Issue 29711 analytics matcher layer (#29755)

This is an implementation for the matcher layer on analytics

dotCMS/src/main/java/com/dotcms/analytics/track/AnalyticsTrackWebInterceptor.java

@@ -0,0 +1,105 @@
+package com.dotcms.analytics.track;
+
+import com.dotcms.filters.interceptor.Result;
+import com.dotcms.filters.interceptor.WebInterceptor;
+import com.dotcms.jitsu.EventLogSubmitter;
+import com.dotcms.util.CollectionsUtils;
+import com.dotcms.util.WhiteBlackList;
+import com.dotmarketing.util.Config;
+import com.dotmarketing.util.Logger;
+import com.liferay.util.StringPool;
+
+import javax.servlet.http.HttpServletRequest;
+import javax.servlet.http.HttpServletResponse;
+import java.io.IOException;
+import java.util.Map;
+import java.util.Optional;
+import java.util.concurrent.ConcurrentHashMap;
+import java.util.function.Predicate;
+
+/**
+ * Web Interceptor to track analytics
+ * @author jsanca
+ */
+public class AnalyticsTrackWebInterceptor  implements WebInterceptor {
+
+    private final static Map<String, RequestMatcher> requestMatchersMap = new ConcurrentHashMap<>();
+
+    private final EventLogSubmitter submitter;
+
+    /// private static final String[] DEFAULT_BLACKLISTED_PROPS = new String[]{"^/api/*"};
+    private static final String[] DEFAULT_BLACKLISTED_PROPS = new String[]{StringPool.BLANK};
+    private final WhiteBlackList whiteBlackList = new WhiteBlackList.Builder()
+            .addWhitePatterns(Config.getStringArrayProperty("ANALYTICS_WHITELISTED_KEYS",
+                    new String[]{StringPool.BLANK})) // allows everything
+            .addBlackPatterns(CollectionsUtils.concat(Config.getStringArrayProperty(  // except this
+                    "ANALYTICS_BLACKLISTED_KEYS", new String[]{}), DEFAULT_BLACKLISTED_PROPS)).build();
+
+    public AnalyticsTrackWebInterceptor() {
+
+        submitter = new EventLogSubmitter();
+        addRequestMatcher(
+                new PagesAndUrlMapsRequestMatcher(),
+                new FilesRequestMatcher(),
+                new RulesRedirectsRequestMatcher(),
+                new VanitiesRequestMatcher());
+    }
+
+    /**
+     * Add a request matchers
+     * @param requestMatchers
+     */
+    public static void addRequestMatcher(final RequestMatcher... requestMatchers) {
+        for (final RequestMatcher matcher : requestMatchers) {
+            requestMatchersMap.put(matcher.getId(), matcher);
+        }
+    }
+
+    /**
+     * Remove a request matcher by id
+     * @param requestMatcherId
+     */
+    public static void removeRequestMatcher(final String requestMatcherId) {
+
+        requestMatchersMap.remove(requestMatcherId);
+    }
+
+    @Override
+    public Result intercept(final HttpServletRequest request, final HttpServletResponse response) throws IOException {
+
+        if (whiteBlackList.isAllowed(request.getRequestURI())) {
+            final Optional<RequestMatcher> matcherOpt = this.anyMatcher(request, response, RequestMatcher::runBeforeRequest);
+            if (matcherOpt.isPresent()) {
+
+                Logger.debug(this, () -> "intercept, Matched: " + matcherOpt.get().getId() + " request: " + request.getRequestURI());
+                //fireNextStep(request, response);
+            }
+        }
+
+        return Result.NEXT;
+    }
+
+    @Override
+    public boolean afterIntercept(final HttpServletRequest request, final HttpServletResponse response) {
+
+        if (whiteBlackList.isAllowed(request.getRequestURI())) {
+            final Optional<RequestMatcher> matcherOpt = this.anyMatcher(request, response, RequestMatcher::runAfterRequest);
+            if (matcherOpt.isPresent()) {
+
+                Logger.debug(this, () -> "afterIntercept, Matched: " + matcherOpt.get().getId() + " request: " + request.getRequestURI());
+                //fireNextStep(request, response);
+            }
+        }
+
+        return true;
+    }
+
+    private Optional<RequestMatcher> anyMatcher(final HttpServletRequest request, final HttpServletResponse response, Predicate<? super RequestMatcher> filterRequest) {
+
+        return requestMatchersMap.values().stream()
+                .filter(filterRequest)
+                .filter(matcher -> matcher.match(request, response))
+                .findFirst();
+    }
+
+}

