From 47a49263aedc01410f830fdb57812bd2e3185b24 Mon Sep 17 00:00:00 2001
From: Patrick Pelissier <patrick.pelissier@gmail.com>
Date: Wed, 2 May 2018 19:36:51 +0200
Subject: [PATCH] Add function _parse_str for string.

---
 m-core.h             |  2 +-
 m-string.h           | 44 ++++++++++++++++++++++++++++++++++++++++++++
 tests/test-mstring.c | 10 ++++++++++
 3 files changed, 55 insertions(+), 1 deletion(-)

diff --git a/m-core.h b/m-core.h
index 99598b52..1d47a5a6 100644
--- a/m-core.h
+++ b/m-core.h
@@ -1483,8 +1483,8 @@ m_core_hash (const void *str, size_t length)
 #define M_POP_POP(a)             ,a,
 #define M_REVERSE_REVERSE(a)     ,a,
 #define M_GET_STR_GET_STR(a)     ,a,
+#define M_PARSE_STR_PARSE_STR(a) ,a,
 #define M_OUT_STR_OUT_STR(a)     ,a,
-#define M_SET_STR_SET_STR(a)     ,a,
 #define M_IN_STR_IN_STR(a)       ,a,
 #define M_SEPARATOR_SEPARATOR(a) ,a,
 #define M_EXT_ALGO_EXT_ALGO(a)   ,a,
diff --git a/m-string.h b/m-string.h
index df4ed27a..8ea382da 100644
--- a/m-string.h
+++ b/m-string.h
@@ -890,6 +890,50 @@ string_in_str(string_t v, FILE *f)
   return c == '"';
 }
 
+static inline bool
+string_parse_str(string_t v, const char **str)
+{
+  STRINGI_CONTRACT(v);
+  M_ASSUME (str != NULL);
+  int c = *((*str)++);
+  if (c != '"') return false;
+  string_clean(v);
+  c = *((*str)++);
+  while (c != '"' && c != 0) {
+    if (M_UNLIKELY (c == '\\')) {
+      c = *((*str)++);
+      switch (c) {
+      case 'n':
+      case 't':
+      case 'r':
+      case '\\':
+      case '\"':
+        // This string acts as a perfect hashmap which supposes an ASCII mapping
+        // and (c^(c>>5)) is the hash function
+        c = " \r \" \n\\\t"[(c^(c>>5))& 0x07];
+        break;
+      default:
+        if (!(c >= '0' && c <= '7'))
+          return false;
+        int d1 = c - '0';
+        c = *((*str)++);
+        if (!(c >= '0' && c <= '7'))
+          return false;
+        int d2 = c - '0';
+        c = *((*str)++);
+        if (!(c >= '0' && c <= '7'))
+          return false;
+        int d3 = c - '0';
+        c = (d1 << 6) + (d2 << 3) + d3;
+        break;
+      }
+    }
+    string_push_back (v, c);
+    c = *((*str)++);
+  }
+  return c == '"';
+}
+
 /* UTF8 Handling */
 typedef enum {
   STRINGI_UTF8_STARTING = 0,
diff --git a/tests/test-mstring.c b/tests/test-mstring.c
index 8e382761..2a7337a4 100644
--- a/tests/test-mstring.c
+++ b/tests/test-mstring.c
@@ -288,6 +288,16 @@ static void test0(void)
   string_get_str(s2, s1, true);
   assert (string_cmp_str (s2, "\"Hello world\"\"Hello \\\"world\\\"\"") == 0);
 
+  const char *sp = string_get_cstr(s2);
+  b = string_parse_str(s1, &sp);
+  assert (b == true);
+  assert (strcmp(sp, "\"Hello \\\"world\\\"\"") == 0);
+  assert (string_cmp_str(s1, "Hello world") == 0);
+  b = string_parse_str(s1, &sp);
+  assert (b == true);
+  assert (strcmp(sp, "") == 0);
+  assert (string_cmp_str(s1, "Hello \"world\"") == 0);
+  
   string_set_str(s1, "Hello \"world\"");
   FILE *f = fopen ("a-mstring.dat", "wt");
   assert (f != NULL);