From 01441fc9f21fb4e8f7fa0755bfcd868de7f8e564 Mon Sep 17 00:00:00 2001
From: richet <richetyann@gmail.com>
Date: Fri, 5 Feb 2021 16:21:54 +0100
Subject: [PATCH] imp. isBinary

---
 src/main/java/org/funz/util/Disk.java |  29 ++++++-
 src/test/java/org/funz/DiskTest.java  |  18 +++--
 src/test/resources/out.txt            | 105 ++++++++++++++++++++++++++
 3 files changed, 145 insertions(+), 7 deletions(-)
 create mode 100644 src/test/resources/out.txt

diff --git a/src/main/java/org/funz/util/Disk.java b/src/main/java/org/funz/util/Disk.java
index 03dc5a1..56b2587 100644
--- a/src/main/java/org/funz/util/Disk.java
+++ b/src/main/java/org/funz/util/Disk.java
@@ -19,6 +19,7 @@
 import java.util.List;
 import org.apache.commons.io.FileUtils;
 import org.funz.Protocol;
+import java.nio.charset.Charset;
 
 public class Disk {
 
@@ -89,6 +90,9 @@ protected static boolean isStringChar(char ch) {
         }
         switch (ch) {
             case ' ':
+            case '\t':
+            case '\r':
+            case '\n':
             case '/':
             case '+':
             case '*':
@@ -123,24 +127,45 @@ protected static boolean isStringChar(char ch) {
             case '\\':
                 return true;
         }
+        //System.out.println(""+ch+"");
         return false;
     }
 
     public static boolean isBinary(File f) {
+        for (Charset cs: new Charset[]{Charset.forName("UTF-8"),Charset.forName("ISO-8859-1"),Charset.forName("ISO-8859-15")}) {
+            // Consider as binary if no charset get it as ASCII...
+            if (!isBinary(f, cs)) return false;
+        }
+        return true;
+    }
+
+    public static boolean isBinary(File f, Charset inputCharset) {
         boolean isbin = false;
         java.io.InputStream in = null;
 
         try {
             in = new FileInputStream(f);
-            BufferedReader r = new BufferedReader(new InputStreamReader(in));
+            BufferedReader r = new BufferedReader(new InputStreamReader(in, inputCharset));
 
-            int sample = (int) Math.min(255, f.length());
+            int sample = (int) Math.min(1024, f.length());
 
             char[] cc = new char[sample]; //do a peek
             r.read(cc, 0, sample);
 
             double prob_bin = 0;
 
+//            int word_size = 10; // will consider word as binary if >50% chars are binary
+//            for (int i = 0; i < cc.length-word_size; i=i+word_size) {
+//                double bins = 0;
+//                for (int j=0; j<word_size; j++) {
+//                    if (!isStringChar((char) cc[i+j])) bins++;
+//                }
+//                if (bins/word_size > 0.5) {
+//                    String b = ">";
+//                    for (int j=0; j<word_size; j++) b=b+" "+(int)cc[i+j];
+//                    System.err.println(b+"<");
+//                }
+//            }
             for (int i = 0; i < cc.length; i++) {
                 char j = (char) cc[i];
                 if (!isStringChar(j)) {
diff --git a/src/test/java/org/funz/DiskTest.java b/src/test/java/org/funz/DiskTest.java
index dc26a8a..2e1e403 100644
--- a/src/test/java/org/funz/DiskTest.java
+++ b/src/test/java/org/funz/DiskTest.java
@@ -32,18 +32,26 @@ public void testlistRecursiveFiles() {
     @Test
     public void testIsBinary() {
         File java_src = new File("src/main/java/org/funz/util/Disk.java");
-        assert !Disk.isBinary(java_src) : "bad binary inference";
+        System.err.println(java_src);
+        assert !Disk.isBinary(java_src) : "bad binary inference: "+java_src;
         File java_class = new File("build/org/funz/util/Disk.class");
-        assert Disk.isBinary(java_class) : "bad binary inference";
+        System.err.println(java_class);
+        assert Disk.isBinary(java_class) : "bad binary inference: "+java_class;
         File R_src = new File("src/test/resources/branin.R");
-        assert !Disk.isBinary(R_src) : "bad binary inference";
+        System.err.println(R_src);
+        assert !Disk.isBinary(R_src) : "bad binary inference: "+R_src;
+        File castem_src = new File("src/test/resources/out.txt");
+        System.err.println(castem_src);
+        assert !Disk.isBinary(castem_src) : "bad binary inference: "+castem_src;
 
