Rework delete logic a bit

It now handles a range of characters and correctly merges lines. It is
not the most slick implementation but it works as a start.

5 files changed, 129 insertions, 56 deletions

diff --git a/src/buffer.c b/src/buffer.c
index e08bca5..fb071da 100644
--- a/src/buffer.c
+++ b/src/buffer.c
@@ -98,14 +98,13 @@ void moveh(struct buffer *buffer, int coldelta) {
   }
 }
 
+void buffer_forward_delete_char(struct buffer *buffer) {
+  text_delete(buffer->text, buffer->dot_line, buffer->dot_col, 1);
+}
+
 void buffer_backward_delete_char(struct buffer *buffer) {
-  // TODO: merge lines
-  if (text_line_length(buffer->text, buffer->dot_line) == 0) {
-    text_delete_line(buffer->text, buffer->dot_line);
-  } else if (buffer->dot_col > 0) {
-    text_delete(buffer->text, buffer->dot_line, buffer->dot_col - 1, 1);
-  }
   moveh(buffer, -1);
+  buffer_forward_delete_char(buffer);
 }
 
 void buffer_backward_char(struct buffer *buffer) { moveh(buffer, -1); }
diff --git a/src/buffer.h b/src/buffer.h
index 1b73505..66096b9 100644
--- a/src/buffer.h
+++ b/src/buffer.h
@@ -40,6 +40,7 @@ void buffer_add_keymap(struct buffer *buffer, struct keymap *keymap);
 
 int buffer_add_text(struct buffer *buffer, uint8_t *text, uint32_t nbytes);
 
+void buffer_forward_delete_char(struct buffer *buffer);
 void buffer_backward_delete_char(struct buffer *buffer);
 void buffer_backward_char(struct buffer *buffer);
 void buffer_forward_char(struct buffer *buffer);
diff --git a/src/text.c b/src/text.c
index 4162e94..64cbd7b 100644
--- a/src/text.c
+++ b/src/text.c
@@ -48,6 +48,28 @@ void text_destroy(struct text *text) {
   free(text->lines);
 }
 
+// given `char_idx` as a character index, return the byte index
+uint32_t charidx_to_byteidx(struct line *line, uint32_t char_idx) {
+  if (char_idx > line->nchars) {
+    return line->nbytes;
+  }
+  return utf8_nbytes(line->data, char_idx);
+}
+
+// TODO: grapheme clusters
+// given `byte_idx` as a byte index, return the character index
+uint32_t byteidx_to_charidx(struct line *line, uint32_t byte_idx) {
+  if (byte_idx > line->nbytes) {
+    return line->nchars;
+  }
+
+  return utf8_nchars(line->data, byte_idx);
+}
+
+uint32_t char_byte_size(struct line *line, uint32_t byte_idx) {
+  return utf8_nbytes(line->data + byte_idx, 1);
+}
+
 void append_to_line(struct line *line, uint32_t col, uint8_t *text,
                     uint32_t len, uint32_t nchars) {
 
@@ -58,13 +80,19 @@ void append_to_line(struct line *line, uint32_t col, uint8_t *text,
   line->nbytes += len;
   line->nchars += nchars;
   line->flags = LineChanged;
-  line->data = realloc(line->data, line->nbytes + len);
+  line->data = realloc(line->data, line->nbytes);
+
+  uint32_t bytei = charidx_to_byteidx(line, col);
 
   // move chars out of the way
-  memmove(line->data + col + 1, line->data + col, line->nbytes - (col + 1));
+  if (col + nchars < line->nchars) {
+    uint32_t nextcbytei = charidx_to_byteidx(line, col + nchars);
+    memmove(line->data + nextcbytei, line->data + bytei,
+            line->nbytes - nextcbytei);
+  }
 
   // insert new chars
-  memcpy(line->data + col, text, len);
+  memcpy(line->data + bytei, text, len);
 }
 
 uint32_t text_line_length(struct text *text, uint32_t lineidx) {
@@ -77,28 +105,6 @@ uint32_t text_line_size(struct text *text, uint32_t lineidx) {
 
 uint32_t text_num_lines(struct text *text) { return text->nlines; }
 
-// given `char_idx` as a character index, return the byte index
-uint32_t charidx_to_byteidx(struct line *line, uint32_t char_idx) {
-  if (char_idx > line->nchars) {
-    return line->nchars;
-  }
-  return utf8_nbytes(line->data, char_idx);
-}
-
-// TODO: grapheme clusters
-// given `byte_idx` as a byte index, return the character index
-uint32_t byteidx_to_charidx(struct line *line, uint32_t byte_idx) {
-  if (byte_idx > line->nbytes) {
-    return line->nchars;
-  }
-
-  return utf8_nchars(line->data, byte_idx);
-}
-
-uint32_t char_byte_size(struct line *line, uint32_t byte_idx) {
-  return utf8_nbytes(line->data + byte_idx, 1);
-}
-
 void split_line(uint32_t col, struct line *line, struct line *next) {
   uint8_t *data = line->data;
   uint32_t nbytes = line->nbytes;
@@ -169,7 +175,7 @@ void new_line_at(struct text *text, uint32_t line, uint32_t col) {
   split_line(col, pl, cl);
 }
 
