recommend-emr-dataprocess/src/main/java/examples/extractor/v20250218/ExtractFeature20250218.java

package examples.extractor.v20250218;

import examples.extractor.ExtractorUtils;
import examples.extractor.RankExtractorFeature_20240530;
import examples.utils.FestiveUtil;
import examples.utils.SimilarityUtils;
import org.apache.commons.lang3.StringUtils;

import java.time.Instant;
import java.time.LocalDateTime;
import java.time.ZoneId;
import java.time.format.DateTimeFormatter;
import java.util.Arrays;
import java.util.HashMap;
import java.util.List;
import java.util.Map;

public class ExtractFeature20250218 {

    private ExtractFeature20250218() {
    }

    public static void handleB1(Map<String, Object> b1Feature, Map<String, Object> featureMap) {
        List<String> times = Arrays.asList("1h", "3h", "6h", "12h", "24h", "72h", "168h");
        List<String> indexList = Arrays.asList("is_share", "share_cnt", "is_return_1", "return_1_uv", "str_one", "ros_one", "str", "ros", "str_plus", "ros_minus", "rovn");
        for (String time : times) {
            for (String index : indexList) {
                double value = Double.parseDouble(b1Feature.getOrDefault(index + "_" + time, "0").toString());
                featureMap.put("b1_" + index + "_" + time, value);
            }

            double rovn = Double.parseDouble(b1Feature.getOrDefault("rovn_" + time, "0").toString());
            double returnNUv = Double.parseDouble(b1Feature.getOrDefault("return_1_uv_" + time, "0").toString());

            featureMap.put("b1_rovn*log(r)_" + time, rovn * RankExtractorFeature_20240530.calLog(returnNUv));
        }

    }

    public static void handleB2ToB11AndB13(Map<String, Map<String, Object>> videoFeature, Map<String, Object> featureMap) {
        List<String> times = Arrays.asList("1h", "3h", "6h", "12h", "24h", "72h", "168h");
        List<String> indexList = Arrays.asList("is_share", "share_cnt", "is_return_1", "return_n_uv", "str_one", "ros_one", "str", "ros", "str_plus", "ros_minus", "rovn");
        for (Map.Entry<String, Map<String, Object>> entry : videoFeature.entrySet()) {
            String key = entry.getKey();
            Map<String, Object> feature = entry.getValue();
            for (String time : times) {
                for (String index : indexList) {
                    double value = Double.parseDouble(feature.getOrDefault(index + "_" + time, "0").toString());
                    featureMap.put(key + "_" + index + "_" + time, value);
                }

                double rovn = Double.parseDouble(feature.getOrDefault("rovn_" + time, "0").toString());
                double returnNUv = Double.parseDouble(feature.getOrDefault("return_n_uv_" + time, "0").toString());

                featureMap.put(key + "_rovn*log(r)_" + time, rovn * RankExtractorFeature_20240530.calLog(returnNUv));
            }

        }
    }

    public static void handleB12(Map<String, Object> b12Feature, Map<String, Object> featureMap) {
        List<String> times = Arrays.asList("7d", "14d", "30d", "60d");
        List<String> indexList = Arrays.asList("is_share", "share_cnt", "is_return_1", "return_n_uv", "str_one", "ros_one", "str", "ros", "str_plus", "ros_minus", "rovn");
        for (String time : times) {
            for (String index : indexList) {
                double value = Double.parseDouble(b12Feature.getOrDefault(index + "_" + time, "0").toString());
                featureMap.put("b12_" + index + "_" + time, value);
            }
            double rovn = Double.parseDouble(b12Feature.getOrDefault("rovn_" + time, "0").toString());
            double returnNUv = Double.parseDouble(b12Feature.getOrDefault("return_n_uv_"+time, "0").toString());
            featureMap.put("b12_rovn*log(r)_" + time, rovn * RankExtractorFeature_20240530.calLog(returnNUv));
        }
    }

