implement validations of pagination URLs

java/org/chromium/distiller/PageParameterDetector.java

Index: java/org/chromium/distiller/PageParameterDetector.java
diff --git a/java/org/chromium/distiller/PageParameterDetector.java b/java/org/chromium/distiller/PageParameterDetector.java
index 0bf6dfcb7a57c7d290bb751bcc82e8d6d8f1767c..5f0d03e3928c1bf249d46d4b69d9ceb12fa31169 100644
--- a/java/org/chromium/distiller/PageParameterDetector.java
+++ b/java/org/chromium/distiller/PageParameterDetector.java
@@ -1,4 +1,4 @@
-// Copyright 2016 The Chromium Authors. All rights reserved.
+// Copyright 2015 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
@@ -20,17 +20,15 @@ import java.util.Set;
  * Background:
  *   The long article/news/forum thread/blog document may be partitioned into several partial pages
  *   by webmaster.  Each partial page has outlinks pointing to the adjacent partial pages.  The
- *   anchor text of those outlinks is numeric.  Meanwhile, there may be a page which contains the
- *   whole content, called "single page".
+ *   anchor text of those outlinks is numeric.
  *
  * Definitions:
- * A single page document is a document that contains the whole content.
  * A paging document is one of the partial pages.
  * "digital" means the text contains only digits.
  * A page pattern is a paging URL whose page parameter value is replaced with a place holder
  * (PAGE_PARAM_PLACEHOLDER).
- * Example: if the original url is "http: *www.foo.com/a/b-3.html", the page pattern is
- * "http: *www.foo.com/a/b-[*!].html".
+ * Example: if the original url is "http://www.foo.com/a/b-3.html", the page pattern is
+ * "http://www.foo.com/a/b-[*!].html".
  *
  * This class extracts the page parameter from a document's outlinks.
  * The basic idea:
@@ -39,60 +37,82 @@ import java.util.Set;
  *       (either a query value or a path component) in URL.  If one part of a URL is always a linear
  *        map from its digital anchor text, we guess the part is the page parameter of the URL.
  *
- * As an example, consider a document http: *a/b?c=1&p=10, which contains the following digital
+ * As an example, consider a document http://a/b?c=1&p=10, which contains the following digital
  * outlinks:
- *   <a href=http: *a/b?c=1&p=20>3</a>
- *   <a href=http: *a/b?c=1&p=30>4</a>
- *   <a href=http: *a/b?c=1&p=40>5</a>
- *   <a href=http: *a/b?c=1&p=all>single page</a>
+ *   <a href=http://a/b?c=1&p=20>3</a>
+ *   <a href=http://a/b?c=1&p=30>4</a>
+ *   <a href=http://a/b?c=1&p=40>5</a>
  * This class finds that the "p" parameter is always equal to "anchor text" * 10 - 10, and so
- * guesses it is the page parameter.  The associated page pattern is http: *a/b?c=1&p=[*!].
- * Then, this class extracts the single page based on page parameter info.  The single page url is
- * http: *a/b?c=1&p=all.
+ * guesses it is the page parameter.  The associated page pattern is http://a/b?c=1&p=[*!].
  */
 public class PageParameterDetector {
-    private static final String PAGE_PARAM_PLACEHOLDER = "[*!]";
+    static final String PAGE_PARAM_PLACEHOLDER = "[*!]";
+    static final int PAGE_PARAM_PLACEHOLDER_LEN = PAGE_PARAM_PLACEHOLDER.length();
 
     /**
-     * Stores information about the link (anchor) after the page parameter is detected:
-     * - the page number (as represented by the original plain text) for the link
-     * - the original page parameter numeric component in the URL (this component would be replaced
-     *   by PAGE_PARAM_PLACEHOLDER in the URL pattern)
-     * - the position of this link in the list of ascending numbers.
+     * The interface that page pattern handlers must implement to detect page parameter from
+     * potential pagination URLs.
      */
-    static class LinkInfo {
-        private int mPageNum;
-        private int mPageParamValue;
-        private int mPosInAscendingList;
+    interface PagePattern {
+       /**
+        * Returns the string of the URL page pattern.
+        */
+        String toString();
+
+       /**
+        * Returns the page number extracted from the URL during creation of object that implements
+        * this interface.
+        */
+        int getPageNumber();
+        
+        /**
+         * Validates this page pattern according to the current document URL through a pipeline of
+         * rules.
+         *
+         * Returns true if page pattern is valid.
+         *
+         * @param docUrl the current document URL
+         */
+        boolean isValidFor(ParsedUrl docUrl);
 
