[12_23] url: support customized protocol

Kernel/Types/analyze.cpp

@@ -50,6 +50,15 @@ is_alpha (string s) {
   return true;
 }
 
+bool
+is_alphanum (string s) {
+  int i;
+  if (N (s) == 0) return false;
+  for (i= 0; i < N (s); i++)
+    if (!(is_alpha (s[i]) || is_digit (s[i]))) return false;
+  return true;
+}
+
 bool
 is_locase_alpha (string s) {
   int i;
@@ -1439,6 +1448,11 @@ occurs (string what, string in) {
   return search_forwards (what, 0, in) >= 0;
 }
 
+bool
+contains (string s, string what) {
+  return search_forwards (what, 0, s) >= 0;
+}
+
 int
 search_backwards (string s, int pos, string in) {
   while (pos >= 0) {

Kernel/Types/analyze.hpp

@@ -97,6 +97,8 @@ bool is_iso_upcase (char c);
  */
 bool is_alpha (string s);
 
+bool is_alphanum (string s);
+
 /**
  * @brief Checks if a string contains only lowercase alphabetic characters.
  *
@@ -706,6 +708,8 @@ int count_occurrences (string what, string in);
  */
 bool occurs (string what, string in);
 
+bool contains (string s, string what);
+
 /**
  * Finds the length of the longest string that is both a suffix of the first
  * string and a prefix of the second string.

System/Classes/url.cpp

@@ -243,15 +243,29 @@ url_default (string name, int type= URL_SYSTEM) {
 
 url
 url_general (string name, int type= URL_SYSTEM) {
-  if (starts (name, "local:")) return file_url (name (6, N (name)));
-  if (starts (name, "file://")) return file_url (name (7, N (name)));
-  if (starts (name, "http://")) return http_url (name (7, N (name)));
-  if (starts (name, "https://")) return https_url (name (8, N (name)));
-  if (starts (name, "ftp://")) return ftp_url (name (6, N (name)));
-  if (starts (name, "//")) return blank_url (name (2, N (name)));
-  if (heuristic_is_path (name, type)) return url_path (name, type);
+  if (contains (name, "://")) {
+    array<string> tokens  = tokenize (name, "://");
+    string        protocol= tokens[0];
+    string        path    = tokens[1];
+    if (N (tokens) == 2 && is_alphanum (protocol)) {
+      if (protocol == "file") return file_url (path);
+      if (protocol == "http") return http_url (path);
+      if (protocol == "https") return https_url (path);
+      if (protocol == "ftp") return ftp_url (path);
+      return url_root (tokens[0]) * url_get_name (tokens[1], type);
+    }
+  }
+  if (starts (name, "local:")) {
+    return file_url (name (6, N (name)));
+  }
+  if (starts (name, "//")) {
+    return blank_url (name (2, N (name)));
+  }
+  if (heuristic_is_path (name, type)) {
+    return url_path (name, type);
+  }
   if (heuristic_is_default (name, type)) return url_default (name, type);
-  if (heuristic_is_mingw_default (name, type))
+  if (os_mingw () && heuristic_is_mingw_default (name, type))
     return url_mingw_default (name, type);
   if (type != URL_CLEAN_UNIX) {
     if (heuristic_is_http (name)) return http_url (name);

tests/Kernel/Types/analyze_test.cpp

@@ -1,7 +1,7 @@
 #include "a_lolly_test.hpp"
 #include "analyze.hpp"
 
-TEST_CASE ("test is alpha") {
+TEST_CASE ("is_alpha") {
   for (unsigned char c= 0; c < 255; c++) {
     if ((c >= 65 && c <= 90) || (c >= 97 && c <= 122)) {
       CHECK (is_alpha (c));
@@ -12,6 +12,57 @@ TEST_CASE ("test is alpha") {
   }
 }
 
