From f552f2d4de965e61c0f0ad762041124bf5c13a12 Mon Sep 17 00:00:00 2001
From: Roman Chyla <roman.chyla@gmail.com>
Date: Thu, 17 Feb 2022 14:11:40 -0500
Subject: [PATCH] feat: use proper generics

---
 .../author/AuthorQueryVariations.java         | 342 +++++++++---------
 .../solr/analysis/author/AuthorUtils.java     |  10 +-
 .../flexible/aqp/TestAqpExtendedLGSimple.java |   6 +-
 3 files changed, 174 insertions(+), 184 deletions(-)

diff --git a/contrib/adsabs/src/java/org/apache/solr/analysis/author/AuthorQueryVariations.java b/contrib/adsabs/src/java/org/apache/solr/analysis/author/AuthorQueryVariations.java
index 9fa001d11..11cc5b33f 100644
--- a/contrib/adsabs/src/java/org/apache/solr/analysis/author/AuthorQueryVariations.java
+++ b/contrib/adsabs/src/java/org/apache/solr/analysis/author/AuthorQueryVariations.java
@@ -1,191 +1,181 @@
 package org.apache.solr.analysis.author;
 
-
 import java.util.HashMap;
 import java.util.HashSet;
 import java.util.LinkedHashSet;
+import java.util.Map;
+import java.util.Set;
 
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
 /**
- * Creates a variations of the author names, it receives eg:
- * John, K
+ * Creates a variations of the author names, it receives eg: John, K
  */
 
