Syntax highlight is a go

14 files changed, 794 insertions, 12 deletions

diff --git a/src/dged/buffer.c b/src/dged/buffer.c
index 89e7dad..0a7dd16 100644
--- a/src/dged/buffer.c
+++ b/src/dged/buffer.c
@@ -73,21 +73,27 @@ static struct kill_ring {
   }
 
 typedef VEC(struct create_hook) create_hook_vec;
+typedef VEC(struct destroy_hook) destroy_hook_vec;
 typedef VEC(struct insert_hook) insert_hook_vec;
 typedef VEC(struct update_hook) update_hook_vec;
+typedef VEC(struct reload_hook) reload_hook_vec;
 typedef VEC(struct delete_hook) delete_hook_vec;
+typedef VEC(struct render_hook) render_hook_vec;
 
 DECLARE_HOOK(create, create_hook_cb, create_hook_vec);
+DECLARE_HOOK(destroy, destroy_hook_cb, destroy_hook_vec);
 DECLARE_HOOK(insert, insert_hook_cb, insert_hook_vec);
 DECLARE_HOOK(update, update_hook_cb, update_hook_vec);
+DECLARE_HOOK(reload, reload_hook_cb, reload_hook_vec);
+DECLARE_HOOK(render, render_hook_cb, render_hook_vec);
 DECLARE_HOOK(delete, delete_hook_cb, delete_hook_vec);
 
 static create_hook_vec g_create_hooks;
 uint32_t g_create_hook_id;
 
 struct hooks {
-  create_hook_vec create_hooks;
-  uint32_t create_hook_id;
+  destroy_hook_vec destroy_hooks;
+  uint32_t destroy_hook_id;
 
   insert_hook_vec insert_hooks;
   uint32_t insert_hook_id;
@@ -95,6 +101,12 @@ struct hooks {
   update_hook_vec update_hooks;
   uint32_t update_hook_id;
 
+  reload_hook_vec reload_hooks;
+  uint32_t reload_hook_id;
+
+  render_hook_vec render_hooks;
+  uint32_t render_hook_id;
+
   delete_hook_vec delete_hooks;
   uint32_t delete_hook_id;
 };
@@ -108,7 +120,16 @@ void buffer_remove_create_hook(uint32_t hook_id, remove_hook_cb callback) {
   remove_create_hook(&g_create_hooks, hook_id, callback);
 }
 
+uint32_t buffer_add_destroy_hook(struct buffer *buffer,
+                                 destroy_hook_cb callback, void *userdata) {
+  return insert_destroy_hook(&buffer->hooks->destroy_hooks,
+                             &buffer->hooks->destroy_hook_id, callback,
+                             userdata);
+}
+
 void buffer_static_init() {
+  VEC_INIT(&g_create_hooks, 8);
+
   settings_register_setting(
       "editor.tab-width",
       (struct setting_value){.type = Setting_Number, .number_value = 4});
@@ -119,6 +140,7 @@ void buffer_static_init() {
 }
 
 void buffer_static_teardown() {
+  VEC_DESTROY(&g_create_hooks);
   for (uint32_t i = 0; i < KILL_RING_SZ; ++i) {
     if (g_kill_ring.buffer[i].allocated) {
       free(g_kill_ring.buffer[i].text);
@@ -142,7 +164,10 @@ static struct buffer create_internal(const char *name, char *filename) {
   b.hooks = calloc(1, sizeof(struct hooks));
   VEC_INIT(&b.hooks->insert_hooks, 8);
   VEC_INIT(&b.hooks->update_hooks, 8);
+  VEC_INIT(&b.hooks->reload_hooks, 8);
+  VEC_INIT(&b.hooks->render_hooks, 8);
   VEC_INIT(&b.hooks->delete_hooks, 8);
+  VEC_INIT(&b.hooks->destroy_hooks, 8);
 
   undo_init(&b.undo, 100);
 
@@ -404,10 +429,17 @@ void buffer_reload(struct buffer *buffer) {
       sb.st_mtim.tv_nsec != buffer->last_write.tv_nsec) {
     text_clear(buffer->text);
     buffer_read_from_file(buffer);
+    VEC_FOR_EACH(&buffer->hooks->reload_hooks, struct reload_hook * h) {
+      h->callback(buffer, h->userdata);
+    }
   }
 }
 
 void buffer_destroy(struct buffer *buffer) {
+  VEC_FOR_EACH(&buffer->hooks->destroy_hooks, struct destroy_hook * h) {
+    h->callback(buffer, h->userdata);
+  }
+
   text_destroy(buffer->text);
   buffer->text = NULL;
 
@@ -418,7 +450,10 @@ void buffer_destroy(struct buffer *buffer) {
   buffer->filename = NULL;
 
   VEC_DESTROY(&buffer->hooks->update_hooks);
+  VEC_DESTROY(&buffer->hooks->render_hooks);
+  VEC_DESTROY(&buffer->hooks->reload_hooks);
   VEC_DESTROY(&buffer->hooks->insert_hooks);
+  VEC_DESTROY(&buffer->hooks->destroy_hooks);
   VEC_DESTROY(&buffer->hooks->delete_hooks);
   free(buffer->hooks);
 
@@ -460,8 +495,12 @@ struct location buffer_add(struct buffer *buffer, struct location at,
                        (struct undo_boundary){.save_point = false});
   }
 
