testsuite/src/com/piclab/wikitest/WikiSearchThread.java

   1
   2 /*
   3  * WikiSearchThread is a background thread that does searches.
   4  * See WikiFetchThread for more details on design.
   5  */
   6
   7 package com.piclab.wikitest;
   8 import com.meterware.httpunit.*;
   9
  10 public class WikiSearchThread extends Thread {
  11
  12 private WebConversation m_conv;
  13 private int m_totalsearches;
  14 private long m_totaltime;
  15 private volatile boolean m_running;
  16
  17 public WikiSearchThread() {
  18         m_conv = new WebConversation();
  19         m_totalsearches = 0;
  20         m_totaltime = 0;
  21 }
  22
  23 public int getSearches() { return m_totalsearches; }
  24 public long getTime() { return m_totaltime; }
  25 public void requestStop() { m_running = false; }
  26
  27 /*
  28  * First, a list of miscellaneous words that do appear in the
  29  * titles and text of the testing articles.
  30  */
  31
  32 public static String[] searchterms = {
  33         "agriculture", "husbandry", "corn", "vegetable", "farming", "nitrogen",
  34         "anthropology", "human", "culture", "language", "networks", "society",
  35         "archaeology", "archeology", "history", "marxism", "ownership", "land",
  36         "architecture", "building", "landscape", "furniture", "carpentry", "roman",
  37         "astronomy", "astrophysics", "astrology", "star", "cosmology", "galaxy",
  38         "biology", "life", "monet", "evolution", "species", "animal", "plant",
  39         "business", "industry", "capitalism", "commerce", "company", "corporation",
  40         "chemistry", "atom", "organic", "element", "alchemy", "polymer",
  41         "classics", "greece", "rome", "latin", "literature", "mythology", "art",
  42         "communication", "media", "television", "radio", "film", "mail",
  43         "computer", "engineering", "linguistics", "algorithm", "graphics", "logic",
  44         "cooking", "food", "heat", "cuisine", "ethnic", "nutrition", "flavor",
  45         "critical", "theory", "frankfurt", "postmodernism", "weber",
  46         "dance", "rhythm", "music", "recreation", "performance", "ballet",
  47         "earth", "science", "geology", "weather", "fossil", "ocean", "environment",
  48         "economic", "scarcity", "communism", "socialism", "utility", "money",
  49         "education", "teaching", "knowledge", "reading", "testing", "school",
  50         "technology", "civil", "mechanical", "nuclear", "process", "control",
  51         "entertainment", "animation", "sport", "humor", "illusion", "theater",
  52         "family", "consumer", "parenting", "sewing", "homemaker", "decoration",
  53         "movie", "cinema", "director", "actor", "genre", "studio", "festival",
  54         "game", "card", "board", "competition", "probability", "drinking", "dice",
  55         "geography", "map", "projection", "continent", "island", "river", "sea",
  56         "handicraft", "bead", "marquetry", "paper", "wood", "garden", "metal",
  57         "history", "etymology", "orthodox", "controversy", "pasteur", "method",
  58         "hobby", "pastime", "professional", "amateur", "collecting", "genealogy",
  59         "language", "othography", "writing", "alphabet", "phonetic", "speech",
  60         "law", "taboo", "more", "jurisdiction", "legislature", "judge", "police",
  61         "library", "information", "book", "journal", "pediodical", "database",
  62         "philology", "syntax", "semantic", "lexicology", "comparative", "cipher",
  63         "letter", "rhetoric", "bible", "poem", "novel", "epic", "essay", "drama",
  64         "math", "mathematics","statistics", "number", "algebra", "calculus",
  65         "music", "melody", "instrument", "ensemble", "orchestra", "harmony",
  66         "opera", "costume", "dialogue", "acting", "voice", "libretto", "stage",
  67         "painting", "glaze", "acrylic", "mural", "portrait", "canvas", "fresco",
  68         "philosophy", "concept", "dialectic", "beauty", "ethic", "aristotle",
  69         "physics", "matter", "space", "energy", "quantum", "particle", "momentum",
  70         "poker", "stud", "wager", "gambling", "joker", "flush", "chip", "deal",
  71         "political", "politics", "government", "violence", "democracy", "fascism",
  72         "psychology", "freud", "ethology", "medicine", "therapy", "drug", "health",
  73         "public", "policy", "activism", "defense", "tax", "administration",
  74         "recreation", "weekend", "holiday", "vacation", "leisure", "sex",
  75         "religion", "christianity", "judaism", "islam", "deity", "faith", "priest",
  76         "sculpture", "clay", "marble", "mobile", "kinetic", "statue", "bust",
  77         "sociology", "kinship", "criminology", "race", "revolution", "gender",
  78         "sport", "equipment", "injury", "spectator", "football", "baseball",
  79         "invention", "recording", "cryptography", "metallurgy", "hydraulic",
  80         "theatre", "mime", "tennessee", "lighting", "scenery", "improvisation",
  81         "tourism", "travel", "sightseeing", "hotel", "camping", "cruise",
  82         "transport", "vehicle", "airline", "train", "ferry", "subway", "car",
  83         "visual", "design", "photography", "fashion", "tattoo", "textile"
  84 };
  85
  86 /*
  87  * Then, a list of miscellaneous words that may or may not appear
  88  * in the test articles.
