ICU-64252.0.1.tar.gz

[apple/icu.git] / icuSources / test / cintltst / custrtst.c

diff --git a/icuSources/test/cintltst/custrtst.c b/icuSources/test/cintltst/custrtst.c
index fe84080afb0a09dd04f6dc4bf2b5e8f7333710b8..1c116ea6d87d657f0db9b31e61ee82e63c15e42d 100644 (file)
--- a/icuSources/test/cintltst/custrtst.c
+++ b/icuSources/test/cintltst/custrtst.c
@@ -1,12 +1,14 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html

 /*
 ******************************************************************************
 *
 /*
 ******************************************************************************
 *

-*   Copyright (C) 2002-2014, International Business Machines
+*   Copyright (C) 2002-2016, International Business Machines

 *   Corporation and others.  All Rights Reserved.
 *
 ******************************************************************************
 *   file name:  custrtst.c
 *   Corporation and others.  All Rights Reserved.
 *
 ******************************************************************************
 *   file name:  custrtst.c

-*   encoding:   US-ASCII
+*   encoding:   UTF-8

 *   tab size:   8 (not used)
 *   indentation:4
 *
 *   tab size:   8 (not used)
 *   indentation:4
 *


@@ -36,6 +38,7 @@ static void TestSurrogateSearching(void);
 static void TestUnescape(void);
 static void TestCountChar32(void);
 static void TestUCharIterator(void);
 static void TestUnescape(void);
 static void TestCountChar32(void);
 static void TestUCharIterator(void);

+static void TestIsWellFormed(void);

 
 void addUStringTest(TestNode** root);
 
 
 void addUStringTest(TestNode** root);
 


@@ -48,6 +51,7 @@ void addUStringTest(TestNode** root)
     addTest(root, &TestUnescape, "tsutil/custrtst/TestUnescape");
     addTest(root, &TestCountChar32, "tsutil/custrtst/TestCountChar32");
     addTest(root, &TestUCharIterator, "tsutil/custrtst/TestUCharIterator");
     addTest(root, &TestUnescape, "tsutil/custrtst/TestUnescape");
     addTest(root, &TestCountChar32, "tsutil/custrtst/TestCountChar32");
     addTest(root, &TestUCharIterator, "tsutil/custrtst/TestUCharIterator");

+    addTest(root, &TestIsWellFormed, "tsutil/custrtst/TestIsWellFormed");

 }
 
 /* test data for TestStringFunctions ---------------------------------------- */
 }
 
 /* test data for TestStringFunctions ---------------------------------------- */


@@ -230,7 +234,7 @@ static void TestStringFunctions()
         if(temp[k] != 0xa4)
             log_err("something threw an error in u_strncpy()\n");
 
         if(temp[k] != 0xa4)
             log_err("something threw an error in u_strncpy()\n");
 

-        u_memset(temp, 0x3F, (sizeof(temp) / sizeof(UChar)) - 1);
+        u_memset(temp, 0x3F, UPRV_LENGTHOF(temp) - 1);

         u_uastrncpy(temp, raw[i][j], k-1);
         if(u_strncmp(temp, dataTable[i][j],k-1)!=0)
             log_err("something threw an error in u_uastrncpy(k-1)\n");
         u_uastrncpy(temp, raw[i][j], k-1);
         if(u_strncmp(temp, dataTable[i][j],k-1)!=0)
             log_err("something threw an error in u_uastrncpy(k-1)\n");


@@ -238,7 +242,7 @@ static void TestStringFunctions()
         if(temp[k-1] != 0x3F)
             log_err("something threw an error in u_uastrncpy(k-1)\n");
 
         if(temp[k-1] != 0x3F)
             log_err("something threw an error in u_uastrncpy(k-1)\n");
 