+  uint32_t begin_idx = text_global_idx(buffer->text, initial.line, initial.col);
+  uint32_t end_idx = text_global_idx(buffer->text, final.line, final.col);
+
   VEC_FOR_EACH(&buffer->hooks->insert_hooks, struct insert_hook * h) {
-    h->callback(buffer, region_new(initial, final), h->userdata);
+    h->callback(buffer, region_new(initial, final), begin_idx, end_idx,
+                h->userdata);
   }
 
   buffer->modified = true;
@@ -592,7 +631,7 @@ struct location buffer_clamp(struct buffer *buffer, int64_t line, int64_t col) {
 
 struct location buffer_end(struct buffer *buffer) {
   uint32_t nlines = buffer_num_lines(buffer);
-  return (struct location){nlines, buffer_num_chars(buffer, nlines)};
+  return (struct location){.line = nlines, .col = 0};
 }
 
 uint32_t buffer_num_lines(struct buffer *buffer) {
@@ -764,12 +803,17 @@ struct location buffer_delete(struct buffer *buffer, struct region region) {
   undo_push_boundary(&buffer->undo,
                      (struct undo_boundary){.save_point = false});
 
+  uint32_t begin_idx =
+      text_global_idx(buffer->text, region.begin.line, region.begin.col);
+  uint32_t end_idx =
+      text_global_idx(buffer->text, region.end.line, region.end.col);
+
   text_delete(buffer->text, region.begin.line, region.begin.col,
               region.end.line, region.end.col);
   buffer->modified = true;
 
   VEC_FOR_EACH(&buffer->hooks->delete_hooks, struct delete_hook * h) {
-    h->callback(buffer, region, h->userdata);
+    h->callback(buffer, region, begin_idx, end_idx, h->userdata);
   }
 
   return region.begin;
@@ -858,6 +902,28 @@ void buffer_remove_update_hook(struct buffer *buffer, uint32_t hook_id,
   remove_update_hook(&buffer->hooks->update_hooks, hook_id, callback);
 }
 
+uint32_t buffer_add_render_hook(struct buffer *buffer, render_hook_cb hook,
+                                void *userdata) {
+  return insert_render_hook(&buffer->hooks->render_hooks,
+                            &buffer->hooks->render_hook_id, hook, userdata);
+}
+
+void buffer_remove_render_hook(struct buffer *buffer, uint32_t hook_id,
+                               remove_hook_cb callback) {
+  remove_render_hook(&buffer->hooks->render_hooks, hook_id, callback);
+}
+
+uint32_t buffer_add_reload_hook(struct buffer *buffer, reload_hook_cb hook,
+                                void *userdata) {
+  return insert_reload_hook(&buffer->hooks->reload_hooks,
+                            &buffer->hooks->reload_hook_id, hook, userdata);
+}
+
+void buffer_remove_reload_hook(struct buffer *buffer, uint32_t hook_id,
+                               remove_hook_cb callback) {
+  remove_reload_hook(&buffer->hooks->reload_hooks, hook_id, callback);
+}
+
 struct cmdbuf {
   struct command_list *cmds;
   struct location origin;
@@ -999,6 +1065,11 @@ void buffer_render(struct buffer *buffer, struct buffer_render_params *params) {
     return;
   }
 
+  VEC_FOR_EACH(&buffer->hooks->render_hooks, struct render_hook * h) {
+    h->callback(buffer, h->userdata, params->origin, params->width,
+                params->height);
+  }
+
   struct setting *show_ws = settings_get("editor.show-whitespace");
 
   struct cmdbuf cmdbuf = (struct cmdbuf){
diff --git a/src/dged/buffer.h b/src/dged/buffer.h
index 1e00b9d..2e71fb3 100644
--- a/src/dged/buffer.h
+++ b/src/dged/buffer.h
@@ -418,8 +418,26 @@ uint32_t buffer_add_update_hook(struct buffer *buffer, update_hook_cb hook,
 void buffer_remove_update_hook(struct buffer *buffer, uint32_t hook_id,
                                remove_hook_cb callback);
 
+/** Buffer render hook callback function */
+typedef void (*render_hook_cb)(struct buffer *buffer, void *userdata,
+                               struct location origin, uint32_t width,
+                               uint32_t height);
+
+uint32_t buffer_add_render_hook(struct buffer *buffer, render_hook_cb hook,
+                                void *userdata);
+void buffer_remove_render_hook(struct buffer *buffer, uint32_t hook_id,
+                               remove_hook_cb callback);
+
+/** Buffer reload hook callback function */
+typedef void (*reload_hook_cb)(struct buffer *buffer, void *userdata);
+uint32_t buffer_add_reload_hook(struct buffer *buffer, reload_hook_cb hook,
+                                void *userdata);
+void buffer_remove_reload_hook(struct buffer *buffer, uint32_t hook_id,
+                               remove_hook_cb callback);
+
 /** Buffer insert hook callback function */
 typedef void (*insert_hook_cb)(struct buffer *buffer, struct region inserted,
+                               uint32_t begin_idx, uint32_t end_idx,
                                void *userdata);
 
 uint32_t buffer_add_insert_hook(struct buffer *buffer, insert_hook_cb callback,
@@ -429,6 +447,7 @@ void buffer_remove_insert_hook(struct buffer *buffer, uint32_t hook_id,
 