dotCMS/src/main/java/com/dotcms/analytics/track/FilesRequestMatcher.java

@@ -0,0 +1,48 @@
+package com.dotcms.analytics.track;
+
+import com.dotcms.visitor.filter.characteristics.Character;
+import com.dotcms.visitor.filter.characteristics.CharacterWebAPI;
+import com.dotmarketing.business.web.WebAPILocator;
+import com.dotmarketing.filters.CMSFilter;
+
+import javax.servlet.http.HttpServletRequest;
+import javax.servlet.http.HttpServletResponse;
+import java.util.Objects;
+
+/**
+ * Matcher for pages or files
+ * @author jsanca
+ */
+public class FilesRequestMatcher implements RequestMatcher {
+
+    private final CharacterWebAPI characterWebAPI;
+
+    public FilesRequestMatcher() {
+        this(WebAPILocator.getCharacterWebAPI());
+    }
+
+    public FilesRequestMatcher(final CharacterWebAPI characterWebAPI) {
+        this.characterWebAPI = characterWebAPI;
+    }
+
+    @Override
+    public boolean runBeforeRequest() {
+        return true;
+    }
+
+    @Override
+    public boolean match(final HttpServletRequest request, final HttpServletResponse response) {
+
+        final Character character = this.characterWebAPI.getOrCreateCharacter(request, response);
+        if (Objects.nonNull(character)) {
+
+            final CMSFilter.IAm iAm = (CMSFilter.IAm) character.getMap().
+                    getOrDefault("iAm", CMSFilter.IAm.NOTHING_IN_THE_CMS);
+
+            // should we have a fallback when nothing is returned???
+            return iAm == CMSFilter.IAm.FILE;
+        }
+
+        return false;
+    }
+}

dotCMS/src/main/java/com/dotcms/analytics/track/PagesAndUrlMapsRequestMatcher.java

@@ -0,0 +1,48 @@
+package com.dotcms.analytics.track;
+
+import com.dotcms.visitor.filter.characteristics.Character;
+import com.dotcms.visitor.filter.characteristics.CharacterWebAPI;
+import com.dotmarketing.business.web.WebAPILocator;
+import com.dotmarketing.filters.CMSFilter;
+
+import javax.servlet.http.HttpServletRequest;
+import javax.servlet.http.HttpServletResponse;
+import java.util.Objects;
+
+/**
+ * Matcher for pages or files
+ * @author jsanca
+ */
+public class PagesAndUrlMapsRequestMatcher implements RequestMatcher {
+
+    private final CharacterWebAPI characterWebAPI;
+
+    public PagesAndUrlMapsRequestMatcher() {
+        this(WebAPILocator.getCharacterWebAPI());
+    }
+
+    public PagesAndUrlMapsRequestMatcher(final CharacterWebAPI characterWebAPI) {
+        this.characterWebAPI = characterWebAPI;
+    }
+
+    @Override
+    public boolean runBeforeRequest() {
+        return true;
+    }
+
+    @Override
+    public boolean match(final HttpServletRequest request, final HttpServletResponse response) {
+
+        final Character character = this.characterWebAPI.getOrCreateCharacter(request, response);
+        if (Objects.nonNull(character)) {
+
+            final CMSFilter.IAm iAm = (CMSFilter.IAm) character.getMap().
+                    getOrDefault("iAm", CMSFilter.IAm.NOTHING_IN_THE_CMS);
+
+            // should we have a fallback when nothing is returned???
+            return iAm == CMSFilter.IAm.PAGE; // this captures also url maps
+        }
+
+        return false;
+    }
+}

