upgrade to the new lexer complete

4coder_default_bindings.cpp

@@ -378,8 +378,8 @@ default_keys(Bind_Helper *context){
     bind(context, 's', MDFR_CTRL, cmdid_save);
     bind(context, 'u', MDFR_CTRL, to_uppercase);
     bind(context, 'U', MDFR_CTRL, rewrite_as_single_caps);
-    bind(context, 'v', MDFR_CTRL, paste);
+    bind(context, 'v', MDFR_CTRL, paste_and_indent);
-    bind(context, 'V', MDFR_CTRL, paste_next);
+    bind(context, 'V', MDFR_CTRL, paste_next_and_indent);
     bind(context, 'x', MDFR_CTRL, cut);
     bind(context, 'y', MDFR_CTRL, cmdid_redo);
     bind(context, 'z', MDFR_CTRL, cmdid_undo);

4coder_default_include.cpp

@@ -15,6 +15,10 @@
 # define DEFAULT_INDENT_FLAGS 0
 #endif
 
+#ifndef DEF_TAB_WIDTH
+# define DEF_TAB_WIDTH 4
+#endif
+
 
 //
 // Memory
@@ -694,6 +698,205 @@ CUSTOM_COMMAND_SIG(move_right){
                          true);
 }
 
+//
+// Auto Indenting and Whitespace
+//
+
+static int
+seek_line_end(Application_Links *app, Buffer_Summary *buffer, int pos){
+    char chunk[1024];
+    int chunk_size = sizeof(chunk);
+    Stream_Chunk stream = {0};
+    
+    int still_looping;
+    char at_pos;
+    
+    if (init_stream_chunk(&stream, app, buffer, pos, chunk, chunk_size)){
+        still_looping = 1;
+        do{
+            for (; pos < stream.end; ++pos){
+                at_pos = stream.data[pos];
+                if (at_pos == '\n'){
+                    goto double_break;
+                }
+            }
+            still_looping = forward_stream_chunk(&stream);
+        }while(still_looping);
+        double_break:;
+        
+        if (pos > buffer->size){
+            pos = buffer->size;
+        }
+    }
+    
+    return(pos);
+}
+
+static int
+seek_line_beginning(Application_Links *app, Buffer_Summary *buffer, int pos){
+    char chunk[1024];
+    int chunk_size = sizeof(chunk);
+    Stream_Chunk stream = {0};
+    
+    int still_looping;
+    char at_pos;
+    
+    --pos;
+    if (init_stream_chunk(&stream, app, buffer, pos, chunk, chunk_size)){
+        still_looping = 1;
+        do{
+            for (; pos >= stream.start; --pos){
+                at_pos = stream.data[pos];
+                if (at_pos == '\n'){
+                    goto double_break;
+                }
+            }
+            still_looping = backward_stream_chunk(&stream);
+        }while(still_looping);
+        double_break:;
+        
+        if (pos != 0){
+            ++pos;
+        }
+        if (pos < 0){
+            pos = 0;
+        }
+    }
+    
+    return(pos);
+}
+
+static void
+move_past_lead_whitespace(Application_Links *app, View_Summary *view, Buffer_Summary *buffer){
+    refresh_view(app, view);
+    
+    int new_pos = seek_line_beginning(app, buffer, view->cursor.pos);
+    char space[1024];
+    Stream_Chunk chunk = {0};
+    int still_looping = false;
+    
+    int i = new_pos;
+    if (init_stream_chunk(&chunk, app, buffer, i, space, sizeof(space))){
+        do{
+            for (; i < chunk.end; ++i){
+                char at_pos = chunk.data[i];
+                if (at_pos == '\n' || !char_is_whitespace(at_pos)){
+                    goto break2;
+                }
+            }
+            still_looping = forward_stream_chunk(&chunk);
+        }while(still_looping);
+        break2:;
+        
+        if (i > view->cursor.pos){
+            app->view_set_cursor(app, view, seek_pos(i), true);
+        }
+    }
+}
+
+CUSTOM_COMMAND_SIG(auto_tab_line_at_cursor){
+    unsigned int access = AccessOpen;
+    View_Summary view = app->get_active_view(app, access);
+    Buffer_Summary buffer = app->get_buffer(app, view.buffer_id, access);
+    
+    app->buffer_auto_indent(app, &buffer,
+                            view.cursor.pos, view.cursor.pos,
+                            DEF_TAB_WIDTH,
+                            DEFAULT_INDENT_FLAGS);