 /** Buffer delete hook callback function */
 typedef void (*delete_hook_cb)(struct buffer *buffer, struct region removed,
+                               uint32_t begin_idx, uint32_t end_idx,
                                void *userdata);
 
 uint32_t buffer_add_delete_hook(struct buffer *buffer, delete_hook_cb callback,
@@ -436,6 +455,11 @@ uint32_t buffer_add_delete_hook(struct buffer *buffer, delete_hook_cb callback,
 void buffer_remove_delete_hook(struct buffer *buffer, uint32_t hook_id,
                                remove_hook_cb callback);
 
+/** Buffer destroy hook callback function */
+typedef void (*destroy_hook_cb)(struct buffer *buffer, void *userdata);
+uint32_t buffer_add_destroy_hook(struct buffer *buffer,
+                                 destroy_hook_cb callback, void *userdata);
+
 /** Buffer create hook callback function */
 typedef void (*create_hook_cb)(struct buffer *buffer, void *userdata);
 
diff --git a/src/dged/display.h b/src/dged/display.h
index 2fc807b..aae2614 100644
--- a/src/dged/display.h
+++ b/src/dged/display.h
@@ -119,6 +119,25 @@ struct command_list *command_list_create(uint32_t capacity,
  */
 void command_list_set_show_whitespace(struct command_list *list, bool show);
 
+enum colors {
+  Color_Black = 0,
+  Color_Red,
+  Color_Green,
+  Color_Yellow,
+  Color_Blue,
+  Color_Magenta,
+  Color_Cyan,
+  Color_White,
+  Color_BrightBlack = 8,
+  Color_BrightRed,
+  Color_BrightGreen,
+  Color_BrightYellow,
+  Color_BrightBlue,
+  Color_BrightMagenta,
+  Color_BrightCyan,
+  Color_BrightWhite
+};
+
 /**
  * Set background color
  *
diff --git a/src/dged/lang.c b/src/dged/lang.c
index 1bfe822..fb740e8 100644
--- a/src/dged/lang.c
+++ b/src/dged/lang.c
@@ -49,7 +49,8 @@ void languages_init(bool register_default) {
     define_lang("C++", "cxx", ".*\\.(cpp|cxx|cc|c++|hh|h)", 2, "clangd");
     define_lang("Rust", "rs", ".*\\.rs", 4, "rust-analyzer");
     define_lang("Nix", "nix", ".*\\.nix", 4, "rnix-lsp");
-    define_lang("Makefile", "make", ".*(Makefile|\\.mk)", 4, NULL);
+    define_lang("Make", "make", ".*(Makefile|\\.mk)", 4, NULL);
+    define_lang("Python", "python", ".*\\.py", 4, NULL);
   }
 }
 
diff --git a/src/dged/minibuffer.c b/src/dged/minibuffer.c
index eacf8da..634a864 100644
--- a/src/dged/minibuffer.c
+++ b/src/dged/minibuffer.c
@@ -2,6 +2,7 @@
 #include "binding.h"
 #include "buffer.h"
 #include "buffer_view.h"
+#include "buffers.h"
 #include "command.h"
 #include "display.h"
 
@@ -20,6 +21,8 @@ static struct minibuffer {
   bool clear;
   struct window *prev_window;
 
+  struct buffer *message_buffer;
+
 } g_minibuffer = {0};
 
 uint32_t minibuffer_draw_prompt(struct command_list *commands) {
@@ -85,7 +88,7 @@ void update(struct buffer *buffer, void *userdata) {
   }
 }
 
-void minibuffer_init(struct buffer *buffer) {
+void minibuffer_init(struct buffer *buffer, struct buffers *buffers) {
   if (g_minibuffer.buffer != NULL) {
     return;
   }
@@ -96,6 +99,9 @@ void minibuffer_init(struct buffer *buffer) {
   g_minibuffer.clear = false;
   g_minibuffer.prompt_active = false;
   buffer_add_update_hook(g_minibuffer.buffer, update, &g_minibuffer);
+
+  g_minibuffer.message_buffer =
+      buffers_add(buffers, buffer_create("*messages*"));
 }
 
 void echo(uint32_t timeout, const char *fmt, va_list args) {
@@ -107,13 +113,37 @@ void echo(uint32_t timeout, const char *fmt, va_list args) {
   g_minibuffer.expires.tv_sec += timeout;
   g_minibuffer.clear = false;
 
-  char buff[2048];
+  static char buff[2048];
   size_t nbytes = vsnprintf(buff, 2048, fmt, args);
 
   // vsnprintf returns how many characters it would have wanted to write in case
   // of overflow
   buffer_set_text(g_minibuffer.buffer, (uint8_t *)buff,
                   nbytes > 2048 ? 2048 : nbytes);
+
+  // we can get messages before this is set up
+  if (g_minibuffer.message_buffer != NULL) {
+    buffer_add(g_minibuffer.message_buffer,
+               buffer_end(g_minibuffer.message_buffer), (uint8_t *)buff,
+               nbytes > 2048 ? 2048 : nbytes);
+  }
+}
+
+void message(const char *fmt, ...) {
+  // we can get messages before this is set up
+  if (g_minibuffer.message_buffer == NULL) {
+    return;
+  }
+
+  va_list args;
+  va_start(args, fmt);
+  static char buff[2048];
+  size_t nbytes = vsnprintf(buff, 2048, fmt, args);
+  va_end(args);
+
+  buffer_add(g_minibuffer.message_buffer,
+             buffer_end(g_minibuffer.message_buffer), (uint8_t *)buff,
+             nbytes > 2048 ? 2048 : nbytes);
 }
 
 void minibuffer_destroy() {
diff --git a/src/dged/minibuffer.h b/src/dged/minibuffer.h
index b7c5171..3e94f8c 100644
--- a/src/dged/minibuffer.h
+++ b/src/dged/minibuffer.h
@@ -4,6 +4,7 @@
 #include <time.h>
 
 struct buffer;
+struct buffers;
 struct command_ctx;
 struct command_list;
 struct keymap;
@@ -15,7 +16,7 @@ struct keymap;
  * nothing if called more than once.
  * @param buffer underlying buffer to use for text IO in the minibuffer.
  */
-void minibuffer_init(struct buffer *buffer);
+void minibuffer_init(struct buffer *buffer, struct buffers *buffers);
 