-        u_memset(temp, 0x3F, (sizeof(temp) / sizeof(UChar)) - 1);
+        u_memset(temp, 0x3F, UPRV_LENGTHOF(temp) - 1);

         u_uastrncpy(temp, raw[i][j], k+1);
         if(u_strcmp(temp, dataTable[i][j])!=0)
             log_err("something threw an error in u_uastrncpy(k+1)\n");
         u_uastrncpy(temp, raw[i][j], k+1);
         if(u_strcmp(temp, dataTable[i][j])!=0)
             log_err("something threw an error in u_uastrncpy(k+1)\n");


@@ -246,7 +250,7 @@ static void TestStringFunctions()
         if(temp[k] != 0)
             log_err("something threw an error in u_uastrncpy(k+1)\n");
 
         if(temp[k] != 0)
             log_err("something threw an error in u_uastrncpy(k+1)\n");
 

-        u_memset(temp, 0x3F, (sizeof(temp) / sizeof(UChar)) - 1);
+        u_memset(temp, 0x3F, UPRV_LENGTHOF(temp) - 1);

         u_uastrncpy(temp, raw[i][j], k);
         if(u_strncmp(temp, dataTable[i][j], k)!=0)
             log_err("something threw an error in u_uastrncpy(k)\n");
         u_uastrncpy(temp, raw[i][j], k);
         if(u_strncmp(temp, dataTable[i][j], k)!=0)
             log_err("something threw an error in u_uastrncpy(k)\n");


@@ -362,7 +366,7 @@ static void TestStringFunctions()
             currToken++;
         }
 
             currToken++;
         }
 

-        if (currToken != sizeof(tokens)/sizeof(tokens[0])) {
+        if (currToken != UPRV_LENGTHOF(tokens)) {

             log_err("Didn't get correct number of tokens\n");
         }
         state = delimBuf;       /* Give it an "invalid" saveState */
             log_err("Didn't get correct number of tokens\n");
         }
         state = delimBuf;       /* Give it an "invalid" saveState */


@@ -426,7 +430,7 @@ static void TestStringFunctions()
         UCharIterator iter1, iter2;
         int32_t len1, len2, r1, r2;
 
         UCharIterator iter1, iter2;
         int32_t len1, len2, r1, r2;
 

-        for(i=0; i<(sizeof(strings)/sizeof(strings[0])-1); ++i) {
+        for(i=0; i<(UPRV_LENGTHOF(strings)-1); ++i) {

             if(u_strcmpCodePointOrder(strings[i], strings[i+1])>=0) {
                 log_err("error: u_strcmpCodePointOrder() fails for string %d and the following one\n", i);
             }
             if(u_strcmpCodePointOrder(strings[i], strings[i+1])>=0) {
                 log_err("error: u_strcmpCodePointOrder() fails for string %d and the following one\n", i);
             }


@@ -1093,18 +1097,18 @@ TestUnescape() {
         0x50, 0x72, 0x69, 0x76, 0x61, 0x74, 0x65, 0x73, 0x20,
         0x5a, 0x65, 0x69, 0x63, 0x68, 0x65, 0x6e, 0x3a, 0x20, 0xdbc8, 0xdf45, 0x1b, 0x03, 0x0a, 0x20, 0x1b, 0x263A, 0
     };
         0x50, 0x72, 0x69, 0x76, 0x61, 0x74, 0x65, 0x73, 0x20,
         0x5a, 0x65, 0x69, 0x63, 0x68, 0x65, 0x6e, 0x3a, 0x20, 0xdbc8, 0xdf45, 0x1b, 0x03, 0x0a, 0x20, 0x1b, 0x263A, 0
     };

-    static const int32_t explength = sizeof(expect)/sizeof(expect[0])-1;
+    static const int32_t explength = UPRV_LENGTHOF(expect)-1;

     int32_t length;
 
     /* test u_unescape() */
     int32_t length;
 
     /* test u_unescape() */

-    length=u_unescape(input, buffer, sizeof(buffer)/sizeof(buffer[0]));
+    length=u_unescape(input, buffer, UPRV_LENGTHOF(buffer));

     if(length!=explength || u_strcmp(buffer, expect)!=0) {
         log_err("failure in u_unescape(): length %d!=%d and/or incorrect result string\n", length,
                 explength);
     }
 