-        LinkInfo(int pageNum, int pageParamValue, int posInAscendingList) {
-            mPageNum = pageNum;
-            mPageParamValue = pageParamValue;
-            mPosInAscendingList = posInAscendingList;
-        }
-    }  // LinkInfo
+        /**
+         * Returns true if a URL matches this page pattern based on a pipeline of rules.
+         *
+         * @param url the URL to evalutate
+         */
+        boolean isPagingUrl(String url);
+    }
 
     /**
-     * Stores a map of URL pattern to its associated list of LinkInfo's.
+     * Stores a map of URL pattern to its associated list of PageLinkInfo's.
      */
     private static class PageCandidatesMap {
-        private final Map<String, List<LinkInfo>> map = new HashMap<String, List<LinkInfo>>();
+        private static class Info {
+            private final PagePattern mPattern;
+            private final List<PageLinkInfo> mLinks;
+
+            Info(PagePattern pattern, PageLinkInfo link) {
+                mPattern = pattern;
+                mLinks = new ArrayList<PageLinkInfo>();
+                mLinks.add(link);
+            }
+        }
+
+        private final Map<String, Info> map = new HashMap<String, Info>();
 
         /**
-         * Adds urlPattern with its LinkInfo into the map.  If the urlPattern already exists, adds
-         * the link to the list of LinkInfo's.  Otherwise, creates a new map entry.
+         * Adds urlPattern with its PageLinkInfo into the map.  If the urlPattern already exists,
+         * adds the link to the list of LinkInfo's.  Otherwise, creates a new map entry.
          */
-        private void add(String urlPattern, LinkInfo link) {
-            if (map.containsKey(urlPattern)) {
-                map.get(urlPattern).add(link);
+        private void add(PagePattern pattern, PageLinkInfo link) {
+            final String patternStr = pattern.toString();
+            if (map.containsKey(patternStr)) {
+                map.get(patternStr).mLinks.add(link);
             } else {
-                List<LinkInfo> links = new ArrayList<LinkInfo>();
-                links.add(link);
-                map.put(urlPattern, links);
+                map.put(patternStr, new Info(pattern, link));
             }
         }
-
-    }  // PageCandidatesMap
+    }
 
     // All the known bad page param names.
     private static Set<String> sBadPageParamNames = null;
@@ -105,12 +125,12 @@ public class PageParameterDetector {
      * - the name of a query name-value component is not one of sBadPageParamNames, and
      * - the value of the query component is a plain number (>= 0).
      * E.g. a URL query with 3 plain number query values will generate 3 URL page patterns with 3
-     * LinkInfo's, and hence 3 page parameter candidates.
+     * PageLinkInfo's, and hence 3 page parameter candidates.
      *
      * @param url ParsedUrl of the URL to process
      * @param pageNum the page number as represented in original plain text
      * @param posInAscendingNumbers position of this page number in the list of ascending numbers
-     * @param pageCandidates the map of URL pattern to its associated list of LinkInfo's
+     * @param pageCandidates the map of URL pattern to its associated list of PageLinkInfo's
      */
     private static void extractPageParamCandidatesFromQuery(ParsedUrl url, int pageNum,
             int posInAscendingNumbers, PageCandidatesMap pageCandidates) {
@@ -118,19 +138,13 @@ public class PageParameterDetector {
         if (queryParams.length == 0) return;  // No query.
 
         for (String[] nameValue : queryParams) {
-            final String queryName = nameValue[0];
-            final String queryValue = nameValue[1];
-            if (!queryName.isEmpty() && !queryValue.isEmpty() &&
-                    StringUtil.isStringAllDigits(queryValue) && !isPageParamNameBad(queryName)) {
-                int value = StringUtil.toNumber(queryValue);
-                if (value >= 0) {
-                    pageCandidates.add(
-                            url.replaceQueryValue(queryName, queryValue, PAGE_PARAM_PLACEHOLDER),
-                            new LinkInfo(pageNum, value, posInAscendingNumbers));
-                }
+            PagePattern pattern = QueryParamPagePattern.create(url, nameValue[0], nameValue[1]);
+            if (pattern != null) {
+                pageCandidates.add(pattern,
+                        new PageLinkInfo(pageNum, pattern.getPageNumber(), posInAscendingNumbers));
             }
         }
-    }  // extractPageParamCandidatesFromQuery
+    }
 
     private static RegExp sDigitsRegExp = null;  // Match at least 1 digit.
 