-void text_delete_line(struct text *text, uint32_t line) {
+void delete_line(struct text *text, uint32_t line) {
   // always keep a single line
   if (text->nlines == 1) {
     return;
@@ -230,20 +236,49 @@ void text_append(struct text *text, uint32_t line, uint32_t col, uint8_t *bytes,
 
 void text_delete(struct text *text, uint32_t line, uint32_t col,
                  uint32_t nchars) {
-  struct line *lp = &text->lines[line];
-  uint32_t max_chars = nchars > lp->nchars ? lp->nchars : nchars;
-  if (lp->nchars > 0) {
 
-    // get byte index and size for char to remove
-    uint32_t bytei = charidx_to_byteidx(lp, col);
-    uint32_t nbytes = utf8_nbytes(lp->data + bytei, max_chars);
+  // delete chars from current line
+  struct line *lp = &text->lines[line];
+  uint32_t chars_initial_line =
+      col + nchars > lp->nchars ? (lp->nchars - col) : nchars;
+  uint32_t bytei = charidx_to_byteidx(lp, col);
+  uint32_t nbytes = utf8_nbytes(lp->data + bytei, chars_initial_line);
+
+  memcpy(lp->data + bytei, lp->data + bytei + nbytes,
+         lp->nbytes - (bytei + nbytes));
+
+  lp->nbytes -= nbytes;
+  lp->nchars -= chars_initial_line;
+  lp->flags |= LineChanged;
+
+  uint32_t initial_line = line;
+  uint32_t left_to_delete = nchars - chars_initial_line;
+
+  // grab remaining chars from last line to delete from (if any)
+  uint32_t src_col = 0;
+  while (left_to_delete > 0 && line < text->nlines) {
+    ++line;
+    --left_to_delete; // newline char
+
+    struct line *lp = &text->lines[line];
+    uint32_t deleted_in_line =
+        left_to_delete > lp->nchars ? lp->nchars : left_to_delete;
+    src_col = deleted_in_line;
+    left_to_delete -= deleted_in_line;
+  }
 
-    memcpy(lp->data + bytei, lp->data + bytei + nbytes,
-           lp->nbytes - (bytei + nbytes));
+  if (line != initial_line) {
+    struct line *lp = &text->lines[line];
+    uint32_t bytei = charidx_to_byteidx(lp, src_col);
+    if (src_col < lp->nchars) {
+      append_to_line(&text->lines[initial_line], col, lp->data + bytei,
+                     lp->nbytes - bytei, lp->nchars - src_col);
+    }
+  }
 
-    lp->nbytes -= nbytes;
-    lp->nchars -= max_chars;
-    lp->flags |= LineChanged;
+  // delete all lines from current line + 1 to (and including) last line
+  for (uint32_t li = initial_line + 1; li <= line; ++li) {
+    delete_line(text, li);
   }
 }
 
@@ -272,11 +307,16 @@ uint32_t text_render(struct text *text, uint32_t line, uint32_t nlines,
   return ncmds;
 }
 
+void text_for_each_chunk(struct text *text, chunk_cb callback) {
+  // if representation of text is changed, this can be changed as well
+  text_for_each_line(text, 0, text->nlines, callback);
+}
+
 void text_for_each_line(struct text *text, uint32_t line, uint32_t nlines,
-                        line_cb callback) {
+                        chunk_cb callback) {
   for (uint32_t li = line; li < (line + nlines); ++li) {
     struct line *src_line = &text->lines[li];
-    struct txt_line line = (struct txt_line){
+    struct text_chunk line = (struct text_chunk){
         .text = src_line->data,
         .nbytes = src_line->nbytes,
         .nchars = src_line->nchars,
@@ -285,9 +325,9 @@ void text_for_each_line(struct text *text, uint32_t line, uint32_t nlines,
   }
 }
 
-struct txt_line text_get_line(struct text *text, uint32_t line) {
+struct text_chunk text_get_line(struct text *text, uint32_t line) {
   struct line *src_line = &text->lines[line];
-  return (struct txt_line){
+  return (struct text_chunk){
       .text = src_line->data,
       .nbytes = src_line->nbytes,
       .nchars = src_line->nchars,
diff --git a/src/text.h b/src/text.h
index 21de899..399eaf4 100644
--- a/src/text.h
+++ b/src/text.h
@@ -15,7 +15,6 @@ void text_append(struct text *text, uint32_t line, uint32_t col, uint8_t *bytes,
 
 void text_delete(struct text *text, uint32_t line, uint32_t col,
                  uint32_t nchars);
-void text_delete_line(struct text *text, uint32_t line);
 
 uint32_t text_render(struct text *text, uint32_t line, uint32_t nlines,
                      struct render_cmd *cmds, uint32_t max_ncmds);
@@ -24,16 +23,18 @@ uint32_t text_num_lines(struct text *text);
 uint32_t text_line_length(struct text *text, uint32_t lineidx);
 uint32_t text_line_size(struct text *text, uint32_t lineidx);
 