 /**
  * Destroy the minibuffer
@@ -28,6 +29,8 @@ struct text_chunk minibuffer_content();
 
 struct buffer *minibuffer_buffer();
 
+void message(const char *fmt, ...);
+
 /**
  * Echo a message to the minibuffer.
  *
diff --git a/src/dged/path.h b/src/dged/path.h
index da62457..8ea9977 100644
--- a/src/dged/path.h
+++ b/src/dged/path.h
@@ -38,3 +38,28 @@ static char *to_abspath(const char *path) {
     return exp;
   }
 }
+
+static const char *join_path_with_delim(const char *p1, const char *p2,
+                                        const char delim) {
+  size_t len1 = strlen(p1);
+  size_t len2 = strlen(p2);
+
+  char *path = malloc(len1 + len2 + 2);
+  uint32_t idx = 0;
+  memcpy(&path[idx], p1, len1);
+  idx += len1;
+  path[idx++] = '/';
+  memcpy(&path[idx], p2, len2);
+  idx += len2;
+  path[idx++] = '\0';
+
+  return path;
+}
+
+static const char *join_path(const char *p1, const char *p2) {
+#ifdef __unix__
+  return join_path_with_delim(p1, p2, '/');
+#elif defined(_WIN32) || defined(WIN32)
+  return join_path_with_delim(p1, p2, '\\');
+#endif
+}
diff --git a/src/dged/syntax.c b/src/dged/syntax.c
new file mode 100644
index 0000000..403cabe
--- /dev/null
+++ b/src/dged/syntax.c
@@ -0,0 +1,563 @@
+#include "syntax.h"
+
+#include <ctype.h>
+#include <dlfcn.h>
+#include <errno.h>
+#include <fcntl.h>
+#include <regex.h>
+#include <string.h>
+#include <sys/mman.h>
+#include <sys/stat.h>
+#include <unistd.h>
+
+#include <tree_sitter/api.h>
+
+#include "buffer.h"
+#include "display.h"
+#include "hash.h"
+#include "hashmap.h"
+#include "minibuffer.h"
+#include "path.h"
+#include "text.h"
+#include "vec.h"
+
+static char *treesitter_path = NULL;
+static bool treesitter_path_allocated = false;
+static const char *parser_filename = "parser";
+static const char *highlight_path = "queries/highlights.scm";
+
+HASHMAP_ENTRY_TYPE(re_cache_entry, regex_t);
+
+struct highlight {
+  TSParser *parser;
+  TSTree *tree;
+  TSQuery *query;
+  HASHMAP(struct re_cache_entry) re_cache;
+  void *dlhandle;
+};
+
+static void delete_parser(struct buffer *buffer, void *userdata) {
+  struct highlight *highlight = (struct highlight *)userdata;
+
+  if (highlight->query != NULL) {
+    ts_query_delete(highlight->query);
+  }
+
+  HASHMAP_FOR_EACH(&highlight->re_cache, struct re_cache_entry * entry) {
+    regfree(&entry->value);
+  }
+
+  HASHMAP_DESTROY(&highlight->re_cache);
+
+  ts_tree_delete(highlight->tree);
+  ts_parser_delete(highlight->parser);
+
+  dlclose(highlight->dlhandle);
+
+  free(highlight);
+}
+
+static const char *read_text(void *payload, uint32_t byte_offset,
+                             TSPoint position, uint32_t *bytes_read) {
+
+  struct text *text = (struct text *)payload;
+
+  if (position.row < text_num_lines(text)) {
+    struct text_chunk chunk = text_get_line(text, position.row);
+
+    // empty lines
+    if (chunk.nbytes == 0 || position.column >= chunk.nchars) {
+      *bytes_read = 1;
+      return "\n";
+    }
+
+    uint32_t bytei = text_col_to_byteindex(text, chunk.line, position.column);
+    *bytes_read = chunk.nbytes - bytei;
+    return (const char *)chunk.text + bytei;
+  }
+
+  // eof
+  *bytes_read = 0;
+  return NULL;
+}
+
+static const char *lang_folder(struct buffer *buffer) {
+  const char *langname = buffer->lang.name;
+
+  size_t tspath_len = strlen(treesitter_path);
+  size_t lang_len = strlen(langname);
+
+  char *fld = malloc(tspath_len + lang_len + 2);
+  uint32_t idx = 0;
+  memcpy(&fld[idx], treesitter_path, tspath_len);
+  idx += tspath_len;
+  fld[idx++] = '/';
+  for (uint32_t i = 0; i < lang_len; ++i) {
+    fld[idx + i] = tolower(langname[i]);
+  }
+  idx += lang_len;
+  fld[idx++] = '\0';
+
+  return fld;
+}
+
+static TSQuery *setup_queries(const char *lang_root, TSTree *tree) {
+  const char *filename = join_path(lang_root, highlight_path);
+
+  // read queries from file
+  int fd = open(filename, O_RDONLY);
+  free((void *)filename);
+  if (fd < 0) {
+    return NULL;
+  }
+
+  size_t len = lseek(fd, 0, SEEK_END);
+  void *data = mmap(0, len, PROT_READ, MAP_PRIVATE, fd, 0);
+  close(fd);
+
+  if (data == NULL) {
+    return NULL;
+  }
+
+  // run queries
+  TSQueryError error = TSQueryErrorNone;
+  uint32_t error_offset = 0;
+  TSQuery *q = ts_query_new(ts_tree_language(tree), (char *)data, len,
+                            &error_offset, &error);
+
+  munmap(data, len);
+
+  if (error != TSQueryErrorNone) {
+    const char *msg = "unknown error";
+    switch (error) {
+    case TSQueryErrorSyntax:
+      msg = "syntax error";
+      break;
+    case TSQueryErrorNodeType:
+      msg = "node type";
+      break;
+    case TSQueryErrorField:
+      msg = "error field";
+      break;
+    case TSQueryErrorCapture:
+      msg = "capture";
+      break;
+    }
+    message("ts query error at byte offset %d: %s", error_offset, msg);
+    return NULL;
+  }
+
+  return q;
+}
+
+#define s8(s) ((struct s8){s, strlen(s)})
+
+struct s8 {
+  char *s;
+  uint32_t l;
+};
+
+static bool s8eq(struct s8 s1, struct s8 s2) {
+  return s1.l == s2.l && memcmp(s1.s, s2.s, s1.l) == 0;
+}
+
+char *s8tocstr(struct s8 s) {
+  char *cstr = (char *)malloc(s.l + 1);
+  memcpy(cstr, s.s, s.l);
+  cstr[s.l] = '\0';
+  return cstr;
+}
+
+static bool eval_match(struct s8 capname, uint32_t argc, struct s8 argv[],
+                       struct s8 value, void *data) {
+  regex_t *regex = (regex_t *)data;
+  if (regex == NULL) {
+    return false;
+  }
+
+  char *text = s8tocstr(value);
+  bool match = regexec(regex, text, 0, NULL, 0) == 0;
+
+  free(text);
+  return match;
+}
+
+static void cleanup_match(void *data) {
+  regex_t *regex = (regex_t *)data;
+  if (regex != NULL) {
+    regfree(regex);
+    free(regex);
+  }
+}
+
+struct predicate {
+  bool (*fn)(struct s8, uint32_t, struct s8[], struct s8, void *);
+  void (*cleanup)(void *);
+  uint32_t argc;
+  struct s8 argv[32];
+  void *data;
+};
+
+typedef VEC(struct predicate) predicate_vec;