    public static void handleVideoBasicFeature(Map<String, Object> videoFeature, long ts, Map<String, Object> featureMap) {
        Double totalTime = Double.parseDouble(videoFeature.getOrDefault("total_time", "0").toString());
        Double width = Double.parseDouble(videoFeature.getOrDefault("width", "0d").toString());
        Double height = Double.parseDouble(videoFeature.getOrDefault("height", "0d").toString());
        Double size = Double.parseDouble(videoFeature.getOrDefault("size", "0d").toString());
        Double bit_rate = Double.parseDouble(videoFeature.getOrDefault("bit_rate", "0d").toString());
        String festiveLabel1 = videoFeature.getOrDefault("festive_label1", "").toString();
        String festiveLabel2 = videoFeature.getOrDefault("festive_label2", "").toString();


        featureMap.put("total_time", totalTime);
        featureMap.put("width", width);
        featureMap.put("height", height);
        featureMap.put("size", size);
        featureMap.put("bit_rate", bit_rate);
        featureMap.put("width/height", ExtractorUtils.divisionDouble(width, height));
        featureMap.put("is_festive", 0);
        featureMap.put("is_greeting", 0);
        if (StringUtils.equals(festiveLabel1, "节假日")) {
            featureMap.put("is_festive", 1);
        } else if (StringUtils.equals(festiveLabel1, "问候语")) {
            featureMap.put("is_greeting", 1);
        }

        featureMap.put("hour", ExtractorUtils.getHourByTimestamp(ts));
        featureMap.put("day_of_week", ExtractorUtils.getDayOfWeekByTimestamp(ts));

        long createTs = Long.parseLong(videoFeature.getOrDefault("gmt_create_timestamp", "0").toString());
        featureMap.put("create_ts_diff", ExtractorUtils.getDaysBetween(createTs, ts));

        String date = LocalDateTime.ofInstant(Instant.ofEpochSecond(ts), ZoneId.systemDefault()).format(DateTimeFormatter.ofPattern("yyyy-MM-dd"));
        String festiveByDate = FestiveUtil.getFestiveByDate(date);
        featureMap.put("today_is_fes", 0);
        featureMap.put("video_fes_eq", 0);
        if (StringUtils.isNotBlank(festiveByDate)) {
            featureMap.put("today_is_fes", 1);
            if (StringUtils.equals(festiveByDate, festiveLabel2)) {
                featureMap.put("video_fes_eq", 1);
            }
        }

    }

    public static void handleC1(Map<String, Object> c1Feature, Map<String, Object> featureMap) {
        List<String> times = Arrays.asList("12h", "24h", "72h", "168h");
        List<String> indexList = Arrays.asList("is_share", "share_cnt", "is_return_1", "return_1_uv", "click", "str_one", "ros_one", "str", "ros", "str_plus", "ros_minus", "rovn");
        for (String time : times) {
            for (String index : indexList) {
                double value = Double.parseDouble(featureMap.getOrDefault(index + "_" + time, "0").toString());
                featureMap.put("c1_" + index + "_" + time, value);
            }
            double rovn = Double.parseDouble(c1Feature.getOrDefault("rovn_" + time, "0").toString());
            double returnNUv = Double.parseDouble(c1Feature.getOrDefault("return_1_uv_"+time, "0").toString());
            featureMap.put("c1_rovn*log(r)_" + time, rovn * RankExtractorFeature_20240530.calLog(returnNUv));
        }
    }

    public static void handleC2ToC3(Map<String, Object> c2Feature, Map<String, Object> c3Feature, Map<String, Object> featureMap) {
        Map<String, Map<String, Object>> featureMaps = new HashMap<>();
        featureMaps.put("c2", c2Feature);
        featureMaps.put("c3", c3Feature);

        List<String> times = Arrays.asList("12h", "24h", "72h", "168h");
        List<String> indexList = Arrays.asList("is_share", "share_cnt", "is_return_1", "return_n_uv", "click");

        for (Map.Entry<String, Map<String, Object>> entry : featureMaps.entrySet()) {
            String key = entry.getKey();
            Map<String, Object> feature = entry.getValue();
            for (String time : times) {
                for (String index : indexList) {
                    double value = Double.parseDouble(feature.getOrDefault(index + "_" + time, "0").toString());
                    featureMap.put(key + "_" + index + "_" + time, value);
                }
            }
        }
    }