         if (OS.isFamilyWindows()) {
             File cmd = new File("C:\\Windows\\system32\\cmd.exe");
-            assert Disk.isBinary(cmd) : "bad binary inference";
+            System.err.println(cmd);
+            assert Disk.isBinary(cmd) : "bad binary inference: "+cmd;
         } else {
             File bash = new File("/bin/bash");
-            assert Disk.isBinary(bash) : "bad binary inference";
+            System.err.println(bash);
+            assert Disk.isBinary(bash) : "bad binary inference: "+bash;
         }
     }
 }
diff --git a/src/test/resources/out.txt b/src/test/resources/out.txt
new file mode 100644
index 0000000..0ee8a79
--- /dev/null
+++ b/src/test/resources/out.txt
@@ -0,0 +1,105 @@
+SBATCH: -p seq -n 1 -t 00-01:00:00
+0:00
+Exit code: 0
+/12
+ ________________________________________________________________________
+  
+  
+ Documentation Generale sur Cast3M :
+   http://www-cast3m.cea.fr/index.php?xml=maj2011
+  
+ Nouveautes de Cast3M 2016 :
+   http://www-cast3m.cea.fr/html/ClubCast3m/club2016/
+          Club_Cast3m_2016_Millard.pdf
+  
+ ________________________________________________________________________
+  
+ $  * *SBATCH -p seq 
+ $  * *SBATCH -n 1 
+ $  * *SBATCH -t 00-01:00:00 
+ $  * * POUTRE CONSOLE EN FLEXION SIMPLE 
+ $  * * 
+ $  * *                          | F 
+ $  * *                          | 
+ $  * *  |                       | 
+ $  * *  |P1                     v P2 
+ $  * *  |------------------------ 
+ $  * *  | 
+ $  * *  | 
+ $  * * 
+ $  * * Formule analytique de la flèche en fonction de l'effort : 
+ $  * * u = F*L^3/(3*E*I) 
+ $  * * où F est la force, L la longueur, 
+ $  * *    E le module de Young (210e9 pour l'acier), 
+ $  * *    I le moment d'inertie de la section de la poutre (largeur*hauteur^3/12) 
+ $  *   
+ $  * * OPTIONS DU CALCUL ******************************************** 
+ $  *   
+ $  * * Dimension 2 et éléments finis de type SEG2 
+ $  * OPTI DIME 2 ELEM SEG2; 
+ $  *   
+ $  * * ACQUISITION DES VARIABLES D'ENTREE *************************** 
+ $  *   
+ $  * ** OPTI ACQU 'input.txt'; 
+ $  * ** ACQU long*FLOTTANT haut*FLOTTANT larg*FLOTTANT F*FLOTTANT; 
+ $  * long = 0.1; 
+ $  * haut = 0.001; 
+ $  * larg = 0.01; 
+ $  * F = 1.; 
+ $  *   
+ $  * * Section 
+ $  * sect = haut*larg; 
+ $  * * Inertie 
+ $  * inrz = (larg*(haut**3))/12.; 
+ $  *   
+ $  * * MAILLAGE ****************************************************** 
+ $  *   
+ $  * * Création des noeuds 
+ $  * P1 = 0. 0.; 
+ $  * P2 = long 0.; 
+ $  * * Création de l'élément fini 
+ $  * poutre = DROI 1 P1 P2; 
+ $  *   
+ $  * * MODELE ET MATERIAU ******************************************** 
+ $  *   
+ $  * mo_pout = MODE poutre 'MECANIQUE' 'ELASTIQUE' 'ISOTROPE' 'POUT'; 
+ $  * ma_pout = MATE mo_pout 'YOUN' 210e9 'NU' 0.3 
+ $  * 	'SECT' sect 'INRZ' inrz; 
+ $  *   
+ $  * * CONDITIONS AUX LIMITES **************************************** 
+ $  *   
+ $  * * Encastrement du point P1 
+ $  * enca = BLOQ 'DEPL' 'ROTA' P1; 
+ $  *   
+ $  * * Calcul de la matrice de rigidité 
+ $  * rigi1 = RIGI mo_pout ma_pout; 
+ $  * rigitot = enca et rigi1; 
+ $  *   
+ $  * * Chargement 
+ $  * char1 = FORC 'FY' (-1.*F) P2; 
+ $  *   
+ $  * * CALCUL ******************************************************** 
+ $  *   
+ $  * * Résolution 
+ $  * dep1 = RESO rigitot char1; 
+ Taille de la matrice: 6 Facteur: 1.3867     Performance (Gflop/s): 0.0000
+ $  *   
+ $  * * Extraction du déplacement au point d'intérêt 
+ $  * dep_P2 = EXTR dep1 'UY' P2; 
+ $  *   
+ $  * * Export du résultat dans un fichier texte 
+ $  * ** OPTI SORT 'output.txt'; 
+ $  * ** SORT 'CHAI' dep_P2; 
+ $  * MESS 'dep_P2=' dep_P2; 
+ dep_P2=-1.90476E-03
+ $  *   
+ $  * * Remarque : 
+ $  * * si dep_P2 était une liste (PROG) il faudrait faire 
+ $  * * SORT 'EXCE' dep_P2; 
+ $  *   
+ $  * *  FIN DE l'EXECUTION ******************************************* 
+ $  *   
+ $  * FIN; 
+  
+__________________________________________________________________________
+