+
+static regex_t *compile_re_cached(struct highlight *h, struct s8 expr) {
+  char *val = s8tocstr(expr);
+  HASHMAP_GET(&h->re_cache, struct re_cache_entry, val, regex_t * re);
+  if (re == NULL) {
+    regex_t new_re;
+    if (regcomp(&new_re, val, 0) == 0) {
+      HASHMAP_APPEND(&h->re_cache, struct re_cache_entry, val,
+                     struct re_cache_entry * new);
+      if (new != NULL) {
+        new->value = new_re;
+        re = &new->value;
+      }
+    }
+  }
+
+  free(val);
+  return re;
+}
+
+static predicate_vec create_predicates(struct highlight *h,
+                                       uint32_t pattern_index) {
+  predicate_vec predicates;
+
+  uint32_t npreds = 0;
+  const TSQueryPredicateStep *predicate_steps =
+      ts_query_predicates_for_pattern(h->query, pattern_index, &npreds);
+
+  VEC_INIT(&predicates, 8);
+
+  bool result = true;
+  struct s8 capname;
+  struct s8 args[32] = {0};
+  uint32_t argc = 0;
+  for (uint32_t predi = 0; predi < npreds; ++predi) {
+    const TSQueryPredicateStep *step = &predicate_steps[predi];
+    switch (step->type) {
+    case TSQueryPredicateStepTypeCapture:
+      capname.s = (char *)ts_query_capture_name_for_id(h->query, step->value_id,
+                                                       &capname.l);
+      break;
+
+    case TSQueryPredicateStepTypeString:
+      args[argc].s = (char *)ts_query_string_value_for_id(
+          h->query, step->value_id, &args[argc].l);
+      ++argc;
+      break;
+
+    case TSQueryPredicateStepTypeDone:
+      if (s8eq(args[0], s8("match?"))) {
+        VEC_APPEND(&predicates, struct predicate * pred);
+        pred->fn = eval_match;
+        pred->cleanup = NULL;
+        pred->argc = 1;
+
+        // cache the regex
+        pred->data = compile_re_cached(h, args[1]);
+
+        memset(pred->argv, 0, sizeof(struct s8) * 32);
+        memcpy(pred->argv, args, sizeof(struct s8));
+      }
+
+      argc = 0;
+      break;
+    }
+  }
+
+  return predicates;
+}
+
+static void update_parser(struct buffer *buffer, void *userdata,
+                          struct location origin, uint32_t width,
+                          uint32_t height) {
+
+  struct highlight *h = (struct highlight *)userdata;
+
+  if (h->query == NULL) {
+    return;
+  }
+
+  // take results and set text properties
+  // TODO: can reuse the cursor
+  TSQueryCursor *cursor = ts_query_cursor_new();
+  uint32_t end_line = origin.line + height >= buffer_num_lines(buffer)
+                          ? buffer_num_lines(buffer) - 1
+                          : origin.line + height;
+  ts_query_cursor_set_point_range(
+      cursor, (TSPoint){.row = origin.line, .column = origin.col},
+      (TSPoint){.row = end_line, .column = buffer_num_chars(buffer, end_line)});
+  ts_query_cursor_exec(cursor, h->query, ts_tree_root_node(h->tree));
+
+  TSQueryMatch match;
+  while (ts_query_cursor_next_match(cursor, &match)) {
+    predicate_vec predicates = create_predicates(h, match.pattern_index);
+
+    for (uint32_t capi = 0; capi < match.capture_count; ++capi) {
+      const TSQueryCapture *cap = &match.captures[capi];
+      TSPoint start = ts_node_start_point(cap->node);
+      TSPoint end = ts_node_end_point(cap->node);
+
+      struct s8 cname;
+      cname.s =
+          (char *)ts_query_capture_name_for_id(h->query, cap->index, &cname.l);
+
+      bool predicates_match = true;
+      VEC_FOR_EACH(&predicates, struct predicate * pred) {
+        struct text_chunk txt = text_get_region(
+            buffer->text, start.row, start.column, end.row, end.column);
+        predicates_match &=
+            pred->fn(cname, pred->argc, pred->argv,
+                     (struct s8){.s = txt.text, .l = txt.nbytes}, pred->data);
+
+        if (txt.allocated) {
+          free(txt.text);
+        }
+      }
+
+      if (!predicates_match) {
+        continue;
+      }
+
+      bool highlight = false;
+      uint32_t color = 0;
+      if (s8eq(cname, s8("keyword"))) {
+        highlight = true;
+        color = Color_Blue;
+      } else if (s8eq(cname, s8("operator"))) {
+        highlight = true;
+        color = Color_Magenta;
+      } else if (s8eq(cname, s8("delimiter"))) {
+        highlight = false;
+      } else if (s8eq(cname, s8("string")) ||
+                 s8eq(cname, s8("string.special")) ||
+                 s8eq(cname, s8("string.special.path")) ||
+                 s8eq(cname, s8("string.special.uri"))) {
+        highlight = true;
+        color = Color_Green;
+      } else if (s8eq(cname, s8("constant"))) {
+        highlight = true;
+        color = Color_Yellow;
+      } else if (s8eq(cname, s8("attribute"))) {
+        highlight = true;
+        color = Color_Yellow;
+      } else if (s8eq(cname, s8("number"))) {
+        highlight = false;
+      } else if (s8eq(cname, s8("function")) ||
+                 s8eq(cname, s8("function.macro")) ||
+                 s8eq(cname, s8("function.method")) ||
+                 s8eq(cname, s8("function.builtin")) ||
+                 s8eq(cname, s8("function.special"))) {
+        highlight = true;
+        color = Color_Yellow;
+      } else if (s8eq(cname, s8("property"))) {
+        highlight = false;
+      } else if (s8eq(cname, s8("label"))) {
+        highlight = false;
+      } else if (s8eq(cname, s8("type")) || s8eq(cname, s8("type.builtin"))) {
+        highlight = true;
+        color = Color_Cyan;
+      } else if (s8eq(cname, s8("variable")) ||
+                 s8eq(cname, s8("variable.builtin")) ||
+                 s8eq(cname, s8("variable.parameter"))) {
+        highlight = false;
+      } else if (s8eq(cname, s8("comment"))) {
+        highlight = true;
+        color = Color_BrightBlack;
+      }
+
+      if (!highlight) {
+        continue;
+      }
+
+      buffer_add_text_property(
+          buffer, (struct location){.line = start.row, .col = start.column},
+          (struct location){.line = end.row, .col = end.column - 1},
