diff options
-rw-r--r-- | CMakeLists.txt | 2 | ||||
-rw-r--r-- | cards_filter.rb | 41 | ||||
-rw-r--r-- | cardset.cpp | 61 | ||||
-rw-r--r-- | cardset.h | 6 | ||||
-rw-r--r-- | wizard.cpp | 243 |
5 files changed, 202 insertions, 151 deletions
diff --git a/CMakeLists.txt b/CMakeLists.txt index 82e9d64..acc7ea3 100644 --- a/CMakeLists.txt +++ b/CMakeLists.txt | |||
@@ -17,6 +17,7 @@ include_directories( | |||
17 | vendor/asio/asio/include | 17 | vendor/asio/asio/include |
18 | vendor/base64/include | 18 | vendor/base64/include |
19 | ${tesseract_INCLUDE_DIRS} | 19 | ${tesseract_INCLUDE_DIRS} |
20 | #/Users/hatkirby/Repos/tesseract/include | ||
20 | ${lept_INCLUDE_DIRS} | 21 | ${lept_INCLUDE_DIRS} |
21 | ${GraphicsMagick_INCLUDE_DIRS} | 22 | ${GraphicsMagick_INCLUDE_DIRS} |
22 | ${curlcpp_INCLUDE_DIRS} | 23 | ${curlcpp_INCLUDE_DIRS} |
@@ -26,6 +27,7 @@ include_directories( | |||
26 | 27 | ||
27 | link_directories( | 28 | link_directories( |
28 | ${tesseract_LIBRARY_DIRS} | 29 | ${tesseract_LIBRARY_DIRS} |
30 | #/Users/hatkirby/Repos/tesseract/build/here/usr/local/lib | ||
29 | ${lept_LIBRARY_DIRS} | 31 | ${lept_LIBRARY_DIRS} |
30 | ${GraphicsMagick_LIBRARY_DIRS} | 32 | ${GraphicsMagick_LIBRARY_DIRS} |
31 | ${curlcpp_LIBRARY_DIRS} | 33 | ${curlcpp_LIBRARY_DIRS} |
diff --git a/cards_filter.rb b/cards_filter.rb new file mode 100644 index 0000000..aba66e3 --- /dev/null +++ b/cards_filter.rb | |||
@@ -0,0 +1,41 @@ | |||
1 | require 'json' | ||
2 | |||
3 | input = File.open(ARGV[0]) do |file| | ||
4 | JSON.load(file) | ||
5 | end | ||
6 | |||
7 | filtered = input.select do |cardJson| | ||
8 | cardJson["object"] == "card" && | ||
9 | # It needs to have a downloadable image | ||
10 | cardJson.count("image_uris") && | ||
11 | # Make sure we can support the card layout | ||
12 | ["normal", "leveler", "saga"].include?(cardJson["layout"]) && | ||
13 | # We only support modern and m15 frames | ||
14 | ["2015", "2003"].include?(cardJson["frame"]) && | ||
15 | # Digital cards look slightly different so ignore them | ||
16 | !cardJson["digital"] && | ||
17 | # Only use english printings | ||
18 | cardJson["lang"] == "en" && | ||
19 | # Currently not supporting silver bordered cards | ||
20 | cardJson["border_color"] != "silver" && | ||
21 | # It is hard to read the name of a planeswalker | ||
22 | !cardJson["type_line"].include?("Planeswalker") && | ||
23 | # This cuts out checklists and special tokens | ||
24 | cardJson["type_line"] != "Card" && | ||
25 | # Amonkhet invocations are impossible | ||
26 | cardJson["set"] != "mp2" && | ||
27 | # Unknown Event is not a real thing huh | ||
28 | cardJson["set"] != "da1" | ||
29 | end.map do |cardJson| | ||
30 | { | ||
31 | "id" => cardJson["id"], | ||
32 | "name" => cardJson["name"], | ||
33 | "imageUri" => cardJson["image_uris"]["png"], | ||
34 | "frame" => cardJson["frame"], | ||
35 | "artist" => cardJson["artist"] | ||
36 | } | ||
37 | end | ||
38 | |||
39 | File.open(ARGV[1], 'w') do |file| | ||
40 | JSON.dump(filtered, file) | ||
41 | end | ||
diff --git a/cardset.cpp b/cardset.cpp index 4003eed..8097834 100644 --- a/cardset.cpp +++ b/cardset.cpp | |||
@@ -14,53 +14,28 @@ cardset::cardset(std::string filename) { | |||
14 | nlohmann::json cardsJson = nlohmann::json::parse(contents.str()); | 14 | nlohmann::json cardsJson = nlohmann::json::parse(contents.str()); |
15 | 15 | ||
16 | for (const auto& cardJson : cardsJson) { | 16 | for (const auto& cardJson : cardsJson) { |
