about summary refs log tree commit diff stats
path: root/linkify/003linkify.cc
diff options
context:
space:
mode:
authorKartik K. Agaram <vc@akkartik.com>2016-12-26 20:44:10 -0800
committerKartik K. Agaram <vc@akkartik.com>2016-12-26 20:58:37 -0800
commit201458e3bd2f1d79a0ea0b853552e9df267e92b1 (patch)
tree0a4f13662cde7f92ae0bcf641c9733e2a0fcd6ef /linkify/003linkify.cc
parente35c2d6857e1ed916221faae707e3c53ff8ed042 (diff)
downloadmu-201458e3bd2f1d79a0ea0b853552e9df267e92b1.tar.gz
3713 - cross-link calls with definitions in html
Diffstat (limited to 'linkify/003linkify.cc')
-rw-r--r--linkify/003linkify.cc186
1 files changed, 186 insertions, 0 deletions
diff --git a/linkify/003linkify.cc b/linkify/003linkify.cc
new file mode 100644
index 00000000..953036db
--- /dev/null
+++ b/linkify/003linkify.cc
@@ -0,0 +1,186 @@
+// Read a tabular cross-reference file generated by ctags, then read a list of
+// html files generated by Vim's TOhtml command on C++ code. Link words
+// in the html files to cross-references from ctags.
+
+// Still plenty of holes:
+// - unnecessarily linking definition location to itself
+// - can't detect strings in spite of attempt to support them below, because
+//   Vim's generated html turns quotes into html entities
+// - distinguishing function and variable names
+// - distinguishing Mu code in C++ files
+// - distinguishing between function overloads
+//   - if there's duplicate tags we aren't smart enough to distinguish between
+//     them yet, so we simply don't add any link at all
+//   - but even that's not perfect, because sometimes the tags file has a
+//     single definition but there's still multiple overloads (say I defined
+//     'clear()' on some type, and it's already defined on STL classes)
+
+struct syminfo {
+  string filename;
+  int line_num;
+  syminfo() :line_num(0) {}
+};
+
+bool has_data(istream& in) {
+  in.peek();
+  if (in.eof()) return false;
+  assert(in);
+  return true;
+}
+
+bool starts_with(const string& s, const string& pat) {
+  string::const_iterator a=s.begin(), b=pat.begin();
+  for (/*nada*/;  a!=s.end() && b!=pat.end();  ++a, ++b)
+    if (*a != *b) return false;
+  return b == pat.end();
+}
+
+void read_tags(const string& filename, map<string, syminfo>& info) {
+  ifstream in(filename);
+//&   cerr << "reading " << filename << '\n';
+  string dummy;
+  while (has_data(in)) {
+    string symbol;  in >> symbol;
+//&     cerr << symbol << '\n';
+    if (info.find(symbol) != info.end()) {
+      info[symbol].line_num = -1;
+      info[symbol].filename.clear();
+    }
+    else {
+      in >> dummy;
+      in >> info[symbol].line_num;
+      in >> info[symbol].filename;
+    }
+    getline(in, dummy);  // skip rest of line
+//&     cerr << symbol << ": " << info[symbol].filename << ':' << info[symbol].line_num << '\n';
+  }
+  in.close();
+}
+
+void replace_tags_in_file(const string& filename, const map<string, syminfo>& info) {
+//&   cerr << info.size() << " symbols\n";
+  ifstream in(filename);
+  ofstream out(filename+".out");
+  while (has_data(in)) {
+    // send lines that don't start with '<span' straight through
+    string line;
+    getline(in, line);
+    if (!starts_with(line, "<span ")) {
+      out << line << '\n';
+    }
+    else {
+      static int span_size = string("</span>").size();
+      int skip_first_span = line.find("</span>") + span_size;
+      out << line.substr(0, skip_first_span);
+      istringstream in2(line.substr(skip_first_span));
+      in2 >> std::noskipws;
+      while (has_data(in2)) {
+        if (isspace(in2.peek())) {
+//&           cerr << "space\n";
+          char c;  in2 >> c;
+          out << c;
+        }
+        // within a line, send straight through all characters inside '<..>'
+        else if (in2.peek() == '<') {
+//&           cerr << "tag\n";
+          char c = '\0';
+          while (in2 >> c) {
+//&             cerr << "span: " << c << '\n';
+            out << c;
+            if (c == '>') break;
+          }
+//&           cerr << "end tag\n";
+        }
+        else {
+          // send straight through all characters inside strings (handling escapes)
+          char c = in2.get();
+          if (c == '"') {
+//&             cerr << "string\n";
+            out << c;
+            while (in2 >> c) {
+              out << c;
+              if (c == '\\') {
+                in2 >> c;  out << c;
+              }
+              else if (c == '"') {
+                break;
+              }
+            }
+          }
+          else if (c == '\'') {
+//&             cerr << "character\n";
+            out << c;
+            while (in2 >> c) {
+              out << c;
+              if (c == '\\') {
+                in2 >> c;  out << c;
+              }
+              else if (c == '\'') {
+                break;
+              }
+            }
+          }
+          // send straight through any characters after '//' (comments)
+          else if (c == '#') {
+//&             cerr << "comment\n";
+            out << c;
+            while (in2 >> c) out << c;
+          }
+          // send straight through any characters after '//' (comments)
+          else if (c == '/' && in2.peek() == '/') {
+//&             cerr << "comment\n";
+            out << c;
+            while (in2 >> c) out << c;
+          }
+          else {
+//&             cerr << "rest\n";
+            if (c == ',' || c == ':') {
+              out << c;
+              continue;
+            }
+            ostringstream out2;
+            out2 << c;
+            while (in2 >> c) {
+              if (isspace(c) || c == '<' || c == '"' || c == '\'' || c == '/' || c == ',' || c == ':') {  // keep sync'd with other clauses above
+                in2.putback(c);
+                break;
+              }
+              out2 << c;
+            }
+            string symbol = out2.str();
+            if (symbol == "equal" || symbol == "index" || symbol == "put-index" || symbol == "length") {
+//&               cerr << "  blacklisted\n";
+              out << symbol;
+            }
+            else if (info.find(symbol) == info.end()) {
+//&               cerr << "  no info\n";
+              out << symbol;
+            }
+            else {
+              const syminfo& s = info.find(symbol)->second;
+              if (s.filename.empty()) {
+//&                 cerr << "  empty info\n";
+                out << symbol;
+              }
+              else {
+//&                 cerr << "  link\n";
+                out << "<a href='" << s.filename << ".html#L" << s.line_num << "'>" << symbol << "</a>";
+              }
+            }
+          }  // end rest
+        }
+      }  // done parsing line
+      out << '\n';
+    }
+  }
+  in.close();  out.close();
+}
+
+int linkify(int argc, const char* argv[]) {
+  map<string, syminfo> info;
+  read_tags(argv[1], info);
+  for (int i = 2;  i < argc;  ++i) {
+    replace_tags_in_file(argv[i], info);
+  }
+  return 0;
+}