    public static void handleC4(Map<String, Object> c4Feature, Map<String, Object> featureMap) {
        List<String> times = Arrays.asList("24h", "72h", "168h");
        List<String> indexList = Arrays.asList("str_one", "ros_one", "str", "ros", "str_plus", "ros_minus", "rovn");

        for (String time : times) {
            for (String index : indexList) {
                double value = Double.parseDouble(c4Feature.getOrDefault("avg_" + index + "_" + time, "0").toString());
                featureMap.put("c4_avg_" + index + "_" + time, value);

                double max = Double.parseDouble(c4Feature.getOrDefault("max_" + index + "_" + time, "0").toString());
                double min = Double.parseDouble(c4Feature.getOrDefault("min_" + index + "_" + time, "0").toString());

                featureMap.put("c4_diff_" + index + "_" + time, max - min);
            }
        }

    }

    public static void handleC5ToC6(Map<String, Object> c5Feature, Map<String, Object> c6Feature, Map<String, Object> videoMap, Map<String, Object> featureMap) {
        Map<String, Map<String, Object>> featureMaps = new HashMap<>();
        featureMaps.put("c5", c5Feature);
        featureMaps.put("c6", c6Feature);
        List<String> times = Arrays.asList("tags_1d", "tags_3d", "tags_7d");

        String title = videoMap.getOrDefault("title", "").toString();

        for (Map.Entry<String, Map<String, Object>> entry : featureMaps.entrySet()) {
            String key = entry.getKey();
            Map<String, Object> feature = entry.getValue();
            for (String time : times) {
                String tags = feature.getOrDefault(time, "").toString();
                Double[] scores = ExtractorUtils.funcC34567ForTagsNew(tags, title);
                featureMap.put(key + "_matchnum" + "_" + time, scores[0]);
                featureMap.put(key + "_maxscore" + "_" + time, scores[1]);
                featureMap.put(key + "_avgscore" + "_" + time, scores[2]);
            }
        }

    }

    public static Map<String, Map<String, String[]>> handleC7ToC8(Map<String, Object> c7Feature, Map<String, Object> c8Feature) {
        Map<String, Map<String, String[]>> resultMap = new HashMap<>();

        Map<String, Map<String, Object>> featureMaps = new HashMap<>();
        featureMaps.put("c7", c7Feature);
        featureMaps.put("c8", c8Feature);
        List<String> indexList = Arrays.asList("share", "return");
        for (Map.Entry<String, Map<String, Object>> entry : featureMaps.entrySet()) {
            String key = entry.getKey();
            Map<String, Object> feature = entry.getValue();
            for (String index : indexList) {
                if (feature.containsKey(index)) {
                    Map<String, String[]> cfMap = new HashMap<>();
                    String[] entries = feature.get(index).toString().split(",");
                    for (String e : entries) {
                        String[] rList = e.split(":");
                        if (rList.length >= 4) {
                            String vid = rList[0];
                            String value1 = rList[1];
                            String value2 = rList[2];
                            String value3 = rList[3];
                            String[] strs = {value1, value2, value3};
                            cfMap.put(vid, strs);
                        }
                    }
                    resultMap.put(key, cfMap);
                }
            }
        }

        return resultMap;
    }

    public static void useC7ToC8(Map<String, Map<String, String[]>> map, String vid, Map<String, Object> featureMap) {
        if (StringUtils.isBlank(vid)) {
            return;
        }
        for (String key : Arrays.asList("c6", "c7")) {
            for (String action : Arrays.asList("share", "return")) {
                String featureKey = key + "_" + action;
                if (map.containsKey(featureKey)) {
                    Map<String, String[]> cfMap = map.get(featureKey);
                    String[] scores = cfMap.get(vid);
                    featureMap.put(featureKey + "_score", Double.parseDouble(scores[0]));
                    featureMap.put(featureKey + "_num", Double.parseDouble(scores[1]));
                    featureMap.put(featureKey + "_rank", ExtractorUtils.reciprocal(Double.parseDouble(scores[2])));
                }
            }
        }
    }

    public static void handleD3(Map<String, Object> d3Feature, Map<String, Object> featureMap) {
        for (String index : Arrays.asList("exp", "return_n", "rovn")) {
            double value = Double.parseDouble(d3Feature.getOrDefault(index, "0").toString());
            featureMap.put("d3_" + index, value);
        }
    }