17 | if ( | 17 | // We're pre-processing the cardset now to reduce the work done here (but |
18 | // The object needs to be a card | 18 | // mostly the space taken by the data). |
19 | cardJson["object"] == "card" && | 19 | card_frame frame; |
20 | // It needs to have a downloadable image | 20 | |
21 | cardJson.count("image_uris") && | 21 | if (cardJson["frame"] == "2015") { |
22 | // Make sure we can support the card layout | 22 | frame = card_frame::m2015; |
23 | (cardJson["layout"] == "normal" || cardJson["layout"] == "leveler" || | 23 | } else if (cardJson["frame"] == "2003") { |
24 | cardJson["layout"] == "saga") && | 24 | frame = card_frame::modern; |
25 | // Digital cards look slightly different so ignore them | 25 | } else { |
26 | !cardJson["digital"] && | 26 | continue; |
27 | // Only use english printings | 27 | } |
28 | cardJson["lang"] == "en" && | ||
29 | // Currently not supporting silver bordered cards | ||
30 | cardJson["border_color"] != "silver" && | ||
31 | // It is hard to read the name of a planeswalker | ||
32 | cardJson["type_line"].get<std::string>().find("Planeswalker") == | ||
33 | std::string::npos && | ||
34 | // This cuts out checklists and special tokens | ||
35 | cardJson["type_line"] != "Card" && | ||
36 | // Amonkhet invocations are impossible | ||
37 | cardJson["set"] != "mp2" && | ||
38 | // Unknown Event is not a real thing huh | ||
39 | cardJson["set"] != "da1" && | ||
40 | // Ignore cards with the special legendary flare | ||
41 | (!cardJson.count("frame_effects") || | ||
42 | !cardJson["frame_effects"].count("legendary"))) { | ||
43 | card_frame frame; | ||
44 | |||
45 | if (cardJson["frame"] == "2015") { | ||
46 | frame = card_frame::m2015; | ||
47 | } else if (cardJson["frame"] == "2003") { | ||
48 | frame = card_frame::modern; | ||
49 | } else { | ||
50 | continue; | ||
51 | } | ||
52 | 28 | ||
53 | size_t cardId = cards_.size(); | 29 | size_t cardId = cards_.size(); |
54 | cards_.emplace_back(cardId, cardJson["name"], | 30 | cards_.emplace_back(cardId, cardJson["name"], cardJson["imageUri"], frame, |
55 | cardJson["image_uris"]["png"], frame, cardJson["id"]); | 31 | cardJson["id"], cardJson["artist"]); |
56 | 32 | ||
57 | std::string canon = hatkirby::lowercase(cardJson["name"]); | 33 | std::string canon = hatkirby::lowercase(cardJson["name"]); |
58 | 34 | ||
59 | for (int i = 0; i < canon.length(); i++) { | 35 | for (int i = 0; i < canon.length(); i++) { |
60 | titles_.add(canon, {cardId, i}, i); | 36 | titles_.add(canon, {cardId, i}, i); |
61 | 37 | ||
62 | chars_.insert(canon.at(i)); | 38 | chars_.insert(canon.at(i)); |
63 | } | ||
64 | } | 39 | } |
65 | } | 40 | } |
66 | } | 41 | } |
diff --git a/cardset.h b/cardset.h index 217d859..2e39f5c 100644 --- a/cardset.h +++ b/cardset.h | |||
@@ -17,14 +17,16 @@ struct card { | |||
17 | std::string imageUri; | 17 | std::string imageUri; |
18 | card_frame frame; | 18 | card_frame frame; |
19 | std::string uuid; | 19 | std::string uuid; |
20 | std::string artist; | ||
20 | 21 | ||
21 | card(size_t id, std::string name, std::string imageUri, card_frame frame, | 22 | card(size_t id, std::string name, std::string imageUri, card_frame frame, |
22 | std::string uuid) | 23 | std::string uuid, std::string artist) |
23 | : id(id), | 24 | : id(id), |
24 | name(std::move(name)), | 25 | name(std::move(name)), |
25 | imageUri(std::move(imageUri)), | 26 | imageUri(std::move(imageUri)), |
26 | frame(frame), | 27 | frame(frame), |