+TEST_CASE ("is_digit") {
+  for (unsigned char c= 0; c < 255; c++) {
+    if (c >= 48 && c <= 57) {
+      CHECK (is_digit (c));
+    }
+    else {
+      CHECK (!is_digit (c));
+    }
+  }
+}
+
+TEST_CASE ("is_space") {
+  for (unsigned char c= 0; c < 255; c++) {
+    if ((c == 9) || (c == 10) || (c == 13) || (c == 32)) {
+      CHECK (is_space (c));
+    }
+    else {
+      CHECK (!is_space (c));
+    }
+  }
+}
+
+TEST_CASE ("is_binary_digit") {
+  for (unsigned char c= 0; c < 255; c++) {
+    if ((c == '0') || (c == '1')) {
+      CHECK (is_binary_digit (c));
+    }
+    else {
+      CHECK (!is_binary_digit (c));
+    }
+  }
+}
+
+TEST_CASE ("is_alpha") {
+  CHECK (is_alpha ("a"));
+  CHECK (is_alpha ("abc"));
+  CHECK (is_alpha ("Hello"));
+  CHECK (!is_alpha ("!"));
+  CHECK (!is_alpha ("abc123"));
+  CHECK (!is_alpha (""));
+}
+
+TEST_CASE ("is_alphanum") {
+  CHECK (is_alphanum ("s3"));
+  CHECK (is_alphanum ("ipv6"));
+  CHECK (is_alphanum ("abc"));
+  CHECK (is_alphanum ("123"));
+  CHECK (!is_alphanum (""));
+  CHECK (!is_alphanum ("!"));
+}
+
 TEST_CASE ("cjk_unified_ideographs") {
   CHECK (is_cjk_unified_ideographs ("<#4E2D>"));
   CHECK (has_cjk_unified_ideographs ("<#4E2D>"));
@@ -103,44 +154,28 @@ TEST_CASE ("test_read_word") {
   CHECK_EQ (i, 0);
 }
 
-TEST_CASE ("test_is_digit") {
-  for (unsigned char c= 0; c < 255; c++) {
-    if (c >= 48 && c <= 57) {
-      CHECK (is_digit (c));
-    }
-    else {
-      CHECK (!is_digit (c));
-    }
-  }
-}
-
-TEST_CASE ("test_is_space") {
-  for (unsigned char c= 0; c < 255; c++) {
-    if ((c == 9) || (c == 10) || (c == 13) || (c == 32)) {
-      CHECK (is_space (c));
-    }
-    else {
-      CHECK (!is_space (c));
-    }
-  }
-}
-
-TEST_CASE ("test_is_binary_digit") {
-  for (unsigned char c= 0; c < 255; c++) {
-    if ((c == '0') || (c == '1')) {
-      CHECK (is_binary_digit (c));
-    }
-    else {
-      CHECK (!is_binary_digit (c));
-    }
-  }
+TEST_CASE ("contains/occurs") {
+  CHECK (contains ("abc", "a"));
+  CHECK (contains ("abc", "ab"));
+  CHECK (contains ("abc", "bc"));
+  CHECK (!contains ("abc", "B"));
+  CHECK (contains ("", ""));
+  CHECK (contains ("abc", ""));
+  CHECK (!contains ("abc", " "));
+  CHECK (contains ("hello world", " "));
 }
 
 TEST_CASE ("replace") {
   CHECK_EQ (replace ("a-b", "-", "_") == "a_b", true);
   CHECK_EQ (replace ("a-b-c", "-", "_") == "a_b_c", true);
 }
 
+TEST_CASE ("tokenize") {
+  CHECK_EQ (tokenize ("hello world", " "), array<string> ("hello", "world"));
+  CHECK_EQ (tokenize ("zotero://select/library/items/2AIFJFS7", "://"),
+            array<string> ("zotero", "select/library/items/2AIFJFS7"));
+}
+
 TEST_CASE ("roman_nr") {
   SUBCASE ("0-9") {
     string_eq (roman_nr (0), "o");

tests/System/Classes/url_test.cpp

@@ -167,4 +167,14 @@ TEST_CASE ("as_string") {
 #endif
 }
 
+TEST_CASE ("unknown protocol like zotero") {
+  url zotero_u= url_system ("zotero://select/library/items/2AIFJFS7");
+  string_eq (zotero_u.protocol (), "zotero");
+  CHECK (!is_or (zotero_u));
+
+  url tmfs_u= url_system ("tmfs://git/status");
+  string_eq (tmfs_u.protocol (), "tmfs");
+  CHECK (!is_or (tmfs_u));
+}
+
 TEST_MEMORY_LEAK_ALL