immortal

README.md

@@ -4,11 +4,7 @@ is a spartan fast configuration language.
 
 It has numbers, strings, lists, and sections.
 
-The reference implementation is WIP. I just got to finishing the data type and
+The reference implementation is WIP. There will most certainly be memory leaks.
-pretty printer today, a parser will be provided soon. I didn’t quite create a
-correct implementation today. There will most certainly also be memory leaks.
-I promised a friend I would share this “later today”, though, and I’m not one
-to break promises just because my code is crap. You have been warned.
 
 You’re not alone: I also wish the code were documented.
 
@@ -19,8 +15,8 @@ You’re not alone: I also wish the code were documented.
 ```
 # lists start with a name, then an indent, and then a value
 my_list
-  - "value"
+  -"value"
-  - 12
+  -12
 
 # strings are quoted
 my_string "this is a string"
@@ -41,7 +37,7 @@ possible, to write a simple, fast implementation in a few hundred lines of C
 (QED), and that might be worth a bit of reduction.
 
 See the [`examples/`](examples/) directory for an example of how to use the
-pretty printer.
+pretty printer and parser APIs.
 
 <hr/>
 

cfg.h

@@ -1,4 +1,5 @@
 #include <assert.h>
+#include <ctype.h>
 #include <stdbool.h>
 #include <stdlib.h>
 #include <stdio.h>
@@ -58,7 +59,7 @@ char* string_cstr(string* s) {
 }
 
 string string_from_cstr(char* c) {
-  int l = strlen(c);
+  size_t l = strlen(c);
   string s = new_string_sized(l);
   s.len = l;
   memcpy(s.str, c, s.len);
@@ -87,7 +88,7 @@ string string_slice(string* s, size_t front, size_t back) {
   assert(front < back);
   assert(front < s->len);
   assert(back < s->len);
-  int len = back - front;
+  size_t len = back - front;
   string res = new_string_sized(len);
   res.len = len;
   memcpy(res.str, s->str+front, len);
@@ -99,33 +100,44 @@ bool string_compare(string* a, string* b) {
   return strncmp(a->str, b->str, a->len) == 0;
 }
 
-ssize_t string_count_char(string* s, char c) {
+ssize_t string_count_escaped_chars(string* s) {
-  int i;
+  size_t i;
   ssize_t res = 0;
-  for (i = 0; i < s->len; i++) if (s->str[i] == c) res++;
+  for (i = 0; i < s->len; i++) if (s->str[i] == '"' || s->str[i] == '\\') res++;
   return res;
 }
 
 string string_quote(string* replace) {
-  int i = 0;
+  size_t i = 0;
-  ssize_t quotes = string_count_char(replace, '"');
+  ssize_t quotes = string_count_escaped_chars(replace);
   string res = new_string_sized(replace->len+quotes);
   res.len = res.cap;
 
   quotes = 0;
   for (i = 0; i < replace->len; i++) {
     if (replace->str[i] == '"') res.str[i+(quotes++)] = '\\';
+    else if (replace->str[i] == '\\') res.str[i+(quotes++)] = '\\';
     res.str[i+quotes] = replace->str[i];
   }
 
   return res;
 }
 
+ssize_t string_find_whitespace(string* s) {
+  ssize_t i;
+  char c;
+  for (i = 0; i < s->len; i++) {
+    c = s->str[i];
+    if (c == ' ' || c == '\n' || c == '\t') return i;
+  }
+  return -1;
+}
+
 // FNV-1a
 size_t string_hash(string* s) {
   size_t hash = 2166136261;
 
-  for (int i = 0; i < s->len; i++) {
+  for (size_t i = 0; i < s->len; i++) {
     hash ^= s->str[i];
     hash *= 16777619;
   }
@@ -213,7 +225,7 @@ typedef struct map_##key_type##_##val_type { \
 } map_##key_type##_##val_type; \
  \
 map_##key_type##_##val_type new_map_##key_type##_##val_type(size_t size) { \
-  int i; \
+  size_t i; \
   map_##key_type##_##val_type m;\
   m.entries = \
     new_list_sized_list_entry_##key_type##_##val_type(size); \