+          (struct text_property){
+              .type = TextProperty_Colors,
+              .colors =
+                  (struct text_property_colors){
+                      .set_fg = true,
+                      .fg = color,
+                  },
+          });
+    }
+
+    VEC_FOR_EACH(&predicates, struct predicate * pred) {
+      if (pred->cleanup != NULL) {
+        pred->cleanup(pred->data);
+      }
+    }
+    VEC_DESTROY(&predicates);
+  }
+
+  ts_query_cursor_delete(cursor);
+}
+
+static void text_removed(struct buffer *buffer, struct region removed,
+                         uint32_t begin_idx, uint32_t end_idx, void *userdata) {
+  struct highlight *h = (struct highlight *)userdata;
+
+  TSInputEdit edit = {
+      .start_point =
+          (TSPoint){.row = removed.begin.line, .column = removed.begin.col},
+      .old_end_point =
+          (TSPoint){.row = removed.end.line, .column = removed.end.col},
+      .new_end_point =
+          (TSPoint){.row = removed.begin.line, .column = removed.begin.col},
+      .start_byte = begin_idx,
+      .old_end_byte = end_idx,
+      .new_end_byte = begin_idx,
+  };
+
+  ts_tree_edit(h->tree, &edit);
+  TSInput i = (TSInput){
+      .payload = buffer->text,
+      .read = read_text,
+      .encoding = TSInputEncodingUTF8,
+  };
+
+  TSTree *new_tree = ts_parser_parse(h->parser, h->tree, i);
+  if (new_tree != NULL) {
+    ts_tree_delete(h->tree);
+    h->tree = new_tree;
+  }
+}
+
+static void buffer_reloaded(struct buffer *buffer, void *userdata) {
+  struct highlight *h = (struct highlight *)userdata;
+
+  TSInput i = (TSInput){
+      .payload = buffer->text,
+      .read = read_text,
+      .encoding = TSInputEncodingUTF8,
+  };
+
+  TSTree *new_tree = ts_parser_parse(h->parser, NULL, i);
+  if (new_tree != NULL) {
+    ts_tree_delete(h->tree);
+    h->tree = new_tree;
+  }
+}
+
+static void text_inserted(struct buffer *buffer, struct region inserted,
+                          uint32_t begin_idx, uint32_t end_idx,
+                          void *userdata) {
+  struct highlight *h = (struct highlight *)userdata;
+
+  TSInputEdit edit = {
+      .start_point =
+          (TSPoint){.row = inserted.begin.line, .column = inserted.begin.col},
+      .old_end_point =
+          (TSPoint){.row = inserted.begin.line, .column = inserted.begin.col},
+      .new_end_point =
+          (TSPoint){.row = inserted.end.line, .column = inserted.end.col},
+      .start_byte = begin_idx,
+      .old_end_byte = begin_idx,
+      .new_end_byte = end_idx,
+  };
+
+  ts_tree_edit(h->tree, &edit);
+  TSInput i = (TSInput){
+      .payload = buffer->text,
+      .read = read_text,
+      .encoding = TSInputEncodingUTF8,
+  };
+
+  TSTree *new_tree = ts_parser_parse(h->parser, h->tree, i);
+  if (new_tree != NULL) {
+    ts_tree_delete(h->tree);
+    h->tree = new_tree;
+  }
+}
+
+static void create_parser(struct buffer *buffer, void *userdata) {
+  const char *lang_root = lang_folder(buffer);
+  const char *filename = join_path(lang_root, parser_filename);
+
+  void *h = dlopen(filename, RTLD_LAZY);
+  free((void *)filename);
+  if (h == NULL) {
+    free((void *)lang_root);
+    return;
+  }
+
+  const char *langname = buffer->lang.name;
+  size_t lang_len = strlen(langname);
+
+  const char *prefix = "tree_sitter_";
+  size_t prefix_len = strlen(prefix);
+  char *function = malloc(prefix_len + lang_len + 1);
+  memcpy(function, prefix, prefix_len);
+  for (uint32_t i = 0; i < lang_len; ++i) {
+    function[prefix_len + i] = tolower(langname[i]);
+  }
+  function[prefix_len + lang_len] = '\0';
+  TSLanguage *(*langsym)() = dlsym(h, function);
+
+  free(function);
+  if (langsym == NULL) {
+    free((void *)lang_root);
+    dlclose(h);
+    return;
+  }
+
+  struct highlight *hl =
+      (struct highlight *)calloc(1, sizeof(struct highlight));
+  hl->parser = ts_parser_new();
+  ts_parser_set_language(hl->parser, langsym());
+
+  TSInput i = (TSInput){
+      .payload = buffer->text,
+      .read = read_text,
+      .encoding = TSInputEncodingUTF8,
+  };
+  hl->tree = ts_parser_parse(hl->parser, NULL, i);
+  hl->query = setup_queries(lang_root, hl->tree);
+  hl->dlhandle = h;
+  HASHMAP_INIT(&hl->re_cache, 64, hash_name);
+
+  free((void *)lang_root);
+
+  minibuffer_echo_timeout(4, "syntax set up for %s", langname);
+
+  buffer_add_reload_hook(buffer, buffer_reloaded, hl);
+  buffer_add_delete_hook(buffer, text_removed, hl);
+  buffer_add_insert_hook(buffer, text_inserted, hl);
+  buffer_add_render_hook(buffer, update_parser, hl);
+  buffer_add_destroy_hook(buffer, delete_parser, hl);
+}
+
+#define xstr(s) str(s)
+#define str(s) #s
+
+void syntax_init() {
+  treesitter_path = getenv("TREESITTER_GRAMMARS");
+  if (treesitter_path == NULL) {
+    treesitter_path = (char *)join_path(xstr(DATADIR), "grammars");
+    treesitter_path_allocated = true;
+  }
+
+  struct stat buffer;
+  if (stat(treesitter_path, &buffer) != 0) {
+    minibuffer_echo_timeout(4,
+                            "failed to initialize syntax, TREESITTER_GRAMMARS "
+                            "not set and grammars dir does not exist at %s.",
+                            treesitter_path);
+
+    if (treesitter_path_allocated) {
+      free(treesitter_path);
+    }
+
+    return;
+  }
+
+  buffer_add_create_hook(create_parser, NULL);
+}
+
+void syntax_teardown() {
+  if (treesitter_path_allocated) {
+    free(treesitter_path);
+  }
+}
diff --git a/src/dged/syntax.h b/src/dged/syntax.h
new file mode 100644
index 0000000..6a2d4a3
--- /dev/null
+++ b/src/dged/syntax.h
@@ -0,0 +1,7 @@
+#ifndef _SYNTAX_H
+#define _SYNTAX_H
+
+void syntax_init();
+void syntax_teardown();
+
+#endif
diff --git a/src/dged/text.c b/src/dged/text.c
index 4d9a073..bc2b1fc 100644
--- a/src/dged/text.c
+++ b/src/dged/text.c
@@ -101,6 +101,26 @@ uint32_t text_byteindex_to_col(struct text *text, uint32_t line,
   return byteidx_to_charidx(&text->lines[line], byteindex);
 }
 
