Merge branch 'master' of github:stedolan/jq

Conflicts:
	c/builtin.c

6 files changed, 163 insertions, 66 deletions

diff --git a/c/builtin.c b/c/builtin.c
index 9be18208..e8c51cd7 100644
--- a/c/builtin.c
+++ b/c/builtin.c
@@ -143,7 +143,15 @@ static void f_length(jv input[], jv output[]) {
   }
 }
 
-static struct cfunction function_list[] = {
+static void f_tostring(jv input[], jv output[]) {
+  if (jv_get_kind(input[0]) == JV_KIND_STRING) {
+    output[0] = input[0];
+  } else {
+    output[0] = jv_dump_string(input[0], 0);
+  }
+}
+
+struct cfunction function_list[] = {
   {f_true, "true", CALL_BUILTIN_1_1},
   {f_false, "false", CALL_BUILTIN_1_1},
   {f_null, "null", CALL_BUILTIN_1_1},
@@ -152,6 +160,7 @@ static struct cfunction function_list[] = {
   {f_multiply, "_multiply", CALL_BUILTIN_3_1},
   {f_divide, "_divide", CALL_BUILTIN_3_1},
   {f_tonumber, "tonumber", CALL_BUILTIN_1_1},
+  {f_tostring, "tostring", CALL_BUILTIN_1_1},
   {f_equal, "_equal", CALL_BUILTIN_3_1},
   {f_length, "length", CALL_BUILTIN_1_1},
 };
diff --git a/c/jv.c b/c/jv.c
index 786e7b26..51070636 100644
--- a/c/jv.c
+++ b/c/jv.c
@@ -340,6 +340,7 @@ typedef struct {
   // high 31 bits are length, low bit is a flag 
   // indicating whether hash has been computed.
   uint32_t length_hashed;
+  uint32_t alloc_length;
   char data[];
 } jvp_string;
 
@@ -350,19 +351,19 @@ static jvp_string* jvp_string_ptr(jv_complex* a) {
 static jvp_string* jvp_string_alloc(uint32_t size) {
   jvp_string* s = malloc(sizeof(jvp_string) + size + 1);
   s->refcnt.count = 1;
-  s->length_hashed = size << 1;
+  s->alloc_length = size;
   return s;
 }
 
 static jv_complex jvp_string_new(const char* data, uint32_t length) {
   jvp_string* s = jvp_string_alloc(length);
+  s->length_hashed = length << 1;
   memcpy(s->data, data, length);
   s->data[length] = 0;
   jv_complex r = {&s->refcnt, {0,0}};
   return r;
 }
 
-
 static void jvp_string_free(jv_complex* s) {
   if (jvp_refcnt_dec(s)) {
     jvp_string* str = jvp_string_ptr(s);
@@ -385,16 +386,37 @@ static uint32_t jvp_string_length(jvp_string* s) {
   return s->length_hashed >> 1;
 }
 
-static jv_complex jvp_string_concat(jvp_string* a, jvp_string* b) {
-  uint32_t la = jvp_string_length(a), lb = jvp_string_length(b);
-  jvp_string* s = jvp_string_alloc(la + lb);
-  memcpy(s->data, a->data, la);
-  memcpy(s->data + la, b->data, lb);
-  s->data[la + lb] = 0;
-  jv_complex r = {&s->refcnt, {0,0}};
+static uint32_t jvp_string_remaining_space(jvp_string* s) {
+  uint32_t r = s->alloc_length - jvp_string_length(s);
+  assert(r >= 0);
   return r;
 }
 
+static void jvp_string_append(jv_complex* string, const char* data, uint32_t len) {
+  jvp_string* s = jvp_string_ptr(string);
+  uint32_t currlen = jvp_string_length(s);
+    
+  if (jvp_refcnt_unshared(string) &&
+      jvp_string_remaining_space(s) >= len) {
+    // the next string fits at the end of a
+    memcpy(s->data + currlen, data, len);
+    s->data[currlen + len] = 0;
+    s->length_hashed = (currlen + len) << 1;
+  } else {
+    // allocate a bigger buffer and copy
+    uint32_t allocsz = (currlen + len) * 2;
+    if (allocsz < 32) allocsz = 32;
+    jvp_string* news = jvp_string_alloc(allocsz);
+    news->length_hashed = (currlen + len) << 1;
+    memcpy(news->data, s->data, currlen);
+    memcpy(news->data + currlen, data, len);
+    news->data[currlen + len] = 0;
+    jvp_string_free(string);
+    jv_complex r = {&news->refcnt, {0,0}};
+    *string = r;
+  }
+}
+
 static const uint32_t HASH_SEED = 0x432A9843;
 