    public static void handleD1(Map<String, Object> d4Feature, Map<String, Object> featureMap) {
        double rosCfScores = Double.parseDouble(d4Feature.getOrDefault("ros_cf_score", "0").toString());
        featureMap.put("d1_ros_cf_score", rosCfScores);
        double rovCfScores = Double.parseDouble(d4Feature.getOrDefault("rov_cf_score", "0").toString());
        featureMap.put("d1_rov_cf_score", rovCfScores);

        double rosCfRank = Double.parseDouble(d4Feature.getOrDefault("ros_cf_rank", "0").toString());
        featureMap.put("d1_ros_cf_rank", ExtractorUtils.reciprocal(rosCfRank));
        double rovCfRank = Double.parseDouble(d4Feature.getOrDefault("rov_cf_rank", "0").toString());
        featureMap.put("d1_rov_cf_rank", ExtractorUtils.reciprocal(rovCfRank));
    }

    public static void handleD2(Map<String, Object> d5Feature, Map<String, Object> featureMap) {
        double score = Double.parseDouble(d5Feature.getOrDefault("score", "0").toString());
        featureMap.put("d2_score", score);

        double rank = Double.parseDouble(d5Feature.getOrDefault("rank", "0").toString());
        featureMap.put("d2_rank", ExtractorUtils.reciprocal(rank));
    }

    public static void handleVideoSimilarity(Map<String, Object> videoFeature, Map<String, Object> headVideoFeature, Map<String, Object> featureMap) {
        String headVideoTitle = headVideoFeature.getOrDefault("title", "").toString();
        String headVideoMergeCate2 = headVideoFeature.getOrDefault("merge_second_level_cate", "").toString();
        String headVideoMergeCate1 = headVideoFeature.getOrDefault("merge_first_level_cate", "").toString();
        String headVideoFestiveLabel2 = headVideoFeature.getOrDefault("festive_label2", "").toString();


        String videoTitle = videoFeature.getOrDefault("title", "").toString();
        String videoMergeCate2 = videoFeature.getOrDefault("merge_second_level_cate", "").toString();
        String videoMergeCate1 = videoFeature.getOrDefault("merge_first_level_cate", "").toString();
        String videoFestiveLabel2 = videoFeature.getOrDefault("festive_label2", "").toString();

        double titleSimilarity = ExtractFeature20250218.calcTxtSimilarity(headVideoTitle, videoTitle);
        double headTitleAndMerge1Similarity = ExtractFeature20250218.calcTxtSimilarity(headVideoTitle, videoMergeCate1);
        double headTitleAndMerge2Similarity = ExtractFeature20250218.calcTxtSimilarity(headVideoTitle, videoMergeCate2);
        double headTitleAndFestiveSimilarity = ExtractFeature20250218.calcTxtSimilarity(headVideoTitle, videoFestiveLabel2);
        double merge1Similarity = ExtractFeature20250218.calcTxtSimilarity(headVideoMergeCate1, videoMergeCate1);
        double merge2Similarity = ExtractFeature20250218.calcTxtSimilarity(headVideoMergeCate2, videoMergeCate2);
        double festiveSimilarity = ExtractFeature20250218.calcTxtSimilarity(headVideoFestiveLabel2, videoFestiveLabel2);

        featureMap.put("title_sim", titleSimilarity);
        featureMap.put("head_title_merge1_sim", headTitleAndMerge1Similarity);
        featureMap.put("head_title_merge2_sim", headTitleAndMerge2Similarity);
        featureMap.put("head_title_festive_sim", headTitleAndFestiveSimilarity);
        featureMap.put("merge1_sim", merge1Similarity);
        featureMap.put("merge2_sim", merge2Similarity);
        featureMap.put("festive_sim", festiveSimilarity);

    }

    private static double calcTxtSimilarity(String txt1, String txt2) {
        if (StringUtils.isBlank(txt1) || StringUtils.isBlank(txt2)) {
            return 0d;
        }
        return SimilarityUtils.word2VecSimilarity(txt1, txt2);
    }
}