+uint32_t text_global_idx(struct text *text, uint32_t line, uint32_t col) {
+  uint32_t byteoff = 0;
+  uint32_t nlines = text_num_lines(text);
+  for (uint32_t l = 0; l < line && l < nlines; ++l) {
+    byteoff += text_line_size(text, l) + 1;
+  }
+
+  uint32_t l = line < nlines ? line : nlines - 1;
+  uint32_t nchars = text_line_length(text, l);
+  uint32_t c = col < nchars ? col : nchars;
+  byteoff += text_col_to_byteindex(text, l, c);
+
+  if (col > nchars) {
+    // account for newline
+    ++byteoff;
+  }
+
+  return byteoff;
+}
+
 void append_empty_lines(struct text *text, uint32_t numlines) {
 
   if (text->nlines + numlines >= text->capacity) {
diff --git a/src/dged/text.h b/src/dged/text.h
index e3bb3e4..922014e 100644
--- a/src/dged/text.h
+++ b/src/dged/text.h
@@ -34,6 +34,7 @@ uint32_t text_line_size(struct text *text, uint32_t lineidx);
 uint32_t text_col_to_byteindex(struct text *text, uint32_t line, uint32_t col);
 uint32_t text_byteindex_to_col(struct text *text, uint32_t line,
                                uint32_t byteindex);
+uint32_t text_global_idx(struct text *text, uint32_t line, uint32_t col);
 
