Changeset 33982

Timestamp:

2020-02-26T21:59:55+13:00 (4 years ago)

Author:

ak19

Message:

SummaryTool.java now processed the handcrafted UNIQUE domains counts file for the manually shortlisted domains where at least one webpage contained at least one genuine MRI sentence.

File:

• other-projects/maori-lang-detection/src/org/greenstone/atea/SummaryTool.java (modified) (4 diffs)

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

@@ -391 +391 @@
         //String containsMRIFile = listing.produceURLsForPagesContainingMRI(domainsFile);
 
-
-        // TODO: for NZ, do IS_MRI. For overseas still CONTAINS_MRI
-        // then also do the shuffle to gen X num of random web page URLs.
-        //String filename = listing.webPagesOfAllNZSitesAndDomainListing(domainsFile);
+        // 1. generate full listing of all web page URLs for all domains in input domain file
+        // 2. AND generate random listing of web page URLs from (1) of appropriate sample size
         listing.mriWebPageListingForDomainListing(domainsFile);
 
@@ -418 +416 @@
         // TODO: generate the tables
 
-        String[] tableFileNames = mongodb.writeTables(outFolder);
+        String[] tableFiles = mongodb.writeTables(outFolder);
+        //List tableFileNames = Arrays.asList(tableFiles);
+        List<String> tableFileNames = new ArrayList<String>();
+        for(int i = 0; i < tableFiles.length; i++) {
+            tableFileNames.add(tableFiles[i]);
+        }
+
+        // Add in the manually created counts (.json) file of manually shortlisted
+        // (unique) domains which genuinely contain at least one MRI sentence.
+        // It may not exist unless the user created this file by hand
+        // AND copied it into the outFolder, "mongodb-data-auto"
+        tableFileNames.add("6counts_sitesWithPagesContainingMRI_manualShortlist");
+            
+        
         // for each table file name, generate the geojson-features .json file
         // that GEOJSON_MAP_TOOL_URL takes as input to produce a map.
@@ -424 +435 @@
         ///CountryCodeCountsMapData.openFirefox();
         
-        for(int i = 1; i < tableFileNames.length; i++) { // empty element at 0
-            String tablefilename = tableFileNames[i] + ".json"; // filenames have no suffix
+        for(int i = 1; i < tableFileNames.size(); i++) { // empty element at 0
+            String tablefilename = tableFileNames.get(i) + ".json"; //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.");
+            logger.error("@@@ Can't generate map data for this.");
             continue;
             }
@@ -436 +447 @@
             CountryCodeCountsMapData mapData
             = new CountryCodeCountsMapData(countsTableFilename);
-            String geoJsonFilename = mapData.writeFeaturesGeoJsonToFile(CountryCodeCountsMapData.SUPPRESS_MAPDATA_DISPLAY);
-            
-            // Ensuring the geo-json file generated exists
-            //String geoJsonFilename = outFolder + File.separator
-            //+ GEOJSON_FEATURES_FILE_PREFIX + tablefilename;           
+            String geoJsonFilename = mapData.writeFeaturesGeoJsonToFile(CountryCodeCountsMapData.SUPPRESS_MAPDATA_DISPLAY);         
+                    
             File geoJsonFile = new File(geoJsonFilename);
             if(!geoJsonFile.exists()) {