     /* try preflighting */
     if(length!=explength || u_strcmp(buffer, expect)!=0) {
         log_err("failure in u_unescape(): length %d!=%d and/or incorrect result string\n", length,
                 explength);
     }
 
     /* try preflighting */

-    length=u_unescape(input, NULL, sizeof(buffer)/sizeof(buffer[0]));
+    length=u_unescape(input, NULL, UPRV_LENGTHOF(buffer));

     if(length!=explength || u_strcmp(buffer, expect)!=0) {
         log_err("failure in u_unescape(preflighting): length %d!=%d\n", length, explength);
     }
     if(length!=explength || u_strcmp(buffer, expect)!=0) {
         log_err("failure in u_unescape(preflighting): length %d!=%d\n", length, explength);
     }


@@ -1160,7 +1164,7 @@ TestCountChar32() {
         buffer[length]=0;
         for(i=0; i<=length; ++i) {
             for(number=-1; number<=((length-i)+2); ++number) {
         buffer[length]=0;
         for(i=0; i<=length; ++i) {
             for(number=-1; number<=((length-i)+2); ++number) {

-                _testStrHasMoreChar32Than(string+i, i, -1, number);
+                _testStrHasMoreChar32Than(buffer+i, i, -1, number);

             }
         }
         --length;
             }
         }
         --length;


