mds-kbdc: process includes

Signed-off-by: Mattias Andrée <maandree@operamail.com>

13 files changed, 256 insertions, 11 deletions

diff --git a/doc/info/mds.texinfo b/doc/info/mds.texinfo
index a4bb672..d2da0c4 100644
--- a/doc/info/mds.texinfo
+++ b/doc/info/mds.texinfo
@@ -5900,6 +5900,10 @@ two zeroes, you do not write `\100' as that would expand
 to the value of the hundredth variable. Instead you
 write `\1.00'.
 
+Use of function calls and variables inside
+@code{include}-statments invokes undefined behaviour.
+Escaped numerals greater than or equal to the 31th
+power of 2 also invoke undefined behaviour.
 
 
 @node Builtin Functions
diff --git a/src/mds-kbdc/make-tree.c b/src/mds-kbdc/make-tree.c
index fab402b..5c644b4 100644
--- a/src/mds-kbdc/make-tree.c
+++ b/src/mds-kbdc/make-tree.c
@@ -662,6 +662,9 @@ int parse_to_tree(const char* restrict filename, mds_kbdc_parsed_t* restrict res
   size_t line_i, line_n;
   const char** keyword_stack = NULL;
   mds_kbdc_tree_t*** tree_stack = NULL;
+  char* cwd = NULL;
+  char* old = NULL;
+  size_t cwd_size = 4096 >> 1;
   size_t stack_ptr = 0;
   int saved_errno, in_array = 0;
   
@@ -672,7 +675,29 @@ int parse_to_tree(const char* restrict filename, mds_kbdc_parsed_t* restrict res
    * can be misleading as the program can have changed working
    * directory to be able to resolve filenames. */
   result->pathname = realpath(filename, NULL); /* XXX use absolute path */
-  fail_if (result->pathname == NULL);
+  if (result->pathname == NULL)
+    {
+      fail_if (errno != ENOENT);
+      saved_errno = errno;
+      
+      /* Get the current working directory. */
+      /* glibc offers ways to do this in just one function call,
+       * but we will not assume that glibc is used here. */
+      for (;;)
+	{
+	  fail_if (xxrealloc(old, cwd, cwd_size <<= 1, char));
+	  if (getcwd(cwd, cwd_size))
+	    break;
+	  else
+	    fail_if (errno != ERANGE);
+	}
+      
+      result->pathname = strdup(filename);
+      fail_if (result->pathname == NULL);
+      NEW_ERROR_(result, ERROR, 0, 0, 0, 0, 1, "no such file or directory in ‘%s’", cwd);
+      free(cwd);
+      return 0;
+    }
   
   /* Check that the file exists and can be read. */
   if (access(result->pathname, R_OK) < 0)
@@ -1043,6 +1068,8 @@ int parse_to_tree(const char* restrict filename, mds_kbdc_parsed_t* restrict res
   saved_errno = errno;
   free(keyword_stack);
   free(tree_stack);
+  free(cwd);
+  free(old);
   return errno = saved_errno, -1;
 }
 
diff --git a/src/mds-kbdc/process-includes.c b/src/mds-kbdc/process-includes.c
index 9677671..8c4bbbf 100644
--- a/src/mds-kbdc/process-includes.c
+++ b/src/mds-kbdc/process-includes.c
@@ -16,6 +16,7 @@
  * along with this program.  If not, see <http://www.gnu.org/licenses/>.
  */
 #include "process-includes.h"
+/* TODO we need to deal with mutually recursive includes */
 
 #include "make-tree.h"
 #include "simplify-tree.h"
@@ -68,26 +69,34 @@ static int transfer_errors(mds_kbdc_parsed_t* restrict subresult, mds_kbdc_tree_
 {
   mds_kbdc_parse_error_t** errors = NULL;
   mds_kbdc_parse_error_t* suberror;
-  size_t errors_ptr = 0, i;
+  size_t errors_ptr = 0;
   int saved_errno;
   
-  /* List errors backwards, so that we can easily insert “included from here”-notes. */
+  /* Allocate temporary list for errors. */
+  if (subresult->errors_ptr == 0)
+    return 0;
   fail_if (xmalloc(errors, subresult->errors_ptr * 2, mds_kbdc_parse_error_t*));
+  
+  /* List errors backwards, so that we can easily handle errors and add “included from here”-note. */
   while (subresult->errors_ptr--)
     {
       suberror = subresult->errors[subresult->errors_ptr];
+      
+      /* If it is more severe than a note, we want to say there it was included. */
       if (suberror->severity > MDS_KBDC_PARSE_ERROR_NOTE)
 	{
 	  NEW_ERROR(tree, NOTE, "included from here");
 	  errors[errors_ptr++] = error;
 	  result->errors[--(result->errors_ptr)] = NULL;
 	}
+      
+      /* Include error. */
       errors[errors_ptr++] = suberror;
       subresult->errors[subresult->errors_ptr] = NULL;
     }
   
