Consolidate reading of input chunks in parser

include/tree_sitter/parser.h

@@ -38,12 +38,11 @@ typedef struct {
 
 typedef struct {
     ts_input input;
-
     const char *chunk;
     size_t chunk_start;
     size_t chunk_size;
     size_t position_in_chunk;
-    
+
     size_t token_end_position;
     size_t token_start_position;
 
@@ -56,29 +55,30 @@ typedef struct {
 
 static void ts_lex(ts_parser *parser);
 static const ts_symbol * ts_recover(ts_state state, ts_state *to_state, size_t *count);
+static void ts_parser_advance(ts_parser *);
 
 static ts_parser ts_parser_make(ts_input input) {
-    size_t bytes_read = 0;
-    const char *chunk = input.read_fn(input.data, &bytes_read);
     ts_parser result = {
         .input = input,
+        .chunk = NULL,
+        .chunk_start = 0,
+        .chunk_size = 0,
+        .position_in_chunk = 0,
+
         .token_start_position = 0,
         .token_end_position = 0,
 
-        .chunk = chunk,
-        .chunk_size = bytes_read,
-        .chunk_start = 0,
-        .position_in_chunk = 0,
-
         .lookahead_node = NULL,
         .prev_lookahead_node = NULL,
         .lex_state = 0,
         .stack = calloc(INITIAL_STACK_SIZE, sizeof(ts_stack_entry)),
         .stack_size = 0,
     };
+
+    ts_parser_advance(&result);
     return result;
 }
-    
+
 static size_t ts_parser_position(const ts_parser *parser) {
     return parser->chunk_start + parser->position_in_chunk;
 }
@@ -157,7 +157,7 @@ static void ts_parser_reduce(ts_parser *parser, ts_symbol symbol, int immediate_
 static const char empty_chunk[1] = { '\0' };
 
 static void ts_parser_advance(ts_parser *parser) {
-    if (parser->position_in_chunk < parser->chunk_size - 1) {
+    if (parser->position_in_chunk + 1 < parser->chunk_size) {
         parser->position_in_chunk++;
     } else {
         parser->chunk_start += parser->chunk_size;

spec/runtime/helpers/spy_reader.cc

@@ -15,13 +15,13 @@ static const char * spy_read(void *data, size_t *bytes_read) {
 }
 
 static int spy_seek(void *data, size_t position) {
-    SpyReader *reader = static_cast<SpyReader *>(data); 
+    SpyReader *reader = static_cast<SpyReader *>(data);
     reader->position = position;
     return 0;
 }
 
 static void spy_release(void *data) {
-    SpyReader *reader = static_cast<SpyReader *>(data); 
+    SpyReader *reader = static_cast<SpyReader *>(data);
     delete reader;
 }
 

spec/runtime/parser_spec.cc

@@ -7,12 +7,12 @@ START_TEST
 
 describe("reading from an input", [&]() {
     ts_document *doc;
-    
+
     before_each([&]() {
         doc = ts_document_make();
         ts_document_set_parser(doc, ts_parse_config_json);
     });
-    
+
     after_each([&]() {
         ts_document_free(doc);
     });
@@ -20,7 +20,7 @@ describe("reading from an input", [&]() {
     it("reads the entire input", [&]() {
         SpyReader reader("\"ok go do it!\"", 3);
         ts_document_set_input(doc, reader.input);
-        
+
         AssertThat(string(ts_document_string(doc)), Equals("(value (string))"));
         AssertThat(reader.chunks_read, Equals(vector<string>({
             "\"ok",

src/compiler/prepared_grammar.h

@@ -9,7 +9,7 @@
 
 namespace tree_sitter {
     class PreparedGrammar : public Grammar {
-    public: 
+    public:
         PreparedGrammar(std::string start_rule_name,
                         const std::map<const std::string, const rules::rule_ptr> &rules,
                         const std::map<const std::string, const rules::rule_ptr> &aux_rules);