@@ -1509,3 +1513,127 @@ TestUCharIterator() {
 
     /* ### TODO test other iterators: CharacterIterator, Replaceable */
 }
 
     /* ### TODO test other iterators: CharacterIterator, Replaceable */
 }

+
+static const UChar valid0[] = { 0 }; // test empty string
+static const UChar valid1[] = { 0x0061,0x270C,0xFE0E, // victory hand with text variation selector
+                                0x0062,0x270C,0xFE0F, // victory hand with emoji variation selector
+                                0x0063,0x270C,0xD83C,0xDFFD, // victory hand with skin tone modifier
+                                0x0064,0x270C,0xFE0F,0xD83C,0xDFFD, // victory hand with emoji variation selector and skin tone modifier (obsolete sequence)
+                                0x0065,0xD83D,0xDC69,0xD83C,0xDFFD,0x200D,0xD83C,0xDFEB, // woman teacher (ZWJ sequence) with skin tone
+                                0x0066,0xD83D,0xDC69,0x200D,0xD83D,0xDC69,0x200D,0xD83D,0xDC67, // family (woman, woman, girl - ZWJ sequence)
+                                0x0067,0x0030,0xFE0F,0x20E3, // keypad 0 (emoji combining seq)
+                                0x0068,0xD83C,0xDDEC,0xD83C,0xDDE7, // flag of UK (regional indicator pair)
+                                0x0069,0xD83C,0xDFF4,0xDB40,0xDC67,0xDB40,0xDC62,0xDB40,0xDC65,0xDB40,0xDC6E,0xDB40,0xDC67,0xDB40,0xDC7F, // flag of England (tag seq)
+                                0x006A,0 };
+static const UChar valid2[] = { 0x006B,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,
+                                0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300, // 29 combining marks
+                                0x006C,0 };
+static const UChar valid3[] = { // sample from Bill Siegrist, 100 UTF16 units
+                                0xD83D,0xDC2E,                                    // U+1F42E   🐮
+                                0xD83D,0xDC3C,                                    // U+1F43C   🐼
+                                0xD83D,0xDC39,                                    // U+1F439   🐹
+                                0xD83D,0xDC31,                                    // U+1F431   🐱
+                                0xD83D,0xDE4B,0x200D,0x2640,0xFE0F,               // U+1F64B  U+200D U+2640  U+FE0F    🙋‍♀️
+                                0xD83D,0xDE47,0xD83C,0xDFFC,0x200D,0x2642,0xFE0F, // U+1F647 U+1F3FC U+200D  U+2642 U+FE0F     🙇🏼‍♂️       
+                                0xD83D,0xDE46,0x200D,0x2642,0xFE0F,               // U+1F646  U+200D U+2642  U+FE0F    🙆‍♂️
+                                0xD83E,0xDDDA,0xD83C,0xDFFF,0x200D,0x2640,0xFE0F, // U+1F9DA U+1F3FF U+200D  U+2640 U+FE0F 🧚🏿‍♀️
+                                0xD83E,0xDDD6,0xD83C,0xDFFE,0x200D,0x2642,0xFE0F, // U+1F9D6 U+1F3FE U+200D  U+2642 U+FE0F 🧖🏾‍♂️
+                                0xD83E,0xDDD6,0xD83C,0xDFFE,0x200D,0x2642,0xFE0F, // U+1F9D6 U+1F3FE U+200D  U+2642 U+FE0F     🧖🏾‍♂️
+                                0xD83E,0xDDDB,0xD83C,0xDFFC,0x200D,0x2642,0xFE0F, // U+1F9DB U+1F3FC U+200D  U+2642 U+FE0F     🧛🏼‍♂️
+                                0xD83E,0xDDD9,0x200D,0x2640,0xFE0F,               // U+1F9D9  U+200D U+2640  U+FE0F    🧙‍♀️
+                                0xD83D,0xDC68,0xD83C,0xDFFE,0x200D,0x2696,0xFE0F, // U+1F468 U+1F3FE U+200D  U+2696 U+FE0F     👨🏾‍⚖️
+                                0xD83D,0xDC69,0xD83C,0xDFFC,0x200D,0xD83D,0xDD27, // U+1F469 U+1F3FC U+200D U+1F527    👩🏼‍🔧
+                                0xD83D,0xDC69,0xD83C,0xDFFF,0x200D,0xD83C,0xDFEB, // U+1F469 U+1F3FF U+200D U+1F3EB    👩🏿‍🏫
+                                0xD83D,0xDC68,0xD83C,0xDFFE,0x200D,0xD83D,0xDCBB, // U+1F468 U+1F3FE U+200D U+1F4BB    👨🏾‍💻
+                                0xD83D,0xDC69,0xD83C,0xDFFD,0x200D,0xD83D,0xDD2C, // U+1F469 U+1F3FD U+200D U+1F52C    👩🏽‍🔬
+                                0xD83D,0xDC68,0xD83C,0xDFFC,0x200D,0xD83D,0xDE92, // U+1F468 U+1F3FC U+200D U+1F692    👨🏼‍🚒
