Work toward implementing JSON5 whitespace. Untested and incomplete.

FossilOrigin-Name: d262c059455ebe0650a45a6c1c04d1baf9609c635df352732dd192426e1bdc39

diff --git a/manifest b/manifest
index 4f17dbdff9bce97450d84a17a4a5f833b380b348..a607ef397a3659b089bd0a249e658869a6be964d 100644 (file)
--- a/manifest
+++ b/manifest
@@ -1,5 +1,5 @@
-C Fix\sto\sthe\sjson-speed-check.sh\sscript.
-D 2023-04-26T13:52:20.152
+C Work\stoward\simplementing\sJSON5\swhitespace.\s\sUntested\sand\sincomplete.
+D 2023-04-26T15:19:19.822
 F .fossil-settings/empty-dirs dbb81e8fc0401ac46a1491ab34a7f2c7c0452f2f06b54ebb845d024ca8283ef1
 F .fossil-settings/ignore-glob 35175cdfcf539b2318cb04a9901442804be81cd677d8b889fcc9149c21f239ea
 F LICENSE.md df5091916dbb40e6e9686186587125e1b2ff51f022cc334e886c19a0e9982724
@@ -592,7 +592,7 @@ F src/hash.h 3340ab6e1d13e725571d7cee6d3e3135f0779a7d8e76a9ce0a85971fa3953c51
 F src/hwtime.h b638809e083b601b618df877b2e89cb87c2a47a01f4def10be4c4ebb54664ac7
 F src/in-operator.md 10cd8f4bcd225a32518407c2fb2484089112fd71
 F src/insert.c a8de1db43335fc4946370a7a7e47d89975ad678ddb15078a150e993ba2fb37d4
-F src/json.c 1d049d073c0197d55e0046877cb87ad7b462419748cf08663d3baf04db676efb
+F src/json.c d0f9a0c630bdea3721e28b5ba2fce119c0e732ad48f6674a8f1644f2ec45ffb1
 F src/legacy.c d7874bc885906868cd51e6c2156698f2754f02d9eee1bae2d687323c3ca8e5aa
 F src/loadext.c be5af440f3192c58681b5d43167dbca3ccbfce394d89faa22378a14264781136
 F src/main.c 09bc5191f75dc48fc4dfddda143cb864c0c3dbc3297eb9a9c8e01fea58ff847d
@@ -1253,7 +1253,7 @@ F test/json/README.md 9d117e0d6da9eee96de2fc8e32b603208b0e5b460ff99e5de3407bb713
 F test/json/json-generator.tcl 229bd293f1865f787c160886cadd282631721925cca2947aaa54bbcd7f65cef7
 F test/json/json-q1.txt 335a7c8ab291d354f33b7decc9559e99a2823d4142291c4be7aa339a631f3c2d
 F test/json/json-speed-check.sh 362f9c5c4a69c6f2c6fa98f538fb2e07e084dc0bef74f8bbd92cc1666e8415e5 x
-F test/json101.test 7241ab038c75ba780b009c6f37e4b660e5e63d1d74c110336a4779a2db8653f7
+F test/json101.test de9c93169b84ac96fd5836c638a2ae1f00e4afbd4003c6b596692d7f05e1cd69
 F test/json102.test 327e77275f338c028faefa2da5164daf6b142a165e3015ff2a6e4251ddc6a0ac
 F test/json103.test 53df87f83a4e5fa0c0a56eb29ff6c94055c6eb919f33316d62161a8880112dbe
 F test/json104.test a502dc01853aada95d721b3b275afbe2dc18fffdac1fea6e96fb20c13586bbb5
@@ -2063,8 +2063,8 @@ F vsixtest/vsixtest.tcl 6a9a6ab600c25a91a7acc6293828957a386a8a93
 F vsixtest/vsixtest.vcxproj.data 2ed517e100c66dc455b492e1a33350c1b20fbcdc
 F vsixtest/vsixtest.vcxproj.filters 37e51ffedcdb064aad6ff33b6148725226cd608e
 F vsixtest/vsixtest_TemporaryKey.pfx e5b1b036facdb453873e7084e1cae9102ccc67a0
-P 3051d6a7c12cbf86634d8fab64f3f3bc9a1eb184260e56a82eb543cb634ba7ef
-R 76520c4f968a15591ad7c027fe9539e0
+P d839c9544d7f28c1abc779eb2d40f95c1a9386984656fbd29d19b1e7830171bc
+R 9b88e23f3662645c8ffe23999211d809
 U drh
-Z 781a4e98041d9791d0ec4084972b315b
+Z 75164b4fdafac0aaf63d4b3999912d9c
 # Remove this line to create a well-formed Fossil manifest.