+    move_past_lead_whitespace(app, &view, &buffer);
+}
+
+CUSTOM_COMMAND_SIG(auto_tab_whole_file){
+    unsigned int access = AccessOpen;
+    View_Summary view = app->get_active_view(app, access);
+    Buffer_Summary buffer = app->get_buffer(app, view.buffer_id, access);
+    
+    app->buffer_auto_indent(app, &buffer,
+                            0, buffer.size,
+                            DEF_TAB_WIDTH,
+                            DEFAULT_INDENT_FLAGS);
+}
+
+CUSTOM_COMMAND_SIG(auto_tab_range){
+    unsigned int access = AccessOpen;
+    View_Summary view = app->get_active_view(app, access);
+    Buffer_Summary buffer = app->get_buffer(app, view.buffer_id, access);
+    Range range = get_range(&view);
+    
+    app->buffer_auto_indent(app, &buffer,
+                            range.min, range.max,
+                            DEF_TAB_WIDTH,
+                            DEFAULT_INDENT_FLAGS);
+    move_past_lead_whitespace(app, &view, &buffer);
+}
+
+CUSTOM_COMMAND_SIG(write_and_auto_tab){
+    exec_command(app, write_character);
+    exec_command(app, auto_tab_line_at_cursor);
+}
+
+CUSTOM_COMMAND_SIG(clean_all_lines){
+    // TODO(allen): This command always iterates accross the entire
+    // buffer, so streaming it is actually the wrong call.  Rewrite this
+    // to minimize calls to app->buffer_read_range.
+    View_Summary view = app->get_active_view(app, AccessOpen);
+    Buffer_Summary buffer = app->get_buffer(app, view.buffer_id, AccessOpen);
+    
+    int line_count = buffer.line_count;
+    int edit_max = line_count;
+    
+    if (edit_max*sizeof(Buffer_Edit) < app->memory_size){
+        Buffer_Edit *edits = (Buffer_Edit*)app->memory;
+        
+        char data[1024];
+        Stream_Chunk chunk = {0};
+        
+        int i = 0;
+        if (init_stream_chunk(&chunk, app, &buffer,
+                              i, data, sizeof(data))){
+            Buffer_Edit *edit = edits;
+            
+            int buffer_size = buffer.size;
+            int still_looping = true;
+            int last_hard = buffer_size;
+            do{
+                for (; i < chunk.end; ++i){
+                    char at_pos = chunk.data[i];
+                    if (at_pos == '\n'){
+                        if (last_hard+1 < i){
+                            edit->str_start = 0;
+                            edit->len = 0;
+                            edit->start = last_hard+1;
+                            edit->end = i;
+                            ++edit;
+                        }
+                        last_hard = buffer_size;
+                    }
+                    else if (char_is_whitespace(at_pos)){
+                        // NOTE(allen): do nothing
+                    }
+                    else{
+                        last_hard = i;
+                    }
+                }
+                
+                still_looping = forward_stream_chunk(&chunk);
+            }while(still_looping);
+            
+            if (last_hard+1 < buffer_size){
+                edit->str_start = 0;
+                edit->len = 0;
+                edit->start = last_hard+1;
+                edit->end = buffer_size;
+                ++edit;
+            }
+            
+            int edit_count = (int)(edit - edits);
+            app->buffer_batch_edit(app, &buffer, 0, 0, edits, edit_count, BatchEdit_PreserveTokens);
+        }
+    }
+}
+
 //
 // Clipboard
 //
@@ -843,6 +1046,16 @@ CUSTOM_COMMAND_SIG(paste_next){
     }
 }
 
+CUSTOM_COMMAND_SIG(paste_and_indent){
+    exec_command(app, paste);
+    exec_command(app, auto_tab_range);
+}
+
+CUSTOM_COMMAND_SIG(paste_next_and_indent){
+    exec_command(app, paste_next);
+    exec_command(app, auto_tab_range);
+}
+
 //
 // Fancy Editing
 //
@@ -1032,70 +1245,6 @@ CUSTOM_COMMAND_SIG(seek_whitespace_down){
                          true);
 }
 