   /* Append errors. */
-  for (i = 0; i < errors_ptr; errors[i++] = NULL)
+  for (; errors_ptr--; errors[errors_ptr] = NULL)
     {
       if (result->errors_ptr + 1 >= result->errors_size)
 	{
@@ -98,8 +107,8 @@ static int transfer_errors(mds_kbdc_parsed_t* restrict subresult, mds_kbdc_tree_
 	  result->errors = new_errors;
 	  result->errors_size = new_errors_size;
 	}
-  
-      result->errors[result->errors_ptr++] = errors[i];
+      
+      result->errors[result->errors_ptr++] = errors[errors_ptr];
       result->errors[result->errors_ptr] = NULL;
     }
   
@@ -108,10 +117,7 @@ static int transfer_errors(mds_kbdc_parsed_t* restrict subresult, mds_kbdc_tree_
  pfail:
   saved_errno = errno;
   while (errors_ptr--)
-    if (errors[errors_ptr] == NULL)
-      break;
-    else
-      mds_kbdc_parse_error_free(errors[errors_ptr]);
+    mds_kbdc_parse_error_free(errors[errors_ptr]);
   free(errors);
   return errno = saved_errno, -1;
 }
@@ -150,7 +156,7 @@ static int process_include(mds_kbdc_tree_include_t* restrict tree)
    * but we will not assume that glibc is used here. */
   for (;;)
     {
-      fail_if (!xxrealloc(old, cwd, cwd_size <<= 1, char));
+      fail_if (xxrealloc(old, cwd, cwd_size <<= 1, char));
       if (getcwd(cwd, cwd_size))
 	break;
       else
@@ -165,6 +171,10 @@ static int process_include(mds_kbdc_tree_include_t* restrict tree)
   our_result = result;
   
   /* Process include. */
+  old = tree->filename, tree->filename = NULL;
+  tree->filename = parse_raw_string(old);
+  fail_if (tree->filename == NULL);
+  free(old), old = NULL;
   process (parse_to_tree(tree->filename, &subresult));
   process (simplify_tree(&subresult));
   process (process_includes(&subresult));
diff --git a/src/mds-kbdc/raw-data.c b/src/mds-kbdc/raw-data.c
index b0985fe..1803cd5 100644
--- a/src/mds-kbdc/raw-data.c
+++ b/src/mds-kbdc/raw-data.c
@@ -18,6 +18,7 @@
 #include "raw-data.h"
 
 #include "globals.h"
+#include "string.h"
 
 #include <libmdsserver/macros.h>
 
@@ -29,6 +30,7 @@
 #include <unistd.h>
 #include <stdio.h>
 #include <string.h>
+#include <stdint.h>
 
 
 
@@ -54,6 +56,8 @@ void mds_kbdc_source_code_initialise(mds_kbdc_source_code_t* restrict this)
  */
 void mds_kbdc_source_code_destroy(mds_kbdc_source_code_t* restrict this)
 {
+  if (this == NULL)
+    return;
   free(this->lines),        this->lines        = NULL;
   free(this->real_lines),   this->real_lines   = NULL;
   free(this->content),      this->content      = NULL;
@@ -68,6 +72,8 @@ void mds_kbdc_source_code_destroy(mds_kbdc_source_code_t* restrict this)
  */
 void mds_kbdc_source_code_free(mds_kbdc_source_code_t* restrict this)
 {
+  if (this == NULL)
+    return;
   free(this->lines);
   free(this->real_lines);
   free(this->content);
@@ -371,3 +377,101 @@ int read_source_lines(const char* restrict pathname, mds_kbdc_source_code_t* res
   return -1;
 }
 
