        int edit_rate,
        vector<VerificationNote>& notes,
        std::function<bool (shared_ptr<Reel>)> check,
        std::function<optional<string> (shared_ptr<Reel>)> xml,
-       std::function<int64_t (shared_ptr<Reel>)> duration
+       std::function<int64_t (shared_ptr<Reel>)> duration,
+       std::function<std::string (shared_ptr<Reel>)> id
        )
 {
        /* end of last subtitle (in editable units) */
@@ -870,9 +878,19 @@ verify_text_details (
        auto empty_text = false;
        /* current reel start time (in editable units) */
        int64_t reel_offset = 0;
-
-       std::function<void (cxml::ConstNodePtr, optional<int>, optional<Time>, int, bool)> parse;
-       parse = [&parse, &last_out, &too_short, &too_close, &too_early, &empty_text, &reel_offset](cxml::ConstNodePtr node, optional<int> tcr, optional<Time> start_time, int er, bool first_reel) {
+       optional<string> missing_load_font_id;
+
+       std::function<void (cxml::ConstNodePtr, optional<int>, optional<Time>, int, bool, bool&, vector<string>&)> parse;
+
+       parse = [&parse, &last_out, &too_short, &too_close, &too_early, &empty_text, &reel_offset, &missing_load_font_id](
+               cxml::ConstNodePtr node,
+               optional<int> tcr,
+               optional<Time> start_time,
+               int er,
+               bool first_reel,
+               bool& has_text,
+               vector<string>& font_ids
+               ) {
                if (node->name() == "Subtitle") {
                        Time in (node->string_attribute("TimeIn"), tcr);
                        if (start_time) {
@@ -912,10 +930,22 @@ verify_text_details (
                        if (!node_has_content(node)) {
                                empty_text = true;
                        }
+                       has_text = true;
+               } else if (node->name() == "LoadFont") {
+                       if (auto const id = node->optional_string_attribute("Id")) {
+                               font_ids.push_back(*id);
+                       } else if (auto const id = node->optional_string_attribute("ID")) {
+                               font_ids.push_back(*id);
+                       }
+               } else if (node->name() == "Font") {
+                       if (auto const font_id = node->optional_string_attribute("Id")) {
+                               if (std::find_if(font_ids.begin(), font_ids.end(), [font_id](string const& id) { return id == font_id; }) == font_ids.end()) {
+                                       missing_load_font_id = font_id;
+                               }
+                       }
                }
-
                for (auto i: node->node_children()) {
-                       parse(i, tcr, start_time, er, first_reel);
+                       parse(i, tcr, start_time, er, first_reel, has_text, font_ids);
                }
        };
 
@@ -937,24 +967,32 @@ verify_text_details (
                shared_ptr<cxml::Document> doc;
                optional<int> tcr;
                optional<Time> start_time;
-               try {
+               switch (standard) {
+               case dcp::Standard::INTEROP:
+                       doc = make_shared<cxml::Document>("DCSubtitle");
+                       doc->read_string (*reel_xml);
+                       break;
+               case dcp::Standard::SMPTE:
                        doc = make_shared<cxml::Document>("SubtitleReel");
                        doc->read_string (*reel_xml);
                        tcr = doc->number_child<int>("TimeCodeRate");
-                       auto start_time_string = doc->optional_string_child("StartTime");
-                       if (start_time_string) {
+                       if (auto start_time_string = doc->optional_string_child("StartTime")) {
                                start_time = Time(*start_time_string, tcr);
                        }
-               } catch (...) {
-                       doc = make_shared<cxml::Document>("DCSubtitle");
-                       doc->read_string (*reel_xml);
+                       break;
                }
-               parse (doc, tcr, start_time, edit_rate, i == 0);
+               bool has_text = false;
+               vector<string> font_ids;
+               parse(doc, tcr, start_time, edit_rate, i == 0, has_text, font_ids);
                auto end = reel_offset + duration(reels[i]);
                if (last_out && *last_out > end) {
                        reel_overlap = true;
                }
                reel_offset = end;
+
+               if (standard == dcp::Standard::SMPTE && has_text && font_ids.empty()) {
+                       notes.push_back(dcp::VerificationNote(dcp::VerificationNote::Type::ERROR, VerificationNote::Code::MISSING_LOAD_FONT).set_id(id(reels[i])));
+               }
        }
 
        if (last_out && *last_out > reel_offset) {
@@ -990,6 +1028,10 @@ verify_text_details (
                        VerificationNote::Type::WARNING, VerificationNote::Code::EMPTY_TEXT
                });
        }
+
+       if (missing_load_font_id) {
+               notes.push_back(dcp::VerificationNote(VerificationNote::Type::ERROR, VerificationNote::Code::MISSING_LOAD_FONT_FOR_FONT).set_id(*missing_load_font_id));
+       }
 }
 