-static int
-seek_line_end(Application_Links *app, Buffer_Summary *buffer, int pos){
-    char chunk[1024];
-    int chunk_size = sizeof(chunk);
-    Stream_Chunk stream = {0};
-    
-    int still_looping;
-    char at_pos;
-    
-    if (init_stream_chunk(&stream, app, buffer, pos, chunk, chunk_size)){
-        still_looping = 1;
-        do{
-            for (; pos < stream.end; ++pos){
-                at_pos = stream.data[pos];
-                if (at_pos == '\n'){
-                    goto double_break;
-                }
-            }
-            still_looping = forward_stream_chunk(&stream);
-        }while(still_looping);
-        double_break:;
-        
-        if (pos > buffer->size){
-            pos = buffer->size;
-        }
-    }
-    
-    return(pos);
-}
-
-static int
-seek_line_beginning(Application_Links *app, Buffer_Summary *buffer, int pos){
-    char chunk[1024];
-    int chunk_size = sizeof(chunk);
-    Stream_Chunk stream = {0};
-    
-    int still_looping;
-    char at_pos;
-    
-    --pos;
-    if (init_stream_chunk(&stream, app, buffer, pos, chunk, chunk_size)){
-        still_looping = 1;
-        do{
-            for (; pos >= stream.start; --pos){
-                at_pos = stream.data[pos];
-                if (at_pos == '\n'){
-                    goto double_break;
-                }
-            }
-            still_looping = backward_stream_chunk(&stream);
-        }while(still_looping);
-        double_break:;
-        
-        if (pos != 0){
-            ++pos;
-        }
-        if (pos < 0){
-            pos = 0;
-        }
-    }
-    
-    return(pos);
-}
-
 CUSTOM_COMMAND_SIG(seek_end_of_line){
     unsigned int access = AccessProtected;
     View_Summary view = app->get_active_view(app, access);
@@ -1163,38 +1312,6 @@ CUSTOM_COMMAND_SIG(write_increment){
     write_string(app, make_lit_string("++"));
 }
 
-#ifndef DEF_TAB_WIDTH
-# define DEF_TAB_WIDTH 4
-#endif
-
-static void
-move_past_lead_whitespace(Application_Links *app, View_Summary *view, Buffer_Summary *buffer){
-    refresh_view(app, view);
-    
-    int new_pos = seek_line_beginning(app, buffer, view->cursor.pos);
-    char space[1024];
-    Stream_Chunk chunk = {0};
-    int still_looping = false;
-    
-    int i = new_pos;
-    if (init_stream_chunk(&chunk, app, buffer, i, space, sizeof(space))){
-        do{
-            for (; i < chunk.end; ++i){
-                char at_pos = chunk.data[i];
-                if (at_pos == '\n' || !char_is_whitespace(at_pos)){
-                    goto break2;
-                }
-            }
-            still_looping = forward_stream_chunk(&chunk);
-        }while(still_looping);
-        break2:;
-        
-        if (i > view->cursor.pos){
-            app->view_set_cursor(app, view, seek_pos(i), true);
-        }
-    }
-}
-
 static void
 long_braces(Application_Links *app, char *text, int size){
     unsigned int access = AccessOpen;
@@ -1873,113 +1990,6 @@ CUSTOM_COMMAND_SIG(execute_previous_cli){
     }
 }
 