+                                0 };
+static const UChar valid4[] = { 0x0061,0x202A,0x0062,0x202A,0x0063,0x202A,0x0064,0x202A,0x0065,0x202A,0x0066,0x202A,0x0067,0x202A,0x0068,0x202A, // to level 16
+                                0x0061,0x202A,0x0062,0x202A,0x0063,0x202A,0x0064,0x202A,0x0065,0x202A,0x0066,0x202A,0x0067,0x202A,0x0068,0x202A, // to level 32
+                                0x0061,0x202A,0x0062,0x202A,0x0063,0x202A,0x0064,0x202A,0x0065,0x202A,0x0066,0x202A,0x0067,0x202A,0x0068,0x202A, // to level 48
+                                0x0061,0x202A,0x0062,0x202A,0x0063,0x202A,0x0064,0x202A,0x0065,0x202A,0x0066,0x202A,0x0067,0x202A,0x0068,0x202A, // to level 64
+                                0x0061,0x202A,0x0062,0x202A,0x0063,0x202A,0x0064,0x202A,0x0065,0x202A,0x0066,0x202A,0x0067,0x202A,0x0068,0x202A, // to level 80
+                                0x0061,0x202A,0x0062,0x202A,0x0063,0x202A,0x0064,0x202A,0x0065,0x202A,0x0066,0x202A,0x0067,0x202A,0x0068,0x202A, // to level 96
+                                0x0061,0x202A,0x0062,0x202A,0x0063,0x202A,0x0064,0x202A,0x0065,0x202A,0x0066,0x202A,0x0067,0x202A,0x0068,0x202A, // to level 112
+                                0x0061,0x2066,0x0062,0x202A,0x0063,0x202A,0x0064,0x202A,0x0065,0x202A,0x0066,0x2069,                             // to level 122 in LRI then pop to 112
+                                0x0061,0x202A,0x0062,0x202A,0x0063,0x202A,0x0064,0x202A,0x0065,0x202A,0x0066,                                    // to level 122 again
+                                0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C, // pop to level 90
+                                0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C, // pop to level 58
+                                0x0061,0x202A,0x0062,0x202A,0x0063,0x202A,0x0064,0x202A,0x0065,0x202A,0x0066,0x202A,0x0067,0x202A,0x0068,0x202A, // to level 74
+                                0x0061,0x202A,0x0062,0x202A,0x0063,0x202A,0x0064,0x202A,0x0065,0x202A,0x0066,0x202A,0x0067,0x202A,0x0068,0x202A, // to level 90
+                                0x0061,0x202A,0x0062,0x202A,0x0063,0x202A,0x0064,0x202A,0x0065,0x202A,0x0066,0x202A,0x0067,0x202A,0x0068,0x202A, // to level 106
+                                0x000A,                                                                                                          // pop to 0
+                                0x0061,0x202A,0x0062,0x202A,0x0063,0x202A,0x0064,0x202A,0x0065,0x202A,0x0066,0x202A,0x0067,0x202A,0x0068,0x202A, // to level 16
+                                0x0061,0x202A,0x0062,0x202A,0x0063,0x202A,0x0064,0x202A,0x0065,0x202A,0x0066,0x202A,0x0067,0x202A,0x0068,0x202A, // to level 32
+                                0x0061,0x202A,0x0062,0x202A,0x0063,0x202A,0x0064,0x202A,0x0065,0x202A,0x0066,0x202A,0x0067,0x202A,0x0068,0x202A, // to level 48
+                                0x0061,0x202A,0x0062,0x202A,0x0063,0x202A,0x0064,0x202A,0x0065,0x202A,0x0066,0x202A,0x0067,0x202A,0x0068,0x202A, // to level 64
+                                0x0061,0x202A,0x0062,0x202A,0x0063,0x202A,0x0064,0x202A,0x0065,0x202A,0x0066,0x202A,0x0067,0x202A,0x0068,0x202A, // to level 80
+                                0x0061,0x202A,0x0062,0x202A,0x0063,0x202A,0x0064,0x202A,0x0065,0x202A,0x0066,0x202A,0x0067,0x202A,0x0068,0x202A, // to level 96
+                                0x000A,0 };
+
+static const UChar malformed1[] = { 0x0061,0xFFFF,0 }; // non-character (BMP)
+static const UChar malformed2[] = { 0x0062,0xD87F,0xDFFE,0 }; // non-character (supplemental)
+static const UChar malformed3[] = { 0x0063,0xD7FC,0 }; // unassigned
+static const UChar malformed4[] = { 0x0064,0xD800,0 }; // unpaired high surrogate
+static const UChar malformed5[] = { 0x0065,0xDC00,0 }; // unpaired low surrogate