@@ -238,7 +250,7 @@ void map_put_##key_type##_##val_type(map_##key_type##_##val_type* m, key_type ke
 } \
  \
 val_type map_get_##key_type##_##val_type(map_##key_type##_##val_type* m, key_type key, val_type dflt) { \
-  int i; \
+  size_t i; \
   size_t hashed = hash(&key) % m->size; \
   list_entry_##key_type##_##val_type bucket = \
     list_nth_list_entry_##key_type##_##val_type(&m->entries, hashed); \
@@ -252,7 +264,7 @@ val_type map_get_##key_type##_##val_type(map_##key_type##_##val_type* m, key_typ
 } \
  \
 list_##key_type map_keys_##key_type##_##val_type(map_##key_type##_##val_type* m) { \
-  int i, j; \
+  size_t i, j; \
   list_entry_##key_type##_##val_type bucket; \
   list_##key_type res = new_list_##key_type(); \
   list_list_entry_##key_type##_##val_type entries = m->entries; \
@@ -268,7 +280,7 @@ list_##key_type map_keys_##key_type##_##val_type(map_##key_type##_##val_type* m)
 } \
  \
 list_##val_type map_vals_##key_type##_##val_type(map_##key_type##_##val_type* m) { \
-  int i, j; \
+  size_t i, j; \
   list_entry_##key_type##_##val_type bucket; \
   list_##val_type res = new_list_##val_type(); \
   list_list_entry_##key_type##_##val_type entries = m->entries; \
@@ -284,7 +296,7 @@ list_##val_type map_vals_##key_type##_##val_type(map_##key_type##_##val_type* m)
 } \
  \
 list_entry_##key_type##_##val_type map_entries_##key_type##_##val_type(map_##key_type##_##val_type* m) { \
-  int i, j; \
+  size_t i, j; \
   list_entry_##key_type##_##val_type bucket; \
   list_entry_##key_type##_##val_type res = new_list_entry_##key_type##_##val_type(); \
   list_list_entry_##key_type##_##val_type entries = m->entries; \
@@ -353,10 +365,10 @@ bool config_value_compare(config_value* a, config_value* b) {
   return false;
 }
 
-string config_str(config* c, unsigned int indent);
+string config_str(config* c, size_t indent);
 
-string config_value_str(config_value* c, unsigned int indent) {
+string config_value_str(config_value* c, size_t indent) {
-  int i, j;
+  size_t i, j;
   char tmp[128];
   string sub;
   string res = new_string();
@@ -382,7 +394,7 @@ string config_value_str(config_value* c, unsigned int indent) {
       for (i = 0; i < c->l->len; i++) {
         string_cappend(&res, "\n");
         for (j = 0; j < indent; j++) string_cappend(&res, " ");
-        string_cappend(&res, "  - ");
+        string_cappend(&res, "  -");
         val = list_nth_config_value(c->l, i);
         sub = config_value_str(&val, 0);
         string_append(&res, &sub);
@@ -428,8 +440,8 @@ config new_config() {
   return c;
 }
 
-string config_str(config* c, unsigned int indent) {
+string config_str(config* c, size_t indent) {
-  int i, j;
+  size_t i, j;
   string key;
   config_value val;
   list_string keys = map_keys_string_config_value(&c->values);
@@ -467,3 +479,233 @@ void config_add_number(config* c, string label, double d) {
 void config_add_list(config* c, string label, list_config_value* l) {
   map_put_string_config_value(&c->values, label, config_list(l));
 }
+
+// Parser
+
+typedef struct parsed_value {
+  bool ok;
+  size_t consumed;
+  union {
+    string err;
+    config_value v;
+  };
+} parsed_value;
+
+typedef struct parsed_config {
+  bool ok;
+  size_t consumed;
+  union {
+    string err;
+    config c;
+  };
+} parsed_config;
+
+bool config_parse_line(string* s, size_t* line, size_t* col) {
+  bool seen = false;
+  while (*s->str == '\n') {
+    seen = true;
+    s->str++;
+    (*line)++;
+    *col = 1;
+  }
+  return seen;