-//
-// Auto Indenting and Whitespace
-//
-
-CUSTOM_COMMAND_SIG(auto_tab_line_at_cursor){
-    unsigned int access = AccessOpen;
-    View_Summary view = app->get_active_view(app, access);
-    Buffer_Summary buffer = app->get_buffer(app, view.buffer_id, access);
-    
-    app->buffer_auto_indent(app, &buffer,
-                            view.cursor.pos, view.cursor.pos,
-                            DEF_TAB_WIDTH,
-                            DEFAULT_INDENT_FLAGS);
-    move_past_lead_whitespace(app, &view, &buffer);
-}
-
-CUSTOM_COMMAND_SIG(auto_tab_whole_file){
-    unsigned int access = AccessOpen;
-    View_Summary view = app->get_active_view(app, access);
-    Buffer_Summary buffer = app->get_buffer(app, view.buffer_id, access);
-    
-    app->buffer_auto_indent(app, &buffer,
-                            0, buffer.size,
-                            DEF_TAB_WIDTH,
-                            DEFAULT_INDENT_FLAGS);
-}
-
-CUSTOM_COMMAND_SIG(auto_tab_range){
-    unsigned int access = AccessOpen;
-    View_Summary view = app->get_active_view(app, access);
-    Buffer_Summary buffer = app->get_buffer(app, view.buffer_id, access);
-    Range range = get_range(&view);
-    
-    app->buffer_auto_indent(app, &buffer,
-                            range.min, range.max,
-                            DEF_TAB_WIDTH,
-                            DEFAULT_INDENT_FLAGS);
-    move_past_lead_whitespace(app, &view, &buffer);
-}
-
-CUSTOM_COMMAND_SIG(write_and_auto_tab){
-    exec_command(app, write_character);
-    exec_command(app, auto_tab_line_at_cursor);
-}
-
-CUSTOM_COMMAND_SIG(clean_all_lines){
-    // TODO(allen): This command always iterates accross the entire
-    // buffer, so streaming it is actually the wrong call.  Rewrite this
-    // to minimize calls to app->buffer_read_range.
-    View_Summary view = app->get_active_view(app, AccessOpen);
-    Buffer_Summary buffer = app->get_buffer(app, view.buffer_id, AccessOpen);
-    
-    int line_count = buffer.line_count;
-    int edit_max = line_count;
-    
-    if (edit_max*sizeof(Buffer_Edit) < app->memory_size){
-        Buffer_Edit *edits = (Buffer_Edit*)app->memory;
-        
-        char data[1024];
-        Stream_Chunk chunk = {0};
-        
-        int i = 0;
-        if (init_stream_chunk(&chunk, app, &buffer,
-                              i, data, sizeof(data))){
-            Buffer_Edit *edit = edits;
-            
-            int buffer_size = buffer.size;
-            int still_looping = true;
-            int last_hard = buffer_size;
-            do{
-                for (; i < chunk.end; ++i){
-                    char at_pos = chunk.data[i];
-                    if (at_pos == '\n'){
-                        if (last_hard+1 < i){
-                            edit->str_start = 0;
-                            edit->len = 0;
-                            edit->start = last_hard+1;
-                            edit->end = i;
-                            ++edit;
-                        }
-                        last_hard = buffer_size;
-                    }
-                    else if (char_is_whitespace(at_pos)){
-                        // NOTE(allen): do nothing
-                    }
-                    else{
-                        last_hard = i;
-                    }
-                }
-                
-                still_looping = forward_stream_chunk(&chunk);
-            }while(still_looping);
-            
-            if (last_hard+1 < buffer_size){
-                edit->str_start = 0;
-                edit->len = 0;
-                edit->start = last_hard+1;
-                edit->end = buffer_size;
-                ++edit;
-            }
-            
-            int edit_count = (int)(edit - edits);
-            app->buffer_batch_edit(app, &buffer, 0, 0, edits, edit_count, BatchEdit_PreserveTokens);
-        }
-    }
-}
-
 //
 // Default Building Stuff
 //

4coder_jump_parsing.cpp

@@ -14,15 +14,6 @@ jump_to_location(Application_Links *app, View_Summary *view, Jump_Location *l){
     app->view_set_cursor(app, view, seek_line_char(l->line, l->column), true);
 }
 
-static int
-gcc_style_verify(String line, int colon_pos){
-    int result = false;
-    if (colon_pos < line.size){
-        result = true;
-    }
-    return(result);
-}
-
 static int
 ms_style_verify(String line, int paren_pos){
     int result = false;
@@ -43,13 +34,16 @@ parse_error(String line, Jump_Location *location,
             int skip_sub_errors, int *colon_char){
     int result = false;
     
+    String original_line = line;
+    line = skip_chop_whitespace(line);
+    
     int colon_pos = find(line, 0, ')');
     if (ms_style_verify(line, colon_pos)){
         colon_pos = find(line, colon_pos, ':');
         if (colon_pos < line.size){
             String location_str = substr(line, 0, colon_pos);
             