diff --git a/manifest.uuid b/manifest.uuid
index d8bd0e1a4d70fc5e46382e69c872530141aa81f1..0de6b4f785932b186b9ccac4d0da232b2597352a 100644 (file)
--- a/manifest.uuid
+++ b/manifest.uuid
@@ -1 +1 @@
-d839c9544d7f28c1abc779eb2d40f95c1a9386984656fbd29d19b1e7830171bc
\ No newline at end of file
+d262c059455ebe0650a45a6c1c04d1baf9609c635df352732dd192426e1bdc39
\ No newline at end of file

diff --git a/src/json.c b/src/json.c
index b201fdde5c127a8f251bcbee200947166452e8e1..58f89b1aaa2f422911f3d6c80743295c16c19c7c 100644 (file)
--- a/src/json.c
+++ b/src/json.c
@@ -758,6 +758,124 @@ static int jsonIs4Hex(const char *z){
   return 1;
 }
 
+/*
+** Return the number of bytes of JSON5 whitespace at the beginning of
+** the input string z[].
+**
+** JSON5 whitespace consists of any of the following characters:
+**
+**    Unicode  UTF-8         Name
+**    U+0009   09            horizontal tab
+**    U+000a   0a            line feed
+**    U+000b   0b            vertical tab
+**    U+000c   0c            form feed
+**    U+000d   0d            carriage return
+**    U+0020   20            space
+**    U+00a0   c2 a0         non-breaking space
+**    U+1680   e1 9a 80      ogham space mark
+**    U+2000   e2 80 80      en quad
+**    U+2001   e2 80 81      em quad
+**    U+2002   e2 80 82      en space
+**    U+2003   e2 80 83      em space
+**    U+2004   e2 80 84      three-per-em space
+**    U+2005   e2 80 85      four-per-em space
+**    U+2006   e2 80 86      six-per-em space
+**    U+2007   e2 80 87      figure space
+**    U+2008   e2 80 88      punctuation space
+**    U+2009   e2 80 89      thin space
+**    U+200a   e2 80 8a      hair space
+**    U+2028   e2 80 a8      line separator
+**    U+2029   e2 80 a9      paragraph separator
+**    U+202f   e2 80 af      narrow no-break space (NNBSP)
+**    U+205f   e2 81 9f      medium mathematical space (MMSP)
+**    U+3000   e3 80 80      ideographical space
+**    U+FEFF   ef bb bf      byte order mark
+**
+** In addition, comments between '/', '*' and '*', '/' and
+** from '/', '/' to end-of-line are also considered to be whitespace.
+*/
+static int json5Whitespace(const char *zIn){
+  int n = 0;
+  const u8 *z = (u8*)zIn;
+  while( 1 /*exit by "goto whitespace_done"*/ ){
+    switch( z[n] ){
+      case 0x09:
+      case 0x0a:
+      case 0x0b:
+      case 0x0c:
+      case 0x0d:
+      case 0x20: {
+        n++;
+        break;
+      }
+      case '/': {
+        if( z[n+1]=='*' && z[n+2]!=0 ){
+          int j;
+          for(j=n+3; z[j]!='/' || z[j-1]!='*'; j++){
+            if( z[j]==0 ) goto whitespace_done;
+          }
+          n += j;
+          break;
+        }else if( z[n+1]=='/' ){
+          int j;
+          for(j=n+2; z[j] && z[j]!='\n'; j++){}
+          n += j;
+          break;
+        }
+        goto whitespace_done;
+      }
+      case 0xc2: {
+        if( z[n+1]==0xa0 ){
+          n += 2;
+          break;
+        }
+        goto whitespace_done;
+      }
+      case 0xe1: {
+        if( z[n+1]==0x9a && z[n+2]==0x80 ){
+          n += 3;
+          break;
+        }
+        goto whitespace_done;
+      }
+      case 0xe2: {
+        if( z[n+1]==0x80 ){
+          u8 c = z[n+2];
+          if( c<0x80 ) goto whitespace_done;
+          if( c<=0x8a || c==0xa8 || c==0xa9 || c==0xaf ){
+            n += 3;
+            break;
+          }
+        }else if( z[n+1]==0x81 && z[n+2]==0x9f ){
+          n += 3;
+          break;
+        }
+        goto whitespace_done;
+      }
+      case 0xe3: {
+        if( z[n+1]==0x80 && z[n+2]==0x80 ){
+          n += 3;
+          break;
+        }
+        goto whitespace_done;
+      }
+      case 0xef: {
+        if( z[n+1]==0xbb && z[n+2]==0xbf ){
+          n += 3;
+          break;
+        }
+        goto whitespace_done;
+      }
+      default: {
+        goto whitespace_done;
+      }
+    }
+  }
+  whitespace_done:
+  return n;
+}
+
+
 #ifdef SQLITE_ENABLE_JSON_NAN_INF
 /*
 ** Extra floating-point literals to allow in JSON.
@@ -795,7 +913,9 @@ static int jsonParseValue(JsonParse *pParse, u32 i){
   JsonNode *pNode;
   const char *z = pParse->zJson;
   while( fast_isspace(z[i]) ){ i++; }
-  if( (c = z[i])=='{' ){
+json_parse_restart:
+  c = z[i];
+  if( c=='{' ){
     /* Parse object */
     iThis = jsonParseAddNode(pParse, JSON_OBJECT, 0, 0);
     if( iThis<0 ) return -1;
@@ -960,6 +1080,10 @@ static int jsonParseValue(JsonParse *pParse, u32 i){
     return -3;  /* End of [...] */
   }else if( c==0 ){
     return 0;   /* End of file */
+  }else if( (j = json5Whitespace(&z[i]))>0 ){
+    i += j;
+    pParse->has5 = 1;
+    goto json_parse_restart;
   }else{
 #ifdef SQLITE_ENABLE_JSON_NAN_INF
     int k, nn;