+static const UChar malformed6[] = { 0x0066,0xFE0F,0 }; // emoji variation selector on non-emoji
+static const UChar malformed7[] = { 0x0067,0xDB40,0xDC67,0xDB40,0xDC7F,0 }; // tag sequence on non-emoji
+static const UChar malformed8[] = { 0xDB40,0xDC67,0xDB40,0xDC7F,0 }; // tag sequence with no base
+static const UChar malformed9[] = { 0x0068,0xD83C,0xDFF4,0xDB40,0xDC67,0xDB40,0xDC62,0xDB40,0xDC65,0xDB40,0xDC6E,0xDB40,0xDC67,0x0069,0 }; // tag sequence with no term
+static const UChar malformedA[] = { 0x006A,0xD83C,0xDFF4,0xDB40,0xDC7F,0 }; // tag sequence with no tag_spec, just term
+static const UChar malformedB[] = { 0x006B,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,
+                                    0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300,0x0300, // 31 combining marks
+                                    0x006C,0 };
+static const UChar malformedC[] = { 0x0061,0x202A,0x0062,0x202A,0x0063,0x202A,0x0064,0x202A,0x0065,0x202A,0x0066,0x202A,0x0067,0x202A,0x0068,0x202A, // to level 16
+                                    0x0061,0x202A,0x0062,0x202A,0x0063,0x202A,0x0064,0x202A,0x0065,0x202A,0x0066,0x202A,0x0067,0x202A,0x0068,0x202A, // to level 32
+                                    0x0061,0x202A,0x0062,0x202A,0x0063,0x202A,0x0064,0x202A,0x0065,0x202A,0x0066,0x202A,0x0067,0x202A,0x0068,0x202A, // to level 48
+                                    0x0061,0x202A,0x0062,0x202A,0x0063,0x202A,0x0064,0x202A,0x0065,0x202A,0x0066,0x202A,0x0067,0x202A,0x0068,0x202A, // to level 64
+                                    0x0061,0x202A,0x0062,0x202A,0x0063,0x202A,0x0064,0x202A,0x0065,0x202A,0x0066,0x202A,0x0067,0x202A,0x0068,0x202A, // to level 80
+                                    0x0061,0x202A,0x0062,0x202A,0x0063,0x202A,0x0064,0x202A,0x0065,0x202A,0x0066,0x202A,0x0067,0x202A,0x0068,0x202A, // to level 96
+                                    0x0061,0x202A,0x0062,0x202A,0x0063,0x202A,0x0064,0x202A,0x0065,0x202A,0x0066,0x202A,0x0067,0x202A,0x0068,0x202A, // to level 112
+                                    0x0061,0x202A,0x0062,0x202A,0x0063,0x202A,0x0064,0x202A,0x0065,0x202A,0x0066,0x202A,0x0067,0x202A,0x0068,0x202A, // to level 128 (error)
+                                    0x0061,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0x202C,0 };    // start PDFs, too late
+
+typedef struct {
+    const char* descrip;
+    const UChar* string;
+    UBool result;
+} StringAndResult;
+
+static const StringAndResult wellFormedTests[] = {
+    { "valid0",     valid0,     TRUE },
+    { "valid1",     valid1,     TRUE },
+    { "valid2",     valid2,     TRUE },
+    { "valid3",     valid3,     TRUE },
+    { "valid4",     valid4,     TRUE },
+    { "malformed1", malformed1, FALSE },
+    { "malformed2", malformed2, FALSE },
+    { "malformed3", malformed3, FALSE },
+    { "malformed4", malformed4, FALSE },
+    { "malformed5", malformed5, FALSE },
+    { "malformed6", malformed6, FALSE },
+    { "malformed7", malformed7, FALSE },
+    { "malformed8", malformed8, FALSE },
+    { "malformed9", malformed9, FALSE },
+    { "malformedA", malformedA, FALSE },
+    { "malformedB", malformedB, FALSE },
+    { "malformedC", malformedC, FALSE },
+    { NULL, NULL, 0 }
+};
+
+static void
+TestIsWellFormed() {
+    const StringAndResult* testPtr;
+    for (testPtr = wellFormedTests; testPtr->descrip != NULL; testPtr++) {
+        UBool result = u_strIsWellFormed(testPtr->string, -1);
+        if (result != testPtr->result) {
+            log_err("test %s with length -1, expected %d, got %d\n", testPtr->descrip, testPtr->result, result);
+        }
+
+        int32_t length = u_strlen(testPtr->string);
+        result = u_strIsWellFormed(testPtr->string, length);
+        if (result != testPtr->result) {
+            log_err("test %s with length %d, expected %d, got %d\n", testPtr->descrip, length, testPtr->result, result);
+        }
+    }
+}