-            if (!(skip_sub_errors && line.str[0] == ' ')){
+            if (!(skip_sub_errors && original_line.str[0] == ' ')){
                 location_str = skip_chop_whitespace(location_str);
                 
                 int paren_pos = find(location_str, 0, '(');
@@ -92,18 +86,16 @@ parse_error(String line, Jump_Location *location,
     
     else{
         int colon_pos1 = find(line, 0, ':');
-        if (colon_pos1 == 1){
+        if (line.size > colon_pos1+1){
-            if (line.size > colon_pos1+1){
+            if (char_is_slash(line.str[colon_pos1+1])){
-                if (char_is_slash(line.str[colon_pos1+1])){
+                colon_pos1 = find(line, colon_pos1+1, ':');
-                    colon_pos1 = find(line, colon_pos1+1, ':');
-                }
             }
         }
         
         int colon_pos2 = find(line, colon_pos1+1, ':');
         int colon_pos3 = find(line, colon_pos2+1, ':');
         
-        if (gcc_style_verify(line, colon_pos3)){
+        if (colon_pos3 < line.size){
             String filename = substr(line, 0, colon_pos1);
             String line_number = substr(line, colon_pos1+1, colon_pos2 - colon_pos1 - 1);
             String column_number = substr(line, colon_pos2+1, colon_pos3 - colon_pos2 - 1);
@@ -119,8 +111,14 @@ parse_error(String line, Jump_Location *location,
             }
         }
         else{
-            int colon_pos1 = find(line, 0, ':');
+            colon_pos1 = find(line, 0, ':');
-            int colon_pos2 = find(line, colon_pos1+1, ':');
+            if (line.size > colon_pos1+1){
+                if (char_is_slash(line.str[colon_pos1+1])){
+                    colon_pos1 = find(line, colon_pos1+1, ':');
+                }
+            }
+            
+            colon_pos2 = find(line, colon_pos1+1, ':');
             
             if (colon_pos2 < line.size){
                 String filename = substr(line, 0, colon_pos1);

4cpp_lexer.h

@@ -869,11 +869,11 @@ cpp_read_block_comment(char *data, int size, int pos){
         if (data[pos] == '*' &&
             pos + 1 < size &&
             data[pos+1] == '/'){
+            pos += 2;
             break;
         }
         ++pos;
     }
-    pos += 2;
     result.token.size = pos - result.token.start;
     result.token.type = CPP_TOKEN_COMMENT;
     result.pos = pos;

4ed_app_target.cpp

@@ -27,9 +27,14 @@
 
 #include "4coder_table.cpp"
 
+#define USE_NEW_LEXER 1
+
+#if USE_NEW_LEXER
+#include "test/4cpp_new_lexer.h"
+#else
 #define FCPP_LEXER_IMPLEMENTATION
-//#include "test/4cpp_new_lexer.h"
 #include "4cpp_lexer.h"
+#endif
 
 #include "4ed_template.cpp"
 

4ed_file_view.cpp

@@ -1183,7 +1183,7 @@ Job_Callback_Sig(job_full_lex){
     i32 buffer_size = file->state.buffer.size;
     buffer_size = (buffer_size + 3)&(~3);
     
-#if 0
+#if USE_NEW_LEXER
     while (memory->size < buffer_size*2){
         system->grow_thread_memory(memory);
     }
@@ -1202,7 +1202,7 @@ Job_Callback_Sig(job_full_lex){
     do{
         i32 result = 
             cpp_lex_size_nonalloc(&lex,
-                                  cpp_file.data, cpp_file.size, cpp_file.size,
+                                  text_data, text_size, text_size,
                                   &tokens, 2048);
         
         switch (result){
@@ -1376,9 +1376,12 @@ file_relex_parallel(System_Functions *system,
         relex_space.max_count = state.space_request;
         relex_space.tokens = push_array(part, Cpp_Token, relex_space.max_count);
         