 static uint32_t rotl32 (uint32_t x, int8_t r){
@@ -507,15 +529,21 @@ const char* jv_string_value(jv j) {
 }
 
 jv jv_string_concat(jv a, jv b) {
-  jv j;
-  j.kind = JV_KIND_STRING;
-  j.val.complex = jvp_string_concat(jvp_string_ptr(&a.val.complex),
-                                    jvp_string_ptr(&b.val.complex));
-  jv_free(a);
+  jvp_string* sb = jvp_string_ptr(&b.val.complex);
+  jvp_string_append(&a.val.complex, sb->data, jvp_string_length(sb));
   jv_free(b);
-  return j;
+  return a;
 }
 
+jv jv_string_append_buf(jv a, const char* buf, int len) {
+  jvp_string_append(&a.val.complex, buf, len);
+  return a;
+}
+
+jv jv_string_append_str(jv a, const char* str) {
+  return jv_string_append_buf(a, str, strlen(str));
+}
+                        
 jv jv_string_fmt(const char* fmt, ...) {
   int size = 1024;
   while (1) {
diff --git a/c/jv.h b/c/jv.h
index ec23262c..e67d614a 100644
--- a/c/jv.h
+++ b/c/jv.h
@@ -80,6 +80,8 @@ uint32_t jv_string_hash(jv);
 const char* jv_string_value(jv);
 jv jv_string_concat(jv, jv);
 jv jv_string_fmt(const char*, ...);
+jv jv_string_append_buf(jv a, const char* buf, int len);
+jv jv_string_append_str(jv a, const char* str);
 
 jv jv_object();
 jv jv_object_get(jv object, jv key);
@@ -103,7 +105,8 @@ jv jv_object_iter_value(jv, int);
 int jv_get_refcnt(jv);
 
 enum { JV_PRINT_PRETTY = 1, JV_PRINT_ASCII = 2 };
-void jv_dump(jv, int);
+void jv_dump(jv, int flags);
+jv jv_dump_string(jv, int flags);
 
 jv jv_parse(const char* string);
 jv jv_parse_sized(const char* string, int length);
diff --git a/c/jv_print.c b/c/jv_print.c
index f1ec92fb..64bf178a 100644
--- a/c/jv_print.c
+++ b/c/jv_print.c
@@ -1,47 +1,70 @@
 #include "jv.h"
 #include <stdio.h>
 #include <float.h>
-#include <math.h>
+#include <string.h>
 
 #include "jv_dtoa.h"
 #include "jv_unicode.h"
 
