further improvements

email_parser/emailparser.cpp

@@ -34,10 +34,17 @@ std::map<std::string, std::string> EmailParser::parse(std::string &email_content
 
     std::string parameters;
     parameters = search_in_images(email_content);
-    if (parameters != "") return get_params_from_string(parameters);
+
+    if (parameters != "") {
+        std::cout << parameters << std::endl;
+        return get_params_from_string(parameters);
+    }
 
     parameters = search_in_text(email_content);
-    if (parameters != "") return get_params_from_string(parameters);
+    if (parameters != "") {
+        std::cout << parameters << std::endl;
+        return get_params_from_string(parameters);
+    }
 
     /* If the code has reached this part and found nothing, it's most likely that there are no QR codes at all. */
 
@@ -74,7 +81,6 @@ std::string EmailParser::find_check_parameters(std::string &part) {
 }
 
 std::string EmailParser::extract_qr_url_from_img(std::string &part) {
-    std::string url = "";
     boost::regex img_tag_regex("<img[^\\n\\r<]*>");
     boost::regex img_url_str("https?:\\/\\/.*(qr(code)?)[^\\n\\r\"]+", boost::regex::icase);
 
@@ -85,7 +91,22 @@ std::string EmailParser::extract_qr_url_from_img(std::string &part) {
         }
 
     }
-    return url;
+    return "";
+}
+
+std::vector<std::string> EmailParser::extract_qr_embeddings_from_part(std::string &part) {
+    std::vector<std::string> embeddings = {};
+    boost::regex img_tag_regex("<img[^\\n\\r<]*>");
+    boost::regex img_base64_str("data:image\\/(png|jpg);base64,[\\w+\\/=]+", boost::regex::icase);
+
+    for (boost::sregex_iterator it{part.begin(), part.end(), img_tag_regex}, end{}; it != end; it++) {
+        std::string img_tag = it->str();
+        for (boost::sregex_iterator it2{img_tag.begin(), img_tag.end(), img_base64_str}, end2{}; it2 != end2; it2++) {
+            embeddings.push_back(split(it2->str(), ",")[1]);
+        }
+
+    }
+    return embeddings;
 }
 
 std::string EmailParser::search_in_images(std::string &content) {
@@ -106,24 +127,7 @@ std::string EmailParser::search_in_images(std::string &content) {
         part.erase(std::remove(part.begin(), part.end(), '\r'), part.end());
         part.erase(std::remove(part.begin(), part.end(), '\n'), part.end());
         std::string decoded = base64_decode(part);
-        cv::Mat image;
-
-        if (decoded.substr(1, 3) == "PNG" || decoded.substr(1, 3) == "JPG") {
-            std::vector<uchar> data(decoded.begin(), decoded.end());
-            image = cv::imdecode(cv::Mat(data), 1);
-        } else if (decoded.substr(0, 3) == "GIF") {
-            std::string gif_file_path = get_application_home_path() + "/temp.gif";
-
-            std::ofstream gif_output(gif_file_path, std::ios::binary);
-            gif_output << decoded;
-            gif_output.close();
-            cv::VideoCapture gif(gif_file_path, cv::CAP_FFMPEG);
-            gif.read(image);
-        }
-
-        cv::QRCodeDetector qrDecoder = cv::QRCodeDetector();
-        std::string decoded_qr = qrDecoder.detectAndDecode(image);
-        return find_check_parameters(decoded_qr);
+        return handle_image(decoded);
     }
     return "";
 }
@@ -158,22 +162,52 @@ std::string EmailParser::search_in_text(std::string &content) {
 
         // If there's no, try search anything that looks like a link to a qr code.
         std::string url = extract_qr_url_from_img(part);
-        Net n;
-        std::string path = get_path_relative_to_home(".local/share/checks_parser/tmp");
-        n.get_file(url, path);
+        if (url != "") {
+            Net n;
+            std::string path = get_path_relative_to_home(".local/share/checks_parser/tmp");
+            n.get_file(url, path);
 
-        std::string qr_code_contents = read_file(path);
+            std::string qr_code_contents = read_file(path);
 