dotCMS/src/main/java/com/dotcms/analytics/track/RequestMatcher.java

@@ -0,0 +1,126 @@
+package com.dotcms.analytics.track;
+
+import com.dotmarketing.util.Config;
+import com.dotmarketing.util.RegEX;
+import io.vavr.control.Try;
+
+import javax.servlet.http.HttpServletRequest;
+import javax.servlet.http.HttpServletResponse;
+import java.io.UnsupportedEncodingException;
+import java.net.URLDecoder;
+import java.util.Objects;
+import java.util.Set;
+
+/**
+ * Matcher to include the tracking for analytics of some request.
+ *
+ * @author jsanca
+ */
+public interface RequestMatcher {
+
+
+    String CHARSET = Try.of(()->Config.getStringProperty("CHARSET", "UTF-8")).getOrElse("UTF-8");
+
+    /**
+     * Return true if match the request with the patterns and methods
+     * @param request {@link HttpServletRequest}
+     * @param response {@link HttpServletResponse}
+     * @return boolean true if the request match the patterns and methods
+     */
+    default boolean match(final HttpServletRequest request, final HttpServletResponse response) {
+
+        final Set<String> patterns = getMatcherPatterns();
+        final Set<String> methods  = getAllowedMethods();
+        return  Objects.nonNull(patterns) && !patterns.isEmpty() &&
+                Objects.nonNull(methods) &&  !methods.isEmpty()  &&
+                isAllowedMethod (methods, request.getMethod())   &&
+                match(request, response, patterns);
+    }
+
+    /**
+     * Match the request with the patterns
+     * @param request {@link HttpServletRequest}
+     * @param response {@link HttpServletResponse}
+     * @param patterns Set of patterns
+     * @return boolean true if any of the patterns match the request
+     */
+    default boolean match(final HttpServletRequest request, final HttpServletResponse response, final Set<String> patterns) {
+
+        final String requestURI = request.getRequestURI();
+        return patterns.stream().anyMatch(pattern -> match(requestURI, pattern));
+    }
+
+    /**
+     * Means the matcher should run before request
+     * @return boolean
+     */
+    default boolean runBeforeRequest() {
+        return false;
+    }
+
+    /**
+     * Means the matcher should run after request
+     * @return boolean
+     */
+    default boolean runAfterRequest() {
+        return false;
+    }
+    /**
+     * Match the request URI with the pattern
+     * @param requestURI String
+     * @param pattern String
+     * @return boolean true if the pattern match the request URI
+     */
+    default boolean match (final String requestURI, final String pattern) {
+
+        String uftUri = null;
+
+        try {
+
+            uftUri = URLDecoder.decode(requestURI, CHARSET);
+        } catch (UnsupportedEncodingException e) {
+
+            uftUri = requestURI;
+        }
+
+        return RegEX.containsCaseInsensitive(uftUri, pattern.trim());
+    } // match.
+
+    /**
+     * Determinate if the method is allowed
+     * @param methods Set of methods
+     * @param method String current request method
+     * @return boolean true if the method is allowed
+     */
+    default boolean isAllowedMethod(final Set<String> methods, final String method) {
+
+            return methods.contains(method);
+    }
+
+    /**
+     * Returns a set of patterns for the matcher
+     * @return Set by default empty
+     */
+    default Set<String> getMatcherPatterns() {
+
+        return Set.of();
+    }
+
+    /**
+     * Returns the request methods allowed for this matcher.
+     * @return Set by default empty
+     */
+    default Set<String> getAllowedMethods() {
+
+        return Set.of();
+    }
+
+    /**
+     * Return an id for the Matcher, by default returns the class name.
+     * @return
+     */
+    default String getId() {
+
+        return this.getClass().getName();
+    }
+}

dotCMS/src/main/java/com/dotcms/analytics/track/RulesRedirectsRequestMatcher.java

@@ -0,0 +1,24 @@
+package com.dotcms.analytics.track;
+
+import javax.servlet.http.HttpServletRequest;
+import javax.servlet.http.HttpServletResponse;
+import java.util.Objects;
+
+/**
+ * Matcher for vanity urls and rules redirect
+ * @author jsanca
+ */
+public class RulesRedirectsRequestMatcher implements RequestMatcher {
+
+    @Override
+    public boolean runAfterRequest() {
+        return true;
+    }
+
+    @Override
+    public boolean match(final HttpServletRequest request, final HttpServletResponse response) {
+
+        final String ruleRedirect = response.getHeader("X-DOT-SendRedirectRuleAction");
+        return Objects.nonNull(ruleRedirect) && "true".equals(ruleRedirect);
+    }
+}

dotCMS/src/main/java/com/dotcms/analytics/track/VanitiesRequestMatcher.java

@@ -0,0 +1,27 @@
+package com.dotcms.analytics.track;
+
+import com.dotmarketing.filters.Constants;
+
+import javax.servlet.http.HttpServletRequest;
+import javax.servlet.http.HttpServletResponse;
+import java.util.Objects;
+
+/**
+ * Matcher for vanity urls and rules redirect
+ * @author jsanca
+ */
+public class VanitiesRequestMatcher implements RequestMatcher {
+
+    @Override
+    public boolean runAfterRequest() {
+        return true;
+    }
+
+    @Override
+    public boolean match(final HttpServletRequest request, final HttpServletResponse response) {
+
+        final Object vanityHasRun = request.getAttribute(Constants.VANITY_URL_OBJECT);
+
+        return  Objects.nonNull(vanityHasRun);
+    }
+}