+
+/**
+ * Encode a character in UTF-8
+ * 
+ * @param   buffer     The buffer where the character should be stored
+ * @param   character  The character
+ * @return             The of the character in `buffer`, `NULL` on error
+ */
+static char* encode_utf8(char* buffer, char32_t character)
+{
+  char32_t text[2];
+  char* restrict str;
+  char* restrict str_;
+  
+  text[0] = character;
+  text[1] = -1;
+  
+  if (str_ = str = string_encode(text), str == NULL)
+    return NULL;
+  
+  while (*str)
+    *buffer++ = *str++;
+  
+  free(str_);
+  return buffer;
+}
+
+
+/**
+ * Parse a quoted and escaped string that may not include function calls or variable dereferences
+ * 
+ * @param   string  The string
+ * @return          The string in machine-readable format, `NULL` on error
+ */
+char* parse_raw_string(const char* restrict string)
+{
+#define r(lower, upper)  (((lower) <= c) && (c <= (upper)))
+  
+  char* rc;
+  char* p;
+  int escape = 0;
+  char32_t buf;
+  
+  /* We know that the output string can only be shorter because
+   * it is surrounded by 2 quotes and escape can only be longer
+   * then what they escape, for example \uA0, is four characters,
+   * but when parsed it generateds 2 bytes in UTF-8, and their
+   * is not code point whose UTF-8 encoding is longer than its
+   * hexadecimal representation. */
+  p = rc = malloc(strlen(string) * sizeof(char));
+  if (rc == NULL)
+    return NULL;
+  
+  while (*string)
+    {
+      char c = *string++;
+      
+      if (escape > 1)
+	{
+	  if      ((escape ==  8) && r('0', '7'))  buf = (buf << 3) | (c & 15);
+	  else if ((escape == 16) && r('0', '9'))  buf = (buf << 4) | (c & 15);
+	  else if ((escape == 16) && r('a', 'f'))  buf = (buf << 4) | ((c & 15) + 9);
+	  else if ((escape == 16) && r('A', 'F'))  buf = (buf << 4) | ((c & 15) + 9);
+	  else
+	    goto end_of_escape;
+	  continue;
+	end_of_escape:
+	  escape = 0;
+	  p = encode_utf8(p, buf);
+	  if (p == NULL)
+	    goto fail;
+	  if (c != '.')
+	    *p++ = c;
+	}
+      else if (escape == 1)
+	{
+	  escape = 0, buf = 0;
+	  switch (c)
+	    {
+	    case '0':  escape = 8;   break;
+	    case 'u':  escape = 16;  break;
+	    default:   *p++ = c;     break;
+	    }
+	}
+      else if (c == '\\')
+	escape = 1;
+      else if (c != '\"')
+	*p++ = c;
+    }
+  
+  *p = '\0';
+  return rc;
+ fail:
+  free(rc);
+  return NULL;
+#undef r
+}
+
diff --git a/src/mds-kbdc/raw-data.h b/src/mds-kbdc/raw-data.h
index 4bc7355..a2f1edc 100644
--- a/src/mds-kbdc/raw-data.h
+++ b/src/mds-kbdc/raw-data.h
@@ -104,5 +104,14 @@ size_t get_end_of_call(char* restrict content, size_t offset, size_t size) __att
 int read_source_lines(const char* restrict pathname, mds_kbdc_source_code_t* restrict source_code);
 
 
+/**
+ * Parse a quoted and escaped string that may not include function calls or variable dereferences
+ * 
+ * @param   string  The string
+ * @return          The string in machine-readable format, `NULL` on error
+ */
+char* parse_raw_string(const char* restrict string);
+
+
 #endif
 