-        std::vector<uchar> data(qr_code_contents.begin(), qr_code_contents.end());
-        cv::Mat image = cv::imdecode(cv::Mat(data), 1);
-
-        cv::QRCodeDetector qrDecoder = cv::QRCodeDetector();
-        std::string decoded_qr = qrDecoder.detectAndDecode(image);
-        parameters = find_check_parameters(decoded_qr);
+            std::vector<uchar> data(qr_code_contents.begin(), qr_code_contents.end());
+            cv::Mat image = cv::imdecode(cv::Mat(data), 1);
 
+            cv::QRCodeDetector qrDecoder = cv::QRCodeDetector();
+            std::string decoded_qr = qrDecoder.detectAndDecode(image);
+            parameters = find_check_parameters(decoded_qr);
+        }
         if (parameters != "") return parameters;
 
         // if there's no any link that looks like a link to QR code, maybe the qr code is encoded as base64 inside an img tag.
+
+        std::vector<std::string> embeddings = extract_qr_embeddings_from_part(part);
+        for (std::string &embedding : embeddings) {
+            std::string decoded = base64_decode(embedding);
+            parameters = handle_image(decoded);
+            if (parameters != "") return parameters;
+        }
     }
     return "";
 }
+
+std::string EmailParser::handle_image(std::string &content) {
+    cv::Mat image;
+
+    if (content.substr(1, 3) == "PNG" || content.substr(1, 3) == "JPG") {
+        std::vector<uchar> data(content.begin(), content.end());
+        image = cv::imdecode(cv::Mat(data), 1);
+    } else if (content.substr(0, 3) == "GIF") {
+        std::string gif_file_path = get_application_home_path() + "/temp.gif";
+
+        std::ofstream gif_output(gif_file_path, std::ios::binary);
+        gif_output << content;
+        gif_output.close();
+        cv::VideoCapture gif(gif_file_path, cv::CAP_FFMPEG);
+        gif.read(image);
+    }
+    if (image.empty()) return "";
+
+    cv::QRCodeDetector qrDecoder = cv::QRCodeDetector();
+    std::string decoded_qr = qrDecoder.detectAndDecode(image);
+    return find_check_parameters(decoded_qr);
+}

email_parser/emailparser.h

@@ -13,9 +13,12 @@ public:
     std::vector<std::pair<int, int>> find_parts(const boost::regex &start_regex, const boost::regex &end_regex, const std::string &content);
     std::string find_check_parameters(std::string &part);
     std::string extract_qr_url_from_img(std::string &part);
+    std::vector<std::string> extract_qr_embeddings_from_part(std::string &part);
 
     std::string search_in_images(std::string &content);
     std::string search_in_text(std::string &content);
+
+    std::string handle_image(std::string &content);
 };
 
 #endif // CHECKS_PARSER_EMAIL_PARSER

main.cpp

@@ -34,12 +34,15 @@
 int main(int argc, char *argv[]) {
 
     EmailParser p;
+    p.parse_file("/home/leca/example_email_receipts/lamoda.eml");
+    // p.parse_file("/home/leca/example_email_receipts/lamoda2.eml");
     p.parse_file("/home/leca/example_email_receipts/читай_город.eml");
     p.parse_file("/home/leca/example_email_receipts/lenta.eml");
     p.parse_file("/home/leca/example_email_receipts/magnit.eml");
     p.parse_file("/home/leca/example_email_receipts/pyaterochka.eml");
     p.parse_file("/home/leca/example_email_receipts/rzd.eml");
     p.parse_file("/home/leca/example_email_receipts/russteels.eml");
+    p.parse_file("/home/leca/example_email_receipts/avtodor.eml");
     return 0;
     curl_global_init(CURL_GLOBAL_ALL);
     qRegisterMetaType<Check>("Check");

translations/en_US.ts

@@ -661,7 +661,7 @@
 <context>
     <name>QObject</name>
     <message>
-        <location filename="../main.cpp" line="74"/>
+        <location filename="../main.cpp" line="77"/>
         <source>Using locale: </source>
         <translation>Using locale: </translation>
     </message>

translations/ru_RU.ts

@@ -637,7 +637,7 @@
 <context>
     <name>QObject</name>
     <message>
-        <location filename="../main.cpp" line="74"/>
+        <location filename="../main.cpp" line="77"/>
         <source>Using locale: </source>
         <translation>Использую локаль: </translation>
     </message>