-        //        char *spare = push_array(part, char, cpp_file.size);
+#if USE_NEW_LEXER
-        //        if (cpp_relex_nonalloc_main(&state, &relex_space, &relex_end, spare)){
+        char *spare = push_array(part, char, size);
+        if (cpp_relex_nonalloc_main(&state, &relex_space, &relex_end, spare)){
+#else
         if (cpp_relex_nonalloc_main(&state, &relex_space, &relex_end)){
+#endif
             inline_lex = 0;
         }
         else{

build.bat

@@ -1,6 +1,6 @@
 @echo off
 
-REM "build_exp.bat" /Zi
+REM "build_exp.bat" /O2
 "build_all.bat" /DFRED_SUPER /DFRED_INTERNAL /Zi
 REM "build_all.bat" /DFRED_INTERNAL /Zi
 REM "build_all.bat" /O2 /Zi

test/experiment.cpp

@@ -206,6 +206,19 @@ end_t(Times *t){
     *t = time;
 }
 
+static int
+equivalent_comments(Cpp_Token *a, Cpp_Token *b, char *data){
+    String s_a = make_string(data + a->start, a->size);
+    String s_b = make_string(data + b->start, b->size);
+    
+    s_a = skip_chop_whitespace(s_a);
+    s_b = skip_chop_whitespace(s_b);
+    
+    int result = match(s_a, s_b);
+    
+    return(result);
+}
+
 static void
 run_experiment(Experiment *exp, char *filename, int verbose,
                int chunks, int max_tokens){
@@ -266,16 +279,17 @@ run_experiment(Experiment *exp, char *filename, int verbose,
                     }
                     else{
                         start = __rdtsc();
-                        new_lex::cpp_lex_nonalloc(&ld,
+                        new_lex::cpp_lex_size_nonalloc(
-                                                  (char*)file_data.data, file_data.size,
+                            &ld, (char*)file_data.data, file_data.size,
-                                                  &exp->testing_stack);
+                            file_data.size, &exp->testing_stack);
+                        
                         time.fsm += (__rdtsc() - start);
                     }
                 }
                 else{
                     if (chunks){
                         start = __rdtsc();
-                        int relevant_size = file_data.size + 1;
+                        int relevant_size = file_data.size;
                         is_last = 0;
                         for (k = 0; k < relevant_size; k += chunks){
                             chunk_size = chunks;
@@ -289,7 +303,8 @@ run_experiment(Experiment *exp, char *filename, int verbose,
                             do{
                                 result =
                                     new_lex::cpp_lex_size_nonalloc(&ld,
-                                                                   (char*)file_data.data + k, chunk_size, file_data.size,
+                                                                   (char*)file_data.data + k, chunk_size,
+                                                                   file_data.size,
                                                                    &exp->testing_stack,
                                                                    max_tokens);
                                 if (result == new_lex::LexFinished ||
@@ -348,15 +363,27 @@ run_experiment(Experiment *exp, char *filename, int verbose,
                 }
                 
                 if (correct->start != testing->start || correct->size != testing->size){
-                    pass = 0;
+                    
-                    if (verbose >= 1){
+                    int mismatch = 1;
-                        printf("token range mismatch at token %d\n"
+                    if (correct->type == testing->type &&
-                               "    %d:%d original %d:%d testing\n"
+                        (correct->type == CPP_TOKEN_COMMENT ||
-                               "    %.*s original %.*s testing\n",
+                         correct->type == CPP_TOKEN_ERROR_MESSAGE)){
-                               j,
+                        if (equivalent_comments(correct, testing, data)){
-                               correct->start, correct->size, testing->start, testing->size,
+                            mismatch = 0;
-                               correct->size, data + correct->start,
+                        }
-                               testing->size, data + testing->start);
+                    }
+                    
+                    if (mismatch){
+                        pass = 0;
+                        if (verbose >= 1){
+                            printf("token range mismatch at token %d\n"
+                                   "    %d:%d original %d:%d testing\n"
+                                   "    %.*s original %.*s testing\n",
+                                   j,
+                                   correct->start, correct->size, testing->start, testing->size,
+                                   correct->size, data + correct->start,
+                                   testing->size, data + testing->start);
+                        }
                     }
                 }
                 
@@ -402,12 +429,12 @@ show_time(Times t, int repeats, char *type){
 
 int main(){
     int repeats = 1;
-    int verbose_level = 1;
+    int verbose_level = -1;
-    int chunk_start = 64;
+    int chunk_start = 0;
-    int chunk_end = 64;
+    int chunk_end = 16;
-#define TEST_FILE "lexer_test2.cpp"
+#define TEST_FILE "parser_test_gcc.cpp"
-#define SINGLE_ITEM 1
+#define SINGLE_ITEM 0
-    int token_limit = 1;
+    int token_limit = 0;
     
     int chunks = (chunk_start > 0 && chunk_start <= chunk_end);
     int c = 0;