-static void jv_dump_string(jv str, int ascii_only) {
+static void put_buf(const char* s, int len, FILE* fout, jv* strout) {
+  if (strout) {
+    *strout = jv_string_append_buf(*strout, s, len);
+  } else {
+    fwrite(s, 1, len, fout);
+  }
+}
+
+static void put_char(char c, FILE* fout, jv* strout) {
+  put_buf(&c, 1, fout, strout);
+}
+
+static void put_str(const char* s, FILE* fout, jv* strout) {
+  put_buf(s, strlen(s), fout, strout);
+}
+
+static void put_space(int n, FILE* fout, jv* strout) {
+  while (n--) {
+    put_char(' ', fout, strout);
+  }
+}
+
+static void jvp_dump_string(jv str, int ascii_only, FILE* F, jv* S) {
   assert(jv_get_kind(str) == JV_KIND_STRING);
   const char* i = jv_string_value(str);
   const char* end = i + jv_string_length(jv_copy(str));
   int c = 0;
+  char buf[32];
   while ((i = jvp_utf8_next(i, end, &c))) {
     assert(c != -1);
     int unicode_escape = 0;
     if (0x20 <= c && c <= 0x7E) {
       // printable ASCII
       if (c == '"' || c == '\\') {
-        putchar('\\');
+        put_char('\\', F, S);
       }
-      putchar(c);
+      put_char(c, F, S);
     } else if (c < 0x20 || c == 0x7F) {
       // ASCII control character
       switch (c) {
       case '\b':
-        putchar('\\');
-        putchar('b');
+        put_char('\\', F, S);
+        put_char('b', F, S);
         break;
       case '\t':
-        putchar('\\');
-        putchar('t');
+        put_char('\\', F, S);
+        put_char('t', F, S);
         break;
       case '\r':
-        putchar('\\');
-        putchar('r');
+        put_char('\\', F, S);
+        put_char('r', F, S);
         break;
       case '\n':
-        putchar('\\');
-        putchar('n');
+        put_char('\\', F, S);
+        put_char('n', F, S);
         break;
       case '\f':
-        putchar('\\');
-        putchar('f');
+        put_char('\\', F, S);
+        put_char('f', F, S);
         break;
       default:
         unicode_escape = 1;
@@ -52,13 +75,14 @@ static void jv_dump_string(jv str, int ascii_only) {
     }
     if (unicode_escape) {
       if (c <= 0xffff) {
-        printf("\\u%04x", c);
+        sprintf(buf, "\\u%04x", c);
       } else {
         c -= 0x10000;
-        printf("\\u%04x\\u%04x", 
-               0xD800 | ((c & 0xffc00) >> 10),
-               0xDC00 | (c & 0x003ff));
+        sprintf(buf, "\\u%04x\\u%04x", 
+                0xD800 | ((c & 0xffc00) >> 10),
+                0xDC00 | (c & 0x003ff));
       }
+      put_str(buf, F, S);
     }
   }
   assert(c != -1);
@@ -66,78 +90,97 @@ static void jv_dump_string(jv str, int ascii_only) {
 
 enum { INDENT = 2 };
 
