Changeset 069dddc59c195676361a8943bc9c92c2cb26c09a

Timestamp:

06/12/2008 10:36:43 AM (2 months ago)

Author:

mitchell <mitchell@frost.(none)>

git-committer:

mitchell <mitchell@frost.(none)> 1213292203 -0400

git-parent:

[01610779fcf24388e97f8de5cc385b8ffdbcd1d1]

git-author:

mitchell <mitchell@frost.(none)> 1213292203 -0400

Message:

Integrated entity parsing into Ohcount.
See the documentation for Ohcount::parse_entities.

Files:

• bin/ohcount (modified) (4 diffs)
• ext/ohcount_native/common.h (modified) (1 diff)
• ext/ohcount_native/parser.c (modified) (2 diffs)
• ext/ohcount_native/ragel_parser.c (modified) (6 diffs)
• ext/ohcount_native/ragel_parser.h (modified) (1 diff)
• ext/ohcount_native/ruby_binding.c (modified) (1 diff)

bin/ohcount

@@ -71 +71 @@
-  end
-
+  # Entities
+  # TODO: do more fun stuff rather than print entities and positions
+  def entities
+    files.each do |file|
+      sfc = Ohcount::SimpleFileContext.new(file, files)
+      polyglot = Ohcount::Detector.detect(sfc)
+      if polyglot
+        Ohcount::parse_entities(sfc.contents, polyglot) do |language, entity, s, e|
+          puts "#{language}\t#{entity}\t#{s}\t#{e}"
+        end
+      end
+    end
+  end
+
-  def help
-    puts <<HELP
@@ -81 +95 @@
-   -a, --annotate
-   -d, --detect
+   -e, --entities
-   -h, --help
-   -s, --summary
@@ -96 +111 @@
-   For each source code file found, the file name will be emitted to
-   stdout prefixed with a tab-delimited language name.
+
+-e, --entities
+
+   For now, simply prints the positions of entities parsed in all
+   source files found within the given paths.
-
--h, --help                      Display this message
@@ -245 +265 @@
-    when '-i', '--individual'
-      self.subcommand = :individual
+    when '-e', '--entities'
+      self.subcommand = :entities
-    when '-?', '-h', '--help'
-      self.subcommand = :help

ext/ohcount_native/common.h

@@ -38 +38 @@
-#include "language_breakdown.h"
-#include "parser.h"
+#include "ragel_parser.h"
-
-/*******************************************

ext/ohcount_native/parser.c

@@ -11 +11 @@
-#include "ruby.h"
-#include "common.h"
-#include "ragel_parser.h"
-
-
@@ -466 +465 @@
-#endif
-
-
+1, 
-    return;
-

ext/ohcount_native/ragel_parser.c

@@ -137 +137 @@
-};
-
-
+*
-LanguageBreakdown *get_language_breakdown(char *name) {
-  int i;
@@ -150 +150 @@
-}
-
-
+*
-void ragel_parse_yield_line(const char *lang, const char *entity, int s, int e) {
-  if (rb_block_given_p()) {
@@ -167 +167 @@
-}
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
- *
- * Entities are defined in the parser and are things like comments, strings,
- * keywords, etc.
- * This callback yields for a Ruby block if necessary:
-
+
+
- * @param *lang The language associated with the entity.
- * @param *entity The entity discovered. There are 3 additional entities used
@@ -193 +208 @@
-    lb->blank_count++;
-    ragel_parse_yield_line(lang, entity, s, e);
+  } else {
+    ragel_parse_yield_entity(lang, entity, s, e);
-  }
-}
-
-
+
+
- *
- * @param *parse_result An allocated, empty ParseResult to hold parse results.
+ * @param count An integer flag indicating whether to count lines or parse
+ *   entities.
- * @param *buffer A pointer to the buffer or character in the buffer to start
- *   parsing at.
@@ -205 +225 @@
- * @return 1 if a Ragel parser is found, 0 otherwise.
- */
-
+ int count,
-                       char *buffer, int buffer_len, char *lang) {
-  pr = parse_result;
@@ -214 +234 @@
-  for (i = 0; strlen(languages[i].name) != 0; i++)
-    if (strcmp(languages[i].name, lang) == 0) {
-1
+count
-      return 1;
-    }

ext/ohcount_native/ragel_parser.h

@@ -1 @@
-
+int count, 

ext/ohcount_native/ruby_binding.c

@@ -178 +178 @@
-
-
+/**
+ * Parses a source file's entities (if available).
+ * An entity is each comment, string, number, keyword, etc. that occurs in a
+ * source file.
+ *
+ * You must pass a Ruby block to this function of the form:
+ *   |language, entity, s, e| where:
+ *     language is the language's name (symbol) e.g. ruby.
+ *     entity is the entity's name (symbol) e.g. comment.
+ *     s is the entity's start position in the buffer (number).
+ *     e is the entity's end position in the buffer non-inclusive (number).
+ * If an entity parser is not available for the given language, the block will
+ * never be called. There is currently no way to dynamically test if a language
+ * has an entity parser.
+ *
+ * @param buffer The buffer to parse.
+ * @param language String language name to parse the buffer as. If you are
+ *   unsure which language name is correct, use Ohcount::Detector.detect(file).
+ * @return nil
+ *
+ * @usage
+ *
+ *   # Print each entity and its position in the buffer
+ *   buffer = File.read("helloworld.c")
+ *   Ohcount::parse_entities(buffer, 'c') do |lang, entity, s, e|
+ *     puts "#{lang}\t#{entity}\t#{s}\t#{e}"
+ *   end
+ */
+static VALUE _ohcount_parse_entities(VALUE self, VALUE buffer, VALUE polyglot_name_value) {
+  char *polyglot_name = RSTRING(polyglot_name_value)->ptr;
+  ParseResult pr;
+  if (!ragel_parser_parse(&pr, 0, RSTRING(buffer)->ptr, RSTRING(buffer)->len, polyglot_name))
+    rb_raise(rb_eStandardError,"Polyglot name invalid: '%s'", polyglot_name);
+  return Qnil;
+}
+
+
-void Init_ohcount_native () {
-  rb_module_ohcount = rb_define_module("Ohcount");
-  rb_define_module_function(rb_module_ohcount, "parse", _ohcount_parse, 2);
+  rb_define_module_function(rb_module_ohcount, "parse_entities", _ohcount_parse_entities, 2);
-  rb_define_module_function(rb_module_ohcount, "polyglots", _ohcount_polyglots, 0);
-