+}
+
+ssize_t config_parse_indent(string* s, size_t* col) {
+  ssize_t count = 0;
+  while (*s->str == ' ') {
+    s->str++;
+    count++;
+    (*col)++;
+  }
+  return count;
+}
+
+parsed_value config_value_parse_error(char* msg, size_t line, size_t col) {
+  parsed_value res;
+  char tmp[128];
+  res.ok = false;
+  snprintf(tmp, 128, "%zu:%zu: %s", line, col, msg);
+  res.err = string_from_cstr(tmp);
+  return res;
+}
+
+parsed_value config_parse_value(string*, size_t*, size_t*, size_t);
+parsed_config config_parse_internal(string*, size_t*, size_t*, size_t);
+
+parsed_value config_parse_list(string* s, size_t* line, size_t* col, size_t indent, size_t consumed) {
+  parsed_value elem;
+  size_t ind;
+  list_config_value* l = malloc(sizeof(list_config_value));
+  *l =new_list_config_value();
+
+  while(true) {
+    elem = config_parse_value(s, line, col, indent);
+
+    if (!elem.ok) return elem;
+
+    list_push_config_value(l, elem.v);
+    consumed += elem.consumed;
+    s->str += elem.consumed;
+
+    if (!config_parse_line(s, line, col)) return config_value_parse_error("expected newline", *line, *col);
+
+    ind = config_parse_indent(s, col);
+
+    if(ind < indent) { s->str -= ind; col -= ind; break; }
+    if (ind > indent) return config_value_parse_error("unexpected indent", *line, *col);
+    consumed += ind;
+
+    if (s->str[0] != '-') return config_value_parse_error("expected hyphen to start list element", *line, *col);
+    s->str++;
+    consumed += 1;
+  }
+
+  return (parsed_value){.ok=true, .consumed=consumed, .v=config_list(l)};
+}
+
+parsed_value config_parse_section(string* s, size_t* line, size_t* col, size_t indent, size_t consumed) {
+  config* res;
+  *col -= consumed;
+  s->str -= consumed;
+  parsed_config c = config_parse_internal(s, line, col, indent);
+
+  if (!c.ok) return (parsed_value){.ok=false, .consumed=c.consumed, .err=c.err};
+
+  res = malloc(sizeof(config));
+  *res = c.c;
+  return (parsed_value){.ok=true, consumed=c.consumed, .v=config_section(res)};
+}
+
+parsed_value config_parse_list_or_section(string* s, size_t* line, size_t* col, size_t indent) {
+  size_t ind;
+
+  if (!config_parse_line(s, line, col)) return config_value_parse_error("expected newline", *line, *col);
+
+  ind = config_parse_indent(s, col);
+  if (ind < indent) return config_value_parse_error("expected greater indent", *line, *col);
+  if (ind > indent) return config_value_parse_error("unexpected indent", *line, *col);
+
+  if (s->str[0] == '-') { s->str++; return config_parse_list(s, line, col, indent, ind+1); }
+  return config_parse_section(s, line, col, indent, ind);
+}
+
+parsed_value config_parse_string(string* s, size_t* line, size_t* col) {
+  size_t i;
+  for (i = 1; i < s->len; i++) {
+    (*col)++;
+    if (s->str[i] == '\\') {
+      if (i +1 == s->len) return config_value_parse_error("Unexpected EOF, expected end of string", *line, *col);
+      i++;
+    } else if (s->str[i] == '"') {
+      return (parsed_value){.ok=true, .consumed=i+1, .v=config_string(string_slice(s, 1, i))};
+    } else if (s->str[i] == '\0') {
+      break;
+    } else if (s->str[i] == '\n') {
+      (*line)++;
+      *col = 0;
+    }
+  }
+  return config_value_parse_error("Unterminated string", *line, *col);
+}
+
+parsed_value config_parse_number(string* s, size_t* line, size_t* col) {
+  int i;
+  ssize_t end = string_find_whitespace(s);
+  bool dot = false;
+  string to_parse = string_slice(s, 0, end == -1 ? s->len-1 : end);
+