@@ -141,12 +155,12 @@ public class PageParameterDetector {
      * A page parameter candidate is one where a path component contains consecutive digits which
      * can be converted to a plain number (>= 0).
      * E.g. a URL path with 3 path components that contain plain numbers will generate 3 URL page
-     * patterns with 3 LinkInfo's, and hence 3 page parameter candidates.
+     * patterns with 3 PageLinkInfo's, and hence 3 page parameter candidates.
      *
      * @param url ParsedUrl of the URL to process
      * @param pageNum the page number as represented in original plain text
      * @param posInAscendingNumbers position of this page number in the list of ascending numbers
-     * @param pageCandidates the map of URL pattern to its associated list of LinkInfo's
+     * @param pageCandidates the map of URL pattern to its associated list of PageLinkInfo's
      */
 
     private static void extractPageParamCandidatesFromPath(ParsedUrl url, int pageNum,
@@ -166,61 +180,29 @@ public class PageParameterDetector {
 
             final int matchEnd = sDigitsRegExp.getLastIndex();
             final int matchStart = matchEnd - match.getGroup(1).length();
-
-            if (isLastNumericPathComponentBad(urlStr, pathStart, matchStart, matchEnd)) continue;
-
-            int value = StringUtil.toNumber(urlStr.substring(matchStart, matchEnd));
-            if (value >= 0) {
-                pageCandidates.add(urlStr.substring(0, matchStart) + PAGE_PARAM_PLACEHOLDER +
-                        urlStr.substring(matchEnd),
-                        new LinkInfo(pageNum, value, posInAscendingNumbers));
+            PagePattern pattern = PathComponentPagePattern.create(url, pathStart, matchStart,
+                    matchEnd);
+            if (pattern != null) {
+                pageCandidates.add(pattern,
+                        new PageLinkInfo(pageNum, pattern.getPageNumber(), posInAscendingNumbers));
             }
         }  // while there're matches
-    }  // extractPageParamCandidatesFromPath
+    }
 
     /**
      * Returns true if given name is backlisted as a known bad page param name.
      */
-    private static boolean isPageParamNameBad(String name) {
+    static boolean isPageParamNameBad(String name) {
         initBadPageParamNames();
         return sBadPageParamNames.contains(name.toLowerCase());
-    }  // isPageParamNameBad
-
-    private static RegExp sExtRegExp = null;  // Match trailing .(s)htm(l).
-    private static RegExp sLastPathComponentRegExp = null;  // Match last path component.
+    }
 
     /**
-     * Returns true if:
-     * - the digitStart to digitEnd of urlStr is the last path component, and
-     * - the entire path component is numeric, and
-     * - the previous path component is a bad page param name.
-     * E.g. "www.foo.com/tag/2" will return true because of the above reasons and "tag" is a bad
-     * page param.
+     * Returns true if given string can be converted to a number >= 0.
      */
-    static boolean isLastNumericPathComponentBad(String urlStr, int pathStart,
-            int digitStart, int digitEnd) {
-        if (urlStr.charAt(digitStart - 1) == '/' && // Digit is at start of path component.
-                pathStart < digitStart - 1) { // Not the first path component.
-            String postMatch = urlStr.substring(digitEnd).toLowerCase();
-            // Checks that this is the last path component, and trailing characters, if available,
-            // are (s)htm(l) extensions.
-            if (sExtRegExp == null) sExtRegExp = RegExp.compile("(.s?html?)?$", "i");
-            if (sExtRegExp.test(postMatch)) {
-                // Entire component is numeric, get previous path component.
-               if (sLastPathComponentRegExp == null) {
-                   sLastPathComponentRegExp = RegExp.compile("([^/]*)\\/$", "i");
-                }
-                MatchResult prevPathComponent = sLastPathComponentRegExp.exec(
-                        urlStr.substring(pathStart + 1, digitStart));
-                if (prevPathComponent != null && prevPathComponent.getGroupCount() > 1 &&
-                        isPageParamNameBad(prevPathComponent.getGroup(1))) {
-                    return true;
-                }
-            }  // last numeric path component
-        }
-
-        return false;
-    }  // isLastNumericPathComponentBad
+    static boolean isPlainNumber(String str) {
+        return StringUtil.toNumber(str) >= 0;
+    }
 
     /**
      * If sBadPageParamNames is null, initialize it with all the known bad page param names, in
@@ -259,6 +241,6 @@ public class PageParameterDetector {
         sBadPageParamNames.add("videos");
         sBadPageParamNames.add("w");
         sBadPageParamNames.add("wiki");
-    }  // initBadPageParamNames
+    }
 
 }