 struct text_chunk {
   uint8_t *text;
diff --git a/src/main/cmds.c b/src/main/cmds.c
index c137ed1..5a1d7b6 100644
--- a/src/main/cmds.c
+++ b/src/main/cmds.c
@@ -68,6 +68,7 @@ int32_t run_interactive(struct command_ctx ctx, int argc, const char *argv[]) {
 }
 
 int32_t do_switch_buffer(struct command_ctx ctx, int argc, const char *argv[]) {
+  disable_completion(minibuffer_buffer());
   const char *bufname = argv[0];
   if (argc == 0) {
     // switch back to prev buffer
diff --git a/src/main/completion.c b/src/main/completion.c
index 7382805..414df9c 100644
--- a/src/main/completion.c
+++ b/src/main/completion.c
@@ -203,6 +203,7 @@ static void update_completions(struct buffer *buffer,
 }
 
 static void on_buffer_delete(struct buffer *buffer, struct region deleted,
+                             uint32_t start_idx, uint32_t end_idx,
                              void *userdata) {
   struct active_completion_ctx *ctx = (struct active_completion_ctx *)userdata;
 
@@ -212,6 +213,7 @@ static void on_buffer_delete(struct buffer *buffer, struct region deleted,
 }
 
 static void on_buffer_insert(struct buffer *buffer, struct region inserted,
+                             uint32_t start_idx, uint32_t end_idx,
                              void *userdata) {
   struct active_completion_ctx *ctx = (struct active_completion_ctx *)userdata;
 
diff --git a/src/main/main.c b/src/main/main.c
index 02afd2b..7b9a812 100644
--- a/src/main/main.c
+++ b/src/main/main.c
@@ -20,6 +20,10 @@
 #include "dged/reactor.h"
 #include "dged/settings.h"
 
+#ifdef SYNTAX_ENABLE
+#include "dged/syntax.h"
+#endif
+
 #include "bindings.h"
 #include "cmds.h"
 #include "completion.h"
@@ -213,11 +217,20 @@ int main(int argc, char *argv[]) {
 
   struct buffers buflist = {0};
   buffers_init(&buflist, 32);
+  struct buffer minibuffer = buffer_create("minibuffer");
+  minibuffer.lazy_row_add = false;
+  minibuffer_init(&minibuffer, &buflist);
+
   buffers_add_add_hook(&buflist, watch_file, (void *)reactor);
+#ifdef SYNTAX_ENABLE
+  syntax_init();
+#endif
+
   struct buffer initial_buffer = buffer_create("welcome");
   if (filename != NULL) {
     buffer_destroy(&initial_buffer);
     initial_buffer = buffer_from_file(filename);
+    free((void *)filename);
   } else {
     const char *welcome_txt = "Welcome to the editor for datagubbar 👴\n";
     buffer_set_text(&initial_buffer, (uint8_t *)welcome_txt,
@@ -225,9 +238,6 @@ int main(int argc, char *argv[]) {
   }
 
   struct buffer *ib = buffers_add(&buflist, initial_buffer);
-  struct buffer minibuffer = buffer_create("minibuffer");
-  minibuffer.lazy_row_add = false;
-  minibuffer_init(&minibuffer);
 
   windows_init(display_height(display), display_width(display), ib,
                &minibuffer);
@@ -383,6 +393,11 @@ int main(int argc, char *argv[]) {
   minibuffer_destroy();
   buffer_destroy(&minibuffer);
   buffers_destroy(&buflist);
+
+#ifdef SYNTAX_ENABLE
+  syntax_teardown();
+#endif
+
   display_clear(display);
   display_destroy(display);
   destroy_bindings();