+  for (i = 0; i < to_parse.len; i++) {
+    (*col)++;
+    if (isdigit(to_parse.str[i])) {
+    } else if (!dot && to_parse.str[i] == '.') {
+      dot = true;
+    } else if (i == 0 && to_parse.str[i] == '-') {
+    } else if (to_parse.str[i] == '\0') {
+      break;
+    } else {
+      return config_value_parse_error("Expected number, got unparseable", *line, *col);
+    }
+  }
+
+  return (parsed_value){
+    .ok=true,
+    .consumed=to_parse.len,
+    .v=config_number(strtod(string_cstr(&to_parse), NULL))
+  };
+}
+
+parsed_value config_parse_value(string* s, size_t* line, size_t* col, size_t indent) {
+  if (s->str[0] == '"') return config_parse_string(s, line, col);
+  if (s->str[0] == '\n') return config_parse_list_or_section(s, line, col, indent+2);
+  return config_parse_number(s, line, col);
+}
+
+parsed_config config_parse_error(char* msg, size_t line, size_t col) {
+  parsed_config res;
+  char tmp[128];
+  res.ok = false;
+  snprintf(tmp, 128, "%zu:%zu: %s", line, col, msg);
+  res.err = string_from_cstr(tmp);
+  return res;
+}
+
+void config_parse_trim(string* s, size_t* line, size_t* col, size_t* consumed) {
+  while (*s->str == ' ' || *s->str == '\t') {
+    (*col)++;
+    s->str++;
+    (*consumed)++;
+  }
+}
+
+parsed_config config_parse_internal(string* s, size_t* line, size_t* col, size_t indent) {
+  parsed_config res;
+  parsed_value val;
+  string label;
+  size_t ind;
+  ssize_t whitespace;
+  size_t consumed = 0;
+  map_string_config_value values = new_map_string_config_value(1024);
+
+  while(s->str[0] != '\0') {
+    ind = config_parse_indent(s, col);
+
+    if (ind > indent) return config_parse_error("unexpected indent", *line, *col);
+    if (ind < indent) break;
+
+    consumed += ind;
+
+    whitespace = string_find_whitespace(s);
+
+    if (whitespace == -1) return config_parse_error("expected whitespace, got EOF", *line, (*col)+s->len);
+
+    label = string_slice(s, 0, whitespace);
+    s->str += whitespace;
+    *col += whitespace;
+    consumed += consumed;
+    config_parse_trim(s, line, col, &consumed);
+
+    val = config_parse_value(s, line, col, indent);
+
+    if (!val.ok) return (parsed_config){.ok=false, .err=val.err};
+
+    s->str += val.consumed;
+    consumed += val.consumed;
+
+    map_put_string_config_value(&values, label, val.v);
+
+    if (!config_parse_line(s, line, col) && s->str[0] != '\0') return config_parse_error("expected newline", *line, *col);
+    consumed++;
+  }
+
+  res.c.values = values;
+  res.ok = true;
+  res.consumed = consumed;
+  return res;
+}
+
+parsed_config config_parse(string* s) {
+  size_t line = 1;
+  size_t col = 1;
+  return config_parse_internal(s, &line, &col, 0);
+}

example.cfg

@@ -0,0 +1,4 @@
+x 1.0
+hi
+  name 1
+  val "hi"

examples/parser_simple.c

@@ -0,0 +1,37 @@
+#include "../cfg.h"
+
+int main(int argc, char** argv) {
+  if (argc != 2) {
+    printf("Usage: %s <config>\n", argv[0]);
+    return 1;
+  }
+  FILE* f = fopen(argv[1], "rb");
+
+  fseek(f, 0, SEEK_END);
+  ssize_t s = ftell(f);
+  fseek(f, 0, SEEK_SET);
+
+  char* contents = malloc(s + 1);
+  fread(contents, 1, s, f);
+  fclose(f);
+
+  contents[s] = 0;
+
+  string arg = string_from_cstr(contents);
+  parsed_config c = config_parse(&arg);
+
+  if (!c.ok) {
+    printf("Error: %s\n", c.err.str);
+    return 1;
+  }
+
+  string str = config_str(&c.c, 0);
+  char* cstr = string_cstr(&str);
+
+  printf("%s\n", cstr);
+
+  free_string(str);
+  free(cstr);
+  free(contents);
+  return 0;
+}