 public class AuthorQueryVariations {
 
-  public static final Logger log = LoggerFactory.getLogger(AuthorQueryVariations.class);
-
-  /**
-   * This method takes input string, e.g. "Hector, Gomez Q" and generates variations
-   * of the author name.
-   * 
-   * <pre>
-   *  HECTOR, GOMEZ
-   *	HECTOR, G
-   *  HECTOR,
-   *  HECTOR, GOMEZ Q*
-   *  HECTOR, G Q*
-   * </pre>
-   * 
-   * It is essentially the same output as you get from getSynonymVariations except
-   * a few special cases. These special cases are variations needed for querying the 
-   * index of author names, but not needed or wanted for the process of transforming
-   * the curated synonyms
-   * <p>
-   * Example "h quintero gomez" will output:
-   * <pre>
-   * GOMEZ, H\w* QUINTERO\b.*
-   * GOMEZ,
-   * GOMEZ, H\w*
-   * GOMEZ, H\w* Q\b.*   &lt;-- only this one is extra added
-   * </pre>
-   * 
-   * @param authorString
-   *      author name
-   * @return map with string mappings
-   */
-  public static HashSet<String> getQueryVariationsInclRegex(String authorString) {
-
-    HashMap<String,String> parsedAuthor = null;
-    parsedAuthor = AuthorUtils.parseAuthor(authorString);
-
-    HashSet<String> variations = new LinkedHashSet<String>();
-    if (parsedAuthor == null) {
-      variations.add(authorString);
-      return variations;
-    }
-    generateNameVariations(parsedAuthor, variations);
-    return variations;
-  }
-
-  public static HashSet<String> generateNameVariations(HashMap<String,String> parsedAuthor) {
-    HashSet<String> variations = new LinkedHashSet<String>();
-    return generateNameVariations(parsedAuthor, variations);
-  }
-
-  protected static HashSet<String> generateNameVariations(
-      HashMap<String,String> parsedAuthor,
-      HashSet<String> variations) {
-
-    // get the base variations
-    generateSynonymVariations(parsedAuthor, variations);
-
-    // add the variations that are needed only for the query phase
-    String last = parsedAuthor.get("last");
-    String first = parsedAuthor.get("first");
-    String middle = parsedAuthor.get("middle");
-
-    if (first != null) {
-      if (middle != null) {
-        if (first.length() > 1) {
-          if (middle.length() > 1) {
-            variations.add(last + ", " + first + " " + middle.substring(0,1) + "\\b.*");
-          } else {
-            variations.add(last + ", " + first + " " + middle + ".*");
-            variations.add(last + ", " + first.substring(0,1) + " " + middle + ".*");
-          }
-        } else {
-          if (middle.length() > 1) {
-            variations.add(last + ", " + first + "\\w* " + middle.substring(0,1) + "\\b.*");
-          }
-        }
-      } else {
-        if (first.length() > 1) {
-          variations.add(last + ", " + first.substring(0,1) + "\\b.*");
-        }
-      }
-    }
-
-    return variations;
-  }
-
-
-
-
-  /**
-   * This method takes input string, e.g. "Hector, Gomez Q" and generates variations
-   * of the author name PLUS enhances the variations with regular expression patterns.
-   * 
-   * The process that transforms the curated synonyms uses *only* the variations
-   * generated here. This limited set is also included in the variations used at query
-   * time but DON'T ADD THINGS HERE that are only necessary for the query phase--use
-   * getNameVariations for that
-   * 
-   *  HECTOR, GOMEZ
-   *	HECTOR, G
-   *  HECTOR,
-   *  HECTOR, GOMEZ Q*
-   *  HECTOR, G Q*
-   * 
-   *  
-   * @param authorString
-   *    name in the natural form
-   * @return map with string mappings
-   */
-  public static HashSet<String> getQueryVariations(String authorString) {
-
-    HashMap<String,String> parsedAuthor = null;
-    parsedAuthor = AuthorUtils.parseAuthor(authorString);
-
-    HashSet<String> variations = new LinkedHashSet<String>();
-    if (parsedAuthor == null) {
-      variations.add(authorString);
-      return variations;
-    }
-    return generateSynonymVariations(parsedAuthor, variations);
-  }
-
-  public static HashSet<String> generateSynonymVariations(HashMap<String,String> parsedAuthor) {
-    HashSet<String> variations = new LinkedHashSet<String>();
-    return generateSynonymVariations(parsedAuthor, variations);
-  }
-
-  protected static HashSet<String> generateSynonymVariations(
-      HashMap<String,String> parsedAuthor,
-      HashSet<String> variations) {
-
-    String last = parsedAuthor.get("Last");
-    String first = parsedAuthor.get("First");
-    String middle = parsedAuthor.get("Middle");
-
-    if (parsedAuthor.size() == 1 && last != null) {
-      variations.add(String.format("%s,.*", last)); // all we got was last name
-    } else {
-      variations.add(String.format("%s,", last));
-    }
-
-    if (first != null) {
-      if (middle != null) {
-        if (first.length() > 1) {
