Improve and simplify text escaping

1 files changed, 47 insertions, 59 deletions

diff --git a/src/gemini2html.c b/src/gemini2html.c
index c71bc58..48c5128 100644
--- a/src/gemini2html.c
+++ b/src/gemini2html.c
@@ -23,37 +23,7 @@
 
 #include "gemini2html.h"
 
-/* Escapes text so it can be safely inserted within HTML tags.
- * Examples:
- *   "5 < 6"    => "5 &lt; 6"
- *   "6 > 5"    => "6 &gt; 5"
- *   "me & you" => "me &amp; you"
- */
-char* escape_html_tag(const char* unescaped, const size_t size) {
-    char* escaped = malloc(5 * size + 1);
-    size_t escaped_pos = 0;
-    char substitute[6] = "";
-
-    for (size_t pos = 0; pos < size; pos++) {
-        switch (unescaped[pos]) {
-            case '<': strcpy(substitute, "&lt;");  break;
-            case '>': strcpy(substitute, "&gt;");  break;
-            case '&': strcpy(substitute, "&amp;"); break;
-            default:
-                strncpy(substitute, unescaped + pos, 1);
-                substitute[1] = '\0';
-                break;
-        }
-        for (size_t i = 0; i < strlen(substitute); i++) {
-            escaped[escaped_pos++] = substitute[i];
-        }
-    }
-    escaped[escaped_pos] = '\0';
-
-    return escaped;
-}
-
-/* Escapes text so it can be safely inserted within HTML tags.
+/* Escapes text so it can be safely inserted within HTML tags and attributes.
  * Examples:
  *   "5 < 6"        => "5 &lt; 6"
  *   "6 > 5"        => "6 &gt; 5"
@@ -61,29 +31,47 @@ char* escape_html_tag(const char* unescaped, const size_t size) {
  *   "i'm so high"  => "I&apos;m so high"
  *   ""hello world" => "&quot;hello world&quot;"
  */
-char* escape_html_attribute(const char* unescaped, const size_t size) {
+char* escape_text(const char* unescaped, const size_t size) {
+    /* In the worst case, every unescaped character corresponds to 6 escape characters.
+     * Examples:
+     *   "'" => "&apos;"
+     *   """ => "&quot;"
+     * 
+     * +1 for the NULL byte at the end of the string
+     */
     char* escaped = malloc(6 * size + 1);
+
     size_t escaped_pos = 0;
-    char substitute[7] = "";
-
-    for (size_t pos = 0; pos < size; pos++) {
-        switch (unescaped[pos]) {
-            case '<':  strcpy(substitute, "&lt;");    break;
-            case '>':  strcpy(substitute, "&gt;");    break;
-            case '&':  strcpy(substitute, "&amp;");   break;
-            case '\'': strcpy(substitute, "&apos;");  break;
-            case '\"': strcpy(substitute, "&quot;");  break;
+    for (size_t unescaped_pos = 0; unescaped_pos < size; unescaped_pos++) {
+        switch (unescaped[unescaped_pos]) {
+            case '<':
+                strncpy(escaped + escaped_pos, "&lt;", 4);
+                escaped_pos += 4;
+                break;
+            case '>':
+                strncpy(escaped + escaped_pos, "&gt;", 4);
+                escaped_pos += 4;
+                break;
+            case '&':
+                strncpy(escaped + escaped_pos, "&amp;", 5);
+                escaped_pos += 5;
+                break;
+            case '\'':
+                strncpy(escaped + escaped_pos, "&apos;", 6);
+                escaped_pos += 6;
+                break;
+            case '\"':
+                strncpy(escaped + escaped_pos, "&quot;", 6);
+                escaped_pos += 6;
+                break;
             default:
-                strncpy(substitute, unescaped + pos, 1);
-                substitute[1] = '\0';
+                strncpy(escaped + escaped_pos, unescaped + unescaped_pos, 1);
+                escaped_pos += 1;
                 break;
         }
-        for (size_t i = 0; i < strlen(substitute); i++) {
-            escaped[escaped_pos++] = substitute[i];
-        }
     }
-    escaped[escaped_pos] = '\0';
 
+    escaped[escaped_pos] = '\0';
     return escaped;
 }
 
@@ -99,8 +87,8 @@ int skip_non_whitespace(const char* str, size_t* pos) {
     while (str[*pos] != '\0' && !(str[*pos] == ' ' || str[*pos] == '\t')) (*pos)++;
 }
 