diff --git a/test-files/mds-kbdc/process-includes/invalid/_error b/test-files/mds-kbdc/process-includes/invalid/_error
new file mode 100644
index 0000000..aa42b42
--- /dev/null
+++ b/test-files/mds-kbdc/process-includes/invalid/_error
@@ -0,0 +1,2 @@
+[.]
+
diff --git a/test-files/mds-kbdc/process-includes/invalid/include_error b/test-files/mds-kbdc/process-includes/invalid/include_error
new file mode 100644
index 0000000..cec9385
--- /dev/null
+++ b/test-files/mds-kbdc/process-includes/invalid/include_error
@@ -0,0 +1,31 @@
+include "_error"
+
+# (include (@ 1 0-7) ‘/home/mattias/git/mds/test-files/mds-kbdc/process-includes/invalid/_error’
+#   (.inner
+#     (map (@ 1 0-0)
+#       (.sequence
+#         (nothing (@ 1 1-2))
+#       )
+#       (.result nil)
+#     )
+#   )
+# )
+# .../test-files/mds-kbdc/process-includes/invalid/_error:1:0–1: warning: alternated value statement is undefined unless the alternatives are identical
+#  [.]
+#  ^
+# .../test-files/mds-kbdc/process-includes/invalid/include_error:1:0–7: note: included from here
+#  include "_error"
+#  ^^^^^^^
+# .../test-files/mds-kbdc/process-includes/invalid/_error:1:0–1: warning: singleton alternation
+#  [.]
+#  ^
+# .../test-files/mds-kbdc/process-includes/invalid/include_error:1:0–7: note: included from here
+#  include "_error"
+#  ^^^^^^^
+# .../test-files/mds-kbdc/process-includes/invalid/_error:1:1–2: error: nothing in value statement
+#  [.]
+#   ^
+# .../test-files/mds-kbdc/process-includes/invalid/include_error:1:0–7: note: included from here
+#  include "_error"
+#  ^^^^^^^
+
diff --git a/test-files/mds-kbdc/process-includes/invalid/include_nonexisting b/test-files/mds-kbdc/process-includes/invalid/include_nonexisting
new file mode 100644
index 0000000..9ff0f5c
--- /dev/null
+++ b/test-files/mds-kbdc/process-includes/invalid/include_nonexisting
@@ -0,0 +1,11 @@
+include "this file does not exist"
+
+# (include (@ 1 0-7) ‘this file does not exist’
+#   (.inner nil)
+# )
+# this file does not exist: error: no such file or directory in ‘.../test-files/mds-kbdc/process-includes/invalid’
+# 
+# .../test-files/mds-kbdc/process-includes/invalid/include_nonexisting:1:0–7: note: included from here
+#  include "this file does not exist"
+#  ^^^^^^^
+
diff --git a/test-files/mds-kbdc/process-includes/valid/_empty b/test-files/mds-kbdc/process-includes/valid/_empty
new file mode 100644
index 0000000..e69de29
--- /dev/null
+++ b/test-files/mds-kbdc/process-includes/valid/_empty
diff --git a/test-files/mds-kbdc/process-includes/valid/_mapping b/test-files/mds-kbdc/process-includes/valid/_mapping
new file mode 100644
index 0000000..6026f43
--- /dev/null
+++ b/test-files/mds-kbdc/process-includes/valid/_mapping
@@ -0,0 +1,2 @@
+<dead letter ´> "e" : "é"
+
diff --git a/test-files/mds-kbdc/process-includes/valid/include_empty_file b/test-files/mds-kbdc/process-includes/valid/include_empty_file
new file mode 100644
index 0000000..c784768
--- /dev/null
+++ b/test-files/mds-kbdc/process-includes/valid/include_empty_file
@@ -0,0 +1,11 @@
+include "_empty"
+
+# (include (@ 1 0-7) ‘.../test-files/mds-kbdc/process-includes/valid/_empty’
+#   (.inner nil)
+# )
+# .../test-files/mds-kbdc/process-includes/valid/_empty: warning: file is empty
+# 
+# .../test-files/mds-kbdc/process-includes/valid/include_empty_file:1:0–7: note: included from here
+#  include "_empty"
+#  ^^^^^^^
+
diff --git a/test-files/mds-kbdc/process-includes/valid/include_include_empty_file b/test-files/mds-kbdc/process-includes/valid/include_include_empty_file
new file mode 100644
index 0000000..9b8e1c1
--- /dev/null
+++ b/test-files/mds-kbdc/process-includes/valid/include_include_empty_file
@@ -0,0 +1,18 @@
+include "include_empty_file"
+
+# (include (@ 1 0-7) ‘.../test-files/mds-kbdc/process-includes/valid/include_empty_file’
+#   (.inner
+#     (include (@ 1 0-7) ‘.../test-files/mds-kbdc/process-includes/valid/_empty’
+#       (.inner nil)
+#     )
+#   )
+# )
+# .../test-files/mds-kbdc/process-includes/valid/_empty: warning: file is empty
+# 
+# .../test-files/mds-kbdc/process-includes/valid/include_include_empty_file:1:0–7: note: included from here
+#  include "include_empty_file"
+#  ^^^^^^^
+# .../test-files/mds-kbdc/process-includes/valid/include_empty_file:1:0–7: note: included from here
+#  include "_empty"
+#  ^^^^^^^
+
diff --git a/test-files/mds-kbdc/process-includes/valid/include_mapping b/test-files/mds-kbdc/process-includes/valid/include_mapping
new file mode 100644
index 0000000..e9d531f
--- /dev/null
+++ b/test-files/mds-kbdc/process-includes/valid/include_mapping
@@ -0,0 +1,16 @@
+include "_mapping"
+
+# (include (@ 1 0-7) ‘/home/mattias/git/mds/test-files/mds-kbdc/process-includes/valid/_mapping’
+#   (.inner
+#     (map (@ 1 0-0)
+#       (.sequence
+#         (keys (@ 1 0-16) ‘<dead letter ´>’)
+#         (string (@ 1 17-20) ‘"e"’)
+#       )
+#       (.result
+#         (string (@ 1 23-27) ‘"é"’)
+#       )
+#     )
+#   )
+# )
+