diff --git a/test/json101.test b/test/json101.test
index 58559aafe566c9e536dabaf6ba58a236e73556f7..3fc3923da30d2253cdbcbf52c80138289160eaa4 100644 (file)
--- a/test/json101.test
+++ b/test/json101.test
@@ -308,26 +308,35 @@ do_execsql_test json-5.8 {
 
 do_execsql_test json-6.1 {
   SELECT json_valid('{"a":55,"b":72,}');
+} {0}
+do_execsql_test json-6.2 {
+  SELECT json_valid5('{"a":55,"b":72,}');
 } {1}
-do_execsql_test json-6.1b {
+do_execsql_test json-6.3 {
+  SELECT json_valid(json('{"a":55,"b":72,}'));
+} {1}
+do_execsql_test json-6.4 {
   SELECT json_valid('{"a":55,"b":72 , }');
+} {0}
+do_execsql_test json-6.5 {
+  SELECT json_valid5('{"a":55,"b":72 , }');
 } {1}
-do_execsql_test json-6.1c {
-  SELECT json_valid('{"a":55,"b":72,,}');
+do_execsql_test json-6.6 {
+  SELECT json_valid5('{"a":55,"b":72,,}');
 } {0}
-do_execsql_test json-6.2 {
+do_execsql_test json-6.7 {
   SELECT json_valid('{"a":55,"b":72}');
 } {1}
-do_execsql_test json-6.3 {
-  SELECT json_valid('["a",55,"b",72,]');
+do_execsql_test json-6.8 {
+  SELECT json_valid5('["a",55,"b",72,]');
 } {1}
-do_execsql_test json-6.3b {
-  SELECT json_valid('["a",55,"b",72 , ]');
+do_execsql_test json-6.9 {
+  SELECT json_valid5('["a",55,"b",72 , ]');
 } {1}
-do_execsql_test json-6.3c {
-  SELECT json_valid('["a",55,"b",72,,]');
+do_execsql_test json-6.10 {
+  SELECT json_valid5('["a",55,"b",72,,]');
 } {0}
-do_execsql_test json-6.4 {
+do_execsql_test json-6.11 {
   SELECT json_valid('["a",55,"b",72]');
 } {1}