-          variations.add(last + ", " + first);
-          variations.add(last + ", " + first.substring(0,1));
-          if (middle.length() > 1) {
-            variations.add(last + ", " + first + " " + middle + "\\b.*");
-            variations.add(last + ", " + first.substring(0,1) + " " + middle.substring(0,1) + "\\b.*");
-          } else if (middle.length() == 1) {
-            //						variations.add(last + ", " + first.substring(0,1) + " " + middle + ".*");
-            //						variations.add(last + ", " + first + " " + middle + ".*");
-          }
-        } else {
-          variations.add(last + ", " + first + "\\w*");
-          if (middle.length() > 1) {
-            variations.add(last + ", " + first + "\\w* " + middle + "\\b.*");
-          } else if (middle.length() == 1) {
-            variations.add(last + ", " + first + "\\w* " + middle + ".*");
-          }
-        }
-      } else {
-        if (first.length() > 1) {
-          variations.add(last + ", " + first + "\\b.*");
-          variations.add(last + ", " + first.substring(0,1));
-        } else if (first.length() == 1) {
-          variations.add(last + ", " + first + ".*");
-        }
-      }
-    }
-
-    return variations;
-  }
+	public static final Logger log = LoggerFactory.getLogger(AuthorQueryVariations.class);
+
+	/**
+	 * This method takes input string, e.g. "Hector, Gomez Q" and generates
+	 * variations of the author name.
+	 * 
+	 * <pre>
+	 *  HECTOR, GOMEZ
+	 *	HECTOR, G
+	 *  HECTOR,
+	 *  HECTOR, GOMEZ Q*
+	 *  HECTOR, G Q*
+	 * </pre>
+	 * 
+	 * It is essentially the same output as you get from getSynonymVariations except
+	 * a few special cases. These special cases are variations needed for querying
+	 * the index of author names, but not needed or wanted for the process of
+	 * transforming the curated synonyms
+	 * <p>
+	 * Example "h quintero gomez" will output:
+	 * 
+	 * <pre>
+	 * GOMEZ, H\w* QUINTERO\b.*
+	 * GOMEZ,
+	 * GOMEZ, H\w*
+	 * GOMEZ, H\w* Q\b.*   &lt;-- only this one is extra added
+	 * </pre>
+	 * 
+	 * @param authorString author name
+	 * @return map with string mappings
+	 */
+	public static HashSet<String> getQueryVariationsInclRegex(String authorString) {
+
+		Map<String, String> parsedAuthor = null;
+		parsedAuthor = AuthorUtils.parseAuthor(authorString);
+
+		HashSet<String> variations = new LinkedHashSet<String>();
+		if (parsedAuthor == null) {
+			variations.add(authorString);
+			return variations;
+		}
+		generateNameVariations(parsedAuthor, variations);
+		return variations;
+	}
+
+	public static Set<String> generateNameVariations(HashMap<String, String> parsedAuthor) {
+		HashSet<String> variations = new LinkedHashSet<String>();
+		return generateNameVariations(parsedAuthor, variations);
+	}
+
+	protected static Set<String> generateNameVariations(Map<String, String> parsedAuthor, Set<String> variations) {
+
+		// get the base variations
+		generateSynonymVariations(parsedAuthor, variations);
+
+		// add the variations that are needed only for the query phase
+		String last = parsedAuthor.get("last");
+		String first = parsedAuthor.get("first");
+		String middle = parsedAuthor.get("middle");
+
+		if (first != null) {
+			if (middle != null) {
+				if (first.length() > 1) {
+					if (middle.length() > 1) {
+						variations.add(last + ", " + first + " " + middle.substring(0, 1) + "\\b.*");
+					} else {
+						variations.add(last + ", " + first + " " + middle + ".*");
+						variations.add(last + ", " + first.substring(0, 1) + " " + middle + ".*");
+					}
+				} else {
+					if (middle.length() > 1) {
+						variations.add(last + ", " + first + "\\w* " + middle.substring(0, 1) + "\\b.*");
+					}
+				}
+			} else {
+				if (first.length() > 1) {
+					variations.add(last + ", " + first.substring(0, 1) + "\\b.*");
+				}
+			}
+		}
+
+		return variations;
+	}
+
+	/**
+	 * This method takes input string, e.g. "Hector, Gomez Q" and generates
+	 * variations of the author name PLUS enhances the variations with regular
+	 * expression patterns.
+	 * 
+	 * The process that transforms the curated synonyms uses *only* the variations
+	 * generated here. This limited set is also included in the variations used at
+	 * query time but DON'T ADD THINGS HERE that are only necessary for the query
+	 * phase--use getNameVariations for that
+	 * 
+	 * HECTOR, GOMEZ HECTOR, G HECTOR, HECTOR, GOMEZ Q* HECTOR, G Q*
+	 * 
+	 * 
+	 * @param authorString name in the natural form
+	 * @return map with string mappings
+	 */
+	public static HashSet<String> getQueryVariations(String authorString) {
+
+		Map<String, String> parsedAuthor = null;
+		parsedAuthor = AuthorUtils.parseAuthor(authorString);
+
+		HashSet<String> variations = new LinkedHashSet<String>();
+		if (parsedAuthor == null) {
+			variations.add(authorString);
+			return variations;
+		}