-struct txt_line {
+struct text_chunk {
   uint8_t *text;
   uint32_t nbytes;
   uint32_t nchars;
 };
 
-typedef void (*line_cb)(struct txt_line *line);
+typedef void (*chunk_cb)(struct text_chunk *chunk);
 void text_for_each_line(struct text *text, uint32_t line, uint32_t nlines,
-                        line_cb callback);
+                        chunk_cb callback);
 
-struct txt_line text_get_line(struct text *text, uint32_t line);
+void text_for_each_chunk(struct text *text, chunk_cb callback);
+
+struct text_chunk text_get_line(struct text *text, uint32_t line);
 
 bool text_line_contains_unicode(struct text *text, uint32_t line);
diff --git a/test/text.c b/test/text.c
index ec99890..9073553 100644
--- a/test/text.c
+++ b/test/text.c
@@ -1,4 +1,5 @@
 #include "assert.h"
+#include "stdio.h"
 #include "test.h"
 
 #include "text.h"
@@ -7,7 +8,7 @@
 #include <string.h>
 #include <wchar.h>
 
-void assert_line_equal(struct txt_line *line) {}
+void assert_line_equal(struct text_chunk *line) {}
 
 void test_add_text() {
   uint32_t lines_added, cols_added;
@@ -25,8 +26,12 @@ void test_add_text() {
   const char *txt2 = "This is line 2\n";
   text_append(t, 1, 0, (uint8_t *)txt2, strlen(txt2), &lines_added,
               &cols_added);
+  ASSERT(text_num_lines(t) == 3,
+         "Expected text to have three lines after second insertion");
   ASSERT_STR_EQ((const char *)text_get_line(t, 1).text, "This is line 2",
                 "Expected line 2 to be line 2");
+
+  text_destroy(t);
 }
 
 void test_delete_text() {
@@ -49,13 +54,36 @@ void test_delete_text() {
   const char *txt2 = "This is line 1\nThis is line 2\nThis is line 3";
   text_append(t, 0, 0, (uint8_t *)txt2, strlen(txt2), &lines_added,
               &cols_added);
+  ASSERT(text_num_lines(t) == 3,
+         "Expected to have three lines after inserting as many");
+
   text_delete(t, 1, 11, 3);
   ASSERT(text_line_length(t, 1) == 11,
          "Expected line to contain 11 chars after deletion");
-  struct txt_line line = text_get_line(t, 1);
+  struct text_chunk line = text_get_line(t, 1);
   ASSERT(strncmp((const char *)line.text, "This is lin", line.nbytes) == 0,
          "Expected deleted characters to be gone in the second line");
 
+  text_delete(t, 1, 0, text_line_length(t, 1) + 1);
+  ASSERT(text_num_lines(t) == 2,
+         "Expected to have two lines after deleting one");
+  struct text_chunk line2 = text_get_line(t, 1);
+  ASSERT(strncmp((const char *)line2.text, "This is line 3", line2.nbytes) == 0,
+         "Expected lines to have shifted upwards after deleting");
+
+  struct text *t3 = text_create(10);
+  const char *delete_me = "This is line🎙\nQ";
+  text_append(t3, 0, 0, (uint8_t *)delete_me, strlen(delete_me), &lines_added,
+              &cols_added);
+  text_delete(t3, 0, 13, 1);
+  struct text_chunk top_line = text_get_line(t3, 0);
+  ASSERT(strncmp((const char *)top_line.text, "This is line🎙Q",
+                 top_line.nbytes) == 0,
+         "Expected text from second line to be appended to first line when "
+         "deleting newline");
+  ASSERT(text_num_lines(t3) == 1,
+         "Expected text to have one line after deleting newline");
+
   // test utf-8
   struct text *t2 = text_create(10);
   const char *txt3 = "Emojis: 🇫🇮 🐮\n";
@@ -71,9 +99,13 @@ void test_delete_text() {
   text_delete(t2, 0, 10, 2);
   ASSERT(text_line_length(t2, 0) == 10,
          "Line length should be 10 after deleting the cow emoji and a space");
-  struct txt_line line2 = text_get_line(t2, 0);
-  ASSERT(strncmp((const char *)line2.text, "Emojis: 🇫🇮", line2.nbytes) == 0,
+  struct text_chunk line3 = text_get_line(t2, 0);
+  ASSERT(strncmp((const char *)line3.text, "Emojis: 🇫🇮", line3.nbytes) == 0,
          "Expected cow emoji plus space to be deleted");
+
+  text_destroy(t);
+  text_destroy(t2);
+  text_destroy(t3);
 }
 
 void run_text_tests() {