-char* escape_and_add_tags(const char* opening_tag, const char* unescaped, const char* closing_tag, const size_t unescaped_size, char* (*escape_fun)(const char*, const size_t), const bool add_newline) {
-    char* escaped = escape_fun(unescaped, unescaped_size);
+char* escape_and_add_tags(const char* opening_tag, const char* unescaped, const char* closing_tag, const size_t unescaped_size, const bool add_newline) {
+    char* escaped = escape_text(unescaped, unescaped_size);
 
     size_t escaped_len = strlen(escaped);
     char* escaped_with_tags = malloc(strlen(opening_tag) + escaped_len + strlen(closing_tag) + 1);
@@ -142,11 +130,11 @@ enum linetype getlinetype(const char* line, const bool pre_is_toggled) {
 
 char* convert_text_line(const char* line, const size_t size) {
     if (line[0] == '\n') return strdup("<br/>\n");
-    else return escape_and_add_tags("<p>", line, "</p>", size, escape_html_tag, true);
+    else return escape_and_add_tags("<p>", line, "</p>", size, true);
 }
 
 char* convert_pre_line(const char* line, const size_t size) {
-    return escape_and_add_tags("", line, "", size, escape_html_tag, true);
+    return escape_and_add_tags("", line, "", size, true);
 }
 
 char* convert_pre_toggle_line(const char* line, const size_t size) {
@@ -168,14 +156,14 @@ char* convert_link_line(const char* line, const size_t size) {
     size_t before_link_text = after_link;
     skip_whitespace(line, &before_link_text);
 
-    char* escaped_href = escape_and_add_tags("<a rel=\"noreferrer noopener\" href=\"", line + before_link, "\">", after_link - before_link, escape_html_attribute, false);
+    char* escaped_href = escape_and_add_tags("<a rel=\"noreferrer noopener\" href=\"", line + before_link, "\">", after_link - before_link, false);
 
     char* escaped_with_tags = NULL;
 
     if (line[before_link_text] == '\0') {
-        escaped_with_tags = escape_and_add_tags(escaped_href, line + before_link, "</a>", after_link - before_link, escape_html_tag, true);
+        escaped_with_tags = escape_and_add_tags(escaped_href, line + before_link, "</a>", after_link - before_link, true);
     } else {
-        escaped_with_tags = escape_and_add_tags(escaped_href, line + before_link_text, "</a>", size - before_link_text, escape_html_tag, true);
+        escaped_with_tags = escape_and_add_tags(escaped_href, line + before_link_text, "</a>", size - before_link_text, true);
     }
 
     free(escaped_href);
@@ -190,7 +178,7 @@ char* convert_h1_line(const char* line, const size_t size) {
     skip_whitespace(line, &pos);
 
     if (line[pos] == '\0') return strdup("<h1></h1>\n");
-    else return escape_and_add_tags("<h1>", line + pos, "</h1>", size - pos, escape_html_tag, true);
+    else return escape_and_add_tags("<h1>", line + pos, "</h1>", size - pos, true);
 }
 
 char* convert_h2_line(const char* line, const size_t size) {
@@ -199,7 +187,7 @@ char* convert_h2_line(const char* line, const size_t size) {
     skip_whitespace(line, &pos);
 
     if (line[pos] == '\0') return strdup("<h2></h2>\n");
-    else return escape_and_add_tags("<h2>", line + pos, "</h2>", size - pos, escape_html_tag, true);
+    else return escape_and_add_tags("<h2>", line + pos, "</h2>", size - pos, true);
 }
 
 char* convert_h3_line(const char* line, const size_t size) {
@@ -208,7 +196,7 @@ char* convert_h3_line(const char* line, const size_t size) {
     skip_whitespace(line, &pos);
 
     if (line[pos] == '\0') return strdup("<h3></h3>\n");
-    else return escape_and_add_tags("<h3>", line + pos, "</h3>", size - pos, escape_html_tag, true);
+    else return escape_and_add_tags("<h3>", line + pos, "</h3>", size - pos, true);
 }
 
 char* convert_ul_line(const char* line, const size_t size) {
@@ -217,7 +205,7 @@ char* convert_ul_line(const char* line, const size_t size) {
     skip_whitespace(line, &pos);
 
     if (line[pos] == '\0') return strdup("<li></li>\n");
-    else return escape_and_add_tags("<li>", line + pos, "</li>", size - pos, escape_html_tag, true);
+    else return escape_and_add_tags("<li>", line + pos, "</li>", size - pos, true);
 }
 
 char* convert_quote_line(const char* line, const size_t size) {
@@ -226,7 +214,7 @@ char* convert_quote_line(const char* line, const size_t size) {
     skip_whitespace(line, &pos);
 
     if (line[pos] == '\0') return strdup("\n");
-    else return escape_and_add_tags("", line + pos, "", size - pos, escape_html_tag, true);
+    else return escape_and_add_tags("", line + pos, "", size - pos, true);
 }
 
 char* convert_line(const char* line, const size_t size, const enum linetype type) {