27 | uuid(std::move(uuid)) {} | 28 | uuid(std::move(uuid)), |
29 | artist(std::move(artist)) {} | ||
28 | }; | 30 | }; |
29 | 31 | ||
30 | class cardset { | 32 | class cardset { |
diff --git a/wizard.cpp b/wizard.cpp index a1d7df4..c99ec79 100644 --- a/wizard.cpp +++ b/wizard.cpp | |||
@@ -55,6 +55,24 @@ void wizard::set_status_callback(status_callback_type callback) { | |||
55 | } | 55 | } |
56 | 56 | ||
57 | Magick::Image wizard::run() { | 57 | Magick::Image wizard::run() { |
58 | tesseract_ptr api{new tesseract::TessBaseAPI()}; | ||
59 | |||
60 | if (api->Init(nullptr, "eng", tesseract::OEM_TESSERACT_LSTM_COMBINED)) { | ||
61 | throw std::runtime_error("Could not initialize tesseract"); | ||
62 | } | ||
63 | |||
64 | std::string allChars; | ||
65 | for (char ch : cards_.getCharacters()) { | ||
66 | allChars.push_back(ch); | ||
67 | if (std::isalpha(ch)) { | ||
68 | allChars.push_back(std::toupper(ch)); | ||
69 | } | ||
70 | } | ||
71 | |||
72 | if (!api->SetVariable("tessedit_char_whitelist", allChars.c_str())) { | ||
73 | throw std::runtime_error("Could not set tesseract allowlist"); | ||
74 | } | ||
75 | |||
58 | std::string text = text_; //"what the heck, it's just some gay guy"; | 76 | std::string text = text_; //"what the heck, it's just some gay guy"; |
59 | // getline(std::cin, text); | 77 | // getline(std::cin, text); |
60 | if (text.back() == '\r') { | 78 | if (text.back() == '\r') { |
@@ -68,104 +86,109 @@ Magick::Image wizard::run() { | |||
68 | 86 | ||
69 | std::string canonText = hatkirby::lowercase(text); | 87 | std::string canonText = hatkirby::lowercase(text); |
70 | designer des(canonText, cards_.getTitles()); | 88 | designer des(canonText, cards_.getTitles()); |
71 | std::list<usage> res = des.generate(rng_); | ||
72 | 89 | ||
73 | Magick::Image endImage; | 90 | Magick::Image endImage; |
74 | bool firstSlice = false; | 91 | bool success = false; |
75 | int ui = 0; | 92 | for (int attempt = 0; attempt < 10; attempt++) { |
76 | 93 | endImage = {}; | |
77 | for (const usage& u : res) { | 94 | |
78 | const card& theCard = cards_.getCard(u.cardId); | 95 | std::list<usage> res = des.generate(rng_); |
79 | const std::string& cardName = theCard.name; | 96 | |
80 | 97 | bool firstSlice = false; | |
81 | std::cout << cardName.substr(0, u.strIndex) << "[" | 98 | int ui = 0; |
82 | << cardName.substr(u.strIndex, u.strLen) << "]" | 99 | |
83 | << cardName.substr(u.strIndex + u.strLen) << std::endl; | 100 | bool failure = false; |
84 | 101 | for (const usage& u : res) { | |
85 | std::cout << "Downloading image..." << std::endl; | 102 | const card& theCard = cards_.getCard(u.cardId); |
86 | if (status_callback_) { | 103 | const std::string& cardName = theCard.name; |
87 | status_callback_("Downloading image..."); | 104 | |
88 | } | 105 | std::cout << cardName.substr(0, u.strIndex) << "[" |
89 | 106 | << cardName.substr(u.strIndex, u.strLen) << "]" | |
90 | Magick::Image cardImg = images_.get(theCard.uuid, theCard.imageUri); | 107 | << cardName.substr(u.strIndex + u.strLen) << std::endl; |
91 | 108 | ||
92 | std::cout << "Reading text..." << std::endl; | 109 | std::cout << "Downloading image..." << std::endl; |
93 | if (status_callback_) { | 110 | if (status_callback_) { |
94 | status_callback_("Reading text..."); | 111 | std::ostringstream status_stream; |
95 | } | 112 | status_stream << "Processing card " << (ui + 1) << "/" << res.size() |
96 | 113 | << " (attempt " << (attempt + 1) << ")..."; | |