 
@@ -1201,34 +1243,31 @@ verify_text_lines_and_characters (
 
 static
 void
-verify_text_details (vector<shared_ptr<Reel>> reels, vector<VerificationNote>& notes)
+verify_text_details(dcp::Standard standard, vector<shared_ptr<Reel>> reels, vector<VerificationNote>& notes)
 {
        if (reels.empty()) {
                return;
        }
 
        if (reels[0]->main_subtitle()) {
-               verify_text_details (reels, reels[0]->main_subtitle()->edit_rate().numerator, notes,
+               verify_text_details(standard, reels, reels[0]->main_subtitle()->edit_rate().numerator, notes,
                        [](shared_ptr<Reel> reel) {
                                return static_cast<bool>(reel->main_subtitle());
                        },
                        [](shared_ptr<Reel> reel) {
-                               auto interop = dynamic_pointer_cast<ReelInteropSubtitleAsset>(reel->main_subtitle());
-                               if (interop) {
-                                       return interop->asset()->raw_xml();
-                               }
-                               auto smpte = dynamic_pointer_cast<ReelSMPTESubtitleAsset>(reel->main_subtitle());
-                               DCP_ASSERT (smpte);
-                               return smpte->asset()->raw_xml();
+                               return reel->main_subtitle()->asset()->raw_xml();
                        },
                        [](shared_ptr<Reel> reel) {
                                return reel->main_subtitle()->actual_duration();
+                       },
+                       [](shared_ptr<Reel> reel) {
+                               return reel->main_subtitle()->id();
                        }
                );
        }
 
        for (auto i = 0U; i < reels[0]->closed_captions().size(); ++i) {
-               verify_text_details (reels, reels[0]->closed_captions()[i]->edit_rate().numerator, notes,
+               verify_text_details(standard, reels, reels[0]->closed_captions()[i]->edit_rate().numerator, notes,
                        [i](shared_ptr<Reel> reel) {
                                return i < reel->closed_captions().size();
                        },
@@ -1237,6 +1276,9 @@ verify_text_details (vector<shared_ptr<Reel>> reels, vector<VerificationNote>& n
                        },
                        [i](shared_ptr<Reel> reel) {
                                return reel->closed_captions()[i]->actual_duration();
+                       },
+                       [i](shared_ptr<Reel> reel) {
+                               return reel->closed_captions()[i]->id();
                        }
                );
        }
@@ -1413,6 +1455,7 @@ verify_reel(
                                }
                        }
                }
+
        }
 
        if (reel->main_sound() && reel->main_sound()->asset_ref().resolved()) {
@@ -1505,6 +1548,15 @@ verify_cpl(
                }
        }
 
+       for (auto version: cpl->content_versions()) {
+               if (version.label_text.empty()) {
+                       notes.push_back(
+                               dcp::VerificationNote(VerificationNote::Type::WARNING, VerificationNote::Code::EMPTY_CONTENT_VERSION_LABEL_TEXT, cpl->file().get()).set_id(cpl->id())
+                               );
+                       break;
+               }
+       }
+
        if (dcp->standard() == Standard::SMPTE) {
                if (!cpl->annotation_text()) {
                        notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISSING_CPL_ANNOTATION_TEXT, cpl->id(), cpl->file().get()});
@@ -1591,7 +1643,7 @@ verify_cpl(
                        );
        }
 
-       verify_text_details(cpl->reels(), notes);
+       verify_text_details(dcp->standard().get_value_or(dcp::Standard::SMPTE), cpl->reels(), notes);
 
        if (dcp->standard() == Standard::SMPTE) {
                if (auto msc = cpl->main_sound_configuration()) {
@@ -2042,6 +2094,14 @@ dcp::note_to_string (VerificationNote note)
                        );
        case VerificationNote::Code::INCORRECT_SUBTITLE_NAMESPACE_COUNT:
                return String::compose("The XML in the subtitle asset %1 has more than one namespace declaration.", note.note().get());
+       case VerificationNote::Code::MISSING_LOAD_FONT_FOR_FONT:
+               return String::compose("A subtitle or closed caption refers to a font with ID %1 that does not have a corresponding <LoadFont> node", note.id().get());
+       case VerificationNote::Code::MISSING_LOAD_FONT:
+               return String::compose("The SMPTE subtitle asset %1 has <Text> nodes but no <LoadFont> node", note.id().get());
+       case VerificationNote::Code::MISMATCHED_ASSET_MAP_ID:
+               return String::compose("The asset with ID %1 in the asset map actually has an id of %2", note.id().get(), note.other_id().get());
+       case VerificationNote::Code::EMPTY_CONTENT_VERSION_LABEL_TEXT:
+               return String::compose("The <LabelText> in a <ContentVersion> in CPL %1 is empty", note.id().get());
        }
 
        return "";