-static void jv_dump_term(struct dtoa_context* C, jv x, int flags, int indent) {
+static void jv_dump_term(struct dtoa_context* C, jv x, int flags, int indent, FILE* F, jv* S) {
   char buf[JVP_DTOA_FMT_MAX_LEN];
   switch (jv_get_kind(x)) {
   case JV_KIND_INVALID:
     assert(0 && "Invalid value");
     break;
   case JV_KIND_NULL:
-    printf("null");
+    put_str("null", F, S);
     break;
   case JV_KIND_FALSE:
-    printf("false");
+    put_str("false", F, S);
     break;
   case JV_KIND_TRUE:
-    printf("true");
+    put_str("true", F, S);
     break;
   case JV_KIND_NUMBER: {
     double d = jv_number_value(x);
     if (d != d) {
       // JSON doesn't have NaN, so we'll render it as "null"
-      printf("null");
+      put_str("null", F, S);
     } else {
       // Normalise infinities to something we can print in valid JSON
       if (d > DBL_MAX) d = DBL_MAX;
       if (d < -DBL_MAX) d = -DBL_MAX;
-      printf("%s", jvp_dtoa_fmt(C, buf, d));
+      put_str(jvp_dtoa_fmt(C, buf, d), F, S);
     }
     break;
   }
   case JV_KIND_STRING:
-    // FIXME: all sorts of broken
-    putchar('"');
-    jv_dump_string(x, 0);
-    putchar('"');
+    put_char('"', F, S);
+    jvp_dump_string(x, 0, F, S);
+    put_char('"', F, S);
     break;
   case JV_KIND_ARRAY: {
     if (jv_array_length(jv_copy(x)) == 0) {
-      printf("[]");
+      put_str("[]", F, S);
       break;
     }
-    printf("[");
-    if (flags & JV_PRINT_PRETTY) printf("\n%*s", indent+INDENT, "");
+    put_str("[", F, S);
+    if (flags & JV_PRINT_PRETTY) {
+      put_char('\n', F, S);
+      put_space(indent+INDENT, F, S);
+    }
     for (int i=0; i<jv_array_length(jv_copy(x)); i++) {
       if (i!=0) {
-        if (flags & JV_PRINT_PRETTY) printf(",\n%*s", indent+INDENT, "");
-        else printf(", ");
+        if (flags & JV_PRINT_PRETTY) {
+          put_str(",\n", F, S);
+          put_space(indent+INDENT, F, S);
+        } else {
+          put_str(", ", F, S);
+        }
       }
-      jv_dump_term(C, jv_array_get(jv_copy(x), i), flags, indent + INDENT);
+      jv_dump_term(C, jv_array_get(jv_copy(x), i), flags, indent + INDENT, F, S);
+    }
+    if (flags & JV_PRINT_PRETTY) {
+      put_char('\n', F, S);
+      put_space(indent, F, S);
     }
-    if (flags & JV_PRINT_PRETTY) printf("\n%*s", indent, "");
-    printf("]");
+    put_char(']', F, S);
     break;
   }
   case JV_KIND_OBJECT: {
     if (jv_object_length(jv_copy(x)) == 0) {
-      printf("{}");
+      put_str("{}", F, S);
       break;
     }
-    printf("{");
-    if (flags & JV_PRINT_PRETTY) printf("\n%*s", indent+INDENT, "");
+    put_char('{', F, S);
+    if (flags & JV_PRINT_PRETTY) {
+      put_char('\n', F, S);
+      put_space(indent+INDENT, F, S);
+    }
     int first = 1;
     jv_object_foreach(i, x) {
       if (!first) {
-        if (flags & JV_PRINT_PRETTY) printf(",\n%*s", indent+INDENT, "");
-        else printf(", ");
+        if (flags & JV_PRINT_PRETTY){
+          put_str(",\n", F, S);
+          put_space(indent+INDENT, F, S);
+        } else {
+          put_str(", ", F, S);
+        }
       }
       first = 0;
-      jv_dump_term(C, jv_object_iter_key(x, i), flags, indent + INDENT);
-      printf(": ");
-      jv_dump_term(C, jv_object_iter_value(x, i), flags, indent + INDENT);
+      jv_dump_term(C, jv_object_iter_key(x, i), flags, indent + INDENT, F, S);
+      put_str(": ", F, S);
+      jv_dump_term(C, jv_object_iter_value(x, i), flags, indent + INDENT, F, S);
     }
-    if (flags & JV_PRINT_PRETTY) printf("\n%*s", indent, "");
-    printf("}");
+    if (flags & JV_PRINT_PRETTY) {
+      put_char('\n', F, S);
+      put_space(indent, F, S);
+    }
+    put_char('}', F, S);
   }
   }
   jv_free(x);
@@ -146,6 +189,15 @@ static void jv_dump_term(struct dtoa_context* C, jv x, int flags, int indent) {
 void jv_dump(jv x, int flags) {
   struct dtoa_context C;
   jvp_dtoa_context_init(&C);
-  jv_dump_term(&C, x, flags, 0);
+  jv_dump_term(&C, x, flags, 0, stdout, 0);
+  jvp_dtoa_context_free(&C);
+}
+
+jv jv_dump_string(jv x, int flags) {
+  struct dtoa_context C;
+  jvp_dtoa_context_init(&C);
+  jv s = jv_string("");
+  jv_dump_term(&C, x, flags, 0, 0, &s);
   jvp_dtoa_context_free(&C);
+  return s;
 }
diff --git a/c/main.c b/c/main.c
index b65d820f..c9188b3e 100644
--- a/c/main.c
+++ b/c/main.c
@@ -77,6 +77,11 @@ void run_tests() {
         printf("\n");
         pass = 0;
       }
+      jv as_string = jv_dump_string(jv_copy(expected), rand());
+      jv reparsed = jv_parse_sized(jv_string_value(as_string), jv_string_length(jv_copy(as_string)));
+      assert(jv_equal(jv_copy(expected), jv_copy(reparsed)));
+      jv_free(as_string);
+      jv_free(reparsed);
       jv_free(expected);
       jv_free(actual);
     }
diff --git a/c/testdata b/c/testdata
index f2eb32a2..8e66a02a 100644
--- a/c/testdata
+++ b/c/testdata
@@ -159,9 +159,9 @@ null
 "asdfasdf"
 {"a":1, "b":2, "c":3}
 
-"asdf" + "jkl;" + .
+"asdf" + "jkl;" + . + . + .
 "some string"
-"asdfjkl;some string"
+"asdfjkl;some stringsome stringsome string"
 
 "\u0000\u0020\u0000" + .
 "\u0000\u0020\u0000"