Changeset 33919 for other-projects/maori-lang-detection/src/org/greenstone/atea/SummaryTool.java

Timestamp:

2020-02-13T22:40:41+13:00 (4 years ago)

Author:

ak19

Message:

SummaryTool now uses the CountryCodeCountsMapData.java class to generate the geojson-features files from the tables it already created using MongoDB query results. Switched over from geojson.tools to geojson.io since the latter allows passing geojson mapdata in the URL. The firefox screenshotting is still not working. But I can't even get complex geojson features to work from the commandline yet, so then there's another possible layer of complexity when running firefox as a Java process. Added jna jar files used by Greenstone's SafeProcess for launching Firefox as a Java process.

File:

• other-projects/maori-lang-detection/src/org/greenstone/atea/SummaryTool.java (modified) (7 diffs)

other-projects/maori-lang-detection/src/org/greenstone/atea/SummaryTool.java

@@ -8 +8 @@
 /**
  * Runs some of the important mongoDB queries I ran.
+ *
+ * This program expects a folder ../mongo-data-auto to exist.
  *
  * TO COMPILE OR RUN, FIRST DO:
@@ -26 +28 @@
 */
 public class SummaryTool {
+    
     static Logger logger = Logger.getLogger(org.greenstone.atea.SummaryTool.class.getName());
+
+    
+    static private final String GEOJSON_FEATURES_FILE_PREFIX = "geojson-features_";
+    
     static private final long FIXED_SEED = 1000;
     
     private final MongoDBQueryer mongodbQueryer;
     private File outFolder;
-
 
     
@@ -101 +107 @@
             // Print out whether there were no isMRI pages for the domain (only containsMRI). A useful thing to know
             if(moreURLs.size() == 0 && filterType == MongoDBQueryer.IS_MRI) {
-            System.out.println("   " + countryCode + " domain " + domain + " had no isMRI webpages - only containsMRI.");
+            System.out.println("   " + countryCode + " domain " + domain + " had no webpages where isMRI=true - only containsMRI.");
             }
 
@@ -149 +155 @@
     //File outFolder = domainsFile.getParentFile();
     String filterName = (filterType == MongoDBQueryer.IS_MRI) ? "isMRI" : "containsMRI";
-    File outFile = new File(outFolder, filterName+"_"+domainsFile.getName());
+    File outFile = new File(outFolder, filterName+"_full_"+domainsFile.getName());
 
     writeURLsToFile(urlsList, outFile, N_totalNumPages);
@@ -158 +164 @@
     int n_numSampleURLs = calcSampleSize(N_totalNumPages);
 
-    System.err.println("*** N, total number of web pages for which " + filterName + "=true from domain shortlist: " + N_totalNumPages);
+    System.err.println("*** N, total number of web pages for which " + filterName + "=true from domain shortlist file: " + N_totalNumPages);
     System.err.println("    (out of " + mongodbQueryer.countOfWebpagesMatching(filterType)
                + " web pages across ALL sites for which " + filterName + " = true)");
@@ -172 +178 @@
     System.out.println("Wrote a sample of n=" + n_numSampleURLs + " of web page URLs "
                + "for the sites in input domainsFile\ninto file: " + Utility.getFilePath(outFile));
+
+    // For N = 6557, z-alpha-over-2 = 1.6449 and m = 0.05 (5%),
+    // n = (z-alpha-over-2^2 x N) / (z-alpha-over-2^2 + 4 x (N-1) x m^2)
+    // = (1.6449^2×6557) ÷ (1.6449^2 + 4 × 6556×0.05^2) = 259.88526851 => 260 rounded up. Check.
     }
 
@@ -384 +394 @@
         // TODO: generate the tables
 
-        mongodb.writeTables(outFolder);
+        String[] tableFileNames = mongodb.writeTables(outFolder);
+        // for each table file name, generate the geojson-features .json file
+        // that GEOJSON_MAP_TOOL_URL takes as input to produce a map.
+
+        for(int i = 1; i < tableFileNames.length; i++) { // empty element at 0
+            String tablefilename = tableFileNames[i] + ".json"; // filenames have no suffix
+            
+            File countsTableFile = new File(outFolder, tablefilename);
+            if(!countsTableFile.exists()) {
+            logger.error("@@@ File " + countsTableFile + " does not exist!");
+            logger.error("@@@ Can't generate map date for this.");
+            continue;
+            }
+            String countsTableFilename = outFolder + File.separator + tablefilename;
+            CountryCodeCountsMapData mapData
+            = new CountryCodeCountsMapData(countsTableFilename);
+            String geoJsonFilename = mapData.writeFeaturesGeoJsonToFile(CountryCodeCountsMapData.SUPPRESS_MAPDATA_DISPLAY);
+            
+
+            /*
+            // Ensure the geo-json file generated exists
+            //String geoJsonFilename = outFolder + File.separator
+            //+ GEOJSON_FEATURES_FILE_PREFIX + tablefilename;
+            
+            File geoJsonFile = new File(geoJsonFilename);
+            if(!geoJsonFile.exists()) {
+            System.err.println("@@@ geoJson file " + geoJsonFilename + " not generated!");
+            continue;
+            }
+            */
+            System.err.println("**** Wrote mapdata to file " + geoJsonFilename);
+            //System.err.println("     Paste the file's geojson contents into "
+            //+ "the editor at " + CountryCodeCountsMapData.GEOJSON_MAP_TOOL_URL
+            //+ " to see the data arranged on a world map");            
+            System.err.println("Total count for query: " + mapData.getTotalCount());
+            
+            /*boolean uriEncoded = true;
+            String mapDataEncodedStr = mapData.getFeaturesGeoJsonString(uriEncoded);
+            System.err.println("Encoded string: " + mapDataEncodedStr);
+            */
+            
+            System.err.println("Data URL string: " + mapData.getAsMapURL());
+            System.err.println();
+            mapData.geoJsonMapScreenshot(outFolder, tablefilename);
+            System.err.println("---");
+
+            // TODO: breaks after first table -> map conversion
+            break;
+        }
+        
         }