  89  */
  90
  91 public static String[] randomterms = {
  92         "abatement", "acacia", "aerate", "allergy", "anvil", "ashtray", "auger",
  93         "badger", "bakery", "benign", "biceps", "bookie", "brazen", "bulldog",
  94         "caliber", "castigate", "centipede", "chemise", "clamor", "cupboard",
  95         "dawn", "debris", "derrick", "dig", "divorce", "doublet", "drummer",
  96         "ebony", "eclipse", "elitist", "emulator", "escrow", "euphoria", "evade",
  97         "famine", "feedback", "fiefdom", "flax", "fox", "freckle", "funnel",
  98         "gallop", "ghetto", "gingham", "gnat", "gossip", "grudge", "guitar",
  99         "halogen", "hedgehog", "heuristic", "hillbilly", "hologram", "hyacinth",
 100         "idealism", "illustrator", "impeach", "income", "injunction", "irony",
 101         "janitor", "jellyfish", "jitterbug", "journalism", "juggling", "jury",
 102         "kamikaze", "kerosene", "kindergarten", "kitten", "klaxon", "knuckle",
 103         "lager", "leech", "lentil", "libido", "locust", "lox", "lullabye", "lyre",
 104         "magenta", "marigold", "mediator", "mileage", "monarch", "municipal",
 105         "navigation", "neurosis", "nicotine", "nostalgia", "nucleus", "nymph",
 106         "oasis", "obscene", "oilcloth", "oratory", "osmosis", "ovary", "owl",
 107         "parsley", "perplex", "phony", "pilgrim", "pliers", "pompadour", "prose",
 108         "quahog", "quaver", "quench", "queue", "quilt", "quince", "quotient",
 109         "railroad", "ravine", "recipe", "rescue", "rig", "roast", "ruthless",
 110         "sarcasm", "sclerosis", "sellout", "shanty", "sigma", "skyhook", "synod",
 111         "tantrum", "tenacious", "thorn", "tithe", "tonsil", "trauma", "tyranny",
 112         "ulcer", "umpire", "unicorn", "unravel", "urinate", "upload", "utensil",
 113         "valence", "veranda", "viewpoint", "volunteer", "vow", "vulnerable",
 114         "wanton", "welcome", "wharf", "whiz", "wilderness", "woofer", "wretch",
 115         "xanthate", "yeast", "yeoman", "yonder", "zealous", "zen", "zonal"
 116 };
 117
 118 public void run() {
 119         int tindex = 0, rindex = 0;
 120         double r;
 121         long start, end;
 122         WebResponse wr;
 123
 124         String prefix = WikiSuite.getServer() + WikiSuite.getScript() + "?search=";
 125         String term = null;
 126
 127         m_running = true;
 128         while ( m_running ) {
 129                 r = Math.random();
 130                 if ( r < 0.1 ) {
 131                         term = searchterms[tindex] + " AND " + randomterms[rindex];
 132                 } else if ( r < 0.3 ) {
 133                         term = searchterms[tindex];
 134                         if ( ++tindex >= searchterms.length ) { tindex = 0; }
 135                         term += " AND " + searchterms[tindex];
 136                 } else if ( r < 0.4 ) {
 137                         term = searchterms[tindex] + " OR " + randomterms[rindex];
 138                 } else if ( r < 0.5 ) {
 139                         term = randomterms[rindex];
 140                         if ( ++rindex >= randomterms.length ) { rindex = 0; }
 141                         term += " OR " + randomterms[rindex];
 142                 } else if ( r < 0.7 ) {
 143                         term = randomterms[rindex];
 144                 } else {
 145                         term = searchterms[tindex];
 146                 }
 147
 148                 start = System.currentTimeMillis();
 149                 try {
 150                         term = java.net.URLEncoder.encode( term, "UTF-8" );
 151                         wr = m_conv.getResponse( prefix + term );
 152                 } catch ( java.io.UnsupportedEncodingException e ) {
 153                         break;
 154                 } catch (Exception e) {
 155                         WikiSuite.warning( "Error (" + e + ") searching for \"" + term + "\"" );
 156                 }
 157                 end = System.currentTimeMillis();
 158
 159                 WikiSuite.finer( "Searched for \"" + term + "\"" );
 160                 ++m_totalsearches;
 161                 m_totaltime += ( end - start );
 162
 163                 if ( ++tindex >= searchterms.length ) { tindex = 0; }
 164                 if ( ++rindex >= randomterms.length ) { rindex = 0; }
 165
 166                 try {
 167                         Thread.sleep( 3000 );
 168                 } catch( InterruptedException e ) {
 169                         break;
 170                 }
 171         }
 172         synchronized (this) { notify(); }
 173 }
 174
 175 }