97 | Magick::Image titleImg = cardImg; | 114 | status_callback_(status_stream.str()); |
98 | titleImg.magick("TIFF"); | 115 | } |
99 | |||
100 | // titleImg.threshold(MaxRGB / 2); | ||
101 | titleImg.write("pre.tif"); | ||
102 | |||
103 | Magick::Geometry margin; | ||
104 | |||
105 | if (theCard.frame == card_frame::m2015) { | ||
106 | margin = Magick::Geometry{595, 46, 57, 54}; | ||
107 | } else if (theCard.frame == card_frame::modern) { | ||
108 | margin = Magick::Geometry{581, 50, 63, 57}; | ||
109 | } | ||
110 | |||
111 | titleImg.crop(margin); | ||
112 | titleImg.zoom({margin.width() * 5, margin.height() * 5}); | ||
113 | |||
114 | // titleImg.quantizeColorSpace(Magick::GRAYColorspace); | ||
115 | // titleImg.quantizeColors(2); | ||
116 | // titleImg.quantize(); | ||
117 | titleImg.backgroundColor("white"); | ||
118 | titleImg.matte(false); | ||
119 | titleImg.resolutionUnits(Magick::PixelsPerInchResolution); | ||
120 | titleImg.density({300, 300}); | ||
121 | titleImg.type(Magick::GrayscaleType); | ||
122 | 116 | ||
123 | titleImg.write("title.tif"); | 117 | Magick::Image cardImg = images_.get(theCard.uuid, theCard.imageUri); |
124 | 118 | ||
125 | Magick::Blob titleBlob; | 119 | std::cout << "Reading text..." << std::endl; |
126 | titleImg.write(&titleBlob); | ||
127 | 120 | ||
128 | pix_ptr titlePix{ | 121 | Magick::Image titleImg = cardImg; |
129 | pixReadMemTiff(reinterpret_cast<const unsigned char*>(titleBlob.data()), | 122 | titleImg.magick("TIFF"); |
130 | titleBlob.length(), 0)}; | ||
131 | 123 | ||
132 | tesseract_ptr api{new tesseract::TessBaseAPI()}; | 124 | // titleImg.threshold(MaxRGB / 2); |
125 | titleImg.write("pre.tif"); | ||
133 | 126 | ||
134 | if (api->Init(nullptr, "eng")) { | 127 | Magick::Geometry margin; |
135 | throw std::runtime_error("Could not initialize tesseract"); | ||
136 | } | ||
137 | 128 | ||
138 | api->SetImage(titlePix.get()); | 129 | if (theCard.frame == card_frame::m2015) { |
139 | api->Recognize(nullptr); | 130 | margin = Magick::Geometry{595, 50, 57, 54}; |
131 | } else if (theCard.frame == card_frame::modern) { | ||
132 | margin = Magick::Geometry{581, 50, 63, 57}; | ||
133 | } | ||
140 | 134 | ||
141 | tesseract::ResultIterator* ri = api->GetIterator(); | 135 | titleImg.crop(margin); |
142 | tesseract::PageIteratorLevel level = tesseract::RIL_TEXTLINE; | 136 | titleImg.zoom({margin.width() * 5, margin.height() * 5}); |
143 | bool foundName = false; | 137 | |
144 | size_t extraChars = 0; | 138 | // titleImg.quantizeColorSpace(Magick::GRAYColorspace); |
139 | // titleImg.quantizeColors(2); | ||
140 | // titleImg.quantize(); | ||
141 | titleImg.backgroundColor("white"); | ||
142 | titleImg.matte(false); | ||
143 | titleImg.resolutionUnits(Magick::PixelsPerInchResolution); | ||
144 | titleImg.density({300, 300}); | ||
145 | titleImg.type(Magick::GrayscaleType); | ||
146 | titleImg.gamma(3); | ||
147 | // titleImg.level(0.2, 0.8); | ||
148 | |||
149 | titleImg.write("title.tif"); | ||
150 | |||
151 | Magick::Blob titleBlob; | ||
152 | titleImg.write(&titleBlob); | ||
153 | |||
154 | pix_ptr titlePix{pixReadMemTiff( | ||
155 | reinterpret_cast<const unsigned char*>(titleBlob.data()), | ||
156 | titleBlob.length(), 0)}; | ||
157 | |||
158 | api->SetImage(titlePix.get()); | ||
159 | api->Recognize(nullptr); | ||
160 | |||
161 | tesseract::ResultIterator* ri = api->GetIterator(); | ||
162 | tesseract::PageIteratorLevel level = tesseract::RIL_TEXTLINE; | ||
