A bunch of stuff on Str

* Fixed __repr__ (somecases missing)
* Added more line breaking chars for splitlines method
* Added a new method to help check if a char is printable or not

Signed-off-by: Sergio Oliveira <[email protected]>

python/common/org/python/types/Str.java

@@ -66,14 +66,76 @@ public Str(org.python.Object[] args, java.util.Map<java.lang.String, org.python.
     //     throw new org.python.exceptions.NotImplementedError("__init__() has not been implemented.");
     // }
 
+    private static boolean isCharPrintable(char c) {
+        // ASCII non-printable
+        if ((int) c <= 0x1f || (int) c >= 0x7f && (int) c <= 0xa0 || (int) c == 0xad) {
+            return false;
+        }
+        if ((int) c == 0x2029) {
+            return false;
+        }
+        if (Character.isISOControl(c)) {
+            return false;
+        }
+        return true;
+    }
+
     @org.python.Method(
             __doc__ = ""
     )
     public org.python.Object __repr__() {
-        String repr = this.value.replaceAll("\\n", "\\\\n").replaceAll("\\r", "\\\\r");
-        return new org.python.types.Str("'" + repr + "'");
+    /*
+    * Reference: https://www.python.org/dev/peps/pep-3138/#id7
+    * TODO: Need to treat the leading surrogate pair characters
+    */
+        StringBuilder sb = new StringBuilder();
+        boolean has_double_quote = false;
+        boolean has_single_quote = false;
+
+        for (char c : this.value.toCharArray()) {
+            if (c == '\'') {
+                has_single_quote = true;
+            } else if (c == '"') {
+                has_double_quote = true;
+            }
+
+            if (c == '\n') {
+                sb.append("\\n");
+            } else if (c == '\t') {
+                sb.append("\\t");
+            } else if (c == '\r') {
+                sb.append("\\r");
+            } else if (c == '\\') {
+                sb.append("\\\\");
+            // ASCII Non-Printable
+            } else if (c <= 0x1f || c >= 0x7f && c <= 0xa0 || c == 0xad) {
+                sb.append(String.format("\\x%02x", (int) c));
+            } else if (!this.isCharPrintable(c)) {
+                sb.append(String.format("\\u%04x", (int) c));
+            } else {
+                sb.append((char) c);
+            }
+        }
+
+        // Decide if we wanna wrap the result with single or double quotes
+        String quote;
+        String repr = sb.toString();
+
+        if (has_single_quote) {
+            if (has_double_quote) {
+                quote = new String("'");
+                repr = repr.replaceAll("'", "\\\\'");
+            } else {
+                quote = new String("\"");
+            }
+        } else {
+            quote = new String("'");
+        }
+
+        return new org.python.types.Str(quote + repr + quote);
     }
 
+
     @org.python.Method(
             __doc__ = ""
     )
@@ -900,7 +962,7 @@ public org.python.Object isnumeric() {
     )
     public org.python.Object isprintable() {
         for (char ch : this.value.toCharArray()) {
-            if (Character.isISOControl(ch)) {
+            if (!this.isCharPrintable(ch)) {
                 return new org.python.types.Bool(false);
             }
         }
@@ -1386,6 +1448,25 @@ public org.python.Object split(org.python.Object sep, org.python.Object maxsplit
         return result_list;
     }
 
+    private static boolean isLineBreak(char character) {
+        // List of line boundaries from https://docs.python.org/3.4/library/stdtypes.html#str.splitlines
+        switch (character) {
+            case '\n':
+            case '\r':
+            case '\u000B':
+            case '\u000C':
+            case '\u001C':
+            case '\u001D':
+            case '\u001E':
+            case '\u0085':
+            case '\u2028':
+            case '\u2029':
+                return true;
+            default:
+                return false;
+        }
+    }
+
     @org.python.Method(
             __doc__ = "S.splitlines([keepends]) -> list of strings\n" +
                     "\n" +
@@ -1415,7 +1496,7 @@ public org.python.Object splitlines(org.python.Object keepends) {
                 next = this.value.charAt(i + 1);
             }
 
-            if (current == '\n' || current == '\r') {
+            if (this.isLineBreak(current)) {
                 end = i;
                 if (current == '\r' && next == '\n') {
                     skip = true;

tests/datatypes/test_str.py

@@ -1,3 +1,6 @@
+
+from unittest import expectedFailure
+
 from .. utils import TranspileTestCase, UnaryOperationTestCase, BinaryOperationTestCase, InplaceOperationTestCase
 
 
@@ -697,25 +700,38 @@ def test_isidentifier(self):
 
     def test_isprintable(self):
         self.assertCodeExecution("""
-        for str_ in [chr(i) for i in range(33)] + ['AAA', 'bcd', '1234', 'eÃⱣỉ', 'ÃⱣỉ', '', '\x07' + 'foo']:
+        for str_ in [chr(i) for i in range(33)] + ['AAA', 'bcd', '1234', 'eÃⱣỉ', 'ÃⱣỉ', '', '\x07' + 'foo', '\u2029']:
             print(str_.isprintable())
             """)
 
+    @expectedFailure
+    def test_isprintable_missing_cases(self):
+        self.assertCodeExecution(r"""
+        tests = ['\u2028']:
+        for test in tests:
+            print(test.isprintable())
+        """)
+
     def test_repr(self):
-        self.assertCodeExecution("""
-        str_ = "\\r\\n"
-        print(repr(str_))
+        self.assertCodeExecution(r"""
+        tests = ["\r\n", "áéíóú", "\u000B", "\u2029", "\\", "'", "\"", "\"'"]
+        for test in tests:
+            print(repr(test))
         """)
 
     def test_splitlines(self):
-        self.assertCodeExecution("""
-        str_ = "aaa\\nbbb\\rccc\\r\\nddd\\n\\reee"
+        self.assertCodeExecution(r"""
+        str_ = "aaa\nbbb\rccc\r\nddd\n\reee"
 
         print(str_.splitlines())
         print(str_.splitlines(True))
-        print('Dont Panic\\n'.splitlines())
-        print('\\n'.splitlines())
+        print("Don't Panic\n".splitlines())
+        print('\n'.splitlines())
         print(''.splitlines())
+
+        s1 = '\r\n\r\n\v\f\x0b\x0c\u2029\x1c\x1d\x1e\x85'
+        print(s1.splitlines())
+        print(s1.splitlines(True))
         """)