+		return generateSynonymVariations(parsedAuthor, variations);
+	}
+
+	public static HashSet<String> generateSynonymVariations(HashMap<String, String> parsedAuthor) {
+		HashSet<String> variations = new LinkedHashSet<String>();
+		return generateSynonymVariations(parsedAuthor, variations);
+	}
+
+	protected static HashSet<String> generateSynonymVariations(Map<String, String> parsedAuthor,
+			Set<String> variations) {
+
+		String last = parsedAuthor.get("Last");
+		String first = parsedAuthor.get("First");
+		String middle = parsedAuthor.get("Middle");
+
+		if (parsedAuthor.size() == 1 && last != null) {
+			variations.add(String.format("%s,.*", last)); // all we got was last name
+		} else {
+			variations.add(String.format("%s,", last));
+		}
+
+		if (first != null) {
+			if (middle != null) {
+				if (first.length() > 1) {
+					variations.add(last + ", " + first);
+					variations.add(last + ", " + first.substring(0, 1));
+					if (middle.length() > 1) {
+						variations.add(last + ", " + first + " " + middle + "\\b.*");
+						variations.add(last + ", " + first.substring(0, 1) + " " + middle.substring(0, 1) + "\\b.*");
+					} else if (middle.length() == 1) {
+						// variations.add(last + ", " + first.substring(0,1) + " " + middle + ".*");
+						// variations.add(last + ", " + first + " " + middle + ".*");
+					}
+				} else {
+					variations.add(last + ", " + first + "\\w*");
+					if (middle.length() > 1) {
+						variations.add(last + ", " + first + "\\w* " + middle + "\\b.*");
+					} else if (middle.length() == 1) {
+						variations.add(last + ", " + first + "\\w* " + middle + ".*");
+					}
+				}
+			} else {
+				if (first.length() > 1) {
+					variations.add(last + ", " + first + "\\b.*");
+					variations.add(last + ", " + first.substring(0, 1));
+				} else if (first.length() == 1) {
+					variations.add(last + ", " + first + ".*");
+				}
+			}
+		}
+
+		return variations;
+	}
 }
diff --git a/contrib/adsabs/src/java/org/apache/solr/analysis/author/AuthorUtils.java b/contrib/adsabs/src/java/org/apache/solr/analysis/author/AuthorUtils.java
index b79a5dec0..bc6c1df28 100644
--- a/contrib/adsabs/src/java/org/apache/solr/analysis/author/AuthorUtils.java
+++ b/contrib/adsabs/src/java/org/apache/solr/analysis/author/AuthorUtils.java
@@ -52,8 +52,8 @@ public static String normalizeAuthor(String a) {
 	/**
 	 * Method used by the tokenizer chain to normalize author names.
 	 * 
-	 * @param a
-	 * @param keepApostrophe
+	 * @param a - author name
+	 * @param keepApostrophe - bool, preserve apostrophe
 	 * @return Normalized string
 	 */
 	public static String normalizeAuthor(String a, boolean keepApostrophe) {
@@ -82,7 +82,7 @@ public static String normalizeAuthor(String a, boolean keepApostrophe) {
 	/**
 	 * Utility method to split string (author name) into constituting parts
 	 * 
-	 * @param a
+	 * @param a - author name
 	 * @return map with 'last', 'first', 'middle' keys
 	 */
 	public static Map<String,String> parseAuthor(String a) {
@@ -102,8 +102,8 @@ public static Map<String,String> parseAuthor(String a, boolean normalize) {
 	 * Utility method employed by AuthorTransliterationTokenizer and also by other components
 	 * inside the parser chain to discover other potential reading of the author's name.
 	 * 
-	 * @param a
-	 * @return
+	 * @param a - author name as string
+	 * @return - list of author name variants in ascii form
 	 */
 	public static ArrayList<String> getAsciiTransliteratedVariants(String a) {
 		HashSet<String> synonyms = new HashSet<String>();
diff --git a/contrib/antlrqueryparser/src/test/org/apache/lucene/queryparser/flexible/aqp/TestAqpExtendedLGSimple.java b/contrib/antlrqueryparser/src/test/org/apache/lucene/queryparser/flexible/aqp/TestAqpExtendedLGSimple.java
index 59c7ec3a5..163c14736 100644
--- a/contrib/antlrqueryparser/src/test/org/apache/lucene/queryparser/flexible/aqp/TestAqpExtendedLGSimple.java
+++ b/contrib/antlrqueryparser/src/test/org/apache/lucene/queryparser/flexible/aqp/TestAqpExtendedLGSimple.java
@@ -36,13 +36,13 @@ public void testExtensions() throws Exception {
     ((AqpQueryTreeBuilder) qp.getQueryBuilder()).setBuilder(AqpNearQueryNode.class,	new AqpNearQueryNodeBuilder());
     
     assertQueryMatch(qp, "this NEAR that", "field",
-        "spanNear([field:this, field:that], 5, true)");
+        "spanNear([field:this, field:that], 5, false)");
 
     assertQueryMatch(qp, "this NEAR3 that", "field",
-    		"spanNear([field:this, field:that], 3, true)");
+    		"spanNear([field:this, field:that], 3, false)");
     
     assertQueryMatch(qp, "this NEAR3 (that OR foo*)", "field",
-				"spanNear([field:this, spanOr([field:that, SpanMultiTermQueryWrapper(field:foo*)])], 3, true)");
+				"spanNear([field:this, spanOr([field:that, SpanMultiTermQueryWrapper(field:foo*)])], 3, false)");
 
     
   }