163 | bool foundName = false; | ||
164 | size_t extraChars = 0; | ||
165 | |||
166 | if (ri) { | ||
167 | do { | ||
168 | const char* line = ri->GetUTF8Text(level); | ||
169 | |||
170 | if (line) { | ||
171 | std::string lineStr(line); | ||
172 | |||
173 | size_t leadin = hatkirby::lowercase(lineStr).find( | ||
174 | hatkirby::lowercase((cardName))); | ||
175 | if (leadin != std::string::npos) { | ||
176 | foundName = true; | ||
177 | extraChars = leadin; | ||
178 | |||
179 | break; | ||
180 | } /* else { | ||
181 | std::cout << "WRONG: " << lineStr << std::endl; | ||
182 | }*/ | ||
183 | } | ||
184 | } while (ri->Next(level)); | ||
185 | } | ||
145 | 186 | ||
146 | if (ri) { | 187 | if (!foundName) { |
147 | do { | 188 | failure = true; |
148 | const char* line = ri->GetUTF8Text(level); | 189 | break; |
149 | 190 | } | |
150 | if (line) { | ||
151 | std::string lineStr(line); | ||
152 | |||
153 | size_t leadin = hatkirby::lowercase(lineStr).find( | ||
154 | hatkirby::lowercase((cardName))); | ||
155 | // if (leadin != std::string::npos) | ||
156 | { | ||
157 | foundName = true; | ||
158 | // extraChars = leadin; | ||
159 | |||
160 | break; | ||
161 | } /* else { | ||
162 | std::cout << "WRONG: " << lineStr << std::endl; | ||
163 | }*/ | ||
164 | } | ||
165 | } while (ri->Next(level)); | ||
166 | } | ||
167 | 191 | ||
168 | if (foundName) { | ||
169 | level = tesseract::RIL_SYMBOL; | 192 | level = tesseract::RIL_SYMBOL; |
170 | 193 | ||
171 | for (int i = 0; i < extraChars; i++) { | 194 | for (int i = 0; i < extraChars; i++) { |
@@ -233,32 +256,40 @@ Magick::Image wizard::run() { | |||
233 | 256 | ||
234 | cardImg.magick("PNG"); | 257 | cardImg.magick("PNG"); |
235 | cardImg.write("slice.png"); | 258 | cardImg.write("slice.png"); |
236 | } else { | ||
237 | std::cout << "Didn't find name" << std::endl; | ||
238 | } | ||
239 | 259 | ||
240 | if (!firstSlice) { | 260 | if (!firstSlice) { |
241 | firstSlice = true; | 261 | firstSlice = true; |
242 | endImage = cardImg; | 262 | endImage = cardImg; |
243 | 263 | ||
244 | if (theCard.frame == card_frame::m2015) { | 264 | if (theCard.frame == card_frame::m2015) { |
245 | // endImage.extent({endImage.columns(), endImage.rows() + 6}, | 265 | // endImage.extent({endImage.columns(), endImage.rows() + 6}, |
246 | // Magick::SouthGravity); | 266 | // Magick::SouthGravity); |
247 | } | 267 | } |
248 | } else { | 268 | } else { |
249 | int xoff = endImage.columns(); | 269 | int xoff = endImage.columns(); |
250 | 270 | ||
251 | endImage.backgroundColor("black"); | 271 | endImage.backgroundColor("black"); |
252 | 272 | ||
253 | endImage.extent({endImage.columns() + cardImg.columns(), cardImg.rows()}, | 273 | endImage.extent( |
254 | Magick::WestGravity); | 274 | {endImage.columns() + cardImg.columns(), cardImg.rows()}, |
275 | Magick::WestGravity); | ||
255 | 276 | ||
256 | endImage.composite(cardImg, xoff, | 277 | endImage.composite(cardImg, xoff, |
257 | (theCard.frame == card_frame::m2015) ? 6 : 0); | 278 | (theCard.frame == card_frame::m2015) ? 6 : 0); |
279 | } | ||
280 | |||
281 | // break; | ||
282 | ui++; | ||
258 | } | 283 | } |
259 | 284 | ||
260 | // break; | 285 | if (!failure) { |
261 | ui++; | 286 | success = true; |
287 | break; | ||
288 | } | ||
289 | } | ||
290 | |||
291 | if (!success) { | ||
292 | throw std::domain_error("Could not generate card"); | ||
262 | } | 293 | } |
263 | 294 | ||
264 | endImage.magick("PNG"); | 295 | endImage.magick("PNG"); |